tt-16-84[1]

THEORY, DESIGN AND OPERATION OF NUTRIENTREMOVAL ACTIVATED SLUDGE PROCESSES

GA Ekama • GVR Marais • IP SiebritzAR Pitman • GFP Keay • L Buchan •A Gerber • M Smollen

WRC Report No. TT 16/84

Water Research Commission te

THEORY, DESIGN AND OPERATION OF NUTRIENT

REMOVAL ACTIVATED SLUDGE PROCESSES

A collaborative information document prepared for the

WATER RESEARCH COMMISSION

by the

UNIVERSITY OF CAPE TOWN,

CITY COUNCIL OF JOHANNESBURG and the

NATIONAL INSTITUTE FOR WATER RESEARCH OF THE CSIR

Contributors

Dr G A Ekama - Division of Water Resources and Public Health Engineering, Department of Civil Engineering, University of Cape Town,Prof G v R Marais Private Bag, Rondebosch, 7700Mr I P Siebritz

Mr A R Pitman - City Council of Johannesburg, P.O. Box 1477, Johannesburg, 2000M r G F P KeayDr L Buchan

Mr A Gerber - National Institute for Water Research, CSlR, P.O. Box 395, Pretoria, 0001Mrs M Smotlen

PRETORIA, 1984

(i)

Obtainable from:Water Research CommissionP O Box 824PRETORIA0001

ISBN 0 908356 13 7

(ii)

FOREWORD

The Republic of South Africa is a water-scarce country as poignantly accentuated by the recent (1983) drought. Itis therefore obvious that everything possible must be done to safeguard our water resources against pollution aswell as to ensure their optimal use. Wastewater treatment has an important role to play in both these objectives.

The discharge of treated effluents back to the river of origin as required by the Water Act, No 54 of 1956, hasnecessitated a high standard of wastewater treatment. Although this requirement has been in effect since 1962when the General and Special Standards were promulgated, the General Standard did not require that phosphatesand nitrogen be removed. As a consequence our water bodies have become enriched with these plant nutrients,and this has resulted in the proliferation of water plants and algae which in turn has given rise to problems withwater purification processes, as well as aesthetic and suspected health problems. In an effort to control eutrophica-tion the authorities have promulgated an effluent standard for dissolved ortho-phosphate of 1 mg/f {as P) which isto be enforced in certain critical catchments as from August 1985.

In order to meet the challenge of producing effluents which comply with the phosphate standard as cost-effectively as possible, as well as producing effluents which are eminently suitable for re-use, the Water ResearchCommission initiated research on nutrient removal from wastewaters in 1974. After nearly ten years' research anddevelopment work, significant progress has been made in this field. Results and findings of the research were pro-gressively published via a number of routes, i.e. technology transfer seminars, workshops and scientific publica-tions. To assist the end user to effectively remove nutrients from effluents it was decided to publish a comprehen-sive information publication directed specifically at user organisations such as iocal authorities and consultingengineering firms. This publication presents practical guidelines for the planning, design and operation of biologicalnutrient removal plants. Possibly the best advertisement for the processes described in this publication is Johan-nesburg's 150 M//d Goudkoppies sewage treatment works where the biological nutrient removal process is pro-ducing an effluent which consistently meets the General and Phosphate Standards.

I recommend this publication to you as a basis for consideration for the design and operation of new plants aswell as the modification and operation of existing plants to meet the General and Phosphate Standards. I furtherextend an invitation to you to consult with experts in this field in the RSA on any site specific problems you may beencountering in process design and operation to meet effluent standards.

VI R HENZENCHAIRMAN: WATER RESEARCH COMMISSION

ACKNOWLEDGEMENTS

Acknowledgements are due to the Co-ordinating Research and Development Committee for Sewage Purificationwho guided the research, the Editorial Panel who guided the writing of this publication and a number of localauthorities and consulting engineering firms for their constructive comments and suggestions received on drafts ofthis document.

Co-ordinating Research and Development Committee for Sewage Purification

D r H N S Wiechers (Chairman)Mr H G J BeekmanMr H J BestDr G G CillieMr G J R EngelsMr G B GordonMr J E HeltingsMr A J Hu5selmannMr W A LombardMr A S LouwProf G v R MaraisMr D W OsbornMr P E OdendaalProf D J J PotgieterMr B StanderMr D F ToerienDr L R J van Vuuren

Editorial Panel

Dr H N S Wiechers (Chairman)Dr G A EkamaMr A GerberMr G F P KeayMr W MalanProf G v R MaraisMr D W OsbornMr A R PitmanProf D J J PotgieterProf W A Pretorius

Water Research CommissionCity Council of Cape TownDepartment of Environment AffairsNational Institute for Water Research, CSIRCity Council of VereenigingCity Council of JohannesburgCity Council of GermistonRand Water BoardS.A. Bureau of StandardsCity Council of BoksburgUniversity of Cape TownCity Council of JohannesburgWater Research CommissionUniversity of PretoriaCity Council of PretoriaTransvaal Provincial AdministrationNational Institute for Water Research, CSIR

Water Research CommissionUniversity of Cape TownNational Institute for Water Research, CSIRCity Council of JohannesburgUniversity of StellenboschUniversity of Cape TownCity Council of JohannesburgCity Council of JohannesburgUniversity of PretoriaUniversity of Pretoria

(iv)

CONTENTS

Page

FOREWORD iii

ACKNOWLEDGEMENTS iv

CONTENTS v

CHAPTER ABSTRACTS AND DETAILED TABLES OF CONTENTS vi through xvi

INTRODUCTION xvii

CHAPTER 1: FUNDAMENTALS OF BIOLOGICAL BEHAVIOUR 1-1Detailed Table of Contents vi

CHAPTER 2: THE NATURE OF MUNICIPAL WASTEWATERS 2-1Detailed Table of Contents vi

CHAPTER 3: INFLUENT WASTEWATER CHARACTERISTICS AND THEIR INFLUENCE ON DESIGN.. . 3-1Detailed Table of Contents vii

CHAPTER 4: CARBONACEOUS MATERIAL REMOVAL 4-1Detailed Table of Contents viii

CHAPTER 5: NITRIFICATION 5-1Detailed Table of Contents ix

CHAPTER 6: BIOLOGICAL NITROGEN REMOVAL 6-1Detailed Table of Contents x

CHAPTER 7: BIOLOGICAL EXCESS PHOSPHATE REMOVAL 7-1Detailed Table of Contents xi

CHAPTER 8: SECONDARY SETTLING TANKS 8-1Detailed Table of Contents xii

CHAPTER 9: MICROBIOLOGICAL ASPECTS 9-1Detailed Table of Contents xiii

CHAPTER TO: PRACTICAL DESIGN CONSIDERATIONS 10-1Detailed Table of Contents xiv

CHAPTER 11: OPERATION OF BIOLOGICAL NUTRIENT REMOVAL PLANTS 11-1Detailed Table of Contents xv

CHAPTER 12: CASE STUDIES OF BIOLOGICAL NUTRIENT REMOVAL 12-1

Detailed Table of Contents xvi

APPENDIX 1 LIST OF SYMBOLS A1-1

APPENDIX 2 BRIEF DESCRIPTION OF THE DETERMINATION OF THE READILY BIODEGRADABLECOD FRACTION OF A MUNICIPAL WASTEWATER A2-1

APPENDIX 3 BRIEF DESCRIPTION OF THE DETERMINATION OF THE MAXIMUM SPECIFIC GROWTH

RATE OF THE NITRIFIERS A3-1

APPENDIX 4 LIST OF RSA BIOLOGICAL NUTRIENT REMOVAL PLANTS A4-1

APPENDIX 5 PHOTOMICROGRAPHS OF BACTERIA FOUND IN NUTRIENT REMOVAL ACTIVATEDSLUDGE PROCESSES A5-1

(v)

ABSTRACTS AND DETAILED TABLES OF CONTENTS

CHAPTER 1

FUNDAMENTALS OF BIOLOGICAL BEHAVIOUR

by

G. v. R. Marais and G.A. Ekama

ABSTRACT

The basic biological behaviour of bacterial growth and energy abstraction is briefly discussed. The various testparameters i.e. the COD, TOD, BOD and TOC, whereby it is attempted to measure the energy available inwastewaters for biological growth are critically examined. It is shown that while the other tests are useful, the CODtest is the best from a biological behaviour point of view. Integration of the COD test parameter into themechanism of biological behaviour is presented.

TABLE OF CONTENTSPage

1. INTRODUCTION 1-1

1.1 Objectives of wastewater treatment 1-2

2. BIOLOGICAL BEHAVIOUR 1-2

2.1 Electron donors and acceptors 1-2

3. CARBONACEOUS ENERGY MEASUREMENT 1-4

3.1 Chemical Oxygen Demand test 1-43.2 Total Oxygen Demand test 1-53.3 Biochemical Oxygen Demand test 1-53.4 Total Organic Carbon test 1-6

4. COD'VSS RATIO 1-7

5. REFERENCES 1-8

CHAPTER 2

THE NATURE OF MUNICIPAL WASTEWATERS

by

G.A. Ekama and G.v.R. Marais

ABSTRACT

The characteristics of municipal wastewaters with respect to their chemical and physical constituents are describ-ed. Under chemical characterization, the different fractions of COD, TKN and total P are defined; under physicalcharacterization, the dissolved, colloidal, suspended and settleable fractions. Approximate ranges of importantwastewater characteristics for normal raw and settled municipal wastewaters in South Africa are given. The effectof primary sedimentation on the wastewater characteristics is briefly discussed.


1. WASTEWATER CHARACTERIZATION 2-1

(vi)

2. CHEMICAL CHARACTERIZATION 2-1

2.1 Carbonaceous materials 2-12.1.1 Biodegradable and unbiodegradable fractions 2-12.1.2 Unbiodegradable fractions 2-22.1.3 Biodegradable fractions 2-22.1.4 Analytical formulation 2-2

2.2 Nitrogenous materials 2-32.2.1 Free and saline and proteinaceous fractions 2-32.2.2 Analytical formulation 2-52.2.3 Specific growth rate of the nitrifiers 2-5

2.3 Phosphorus 2-5

3. PHYSICAL CHARACTERIZATION 2-5

4. QUANTIFICATION OF WASTEWATER CHARACTERISTICS 2-7

5. REFERENCES 2-8

CHAPTER 3

THE INFLUENCE OF WASTEWATER CHARACTERISTICS ON PROCESSDESIGN

by


ABSTRACT

This chapter gives a brief qualitative overview of how the more important characteristics of a wastewater - totalCOD, TKN and P concentrations, COD fractions, TKN/COD, P!COD ratios, maximum specific growth rate of thenitrifiers and temperature, raw or settled wastewater - affect the process design, performance and operation of abiological nutrient removal activated sludge process with respect to sludge age, process volume, anaerobic andanoxic sludge mass fractions, process configutation, oxygen demand, and effluent quality.


1 . INTRODUCTION 3-1

2 . RELATIVE IMPORTANCE OF CHARACTERISTICS 3-1

3 . EFFECTS OF CHARACTERISTICS ON DESIGN 3-13.1 Influent COD and influent flow 3-13.2 Influent TKN, jjnrT.T and temperature 3-23.3 Readily biodegradable COD, slowly biodegradable COD 3-2

3.3.1 Nitrogen removal 3-23.3.2 Excess phosphorus removal 3-63.3.3 Implication on process development 3-8

4. CLOSURE 3-10

5. REFERENCES 3-10

(vii)

CHAPTER 4

ORGANIC MATERIAL REMOVAL

by


ABSTRACT

The steady state equations for the activated sludge process for organic material removal only are presented. Usingthese equations as a basis, the changes in composition of the sludge (active, endogenous and inert fractions) withchange in sludge age are demonstrated. Equations for calculating the process volume requirements, daily averageoxygen demand, daily mass of sludge produced, nitrogen and phosphorus requirements for sludge production aregiven and the changes in these parameters with change in siudge age, temperature and wastewater characteristicsare discussed in detail. Hydraulic control of sludge age as the process control parameter, is compared with the loadfactor approach. Practical considerations regarding the selection of the sludge age are given. The design procedureis demonstrated with worked examples for raw and settled wastewater respectively.


1. INTRODUCTION 4-1

2. BIOLOGICAL KINETICS 4-1

3. PROCESS KINETICS 4-2

3.1 Mixing regimes 4-23.2 Sludge age 4-23.3 Nominal hydraulic retention time 4-33.4 Effluent COD concentration 4-3

4. PROCESS DESIGN EQUATIONS 4-3

5. STEADY STATE DESIGN CHART 4-5

6. PROCESS VOLUME REQUIREMENTS 4-6

7. CARBONACEOUS OXYGEN DEMAND 4-7

8. DAILY SLUDGE PRODUCTION 4-8

9. NUTRIENT REQUIREMENTS FOR SLUDGE PRODUCTION 4-9

10. PROCESS DESIGN AND CONTROL 4-10

10.1 Sludge age vs load factor as control parameter 4-1010.2 Process control 4-1110.3 Hydraulic control of siudge age 4-13

11. SELECTION OF SLUDGE AGE 4-13

11.1 Short sludge ages (1 to 5 days) 4-1311.2 Intermediate sludge ages (10 to 15 days) 4-1411.3 Long sludge ages (20 days or more) 4-15

12. DESIGN EXAMPLE 4-17

12.1 Influent wastewater characteristics 4-1712.2 Temperature effects 4-1712.3 Calculations for organic material degradation 4-18

13. CONCLUSIONS 4-19

14. REFERENCES 4-20

(viii)

CHAPTER 5

NITRIFICATION

by


ABSTRACT

After a brief discussion of the kinetics of nitrification, the steady state process equations for nitrification arepresented. The effect of incorporating nitrification in a design on the selection of the sludge age is discussedtogether with the effects of nitrification on the alkalinity and pH of the wastewater. For nitrification-denitrificationprocesses, the effect of the unaerated zones on nitrification is discussed and an important parameter in the designof these processes i.e. the nitrification capacity is defined and demonstrated for raw and settled wastewater. Thebehaviour of nitrification for typical raw and settled wastewaters with changes in temperature and sludge age andthe effect of nitrification on the daily average oxygen requirement is demonstrated with the aid of worked ex-amples.


1 . INTRODUCTION 5-1

2. BIOLOGICAL KINETICS 5-1

2.1 Growth 5-12.2 Endogenous respiration 5-22.3 Behavioural characteristics 5-2

3. PROCESS KINETICS 5-2

3.1 Effluent ammonia concentration 5-2

4. FACTORS INFLUENCING NITRIFICATION 5-4

4.1 Influent source 5-44.2 Temperature 5-44.3 pH and Alkalinity 5-44.4 Unaerated zones 5-74.5 Dissolved oxygen concentration 5-94.6 Cyclic flow and load 5-10

5. DESIGN CONSIDERATIONS 5-10

5.1 Fate of the influent TKN 5-105.2 Effluent TKN 5-115.3 Nitrification capacity 5-11

6. DESIGN EXAMPLE 5-13

6.1 Effects of nitrification on mixes liquor pH 5-136.2 Minimum sludge age for nitrification 5-146.3 Raw wastewater 5-146.4 Settled wastewater 5-156.5 Nitrification process behaviour 5-15

7. REFERENCES 5-17

(ix)

CHAPTER 6

BIOLOGICAL NITROGEN REMOVAL

by


ABSTRACT

The kinetics of biological denitrification and the different activated sludge process configurations in which it can beachieved, are discussed. Experimental work on laboratory and pilot scale processes is presented and the develop-ment of the general dynamic nitrification-denitrification process model is described. For constant flow and loadconditions, simple design equations are given with which the behaviour of the different denitrification processescan be analysed for different wastewater characteristics. These equations are cast into a step by step design pro-cedure. This design procedure is demonstrated for raw and settled wastewaters with the aid of worked examples.


1. PROCESS FUNDAMENTALS 6-1

1.1 Nitrogen removal by sludge production 6-11.2 Biological denitrification _ 6-11.3 Effect of denitrification on oxygen demand 6-11.4 Effect of denitrification on Alkalinity 6-2

2 .REQUIREMENTS FOR DENITRIFICATION 6-2

2.1 Presence of nitrate 6-22.2 Absence of dissolved oxygen 6-22.3 Facultative bacterial mass 6-22.4 Electron donor (Energy source) 6-2

3. DENITRIFICATION PROCESS CONFIGURATION 6-3

3.1 Wuhrmann configuration 6-33.2 Ludzack-Ettinger configuration 6-33.3 Bardenpho configuration 6-4

4. EXPtRlMENTAL BASIS FOR DENITRIFICATION KINETICS * 6 4

4.1 Constant flow and load conditions 6-54.1.1 Experimental work with plug flow reactors 6-54.1.2 Nitrate removal and substrate utilization 6-94.1.3 Completely mixed anoxic reactors 6-9

4.2 Cyclic flow and load conditions 6-10

5. DENITRIFICATION POTENTIAL 6-11

5.1 Denitrification potential of the primary anoxic reactor 6-115.2 Denitrification potential of the secondary anoxic reactor 6-12

6. DESIGN PROCEDURE FOR N REMOVAL PROCESSES 6-12

6.1 Principles of the design procedure 6-136.2 Can complete denitrification be achieved? 6-136.3 When complete denitrification cannot be achieved 6-16

7. DESIGN EXAMPLES 6-16

7.1 Variability of wastewaster characteristics 6-167.1.1 Variability of nitrification rate 6-167.1.2 Variability of denitrification rates 6-177.1.3 Variability of readily biodegradable COD fraction 6-17

7.2 Effect of temperature 6-187.3 Maximum unaerated sludge mass fraction 6-187.4 Nitrification 6-19

(x)

7.5 Denitrification 6-207.6 Process volume and oxygen demand 6-23

7.6.1 Process volume 6-237.6.2 Daily average total oxygen demand 6-24

7.7 Closure 6-25

8. REFERENCES 6-26

CHAPTER 7

BIOLOGICAL EXCESS PHOSPHORUS REMOVAL

by

G.A. Ekama, G.v.R. Marais and I.P. Siebritz

ABSTRACT

The evolution of a quantitative parametric model for biological excess phosphorus removal is described and it isshown that the readily biodegradable COD fraction of the influent is one of the most important parameters in ex-cess P removal. The equations of the model, which allow prediction of the magnitude of excess P removal in termsof the process design parameters and wastewater characteristics, are presented together with their capabilities andlimitations. The design procedure is critically discussed in the light of recent research findings in the biological ex-cess P removal field. The implications of the model on nutrient removal process design is described. Step by stepdesign procedures for the different nutrient removal processes are set out and a design chart by means of whichthe excess P removal can be rapidly estimated is presented. Rough guidelines for process selection and anaerobicreactor sizing based on wastewater characteristics are set out. The design procedures are demonstrated with theaid of worked examples for raw and settled wastewaters.


1. INTRODUCTION 7-1

2. BACKGROUND 7-1

2.1 Earlier developments 7-12.2 Later developments 7-3

2.2.1 Positive exclusion of nitrate from the anaerobic reactor 7-32.2.2 Phosphorus removal process analyses 7-4

3. EVOLUTION OF EXCESS PHOSPHORUS REMOVAL MODEL 7-5

3.1 Prerequisites for excess phosphorus removal 7-53.2 Magnitude of excess phosphorus removal 7-83.3 Quantitative model for excess phosphorus removal 7-83.4 Comments on the parametric model and recent research developments 7-12

4. IMPLICATIONS OF THE MODEL ON DESIGN 7-16

4.1 Coefficient of excess P removal 7-164.2 Excess P removal propensity factor 7-164.3 Excess P removal Model 7-16

5. DESIGN PROCEDURE 7-17

6. DESIGN CHART 7-21

7. DESIGN EXAMPLES 7-23

7.1 The Phoredox Process 7-237.2 The UCT Process 7-24

(xi)

7.3 Comparison of Phoredox and UCT Processes 7-257.4 Redesign of the UCT Process 7-267.5 Process volume 7-287.6 Oxygen demand 7 29

8. GUIDELINES FOR PROCESS SELECTION 7-29

8.1 Process selection 7-298.2 Anaerobic sludge mass fraction selection 7-308.3 Subdivision of anaerobic reactor 7-30

9. CLOSURE AND CONCLUSIONS 7-30

10. REFERENCES AND BIBLIOGRAPHY 7-32

CHAPTER 8

SECONDARY SETTLING TANK DESIGN

BY

G.A. Ekama, A.R. Pitman, M. Smollen and G.v.R. Marais

ABSTRACT

Settling behaviour of mixed liquor as described by the stirred batch settling test is discussed. Batch settling testdata are then integrated with secondary settling tank behaviour via the flux theory using both graphical andanalytical procedures. A secondary settling tank design and operating chart is presented and its use demonstratedby worked examples. Practical aspects of secondary settling tank design are discussed.


1. INTRODUCTION 8-1

2. SECONDARY SETTLING TANK DESIGN 8 2

3. STIRRED BATCH SETTLING TEST 8-2

4. SOLIDS FLUX 8-3

5. GRAPHICAL APPLICATION OF FLUX THEORY 8-3

5.1 Graphical procedure 8-35.2 Concentration profiles 8-75.3 Dynamic conditions 8-85.4 Discussion 8-9

6. ANALYTICAL APPLICATION OF FLUX THEORY 8-9

6.1 Empirical equation for settling velocity versus sludge concentration 8-96.2 Mathematical properties of theoretical flux equation 8-106.3 Application of flux equation to settling tanks 8-106.4 Settling tank design and operating chart 8-116.5 Design example 8-12

7. PRACTICAL ASPECTS OF FINAL CLARIFIER DESIGN 8-13

8. REFERENCES 8 14

Ixii)

CHAPTER 9

MICROBIOLOGICAL ASPECTS

by

L. Buchan

ABSTRACT

The types of organisms present in sewage processes are described with emphasis placed on their mutual in-terdependence, nutritional requirements and competition for predominance. Aerobic and anaerobic respiration areintroduced with environmental conditions having a marked effect on organism selection and growth. The effect ofsewage as a substrate for organism growth and the succession of organisms in activated sludge processes arediscussed with the changes in biological populations caused by anaerobic and anoxic conditions being importantfactors in biological nutrient removal. Bacteria which accumulate excess phosphate in the form of insolublepolyphosphate granules appear to predominate in biological phosphorus removal plants and many of these havebeen identified as Acinetobacter sp. Various photomicrographs of the predominant bacteria and protozoa inthese plants are also shown (See Appendix 5).


1. EUKARYOTIC AND PROKARYOTIC CELLS 9-1

2. PROTISTS IMPORTANT IN ACTIVATED SLUDGE 9-1

2.1 Protozoa 9-12.1.1 Metabolism and growth 9-1

3. PROKARYOTES 9-1

3.1 Nutritional requirements 9-23.1.1 Carbon 9-23.1.2 Nitrogen and sulphur 9-23 1.3 Growth factors 9-2

3.2 Bacterial energetics 9-23.2.1 Aerobic respriration 9-23.2.2 Anaerobiosis 9-2

3.3 Summary op nutritional categories among bacteria of importance in activated sludge 9-33.3.1 Chemoautotrophs 9-33.3.2 Chemoheterotrophs 9-3

4. SEWAGE AS A SUBSTRATE FOR MICRO-ORGANISMS 9-3

5. ORGANISMS PRESENT IN SEWAGE AND THEIR FATE IN ACTIVATED SLUDGE 9-3

6. SELECTIVE PRESSURES AND THE SUCCESSION OF ORGANISMS IN THE ACTIVATED SLUDGEENVIRONMENT 9-4

6.1 General 9-46.2 Selection of organisms in an aerobic environment 9-46.3 Selection of organisms in a nitrogen removing anoxic/aerobic system 9-46.4 Selection of organisms in an anaerobic/anoxic/aerobic system 9.4

PHOTOMICROGRAPHS see Appendix 5

( XIII I

CHAPTER 10

PRACTICAL DESIGN CONSIDERATIONS

by

GF.P. Keay

ABSTRACT

This chapter lays emphasis on the more practical problems involved in the design of biological nutrient removalplants. Process input parameters such as sewage characteristics and dissolved oxygen input are considered, as areinstrumentation and the various methods of direct and indirect oxygen transfer. The need for backup facilities isemphasized, while the methods of scum handling and waste activated sludge disposal are discussed. Points toconsider in the design of the various zones in the process and final clarifiers are also mentioned. Finalfy, theeconomics of the process, the merits of load balancing and staffing requirements also receive attention.


1. INTRODUCTION 10 1

2. PLANT CONTROL 10 1

2.1 Process input parameters 10 12.2 Process output parameters 10-12.3 Other process considerations 10-2

3. INSTRUMENTATION 10-2

3.1 Flow metering 10-23.2 Oxygenation control 10-23.3 Temperatures 10-23.4 Energy consumption 10-33.5 Sludge blanket sensors 10-33.6 Sampling 10 33.7 Running hour meters 10 3

4. DISSOLVED OXYGEN CONTROL '. 10-3

4.1 Indirect oxygen transfer 10-34.2 Aeration systems 10-4

4.2.1 Mechanical surface aerators 10-44.2.2 Fine bubble diffused aeration 10-44.2.3 Turbine aeration system 10-44.2.4 Pure oxygen systems 10-4

5. HYDRAULICS 10-4

6. "BACKUP" FACILITIES 10-5

7. SCUM HANDLING AND CONTROL 10-5

8. TREATMENT OF WASTE ACTIVATED SLUDGE 10-5

9. SECONDARY CLARIFIERS 10-6

10. BIOLOGICAL REACTOR - 10-6

10.1 Anaerobic zone 10-710.2 Anoxic zone 10-710.3 Aerobic zones 10-7

11. ECONOMICS 10-7

11.1 Typical treatment costs for a biological nutrient removal plant 10-7

(xiv)

11.2 Selection of process configuration 10-811.3 Effect of equalization (flow balancing) on the sizing of units and associated cost implications 10-8

11.3.1 Aeration 10-811.3.2 Biological reactor 10-811.3.3 Secondary sedimentation 10-911.3.4 Pipework, channels, etc 10-911.3.5 Downstream processes 10 911.3.6 Cost implications 10-9

11.4 Effect of load balancing 10-911.5 Full and partial nitrification and denitrification 10 9

12. STAFFING 10-10

ANNEXURES

1. Calculated aerator performance 10-102. Calculation of electricity costs 10-113. Secondary sedimentation requirements without flow equalization 10-114. Cost implications of flow equalization 10 12

CHAPTER 11

OPERATION OF BIOLOGICAL NUTRIENT REMOVAL PLANTS

by

A.R. Pitman

ABSTRACT

Based on lull-scale experience, this chapter gives a guide to the operation of biological nutrient removal plants. Itconsiders sewage characteristics, phosphate release and uptake and desirable operating conditions in variouszones and final clanfiers. The problems of nitrate feedback and unwanted oxygen inputs to anaerobic zones alsoreceive attention. Remedial measures, monitoring procedures and control of unwanted microbial growths arediscussed and typical performance results presented. A method of problem solving via an algorithm is proposed,while the treatment and disposal of waste activated sludge and process liquors are covered. Finally, the implica-tions of complying with effluent Standards are pointed out.


1. BACKGROUND 11-1

2. INFLUENT SEWAGE 11-1

2.1 Characteristics 11-12.2 Storm flows 11-12.3 Improving feed characteristics 11-2

3. BASIC CONCEPTS IN BIOLOGICAL PHOSPHORUS REMOVAL 11-2

3.1 Ability to release phosphate 11-23.2 Phosphorus uptake ability 11-3

4. OPERATING CONDITIONS IN VARIOUS ZONES AND CLARIFIERS 11-3

4.1 Anaerobic zone 11-34.1.1 Nitrate feedback 11-34.1.2 Oxygen input to anaerobic zone 11-44.1.3 Retention time 11-4

4.2 Anoxic zone or primary anoxic zone 11-54.3 Aerobic zone 115

(xv)

4.4 Second anoxic zone 11*74.5 Final aerobic zone 11-84.6 Final clarifiers H-8

5. SLUDGE AGE CONTROL 11-8

6. REMEDIAL MEASURES WHEN NUTRIENT REMOVAL FAILS 11-9

6.1 Supplemeting COD 11-96.2 Chemical addition 11-9

7. UNWANTED MICROBIOLOGICAL GROWTHS 11-9

7.1 Nuisance scums 11-97.2 Filamentous growths 11-10

8. MONITORING 1M0

8.1 Typical results of chemical monitoring 11-11

9. SCHEME FOR PROBLEM SOLVING IN NUTRIENT REMOVAL PLANTS 11-12

10. TREATMENT OF WASTE ACTIVATED SLUDGE 11-12

11. SLUDGE DISPOSAL 11-14

12. DISPOSAL OF LIQUORS 1M4

13. COMPLIANCEWITH STANDARDS 11-14

14. REFERENCE 11-16

CHAPTER 12

SUMMARY OF CASE STUDIES ON BIOLOGICAL NUTRIENT REMOVALAT:

I. JOHANNESBURG'S SEWAGE WORKS by A.R. PITMAN

II. THE NIWR DASPOORT PILOT PLANT by A. GERBER

ABSTRACT

Two summaries of case studies of biological nutrient removal are presented, i.e. full-scale experience at Johan-nesburg's Goudkoppies and Northern VSorks and pilot scale experience at the National Institute for WaterResearch's Daspoort experimental pilot plant. Reference is made to reports in which detailed information on theabovementioned case studies is given, as vsell as to other case studies which have been reported in the literature.


1. INTRODUCTION 12-1

2. CASE STUDY NO I:Operating experiences with nutrient removal activated sludge plants in Johannesburg 12-1

3. CASE STUDY NO II:Biological nutrient removal at the Daspoort sewage works pilot plant 12-4

4. REFERENCES 12-5

(xvi)

INTRODUCTION

The Republic of South Africa is rapidly approaching thepoint of maximum economic exploitation of conven-tional water resources. In addition, the country is facedwith deteriorating water quality which may become alimiting factor in water resources development prior tolimitations on water quantity.

The accelerated deterioration of water quality is aresult of, amongst others, the discharge of ever-increasing quantities of treated effluents to rivers andstreams. Effective removal of pollutants fromwastewalers, particularly the nutrients carbon, nitrogenand phosphorus, is therefore of the utmost importance.Furthermore, high quality effluents may serve as aneconomic source to augment dwindling waterresources.

The need for pollution control and indirect reuse oftreated wastewater is legally expressed in the Water Act(54 of 1956) and standards promulgated in terms of theAct. The General and Special Standards which werepromulgated in 1962, emphasized the adequate removalof carbonaceous material and the transformation of am-monia to nitrate. More recently accelerated eutrophica-tion of many of South Africa's impoundments, with at-tendant health, water quality and aesthetic problems,has become apparent and resulted in the promulgationin 1980 of an upper limit for effluent phosphates for anumber of critical catchments. Furthermore, limitingthe total dissolved solids content of treated effluents isnow also considered of prime importance. The need toremove nutrients and reduce the increase in dissolvedsolids during treatment, places difficult demands onfuture wastewater treatment plants.

Technology for the removal of the nutrientsnitrogen and phosphorus from wastewater has beendeveloped to a high level in the Republic of SouthAfrica. Many organizations have been instrumental inthis development, including research organizations,universities, consulting engineering firms and localauthorities. A wide range of physical, chemical andbiological processes and combinations of these are nowavailable both for upgrading of existing works anddesign of new works. Methods available are: Physical-chemical, of particular use for removal of phosphatesfrom the effluents of existing works - the only viableoption, in fact, for dealing with biological filter ef-fluents*; physical-chemical-biological* *, for theremoval of nitrogen and phosphates, of use where thecharacteristics of the wastewater are unfavourable forthe biological removal of the nutrients; biological, forthe removal of all or major fractions of the nitrogen andphosphates in the influent, no or only minimal additionof chemicals is needed to produce the required effluentphosphate quality. Biological nutrient removal is par-ticularly attractive for the treatment of municipalwastewaters because treatment costs are generally less

than with physical-chemicai methods, the salt concen-tration of the effluent is not raised as with chemicaladdition end the characteristics of the wastewaterusually are amenable to this type of treatment. In conse-quence, the Water Research Commission has beengreatly interested in developing this method of nutrientremoval.

The Water Research Commission has stimulated,co-ordinated and financed research and developmentwork in the field of biological nutrient removal in the ac-tivated sludge process from about 1973. Since that timeit has contracted the National Institute for WaterResearch of the Council for Scientific and IndustrialResearch, the Universities of Cape Town and Pretoriaand the City Council of Johannesburg to do researchand development work in this field on a co-ordinatedjoint venture basis. A significant amount of informationon processes for biological nutrient removal is nowavailable. Much of this information has already beenpublished in local and international journals as well ashaving been presented at a number of conferences,seminars and open days. However, in the light of theurgent need of local authorities and others who have tomeet ever stricter effluent standards, and in particularthe effluent phosphate standard of 1 mg/f soluble or-thophosphate which is to be strictly enforced in anumber of critical catchments from August 1985, it wasdecided to compile a comprehensive information docu-ment on the biological removal of nitrogen andphosphates covering relevant results and findings madeunder Water Research Commission sponsored projects.

This publication has been compiled as a self-contained document which does not require referenceto other publications on wastewater treatment. It is in-tended primarily for the design engineer and manage-ment staff responsible for operation and control ofwastewater purification works. The level of presenta-tion assumes that the reader has had tertiary trainingand/or considerable practical experience in the field ofwastewater treatment.

The basic process for the simultaneous biologicalremoval of phosphates and nitrogen was proposed byBarnard in 1976 and is called the Phoredox process inSouth Africa, and the Bardenpho or Modified Barden-pho in the United States. This process belongs to thesingle sludge, multi-reactor group of processes. For thephosphate removal aspect the fundamental principleembodied in this process is that an anaerobic stateneeds to be created at some point in the process in sucha way that phosphate is released, a consequence ofwhich is that biological uptake of phosphate in excessof normal metabolic requirements is induced when thesludge is aerated subsequently. In the Phoredox pro-cess the endeavour is to create the requisite anaerobicstate by mixing the influent waste stream with the

'Phosphate removal by chemical addition: the reader is referred to the Water Research Commission publication "Guidelines for chemicalremoval of phosphates from municipal effluents."

'For example, the LFB process, the Lime-Fiotation-Biological process developed by the National Institute for Water Research, Pretoria, RSA.

(xvii)

sludge recycled from the secondary settling tankwithout aeration in an anaerobic tank at the head of theworks.

This method for promoting the anaerobic state ap-pears in a number of processes that developed from thePhoredox process in order to accommodate waste-waters with unfavourable characteristics or in whichnitrification-denitrification is not required. One suchdevelopment, known as the UCT process, is describedin detail in this publication. Some of the other processeswhich also incorporate the Phoredox principle, are notdealt with, for example, the AO process is not con-sidered since this process is designed to removephosphate in short sludge age systems where nitrifica-tion is prevented from occurring, and is unlikely to findsignificant application in the RSA. The processdeveloped by Roberts and Kerdachi at Pinetown, Natal,also is not dealt with. This process differs from theothers in that nitrification, denitrification and theanaerobic state, TO induce P release, all take place inone reactor. Also, the process is oxygen-limited, in con-sequence, its biological response with regard to sludgeproduction, oxygen demand, nitrification, etc. is notamenable to description by the kinetic models govern-ing non-oxygen-limited processes such as the multi-reactor Phoredox, UCT and AO processes. The writersof this manuscript felt therefore that the Roberts-Kerdachi process falls outside their ambit of com-petence.

All the above processes are of the main-streamkind. Side-stream processes, of which the Phostrip pro-cess is currently the most widely used, are not discuss-ed in this publication. In side-stream processes theanaerobic state is created by passing a fraction of thesecondary settler underflow through an anaerobic reac-tor to release phosphates through endogenous respira-tion (instead of by the addition of feed waste flow) andprecipitating the released phosphates by lime addition.Insufficient data on the behaviour of this process wasavailable to the authors to allow directives to be put for-ward for design and operation of this process for pro-

ducing effluents complying with South African effluentregulations with regard to both nitrogen andphosphates.

Guidelines for design set out in this publication arebased primarily on about ten years of extensivelaboratory investigations at the University of CapeTown on biological nutrient removal from predominant-ly domestic wastewaters. These guidelines were verifiedon data generated at pilot scale on the National Institutefor Water Research's facility at Daspoort, Pretoria, andat full-scale on the Johannesburg Goudkoppies andNorthern Works. The mathematical model for car-bonaceous material oxidation, nitrification anddenitrification have been shown to give accurate predic-tions of full scale plant response. The theory forbiological phosphate removal is currently still in adevelopment stage with research in this area continu-ing. Nontheless, tentative but conservative guidelinesfor process selection and design of biological phosphateremoval aspects are presented.

Biological nutrient removal in the activated sludgeprocess has only a short history of application. Current-ly in South Africa there are of the order of thirty plantsdesigned for and/or being operated to obtain biologicalnutrient removal, with varying degrees of success.Practical experience on these plants should increaseknowledge on the application of the process to a varietyof wastewaters under different operating conditionsand effluent requirements. Further research is still need-ed and is continuing in a number of areas whereknowledge is still inadequate, for example, themechantsm(s) and factors controlling excess phosphateremoval, the role of the anaerobic stage in the process,settling characteristics of the sludge formed in the pro-cess and factors which control these, as welt as scumformation. This publication therefore does not pretendto be the fina! word on biological nutrient removal; it isan information document with interim guidelines fordesign and operation of biological nutrient removalplants.

H N S WIECHERSWATER RESEARCH COMMISSIONPRETORIA, 1984

(xviii)

CHAPTER ONE

FUNDAMENTALS OF BIOLOGICAL BEHAVIOUR

G.v.R. Marais and G.A. Ekama

1. INTRODUCTION

Life, in order to persist, requires a continuous through-put of (1) matter and (2) energy:

(1) The most prominent elements passing through thelife system are hydrogen (H), oxygen (0), carbon(C), nitrogen (N), phosphorus (P) and sulphur (S),plus a host of minor elements.

(2) Energy is derived principally from three sources andthese form a convenient criterion to categorise theorganisms implicated:

(a) Solar radiation: photo-synthetic autotrophs

(b) Organic compounds: heterotrophs

(c) Inorganic compounds: chemical autotrophs.

(a) In the biosphere the fundamental source ofenergy is solar radiation. The photo-syntheticautotrophic cells fix a small fraction of the solarenergy by forming complex high energy organic(H, C, 0) compounds and producing oxygen. Ofthe matter requirements the elemenis H and 0 areobtained from H20; carbon from C02; andphosphorus, nitrogen, sulphur and the minorelements from the dissolved salts of theseelements. Both H20 and C02 are readily available;sulphur is usually in adequate supply; however theavailability of phosphorus and nitrogen usually islimited - phosphorus does not occur readily insoluble form and nitrogen is useful to the organismprincipally in the NH, and N03 forms. Therestricted availability of these two elements usuallyconstitutes the limiting factor to the mass ofautotrophic life a body of water can generate. Forthis reason, phosphorus and nitrogen are termedeutrophic (life-giving) substances. Phosphorus islimited by the mass available or entering a par-ticular ecosystem, but ammonia can be generatedfrom dissolved molecular nitrogen by certain micro-organisms. In this respect, the control ofphosphorus in the body of water takes on a greatersignificance than the control of nitrogen.

Whereas the eutrophic elements N and P nor-mally are the principal agents of pollution in bodiesof water, by stimulating autotrophic growth, theaction associated with this growth, of producingoxygen, is utilized in the oxidation pond system ofwastewater treatment - the oxygen is utilized byheterotrophic organisms to metabolize the car-bonaceous material in the influent to the pond, in

this fashion to assist in destroying organic energy,as described in (b) below.

(b) The high energy organic compounds synthesiz-ed by the autotrophs form the basic source ofenergy for the heterotrophic cells to synthesize themore complex molecules that constitute the cellmass, including proteins. Only a fraction of theenergy utilized is incorporated in the eel! massgenerated, the balance is lost as heat. The mass oforganisms thus generated in turn form the matterand energy sources (prey) for other organisms thatlive on them (predators), and these in turn becomeprey to yet other predators. Each prey-predatortransformation is accompanied by a substantialloss of energy; in this fashion through the sequen-tial chain of life there is a continuous reduction ofthe organically bound energy originally fixed by thephoto-synthetic autotrophs. When the organicenergy reduces to zero heterotrophic life ceases. Inwastewater treatment plants the full chain of lifedoes not develop so that a substantial fraction ofthe input energy is retained in the life mass in thesystem but this fraction is physically separatedfrom the liquid in the sludge wasted daily (for fur-ther treatment) leaving very little organic energy inthe effluent.

Presence of organic energy in the receivingbody of water gives rise to fungal and otherheterotrophic growths, deoxygenation of the water(due to the metabolic activities of these growths}thereby rendering the water unsuitable for higherlife forms such as fish, anaerobic conditions caus-ing fermentation and redissolution of heavy metalsalts, and so on. These effects reduce the aestheticappearance, recreational use and the re-use of thewater.

(c) Chemical autotrophs derive their energy forgrowth from the oxidation of inorganic com-pounds. In wastewater microbiology the principalspecies of interest in this genera are the nitrifyingbacteria. These bacteria are obligate aerobes (i.e.can grow only when oxygen is present), and deriveenergy by oxidizing saline ammonia to nitrite andnitrate. In wastewater treatment the nitrifiers forma vital link in the removal of the autotrophic ele-ment nitrogen from the water by converting am-monia nitrogen to nitrate nitrogen when, by ap-propriate design, the heterotrophs can be forced toutilize the nitrate in the heterotrophic metabolismof carbonaceous material, in which action thenitrate is reduced to molecular nitrogen andescapes as nitrogen gas. (see later).

1-1

1.1 Objectives of wastewater treatment 2.1 Electron donors and acceptors

From the considerations above the objectives ofwastewater treatment are seen to be twofold:1. Reduction of organic-bound energy to a level such

that the heterotrophic growth and associated de-oxygenation effects in the receiving body of waterare acceptably low.

2. Reduction of the autotrophic substances,phosphates, ammonia and nitrates, to levefs suchthat photo-synthetic autotrophic growths and theircapacity to fix solar energy as organic energy in thereceiving body of water are acceptably low.

2. BIOLOGICAL BEHAVIOUR

Biological metabolism is principally 3 process of energyconversion. In biological wastewater treatment pro-cesses, the input energies are present in the wastewaterin the form of carbonaceous and nitrogenous organiccompounds. These are converted into other energyforms during which some of the energy is lost as heat.

The carbonaceous energy is utilized by theheterotrophic group of organisms. This is a wide spec-trum group, which, given sufficient time and appro-priate environmental conditions, will utilize every typeof organic material. The group is ubiquitous and in anygiven situation those members of the group that obtainmaximum benefit from the specific organic material andenvironmental conditions will develop. As the organicsource of the environmental conditions change, soassociated changes in the heterotrophic organismspecies take place.

The nitrogenous material serves as prime energysource in the free and saline ammonia form. Pro-teinaceous nitrogen is broken down into its ammoniacaland carbonaceous constituents by heterotrophs. Thefree and saline ammonia is utilized as an energy sourceby two specific organisms species, the Nitrosomonasand Nitrobacter. The former converts ammonia tonitrite and the latter nitrite to nitrate, these twoorganism species jointly being called the nitrifiers. Com-pared to most of the members of the heterotrophicgroup, the nitrifiers are slow growing. Because thegroup consists of only two species, each performing aparticular function in the nitrification process, the en-vironmental conditions must be conducive to bothspecies, otherwise one or the other will not propagateand nitrification will be adversely affected. In generalthe environmental conditions for the nitrifiers are muchmore circumscribed than those for the heterotrophs.With the heterotrophs, the variety of organisms is solarge that some group of the species nearly always willadapt to the biodegradable organic material in the in-fluent or changes in the environmental conditions. Withthe nitrifiers, once the environmental conditions changeto beyond a rather restricted range, they cease to grow.

For all the organisms the energy source serves twofunctions:(1} for the supply of material which is transformed to

new cell material i.e. synthesis of new celt mass;(2) for the supply of energy to effect the transforma-

tion.For heterotrophic growth energy is obtained from

the carbonaceous material, as follows: Each organicmolecule is split to give hydrogen ions, carbon dioxidemolecules and electrons. Because it releases electronsthe organic molecule is called an electron donor and onyielding electrons the molecule is said to be oxidized.The electrons (and hydrogen ions) are eventuallytransferred to a molecule that can receive them and thismolecule is called an electron acceptor; on receipt ofelecrons the molecule is said to be reduced. In thisoxidation-reduction (redox) reaction free energy isreleased, that is, energy that can be utilized to do work.

Under aerobic conditions the electron acceptor isoxygen (O2) and it is reduced to water iH7O). If nooxygen is available but nitrate (NOjl or nitrite (NOZ) is,an anoxic state exists; the NO3 and N02 serve as elec-tron acceptors and both are reduced to nitrogen gasand water. If no O2, NO] or NOj is present an anaerobicstate exists and the organism generates its own electronacceptor internally (endogenouslyt, in contrast to exter-nally provided (exogenous) electron acceptors (O2,NO2 and NO3), provided in the aerobic and anoxicstates.*

The quantity of free energy released in a redoxreaction depends on the electron donor and the accep-tor. When carbonaceous material is the donor andoxygen is the final acceptor the free energy is relativelyhigh; with N02 and NO] as acceptor the free energy isabout 5% less than with oxygen (McCarty, 1964). Inter-nally generated electron acceptors give rise to redoxreactions that yield very little free energy.

Taking oxygen as the electron acceptor the freeenergy per electron transferred for different configura-tions of carbonaceous molecules lies within a relativelynarrow band of values. For sewage which contains aninnumerable number of different carbonaceoussubstances, the mean free energy per electron transfer-red is virtually constant, McCarty (1964).

In aerobic synthesis of organism mass a fraction ofthe carbonaceous material (with oxygen as electron ac-ceptor) is oxidized via a complex set of redox reactions,to yield free energy, see Figure 1.1. The free energy (viaADP-ATP exchanges) is utilized to reorganise the re-maining carbonaceous molecules to protoplasmicmaterial, the free energy being lost as heat in thisreorganization.

In the qualitative description above the questionthat arises is: "How does one quantify this synthesisreaction?" The quantification rests on the indestruct-ability (conservation) of matter in chemical reactions.

"In Bacteriology ii the terminal hydrogen ion and electron acceptor is molecular dissolved oxygen, then The process of substrate oxidation istermed aerobic; if it is chemically bound oxygen, i.e. nitrate, nitrite or sulphate, the process is termed anaerobic; if the terminal electron accep-tors originate outside the cell, i.e. exogenous, the process is termed one of respiration; if the terminal electron acceptors originate inside thecell, i.e endogenous, the process ts termed fermentation. In Sanitary Engineering, the usage of some of these terms differs from that inBacteriology. In particular if no dissolved oxygen is present but nitrate is, the process of environment is called anoxic; if neither dissolvedoxygen nor nitrate is present, the process or environment is called anaerobic.

1-2

NADRtD NAD

RED

(2H+;2e")

ADP+P

COQRE(2H+;2e-)

ADPt P

2 cyt bREIJ(2e-)

2cytcR E D

(2e~)

2cyt aRED

ox

ox

ATP

CoQox

ATP^T

2 cyt box

2cyt cox

2H

2 cyt a03

AT

H90 (2H+;2e-)

FIGURE 1.1a: ATP (biological energy] formation by sequentialoxidation-reduction reactions (e transport) in the cytochromes

of the micro organism

OXIDATION OF I SYNTHESIS XCO;

ELECTRON DONOR OF CELL MASS

HEATATP* ADP*P

ADP+P-ATP

'inHEAT

FIGURE 1.1b: Energy change during synthesis

66 e ATP- ADP*P66 H*

l 7 / 2 0 2

ADP+P- ATP(No i l tdron Inl*rchoh9«)

FIGURE 1.1c: Electron changes during synthesis

1-3

During synthesis, in the set of redox reactions, elec-trons and hydrogen ions are transferred from the donorto the acceptor (see Figure 1.1) but, the electrons canbe neither created nor destroyed; however, they can bemeasured, as follows: The number of electrons ac-cepted in the reduction of a molecule of oxygen towater is a defined physical-chemical quantity. If the car-bonaceous material is oxidized completely to CO2, withoxygen as acceptor, the mass of oxygen consumedgives a direct stoichiometric measure of the electronstransferable from the organic molecules, and, as thefree energy per electron (with O2 as acceptor) is virtuallyconstant for all carbonaceous materials, indirectly ameasure of the free energy also is obtained. Thus, in asynthesis reaction, by (1) oxidizing completely the inputbiodegradable material and (2) oxidizing completely thesynthesized material, the ratio

oxygen demand of synthesized materialoxygen demand of input material

_ 02 equivalent of synthesized materialO2 equivalent of input material

_ electrons available in synthesized materialelectrons available in input material

- specific yield coefficient of the reaction.

Very closely the specific yield coefficient above alsoreflects the ratio

free energy in the synthesized materialfree energy in the input material.

The difference between (0; equivalent of inputmaterial) and (02 equivalent of synthesized material),reflects the electrons transferred to oxygen in order torelease free energy to drive the synthesis reaction i.e.,

0z demand = (02 equivalent of input material)- (0 ; equivalent of synthesized

material). (1.1)

This equation is stoichiometricaliy exact and formsthe basis for fundamental investigations into thedescription of the activated sludge process. Very close-ly, in fact for all practical purposes exactly, it alsoreflects the energy changes in aerobic and anoxic redoxreactions. Thus, by the developments above, we cantrace the energy changes by measuring the electrons(as equivalent oxygen) in the input material and theelectrons contained in the synthesized material, the dif-ference between these two being equal to the oxygenutilized during change in state of the material. Electronsare indestructible and can and always should be ac-counted for in any investigation; this brings to the forethe importance of electron mass balances to check thevalidity of investigations both in research and in theoperation of a plant.

During nitrification, ammonia serves as the elec-tron donor and oxygen as the acceptor, to convert NH<to N02 and NOj. Ammonia is present in the influenteither as an NH3 radical in protein molecules(albuminoid ammonJa) or as dissolved ammonia (NH3)and ammonium ions fNH^) (free and saline ammoniarespectively). On conversion of the proteinaceous am-monium radical to the free and saline form (by the

heterotrophs) it becomes available as an electron donorwith oxygen as acceptor to form NO3 and NOj. As themolecular forms of both NH^ and NOj are fixed, oncethe total mass of nitrogen available for nitrification isknown the oxygen necessary to oxidize it is also known,and hence the electron donor capacity.

One last point: The yield as defined earlier, isalways an experimentally derived value. Biochemistry isnow sufficiently advanced to allow the yield to becalculated theoretically fairly accurately by taking intoaccount the stoichiometry of the electron donor and ac-ceptor, the thermodynamic free energy released perelectron transferred and the biochemical pathwaysthrough which the transfer takes place (for example, theEmbden-Meyerhof pathway - Krebs cycle sequence;Payne, 1971). Such an approach, although of import-ance conceptionally, is not so valuable practicallybecause in wastewater treatment plants, the micro-biological mass does not consist of pure culture butcomprises a complex ecological system of primaryorganisms and ones that prey on them.

3. CARBONACEOUS ENERGY MEASUREMENT

3.1 Chemical Oxygen Demand test

The discussion in the previous section, in which it wasshown that the energy changes in biological reactions isreflected in the number of electrons transferred, led tothe conclusion that the electron donor capacity can bemeasured in terms of the oxygen required to oxidize thecarbonaceous matter to C02. Such a measurement isavailable in the Chemical Oxygen Demand (COD) test.

In the COD test the electron donor capacity of car-bonaceous material is measured by oxidizing thematerial by a strong oxidant i.e. a hot dichromatesulphuric acid solution, in which the electron acceptor isoxygen. This oxidant is so strong that oxidation of thecarbonaceous matter is complete, or very near com-plete. From the half reaction for the reduction ofoxygen,

4e- -t 4H* + 0 ; H,0

it can be seen that 1 mole of molecular oxygen (32 g) isequivalent to 4 electron equivalents. Stoichiometricallyin the COD test when the equivalent of 1 g 02 is used upthen the mass of COD oxidized is alsio 1 g, i.e.

1 g COD = 1 g 02 = 1/8 electron equivalent.

The COD test, therefore, supplies the information re-quired to trace the electron and associated energychanges in a process.

Ammonia is not oxidized in the test so that the testreflects only the electron donor capacity of car-bonaceous material.

In practical application some organic material is notoxidized by the COD test under any circumstances,such as the aromatic hydrocarbons and pyridenes,although these are utilized by micro-organisms. Poorestimates of COD can also arise with lower fatty acidssuch as acetate because these compounds are in theunionized form at the low pH at which the test is per-formed - the unionized molecules can be extremely dif-

1-4

ficult to oxidize; in this regard addition of a catalystsilver sulphate overcomes the problem to a large degreeto give virtually 100% oxidation.

In performing a COD test it should be rememberedthat the test involves a time dependent reaction and thespecified reftuxtng period of 2 hours must be adhered toto ensure complete or near complete oxidation. Further-more, the degree of oxidation is subject to mass actioneffects - equivalent masses of the same organic com-pound refluxed at different final test volumes yield dif-ferent results for the COD; the temperature at which therefluxing step takes place also affects the oxidation rate,the refluxing temperature in turn is dependent on theconcentration of sulphuric acid in the test. These fac-tors have all been considered in the development of theCOD test procedure. Hence, the COD test will givereliable results only if the test is done in strict accor-dance with the set procedures, for example, those inStandard Methods 11981). !f these procedures are strict-ly followed our experience is that the COD test ade-quately reflects the electron donor capacity of the sam-ple. Application of COD test results to carbonaceousmaterial mass balances on activated sludge plants atsteady state also indicate that with careful experimentalwork mass balances between 98 and 102 per cent areobtainable i.e. the mass of influent COD per day can beaccounted for by the sum of the daily mass of oxygenutilized and the daily masses of COD in the waste sludgeand effluent. Extensive investigations into CODbalances by Schroeter, Dold and Marais (1983) have in-dicated that where poor balances are obtained, in-variably the causes can be traced to errors in oxygen de-mand for nitrification (to correct the total oxygen de-mand to the carbonaceous oxygen demand) and/orerrors in the oxygen utilization rate measurement.

3.2 Total Oxygen Demand test

In the Total Oxygen Demand (TOD) test the sample isoxidized at high temperature in a combustion oven. TheTOD is the mass of oxygen required for the oxidation ofall oxidizable material contained in a unit volume ofwastewater sample; both carbonaceous and nitro-genous compounds are oxidized. A problem with thistest is that the amount of oxygen available in the com-bustion oven influences the degree of oxidation of am-monia and organic nitrogen - with oxygen in excessthere is a conversion of nitrogenous material to nitricoxide, whereas when oxygen is not in excess, oxidationof the nitrogenous material may be partial only. Further-more, nitrate, nitrite and dissolved oxygen in the sampleinfluence the TOD test result. Because the test determines both the carbonaceous and nitrogenous oxida-tion potential, an additional TKN test is required toisolate the carbonaceous fraction of the TOD. At thepresent stage of development of the test, difficulty isfound in obtaining representative and reproduciblereadings for samples which contain paniculate mat-ter - these samples must be thoroughly homogenisedto obtain representative readings as the sample volumeis extremely small (10-20 ^ i ) . Sample homogenisationalso increases the dissolved oxygen contamination. Thedifficulties of applying the TOD test to wastewater can

be overcome, but the resulting test procedure reducesits attraction for general routine use.

3.3 Biochemical Oxygen Demand test

This test provides a measure of the oxygen consumed inthe biological oxidation of carbonaceous material in asample over 5 days at 20DC. According to Gaudy (1972)the origins of this test can be traced back about a hun-dred years to Frankland who appears to have beenamongst the first to recognise that the observed oxygendepletion in a stored sample is due to the activity ofmicro-organisms. The Royal Commission on SewageDisposal appeared to have developed the concept ofusing the oxygen depletion as a measure of the strengthof pollution. Over the past century extensive researchtowards elucidating the meaning of the test, quantifyingit and integrating it in theoretical descriptions ofwastewater treatment processes, has been undertaken.To date the Biological Oxygen Demand (BOD) test isstill the most popular parameter for assessing the pollu-tion strength of influents and effluents and for describ-ing the behaviour of treatment processes. Its continuedpopularity seems to stem mainly from the body of ex-perience that has built up in its use. Whereas there isjustification for retaining it in effluent descriptions forregulatory purposes, in so far as present day scientificdescriptions of treatment process is concerned fewcogent arguments for its use over that of the COD testcan be produced.

The main deficiency of the BOD test arises fromthe inadequacy with which it assesses the electrondonor capacity of the carbonaceous organic material ina sample. At the end of 5 days some of the biodegrad-able carbonaceous material originally present in thesample remains as unbiodegradable material. Conse-quently, it is not possible to perform a mass balanceeven with respect to the biodegradable input material.Furthermore unbiodegradable material originally pre-sent in the sample remains unaffected by the test sothat there is no procedure, related to the test, wherebyan assessment of this fraction can be obtained, (in con-trast in the COD test, because the total electron donorcapacity is measured, even though the unbiodegradablefractions are not isolated in the test, it is possible toestimate these fractions from the test results and thekinetic behavioural patterns of the treatment process).

Despite the deficiencies listed above, if the B0D5

value constituted a consistent estimate it could still haveprovided a relative measure of practical usefulness, buteven this cannot be guaranteed. Gaudy (1972) in athought provoking review on the BOD gave a summaryof the work on the BOD test (in which he took a leadingpart). He concluded that the standard approach to for-mulating the BOD time curve as a first order reaction isquite invalid: The actual curve arises from two main ef-fects, (1) bacterial synthesis from the biodegradable in-put - a reaction usually complete in one to two days,and (2) predator growth using the synthesized bacteriaas substrate. The second phase, being dependent onthe first, usually lags the first so that a distinct "pauce"or "plateau" in the time curve is exhibited (see Figure1.2). The occurrence and duration of this pause isdependent on the bacteria-prey relationship initially pre-

1-5

280

240 -

4 6TIME (days)

8 10

FIGURE 1.2: Typical experimental BOD lime curve onwastewater sample with heterogeneous seed clearly showing

the "plateau" behaviour after about 1,5 days

sent in the sample, the "strength" of the organicmaterial, presence of inhibitory substances and so on.These compound effects certainly are not reflected by afirst order formulation. Furthermore, should nitrificationoccur during the 5 days the BOD value can be com-pletely misleading; this may not be apparent unless pro-cedures to suppress nitrification are imposed on thetest. Evidently the B0D5 value is subject to a number ofinfluences which can be present in varying degreesresulting in unknown variable effects on the observedvalue.

Perhaps, the neatest summing up of the use of theBOD test is that due to Hoover, Jasewicz and Porges(1953): "The BOD test is paradoxical. It is the basis of allregulatory actions and is used routinely in almost allcontrol and research studies on sewage and industrialwaste treatment. It has been the subject of a tremen-dous amount of research, yet no one appears to con-sider it adequate'y understood or well adapted to hisown work". Considered in the light of the biochemicaldescriptions of biological growth these remarks are stillpertinent today.

3.4 Total Organic Carbon test

In the Total Organic Carbon (TOO test the sample isoxidized in a combustion chamber in the presence ofoxygen and the carbon dioxide content of the burntgas is measured. If the organic carbon content only isrequired, then it is necessary to have a pre-treatmentstage in sample preparation by acidification and C02

purging by sparging, or, to do an additional test inwhich the inorganic carbon only is measured. The olderinstruments can only measure the carbon content ofsoluble organic material but newer models can deal withparticulate organic material.

As a procedure to estimate the electron donorcapacity of the sample the TOC test can be verymisleading. This arises from the fact that the ratio oforganic carbon to electrons is not constant for differentorganic materials. For example taking glucose andglycerol:

The number of available electrons may be deter-mined by(i) measuring the number of moles of 0 ; consumed in

the complete combustion of one mole of the com-pound of interest, and

(ii) multiplying by four (which is the relative number ofelectrons required to reduce one mole of oxygen).

e.g. la) Glucose, C6 H12 O6

CE H12 O6 + 6 O2 -* 6 CO; + 6 H2O6 moles O2 =» 6 x 4 = 24 e~ compoundi.e. 24/6 - 4 e~ available per unit organic C.

4 H : 0(b) Glycerol, C3 H8 03

C3 HE O3 + 7/2 02~* 3 C02 +7/2 moles 02 =*• j x 4 = 14 e" compoundi.e. 14/3 = 4,66 e" available per unit organic C.

Comparing glucose and glyceroi the electron donorcapacity per unit C differs by 17%.

1-6

The analysis above points to the inadvisability ofutilizing the TOC test to describe the energy basedbehaviour of waste water treatment processes. Thisdoes not mean that the test does not have other uses; itserves a valuable function as control parameter or as aparameter in tertiary treatment where the carbon con-tent in the effluent may be of great importance.

4. COD/VSS RATIO

Consider a completely biodegradable soluble substrate,for example glucose. If a sample of glucose solution isinnoculated with bacteria, syntheses of new bacterialmass will occur. The change in soluble COD, ACOD(soluble), will be reflected in an increase in the COD ofthe bacterial mass, ACOD(bacteria), and oxygen utilizedto generate free energy, AOj(utilized),

ACOD(soluble) - ACOD(bacteria) + AO2(utilized)(1.2)

Equation (1.2} reflects the destination of the electronsfrom the donor to the synthesized material and the elec-tron acceptor, oxygen. From extensive work reported inbacteriology, reviewed by Payne (1971), the fraction ofACOD(soluble) appearing as ACOD(bacteria) appears tobe nearly constant. Usually the fraction is expressed asa ratio called the specific yield coefficient Y c o o (see Eq1.1) i.e.

ACOD(bacteria)/ACOD(soluble) = YCOD H.3)

Equation 1.2 can be rewritten in terms of YCOD

ACOD(soluble) = YCOD.ACOD(soluble) + AO:

and recasting Eq (1.4)

AO2 = (1 - YC0D) ACOD(soluble)

fecting it. Theoretically fcv has been evaluated by accep-ting an empirical stoichiometric formulation forbiological sludge, for example that of Hoover andPorges (1952),

CSH7O2N •+ 5 O 2 - 5CO2 + 2H2O -4 NH3

Stoichiometrically,

113 g VSS -* 160 g 0 - 160 g CODi.e. 1 g VSS — 1,42 mg CODi.e. COD/VSS = 1,42 mg COD., mg VSS

Eckenfelder and Weston (1956) presented experimentaldata that supports this theoretical ratio. However, othertheoretical ratios have been proposed and each findsome experimental confirmation, see Table 1.1. Thebasic difficulty in evaluation of frv arises from the factthat fcv is a ratio so that the small random errors in eitherthe COD or the VSS magnify the dispersion of the ratio;a reasonably stable mean estimate of the fcu value of asludge requires at least thirty COD and VSS pair deter-minations. With smaller numbers the mean fcv value canstill vary significantly.

(1.4)

(1.5)

In wastewater treatment kinetics the usage has arisen,probably from the work of bacteriologists, to measurethe mass of volatile solids generated in the synthesisreaction, Xa, rather than the COD of the volatile solidsgenerated, ACOD(bacteria). To retain the usefulness ofEq (1,4), in particular in the form given by Eq (1.5), it isnecessary to relate AXa to ACOD(bacteria). The usagehas developed to express the relationship between AXa

and ACOD(bacteria) by the ratio of the two parameters,called the COD/VSS ratio, defined by the symbol f^,

fcv = COD/VSS - ACOD(bacteria)/AXa

Using this definition the yield coefficient YC0D is ex-pressed in terms of a specific yield coefficient withrespect to volatile solids Yh, and the COD/VSS ratio,

YCOD - fcv-Yh

and inserting in Eq (1.5),

AO2 = (1 - fcvYh) ACOD(soluble)

(1.6)

(1.7)

Equation (1.7) is very important in biological wastewatertreatment kinetics.

Considerable research has been undertaken toestablish a value for f^ and to determine the factors af-

TABLE 1.1EMPIRICAL

Bacterialformulation

C5H7O,NC5H9O3NC7H10O,NC,HBO ;N

THEORETICAL COD VALUES FOR VARIOUSFORMULATIONS FOR MICROBIAL

(AFTER

Molecularweight

113131156114

MCCARTY, 1964).SLUDGE

Theoretical COD in grams

Permole

160160232168

Per gTSS

1,281,101,331,32

Per gVSS

1,421,221,481,47

Per gcarbon

2,672,672,762,80

Considering the COD/VSS ratio of sludges derivedfrom activated sludge processes treating municipal ef-fluents, it should be remembered that not only is activevolatile mass generated but volatile mass is also presentin unbiodegradable participate form in the influent(which accumulates in the sludge mass) and inertbiological residue generated due to endogenous respira-tion (which also accumulates in the sludge mass). Con-sequently the sludge normally consists of three frac-tions, active, inert influent and endogenous. Each ofthese could have a different fcv value, a factor not con-sidered in deriving the theoretical ratios discussedearlier. The proportions of the three fractions in a sludgedepend on the sludge age, the active fraction beingrelatively small at long siudge ages and high at shortsludge ages. For practical purposes an importantaspect, therefore, is whether fcv changes with sludgeage or whether a mean experimentally derived value canbe assigned to all activated sludges. Accordingly Maraisand Ekama (1976) and Schroeter, Dold and Marais(1982) investigated the COD/VSS ratio of activatedsludge on laboratory scale units treating municipalwastewater for sludge ages ranging from 2 to 30 days.

1-7

Their data indicate that despite the significant changesin proportion of the three fractions with changes insludge age, fcv appeared to remain substantially cons-tant, at 1,48 mgCOD/mgVSS. This finding is very im-portant because it allows that COD balances generallycan be performed using VSS measurements on thewaste sludge.

5. REFERENCES

AMERICAN PUBLIC HEALTH ASSOCIATION. AWWA and WPCF(1981) Standard methods for the examination of water andwastewater, 16th ed. Published by APHA, Washington.

ECKENFELDER, W.W. and WESTON, R. F. (1956) Kinetics ofbiological oxidation. In Biological treatment of sewage and in-

dustrial wastes Vol. 1. Eds. B.J. McCabe and W.W. Eckenfelder,Rheinhold Pub). Co. New York. 18-34.

GAUDY, A.F. (1972) Water pollution microbiology. Ed. R. Mitchell,Wiley Interscience, New York.

HOOVER, S.R. JASEWIC2, L. ana PORGES, N. (1953) An inter-pretation ol the BOD Test in terms of endogenous respiration ofbacteria. Sewage and Industrial Wastes, 25, 1163-1173.

HOOVER, S-R. and PORGES, N. (1952) Assimilation of dairy wastesby activated sludge. II. The equations of synthesis and rate ofoxygen utilization. Sew. and Ind. Wastes J., 24, 306 312.

MARAIS, G.v.R. and EKAMA, G.A. (1976) The activated sludge pro-cess Part 1 - Steady state behaviour. Water SA. 2, 164-200.

McCARTY, PL. (1964] Thermodynamics of biological synthesis andgrowth, 2nd Int. Conf. Wat. Poll. Research, Pergamon Press,New York. 169-199.

PAYNE, W.J. 11971) Energy yields and growth of heterouopris. An-nual Review, Microbiology, 17-52.

SCHROETER, W.D., DOLD, P.L. and MARAIS, G.v.R. (1932) TheCOD/VSS ratio of the volatile solids in the activated sludge pro-cess. Res Rept. No. W 45, Dept. of Civil Eng., University of CapeTown.

1-8

CHAPTER TWO

NATURE OF MUNICIPAL WASTEWATERS


1. WASTEWATER CHARACTERIZATION

In so far as it affects biological treatment, municipalwaste flows can be characterized chemically andphysically.

Chemical characterization comprises identificationand measurement of: (a) carbonaceous and nitro-genous constituents in their various fractions, solubleand paniculate, biodegradable and unbiodegradable;and (b) inorganic constituents principally total alkalinity(Alkalinity), total acidity, pH and phosphorus. Inorganicconstituents such as calcium, sodium, magnesium,chloride and sulphate usually are of minor importancewith the exception of calcium which, in certain cir-cumstances, can be of significance in phosphorusremoval by calcium phosphate precipitation.

Physical characterization comprises separation ofthe wastewater into dissolved, suspended and settle-able constituents.

2. CHEMICAL CHARACTERIZATION

2.1 Carbonaceous materials

Characterization of the carbonaceous material in the in-fluent is done via the Chemical Oxygen Demand (COD)

test. For activated sludge process design it is necessaryto identify and know the magnitudes of the various frac-tions of the influent COD as these significantly affectthe response of the process. Research at the Universityof Cape Town has indicated that the subdivisionsshown diagrammatically in Figure 2.1 are essential foraccurate description of the behaviour of the process.

2. 7. 7 Biodegradable and unbiodegradable fractions

From Figure 2.1 the first major subdivision is intobiodegradable COD (Sbl) and unbiodegradable COD(SU1) fractions. This division is important for the follow-ing reasons: In design, knowing biodegradable CODconcentration and the flow per day gives thebiodegradable COD load on the plant; knowing themass of biodegradable COD load per day and selectinga sludge age, the daily carbonaceous oxygen re-quirements and the active mass in the process can beestimated from the kinetic equations governing the pro-cess (Eq 4.10, Chapter 4). Conversely the fraction of thetotal influent COD which is biodegradable can beestimated from the carbonaceous oxygen consumptionrate in a steady state laboratory scale completely mixedactivated sludge process run at one or more sludgeages. The mass of Sbi is then estimated by inserting themeasured oxygen demand into the steady state equa-

INFLUENTCOD ( S t i )

BIODEGRADABLE

COD

UNBIODEGRADABLECOD

SOLUBLEREADILYBIODEGRADABLE

PARTICULATESLOWLYBIODEGRADABLE

SOLUBLEUNBIODEGRADABLECOD

l S u . i >

PARTICULATEUNBIOOEGRADABLECOD

FIGURE 2.1: Division of the total influent COD in municipalwastewaters into its various constituent fractions

2-1

tion for the carbonaceous oxygen demand and by trialand error finding the value of Sbi that balances the equa-tion, (Eq 4.15, Chapter 4). This approach will be ade-quate for developing information for design. Itpresumes that the various constants in the kinetic equa-tions, such as the specific yield constant Yh and thespecific endogenous mass* rate constant bh have beencorrectly evaluated from research studies, (Marais andEkama, 1976). It presumes also that all the bio-degradable COD has been metabolized, a presumptionjustified from research findings if the sludge age isgreater than 3 days at 20°C, (Dold, Ekama and Marais,1980). If nitrification occurs in the process this eitherhas to be allowed for when determining the car-bonaceous oxygen demand from the measured oxygendemand (see Chapter 5) or, nitrification must be sup-pressed by addition of a nitrification inhibitor to the pro-cess, for example by addition of thio-urea at intervals tothe process.

2. 1-2 Unbiodegradable fractions

The next major subdivision is to divide each of the un-biodegradable and biodegradable COD concentrations,Suj and Sbi respectively, into their relevant subconcen-trations.

Consider first the unbiodegradable COD concentra-tion, Sui. This concentration consists of two fractions,an unbiodegradable soluble COD, Susi and an un-biodegradable particulate COD, Su[)1. Both fractions arehypothesised to be unaffected by biological action sothat at steady state the masses that enter equal themasses that leave the process. The Sus, passes out inthe liquid effluent discharge. The Sup; is enmeshed inthe sludge, accumulates in the system to a mass equalto that entering in the influent per day, multiplied by thesludge age, but eventually at steady state the massesentering and leaving the system are the same. The massleaves the system via the daily sludge mass wasted fromthe plant. Thus the Suv, has the principal effect of in-creasing the mixed liquor concentration.

Usually it is more convenient to express the un-biodegradable particulate influent fraction not in termsof its COD but in terms of its influent volatile solids con-centration, X,,. This is readily accomplished by noting inChapter 1 that the COD and volatile solids are defined asfollows:

Xii = Supi/fCv

where fCv = COD/VSS ratio = 1,48 mgCOD/mgVSS.

Measurement of SUSI is relatively simple; alaboratory scale unit is run at say two different sludgeages both greater than 3 days. The COD of the cen-trifugcd effluent or the filtrate through a fine filter,estimates S;JSi. The balance of the influent COD, aftersubtracting Sbi and Su, from the total influent COD, Sti,is the unbiodegradable particulate COD, Supi, i.e.

^ - c: _ c _ c

The reasonableness of the Sup, estimate is checked byrunning a laboratory scale unit at a long sludge age, say,15 days or longer, calculating the total volatile solidsconcentration using the input va'ues for SD. and Supi

(i.e. X,,) [Eq 4.13, Chapter 4) and checking this valueagainst the volatile mass observed. A long sludge age isselected because the total volatile mass of sludgebecomes more sensitive to X,, as the sludge age in-creases. Overall the SuP., SJ; and Sb, must give rise toconsistency at the different sludge ages. Once this con-sistency is judged adequate the COD fractions are ac-ceptable for design.

2. 1.3 Biodegradable fractions

The data obtained above are all on steady state aerobicactivated sludge units. Subdivision of Sbj into readilybiodegradable soluble COD, Sbsi, and slowlybiodegradable paniculate COD, Sbpi, (Fig. 2.1) can beobtained only by dynamic loading of the units, for thefollowing reasons: The fraction Sbsi is rapidly taken upby the sludge in a matter of minutes and metabolized,giving rise to a high unit rate of oxygen demand for syn-thesis. (The kinetics of utilization are in accordance withMonod's equation). The fraction SDpi requires to be ad-sorbed and stored on the organism, broken down tosimpler chemical units by extracellular enzymes andthen absorbed and metabolized by the organism. Theextracellular breakdown is slow and forms the limitingrate in the synthesis reaction, the rate being about 1/7to 1/10 that for the readily biodegradable COD, Sbsi.(The kinetics for synthesis of particulate COD is for-mulated via Levenspiel's active site surface reactiontheory). Under steady state (i.e. constant flow and loadconditions) the stored Sbp is negligibly small at sludgeages greater than about 2 to 3 days. However at shortsludge ages (2-3 days) under cyclic flow conditionsunder square wave loading, 12 h on, 12 h off, duringThe loading period the rate of Sbp adsorption andstorage is so large that the organism cannot deplete itconcurrently, even at the maximum utilization rate. As aconsequence, after the loading period terminates theutilization rate of the stored Sbp continues at the samemaximum rate for a while, and then slowly declines.This behaviour allows the readily biodegradable COD tobe determined; measure the oxygen utilization rate for aperiod before loading terminates and a period after, thedifference in the rates defines the rate due to the Sbs.Details of the test procedure are given in Appendix A2.

The importance of the Sbsi fraction will become evi-dent when denitrification and biological excessphosphorus removal are discussed in Chapters 6 and 7respectively.

2. 1.4 Analytic formulation

For analysis the relationship indicated in Figure 2.1 canbe expressed as follows:

= 0,45 mgVASS/mgCOD utilized, independent of temperature= 0,24 mgVASS/mgVASS/day at 20°C and= 0,24 (1,029l'T-2» at temperature T°C, see Chapter 4.

2-2

Biodegradable and unbiodegradable COD fractions:

Stl = SU1 + Sbl (2.1)

where

Sti - total influent COD concentration (mgCOD/f)SUI = unbiodegradable COD concentration

(mgCOD//)Sbl = biodegradable COD concentration

(mgCOD/i)

Each of the two fractions on the right hand side of Eq2.1 is again subdivided, Figure 2.1.

Unbiodegradable COD fractions:

The unbiodegradable COD concentration consistsof two components, soluble and paniculate, i.e.

where

(2.2)

Susi = unbiodegradable soluble COD concentra-tion (mgCOD/f)

Supi = unbiodegradable particulate COD concen-tration (mgCOD/f).

It is convenient to express SUSI and Supi empirically interms of the total COD concentration S,,, i.e.

^ - f S I? 3)°USI TUS D l l \*..JI

^un i — T,ir, O., l .*tj

where

fus = unbiodegradable soluble COD fraction withrespect to the total COD (mgCOD/mgCOD)

fup = unbiodegradable particulate COD fractionwith respect to the total COD (mgCOD/mgCOD).

Hence from Eq (2.2)

SU1 = (fus + fUp) Stl (2.5)

The particulate unbiodegradable COD concentrationmay be expressed also in terms of volatile solids, Xh:

"ii ~

- f u p . S t l / f c (2.6)

where

X,. = unbiodegradable particulate volatile solidsconcentration in the influent (mgVSS/i)

fcv = COD to VSS ratio of the solids= 1,48mgC0D/mgVSS.

Biodegradable COD fractions:

The biodegradable COD concentration is foundfrom Eq 2.1 as follows:

and from Eq 2.5

- S - S (f -4- f^ t i •J t i " u p • 'us

= Stl(1 - f u p - f j

From Figure 2.1 the biodegradable COD, Sbl, is dividedinto readily biodegradable soluble COD, Sb£l, and slowlybiodegradable paniculate COD, S^,. Each can be ex-pressed in terms of SL| as follows:

Ste. =fbsSbland (2.8a)

p- — ' 1 (2.9a)

where

fb;, = readily biodegradable COD fraction withrespect to the biodegradable COD.

The readily biodegradable COD can also be expressed interms of the total COD, S,,, i.e. substituting for Sbi fromEq 2.7 in Eq 2.8 yields

Sb5i = fbs<1 - f u P " U S , , (2.8b)

~~ ' i (2.9b)

where

(2.7)

fts - readily biodegradable COD fraction wi fhrespect to the total COD.

2.2 Nitrogenous materials

Characterization of the nitrogenous material in the in-fluent is with the Total Kjeldahl Nitrogen (TKN) andKjeldahl free and saline ammonia tests. As for the car-bonaceous material, the nitrogenous material also issubdivided into different fractions but in a differentfashion from that for the carbonaceous material. Thesubdivision is shown in Figure 2.2.

In certain wastewaters nitrate and nitrite might bepresent; the TKN test does not include these. The vastmajority of municipal effluents will not contain nitrate ornitrite because in most sewerage systems the waste-water will be in an unoxygenated state and any nitrateentering the system is likely to be denitrified before itreaches the wastewater treatment plant.

2.2. 1 Free and saline and proteinaceous fractions

The free and saline ammonia fraction, Nai, is determinedby a test of this name. The proteinaceous or organicfraction is determined by difference between the TKN,N,,, and free and saline ammonia values, Nai.

The free and saline ammonia is immediatelyavailable for incorporation into the bacterial protoplasmand for conversion to nitrite or nitrate, if the process isappropriately designed.

The proteinaceous fraction, from Figure 2.2, con-sists of a number of sub-fractions, unbiodegradablesoluble and unbiodegradable particulate organicnitrogen fractions, NU1 and Np, respectively, and abiodegradable organic nitrogen fraction, Noi. The un-biodegradable soluble fraction passes unaffectedthrough the system and is discharged in the effluentsimilar to the soluble unbiodegradable COD. The un-biodegradable particulate organic nitrogen by implica-tion must be part of the unbiodegradable particulateCOD (as XJ, and is treated as such; hence, this frac-tion leaves the process via the sludge wasted daily. The

2-3

INFLUENTTKN ( N f i )

FREE AND SALINEAMMONIA [ N 0 | )

ORGANICALLY BOUNDNITROGEN ( N t i - N 0 | )

JHBlODCGRADABLfSOLUBLE

( N u | )

U N BHXJEGRADA BLEPARTICULATE

BIOOEGRADABLE

FIGURE 2.2' Division of the total influent TKN in municipalwastewater into its various constituent fractions

biodegradable organic nitrogen is broken down by theheterotrophs to free and saline ammonia which with itsinfluent counterpart participates in any furtherbiologically mediated reactions. The breakdown occursvery readily and normally it can be expected to be vir-tually complete for all sludge ages greater than about 3days.

With regard to measurement of these fractions, thefree and saline ammonia (Nai) is measured directly bythe test bearing this name; the total organic nitrogen isfound from the difference between the TKN and freeand saline ammonia test concentrations. Difficulties inquantification arise when subdividing the total organicnitrogen into its biodegradable and unbiodegradablefractions, Noi, Nui and Npl, (see Figure 2.2). NU1 and Npj

are postulated by association to their counterparts inthe COD sub-division. Estimates of their magnitudescan be obtained only by comparing the observedresponse of laboratory scale processes with thatpredicted by the theoretical model incorporating thesefractions. From a large number of such comparisons onprocesses operated within the range of sludge agesfrom 2 to 30 days, under cyclic and steady states in thetemperature range 12 to 25°C, it would appear that N^and NP, are necessary in the model to obtain good andconsistent correlation between observation and predic-tion. However, the fractions are small and it is notnecessary therefore to have accurate estimates. Con-currently Npj is expressed in terms of the un-biodegradable particulate volatile solids in the influent(XJ i.e. Npi is 10% of X,. (see Eq 2.13); from experimen-tal data on the TKN/VSS ratio of the MLVSS in thebiological reactor, using the same influent wastewatersource it was found that the TKN/VSS ratio remainedthe same irrespective of sludge age from 3 to 30 daysdespite the fact that the various VSS fractions changewith sludge age. Hence it is reasonable to accept thatthe TKN/VSS ratio of the inert organic solids is equal tothat of the other VSS fractions at 0,10 mgN/mgVSS.

TABLE 2 1 APPROXIMATE AVERAGE MUNICIPALWASTEWATER CHARACTERISTICS FOR

SETTLED WASTEWATERS IN SOUTH

Wastewater Characteristic

1. Influent COD (mgCOD/i)2. Influent BOD5 (mgBOD/i)3. Influent TKN (mgN/J)4. Influent Phosphorus (mgP.'/l5. Total Suspended Solids (mg.7)6. Seftleable Solids in fmg/f)

in ( m l / / ) "7. Non-Settleable Solids (mg/i)8. TKN/COD ratio (mgN/mgCOD)9. Tolal P/COD ratio

(mgP/mgCOD)10. Unbiodegradable particulate

COD fraction <fjp)11. COD/VSS ratio of unbiodegrad-

able paniculate COD (fcvf(mgCOD.'mgVSS)

12. TKN/VSS ratio of unbiodegrad-able paniculate COD <f-.)fmgN/mgVSS)

13. Unbiodegradable soluble CODfraction (fus)

14. Fraction biodegradable COD

15. Readily biodegradable COD withrespect to total (flE)

16- Maximum specific growth rate ofthe nitrifiers (.'d!

17. Ammonia/TKN fraction (fne)18. Unbiodegradable soluble TKN

fraction (inj19 Minimum temperature (CC)20- Maximum temperature (°C)21. Alkalinity (mg// as CaCOj)

Raw

500-800250-40035-808-18

270-450150 350

6-14100-300

0,07-0,10

RAW AMDAFRICA

Settled

300-S00150-30030-706-15

150-3000-500-2

100-3000,09-0,12*

0,015-0,025 0,020-0,030"

0,070,20

1,45-1,50ave. 1,48

0,09-0,12ave. 0,10

0,04-0.10

0,75-0,85

0,08-0.25

0,20-0,700,60-0,80

0,00-0,0410-1520-30

200-300

0,00-0,10

1,45-1,50ave. 1,48

0,09-0,12ave. 0,10

0,05-0,20*

0.80-0,95

0.10-0,35*

0,20-0,700,70-0,90*

0,00-0,05'10-1520-30

200-300

*Note that the characteristics of settled wastewater shoufd beselected in conformity with the characteristicsraw wastewater and primary sealing tankChapter 4. Section 12.1).

"Volume settled in a 1 i Imhoff cone in 2 hours.

of ihe originalbehaviour (see1

2-4

Furthermore the unbiodegradable soluble organicnitrogen concentration (Nuj) apparently is very low, be-tween 0 and 5% of the influent TKN for raw waste-waters. Later it will be shown (Chapter 5) that the Nul

concentration plays a very minor role in design. Thesuggested fractional values are listed in Table 2.1.

2.2.2 Analytical formulation

The total TKN, N,; is divided into its constituents asfollows, (see Figure 2.2):

12.10)- Nai + NUI + NP1 + N

where

N,, - total influent TKN (mgN/i)Nai - influent ammonia (mgN/jf)NUI = soluble unbiodegradable organic nitrogen in

the influent (mgN/f)Npi = particulate unbiodegradable organic nitrogen

in the influent (mgN/jf)NOI - biodegradable organic nitrogen in the in-

fluent (mgN/i)

It is convenient to express some of the fractions ofnitrogen in terms of the total TKN concentrations

Nai =fnaN11 (2.11)

Nul = fnuN,, (2.12)

where

fna ~ ammonia fraction of the influent TKN(mgN/mgN)

fnu = unbiodegradable soluble organic nitrogenfraction {mgN/mgN)

The unbiodegradable organic nitrogen in the particulatematerial, Npj, is best expressed in terms of the un-biodegradable influent volatile material, Xii( or in termsof its COD counterpart, i.e.

pi " Tn b u p | / t c v — fn Xji (2.13)

where

fn = nitrogen fract ion of the influentbiodegradable volatile particulate material =0,10 mgN/mgXj,

From Eqs 2.10 to 2.13, the biodegradable organicnitrogen (NOI) can be found by subtracting Nul, Nai andNpi from Nti

No, = - f n a - fnu) - fnfupS t i/fc (2.14)

The significance of the various influent TKN fractionson the design of nutrient removal activated sludge pro-cesses is discussed in Chapter 3, Section 3.2 and inChapter 5, Section 5.1.

2.2.3 Specific growth rate of nitrifiers

Experimental observations on a number of differentmunicipal wastewater flows have indicated that the

maximum specific growth rate constant of the nitrifiers,pnm, can differ greatly in value and appears to bespecific to each waste flow. The magnitude of ^nrr, canhave a significant influence on the design of nutrientremoval processes, see Chapters 3, 4, 5 and 6.Therefore, for optimal design it is most desirable to havean estimate of pnrn. This estimate is obtained by runninga completely mixed single reactor at a sludge age ofabout 8 days, at 20°C imposing a sequence of aerobicand anoxic periods on it and measuring the rate ofnitrate increase (or the rate of Alkalinity decrease) dur-ing the aerobic period. The data thus obtained can beanalysed manually to give the ^r im at 20cC, i.e. p*nmj0. Byapplying the well established relationship between ^ n m

and temperature the value at any temperature can bedetermined. Details of the test procedure and analysisof the data are given in Appendix 3.

2.3 Phosphorus

The total phosphorus concentration in municipal waste-waters consists mainly of two fractions, a soluble or-thophosphate (POJ") fraction and an organically boundphosphorus fraction which may be soluble or particulatein form. In both settled and raw municipal wastewaters,the orthophosphate fraction predominates, ranging be-tween 70 to 90 per cent of the total phosphorus.

The orthophosphate concentration* can bemeasured by a colorimetric test of that name. However,this measure records only the orthophosphate concen-tration of the influent, which may underestimate thetotal phosphorus in the influent by about 10 to 20 percent or more. The organically bound phosphorus is con-verted to orthophosphate in the activated sludge pro-cess so that the orthophosphate concentration in the in-fluent underestimates the phosphorus to be removed inthe process. For example, if the orthophosphate con-centration is 8 mgP/ i and the total-P is 10 mgP/i thenfor an effluent standard of 1 mgP// orthophosphate, ifthe plant is designed to remove 7 mgP/jt on the basis ofthe ortho-P influent concentration, the effluent actuallywill contain 3 mgP/l ortho-P. The process should bedesigned to remove 9 mgP/i on the basis of the total-Pinfluent concentration. Consequently, for the influentthe only reliable test is the total-P test. For the effluent,the filtered ortho-P concentration very closely approx-imates the filtered total-P concentration and this ortho-P test would appear to be adequate. However, in per-forming the ortho-P test, a problem sometimes is en-countered in that the natural colour in some waste-waters interferes with the colorimetric test giving rise tospurious results.**

3. PHYSICAL CHARACTERIZATION

Physical characterization of the influent exists mainly inestimating the expected performance of the primarysettling tank. The function of the settling tank is toseparate in some degree the settleable particulate

"All phosphorus and phosphate concentrations are measured relative to the phosphorus atom i.e. mgP/t.••This is a problem encountered in treated effluents from the Western Cape, but apparently is absent from treated effluents of Johannesburg

and Pretoria.

2-5

co II o idol futptndtd

!O- 3

10- 8

10- 4

10- 7

10

10- 6

10- 2

- 5

10- I

- 410 10 10

rt movable by

- 3

coagulation

10

10- 2

100H —

-I m m10

t t t t l t a b l t

FIGURE 2 3: Particle size range and classification

material from the influent flow to give a supernatantoverflow having reduced values of COD, TKN andphosphorus.

Physical characterization basically is made on therelative size of the particles in the wasteflow. Here threebroad divisions are recognized, dissolved, colloidal andsuspended solids, see Figure 2.3. Generally material ofparticle size smaller than 10~3 to 1 ^m is accepted as col-loidal and material with particle size larger than 1 m assuspended. To obtain estimates of the suspended mate-rial pore size membrane filters other than 1 m are inuse which can lead to significantly different results sothat when reporting or interpreting data, careful noteshould be made of the pore size of the filter membranes(and papers) on which the data are based.

The fraction of suspended solids (i.e. retained on a1 /am filter) that can be removed by gravity settling iscalled the settleable solids and usually ranges inequivalent pore size from 10 to 50 jjm depending on par-ticle density. However, a more practical procedure toestimate the mass of senleable solids in a sample ofwastewater, is to use the Imhoff cone: A one litre Im-hoff cone is filled with a sample and allowed to settle fortwo hours. The volume of solid material settled into thebottom of the cone gives a measure (in m i / i } of thequantity of material that can be removed by gravitysedimentation. From a targe number of tests it has beenfound that 1 mi/l of settleable solids is approximatelyequivalent to a dry mass* of 25 mg/f ; with thisequivalence it is possible to obtain an estimate of themass of settleable solids using the Imhoff test.

The settleable solids can be measured either astotal solids where total solids include both the organicand inorganic fractions, or, as volatile or organic solidswhere the volatile solid is the difference between thetotal solids and the ash remaining after the total solidsare incinerated in an oven at 600°C for half an hour. Atthis temperature most inorganic (mineral) salts, exceptmagnesium carbonate, do not decompose or volatilize.The ratio Volatile Solids:Total Solids of the sertleablesolids from municipal wasteflows is about 0,75 i.e. 75%volatile, 25% mineral so that Volatile Solids = 0,75 TotalSolids. The COD associated with the volatile solids can

be estimated by multiplying the volatile solid mass by1,48 i.e. COD £ 1,48 Volatile Solids.

The Imhoff cone test serves as a means wherebythe percentage removal of Total Phosphorus and TKNin the primary settling tank can be estimated. This isdone by determining total-P and TKN concentrations inthe waste sample before settling and in the supernatantafter settling in the cone.

In a primary settling tank the degree of COD andsettleable solids removed by gravity sedimentationdepends on the upflow velocity (overflow rate) andretention time. For relatively deep tanks (>3 m) withmean retention times of 2 to 3 h and mean overflowrates of 1,5 to 2 m/h between 50 and 80% of thesettleable solids (as given by the Imhoff cone test) and30 to 50% (mean of 40%) of the total COD will beremoved. Increasing the retention time above 3 h doesnot improve the removals significantly. However, lower-ing the retention time to say one hour (equivalent to3 m/h mean overflow rate) reduces the removal toabout half of that at 3 h retention time.

With regard to the TKN and P removal in primarytanks this can vary depending on state of the wasteflow. It seems that in fresh wastewater a larger fractionof the TKN and P may be bound to the settleable solidsthan if the wastewater is old and solubilization of theTKN and total P fractions has occurred. Unfortunatelyinsufficient information on these aspects is available tolay down criteria for design. For design it would seempreferable to assume that solubilization does occur sothat only a minor fraction of the TKN and P respectivelyis removed by primary sedimentation, between 15 and25%, even though there are reported removals of up to40%.

The discussion above indicates that with primarysedimentation a significant reduction is obtained inCOD (40%) but a smaller reduction of TKN and total P(15 to 20%). This has the effect that settled wastewaterhas increased TKN/COD and P-COD ratios comparedto the raw wastewaters. Note also that the readilybiodegradable COD concentration (SbJ which is com-pletely in the dissolved form is not reduced so that theconcentrations in the raw and settled flow are the same.

'Mass retained on filter paper {Whatmans No. 1] after drying for 12 hours at 105cC. Measurements taken from raw wastewater ai the CapeFiats treatment works by Ekarna, Marais and coworkeis.

2-6

hence the fraction S t l/S t l increases for settled sewage.With regard to the unbiodegradable participate CODthis fraction appears to be largely removed by primarysedimentation so that settled wastewaters contain lowfractions of unbiodegradable paniculate COD.

4. QUANTIFICATION OF WASTEWATERCHARACTERISTICS

In the above section the various chemical and physicalwastewater characteristics were defined. In this section

these characteristics are quantified to facilitate their usein design. The influence of the various characteristicson the design of an activated sludge process is discuss-ed in Chapter 3.

The ranges of the magnitudes of the variouschemical and physical characteristics of approximatelynormal raw and settled municipal wastewaters is suchthat it is difficult to give smaller ranges for thesecharacteristics. Generally it has been found that thegreater the diversity and larger the community fromwhich the wastewater originates, the less thecharacteristics will vary during the life of the plant. Also

' U l

S1i SDi

33

7 %

(/)

2 0 %

Solubleunbiodegradable

Particulateunbiodegrodoble

Particulatebiodegradable

Readilybiodegradable

FIGURE 2.4: Diagrammatic representation of the differentfractions of the total influent COD concentration. Numerical

values apply to raw municipal wastewater

organicnitrogen

Nul • fn u

N t l

i t «

Unbiodtgrodoblt

organic nitrogtnUnbiodtgradobktparticulattorganic nltrogtn

Biodtgradobt*organic nitrog*n

Fi»«t and»alintammonia

E q ( 2 . l 3 )

Eq (2 . l 4 )

FIGURE 2.5: Diagrammatic representation of the various frac-tions of the influent TKN in municipal wastewater. Numerical

values apply to raw municipal wastewater

2 7

the greater The diversity and larger the community, thesmaller the daily variations in wastewater characteristicsand flow.

A diagrammatic representation of the magnitudesof the different COD and TKN fractions in an approx-imately normal raw municipal wastewater is shown inFigures 2.4 and 2.5 respectively. With regard to theCOD fractions, Figure 2.4 shows that about 7% of thetotal COD is soluble unbiodegradable, 13% paniculateunbiodegradable, 60% paniculate slowly biodegradableand 20% soluble readily biodebradable. With regard lothe TKN fractions, Figure 2.5 shows that about 3% ofthe total TKN is soluble unbiodegradable organicnitrogen, 10% paniculate unbiodegradable organicnitrogen associated with the paniculate unbiodegrad-able COD, 12% biodegradable organic nitrogen and75% free and saline ammonia.

Caution should be exercised when selecting an un-biodebradable paniculate COD fraction (fuP): socialorganization of the community for whom the waste-water treatment process is envisaged must be criticallyexamined. For example, analyses of data reported bySutton ef a/. (1979) from Canada, indicate that the fup

fraction is 0,25 for raw wastewater. Such a high valueprobably arises from the use of garbage grinders. Asimilarly high fup is used in the design procedure followed by Black and Veatch (1979) [based on McKinney's

activated sludge model] which includes an unbiode-gradable organic material fraction which is equivalent toan fup value of 0,23. Wastewaters with high fup fractionswill result in greater sludge production than those withlow fractions and selecting too low a value will result inunderdesign of the treatment process and sludgedisposal facilities. This demonstrates the pitfalls inselecting wastewater characteristics uncritically andwithout recognition of the factors which contribute totheir nature and composition. It also demonstrates thedangers of applying uncritically, technology developedoverseas to local conditions and vice versa.

5. REFERENCES

BLACK and VEATCH (1979) Design procedure for complete mixingactivated sludge process. Civil-Environmental Division, P.O. Box8405, Kansas City, Mo.64114.

DOLD, P.L., fcKAMA, G.A. and MARAIS, V.v.R. (19801 A generalmodel for the activated sludge process. Prog. Wat. Tech., 12,47-77.

MARAIS, G.v.R. and EKAMA, G.A. (1S76) The activated sludge pro-cess pan 1 - Steady state behaviour. Water SA, 2, 4, 163-200.

LEVENSPIEL. 0- (1972) Chemical Reaction Engineering 2nd Ed.,John Wiley, New York, 465-469.

SUTTON, P.M., JANK, B.E., MONAGHAN, B.A. and MURPHY,K.I. 11979) Single sludge nitrogen removal systems. ResearchReport No 88, Water Technology Centre, Burlington, Canada.

CHAPTER THREE

INFLUENCE OF WASTEWATER CHARACTERISTICS ON PROCESS DESIGN

by


1. INTRODUCTION

In this chapter considerations in the design of nutrientremoval activated sludge processes are qualitativelydescribed to provide a background for the subsequentchapters in which the process design for COD removalonly (Chapter 4), nitrification (Chapter 5), nitrification-denitrification (Chapter 6) and biological phosphorusremoval (Chapter 7) are quantitatively set out - in thesechapters detailed equations, design charts, samplecalculations and recommendations for design arepresented.

2. RELATIVE IMPORTANCE OFCHARACTERISTICS

In the design of activated sludge processes thecharacteristics of the wastewater are of prime impor-tance. These characteristics govern both the selectionof the process and the removals of nitrogen andphosphorus attainable in the process.

Influent wastewater characteristics of crucial im-portance are:

(1) Mean influent COD and TKN concentrations (Stl

and N,, respectively) and mean daily flow (Q)(2) Influent readily biodegradable COD fraction (fbs)(3) Influent TKN/ COD concentration ratio (Nlt/St(l(4) Influent {Total PI/COD concentration ratio (P,,/SJ(5) Maximum specific growth rate of the nitrifiers at

the reference temperature of 20°C (ynrTlZ0)(6) Average minimum and maximum temperatures

(Tmin and TmaJ in the process.

Influent wastewater characteristics of lesser importanceare:

(7) Unbiodegradable soluble COD fraction (fus)(8) Unbiodegradable paniculate COD fraction (fup).

Influent characteristics normally not important in designare:

(9) Unbiodegradable soluble organic nitrogen fraction(fnu) and

(10) Ammonia fraction of the TKN (fna).

3. EFFECTS OF CHARACTERISTICS ON DESIGN

3.1 Influent COD and influent f low

The mean influent COD concentration, Si;,* and meanflow per day, Q, define the mass COD load day on theplant, M(Stl), by

M(Sti) = St l.Q

This COD mass consists of various fractions, as set outin Chapter 2, and these fractions are different for un-settled and settled municipal waste flows. However, toindicate the effect of the COD on the plant design, ac-cept a specific waste flow influent. Then, for a specifiedsludge age, once M{S,,) is known, theoretically themass of sludge in the reactor, M(XV) or MfXjt, and themass of oxygen required daily for carbonaceousdegradation, M(OC), also are known. The sludge agemust be specified as the mass is very sensitive to thesludge age. Knowing the mass of sludge, the volume ofthe reactor is found from the specified mixed liquor con-centration, Xv or Xt, i.e. the mass of sludge is "diluted"to the required concentration by providing the ap-propriate volume,

Vp = M(XJ/X, or M(Xt)/X,

Knowing the volume, one can now calculate, if sodesired (it is not essential! the nominal hydraulic reten-tion time, Rhn, of the reactor,

Rhn - Vp/Q

From a process design point of view the hydraulic reten-tion time is of little interest: For two waste flows, havingthe same COD fractional constitution, provided M(S,,)and Xv are the same for both, the volumes of the twoplants and the oxygen demand per day will be identicalirrespective of whether the first flow consists of a lowS,; and high Q and the second of a high S,, and low Q.The nominal hydraulic retention time will differ, the firstplant having a short and the second a long retentiontime respectively but this does not affect the degrada-tion action, except insignificantly. The influent flow isimportant in the design of the settling tanks - for thesame upflow velocity (overflow rate) in the secondary

•The mean influent COD, Sll( is ihe weighted mean concentration; it ts not the mean of the COD concentrations measured over the day on theinfluent. The magnitude of St, is found from measuring every hour, say, the COD, Suit), and flow, Qlt), and forming

Sx, = [ I S,i(t).Q(t)]/[ I Q(tl]

If the wastewater is settled, Si, refers to the measurements on the effluent from the primary settling tank.

3-1

settling tank, the size for the first flow (high) will belarger than that for the second flow (low). Thus know-ing only the COD load (and its fractional components),and the sludge age, the volume of the plant and themean carbonaceous oxygen requirements can beestimated rapidly, also, the mass of sludge to be wastedper day. Knowing the mean flow, the size of the settlingtanks can be estimated. Once these basic estimates areavailable, the effects of cyclicity in the flow and load onthe peak carbonaceous oxygen demand and peak flowthrough the settling tank, then can be superimposed.

It was stated above that the mass of sludge, M(XV),depends on M(S.,) and the sludge age. As the sludgeage increases so the M(XV) increases likewise, the massbeing roughiy proportional to the sludge age; this is il-lustrated in Figure 3.1 where the mass of sludge in theprocess per unit influent COD. for unsettled and settled"normal" municipal waste flows, are shown plotted ver-sus sludge age. Consequently the longer the sludgeage, ihe bigger the reactor volume required to deal withthe same IVKSJ. In Figure 3.1 are also shown plottedthe carbonaceous oxygen demands per unit COD inputfor settled and unsettled waste flows. As for M(XV) therespective magnitudes of M(OC) do not differ greatlybetween settled and unsettled wastes. The differencearises principally from the fraction of unbiodegradableparticuiate COD, fup, in the two wastes being higher inunsettled than in settled waste. However, the totalmasses of sludge and carbonaceous oxygen demandsfor the two flows will differ appreciably because approx-imately 40% of the COD is removed in the primarysettling tanks.

10

toUJ

8ft"

?UJ

a_ j

to

u.o

IAS

S

a.

ao

o

)iol

co-ao_o

OO

CO

>

- 0

Raw wostewaterSettled wcstewoter

1,0

0 10 20 30SLUDGE AGE ( d )

FIGURE 3.1: Mass of volatile solids and mass of carbonaceousoxygen demand per day per kg COD load per day versus sludge

age for settled and unsettled municipal waste flows.

3.2 Influent TKN, ^nrrtl and temperature

In the discussion above, to calculate the sludge mass itwas stated that the sludge age must be specified. Selec-tion of the sludge age probably is the most importantdecision the designer must make, particularly when theobjective includes nutrient removal. If removal of car-bonaceous material only is required, the sludge age canbe quite short; three days will be enough for CODremoval - the sludge age selected is dependent moreon the expected settleability of the sludge than on theCOD removal. If nitrification only is required the sludgeage must be sufficiently long to allow the nitrifyingorganisms to grow and metabolize virtually all theavailable nitrogen. If denitrification is required in addi-tion to nitrification, a fraction of the sludge mass mustbe kept unaerated to denitrify the nitrate generated. Asthe nitrifiers are obligate aerobes, this implies that innitrification-denitrification plants the sludge age mustbe increased above that where nitrification only is re-quired, to ensure that the aerobic sludge age is still ade-quate for nitrification.

The maximum specific growth rate constant of thenitrifiers, ^nrru defines the minimum aerobic sludge agefor the initiation of nitrification. This minimum sludgeage must be increased by a further 25 per cent to ensurethat nitrification is near complete. Now ^nm tends to bespecific to the waste flow and can range from 0,20 to0,70/day at 20°C. Also ^ is very sensitive totemperature, the value halving for every 6°C drop intemperature. The "aerobic" sludge age is given by1 /^nm; it is clear therefore that, in a nitrification plant, a' o w Mnm value for the waste flow coupled with a lowtemperature during the winter, will require a long sludgeage to ensure efficient nitrification throughout the year.In nitrification-denitrification plants the process sludgeage needs to be increased further to accommodate theeffects of the unaerated fraction of the sludge mass.These factors in combination can cause that whereas asludge age of 4 days for a Mnm = 0,3 at 20°C will beadequate theoretically for a purely nitrifying aerobicplant, for a nitr i fying-denitr i fying plant withMnm2o = 0.3, Tmin = 12CC and an unaerated sludgemass fraction of 0,5, the minimum process sludge agewil! be about 15-20 days to ensure a high efficiency ofnitrification.

Provided the sludge age is adequate, the mass ofnitrate formed by a process is given by the mass of TKNin the influent minus the mass of nitrogen incorporatedin the sludge mass wasted every day. This gives thenitrification capacity of the plant (usually expressed asnitrate formed/unit influent flow). The higher the TKNconcentration in the influent so, in general, the higherthe nitrification capacity - we can form the rough rulethat the nitrification capacity is proportional to the TKNconcentration in the influent.

3.3 Readily biodegradable COD, slowlybiodegradable COD

3.3. 1 Nitrogen removal

In the previous section it was mentioned that a fractionof the process sludge mass must be unaerated to

3-2

denitrify the nitrate generated in the aerobic fraction,but the magnitude of this unaerated fraction was stillundetermined. It is the intention now to describe quali-tatively how this unaerated fraction is determined andhow the influent characteristics affect the determina-tion.

In the absence of oxygen, nitrate serves as an elec-tron acceptor. With oxygen, the process designer hasthe problem of determining the oxygen demand forknown COD load; in denitrification the problem is todetermine the COD to denitrify a known mass of nitrate.The COD for denitrification comes from three sources(1) influent readily biodegradable soluble COD, (2) in-fluent slowly biodegradable paniculate COD and (3)particulate COD derived from the death of organisms.Usually (2) and (3) are lumped together in formulatingthe reduction of nitrate.

In the nitrification-denitnfication process becausethe same sludge mass generates the nitrate anddenitrifies it (called the single sludge process), thesludge mass normally is subdivided into two reactors inseries, the one aerated (to nitrify) and the otherunaerated (to denitrify).* Two basic configurationshave been developed, the Modified Ludzack-Ettinger(MLE) configuration, Figure 3.2, and the Wuhrmannconfiguration, Figure 3.3. In the MLE configuration, the

ANOXIC AEROBICREACTOR REACTOR

MIXED LIQUOR RECYCLE

IHfLUENT

WASTE FLOW

SETTLER

FIGURE 3.2: Modified Ludzack Ettinger nitrification denitrifi-cation activated sludge process

AEROBICREACTOR

ANOXICREACTOR

WASTE FLCW

SETTLER

FIGURE 3.3: Wuhrmann nitrification-denitrification activatedsludge process.

first reactor is left unaerated, the second is aerated. Thenitrate generated in the second reactor is transferred tothe first via the underflow and an inter-reactor (internal)recycle. In the Wuhrmann configuration the nitrate isgenerated in the first reactor and transferred by the nor-mal serial flow to the second unaerated reactor - an in-ternal recycle is not required, the only recycle being theunderflow recycle to transfer the sludge from the secon-dary settling tank back to the first reactor.

The denitrification action can best be explained byconsidering the unaerated reactors as plug flow reac-tors. (This can be shown to provide valid explanationalso for completely mixed reactors). In the unaeratedreactor of the MLE configuration (called the primaryanoxic reactor), the denitrification takes place at twosimultaneously occurring rates, the first due to the in-fluent readily biodegradable COD, the second due tothe particulate COD derived principally from the in-fluent, see Figure 3.4a. The first rate is very fast, thereaction being complete in 6 to 20 minutes and it utilizesall the readily biodegradable COD. The second rate isslower, at about 111 of the first rate, and continues forthe total time in the reactor. In the unaerated reactor ofthe Wuhrmann configuration, (called the secondaryanoxic reactor), a single denitrification rate is observed,due principally to the slowly biodegradable particulateCOD derived from death of the organisms, (Figure3.4b). This source of COD becomes available at a lowrate and denitrification rate accordingly is low, about2/3 that due to particulate COD in the MLE primaryanoxic reactor.

The nitrate removal can be expressed either as totalmass removed per day or as mass removed per litre offlow, i.e. as a concentration relative to the flowmgA(NO3-N)/j[ flow. The latter method is the most con-venient because it allows ready comparison with theTKN concentration in the influent fiow, mgTKN// flow.We have stated that the removal is due to the completeutilization of the readily biodegradable COD and the par-tial utilization of the biodegradable particulate COD.The denitrification achievable due to the latter can beformulated in terms of a denitrification rate constant, K,the active mass of sludge generated per litre of influentand the fractional mass of the process sludge containedin the specific anoxic reactor, as follows: (See Chapter6, Section 4).

In the primary anoxic reactor

ANt = Cx (mass of readily biodegradable COD perlitre of influent) + K2 x (mass of activesludge generated by one litre of influent) x(mass fraction of sludge in the primaryanoxic reactor) (3.1)

In the secondary anoxic reactor

AN3 = K3 x (mass of active sludge generated perlitre of influent) x {mass fraction of sludgein the secondary anoxic reactor) (3.2)

The values of K2 and K3 have been determined ex-perimentally and found to remain virtually constant forsludge ages between 10 and 30 days, at a fixed

' Other processes, in which the sludge is sequentially aerated and left unaerated, also have been proposed and built, either sequentially in time ina single completely mixed reactor or sequentially in space as, for example, in the Carrousel system.

3-3

TIO

N

<iccUJooa

z

a

V i K|Xo

1 K z x o \

1 st j 2nd Phase ^ \

i

TIMEla) Primarv anoxic

TIME(b) Secondary anoxic

FIGURE 3.4: Denltrification time behaviour in plug flow anoxicreactor, (a) primary anoxic and (b) secondary anoxic reactors.

temperature. This is also indicated by simulation usingthe general activated sludge model of Dold, Ekama andMarais (1980) as extended to include denitrification byvan Haandel, Ekama and Marais (1981). The two equa-tions developed above, expressed in technical terms arefor primary and secondary anoxic reactors respectively

AN, = f b 5 (1- f l J D - f u s )S ; i /8 ,6

Ti ' ' *us ' u p ''hnl (3.3)'

Rhril/Rhn and Rhn3/Rhn are the ratios of the nominalretention times of the primary and secondary anoxicreactors respectively with respect to the total. Theratios respectively are equal to the primary and secon-dary anoxic sludge mass fractions. Here as for aerobicbiological degradation, the nominal retention is not thebasic parameter even though one may formulate theequations to include it. From Eqs. 3.3 and 3.4 thedenitrificatton potential is a function of the influentCOD.S,., but is approximately so only, being modifiedby the magnitude of the unbiodegradable fractions ofthe influent COD, fus and fup.

In Eq. 3.1, all the readily biodegradable COD isutilized; the denitrification attained from this source inthe influent is very effective and constitutes about 50%denitrification achieved in "normal" municipal waste-water. The removal due to the particulate biodegradableCOD typified by K2 and K3 in Eqs. 3.1 and 3.2 inter alia,is directly proportional to the anoxic sludge mass frac-tions of the respective anoxic reactors (i.e. f,3 and fx3}.

"These equations are for illustrative purposes only, the meaning of thesymbols can be found in Chapters 4 to 6.

These fractions cannot be increased ad lib; for a fixedsludge age the magnitude of the sum of these fractionsis subject to the proviso that efficient nitrification mustbe maintained and this in turn is fixed by the aerobicsludge age. In this way there is an upper limit to the sumof anoxic sludge mass fractions which limits the nitrateremoval achievabie. Furthermore because K3 is roughly2/3 of K2, the nitrate removal due to K3 in the secondaryanoxic reactor per unit volume is only about 2/3 as ef-fective as that in the primary anoxic reactor so that perunit sludge mass fraction the primary anoxia reactor ismore efficient than the secondary anoxic reactor.

The nitrate removals, ANj and ANJ( are called thedenitrification potentials of the respective reactors. Theterm potential is used because in the primary anoxicreactor (as in an MLE process), for example, thedenitrification potential can be achieved only if suffi-cient nitrate, equal to the potential, is recycled to thisreactor - if less is recycled the full potential of the reac-tor is not realized and the denitrification performance(or denitrification capacity) is less than the denitrifica-tion potential. When the performance is less than thepotential then the performance can be increased by in-creasing the recycle. Now the recycle contains dissolv-ed oxygen in addition to nitrate, and the oxygen utilizesthe COD preferentially; in consequence, recycle ratiosin excess of 4 to 6 usually do not significantly improvethe performance, but even can cause a decrease in per-formance due to the high mass of DO transported to theanoxic reactor via the recycle flow. If more nitrate isrecycled than lhat defined by the potential, only thepotential value is removed. The excess is returned fromthe anoxic reactor to the aerobic reactor; in such asituation the recycle can be reduced appropriatelywithout affecting the nitrogen removal by the process.

The combined primary and secondary anoxicsludge mass fraction i.e. the total unaerated sludge

3 4

mass fraction, cannot be increased ad lib., for tworeasons:

U) Nitrification requires a minimum aerobic sludge ageand to preserve this, as the unaerated mass frac-tion is increased so also the process sludge agemust be Increased to preserve the minimumaerobic sludge age; increasing the sludge age,however, increases the total mass of sludge andhence the total volume of the reactor system. Fur-thermore, any increase has a relatively small effecton nitrate removal because the increase con-tributes to denitrification only through the K2 or K3

components and these are relatively inefficient.Practically, for South African conditions, the upperlimit to a sludge age appears to be about 30 days.

(2) When the nitrification rate is high the aerobicsludge age needed is small so that with a longsludge age it would be possible to have a largeunaerated sludge mass fraction. Experimental data,supported by simulation studies, show that forsludge ages between 20 and 30 days, if theunaerated fraction of a process increases aboveabout 70% at 20°C and about 50% at 12DC thetime available for aerobic degradation is insufficientto process all the particutate influent biodegradableCOD and the process acquires a degradation pat-tern still III understood. Consequently for designunder South African conditions for sludge ages of20 to 30 days, the unaerated mass fraction shouldnot exceed about 55 to 60%.

We have noted earlier that the denitrificationpotentials are approximately proportional to the influentCOD concentration, S1(. Now previously it was pointedout that the nitrification capacity is approximately pro-portional to the influent TKN. Consequently theTKN/COD ratio of the influent is a good parameter interms of which to assess process nitrification-denitrification behaviour because it gives approximatelythe ratio of the nitrate generated to ihe denitrificationpotential. Now accepting that

• ^irir11 will vary around 0,4 per day,

• the sludge age is in the range 20 to 30 days,

• "normal" municipal wastewater flows contain readilybiodegradable COD fractions of about 20% withrespect to the total influent COD,

• the internal recycle ratio is limited to a maximum of6:1, and

• the minimum winter temperature is 14°C,

it would appear that for these conditions, if the influentTKN/COD is greater than about 0,10, the denitfificationpotential is such that it is not possible to denitrify all thenitrate generated even if all the unaerated fraction of theprocess is located in the primary anoxic reactor. Conse-quently in this situation the MLE process is the mostefficient. If the influent TKN/COD ratio is less thanabout 0,09 then complete denitrification is possible.

Even if complete denitrification is possible this cannotbe accomplished by using an MLE process only for thefollowing reason;

Say the steady state nitrate concentration in the aerobicreactor is N mg;7; if an internal recycle ratio of a is im-posed and an underflow recycle ratio of s then(a +s)/(a + s+1) of this nitrate is recycled and1/ta-i-s-t-1) goes out with the effluent. To deal with the1/{a + s+1) fraction a secondary anoxic reactor needsto be incorporated also in the process, to give theBardenpho process first proposed by Barnard, 1973(Figure 3.5). Of course the secondary anoxic reactor perunit volume is not as efficient as the primary becauseK3 < K2 so that the denitrification potential for the sameunaerated total mass fraction will be lower than for anMLE process only, but this must be sacrificed and is thereason why complete denitrification is achievable onlyfor a TKN/COD S 0,09 instead of 0,10. This significantreduction in the upper limit of the TKN/COD is prin-cipally because the a-recycle should not be increasedabove a —6 (to limit the feedback of oxygen to theprimary anoxic reactor). If the oxygen in the recyclecould have been reduced to zero then very higha-recycles could be utilized and it might not benecessary to employ a secondary anoxic reactor.

PRIMARYANOXICREACTOR

AEROBICREACTOR


SECONDARYANOXICREACTOR

REAERATIONREACTOR

WASTE FLOW

t

FIGURE 3.5: Bardenpho process for biological nitrogenremoval - a combination of the modified Ludzack-Ettinger and

Wuhrmann processes.

As the TKN/COD ratio decreases below 0,09 so thesize of the primary anoxic reactor may be decreased andthe second anoxic reactor increased in order to maintaina low recycle, illustrated in Figure 3.6.

INCREASINGTKN/CODRATIO

FIGURE 3.6: Effect of TKN/COD ratio on relative sizes ofprimary and secondary anoxic reactors in the Bardenpho pro

cess.

3-5

3.3.2 Excess phosphorus removal

There is considerable support for the view that undercertain conditions some species of organisms, for ex-ample, the Acinetobacter spp, can take up phosphorusin excess of its metabolic requirements. Some of theacinetobacter species are obligate aerobes, others canutilize nitrate as electron acceptor. The species is slowgrowing with rather specific requirements to the type ofnutrient it can utilize; it can utilize, for example, many ofthe lower fatty acids, but not sugars, such as glucose.Because it is slow growing, in aerobic systems, withmixed cultures of organisms, its growth is suppressedby faster growing organisms, because the faster grow-ing organisms utilize most of the substrate and even-tually dominates over the slow growing ones. This is thereason why it is not easy to find evidence ofacinetobacter in purely aerobic processes.

From the discussion above, how then does it hap-pen that the poly-P organism proliferates in somesystems containing unaerated fractions? Assume thatthe organism already contains some excess phosphate.This phosphate is present as the radical PQ\ but boundtogether to form polyphosphate chains. The bonds be-tween the radicals in the chain contain appreciableenergy. If the acinetobacter is discharged to ananaerobic reactor (i.e. a reactor that has neither nitratenor oxygen present), in which substrate is presentsuitable for growth of Acinetobacter spp, it breaksdown some of the bonds in the polyphosphate chain,releasing phosphate radicals to the surrounding liquid.The bond energy thus released is utilized to absorb,complex and store the substrate in the organism - wesay it sequesters the substrate. In contrast, in theanaerobic zone, because no external electron acceptor(oxygen or nitrate) is available, the non poly-Porganisms cannot utilize the substrate. On leaving theanaerobic zone and entering an anoxic zone (withnitrate present) or an aerobic zone (with oxygenpresent), the poly-P organisms now has electron accep-tors available and metabolizes the stored substrate se-questered for itself in the anaerobic zone i.e. it can growwithout being in competition with the non poly-Porganisms for The balance of the energy still available inthe surrounding medium.

in the anoxic and aerobic zones the poly-Porganisms, beside using the sequestered substrate forgrowth, also use it to take up phosphate radicals (PO )from the surrounding medium, again to form phosphatechains for use subsequently when returned to theanaerobic zone. Thus, by having the propensity to storepolyphosphate the acinetobacter and other poly-Porganisms have a positive advantage over the non poly-P organisms to obtain energy for themselves in the ap-parently inhospitable anaerobic reactor. This explainsthe apparently anomolous growth response of the poly-P organisms.

The behaviour of the poly-P organisms in theanaerobic reactor is not yet well understood but a sum-mary of the behaviour appears to be as follows: When asubstrate is added that can be utilized directly by thepoly-P organisms, such as acetate or butyric acid, therelease of P (to obtain energy for sequestering thesubstrate) is very rapid, and the release is proportional

to the mass of substrate added. Also, the net removal ofP from the anaerobic-anoxic-aerobic system increasesas the substrate added increases, but insufficient datahas been collected to express these responses quan-titatively.

When sewage is added to the anaerobic reactor thebehaviour differs in two aspects from that when acetateis added: (1) only a fraction of the COD added is im-plicated in the P release and P removal; (2) the rate of Prelease is significantly slower with sewage than withacetate addition and the release is of a different kineticorder.

(1) The influent COD that induces P release seems tobe associated with the readily biodegradable CODfraction. It has been observed consistently inlaboratory scale processes that the release andremoval of P increases and decreases as this frac-tion increases and decreases (see Figure 3.7)although it is not yet clear, and highly unlikely, thatall the readily biodegradaie COD is implicated inthis action.

With regard to the paniculate biodegradableCOD, this fraction appears to have little effect on Prelease. This is not an unreasonable conclusionbecause the paniculate COD requires adsorption,storage and extracellular breakdown prior to ab-sorption by the organisms; even under aerobic con-ditions the rate of metabolism of the particulateCOD is only at 1/7 to 1/10 the rate at which readilybiodegradable COD is metabolized, Dold, Ekamaand Marais (1981).

(2) The slower rate of P release with sewage additioncompared to that with acetate addition seems to in-dicate that the substrate suitable for P releasebecomes available indirectly to the poly-Porganisms. This is further supported by the obser-vation that whereas with acetate addition the Prelease appears to be zero order with respect toacetate concentration, (Figure 3.8), with sewageaddition the release appears to conform to a firstorder reaction; in a plug flow anaerobic reactorreceiving sewage the rate of release is high initiallybut dies away steadily along the reactor until it isvinually zero. This is also indicated in a seriessystem of anaerobic reactors (Figure 3.9). The firstorder nature of the reaction in the series system isverified by plotting the log (release/reactor) versusthe cumulative retention time, to give a linear plot,(Figure 3.10). Comparison between the release in aplug flow reactor and a completely mixed reactoralso shows that for the same anaerobic mass frac-tion if the mass fraction is so small that all the P isnot released in the plug flow reactor, the release inthe completely mixed reactor is less than in theplug flow one, a behavioural pattern in conformityto that expected where a first order reaction con-trols.

According to the observed behaviour describ-ed above, at sludge ages about 20 to 25 days, theanaerobic mass fraction of a completely mixedreactor should be about 0,2 to ensure that 90% or

3-6

400-Q

o

3 5 0

u>CO

3 0 0

2 5 0

r«cyd«d = 0

oO

OO

150 a:uJQQCD

J00

a5 0

20 "50 *

DAY OF OPERATION5 0

FIGURE 3.7: Experimental data showing the strongdependence of excess P removal (AP) on readily biodegradable

COD in the influent (Sbsi)-

= 15

zoP 10

o

V)

§ltoO

o

1

1 mtial Acetate

* 5 £-• 10 O• 15 ~• 20 O

- time ofacetateaddition

i

cone, (mg/l)

estimatedvaluet fofcompieteuptake

MLVSS « 1008

• *

• •

* *

i

I

mg/l

•

i

5 -

o 40 120 160eoT I M E (min)

FIGURE 3.8: Phosphorus release in a batch anaerobic reactorwith different concentrations of acetate added. Note themagnitude of the release is proportional to mass of acetateadded but the rate of release is independent of the acetate con-

centration.

3-7

10<noxa.

S t i "BOO mq/\Haw t w l m t i rSIu 4q% ag* " 1 0 4Tt * p » 20 • CS,'bil

T x o 0,042Reoctor I

0,084

2J .

0,1263

J

0,1604

0 50ACTUAL ANAEROBIC

( m l n )

100RETENTION

150TIME

FIGURE 3.9: Phosphorus concentration in a four in seriesanaerobic reactor system, in a modified UCT process, receiving

unsettled municipal sewage influent.

c•

V.

a.

J_

uLJB-

PH

OR

US

1

moI

100

7050

30__

10

7

3

3

2

i

-

-

-

Reoctor

t

. ; •1 Z

i

5

\ ft.

•42 C.00-)

1 2

! ( -

5 < T

> ^ , -

/ 1 ~

Raw «oirt*cl«r _Sludgi agf - 13 4Ttmp • 20" C

t 1 I "

\

>

-

K

0,126 0fl6» o \

3 4 *

0 50 100 150ACTUAL ANAEROEIC RETEIiTION TIME

FIGURE 3.10: Phosphorus release per reactor in a four in seriesanaerobic reactor system in a modified UCT process receiving

unsettled municipal sewaye influent (see also Figure 3.9)

more of the P is released; if it were a plug flow reac-tor the mass fraction needs to be about 0,15 to givethe same percentage release. Because the releaserate is first order there is little to be gained by mak-ing the anaerobic mass fraction greater than about20%. Mass fraction of 10% or less again will be toosmall allowing only about 50 to 60% of the poten-tial release to take place. Then again, subdividingthe anaerobic reactor into say two reactors in seriesgives improved usage of the anaerobic reactormass fraction and requires a mass fraction of about15% to give the same release as a completely mix-ed reactor of 20%. These fractions apply only forsludge ages of 20 days and longer.

The reasons for the different responses observedwith acetate and sewage readily biodegradable CODadditions are not yet clear but the following explanationcould be valid: Most of the fractions of readilybiodegradable COD, apparently, are not in a chemicalform suitable for utilization by the poly-P organisms. Inthe anaerobic zone the non poly-P organisms, for sur-vival, because no external electron acceptors areavailable, partially utilize the readily biodegradableCOD, via the Embden-Meyerhof or other equivalentpathways, to generate a small amount of energy, bycreating internal or endogenous electron acceptors: Theoutcome of this action is a breakdown of the originalCOD molecules to lower fatty acids and other similarforms. These are rejected by the organism and becomeavailable to the poly-P organisms for sequestration.Thus, where substrate is present in a form unavailableto the poly-P organism, the presence of non poly-Porganisms can modify some of the COD fractions toforms suitable for use by the poly-P organisms. Thekinetics of this transformation will govern the kinetics ofsequestration by the poly-P organisms because theCOD breakdown is the rate limiting step. The non pofy-P organisms are, in terms of this hypothesis, essential tothe growth of the poly-P ones.

3.3.3 Implication on process development

The rationalization of the poly-P behaviour leads to anumber of conclusions that are of significance todesigners of P removal processes:

1

(21

In terms of the hypothesis set out above, thegeneration of the lower fatty acids by acid diges-tion of primary settling tank underflow, as propos-ed and implemented by the Municipality of Johan-nesburg, makes good sense and merits seriousresearch attention.

Any external electron acceptor entering theanaerobic reactor, like nitrate or oxygen, will beutilized immediately by the non poly-P organismsacting on the readily biodegradable COD andthereby reduce the COD concentration available forthe generation of substrate suitable for sequestra-tion by the poly-P organisms.

(3) The fact that the whole sequestration action is slowmeans that adequate anaerobic mass must be

3-8

made available to process as much as possible ofthe readily biodegradable COD source. However,because the reaction is first order it also impliesthat there is little merit in enlarging the anaerobicmass ad lib., each increase giving a decreasingreturn on the previous increase. Also, seriesanaerobic reactors are more efficient and hencesmaller anaerobic mass is necesary, to achieve thesame effect than with a single completely mixedanaerobic reactor.

With regard to design implications, the basic re-quirement in a design is to provide an anaerobic zonewhich receives the influent containing the readily bio-degradable COD fraction and so becomes available,directly or indirectly, to the poly-P organisms for se-questration. This zone must be large enough to allowthe sequestration action to be as near complete aspossible in order to obtain maximum effect. Further-more, the zone should not receive any form of the exter-nal electron acceptor (nitrate, nitrite or dissolved oxy-gen) as these cause the fast growing non poly-Porganisms to utilize the substrate before it can be se-questered by the poly-P organisms.

The first explicit design of a configuration, in whichan anaerobic reactor is proposed, was due to Barnard*(1976). He accepted the Bardenpho process (Figure 3.5)as the basic configuration to nitrify and denitrify. A frac-tion of the primary anoxic reactor was isolated to serveas an anaerobic reactor. This reactor was placed at thehead of ihe process and received the underflow recycleand the influent flow (see Figure 3.11). The reasoningbehind this configuration was that the Bardenpho sec-tion of the process would give complete nitrification anddenitrification so that the underflow would contain littleor no nitrale, in this fashion allowing optimal conditionsin the anaerobic reactor and maximum release of P. Theprocess configuration has been variously called theModified Bardenpho, Phoredox or 5 stage Phoredoxprocess.


MIXED UOUOR RECYCLEANAEROBICREACTOR

INFLUENT,

AEROBIC SECONDARYREACTOR ANOXIC

REACTORREAERATIONREACTOR

WASTE FLOWSETTLER

SLUDGE RECYCLE t

FIGURE 311- Phoredox (or modified Bardenpho} process forn i t r i f icat ion-deni t r i f icat ion-bio iogica! excess phosphorus

removal.

In assessing the functioning of the Phoredox pro-cess, the reservation of a portion of the unaerated massfraction to the anaerobic reactor concomitantly reducesthe denitrification potential of the process. Thus, com-

pared to the Bardenpho process for the same totalunaerated mass fraction, the Phoredox has a lowerdenitrification potential. Consequently the maximumTKN/COD ratio the Phoredox process can denitrifycompletely is reduced. For a "standard" raw sewage in-fluent, i.e. 20% of the influent COD is readily bio-degradable, at sludge ages of approximately 20 days,maximum unaerated mass fraction of 0,55 and ananaerobic mass fraction of 0,15, it would seem that themaximum TKN/COD ratio for complete denitrificationshould not exceed 0,08. For settled sewage the readilybiodegradable COD is higher = 25% of the total influentCOD and the maximum TKN/COD ratio is about 0,09.This ratio is very dependent on the fraction of readilybiodegradable COD because this fraction contributessignificantly to the denitrification. Consequently thevalues given here are relative ones only - it is importantto estimate the true readily biodegradable fraction ex-perimentally.

As the TKN COD ratio increases from low values,so it will be found that the relative volumes of theprimary and secondary anoxic reactors change accor-dingly. To achieve optimum denitrification response atlow TKN/COD ratios the primary reactor is small andthe secondary reactor large; at high TKN.'COD ratios,near or at the maximum allowed, the primary reactor islarge and the secondary very small.

Should the TKN/COD ratio exceed the maximum,nitrate will appear in the effluent. In the Phoredox pro-cess this nitrate is recycled in the underflow to theanaerobic reactor where it utilizes preferentially some,or all (depending on the concentration of nitrate) of thereadily biodegradable COD, thereby reducing themagnitude of the COD available for sequestration by thepoly-P organisms. Now the concentration of nitrate inthe underflow recycle is the same as that in the effluentand 1 mg (NOj-NI) removes 8,6 rng readilybiodegradable COD, consequently, a concentration of5 mg 7 (NOj-N) in the underflow with a recycle ratio of1:1 will remove 43 mg// of this COD fraction which, foran influent of 500 COD/7 will reduce the readilybiodegradable COD from approximately 100 to60 mg / i . The only operational means for reducing themass of nitrate transferred to the anaerobic reactor is toreduce the underflow recycle ratio but this procedurehas only limited possibilities because low recycle ratiosmay give rise to difficulties with the secondary settlingtank - at low recycles the sludge mass builds up in thetank and may cause failure of the tank when the SVI ishigh, or loss of sludge in the effluent may occur fromrising sludge due to denitrification in the sludge mass.

The principal difficulty with the Phoredox con-figuration, under high TKN/ COD ratios, arises from thefact that the anaerobic reactor performance is depen-dent on the nitrate in the effluent. To make theanaerobic reactor independent of the effluent nitrate,the University of Cape Town (UCT) process configura-tion was developed, Figure (3.12). The basic denitrifica-tion configuration is the Modified Ludzack-Ettinger pro-cess (Figure 3.2). The anaerobic reactor is, like the

"The basis for proposing the anaerobic reactor, by Barnard, differs fundamentally from the hypothesis suggested in the previous sections. Hesuggested that the redox potential need to be lowered sufficiently to induce P release. He noted the adverse effect on P removal by nitratedischarged to the anaerobic reactor but suggested that this arose from the nitrate limiting the depression of the redox potential. These aspectsare dealt with in greater detail in Chapter 7.

3-9

ANAEROBIC ATOXIC AEROBICREACTOR REACTOR REACTOR

RECYCLE MIKED LIQUOR

FLOW

INFLUE* EFFL JENT

FIGURE 3.12: University of Cape Town (UCT) process torn i t r i f i ca t ion-den it r i f i ca t i onb io log i ca l excess phosphorus

removal

Phoredox, placed at the head of the works but the re-cycle flow to this reactor, the r-recycle, is taken fromthe primary anoxic reactor. By regulating the main inter-nal recycle from the aerobic to the anoxic reactor, thea-recycle, so that the denitrification potential of theprimary anoxic reactor is not exceeded, a nitrate-free ef-fluent is assured to the anaerobic reactor.

In the UCT process, if, say, an r-recycle of 1:1 isused the concentration in the anaerobic reactor is halfthat in the rest of the plant and to get the requiredanaerobic mass fraction the volume of this reactormust, in consequence, be increased to twice thevolumetric size of that in the Phoredox for the sameanaerobic mass fraction. The main advantage of theUCT process is that by operationa! means the anaerobicreactor can be made independent of the nitrate in theeffluent, and this can be maintained up to a TKN/CODratio of about 0,14. As an alternative to the Phoredoxprocess, at low TKN/COD ratios, i.e. TKN/COD < 0,08,the UCT process is also satisfactory although the nitrateremoval will not be as high as in the Phoredox as it is notpossible ever to get zero nitrate in the effluent from aUCT process.

To overcome the need for explicit control of thea-recycle, the Modified UCT process was developed,(Figure 3.13). In this process the primary anoxic reactoris divided inio two fractions, the first having an anoxic

ANAEROBIC ANOXIC AEROBICREACTOR RCACTORS REACTOR

MIXED LIQUOR RECYCLES

WASTE FLOWSETTLER

INFLUENT EFFLUENT

FIGURE 3 13: Modified UCT process.

mass fraction of about 0,1. The a-recycle is dischargedto the second anoxic reactor and remains fixed at say4:1 ratio irrespective of whether nitrate is recycled backto the aerobic reactor or not, provided the a-recycleratio is so large that it brings enough nitrate to the se-cond anoxic reactor to equal or exceed its denitrificationpotential. The underflow recycle 's-recycle) dischargesto the first anoxic reactor, to reduce the nitrate to zero.By means of this configuration zero nitrate in ther-recycie can be maintained for TKN/COD ratios up toabout 0,11.

4. CLOSURE

There are numerous aspects and details in the design ofnutrient removal plants that have not been dealt with inthis short review - attention was focussed on the majoraspects of a fundamental nature only. For details ofdesign reference needs to be made to Chapters 4- to 7.Designs that conform to the recommendations laiddown in these chapters, for treating normal municipalwasieflows, settled or unsettled, will have a high expec-tation of phosphorus removal of about 1,7 to 2 mg PO4-P per 100 mg influent COD, i.e. a AP/COD of 0,017 to0,02 mgP/mg influent COD. This removal ratio canvary, depending principally on the fraction of readilybiodegradable COD in the influent. Should the influentP/COD concentration ratio exceed 0,017 TO 0,02, verylikely all the phosphorus will not be removed biologicallyfrom the wastewaler and provision necessarily will haveto be made for chemical precipitation of The remainingphosphorus; should the influent P/COD be less than0,017 to 0,02, then very likely removals down to0,5 mgP/jf can be achieved for a high proportion of thetime.

The above biological P removal propensity inAP/COD forms an approximate basis for judging theperformance of any designed plant. The plant shouldnot be assessed on the basis of the effluent P concen-tration only, but on the AP per unit influent COD. If theAP/COD achieved is less than 0,017 to 0,02, the causefor the lower removal should be sought in thewastewater characteristics, process design or operationof the plant, or in combinations of these three. Theresults of such investigations would be of great value toresearch workers and designers in the endeavour tobuild plants that perform optimally.

5. REFERENCES

DOLD, P.L., EKAMA, G.A. and MARAIS, G.v.fl. (1980J A generalmodel for the activated sludge process, Prog. Wat, Tech., 12.47-77.

VAN HAANDEL, A.C., EKAMA, G.A. and MARAIS, G.v.R. (1981)The activated sludge process part 3 - Single sludge denitrifica-tion, Water Research. 15, 1135-1152.

3-10

CHAPTER FOUR

CARBONACEOUS MATERIAL REMOVAL

by


1. INTRODUCTION

Basically all aerobic biological treatment processesoperate on the same principles: the various treatmentprocesses i.e. trickling litters, aerated lagoons, conTact-stabilization, extended aeration, etc. differ only in theconditions under which the biological reactions are con-strained to operate. The activated sludge process com-prises the flow regime in the reactor, the sizes, numberand configuration of the reactors, recycle flows, in-fluent flow and other features incorporated eitherdeliberately, or present inadvertently or unavoidably.Whereas the response of the organisms is in accor-dance with their nature, that of the process is governedby both the organism characteristics and the physicalfeatures that define the process.

The ecology of the suspended sludge mass in thereactors of the activated sludge processes is complexcomprising numerous species of bacteria and manyspecies of higher organisms that prey on the bacteria.Despite the complex nature of the ecosystem, it wouldappear that the kinetic behaviour of the process, insofaras it reflects the energy removal from the wastewater,can be modelied as if the system were governed by anequivalent or surrogate bacterial mass, havingcharacteristics derived in part from the observedresponse of the treatment processes. It must be em-phasised that this surrogate mass has characteristicsvery different from those of a pure bacterial culture.This point is mentioned because criticism of this ap-proach has often hinged on the incompatibility of thebehaviour of the equivalent bacterial mass with thatobserved in a pure bacterial culture.

in this chapter the basic kinetics of the car-bonaceous material removal, and its incorporation inthe process equations are briefly stated, and applicationto design is set out in detail. The design equationspresented are for the steady state completely mixedsingle reactor activated sludge process. The steadystate solution is of prime importance because (1) itforms the basis for prelimtnary design of any activatedsludge process be it aerobic, anoxic-aerobic oranaerobic-anoxic-aerobic; (2) the relationships also areuseful in indicating operational control procedures forthe process.

If certain minimum conditions are imposed on theprocess, nitrification will occur. With nitrification theoxygen demand can show a considerable increase, ofup to 40% more than that required for carbonaceousmaterial oxidation. Nitrification is not dealt with in thischapter but is considered in detail in Chapter 5.

2. BIOLOGICAL KINETICS

When a heterotrophic organism population underaerobic conditions is brought into contact withbiodegradable organic material* consisting of solublereadily biodegradable and participate slowlybiodegradable COD fractions (see Chapter 2), itsresponse may be described qualitatively as follows:

(1) The soluble readily biodegradable COD passesdirectly through the cell wall and is metabolized at ahigh rate.

(2) The particulars slowly biodegradable COD is ad-sorbed onto the organisms and storage of CODtakes place. This reaction is very rapid and effec-tively removes all the paniculate and colloidal CODfrom the wastewater. The stored COD is brokendown by extracellular enzymes and transferredthrough the cell wall and metabolized in the samemanner as the readily biodegradable COD fractionin (1) above. The rate of enzymatic breakdown isrelatively slow and constitutes the limiting rate inthe overall synthesis reaction, at about one tenth ofthe rate for the readily biodegradable COD.

(3) A fraction of the COD metabolized is converted in-to new cell material, the balance is utilized togenerate energy to bring about the conversion andeventually, lost as heat. The energy is generated bysupplying an electron acceptor, oxygen, so that theoxygen utilized is directly related to the COD lost.There is strong experimental evidence that the frac-tion of COD converted to metabolic material is con-stant with respect to the COD utilized, Payne(1971}. This fraction is called the specific growthyield coefficient, YC0D {mg COD/mg COD). Usual-ly the coefficient is expressed in terms of thevolatile solids formed per unit of COD utilized, Yh,(mg VSS/mg COD) and is related to YrOD by Eq.1.6, i.e. Yh - YC0D/fcv.

(4) Concomitantly with (3) above, but distinct from it,there is a net loss of live mass called endogenousmass loss. Not all the live mass that disappears perse is lost as energy, a fraction remains (17 to 20 percent, McKinney and Symons, 1964} as unbiode-gradable organic residue, called endogenousresidue. The mass of oxygen to be supplied for en-

*Also called substrate or biodegradable COD.

4-1

dogenous mass loss is proportional to the volatilemass that disappears from the system. This netbehaviour is believed to be the outcome of a morecomplex set of mechanisms involving the growthof primary organisms (e.g. bacteria) and predatorson these organisms (e.g. protozoa). Dold, Ekamaand Marais (1980) proposed and formulated the netbehaviour as a uniform organism mass having adeath-regeneration component, an approachwhich approximates the prey-predator systemunder certain circumstances.

Using the death-regeneration approach Dold et al(1980) in their development of the general activatedsludge process model, and van Haandel and Marais(1981) in their extension of this model to denitrification,have shown that the response of all the known processtypes (including nutrient removal processes) can besimulated very closely, even under cyclic flow and loadconditions. Under constant flow and load, the death-regeneration approach reduces to the endogenousmass loss approach. This latter approach leads torelatively simple process equations (Marais and Ekama,1976} and constitutes the basic design equations in thismonograph, see Eqs. 4.6 to 4.19.

3. PROCESS KINETICS

3.1 Mixing regimes

In the activated sludge process, the mixing regime inthe reactor and the sludge return influence the responseof the process - therefore consideration must be givento reactor kinetics.

There are two extremes of mixing; completely mix-ed and plug flow.

In the completely mixed regime the influent is in-stantaneously and thoroughly mixed (theoretically) withthe reactor contents. Hence the effluent flow from thereactor has the same constitution as the reactor con-tents. The reactor effluent flow passes to a settlingtank; the overflow from the tank is the stabilized wastestream, the underflow is concentrated mixed liquor andrecycled back to the reactor.

In the completely mixed system the rate of returnof the underflow has no effect on the reactor processexcept if an undue sludge build-up occurs in the settlingtank. The shape of the reactor is approximately squareor circular in plan, and mixing is usually by mechanicalaerators but can also be effected by bubble aeration.Examples are extended aeration plants, aeratedlagoons, Pasveer ditch and completely mixed activatedsludge plants.

In a plug flow regime, the reactor usually is a longchannel type basin. The influent is introduced at oneend of the channel, flows along the channel axis and ismixed by air spargers set along one side of the channel.Theoretically each volume element of liquid along theaxis is assumed to remain unmixed with the elementsleading and following. Discharge to the settling tanktakes place at the other end of the channel. To in-noculate the influent waste flow with organisms, theunderflow from the settling tank is returned to the in-

fluent end of the channel. This creates an intermediateflow regime deviating from true plug flow conditionsdepending on the magnitude of the recycled underflow.Conventional activated sludge plants are of the in-termediate flow regime type with recycle ratio varyingfrom 0,25 to 3 times the influent flow rate. If the recyclerate is raised very high, the mixing regime approachesthat of complete mixing.

Intermediate flow regimes are also achieved byhaving completely mixed reactors in series, or by step-aeration. In the latter, the influent is fed at a series ofpoints along the axis of the plug flow type reactor. Bothconfigurations require, for inoculation purposes, recycl-ing of the sludge from the settler to the influent point.

The mean kinetic response of an activated sludgeprocess, i.e. sludge mass, daily sludge production, dailyoxygen demand and effluent quality is adequately, in-deed accurately, given by assuming the process is com-pletely mixed and the influent flow and load are con-stant. This allows the plant volume, the mass of sludgewasted daily and mean daily oxygen demand to bedetermined by relatively simple formulations. Peakoxygen demands which arise under cyclic flow and loadconditions can be estimated subsequently quite ac-curately by applying an adjustment to the mean dailydemand, so also for other parameters. These ad-justments have been developed from simulationstudies, using the general model of Dold, Ekama andMarais (1980), on processes operated under cyclic andunder constant flow and load conditions. For details seeChapter 6, Section 7.6.2.

3.2 Sludge age

A diagrammatic sketch of the completed mixed processwith sludge recycle is given in Figure 4.1. This sketchdiffers from that usually presented in that the sludgewasted each day is abstracted directly from the reac-tor - the common practice is that the waste sludge isabstracted from the secondary settler underflow.Sludge abstraction directly from the reactor leads to amethod of control of the sludge age, called thehydraulic control of sludge age, which has significantadvantages compared to control by wastage via theunderflow, see Section 10.

PROCESSREACTOR

WASTE SLUDGE

L ABSTRACTION

StCONOARYSETTLINGTANK

EFFLUENT

FIGURE 4.1: A diagrammatic representation of the completelymixed activated sludge process with hydraulic control of

sludge age and sludge recycle.

4-2

The sludge sge, Ri( in days, is defined by:

_ mass of sludge in reactor , , . . . . . .R = __— — 2 — (days) (4.1)

mass of sludge wasted per dayBy abstracting the sludge directly from the reactor,

the wasted liquor and reactor liquor concentrations arethe same. If a sludge age of, say, 10 days is required,one tenth of the volume of the reactor must be wastedevery day.* This can be achieved by a constant draw-off at flow rate, q, per day where q.1 day ~ v, thevolume to be wasted,hence

(4.2)

where

V., = volume of the process reactor (ft

3.3 Nominal hydraulic retention time

In activated sludge theory the volume of the process perunit of volume of influent flow is known as the nominalprocess hydraulic retention time i.e.

(4.3)= Vo/Q

where

Rhn= average nominal process hydraulic retentiontime (d)

Q = daily average influnnt flow rate (f/d)

3.4 Effluent COD concentration

Under normal activated sludge process operating condi-tions in South Africa, where the sludge ages are in ex-cess of 10 days (to ensure nitrification and nutrientremoval) the nature of the influent COD in municipalwastewaters is such that the COD concentration in theeffluent is inconsequential in the process design - thesoluble readily biodegradable COD fraction is complete-ly utilized in a very short period of time (less than 1 hour)and the paniculate COD, whether biodegradable or un-biodegradabie, is adsorbed or enmeshed in the sludge

floes, and settles out with the sludge in the secondarysettling tanks. Consequently, the effluent COD concen-tration is comprised virtually wholly of the soluble un-biodegradable COD (from the influent) plus the COD ofthe sludge particles which escape with the effluent dueto imperfect operation of the secondary settling tank(see Chapter 3, Section 3.3). Hence the effluent CODconcentration, S:e, is approximately given by

S1e = Sus (4.4a)

for filtered samples, or,

Sle = SUE + fcv.Xve '4.4b!for unfiltered samples, where

Sus = unbiodegradable COD in the effluent= susi = fU5 SJt (mgCOD/f) (see Eq. 2.3)

Xve = volatile solids concentration in the effluent(mgVSS'f)

fcv = COD/ VSS ratio of the volatile solids= 1,48 mgCOD/mgVSS.

4. PROCESS DESIGN EQUATIONS

Once it is recognized that all the COD in the influent, ex-cept the soluble unbiodegradable COD, is either utilizedby the micro-organisms to form micro-organism mass,or remains in the process and accumulates as inertsludge mass, it follows that the mass of sludge produc-ed and the carbonaceous oxygen demand in the pro-cess are functions of the mass of COD to be treated dai-ly; the greater the daily COD mass load, the greater thesludge production and carbonaceous oxygen demand.

The equations** given below give the masses ofsludge generated in the process and the oxygen de-mand for COD removal as a function of the total CODload, the waste water characteristics i.e. the un-biodegradable soluble and particulate COD fraction (fu5

and f Jp) and the sludge age. The constants in the equa-tions i.e. the specific yield coefficient Y;i, the specificendogenous mass loss rate bh, the unbiodegradablefraction of the active mass f, and the COD,'VSS ratio ofthe sludge fcv, as well as their temperature dependen-cies are given in Table 4.1.

TABLE 4.1 KINETIC CONSTANTS AND THEIR TEMPERATURE DEPENDENCY FOR THE STEADY STATECARBONACOUS DEGRADATION ACTIVATED SLUDGE MODEL [ex MARAIS AND EKAMA, 1976)

Constant

Yield coefficient (mg VSS.-mg COD)Endogenous respiration rate { d}Endogenous residue (mg VSS/mg VSS)COD VSS ratio (mg COD'mg VSS)

mbol

Yh

bhf

f f

TemperatureDependency

Equation

remains constantbhT^bhjoS11-201

remains constantremains constant

0

1,0001,029

-_

StandardValue<20lC)

0,450,240,201 4?

Eq.

(4,5)

"This assumes thai the mass of sludge in ihe secondary settling tsnks is negligible with respect IO thai in the process reactors. This assumptionis valid when the process is operated at high underflow recycle ratios (=1:1} and when the sludge age is longer than about 3 d3ys. (see Sec-tion 10 below).

'The derivation of these equations is given by Marais and Ekama 119761 and a list of symbols used in the equations is given in Appendix 1.

4-3

In the equations below, the prefix M denotes themass of the parameter which follows in brackets.

(1) Influent

M(S,,) = Q.S,, mgCOD/d (4.6)

M(Sbl) = Q.Sbl mgCODd (4.7a)

= Q.S,, (1 - f u s - f u p ) mgCOD/d (4.7b>

= M(S I IJI1- f u t - f u p ) mgCOD/d (4.7c)

M(X,i) = Q.X,, mgVSS/d (4.8a)

= Q.fUDSti/fcv mgVSS/d !4.8b)

= M(S,,)fup/fcv mgVSS/d (4.8c)

(2) Process

M<Xa) = VpXa

M(Xe) - VpXe

M(X() .= VpXi

M(XV> = V pX,

M(OC) = VpOc

(a) The Active Volatile Solids Mass (mg VSS)

M(Xa) = M(Sbi)YhRs/(1+bhR5) (mg)

= (1- f u s - f o p )M(S t l ) YhRs/(1+bhRs)(4.10)

(b) The Endogenous Residue Volatile Solids Mass (mgVSS)

where

f, - MLVSS/MLSS ratio of the sludge.

(f) The Carbonaceous Oxygen Demand (mg 0/d)

M(0c) = M(0 synthesis) + M(O endogenous mass loss)

= (1 -fcvYh) M(Sbl) + fcvd - f ) bn M(Xa) (mg 0/d)

= M(StlH1-fus-fupH<1-fcvYh)

Y RA f it *»u ' h n s 1

bwR(4.15)

ti'V

Knowing the mass of mixed liquor (MX, or MXJ inthe process, the volume of the process is determinedfrom the value specified for the MLSS or MLVSS con-centration, X,, or X, respectively,

Vp = MIX.l/X, - M(XV)/XV (4.16)

mg

mg

mg

mg

mg

VSS

VSS

VSS

VSS

0/d

(4

{4

(4

(4

(4

,9a)

.9b)

.9c)

,9d)

.9e)

M(Xe) = f bhRs M(Xa) (mg) (4.11

(c) The Inert Volatile Solids Mass (mg VSS)

M£X,) = M(XJ Rs img)

= M(S t i)(f t in/fcv)Rs (4.12)

(d) The Total Volatile Suspended Solids Mass (mgVSS)

Knowing the volume Vp, the total nominalhydraulic retention time, Rhn, then is found from ihespecified daily average flow rate Q, using Eq. 4.3.

The above design equations lead to the followingimportant conclusions:

The mass of volatile solids in the reactor is a func-tion only of the mass of COD utilized per day and thesludge age. Consequently, insofar as the mass of sludgeis affected, it is immaterial whether the mass of CODutilized arises from a low daily flow with a high CODconcentration, or a high daily flow with a low COD con*centration; provided M(S,,) is the same in both in-stances, the masses of sludge will be identical.However, the hydraulic retention times will C'fer, beinglong in the first and short in the secon^ instance,respectively. The hydraulic retention time, therefore, isincidental to the COD mass utilized, the MLVSS and thedaily flow — it serves no basic kinetic function in steadystate operation.

The active volatile mass M(X3) in the process is thelive organism mass which performs the b;odegradationprocesses. The remaining volatile masses, M(Xe) andM(X,) are inactive and do not serve any function insofaras the biodegradation mechanisms in the process areconcerned. The active fraction of the sludge mass withrespect to the volatile solids fav is given by

fav = M(Xa)/M(Xv) (4.17)

substituting for M(Xa) and M(XV) and noting that

M(Sbl) - (1 -fUB - fup) M<StL) and

= (fUD/fcv).M(Sti) yields

M(X

= vhRsM(Sb,(1+bh Rs)

(mg)

1 - 1 + f h R |r- - 1 +tbhHs+ (4.18)

(4.13)

(e) The Total Suspended Solids Mass (mg TSS)

M(X,) = MtXJ/fj (4.14)

< ' Tus ' u p '

If the total suspended solids mass (TSS) is used asthe basis for determining the active fraction, the the ac-tive fraction of the sludge mass with respect to the totalsuspended solids fat is given by

fal = frfav (4.19)

where

f, - MLVSS/MLSS ratio of the sludge.

4-4

The design equations set out above form the start-ing point for all the activated sludge processes con-sidered in this manuscript, from the relatively simplesingle reactor completely mixed aerobic process, to themore complex multi-reactor anoxic-aerobic and theanaerobic-anoxic processes. For these more complexsystems the basic equations apply only if the constraintsimposed on the processes are adhered to. Provided thisis done the effects of nitrification and nitrification-denitrification and the associated oxygen demands canbe formulated as additional to the basic equations. Thatthis "simplified" approach is adequate for design hasbeen established by the close correlation achieved whencomparing the mean response of the process predictedby the general theory (and verified experimentally) withthat predicted by the basic and additional equations.

It is important to take cognizance of the con-straints. For example, in anoxic-aerobic processes anupper limit is imposed on the fractional mass of sludgethat may exist unaerated. If exceeded, then inter a//a themass of sludge produced per day increases and the car-bonaceous oxygen demand decreases below thatpredicted by the basic equations. The reason for thesedeviations lies in the kinetics of degradation of the slow-ly biodegradable particulate material - if the aerobicfraction of the sludge mass is too small the particulatematerial is only partially metabolized and accumulates inthe process as additional volatile solids; concomitantlythe carbonaceous oxygen demand is reduced. Evidentlywhere such conditions apply the basic equationsbecome inappropriate; however approximate solutionsare sometimes obtainable by simulation using thegeneral model, (see Arkley and Marais (1982) andSehayek and Marais (1981) for a detailed discussion).

5. STEADY STATE DESIGN CHART

Plots of Eqs. 4.10 to 4.15 and 4.18 to 4.19 are shown inFigure 4.2 for a unit mass of M(Stl). The values of thekinetic constants, Yh, bh, fcv and f are those listed inTable 4.1 for 20°C and have been verified in extensivelaboratory and pilot scale investigations (Ekama andMarais, 1978). The values of the unbiodegradable par-ticulate and soluble COD fractions (fup and fus respec-tively) are listed in Table 4.2 and are for raw and settledmunicipal wastewaters.

The plots in effect give the masses M(Xa), M(Xe),M(XJ, MIXJ and M(XJ contained in the reactor for aunit mass of COD applied per day for sludge ages up to30 days. Also shown is the associated mass of oxygento be supplied per day for carbonaceous materialdegradation and the active fraction of the sludge withrespect to VSS (Xv) and TSS (X.). The volumes of thereactors at two different sludge ages will be in directproportion to the masses M(X.) or M(XJ if the mixed li-quor concentrations, X( or Xv are specified to be thesame in both plants.

TABLE 4.2 INFLUENT WASTEWATER FRACTIONS FORSETTLED AND UNSETTLED SEWAGE

Sewage Fraction

Soluble unbiodegradable fraction, fUi

(mg COD'mg COD)Particulate unbiodegradable fraction, fUE.(mg COD/mg COD)MLVSS/MLSS ratio (f,)

SewageUnsettled Settled

0,05

0,13

0,75

0,08

0,04

0,83

10RAW WASTEWATER

"2 oT =

' Y h

bn' cv

. f

-

120" C0,45

= 0,24

• 1,48- 0,20

I

' up

* « .

f |

/

*->

» 0s 0

= 0

4 -

1

f

,13

0 5

7 5

/

-

0c

10 20SLUDGE AGE(d)

,2

.0

^ • V

0,6

0,4

02.

Q

- *

OE

M^

z

>-g

o

<JOQ

lie

aa.•

8o

Q .

a.r

CO

D

(

tyJtL

CO

u.

tr

10

Q

"> 6u_O

2 4 -

SETTLED WASTEWATER1

' u pfu i

- \ fi\

1 1

"0,04= 0,08

0,83

/ ^ 4 x

i

T

bh

fr

f

>

*1

i

- 20*C= 0,45• 0,24• 1,48

= o,zo

FIGURE 4.2: Active (Xo), endogenous (XJ . inert (X,). volatile(Xv), total (X,), and carbonaceous oxygen demand IOJ per kgCOD applied to the biological process, and active fraction withrespect to volatile solids UaJ and total solids (fot) versus sludgeage for raw (Figure 4.2a, left) and settled [Figure 4.2b, right)

wastewaters at 20°C

LU

,0

0,8

0^

0,4

0,2

a.

S8§£LLJ OU c*< Jt

5CD

10 2O 301"SLUDGE AGE (d)

4-5

The diagrams show that the active mass increasesfairly rapidly with an increase in sludge age up to about10 days sludge age, whereafter it increases onlymarginally. The carbonaceous oxygen demand shows asimilar behaviour; at zero sludge age the oxygen de-mand is that due to synthesis, additional oxygen de-mand at sludge ages greater than zero is due to en-dogenous respiration which is proportional to the activemass. In contrast the fractions of endogenous and inertsolids increase rapidly relative to the active mass atsludge ages greater than 10 days and consequently inprocesses treating raw wastewater at sludge ageslonger than 10 days, only a small fraction of the sludgemass is active.

In design, the following sequence of calculation issuggested:

Select values for fup and fuS which are believed tobest represent these fractions of the wastewater. Selectthe MLVSS/MLSS ratio of the sludge produced. Deter-mine:

(a) SUOI (Eq. 2.4), SUSl (Eq. 2.3), X,, (Eq. 2.6)

(b) MIS,,) and/or M(Sbl)

(c) M<Xa}(Eq. 4.10), M(X,.)(Eq. 4.11), M(X;)(Eq. 4.12),

M(XJ(Eq. 4.13), M(Oc)(Eq. 4.15) and M(X,)(Eq.4.14)

(d) VpIEq. 4.16)

(e) Rhn(Eq. 4.3)

(f) Ste (Eqs. 4.4a or b)

In the design sequence set out above the inputCOD and its characteristics will be governed by thespecific waste flow. The parameter that requires selec-tion is the sludge age; this will depend on the objectivesfor the plant, for COD removal only, nitrification etc.Specification of the sludge age, therefore, is an impor-tant design decision and requires special consideration,see Section 11. For our general purpose however, wewill proceed to show how the various parameters suchas reactor volume, oxygen demand, etc. are affected bythe sludge age.

6. PROCESS VOLUME REQUIREMENTS

In Section 4 above a procedure is set out whereby thevolume of the reactor is estimated. This involves deter-mining the mass of sludge accumulated for a specifiedsludge age and a specified COD mass loading per day;the volume is then determined by diluting the mass ofsludge accumulated to a specified mixed liquor concen-tration. Once the volume is fixed, the retention time, oraeration time, also is fixed (by Eq. 4.3), Hence retentiontime, or aeration time, is immaterial in the design and itis merely a consequence of the kinetics governing theprocess. This point is mentioned specifically becausesome design procedures lay stress on the aeration timeas a design parameter, a procedure that can result inserious miscalculation of Jhe volume requirements.

Compare, for example, two plants operating at thesame sludge age, both receiving the same mass of CODper day but the first at high concentration and low flowand the second at a low concentration and high flow. Ifdesigned on a specified retention time, the volume ofthe first will be much smaller than that of the secondwhereas the sludge mass contained in the reactors willbe identical. Consequently the first plant may have aninordinately high MLSS concentration which may causeproblems in the secondary settling tank.

Retention time therefore is a completely Inap-propriate parameter for design and other purposes suchas a criterion for comparing the tankage requirements ofdifferent plants.

To form an overall impression of the volume re-quirement and in order to make valid comparisons onthe volume requirements between different plants, thevolume per unit COD input is a much superiorparameter. This method involves the specification of anMLVSS or MLSS concentration (Xv or X.) from whichthe reactor volume is determined from the sludge massproduced per unit COD load at the required sludge age(see Figure 4.2). Furthermore, in order to assess the ef-fects of primary sedimentation it is preferable to relatethe volume requirements to the input COD to the treat-ment works; for example for settled wastewater, theCOD load on the process is reduced i.e. S,l(fjrocess,= S,,,work5, (1- f r p s) where frps = fraction of CODremoved by primary sedimentation. It is St;iprocessi whichfixes the sludge mass and because for settledwastewater Sti is up to about 40% less than for rawwastewater, the process volume for settled wastewateraccordingly will be much lower than that for raw waste-water.

From Eqs. 4.13, 4.14 and 4.16 the process volumerequirements on the basis of process COD load is foundby calculating the mass of sludge, M(X.), generatedfrom the total influent COD load M(S,,) and assuming anMLVSS/MLSS ratio for the sludge mass i.e.

f i = 0 . 5 , R d+bhRs) +

fcv'(4.20)

and hence the process volume requirements per unitCOD load to the works for a specified MLSS concentra-tion X, and assumed MLVSS/MLSS ratio fs is given by

• Cl-fus-fUD)Yh(1+fbhRs

(1+bhRs)

(4.21a)

Taking the wastewater characteristics for rawwastewater given in Table 4.2, and the values of thekinetic constants given in Table 4.1, a plot of processvolume requirements in rnVkg COD applied per day ver-sus sludge age for different MLSS concentrations inkg/m3 is given in Figure 4.3. dearly, the lower thespecified MLSS concentration and the longer thesludge age, the larger the process volume per unit CODload.

The process volume requirements may also bedetermined from the equivalent COD load per capita.The vertical axis on the right hand side of Figure 4.3 is

4-6

enxcro*

K r-

r zLLi UJ

2UJ

oroUJor

H

z:o

UJ

2 Q-J OO -1

aUi OIf) OUJ (_O x

3,5

3,0

2,5

2,0

L5

1,0

0,5

0,0

1 1 1 1 1

Raw Wasrawater Load/Capita/d= o,io Kg COD

COD removal by primarysedimentation = 40%

ReGCtor ruLSS Jrconcentration y^ _

Raw S ^^ ^

".:"-5--—:Settled

WaMewatert i

350

300

250

200

!50

100

50

00 5 10 15 20 25 30

SLUDGE AGE (d)

FIGURE 4.3- Process volume requirements in m3/kg COD rawwastewater load versus sludge age at different average pro-cess MLSS concentrations for raw and settled wastewater(assuming 40% COD removal by primary sedimentation). Pro-cess volume requirements in //capita also given based on araw wastewater COD contribution of 0,10 kg COD/capita/d.

a.o

UJ

r.z>-io>L0to

Oor

scaled on the basis that the equivalent COD load percapita for raw wastewater is 0,10 kg COD/d. Hencetreating raw wastewater at a sludge age of 25 days andan MLSS concentration of 5 kg/m\ a process volumeof 145 f per capita is required or 1,45 m3kg COD ap-plied per day.

The comparative process volume requirements forsettled wastewater per unit COD load on the treatmentworks also can be determined from Eq. 4.21a, providedthe COD fraction removed by primary sedimentation isincorporated in Eq. 4.21a by multiplying right-hand sideof Eq. 4.21a by (1 -f rps) and inserting the appropriate fljpand fus values for settled wastewater i.e.

Vp ^ Rs ((1 - f u 5 - f u p )Y h ( i +fbhR.) f ^S t iQ~ X ^ { " (UbhR5) " + f c v

} f 1 " f f p s l

(4.21b)

where S,, = total COD concentration to treatmentworks.

The process volume requirements for settledwastewater are shown also in Figure 4.3 for the f fui

and f, values given in Table 4.2 assuming primarysedimentation removes 40% of the COD load <f,ps =0,40). From Figure 4.3, treating settled wastewater at asludge age of 25 days and an MLSS concentration of5 kg/m3, a process volume of 0,60 mVkg raw COD ioadto the works or 60 £ per capita is required. Comparingthe process volume treating raw or settled wastewatersit can be seen that a significant reduction in processvolume can be obtained through primary sedimenta-tion.

7. CARBONACEOUS OXYGEN DEMAND

The mean daily carbonaceous oxygen demand per unitCOD discharged to the process is shown plotted versussludge age at 20°C for raw and settled wastewater inFigures 4.2a and 4.2b respectively (calculated from Eq.4.15) - f o r treatment of raw wastewater St, = C0D ofthe raw wastewater; for treatment of settled waste-water S,;^(C0D of raw wastewater)(1-frp). Thesediagrams show that for sludge ages longer than 15 daysthe increase In carbonaceous oxygen demand per unitprocess COD ioad {MtOJ/MtSJ} is marginal with in-crease in slucfge age for both raw and settled waste-water. The carbonaceous oxygen demand per unit pro-cess COD load for raw and settled wastewater is usuallywithin 10% of each other, with the demand for settledwastewater being the higher value. This is becausecompared to raw wastewater, more of the total COD insettled wastewater is biodegradable, i.e. O-fu s~fup)usually is greater for settled wastewater than for rawwastewater (see Eq. 2.7) but the difference is small.This can be seen in Figures 4.2a and 4.2b - at 20 dayssludge age, the carbonaceous oxygen demand per unitCOD load is 0,635 kgO/kgCOD for raw wastewater and0,682 kgO/kgCOD for settled wastewater. Althoughthere is only a small difference in carbonaceous oxygendemand per unit process COD had between raw andsettled wastewaters there is a substantial difference inthe O2 demand per unit COD load to the works. Forsettled wastewater, this is given by 0,682 (1 - f ) whichfor frps = 0,40 gives 0,41 kgO/kgCOD toad on the works.

4-7

Clearly primary sedimentation will lead to significantaeration energy savings - because primary settlingtanks remove about 30 to 40% of the raw influent COD,the carbonaceous oxygen demand for settled waste-water will be about 30 to 40% lower than that for rawwastewater.

The carbonaceous oxygen demand is the oxygendemand, for the oxidation of the infiuent COD only. Innutrient removal processes oxygen is also required fornitrification i.e. the oxidation of ammonia to nitrate, andsome oxygen is recovered during denitrification i.e.reduction of nitrate to nitrogen gas. The total oxygendemand for a nutrient removal process therefore isgiven by the sum of the carbonaceous and nitrificationdemands less that recovered by denitrification. The pro-cedures for calculating the oxygen demand for nitrifica-tion and the oxygen recovered by denitrification arediscussed in Chapters 5 and 6 respectively.

Owing to the daily cyclic nature of the COD loadthe carbonaceous oxygen demand will vary con-comitantly. The TKN load will also vary daily cyclically;in nitrification-denitrification plants the oxygen demandfor nitrification and oxygen recovery with denitrificationconjointly with the carbonaceous demand give rise to anet cyciic total demand. From simulations using thegeneral model of Dold et alempirical rules for estimatingthe daily peak and trough net oxygen demands havebeen developed and are discussed in Chapter 6, Section7.6.2.

0,5

•o\

o 'i.03

Q O

0,2a:

CO

in 0,1

to

0,00 10

SLUDGE AGE20(d)

30

FIGURE 4.4: Daily tola! suspended solids (TSS) sludge production in the activated sludge process (secondary sludgel per unitCOD load on the process versus sludge age for raw (fuP = 0,13.fus = 0,05 and fj = 0,75) and settled (fUp = 0,04, fufi = 0,D8 and

f, = 0,83) wastewaters at 20°C

8. DAILY SLUDGE PRODUCTION

The mass of sludge produced per day by the activatedsludge process is equal to the mass of sludge wastedper day and is called waste activated sludge or secon-dary sludge. From the definition of sludge age (see Eq.4.1), the mass of sludge produced per day M(AXt) isgiven by the mass of sludge in the process |M(Xf)} divid-ed by the sludge age (Rs) i.e.

M(AXt) = M(Xt)/Rs mgTSS/d

Substituting Eq. 4.14 for MtX,) and simplifying,yields the sludge produced per day per mg COD load onthe biological process i.e.

M(AX,) _ 1 , ( l - f u p - f , s ) Y h f"M(ST - O (1+bhRs)

n + f b * F V + f J

(4.22)

A plot of the daily total sludge mass (TSS) produc-ed per unit COD load on the biological process (Eq.4.22) versus sludge age (R3) is shown in Figure 4.4 forraw and settled wastewater (see Table 4.2 for waste-water characteristics). From Figure 4.4, it can be seenthat the mass of sludge produced in the activatedsludge process (per unit COD load on the biological pro-cess) decreases as the sludge age increases for bothraw and settled wastewater but the rate of decrease isnegligible at sludge ages longer than 20 days. Treatingsettled wastewater results in lower secondary sludgeproduction per unit COD load on the biological processthan treating raw wastewater. This is because theunbiodegradable particulate COD fraction (fup) in settledwastewater is lower than that in raw wastewater.

Temperature effects on secondary sludge productionare small - sludge production at 14°C* is about 5%greater than at 20°C, a difference which is completelymasked by the uncertainty in the estimates of thewastewater characteristics fup and fus and theMLVSS/MLSS ratio (f,) of the sludge.

Although the sludge production of a plant treatingsettled wastewater is lower than that of a plant treatingraw wastewater, the total sludge mass of sludge to bedisposed of for a plant treating settled wastewater ishigher because the total sludge includes both primaryand secondary sludge - in plants treating raw waste-water, only secondary sludge is generated. The totalsludge produced treating settled wastewater is slightlyhigher (by 10 to 20%) than that treating raw wastewater- this is because some of the settleable solids arebiodegradable and are oxidized in the activated sludgeprocess,

The daily production of secondary and primarysludges (when primary settling tanks are included in thetreatment works) is the mass of sludge that needs to bedisposed of by sludge handling methods. Sludge handl-ing and disposal is not dealt with in this manual but thedesign of nutrient removal activated sludge processesshould not be seen as separate from the design of thesludge disposal unit processes, in fact all unit processesof the treatment works from raw water pumping toultimate disposal of sludge, should be viewed as an in-tegrated system where the design of one unit processdepends on the unit processes before it, and decisions

"Sludge production al 14°C can be calculated by adjusting bn fortemperature using Eq. (4.5l in Table 4.1.

4-8

on its design may affect the design of unit processesfollowing it. The following examples will illustrate this -(1) in sludge handling practice there is evidence that amixture of primary and secondary sludges thickens anddewaters better than secondary sludge alone; however,in biological nutrient removal practice it has been foundthat when treating raw wastewater, better removals ofnitrogen and phosphorus are obtained than whentreating settled wastewater - ease of sludge thickeningand dewatering therefore need to be weighed againstthe higher nutrient removal; (2) when phosphorus-richwaste biological sludges are subjected to anaerobicconditions, the phosphorus is released from the soiidphase to the liquid phase; consequently, supernatantfrom waste sludge gravity thickeners may be phosphaterich and should not be discharged to the activatedsludge process - in design consideration should begiven to chemically treating the gravity thickener super-natant to precipitate the phosphorus or, thickening thesludge by dissolved air flotation to prevent phosphorusrelease; (3) to improve biological excess P removal, theJohannesburg Municipality specifically operates one ofthe anaerobic digesters in an acid condition for the pro-duction of short chain organic compouds which arethen discharged to the anaerobic reactor of the ac-tivated sludge process (see Chapter 3, Section 3.3)- consideration must be given at the design stage tothe desirability of providing a specially designed facilityfor this purpose.

9. NUTRIENT REQUIREMENTS FOR SLUDGEPRODUCTION

All live biological material and many inert organicmaterials contain nitrogen and phosphorus. In thevolatile suspended solids in the activated sludge pro-cess, the nitrogen content (as N with respect to V5S,fn) ranges between 9% and 12% with an average ofabout 10%, and the phosphorus content (as P withrespect to VSS, fp), under purely aerobic conditions,ranges between 1% and 3% with an average of about2,5%.

Under daily average conditions, the mass ofnitrogen incorporated into the sludge mass is equal tothe nitrogen content of the daily waste sludge. Accept-ing that the nitrogen content of the active, endogenousand inert volatile fractions are equal (which in most in-stances is a reasonable assumption), and equating theconcentration of nitrogen required for daily sludge pro-duction per litre influent, Nb, to the nitrogen content ofthe waste sludge yields

Q.NS - fr, M(XV)/RS

Substituting for M(XV) from Eq. 4.10 and solving forthe nitrogen requirements per unit COD, Ns/Stl yields

— = f11 tcv

(4.23)

where

fn = nitrogen content of the volatile solids= 0,10 mg N/mg VSS

To estimate the P requirement for daily sludge pro-duction, it is not correct to assume that the P content ofthe active, endogenous and inert volatile fractions is thesame. The P content of the endogenous and inertvolatile fractions (fp) remain approximately constant at1,5% whereas the P content of the active mass (y) canvary from 3% to as high as 35% depending on the dif-ferent conditions in the process. Indeed, it is the pro-pensity of the active mass to accumulate large quan-tities of P that is exploited in biological P removal pro-cesses. Accepting the different P contents of the activefraction and the endogenous and inert fractions, theconcentration of P per / influent (Ps) required for dailysludge production per mg/ i COD load (S,.) is given by

j

St:

where

' u p

(1+bhR s

(4.24)

y - P content of the active mass (mg P/mgVASS)fp - P content of the inert and endogenous mass

(mg P/mg VSS) = 0,015

The P incorporated into the sludge mass per unitCOD (Py/S,,) is also the P removal from the wastewaterper unit COD,AP/STl. In the activated sludge process,under purely aerobic conditions the y coefficient isabout 0,03 mgP/mgVASS: when an anoxic reactor isincluded in the process, the coefficient is increased toabout 0,06 mgP/mgVASS; when an anaerobic reactoris incorporated in the process, then, depending,amongst others, on the readily biodegradable COD con-centration in the anaerobic reactor, the y coefficient canvary between 0,06 and 0,35 mgP/mgVASS (seeChapter 7, Section 3.3). By stimulating high y coeffi-cients, with the aid of the anaerobic reactor, high Premovals may be effected. This is the basis of biologicalP removal and is discussed in detail in Chapter 7.

A plot of Eqs. 4.23 and 4.24 is given in Figure 4.5for fn = 0,10mg N/mg VSS, y ~ 0,03 mgP/mgVASS,fP = 0,015 mgP/mgVSS, Yh = 0,45 mgVSS/mgCOD,bh = 0,24/d at 20°C, fcv - 1,48 mgCOD/mgVSS, andf -0 ,20 , versus sludge age (Rs) for raw (flIS = 0,05 andfljp = 0,13) and settled ( f ^ -0 ,08 and f^ -0 ,04) waste-waters.

From Figure 4.5, it is evident that greater quantitiesof nitrogen and phosphorus are required for sludge pro-duction from raw than from settled wastewaters. This isbecause greater quantities of sludge are produced permgCOD process load when treating raw wastewaters.Furthermore, the nutrient requirements decrease as thesludge age increases because nett sludge productiondecreases as sludge age increases. Generally, for sludgeages greater than 10 days, the nitrogen removal at-tributable to nett sludge production is less than0,025 mgN/mgCOD applied. As the influent TKN, CODratio for domestic wastewater is in the approximaterange 0,07 to 0,12, only a mine fraction of the influentnitrogen is removed by sludge wastage; additionalnitrogen removal is obtained only by nitrification anddenitrification, the kinetics of which are discussed inChapters 5 and 6. With regard to P removal, as the totalP/COD ratio of domestic wastewaters usually ranges

8

Otr

0,14

0,12

0,10

0,08

0,06

0,04

0,02

NUTRIENT REQUIREMENTS

0,C00

RAW WASTEWATERSETTLED WASTEWATER

\

^APPROXIMATE RANGE OF INFLUENT„ TKN/COD AND P/COD RATIOS OF

MUNICIPAL WASTE WATERS

ADDITIONAL NUTRIENT REMOVALREQUIRED FOR COMPLETEREMOVAL

NUTRIENT REMOVAL ATTRIBUTABLETO SLUDGE PRODUCTION

0,035

0,030

0,025

0,020

0,015

0,010

0,005

5 10SLUDGE

15 20AGE (d )

25

FIGURE 4.5: Normal nutrient nitrogen and phosphorus require-ments per kgCOD load in the activated sludge process versussludge age together with influent nutrient loads per kgCOD.

9E

Xin

between 0,015 to 0,030, only a minor fraction ofphosphorus, 0,005 mgP/mgVSS, can be removed bythe aerobic activated sludge process; excess biologicalP removal is obtained by stimulating high y coefficientsin processes by incorporating an anaerobic reactor.Details of excess biological P removal are considered inChapter 7.

10. PROCESS DESIGN AND CONTROL

The parameter of fundamental importance in the designand control of the activated sludge process is the sludgeage. It is instructive to consider how this parameter in-fluences both the control and design of the process andto contrast it to other design and control parametersthat have been implemented in the past.

10.1 Sludge age versus load factor as controlparameter

A widely accepted basis for design (especially in NorthAmerica and Europe) is the Load Factor (LF). It is alsoknown as the Sludge Loading Rate (SLR) and has beenused principally when the BOD5 forms the influentenergy measurement parameter. The objective of theload factor was to form an estimate of the Food /Micro-organism (F/M) ratio and as a result the load factor wasdefined as

LFmass of BOD load/d

mass of sludge in process

M(SB0D)MtXJorMOC,)

(4.25)

where

M*SB0D) = mass of BOD load per day (mg)

M<Xv),(X<Xt) - mass of volatile (mgVSS} or total(mgTSS) suspended solids in pro-cess respectively.

A difficulty with the approach is that both theM(XJ and MIX,) measurements include unbiodegrad-able and inert material. To represent the F/M ratio cor-rectly, the LF should be defined as follows:

LF = F/M = M(ASB0D)/M(Xa) (4.26)

where

M(ASB0D) = mass change in BOD across plant perday (mgBOD/d)

M(X8) = active mass in process (mgVASS)

As a practical control parameter, the F/M ratio asdefined by Eq. 4.26 is virtually unusable principallybecause M(Xa) cannot be measured directly. It is for thisreason that the LF was approximated by Eq, 4.25* withMIX,) or M(X.| as the solids parameter. In this form,however, the LF is an uncertain parameter againstwhich to correlate the response of the process becauseMIX,) depends on both the biodegradable COD and theinert particulate material in the influent. For example,for the same biodegradable COD concentration, the LFwill differ between settled and raw wastewaters where-as the oxygen demand and M(Xa) will be the same forboth. Furthermore the LF will be different depending onwhether the settleable solids are measured in terms ofMLVSS or MLSS. When the COD forms the influentenergy measurement parameter, the LF is defined by

* Because the effluent BOD usually is very small with respect to the influenl BOD, the change in BOD mass load per day M(ASBOD) is closely ap-proximated by the BOD mass load per day M(SBOD)-

4-10

Eq. 4.27 which is even further out of place because theCOD measure includes the unbiodegradable soluble andparticulate COD fractions

LF = M(S COD'

MlXv)orM{X t|(4.27)

where

MIS COD' ~ mass of COD load per day.

If Eq. 4.27 is written in terms of the mass change ofCOD per day across the plant i.e. M(ASCGDI, then ac-count of only the soluble unbiodegradable COD istaken. The unbiodegradable particulate COD affects theLF by increasing both the M(ASCQ[)) and M(XJ or M(X,)so that different load factors can be obtained dependingon the unbiodegradable particulate COD fraction.

To overcome the problem of the unbiodegradableCOD fractions on the LF, Marais and Ekama (1976)defined the F/ M ratio in terms of COD as the masschange of biodegradable COD per day per unit activevolatile mass and, to distinguish this parameter from theLF, called it the Substrate Utilization Rate with respectto the active mass, SUR;, i.e.

SUR. = F/M = M(ASC0D)/M(Xa) (4.28)

where

MIASCuD) = mass change in biodegradable CODper day.

Again the SURft parameter is of little practical valuebecause neither the M(ASJ (due to the unknown unbio-degradable particulate COD content of the wastewater)nor the M<Xa) can be measured. However, using Eq.4.28 as a basis, Marais and Ekama (1976) showed thatthe Substrate Utilization Rates with respect to the ac-tive mass (SURa), volatile mass (SURJ and total massI SUR,) can all be written in terms of sludge age, but dif-ferent SUR's are obtained for a fixed sludge age depen-ding on the parameters for measuring the sludge con-centration (Xa, Xv or X,) and the influent energy (COD orBOD) (see Figure 4.6).* Similarly for the same sludgeage, different values of the LF or SLR can be obtaineddepending on the measurement parameters for thesludge concentration and influent energy.

The dependency of the SUR's, LF's or SLR's onthe measurement parameters for the sludge concentra-tion and influent energy is undesirable because it pro-vides a permanent source of error and misunderstand-ing at, and between all levels of technical staff concern-ed with the process. But perhaps of greater importanceis the fact that a// the different SUR's, LF's or SLR's arefunctionally related to the sludge age - no additional in-formation can be derived from the SUR's, L F's or SL R 'sthat cannot be derived from the sludge age.

The sludge age therefore can replace completelythe LF, SLR and SUR as a reference parameter. Fur-thermore, the sludge age can be fixed by a simple pro-cess control procedure if the process is appropriatelydesigned. This control procedure is much more prac-tical than procedures based on the LF approach.

0,0

SETTLED WASTEWATER

0 5 10 15 20SLUDGE AGE (d)

25 30

FIGURE 4.6: Comparison between various substrate utilizationrates (SUR) and sludge age.

10.2 Process control

The usual procedure of the activated sludge process in-volves keeping the sludge concentration at some valueeither specified from design considerations or establish-ed from experience on the plant behaviour. Knowingthe sludge concentration (Xv or Xt), to check that the LFis within the desired limits requires extensive testingthroughout the day of the influent COD (or BOD) con-centration and flow pattern to determine the daily COD(or BOD) mass load. To calculate the correspondingsludge age, a history of the daily sludge mass wastageneeds to be kept. Often in conventional plants, thewaste sludge is abstracted from the secondary settlingtank underflow to benefit from the thickening functionof the secondary settlers. However, the sludge concen-tration of the underflow varies considerably with thedaily cyclic flow through the plant (see Figures 4.7 and4.8). Hence, each time sludge is wasted it is necessaryto measure the underflow concentration and calculatethe sludge age.

From the above, it is clear that in order to know thesludge age or the LF accurately, intensive testing of theplant and influent is required. This is manageable onlarge plants where the technical supervision is ade-quate, but on small plants both the Load Factor andsludge age usually are unknown.

The discussion above demonstrates that wastingsludge from the secondary settler underflow does notresult in a simple and accurate control of either thesludge age or the LF. However, by using a differentsfudge wasting procedure, an accurate and simple con-

"The derivation of some of these SUR values is given by Marais and Ekama and those for well senled wastewater (i.e. fUD = 0,000! is shown inFigure 4 6. It should be noted that if volatile men material is present in the influent i.e. fuf, >0,00, e.g. in partially settled and raw wastewaters,a set of SUR values can be calculated for each selected value of fup.

4-11

CO

- I

16000

12000

eooG -cc

zUJ

R 40O0

0000

S e t t l i n g tank

under f low

— a.m.

1 J

noon

Process ~~reoctor

p.m. -

! 1 I 1

OOhOO 06h00 I2KX)

TIME OF DAY

24h00

FIGURE 4 7 Experimental data from a full scale activatedsludge plant illustrating stability of the reactor MLSS concen-tration compared with that in the secondary settler underflow/.

LOW FLOW PERIODi.e. high recycle ratio for

constant underflow rate

HIGH FLOW PERIODi.e. low recycle ratio for

constant underflow rate

Qi

settledsludge

INFLUENT FLOW RATE ( m 5 / d )

UNDERFLOW RATE ( m3/d)

SLUDGE DENSITY IN REACTOR (mgMLSS/l)

SLUDGE DENSITY IN UNDERFLOW (mgMLSS/l)

(Qi +Qr)

OrX t

Qr/Q UNDERFLOW RECYCLE RATIO

FIGURE 4.8: Increased sludge accumulation and higher under-f low sludge concentration for (i) decreased recycle f low for fix-ed influent flow and (it) increased influent flow for fixed recycle

flow.

4-12

trol of the sludge age is possible. This different sludgewastage procedure is called the hydraulic control of thesludge age.

10.3 Hydraulic control of sludge age

Hydraulic control of sludge age was first proposed andimplemented in a generalized form by Garrett in 1958,based on a method of "modified sewage aeration" im-plemented by Setter, Carpenter and Winslow 119451.* Itoperates as follows: Suppose a sludge age of 10 days isspecified. A satellite settling tank or dissolved air flota-tion unit is provided, completely independent of thesecondary settling tank, to which 1/10 of the reactorvolume is pumped every day. The supernatant isdischarged to the effluent channel (or pumped back tothe reactor)," the thickened sludge is pumped to thedigesters or drying beds. That this method gives thecorrect sludge age can be seen from Eq. 4.2 and theprocedure works because the mixed liquor concentra-tion changes insignificantly over the day. This is verifiedby comparing the stability of the concentration of mixedliquor in the reactor and the concentration of siudge inthe underflow over the day in one of the major plants ofJohannesburg, Figure 4.7 (Nicholls, 1975).

Controlling the sludge age in small works may re-quire pumping at regular times, every day or every fewdays, provided the volume pumped is such that the re-quired volume, v, per day is abstracted. On large plantsthe satellite settling tank (or flotation unit) may bedesigned for continuous operation receiving a constantflow, or, operated over a part of the day.

An important point about hydraulic control of thesludge age is that irrespective of the flow through theplant if a fixed fraction of the volume of the reactor isremoved and wasted every day, the sludge age is fixed.If the COD mass load per day on the plant remains cons-tant, the sludge concentration will remain constantautomatically. If the COD mass load increases, thesludge concentration will increase automatically, tomaintain the same sludge age. Thus, by monitoring theMISS concentration and its changes for a fixed sludgeage an indirect measure is obtained of the COD load andload change.

By means of the hydraulic control procedure, thesludge age may be changed by simply changing thevolume wasted per day. If say, the sludge age is reduc-ed from 25 days to 20 days by hydraulic control, the fulleffect of the change will become apparent only afterabout 15 days. Thus the organisms have an opportunityto adapt gradually to the change in load.

Hydraulic control of sludge age is particularly rele-vant to plants with sludge ages longer than about 4 daysbecause for these plants the mass of sludge containedin the secondary settling tanks is a minor fraction of thetotal mass of sludge in the system. At sludge agesshorter than 4 days the mass of sludge in the secondarysettling tanks becomes appreciable with respect to the

total mass of sludge in the system. Hydraulic controlwill have to take cognizance of this and accuracy of thecontrol will require additional testing. It should be notedthat irrespective of the sludge age of the plant, thesatellite settling tank required for hydraulic control willalways be very small with respect to the secondary settl-ing tanks - at 2 days sludge age the retention time isusually about 4 to 6 hours and hence the satellite settl-ing tank has to accommodate 1/12 to 1/8 of the dailyaverage flow; at 25 days sludge age the retention time isabout ] to 1^ days and the satellite settling tank has toaccommodate about 1/20 of the daily average flow; incontrast the secondary settling tank has to accom-modate 2 to 4 times the daily average flow.

With regard to design, hydraulic control of thesludge age has the following implications: It devolves agreater onus on the designer and removes responsibilityfrom the plant operator: It becomes essential that thedesigner calculates the sludge mass more exactly, toprovide sufficient reactor volume under the design loadto allow for the required concentration of MLSS in thereactor at the specified sludge age. Also, the settlingtank design, recycle and aeration capacities must beadequate. If these aspects are catered for adequately,then with hydraulic control of the sludge age, plant con-trol is simplified and, on small scale plants, may even doaway with the requirements for solids and SVI tests ex-cept at long intervals. Hydraulic control of sludge agemakes the Sludge Loading Rate, SLR, (or SubstrateUtilization Rate, SUR), redundant and introduces an en-tirely different attitude to process control.

11. SELECTION OF SLUDGE AGE

Selection of the sludge age is the most fundamental andimportant decision in the design of an activated sludgeprocess. The sludge age of a plant depends on manyfactors inter alia, stability of the process, sludge set-tleability, whether or not the waste sludge should besuitable for direct discharge to drying beds, and mostimportant of all, the quality of effluent required i.e. isCOD removal only acceptable, must be effluent benitrified, is nitrogen and phosphorus removal required.Although the discussion below will be aimed at SouthAfrican conditions, where nitrification is obligatory andstringent standards for the effluent phosphate concen-tration have been promulgated, other situations will bediscussed briefly so as to place South African activatedsludge process technology in perspective with othertechnologies, for example those of Europe and NorthAmerica.

11.1 Short sludge ages (1-5 days)

11.1.1 Conventional plants:

These plants are operated in the conventional con-

1 For a review of the development of hydraulic control of sludge age see Ekama and Marais 11978).

•it should be noted that in a gravity thickener for waste sludge from a nutrient removal plant, considerable concentrations of phosphorus maybe released in the thickener, which, when discharged to the activated sludge process will be counter-productive for excess biological Premoval. The thickener supernatant may have to be treated chemically to precipitate the P before discharge to the activated sludge process(see Section 8 above).

4-13

figuration i.e. a semi plug flow configuration, butmodified systems such as contact stabilization, stepaeration, step feed and others are also implemented.

Short sludge age plants are extensively used inEurope and North America but are virtually unknown inSouth Africa principally because the promulgated stan-dards for effluents cannot be satisfied, i.e. that ofobligatory nitrification.

Short sludge age plants are for the purpose of CODremoval only, towards which objective sludge ages of 1to 3 days are sufficient. BOD or COD reductions, for theconventional system range from 75 to 90%; the actualremoval achieved depends on the influent, the state ofthe sludge and the efficiency of the settling tanks.Predatory activity of the micro-organisms in the sludgeis relatively low which causes turbidity and high effluentCOD.

Nitrification should be absent, or the plant can beoperated to prevent nitrification; this however may bedifficult to achieve with high water temperatures in theTropics.

Advantages with low sludge age plants are: smallprocess volumes for unit input COD (Figure 4.3) withcorresponding low carbonaceous oxygen demand (seeFigure 4.2); nitrification is unlikely, giving a saving in thenitrification oxygen demand (see Figure 6.1); oxygendemand under cyclic flow and load tends to be heavilydamped making control of the oxygen easier than withlonger sludge age plants. Disadvantages are: the massof sludge produced by unit COD input is high (Figure4.4) and the active fraction of the sludge is high (Figure4.2). The sludge treatment facilities accordingly con-stitute an appreciable fraction of the treatment costsalthough the running costs can be reduced by energyrecovery from anaerobic digestion of the sludge (seebelow); the process tends to be unstable and sensitiveto mechanical breakdown.

Because of the relatively low uxygen demand andthe high sludge production per unit input COD it ispossible to make these high rate processes near self suf-ficient with respect to energy requirements - sufficientmethane gas can be produced by anaerobic digestion ofthe waste sludge to generate electricity for oxygen sup-ply and process operation.

11. 1-2 Aerated lagoons:

In aerated lagoons, (as opposed to aerated oxidationponds which supplement the oxygenation by algae) theoxygen requirement in the process is supplied wholly byaerators, usually floating mechanical ones. There areessentially two types of aerated lagoons, suspensionmixed and facultative.

Suspension mixed aerated lagoons have sufficientenergy input per unit volume by the aeration equipmentto keep the sludge in suspension. In facultative lagoonsthis energy input is insufficient and settlement takesplace on the lagoon floor, to form a sludge layer whichdecomposes anaerobically, as in an oxidation pond.

Kinetically, suspension mixed lagoons are mem-bers of the activated sludge process, the only differencebeing in that the former is a flow through system i.e. thesludge age equals the nominal hydraulic retention time,whereas the latter has a sludge recycle from the settling

tank making the sludge age longer than the hydraulicretention time. Accordingly the volume of the aeratedlagoon per unit input COD is large whereas it is small forthe conventional short sludge plant. (For a sludge ageof, say, 1 day the aerated lagoon will have an hydraulicretention time of one day whereas the short sludge ageconventional plant (with a one day sludge age) will havea hydraulic retention time of about 1/8 of a day.

The effluent from a suspension mixed aeratedlagoon has the same constitution as the mixed liquor inthe basin. The COD removed from the system via theoxygen demand is relatively small so that the COD in theeffluent is quite unacceptable for discharge to receivingwaters. In fact the principal objective of all short ageplants is to act as biological flocculators, to physicallychange the influent COD, soluble and paniculate, to aform that allows effective liquid-solid separation. In theconventional short sludge age plant the waste sludge istransferred to the sludge treatment facility; in theaerated lagoon usually a second pond is provided i.e. anoxidation pond or a facultative aerated lagoon, to allowthe now readily settleable paniculate material to settleas a sludge layer and thereby produce a relatively solids-free effluent.

Aerated lagoons find application principal!/ in in-dustrial waste treatment and seasonal waters. Designconsiderations are not dealt with in this manual; theseare discussed by Marais and Ekama (1980) andreference should be made to that report.

11.2 Intermediate sludge ages (10-15 days)

Where nitrification is obligatory, the sludge ages re-quired are 5 to 8 times longer than those for CODremoval only, depending on the temperature. In thetemperate regions where water temperatures may fallbelow 12UC to 14°C the sludge age is unlikely to be lessthan 10 to 1G days. In this range of sludge age, the ef-fluent COD concentration no longer plays a role in thedesign because for sfudge ages longer than about 4days, the effluent COD or BOD concentrations remainapproximately constant (see Section 3.4 above). The ef-fluent ammonia concentration also plays a minor role indesign because nitrification kinetics are such that oncenitrification is achieved, it is virtually complete; eventhough the effluent standards may require an effluentammonia concentration of less than, say, 10 mgN/ i ,once nitrification takes place the concentration is likelyto be less than 4 mgN/i . Consequently for nitrification,the sludge age of the process is fixed completely by therequirement for nitrification. (The method forcalculating the minimum sludge age for nitrification isgiven in Chapter 5, Section 3.)

With low alkalinity wastewaters (like those en-countered in the Western and Southern Cape), nitrifica-tion can cause a significant reduction in effiuent pH,often as low as 5. This not only causes problems withthe nitrification process itself in that only partial nitrifica-tion is achieved, with the likelihood of non-complianceof the 10 mgN/i effluent ammonia standard (seeChapter 5, Section 4.3), but it also tends to favour thedevelopment of poor settling sludges and to produceaggressive effluents that can do considerable damageto concrete surfaces. To reduce these problems (and to

4-14

derive other advantages cited later in this section) thepolicy of deliberate biological denitrification has evolv-ed. However, once biological denitrification is incor-porated in the process, sludge ages longer than 10 to 15days are required and the process falls into the longsludge age category.

Comparing intermediate sludge age plants withhigh rate plants, the oxygen demand per kg COD(including nitrification) is doubled (see Figure 6.1), theprocess volume is 3 to 4 times larger (see Figure 4.3),the daily sludge mass wasted is reduced by 40% (seeFigure 4.4) and active fraction is much lower (see Figure4.2). Intermediate sludge age plants are much morestable than high rate plants, requiring less sophisticatedcontrol techniques or operator intervention therebymaking these plants more suitable for general applica-tion.

At intermediate sludge ages, the active fraction ofthe waste sludge is still too high for direct discharge todrying beds. Consequently some form of waste sludgestabilization would need to be incorporated in the treat-merit works, i.e. either aerobic or anaerobic digestion.The former has the advantage of ease of operation butthe disadvantage of energy costs for oxygen supply; thelatter has the advantage of energy production but thedisadvantage of complexity of operation. Even withenergy recovery by anaerobic digestion of wastesludge, because of the low mass of sludge wasted fromthe activated sludge plant and high oxygen demand perkgCOD load, energy self-sufficiency at intermediatesludge ages is impossible. However, on large plants (ap-proximately 500 000 person equivalent) where technicalsupervision and operator expertise are of a high level,energy cost can be reduced by gas production fromanaerobic digesters and probably can be justifiedeconomically, particularly if energy costs continue to in-crease as they have over the past decade.

In nitrifying aerobic activated sludge plants, there isalways the possibility that denitrification can take piacein the secondary settling tank. This problem is exacer-bated by the process control procedure of abstractingthe waste sludge from the settling tank underflow (seeSection 10 above). This procedure is used to benefitfrom the thickening function of the settling tank: thelower the underflow recycle ratio, the higher the under-flow concentration, the lower the volume of wastesludge. However, low underflow recycle ratios result inlarge sludge accumulations in the secondary settlingtank (see Figure 4.8). Hence the sludge might be retain-ed in the settling tank for a long period of time and thisis likely to lead to denitrification. The degree ofdenitrification is increased as:

(i) the residence time in the settling tank increases -this depends on the recycle ratio and peak flowconditions;

(ii) the active fraction of the sludge increases, i.e. isgreater at shorter sludge ages, Figure 4.2;

(iii) the temperature increases; and

(iv) the mass of biodegradable COD adsorbed on theorganism mass increases (van Haandel, Ekama and

Marais, 1981)-this is generally greatest at thepeak load condition.

If denitrification in the settling tank takes place to asufficient degree, attachment of sludge particles to thenitrogen gas bubbles causes buoying of the particles tothe surface. The flotation effect can cause considerableloss of sludge via the settling tank overflow. Thisbehaviour has been observed, for example, at Bellville,Cape Town, where sludge losses have been observedduring that latter part of summer when the watertemperatures are around 22CC. Sludge loss occurs dailyin the afternoon after the peak flow and load when boththe sludge accumulation in the settling tank and storedCOD adsorbed on the sludge mass are at a maximum.Although the underflow recycle ratio is quite high (s = 2)sufficient sludge still accumulates to make denitrifica-tion significant at high temperatures.

The above discussion on experience with nitrifyingactivated sludge plants points to the following conclu-sions:

The secondary settling tank no longer should servethe dual purpose of solid-liquid separation and thicken-ing. The sludge residence time needs to be minimized,to reduce denitrification, by having a high underflowrecycle ratio, of 1 to 2:1, but this will result in inade-quate thickening of the waste sludge withdrawn fromthe underflow. The thickening aspect can be solved byimplementing hydraulic control of the sludge age, byproviding a small independent satellite settling tank forgravity thickening of the waste sludge (the supernatantbeing returned to the reactor so that inadequateclarification due to denitrification flotation does not af-fect effluent quality) or by replacing the gravity satellitetank with a dissolved air flotation unit." These modifica-tions, however, ameliorate the effects on the secondarysettling tank but do not positively remove the rootcause, i.e. the high nitrate concentration in the mixed li-quor. This can be done only by including denitrificationin the process, which then shifts the process to that ofthe long sludge age type.

11.3 Long sludge ages (20 days or more}

11,3, 1 Aerobic plants:

Long sludge age aerobic plants are usually called ex-tended aeration plants. The sludge age is chosen to beof such a length that the active mass fraction of thewaste sludge is sufficiently reduced to allow dischargedirectly to sludge drying beds after thickening. Whatthis sludge age should be will depend to a degree on theclimatic conditions i.e. whether the sludge can be driedbefore it starts smelling, but probably exceeds 30 to 40days - research into the sludge ages necessary underSouth African conditions is necessary.

Compared to intermediate sludge age plants thetotal oxygen demand (i.e. carbonaceous plus nitrifica-tion) is about equivalent (Figure 6.1); the process

"Thickening by dissolved air flotation now is both technically andeconomically feasible, Bratby and Marais (1976, 1977).

4-15

volume requirement is 50 to 60% higher and the activefraction 50% lower. As nitrification always occurs, pro-blems with rising sludge in the settling tank are to be ex-pected. Control of the sludge age is best via thehydraulic sludge age control method, so that highrecycles from the secondary settling tank can be impos-ed to reduce rising sludge in the tank, as describedunder intermediate sludge age processes above. Pro-blems of low pH when treating low Alkalinitywastewater are to be expected.

Extended aeration plants are very stable in opera-tion and require probably less supervision than anyother activated sludge process. Although the volumerequirements and the oxygen demand per unit inputCOD is large, the relative ease of operation makes thisprocess the indicated one for small communities.However, it must be remembered that this process willdischarge a low COD effluent but with high nitrate andphosphate concentrations. The nitrate concentrationcan be drastically reduced and low pH effluentprevented by incorporating an anoxic zone in the pro-cess without basically disturbing The ease of operation.This modification always should be kept in mind whenconsidering extended aeration as a possible process.

Processes that can be operated in the extendedaeration modes are the completely mixed single reactor,the Orbal multi-channel, oxidation ditch and Carrouselprocesses.

11.3.2 Anoxic-aerobic process

Once sludge ages in the range greater than 20 days areaccepted it is possible usually to incorporate denitrifica-tion in the process without upsetting stability innitrificaion. The reactor is suitably subdivided intounaerated (anoxic) and aerated zones in a variety ofways, the denitrification taking place in the unaeratedzones and the nitrification in the aerated zones to givethe so-called nitrification-denitrification process. Theadvantages are substantial: (1) Nitrate concentration inthe effluent is reduced, the nitrate utilized in theunaerated zones lowering the oxygen demand of theprocess accordingly; with complete denitrification thetotal oxygen demand can be reduced by approximately15 to 20% compared with the requirements for a nitrify-ing process. (2) Loss of sludge in the overflow from thesettling tsnk due to rising sludge is largely eliminated.(3) Alkalinity recovery with denitrification causes pH in-stability with low influent alkalinity wastewaters to bevery unlikely. Normally the sludge ages in nitrification-denitrification plants are insufficient to allow directdischarge of the waste sludge to the drying beds.

Process configurations of anoxic-aerobic systemsare the Modified Ludzack-Ettinger process in which aprimary anoxic reactor is provided, the Bardenphowhich incorporates both primary and secondary anoxicreactors, the Orba! in which anoxic zones are created byreducing the aeration energy input in channels, Car-rousel and oxidation ditch in which the aeration at theaerators is controlled to induce anoxic operation in thesame reactor, or in two interlinking parallel reactors.

Incorporation of biological denitrification in a pro-cess imposes additional constraints on the design. It re-quires the selection of a process sludge age and an

unaerated mass fraction, (to achieve the requireddenitrification or the highest denitrification possible),yet ensuring efficient nitrification under all expectedconditions. (See Chapter 5, Section 4.4, and Chapter 6,Section 6).

11.3.3 Anaerobic-anoxia-aerobic systems

These plants are needed where biological removal ofboth nitrogen and phosphorus is required and are calledbiological excess P removal plants. The phosphorusremoval propensity is created by providing a zone in theprocess that receives the influent wastewater but notoxygen and nitrate (via the recycles of mixed liquor).The incorporation of this zone imposes further con-straints on the design of the system: A sludge age mustbe selected that will ensure efficient nitrification at alltimes, provide for anoxic zones that will give completeor maximal removal of nitrate and an anaerobic zonethat receives the influent flow and recycles of mixed li-quor not containing nitrate generated in the system.The critical aspects of the design are the assurance ofno discharge of nitrate to the anaerobic zone. The suc-cess of the design is sensitive to a number of factors:ihe influent wastewater characteristics, such as thereadily biodegradable COD concentration, theTKN/COD and P/COD ratios, the maximum specificgrowth rate of the nitrifiers, maximum and minimumtemperatures, severity of cyclic load and flow, controlof aeration and control of nitrate and dissolved oxygenin some of the recycles.

Aeration control is a particularly vexing problemunder cyclic load and flow conditions because the pro-cess is affected by too high or too low oxygen concen-trations in the aerobic zone. Too high concentrationscause oxygen to be recycled to the anoxic zone (and theanaerobic zone in the 5 stage Phoredox process),thereby reducing the potential for P and N removal; toolow oxygen concentrations cause the nitrification effi-ciency to decline and poor settling sludges to develop.

Oxygen control setups that have been developedstill require considerable operational attention, par-ticularly the oxygen probes, and at present their installa-tion in smaller plants is to be questioned. These dif-ficulties have prompted research into an alternativesolution - load and flow equalization. In this approachan equalization tank is provided upstream of the plant.The effluent flow from the tank is controlled in such amanner that the cyclic fluctuations in the flow and loadare damped to very small values. The control is via amicrocomputer that calculates the discharge flow sett-ing. The essential inputs to the computer are the depthof liquor in the equalization tank and the flow rate out ofthe tank; both these measurements can be readilymonitored automatically and operated realiabfy overlong periods of time. The microcomputer also hasreached a stage of development that ensures its longterm trouble-free service. The equalization approachwith the microcomputer control, has been tested atGoudkoppies plant, Johannesburg and shows greatpotential for reducing aeration and other control pro-blems in nutrient removal plants. Detailed informationon equalization tanks is available in a report by Dold,Buhr and Marais (1982). Design details of the P removal

4-16

processes, considered in This monograph, are given inChapter 7.

It should be noted that the waste sludge from abiological excess P removal activated sludge processcontains high concentrations of P. If the siudge is allow-ed to become anaerobic or is anaerobically digestedmuch of the phosphorus removed from the wastewaterwill be released from the sludge mass to the bulk liquid;consequently seepage from drying beds may have to bechemically treated to precipitate the phosphorus beforerecycling to the plant. Failure to recognize thisbehaviour of phosphorus-rich biological sludges maylead to inadvertent re-contamination of the effluent withhigh concentrations of phosphorus (see Section 8above).

Examples of N and P removal processes are themulti-reactor Phoredox and UCT processes in which theanaerobic, anoxic and aerobic conditions are induced inseparate reactors interlinked by recycle flows. Otherremoval systems have been proposed, for example thatby Kerdachi and Roberts 11982) in which these condi-tions are imposed within the same reactor.

Where P removal without obligatory nitrogenremoval is required the sludge age can be selected to pre-vent nitrification and in the multi-reactor type planttheoretically only anaerobic and aerobic reactors areneeded, i.e. no anoxic reactor. The sludge age is reducedsubstantially, probably to less than eight days, so that thereactor volume per input COD is much smaller thanwhere nitrification-denitrification is a requirement. How-ever, even with these short sludge ages it would be wise,always, to make provision for denitrification of anynitrate generated and recycled to the anaerobic zone.This provision is of particular relevance in warm climatesor where the maximum specific growth rate of the nitri-fiers is very high so that nitrification can occur at aerobicsiudge ages of, say, 2 days. These short sludge age Premoval plants are not considered in this monograph.

12. DESIGN EXAMPLE

In this section, the procedure for design of organicmaterial (COD) degradation in purely aerobic processes

is demonstrated with numerical examples. Assumingconstant flow and load conditions, calculations arepresented showing how estimates of the processvolume requirements, average daily carbonaceous oxy-gen demand and daily sludge production are obtainedfor the treatment of raw or settled wastewaters basedon the wastewater characteristics.

12.1 Influent Wastewater Characteristics

The daily average characteristics for a particular rawwastewater are given in Table 4.3. In the design exam-ple it is assumed that primary sedimentation tanksremove 40% of the total COD load, 15% of the TKNand total P and 10% of the readily biodegradable COD.The settled wastewater characteristics are also given inTable 4.3. it should be noted that the soluble concentra-tions in the settled wastewater should be approximatelyequal to those in the raw wastewater. For example,from Table 4.3 the soluble unbiodegradable COD con-centration in the raw wastewater is 0,05 mg COD/mgCOD, i.e, 0,05.600-30 mg COD/I (Eq. 2.3). As thesoluble unbiodegradable constituents are not affectedby primary sedimentation, the unbiodegradable soiubleCOD fraction of the settled wastewater accordingly willbe approximately 30/360-0,08 mg COD/mg COD.Similarly, the readily biodegradable COD fraction (fbs) ofthe raw wastewater is given as 0,24. This is equivalentto a concentration of 0,24 (1-fup-fus)S t ( = 0,24(0,82)600= 118 mg COD/i (see Eqs. 2.8 and 2.9). Primarysedimentation is assumed to remove 10% of the readilybiodegradable COD (due to biodegradation), the fbs

fraction in the settled wastewater is given by{(1,0-0,10)118}/{(1 -0,08-0,04)360} - 0,33 (see Eqs.2.8 and 2.9). The examples show that care must betaken that settled wastewater characteristics are not fix-ed at values that are inconsistent with the raw waste-water characteristics and primary settling tank be-haviour.

12.2 Temperature effects

From Table 4.1, the only constant in the organicmaterial degradation theory that is affected by

TABLE 4.3 RAW AND SETTLED WASTEWATER

Parameter

Influeni COD concentrationInfluent TKN concentrationInfluent P concentrationTKN COD ratioP/COD ratioUnbiodegradable soluble CODUnbiodegradable paniculate CODMLVSS- MLSS ratio of sludge producedTemperature - Summer

- WinterpH of wastewaterInfluent flow

Symbol

S,iNt.Pii—

—

' u p

t,Tmax

Tmm—

0

' Influent concentrations appropriately weighted to take account

CHARACTERISTICS (SEE ALSO TABLES

ValueRaw

600*48*10*0,0800,0170,050,130,75

22147,5

13,33

Settled

360*41*8,5"0.1140,0240,080,040,83

22147,5

13,33

of daily cyclic flow variations

5.2 AMD 6.1)

Units

mgCOD/imgN/fmgP/JmgN/mgCODrngP/mgCODmgCOD. mgCODmgCOD/mgCODmgVSS mgTSS°C°C—M//d

4-17

temperature is the specific endogenous respiration rate,bh. This rate reduces by about 3% every 1°C drop intemperature - from Eq. 4.5, (given in Table 4.1), therate at 14°C is 0,20/d and at 22°C is 0,25/d. The effectof the reduction in the rate with decrease in temperatureis that at reduced temperatures the daily sludge produc-tion is marginally increased and the average car-bonaceous oxygen demand is marginally decreased -the differences in sludge production and oxygen de-mand are less than 5% for a 6C C change in temperature.Consequently, the average carbonaceous oxygen de-mand should be calculated at the maximum tempera-ture and the process volume and sludge production atthe minimum temperature in order to find the maximumvalues of these parameters.

12.3 Calculations for organic material degradation

The example will demonstrate the effect of temperatureand sludge age on (i) the mass of sludge in the process,'ii) the average daily carbonaceous oxygen demand, (iii)the active fraction of the sludge and liv) mass of sludgewasted daily. These four parameters will be calculatedfor the raw and settled wastewater at 14°C and 22°C forsludge ages ranging from 5 to 30 days. The values of theconstant Y,, fCv, f and bh (appropriately adjusted fortemperature) are given in Table 4.1.

Mass of COD treated per day - M(Stl) = Q-S,i(4.6)

Mass of biodegradable COD = M{Sbl) =(4.7c)

Mass of unbiodegradable particulate solids: tfUp/fcv (4.8c)

Hence for raw wastewater:

M(SJ =• 13,33 106.600 mg COD/d = 8000 kg COD/d

f.iS = 0,05 mg COD/mg COD

fop = 0,13 mg COD/mg COD

M(Sn.) = (1 -0,05-0,13)8000 =6560 kg COD/d

= 8000-0,13/1,48 =702,7 kg VSS/d

for raw wastewater; and

, .233,

<4.29b,

for settled wastewater.

From Eq. 4.14 the mass of total suspended solids inthe process is given by

M(X;} = M(Xj/0,75for raw wastewater <4.30aJ

= M(Xv)/0,83for settled wastewater <4.30b)

From Eq. 4.15, the average daily carbonaceousoxygen demand is given by

M(Oc)=M(Sbt}{(1-fcvYh) + fc

= 6560 {(0,334)+ 0,533


- ^ (T+ tTRj )

I I (-,j Hs l

= 4224 {(0,334 + 0,533 - p ^ -

for settled wastewater.

}(kgO/d)

(4.31a)

(kgO/d)

C4.31b)

From Eq. 4.18, the active fraction with respect to the

volatile suspended solids (fav) is given by

f a v = 1 /{1+ 0,2 bhTR£ + 0,238 <1+bhTRJ}

= 1 /{1,238 + 0.438 bhT R5} for raw wastewater.

(4.32a)

U - 1/{1+0,2bhT Rs + 0,068(l+bhTRj)

= 1 /{1,068 - 0,268 bhT Rs) for settled wastewater(4.32b)

From Eq. 4.22, the mass of sludge (as total suspend-ed solids) produced per day is given by

0,75 ),2bn7RJ +


__ 8000,0,88-0,45

and for settled sewage:

M(S,,) = 60% of 8000 =4800 kg COD d

fus = 0,08 mg COD.'mg COD

fup = 0,04 mg COD, mg COD

M(Sbl) = (1 - 0,08 - 0.04)4800 = 4224 kg COD/d

M(XJ = 4800.0,04/1,48 = 129,7 kg VSS/d for settled wastewater.

0,131,48'

>}(4.33a)

0,04

(4.33b)

From Eq. 4.13 the mass of volatile solids in the pro- Substituting the b,T value for 14°C li e 0 202/d)cess is given by a n d ^C (i.e. 0,254/d) into Eqs. 4.25 to 4.29 the

parameters M(XV), M(XJ, M(Ot), fev and M(AXt) can beRS calculated for sludge ages 5 to 30 days. The results are

shown plotted in Figure 4.9.

4-18

SLUDGE AGE (d

0,0

t 1

14°

-

i i

!

Cc

•'T 1

-

SETTLEDWASTEWATER

RAWWASTEWfl^ ER

10 15 20 25 30SLUDGE AGE (d ) SLUDGE AGE (d

FIGURE 4.9. Mass of sludge (MLSS and MLVSS), active frac-tion (with respect to TSS). average carbonaceous oxygen de-mand and mass of sludge ias TSS) produced daily versussludge age for raw and settled wastewaters %x W C and 22°C.

Wastewater characteristics given in Table 4.3.

Figure 4.9 confirms that the mass of sludge in theprocess (MLSS or MLVSS), the average carbonaceousoxygen demand and the active fraction (with respect toMLSS or MLVSS) are only marginally affected bytemperature and for these parameters, insofar as designis concerned, are not really of consequence. However,the influent waste type i.e. raw or settled wastewater,has a significant effect; raw wastewater results in mofesludge in the process, a higher oxygen demand and alower active fraction of the sludge than settled wastewater. The difference in effects between raw andsettled wastewater depends wholly on the efficiency ofthe primary settling tanks-the differences apparent inFigure 4.9 arise from a 40% COD removal in primarysedimentation. The greater the efficiency of primarysedimentation the greater the difference between theparameters shown in Figure 4.9.

From Eq. 4.16 for the same MLSS concentration,the process volume is proportional to the mass ofsludge contained in it. Hence for the same MLSS con-centration, the volume of the process treating settledwastewater will be only 40% of that treating raw waste-water at 25 days sludge age. Also, the settled waste-water plant requires only 64% of the oxygen the rawwastewater plant requires. However, the active fractionof the sludge in the settled wastewater plant is 34%,probably too high for direct discharge to drying beds,whereas that from the raw wastewater plant is 22%,probably sufficiently low for discharge to drying bedswithout further stabilization. Clearly the choice oftreating settled sewage as against raw sewage requires

weighing the advantages and disadvantages of eachagainst the other, i.e. settled sewage (smaller processvolume, smaller oxygen demand and smaller secondarysludge production, but, with primary sludge productionand its disposal) against raw sewage (greater processvolume, higher oxygen demand and higher secondarysludge production, but no primary sludge production}.The choice is made more difficult by the fact that higherbiological removals of N and P are achieved with rawwastewater than with settled wastewaters (see Chapter3, Section 3). These aspects are discussed in detail inChapters 6 and 7 and will be demonstrated by extendingthe above design example to include biological nitrogenand phosphorus removal.

13. CONCLUSIONS

The calculations for the daily sludge production, dailyaverage carbonaceous oxygen demand, processvolume and nitrogen and phosphorus requirements forsludge production etc. were demonstrated for a rawand a settled sewage with selected wastewatercharacteristics for the raw sewage and assumed effectsof primary settlement on the raw sewage character-istics. Selection of the numerical magnitudes of thesecharacteristics was based on experience with fivemunicipal waste flows so that the data should not betaken to apply generally. Whenever possible, thecharacteristics should be determined experimentally, orestimated by some semi-experimental procedure sup-

4-19

plemented by experience. Where design information isinadequate it will be necessary to select ranges of valuesfor some parameters, ranges that appear to span acrossthe true values, and to repeat the design a number oftimes using various combinations of the limiting values.In this way a picture will be built up of the sensitivity ofthe plant process response so that appropriate provi-sions can be made in the design. There are so manyinteractive effects that ample design provision in oneaspect, for example, does not necessarily resolve thatproblem in terms of the whole design as it may createproblems in other aspects.

14. REFERENCES

ABKLEY, M. and MARAIS. G.v.R. (1982) Effect of The anoxic sludgemass traction on the activated sludge process. Research ReportNo W 38, Depl. of Civil Eng., Univ of Cape Town.

BRATBY, J. and MARAIS. G.v.R. (1976) A guide for the design ofdissoived-air (pressure) notation systems for activated sludge pro-cesses. Water SA. 2, 2, 86 100.

BRATBY, J. and MARAIS. G.v.R. 11978) Aspects of sludge thicken-ing by disso)ved-air flotation. Wat. Pullut. Control, 77, 3, 421-432.

DOLD, P L.. EKAMA, G.A. and MARAIS. G.v.R. (1980) A generalmodel (or the activated sludge process. Prog. Wat. Tech., 12,47-77.

DOLD, P L . , BUHR, H.O. and MARAIS, G.v.R. (1982) Design andcontrol of equalization tanks. Research Report No. W 42. Dept- ofCivil Eng., Univ of Cape Town.

EKAMA, G.A and MARAIS, G.v.R. (1978i The dynamic behaviour ofthe activated sludge process. Research Report No. W 27, Dept. ofCivil Eng., Univ. of Cape Town.

GARRETT, M.T. (1958) Hydraulic control of activated sludge growthrate. Sew. and Ind. Wastes, 30. 253.

KERDACHI, D. and ROBERTS, M. (1982) Full scale phosphateremoval experiences in the Umhlatuzana works at different sludgeages. Wat. Set. Tech., 15. 3-4, 262-282.

MARAIS, G.v.R. and EKAMA. G.A. <1976i The activated sludge pro-cess part 1 - Steady state behaviour. Water SA, 2, 4, 163-200.

MARAIS, G.v.R. and EKAMA, G.A. (1980) Aerated lagoons.Research Report, Dept. of Civil Eng., Univ. of Cape Town.

McKINNEY, R E. and SYMONS. J.M. 119641 Growth and en-dogenous metabolism - a discussion. Proc. 1st Int. Conf. on Wat.Poltut. Control, London. 1962, Pergamon Press, Oxford.

NICHOLLS, H.A. 119751 Internal Thesis Report, Dept. of Civil Eng.,Univ. of Cape Town.

PAYNE, W.J. (1970). Energy yields and growth of heterotrophs. An-nual reviews - microbiology, 17-51.

SEHAYEK, E. and MARAIS, G.v.R. (1981) Kinetics of biologicalnitrogen removal in the single sludge activated sludge process.Research Report No. W 41, Dept. of Civil Eng., Univ. of CapeTown.

SETTER, L.R., CARPENTER, W.T. and WINSLOW, G.C. 11945)Practical application of modified sewage aeration. Sew. WorksJour., 17, 4, 669.

VAN HAANDEL, A.C., EKAMA. G.A. and MARAIS, G.v.R. (1981)The activated sludge process - 3 - Single sludge denitrification.Water Research, 15, 1135-1152.

VAN HAANDEL, A.C. and MARAIS. G.v.R. (1981) Nitrification anddenitrification in the activated sludge process. Research ReportNo. W 39, Dept. of Civil Eng., Univ. of Cape Town.

4-20

CHAPTER FIVE

NITRIFICATION

by

G.A. Ekama and G.v.R. Marats

1. INTRODUCTION

The term nitrification describes the biological processwhereby free and saline ammonia is oxidized to nitriteand nitrate. Nitrification is mediated by specificautotrophic organisms with behavioural characteristicsthat differ significantly from the heterotrophic ones.The objectives in this chapter are to review briefly thekinetics of nitrification, to highlight the factors that in-fluence this biological reaction and set out the pro-cedure for designing a nitnfying-aerobic activatedsludge process.

It has been well established that nitrification is dueto two specific genera of autotrophic bacteria, theNitrosomonas and Nilrobacter. Nitrification takes placein two sequential oxidation steps: (1) Nitrosomonasconvert free and saline ammonia to nitrite, and (2) Nitro-bacter convert nitrite to nitrate. The nitrifiers utilize am-monia for their synthesis nitrogen requirements, and asa source of energy to bring 3bout synthesis. The am-monia requirement for synthesis, however, is a negligi-ble fraction of the total ammonia processed by theorganisms, at the most 2% (van Haandel and Marais,1981). Consequently it is usual to neglect the metabolicnitrogen requirements* of the nitrifiers and to considerthe nitrifiers simply to act as catalysts for the ammonia-nitrite-nitrate reactions - the reactions are taken asstoichiometric. This greatly simplifies the description ofthe kinetics of the process.

Consider the two basic redox reactions in nitrifica-tion,

O2 {Nitrosomonas)

NO; + 112 0 , (Nitrobacter)

H20(5.1a)

(5.1b)

Stoichiometrically the oxygen requirements for the firstand second reactions are 3,43 and 1,14 mgO. mgN [alsowritten as mgO/mglNHj-N)]. Hence the conversion ofammonia to nitrate, both expressed as N, requires 4,57mgO/mgN utilized. Taking into account the ammoniautilized for metabolic purposes of the nitrifiers, theoxygen requirements per mg!NH3-N) processed areslightly less, with reported values down to 4,3 mgO/mg(NHj-N). However, such a sophisticated approach ismore of academic than practical interest and we will ac-cept the stoichiometric value, 4,57 mgO, mg(NH3-N).

2. BIOLOGICAL KINETICS

2.1 Growth

In order to fo-rnulate the nitrification behaviour it isnecessary to understand the basic biological growthkinetics of nitrification. Here one has to consider thetwo stage nitrification behaviour. The rate of conversionof ammonia to nitrate, by the Nitrosomonas is muchslower than that of nitrite to nitrate, by the Nitrobacter.In fact, under most circumstances in municipal waste-water treatment plants, any nitrite that is formed is con-verted virtually immediately to nitrate. As a conse-quence verv little nitrite is observed in the effluent froma plant operating on an influent that does not containsubstances that inhibit the Nitrobacter. The limiting ratein the nitrification sequence, therefore, is that due tothe Nitrosomonas - one needs to consider the kineticsof this organism only. Because the nitrite produced isvirtually immediately converted to nitrate in terms of thegrowth rate of the Nitrosomonas, it can be assumedthat the conversion is from ammonia to nitratedirectly - on this basis the kinetics of nitrificationreduce to the kinetics of the Nitrosomonas.

Experimental investigations by Downing, Painterand Knowles (1964) showed that the nitrification ratecan be formulated in terms of Monod's relationship.Monod found that: (1) The mass of organismsgenerated is a fixed fraction of the mass of substrate (inthis case ammonia) utilized. (2) The specific rate ofgrowth, that is, the rate of growth per unit mass oforganisms per unit time, is related to the concentrationof substrate surrounding the organisms.

From (1):

M(AXn) = YnM(ANa> (5.2)

where

M(AXn) = mass of nitrifiers generated, (mgVSS)

M(ANa) = mass of ammonia as N utilized,[mglNHj-N)]

Yn = mass of nitrifiers generated per unitmass of ammonia utilized,[mgXn/mg{NH3-N)]

' In nitrification processes, where the influent energy source consists principally of ammonia, this simplification no longer can be applied and thedetermination of the nitrifying sludge mass is necessary to design the process volume (Neytzel-de Wilde, 1977). With municipal wastes undercyclic flow conditions the nitrifying mass also needs to be calculated.

5-1

Consequently one can write

(5.3)dt ' n dt

From (2) Monod developed the following relation-ship:

MnT =

where

MnmT

K(5.4)

nT

= specific growth rate observed at concentration Na, (mgXn/mgXn/d)

= maximum specific growth rate possible<mgXn/mgXr,/d)

K

N

= half saturation constant, i.e. the concentra-tion at which y r i=$yn,n, [mg(NHrN)/j£]

= concentration of ammonia surrounding theorganisms, [mg(NH3-N)/f]

T subscript refers to temperature, °C

The growth rate is given by the product of thespecific growth rate and the organism mass

KnT+Nv

- -V,(5.5)

Instead of expressing the growth rate in terms ofthe mass of organisms, it is often useful also to expressit in terms of the depletion rate of ammonia, as follows:From Eq. 5.3 substituting for dXn/dt in Eq. 5.5

" d T " KnT + Na x " (5-61

Equation 5.6 is sometimes written in an equivalentform as follows: Define the maximum specific substrateutilization rate, KmT = fjnrT,T/Yn then the specificsubstrate utilization rate KT becomes from Eq. 5.4

K T "

and the rate of substrate utilization, from Eq. 5.6

dN= KmT ISLdt KnT + N,

• X r

(5.7)

(5.8)

This form is mentioned here because it is the one oftenused to express heterotrophic growth, and in fact is theform utilized in all the work published by the Maraisgroup, because of its greater convenience for some pur-poses. The point to note is that Eqs. 5.6 and 5.8 are ofequal merit.

The ammonia nitrogen that disappears in thenitrification reaction, reappears as nitrate nitrogenbecause the stoichiometric approach is adopted, i.e.

dNndt dt

Consequently

dNn (H

(5.9)

(5.10)

Nn - nitrate concentration, (mg(NOj-N)//).

The oxygen consumption associated with nitrifica-tion is based on the stoichiometric assumption mentionedearlier, i.e. conversion of 1 mg(NH3-N) to 1 mg(IMOrN)requires 4,57 mgO. Consequently

(5.11)dt dt

2.2 Endogenous respiration

Up to this stage attention has been focussed on thegrowth aspect. However, Herbert (1958) showed thatgenerally an organism mass undergoes a continuousmass loss, called endogenous mass loss. Furthermorefrom the great amount of research undertaken by theMarais group it seems that this loss occurs in-dependently of the growth. This allows straightforwardformulation of endogenous mass loss as,

dXdt

n = - bnT (5.12)

where

bnT - specific endogenous mass loss rate forNitrosomonas (mg/mg/d)

2.3 Behavioural Characteristics

In Figure 5.1 the relationship between the specific growthrate, \iu, the specific substrate utilization rate, K, and thesubstrate concentration, NH is shown, as described byMonod's equation Eqs. 5.4 and 5.7 respectively. The rateconstants selected are finm20 = 0,33, Yn = 0,10 (henceKm = 3,3) and Kn20=1,0. The interesting feature of thisplot is that, because K,, is so small = 1 mg(NH3-(\!}/f, thespecific rate is virtually at a maximum for concentrationsof 2 mg(NH3-N)/jE and higher. However at concentra-tions less than two the rate rapidly declines to zero* so thatthe ammonia is not readily reduced to zero.

where

3. PROCESS KINETICS

As indicated in Chapter 4 the basic process unit is thecompletely mixed activated sludge unit. The processunder steady state provides the information necessaryfor design.

With regard to nitrification, for design, the principalsteady state solution required is that of the effluent am-monia concentration from a completely mixed aerobicreactor. This solution forms the basis for extensiveanalysis of the process behaviour. A comprerhensiveanalysis is not required here; for those interestedreference should be made to Ekama, van Haondel andMarais (1979) and van Haandel and Marais (1981).

3.1 Effluent ammonia concentration

Consider nitrification in a completely mixed reactor,under constant flow and load conditions (see Figure4.1). A mass balance on the accumulation of nitrifiermass M(AXn) is given by

5-2

UJ

\

o w

or

oUJ

8> o

I I 1TEMPERATURE = 20°C

Yn =0,10 mgVSS/mgN ; Kn = l ,0mgN/ lun r n = 0,33 / d ; Km = U n m / Y n

= 3,3 mgN/mgVSS/d

Km »unm

enorUJ

0,6

0,5

0,4

0,3

0,2

0,1 £

orr-

o ^LJ "Or~ \<

or *

0 10 20 30 40 50 60 70AMMONIA CONCENTRATION ( N a ) (mgN/ l )

FIGURE 5.1: Monod's relationship for nitrification at 20°C.

0.0

Ou_oUJQ.in

where

J = V pAX n=[^m T .N a / (K r i T+N a ) ] .X n .V pAt- b n T . X . . V p . A t - X n . q . A t

Vp = process volume (f)q - volumetric withdrawal rate of waste sludge

from the reactor Ii7d)

Dividing by V,, At

AXn/At - b w Na/(KnT + Na)]Xn - bnT Xn - Xn q/Vp

Now

Vp Xn _ Mass of nitrifier sludge in process _ Vp _ _q X,, Mass of nitrifier sludge wasted per day q 5

where

Rs - sludge age (d) (see Eq. 4.2).

Under constant flow and load, when steady state isachieved, AXn'At = 0 and solving for Na yields,

knT(bnT+1/Rs)N, = (5.13)

It is of interest to note that in Eq. 5.13 the ammoniaconcentration (Na) in the reactor (and effluent} is in-dependent of the specific yield constant (Yn) and the in-fluent ammonia concentration {Nfl[). Using the samenitrification kinetic constants as those in Figure 5.1 andtaking b.,T = 0,0, a plot of Eq. 5.13 with Na versus sludgeage Rs is given in Figure 5.2. Mote that at long sludgeages Na is very low and remains so until the sludge ageis lowered to about 3,5 days beiow which, Na increasesrapidly until, in terms of this equation, Na can exceedthe influent concentration, N 3 i - evidently the limit ofapplication of the equation is when Na = N,,. Substitut-

ing Nai for Na in Eq. 5.13 and solving for Rs gives theminimum sludge age, Rsrn, below which theoretically,no nitrification can be achieved. This minimum sludgeage varies slightly with the magnitude of Nai, see Figure5.2 - higher Nai gives a slightly lower Rsrn. Now the ef-fective influent Nai (see later) will rarely be less thanabout 20 m g / i . Noting that KnT = 1 mgN/ l , then

2 60 -

§50

u 40ozoo N

zo

30

2 0

10 -

0 * -

-

1 1

VA

RIO

US

1

<RsmS

t

4

INF

LUE

NT

,

to

-J

"oz

1I

1

I

RS>R

Dato

Yn

1

sm

'ram

= 0

Unm =

Kr

at

I

i =1.

2 0

M

1 1

Downing et al (1964)

05 mgVSS/mgN

0,33 /d

0 mgN/ i

°C

-

Rsm ^ Rsm

1 1 •4 6 8SLUDGE AGE (days)

10 12

FIGURE 5.2: Relationship between sludge age for nitrificationand influent ammonia concentration, Nai-

5-3

Kr,j/Nai is negligibly small with respect to unity.Substituting zero for KnT/Na, in Eq 5.13 and solving forRsm, yields

For practical application, Eq. 5.14 adequately definesthe minimum sludge age for all Na, greater than about5 m g N / i .

With regard to bn?0, this value empirically is takenas constant for all waste flows, at bri20 = 0,u4;d. Its ef-fect is small so that there is no need to enquire closelyinto all the factors affecting it.

With regard to KnT little information on effects ofinhibitory agents is available; very likely KnT will increasewith inhibition.

The virtually constant value for Hsrn (for the ac- 42 Temperaturecepted values of ^nmT and bnT} and the rapid attainmentof high nitrification efficiency at sludge ages slightlygreater than Rsm causes that in a particular plant, as thesludge age is increased, once Rs exceeds R^, high effi-ciency of nitrification will be observed. Consequently,under steady state conditions one would expec* a planteither not to nitrify, or, to nitrify with high efficiencydepending on whether Rs is shorter or longer than Rsm

respectively. This behaviour is well attested by observa-tion on laboratory scale plants.

From the discussion above clearly the magnitudesof (jr,mT and bnT affect Rsrr; it is of importance, therefore,for practical application, to enquire into those factorsthat influence these constants and the nitrificationphenomena in general.

4. FACTORS INFLUENCING NITRIFICATION

A number of factors affect the nitrification rate con-stants, efficiency of nitrification and the maximumsludge age. These are (1) wastewater source, (2)temperature, (3) pH, (4) unaerated zones, (5) dissolvedoxygen concentration, and (6) cyclic flow and toad.

4.1 Influent source

The maximum specific growth rate constant ^nrn j hasbeen observed tn be specific tn the pournn of thp wasteflow and, even then, to vary between different batchesfrom the same source. This specificity is so marked that(jnrriT should be classified as a wastewater characteristic.The effect appears to be of an inhibitory nature due tosome substancefs) in the wastewater. It is not toxicitybecause high efficiency of nitrification can be achievedeven with a low jjnrr. value if the sludge age is increasedsufficiently. These inhibitory substances are more likelyto be present in effluents having some industrial wastecomponents and in general, the higher this fraction, thelower f in m T tends to be, but the specific waste fractionsthat cause the reduction of j jn m T have not been clearlydelineated. Taking the j jn m T value at 20°C as thereference value, _prT-.;0 values have been reported rang-ing from 0,33 to 0,65 per day. These two limits will havea significant effect on the minimum sludge age: Twoplants, having these respective finrr.2O values, will haveRsni values differing by hundred per cent. Clearly due tothe link between the waste flow and M™-T. the latter'svalue should always be estimated experimentally for op-timal design, in the absence of such a measurement, alow value for ^nrTlT necessarily will need to be selectedwhich, for a particular waste flow, should the actual nrT1

be higher, will result in a non-optimal design. An ex-perimental procedure to determine Mnm2o 's 9'ven in Ap-pendix 3.

The jj,irTi and Kr, constants are sensitive to temperaturewith a high temperature coefficient (see Eq. 5.15); semi-empirically br,T is taken also to have temperaturedependency at the same rate as that for heterotrophs(see Eq. 5.16).

(5.15a)

(5.15b)

(5.16)

KnT = Kn 2 0 (1,123) ( T 2 0 '

°r,T = br.2C(1,029)(T-20>

The effect of temperature on ^nrT,T is particularlydramatic, for every 6°C drop in temperature the nnrr.Tvalue will halve which means that the minimum sludgeage for nitrification will double. Design of plants fornitrification, therefore, should be based on theminimum expected process temperature. Note that thetemperature sensitivity of KnT does not affect theminimum sludge age, but it does affect the efficiency ofnitrification - the higher the Kn the higher also the ef-fluent ammonia at R£ > Rsrri (See Section 4.3).

4.3 pH and Alkalinity

The specific growth rate of the nitrifiers ptn is extremelysensitive to the pH of the culture medium. It seems thatthe activities* of both the hydrogen and hydrox/i ions,(H*) and (OH"), act inhibitorily when their respectiveconcentrations increase unduly. This happens wh>en thepH increases above 8,5 [increasing (OH")] or decreasesbelow 7 [increasing (H*)]; optimal nitrification rates areexpected for 7 < pH < 8,5 with sharp declines outsidethis range.

From Eq. 5.4, n is a function of both / j n m and Kn.From an analysis of Eq. 5.13 it was shown earlier thatthe minimum siudge age is dominated by the magnitudeof jjni--r; it is only very weakly influenced by Kr.T. At Rs

;» Rsm however, the effluent ammonia concentration(Na), although low, is relatively speaking, significantlyhigher for larger KnT values: For example if KnT in-creases by a factor of two, the effluent ammonia con-centration will increase correspondingly by the samefactor (see Eq 5.13). Consequently the value of K,,T issignificant in so far as it controls the effluent ammoniaconcentration at Rs ;§> RE .

With regard to experimental estimates of ^nfT1T a nd

KnT, a great amount of research has gone into tracingthe effect pH has on ,unrTlT. These investigations general-ly have not separated out the effects of ^f;,,lT and KT|T sothat most data are in effect lumped parameter estimatesof ^nmT. Almost no information is available on the effectof pH on KrT. The effect of pH on KriT can be surmisedfrom data obtained in a laboratory scale completely mix-

* Equivalent to concentration.

5-4

0 ,0 a 20 ,25SLUDGE AGE (days)

FIGURE 5.3: Relationship between TKN and nitrate concentra-tions and pH with sludge age in an activated plant treatingweakly buffered influent sewage from the Strand-Somerset

West area, Cape.

ed reactor aerobic activated sludge investigation on awaste flow from Somerset West in the Cape Province.The plant was operated at 20°C over a range of sludgeages from 2 to 30 days. The response of the plant interms of pH, filtered effluent TKN and nitrate plus nitriteis shown in Figure 5.3. Concomitantly with nitrificationthe pH declined, stabilizing at about 4 when the sludgeage exceeded 15 days. The TKN stabilized at10 mgN/f. Up to 10 days sludge age, the pH and nitrateconcentration fluctuated in a see-saw fashion. The factthat stable nitrification was achieved only at 10 dayssludge age when the pH was low, indicated that ^nmT

had decreased lo a concomitantly low value. The factthat the effluent TKN (= ammonia) stabilized at the highvalue of 10 mg/ i indicated that KnT had increased. Oncontrolling the pH at 7,0 nitrification commenced atabout 3 days sludge age and the effluent TKN (= am-monia) stabilized at about 4 mg/ / for Rs > 3 days.Evidently the pH and ^nniT reacted interactively so thaithe minimum sludge age would fluctuate with the pHand achieve stability only at the long sludge age of 10 to15 days. Evidently also the KrT value is increased at lowpH values. Due to the lack of information it is difficult tomodel quantitatively the effect of pH on ^nrr:J and KnT.The following analysis, based on various data sourcesand empirical assumptions on the behaviour, demon-strates the general expected trends even though thepredicted values cannot be taken as quantitatively cor-rect. Accepting that nnrr. remains constant for 7,2 < pH< 8,5 and decreases as the pH decreases below 7,2(Downing et al. 1964, Loveless and Painter, 1968) thisbehaviour can be modelled as follows:

(5.17a)

(5.17b)

For 7,2 < pH < 8,5,

PnmpH = Mnm7,2

For 5 < p H <7,2,

MnmpH = MnmT^ns'P*7'21

where

<t>ns ~ pH sensitivity coefficient= 2,35.

In Figure 5.4 the predictions, using Eq. 5.17 arecompared with experimental data on fjnrn collected byMalan and Gouws (1966). This comparison is illustrativeonly as it is not clear from the data of Malan and Gouwswhether the effect of pH on Kn was taken into accountwhen calculating j jn m p H .

Changes in the value of Kn with pH are not at alldocumented in the literature. The best than can be doneis to accept a dependency on pH and check the predic-tions against ammonia effluent concentrations reportedat the different pH values. On this basis the followingequations are suggested,

For 7,2 < pH < 8,5

For 5 < p H < 7,2

K n p h - Kn7p2<Dns '7.2-P"»

where

4>ns = pH sensitivity coefficient= 2,35.

(5.18a)

(5.18b)

5-5

UorUJ

0 0

8 0

6 0

4 0

2 0

r\

I i 1 I 1

/THEORETICAL . /MODEL / .

/ /

- ^^ yi—~T~\ 1 1

1 1

/

MfiLAN 5

1 1

1 1

-

-

GOUWS(1966)

-

6MIXED

7LIQUGR

8PH

FIGURE 5.4. Relationship between max imum specif ic g row thrate for Nitrosomon&s, ^^rr, and mixed liquor pH. Experimental

data after Malan and Gouws !19G6).

The overall effect of pH or 5 < pH < 7,2 andtemperature on n can be modelled by substituting Eqs.5.15, 5.17 and 5.18 into Eq. 5.4 i.e.

MnTpH - Mil 3 t temperature T°C and pH equal to pH.

(5.19)(Kn20*2,35) ( 7-2 2Ql N.

where

MnTpH = specific growth rate at temperature T andpH-pH

fjnm = maximum specific growth rate at tempera-ture 20cC and pH = 7,2

Kn20 = saturation coefficient at temperature 20°Cand pH = 7,2.

Although Eq. 5.19 is qualitative in prediction only, itserves an important function in illustrating the adverseeffects that temperature and pH can have on n: A plotof jir.-jv.ri is shown in Figure 5.5 for selected values of pHand temperature accepting ^nrri20|7 2) = 0,33/d andKn20;7 2. = 1.00 mgN/1. Clearly low pH and lowtemperature can have significant effects on design.

In design the temperature effects have to be ac-cepted as restrictions imposed by the environment.With regard to the effects of low pH, these may, or maynot, occur depending on the alkalinity of the influent.Low influent Alkalinity is likely to give rise to low pH inthe purely aerobic activated sludge process. However, itis possible to limit, or completely obviate, pH reductionby operating the plant as an anoxic-aerobic process.The reasons for this are as follows:

From the overall stoichiometric equations for nitrifi-cation (Eqs. 5.1a and 5.1b), nitrification releaseshydrogen ions which in turn decreases Alkalinity* of themixed liquor. For every 1 mg (NH/-N) that is nitrified7,14 mg Alkalinity (as CaCOj) is destroyed. Based onequilibrium chemistry of the carbonate systemfLoewenthal and Marais 1977), equations linking the pHwith Alkalinity for any partial pressure of carbon dioxidehave been developed by van Haandel and Marais(1981). These relationships are shown plotted in Figure5.6. When the Alkalinity falls below about 40 nig/jf asCaC03 then, irrespective of the partial pressure of car-bon dioxide, the pH becomes unstable and tends todecrease to low values. Generally, if nitrification causesthe Alkalinity to drop below about 40 mg/f (as CaC03)problems associated with low pH very likely will arise inthe plant, such as poor nitrification efficiency, tendencyfor developing bulking sludges, corrosive effluent, etc.

For any particular wastewater the possible effect ofpH can be readily assessed, as follows: Take the exam-

LJ

orxh-

8or

o

oLJJ

CL

0,6

0,5

0,4

0,3

0,2

0,1

0,0

r iU n r n = 0,33 and Kn =

vL -—-—,

i

1,0 at

pH *

~pH=~

1

20° C

7,0'

7,0_ -~

5,5—

5,5 "1

1and pt-

— — •

T =

T-

25— —

2 0 e

= 25i ' •

= 20I

i-7, 2

^C

C• • •

°c—

°c

0 5 10 15 20 25

AMMONIA CONCENTRATION CmgN/!)

FIGURE 5.5: Diagrammatic representation of Eq 5.19 - rela-tionship describing the specific growth rate of Nitrosomanas

for changes in temperature and mixed liquor pH.

'Alkalinity signifies the Total Alkalinity obtained by titrating the sample to an equivalent carbonic acid solution at pH approximately 4,3.

5-6

10

oUJX

0-100

jCARBON DIOXIDECONCENTRATION(mg/1 oi CaC03)

SATURATION ~ 0,3 wg/lCoCO-

0 100ALKALINITY

200 300 400OE CaC05)

FIGURE 5.6; Mixed liquor pH versus alkalinity for differentconcentrations of carbon dioxide.

pie of an influent having an Alkalinity of 200 mg/f asCaCO3; expected production of nitrate is 24 mgN/I .The expected Alkalinity in the effluent wilt be(200-7,14.24) = 29 mg// as CaCO,. From Figure 5.6,such an effluent is likely to have a pH < 7,0.

Wastewaters having low Alkalinity are often en-countered where the municipal supply is drawn fromareas underlain with sandstone.

The abovementioned problems are to be expectedin aerobic nitrifying plants treating municipal effluentsof towns located in the coastal areas of South Africa,from Cape Town to, and including, Natal. The effluentfrom Somerset West mentioned earlier (see Figure 5.3)is a typical example. The only practical approach totreating such effluents is to create an anoxic zone in theprocess to denitrify some or all of the nitrate generated.In contrast to nitrification, denitrification takes uphydrogen ions which is equivalent to generatingAlkalinity. By considering nitrate as electron acceptor, itcan be shown that for every mg (N03-N} denitrified,there is an increase of 3,57 mg Alkalinity as CaCO3.Hence incorporating denitrification in a nitrification pro-cess causes the net loss of Alkalinity to be reducedusually sufficiently to maintain the Alkalinity above40 mg/ / and consequently the pH above 7."

In the example above, where the Alkalinity in theprocess is expected to decline to 29 mg / i as CaC03, if25% of the nitrate could be denitrified the gain inAlkalinity would be (0,25.29.3,57)-26 mg/ i as CaC03

and will result in an Alkalinity of (29+ 26) = 55 mg/ i asCaCO3 in the process. In this event the pH should re-main above 7. For low Alkalinity wastewaters it is im-

perative, therefore, that denitrification be built into nitri-fying plants.

Incorporation of unaerated zones in the process in-fluences the sludge age of the process at which nitrifica-tion takes place so that cognizance must be taken of theeffect of an anoxic or unaerated zone in establishing thesludge age of a nitrifying-denitrjfying plant.

4.4 Unaerated zones

The effect of unaerated zones on nitrification can bereadily formulated if the following assumptions aremade:

(1) Nitrifiers, being obligate aerobes, can grow only inthe aerobic zones of a process.

(2) Endogenous mass loss of the nitrifiers occurs underboth aerobic and unaerated conditions.

(3) The concentration of nitrifiers in the unaerated andaerated zones is essentially equal.

With these assumptions, Ekama, van Haandel andMarais (1979) showed that if a fraction fxl of the totalsludge mass is unaerated, i.e. (1-fM) is aerated, theeffluent ammonia is given by

KnT<DnT+1/R£)(5.20)

Equation 5.20 is identical in structure to Eq. 5.13, ifone views the effect of the unaerated mass (fxl) as

'Quantitative estimation of the denitrification that can be achieved in an anaxic-aerobic process is considered in Chapter 6.

5-7

reducing the value of (infT,T to nrTiT (1 - f ^ KFollowing the same reasoning as that preceding

Eq. 5.14 one can show that the minimum sludge age fornitrification RsrT1 in a process having an unaerated massfraction, fx,, is

R (5.21)

Alternatively, if R£ is specified the minimum aerobicsludge mass fraction (1 - fxm) that must be present fornitrification is found by substituting Rs for Rsm and fxm

for f^ in Eq. 5.21 and solving for (1 -1 x m ) , i.e.

(1 - f x J = (bnT+1/Rs)/>wr (5.22)or equivalently the maximum allowable unaeratedsludge mass fraction is given by solving for fxm in Eq.5.22,

f>fri = 1-ibnT + 1/Rs)/MnmT (5.23)"

For a fixed sludge age, Rs, the design value for theminimum aerobic sludge mass fraction (1 - fxm) alwaysshould be significantly higher than that given by Eq.5.22, because nitrification becomes unstable when theaerated sludge mass fraction decreases to near theminimum value as given by Eq. 5.22. Consequently toensure efficient nitrification 090%) the minimumaerobic sludge mass fraction (1 - f>rn) must be increasedby a factor of safety, Sf to give the minimum designaerobic sludge mass fraction; from Eq. 5.22,

s /M n m T (5.24c.:and the corresponding maximum design unaeratedsludge mass fraction, from Eq. 5.23 is

fKm = 1-S,(bn T + 1/Rs)//wr (5.24b)

With the aid of the temperature dependency equa-tions for nitrification (Eqs. 5.15 and 5.16), a diagram-matic representation of Eq. 5.24 is given in Figure 5.7for S f= 1,25, Hnm2o = 0,65 and 0,33/d at 14CC and 20cC.The two ^nm2o values are the approximale extremes ofthe range of jinm2o values reported in the literature. Thediagram shows that fxm is very sensitive to the maxi-mum specific growth rate of the nitrifiers (fjnmT): unlessa sufficiently large aerobic sludge mass fraction is pro-vided (i.e. 1 -fxfT1, see right hand of Figure 5.7), nitrifica-tion will not take place and consequently nitrogenremoval by denitrification is not possible.

If one fixes the minimum design aerobic sludgemass fraction, the sludge age can be determined by solv-ing for (1 - fx m) in Eq. 5.24a. Knowing Rs the value of theeffluent ammonia concentration (Na) can be determinedfrom Eq. 5.20. By doing a number of solutions of thiskind, at 14°C, it will become apparent that if S, isselected at 1,25 or greater the effluent quality will havestability and range from 2 to 4 mg (NH3-N)/X dependingon the other constants in the formulations. Acceptingthe sludge age as determined at 14°C, then at 20°C the

LOWER"LIMIT

FACTOR OF SAFETY- 1,25

-0,65n m 2 0 -0,33

I I t i

10 15 20 25SLUDGE AGE (d)

30

FIGURE 5.7: Maximum allowable unaerated sludge mass frac-tion and minimum aerated sludge mass fraction required tosustain nitrification versus sludge age for different maximumspecific growth rates of the nitrifiers at 20°C and 14CC and

0,0

0,2

0.4

0,6

0,8

55 40'J.o

UJ z8 2

0 —

8£UJ< Q:

i|a:

1 <

•The usual approach for determining f,m is based on the concept of "aerobic sludge age" i.e. equating the minimum sludge age for nitrificationRsm to the aerobic sludge age of the process R«. This leads to f*m = 1 - 1.'{R&(nnmT-bnT)}. Th's equation overestimates the maximumunaerated sludge mass fXfl! at a particular sludge age because the approach ignores the endogenous mass loss of The nitrifiers in the unaeratedzones. If the endogenous mass loss can be ignored in nitrification kinetics i.e, bn = 0,0, Then the two approaches are equivalent.

5-8

value of Si will be found to have increased significantlyand the effluent ammonia to have decreased to 2 mg/ ior less. Consequently, for design the lower expectedtemperature should be selected to determine the sludgeage and the aerobic mass fraction, If this is done, usingS- = 1,25, one can accept an effluent ammonia value of2-4 mgN/i at the lowest temperature and 1-3 mgN/I at20cC, that is, Eq. 5.20 is bypassed by the approximationabove, provided R£ at least equals 1,25 R^. This ap-proximation is sufficient for design and will be the ap-proach followed in the design example.

4.4. 1 Maximum allowable unaerated mass fraction

The equations above allowed the maximum designunaerated sludge mass fraction to be determined forany selected sludge age (Eq. 5.24b). If one acceptsS, = 1,25 then fxrn or (1 - fxm) can be plotted versus R£

for any selected ^nmT. This is illustrated in Figure 5.7where the design (fxrTi) and (1 - f*m) values are plottedversus Rs for 20°C and T4°C for two selected j in m 2 0

reference values of 0,65 and 0,33 (bn2a w a s fixed at

0,04/d). Evidently from Figure 5.7, for iinrr2o= 0,65 theunaerated mass fraction at 14°C can be approximately0,8 at sludge age of 30 days. Such a high unaeratedmass fraction is the indicated one theoretically atRs = 10 days or longer at 20°C. One may now ask: Is thismass fraction acceptable for design or are other con-straints present that may limit the fraction to lowervalues7 To answer this, two aspects appear to be perti-nent:

(1) Experience with laboratory scale nitrificationdenitrification plants tends to support the opinionthat at high unaerated mass fractions, the processis prone to bulking particularly at low temperatures.Processes at unaerated mass fractions of 0,5 or lessshow much less tendency in this regard.

(2) An upper limit to the unaerated mass fraction Is evi-dent from experimental work on the Ludzack-Ettinger process, and theoretical predictions usingthe general activated sludge model. Experimentallyat 20°C with Rs = 20 days, if f,^ > 0,70, the mass ofsludge generated is found to increase sharply. Thisobservation is indicated also by theoretical simula-tion. Theoretically, this happens for fxm > 0,60 atT = 14°C and Rs = 20 days. The reason is that forsuch high fim, the exposure of the sludge toaerobic conditions becomes insufficient to utilizethe adsorbed and enmeshed paniculate material.This leads to a decrease in active mass and oxygendemand and a build-up of enmeshed material,which causes a progressive decline in process ac-tivity because of the decline in active mass. Whenthis happens, the process still functions in that theCOD is removed from the wastewater, but thedegradation of the COD is reduced; the processbegins to assume the role of a contact reactor of acontact-stabilization process, i.e. a bio-flocculatorwith minimal degradation. This critical state occursat lower f im as the temperature is decreased andthe sludge age is reduced (Arkley and Marais,1982).

From the discussion above it would appear that theunaerated mass fraction should not be allowed to in-crease above a certain upper limit of about 50%, butcertainty not above 60%, as indicated in Figure 5.7.

In design of N and P removal plants the unaeratedsludge mass fraction fvr,, usually needs to be high i.e.>40%. If the pnrn2C value is low i.e. <0,40.'d (which willbe the usual case in design where insufficient informa-tion on the nrr.20 's available) the necessary high f>f,,magnitudes will be obtained only at long sludge ages(see Figure 5.7). For example, if Mnr-2c= 0,35/d, thenwith S<=1,3 at Tmin=14°C, an fsrri=0,45 (Eq. 5.24b}gives a sludge age of 25 days and for fxrT. = 0,55 a sludgeage of 37 days. Long sludge ages require large processvolumes - increasing Rs from 25 to 37 days increasesthe process volume by 40% (see Figure 4 3) whereas f^.increased only 22%. Also, for ihe same P content in thesludge mass, the P removal is reduced as the sludge ageincreases because the mass of sludge wasted dailydecreases as the sludge age increases (see Figure 4.5).Consequently, for low nm2G values, the increase in Nand P removal that can be obtained by increasing theunaerated sludge mass fraction above 0,50 to 0,60might not be economical due to the large processvolumes this will require, and might even be counterproductive insofar as it affects P removal. A sludge ageof 35 days probably is near the limit of eonomic prac-ticality which, for low nml4 = 0,16 values will limit theunaerated mass fraction to about 0,5. At higher (jnmi4values, the sludge ages allowing 50 per cent unaeratedmass fractions decrease significantly again indicatingthe advantages of determining experimentally the valueof /inrr,2o l 0 check whether a higher value is acceptable.

4.5 Dissolved oxygen concentration

High dissolved oxygen concentrations, up to 33 m g / i ,do not appear to affect nitrification rates significantly.Low oxygen concentrations however do reduce thenitrification rate. Stenstrom and Poduska (1980) havesuggested formulating this effect as follows:

OMno Mnmo is , r\

where

(5.25)

O = oxygen concentration in bulk liquid (mgO/i)Ko - half saturation constant (mgO/f)

Hnmo ~ maximum specific growth rate (/d)MnD = specific growth rate at O mg/ / oxygen(/d).

The value attributed to Kc, has ranged from 0,3 to2 mgO/jf, i.e. at DO values below Ko the growth ratewill decline to less than half the rate where oxygen ispresent in adequate concentrations.

The wide range of Ko probably has arisen becausethe concentration of DO in the bulk liquid is notnecessarily the same as inside the biological floe wherethe oxygen consumption takes place. Consequently thevalue will depend on the size of floe, mixing intensityand oxygen diffusion rate into the floe. Furthermore in afull-scale reactor the DO will vary over the reactorvolume due to the discrete points of oxygen input (withmechanical aeration) and the impossibility of achieving

5-9

instantaneous and complete mixing. For these reasonsit is not really possible to establish a generally applicableminimum oxygen value - each reactor will have a valuespecific to the conditions prevailing in it. In nitrifyingreactors with bubble aeration a popular DO lower limit,to ensure unimpeded nitrification, is 2 mgO/i at thesurface of the liquid.

Under cyclic flow and load conditions the dif-ficulties of ensuring an oxygen supply matching theoxygen demand and a lower limit for the DO have beendiscussed previously (see Chapter 4, Section 11.3),Equalization of the lead probably presents the best prac-tical means to facilitate control of the oxygen level in thereactor. In the absence of equalization, mitigation of theadverse effects of low oxygen concentration duringpeak oxygen demand periods is by having siudge agessignificantly longer than the minimum necessary fornitrification.

4 6 Cyclic f low and load

It is well attested experimentally that under cyclic flowand load conditions the nitrification efficiency of an ac-tivated sludge process decreases compared to thatunder steady conditions. This behaviour is also ap-parent in simulation studies using the general activatedsludge model. From the simulation studies, during thehigh flow and/or load period, even though the nitrifiersare operating at their maximum rate, it is not possible tooxidize all the ammonia available, and an increased am-monia concentration is discharged in the effluent. Thisin turn reduces the mass of nitrifiers formed in the pro-cess - equivalently the effect of cyclicity in flow andload is to reduce the aerobic sludge age. The adverse ef-fect of the cyclicity becomes more marked as the frac-tional amplitude of the flow and load cycles increase,and, is ameliorated as the aerobic sludge age increases.Simulation studies of the cyclic effect have indicatedthat an approximate relationship can be established be-tween the effluent ammonia concentration under steadystate and that under cyclic conditions (see van Haandeland Marais 1981). However if the factor of safety onnitrification, S,, is 1,25 or greater the cyclic effect isrelatively small and for design usually can be neglected.

At the risk of being tedious, it should again bepointed out that if the value of ynn> is selected higherthan the actual value, even with a sludge age of 1,25 to1,35 times the theoretical minimum, the plant is likely togive rise to a fluctuating nitrate effluent with reducedmean efficiency in nitrification. Hence a conservativeestimate of f jnm is essential for a safe design.

Under cyclic flow and load conditions, it isnecessary to know the value of both the specific yieldcoefficient of the Nitrosomonas, Y r, the maximumspecific growth rate constant, pnmT, and the half satura-tion constant, Kn; in contrast, only fjnrT,T and Kn arenecessary in steady state calculations. The values ofthese constants have been determined by a variety ofmethods but the constants have a conjoined effect thatmay cause that in a particular method of evaluation dif-ferent sets of values may be obtained. For practical pur-

TABLE 5 1 KINETIC CONSTANTS FOR NITROSOMONASACCEPTED IN THIS MONOGRAPH

Constant Symbol

Specific yieldcoefficient Y^

Endogenousrespiration rate fa-

Half saturationcoefficient K-*

Valueat 20cC

O.'O

0,04

1,00

" Note :ha* K' increases with increase in *

TemperatureCoefficient

1,000

1 029

V-23

Eq No.

5.16

5.15b

poses it is not possible to obtain a complete set ofvalues so that one approach is to accept the value ofone constant and in any particular experimental evalua-tion determine the other constants from the experimen-tal data using this value. The estimates for Yn and Kr ac-cepted in this monograph are listed in Table 5.1; ex-perimental fjnm determination, therefore, is relative tothe value of Yn and K,. listed.

5. DESIGN CONSIDERATIONS

5.1 Fate of the influent TKN

The various fractions of the influent TKN have beendescribed in Chapter 2. Of these the following are of im-portance to our discussion: (1) free and saline ammoniaconcentration (Nai); (2) biodegradable organic nitrogen(NO1); and (3) the soluble unbiodegradable organicnitrogen (Nwi).

The biodegradable organic nitrogen is decomposedin the process by heterotrophic action to free and) salineammonia. Furthermore, on death of the organisms, byendogenous mass loss, the nitrogen released is in theorganic bound form and the biodegradable fraction alsois broken down to free and saline ammonia. These twosources of free and saline ammonia together with that inthe influent are utilized by the organism mass first forthe formation of new organisms and the balance forgeneration of nitrate, if the conditions are favourable.

The 'exact' formulations describing the interactionsbetween the free and saline ammonia and the organicbound nitrogen are complex and can be developed ef-fectively (at present) oniy in terms of the growth- death-regeneration model of Dold, Ekama and Marais (1981).*However under steady state conditions the kineticssimplify and the formulations reduce to describing neteffects. These can be developed readily in terms of thesynthesis - endogenous respiration approach alreadyused to develop the steady state equations for car-bonaceous material degradation, in Chapter 4. Fordesign there is even a further simplification possible: Innitrification systems if, for example, the constraints areput on the system that the minimum design sludge agemust be at least 1,25 to 1,35 times the theoretical

"A more elaborate discussion is available by van Haandel and Marais 11981).

5-10

minimum to initiate nitrification, and, that the pH re-mains above 7, some of the steady state kinetic equa-tions predict effluent concentrations thai fall within anarrow band of values, sufficiently narrow that one canaccept the effluent parameter virtually as constant."Below are set down some of these steady state equa-tions of importance in developing design procedures.Simple approximations adequate for design are in-dicated and some general response graphs are derived.

In the design of those aspects of a process effect-ing nitrification, basically the following information isneeded: (1) effluent TKN concentration; and (2) effluentnitrate concentration.

5.2 Effluent TKN

The effluent TKN will consist of free and saline am-monia (Na), biodegradable soluble organic N (NCl),unbiodegradable soluble organic N (NJ and TKN in theeffluent volatile solids. Equations for the fractions areset out below:

(1) Free and saline ammonia (NJ:

Na is given by Eq, 5.20, i.e.

M _ KnT(b,T+1/Rs)(5.20)

Note that No is independent of the influent am-monia concentration. Note also that the equationapplies only if the sludge age Rs > RMri, howeverthis condition always should be satisfied in design.

(2) Biodegradable organic nitrogen concentration (NJ

The biodegradable organic nitrogen is brokendown to free and saline ammonia by heterotrophicorganisms. Although a complex reaction, involvinga feedback of this nitrogen form from death of theorganisms, the prediction simulated by the generalmodel indicates that the following equation, interms of the influent biodegradable organicnitrogen concentration, is adequate

K, XaR15.26)

hn

where

NOI = influent biodegradable organic nitrogenmg(TKN-N)//. Its value is given by Eq. 2.14.

Kr - kinetic constant for degradation of No. Itsvalue and temperature dependency is givenin Table 5.3.

Xa = active mass concentration in aerobic reac-tor.

Rhri = nominal hydraulic retention time in process.

(3) Soluble unbiodegradable organic nitrogen (NJ

Nu = NU( (5.27)where

NUi - soluble unbiodegradable organic nitrogen,mg(TKN-N)/i. Its value is given by Eq. 2.12.

All these effluent fractions are soluble; consequentlythe total soluble TKN in the effluent (Nt) is given by thesum of these, i.e.

NT = N a +N 0 +N u (filtered N,) (5.28a)

The experimental value of N, will be given by the TKN ofthe filtered effluent sample. If the sample is not filteredthe theoretical TKN of the effluent will be greater by theconcentration of TKN in the volatile solids, i.e.

(unfiltered N:) (5.28b)

where

X^ — effluent volatile solid concentration(mgVSS/jf)

fn - fraction of TKN in VSS = 0,1 mg(TKN-N)/mgVSS.

5.3 Nitrification capacity

The concentration of nitrate generated in the system(NJ is given by the TKN in the influent (NJ less thesoluble TKN in effluent (NJ less the concentration oftotal influent TKN incorporated in the net sludgegenerated each day (Ns) i.e.

Nc - N t l - N t - N s * * (5.29)

The value of Ns is determined from the mass of nitrogenincorporated in the volatile solids per day. It is given byfri times the mass of votatile solids wasted per day(MIXJ/RJ, divided by the mean daily flow, i.e.

~ - (see Chapter 4, Section 9).

where

M(XV) is given by Eq. 4.13.

Note that M(XV) does not include the volatile mass ofnitrifiers as the masses of these, we have statedpreviously, are negligible.

In Eq. 5.29, Nc defines the nitrification capacity ofthe process. A formal definition of nitrification capacity(Nc) is the mass of nitrate produced per unit volume offlow, i.e. mg(NCyN}/f of flow.

Eq. 5.29 shows that the nitrification capacity (Nc) isthe difference between the influent TKN concentration(NJ and the sum of the effluent TKN concentration (Nt)and the nitrogen required for sludge production IN,.).The effluent TKN concentration (NJ depends on the ef-ficiency of nitrification. In calculations, with maximumunaerated siudge mass fraction ( f x r j , (for a selectedsludge age), if the factor of safety (S,) is selectedgreater than 1,25 to 1,35 at the lowest expectedtemperature (Tmin), the efficiency of nitrification alwayswill be high (>90%) and N, generally will be less than 3to 4 mgN/f. Also, with 5, > 1,25at Tmjn, N, will be vir-

* The same approach also applied in the design equations for carbonaceous material degradation that, provided Rb > 3 days usually one can ac-cept the filtered biodegradable COD in the influent as zero [see Chapter A, Section 3.4).

"This equation presumes thai no nitrate is present in the influent and this is true, virtually always, for municipal flows in South Africa.

5-11

tually independent of both the process configurationand the subdivision of the sludge mass into aerated andunaerated fractions. Consequently, for design N. usual-ly need not be calculated explicitly from Eqs. 5.20, 5.26to 5.28, but can be taken to be approximately 3 to4 mgN/t provided that there is reasonable assurancethat the actual ^^20 value will not be less than the valueaccepted for design and that there is sufficient aerationcapacity so that nitrification is not inhibited by an insuf-ficient oxygen supply. Accepting the calculated fKm andselected sludge age (Rs) at the lower temperature, thenat higher temperatures the nitrification efficiency andthe factor of safety (Sf) both will increase so that atsummer temperatures (Tmax), N, will be lower, approx-imately 2 mgN/f.

Dividing Eq. 5.29 by the total influent COD concen-tration (SJ yields the nitrification capacity per mgCODapplied to the biological process, Nc/St l,

IMe/Stl - N t l/S t l-Ns/S t i-N,/Sxi 15.30)

where

Nf/S., = nitrification capacity per mgCOD appliedto the process (mgN/mgCOD)

N,,/Stl = TKN/COD ratio of the wastewater

Ns/S,h = nitrogen required for sludge productionper mgCOD applied (see Eq. 4.23).

In Eq. 5.30 the nitrification capacity, COD ratio (NC/SXJ)of a process can be estimated approximately by evaluat-ing each of the terms in the RHS of the equation asfollows:

N,/S,,: Provided the constraint for efficientnitrification is satisfied at the lowesttemperature (Tmir.) the TKN in the effluent(N,) can be assumed at Tmin to be about4 mgN/i i.e. N t/S t l will range from 0,005to 0,01 for SI( ranging from 800 to 400 re-spectively. At Tmax, Nt = 2 mgN/i .

Ns/S,,: This ratio gives the TKN concentration inthe influent abstracted for sludge produc-tion. Its value will depend on whether settl-ed or raw wastewater is being treated, andthe sludge age. The ratio, Ns/S,,, for thispurpose can be read from Figure 4.5, orcalculated from Eq. 4.23.

Nn/S.,: This value is obtained by measurement ofthe TKN and COD of the influent and canrange from 0,07 for unsettled municipalwaste flows to 0,T2 for settled flows.

A graphical exposition of the relative importance ofthese three ratios on NL/Stl is obtained by plottingNc/S,i versus sludge age for selected influentTKN/COD (N,,/St1J ratios ranging from 0,07 to 0,09 forraw wastewater and 0,09 to 0,11 for settled wastewater.The Nc/S,, values were found from Figure 4.5 (or Eq.4.23) and the N,/Stl values selected constant at 0,007(this is for Sti of approximately 600mgCOD/f). Theplots for 14°C and 20°C are shown in Figures 5.8a and5.8b respectively. Also shown plotted on the diagramsare the minimum sludge ages for unaerated mass frac-tions of 0,0, 0,2, 0,4 and 0,5, at a selected finm20 value of

0,15ooo

5 O.IO

<

0,05

TEMP • I4"CUr>m20 • 0 , 4 ft

b n 2 o r ° t 0 4 / d

S f - -,23

WASTEWATER f fu,CHARACTERISTICS

RAW 0,13 0,09SETTLED 0,04 0,08

SETTLEDWASTEWATER

RAW I _JWASTEWATER

0,000

O

10 15 20SLUDGE AGE id '

FIGURE 5.8a

25

0,15

0,i0

0,05o<

0,00

TEMP • 20 *CUnm20 " ° - 4 /6

b f t20 * 0,04 /dS f « 1,23

L - S-fi *J 1°-• — DVI ° i °

WASTEWATERCHARACTERISTICS

RAWSETTLED

SETTLEDWASTEWATER

'up "o i

0,13 0,050,04 0,08

0 10SLUDGE

15 20AGE (d

FIGURE5.8b

FIGURE 5.8: Nitrification capacity per mg COD applied versussludge age tot different influent TKN/COD ratios at 14DC(Figure 5.8a) and ZOC'C (Figure 5.8b)- Also shown are theminimum sludge ages required to obtain complete nitrificationat different unaerated sludge mass fractions for jj,,m?o =

5-12

0,4. For any particular unaerated mass fraction the plot-ted values of Nc/S,, are valid only at sludge ages greaterlhan the corresponding minimum sludge age. Theseplots allow one to assess the effects of the three ratiosand temperature versus sludge age.

Considering the effects of temperature, to obtaincomplete nitrification at 14°C (for a selected fBrn), thesludge age required is more than double that at 20c'C.The corresponding nitrification capacities per influentCOD at 14CC show a marginal reduction to those at20°C (because the sludge production at 14°C is slightlyhigher than at 20°C).

Considering the effect of sludge age for a selectedNtl/Sl, value, the Nc/S,, increases as the sludge age in-creases because the nitrogen needed for sludge produc-tion decreases with sludge age and thus becomesavailable for nitrification. However, the increase ismarginal for Rs > 10 days.

Considering the influent TKN/COD ratio (N^/S.,)evidently for both settled and unsettled wastewater, atany selected sludge age, N, / S,, is very sensitive to theNtl/Stl value; an increase of 0,01 in Nj,/ S,, causes an in-crease of about 0,01 in Nc/S:, (0,01 mgN/mgCOD). Forthe same Ntl/S., ratio for raw or settled wastewater(compare TKN/COD ratio of 0,09 in Figures 5.8a and b),NL/S,, for unsettled is less than for settled wastewaterbecause more volatile solids are produced per unit CODload from unsettled than from settled wastewater;however, leaving aside this point of difference, the in-crease in nitrate produced per influent COD will, whenthe influent TKN/COD ratio increases, increase the dif-ficulties, or, make it impossible, to obtain completedenitrification. This will become clear when denitrifica-tion is considered in Chapter 6. As primary settling in-creases the influent TKN/COD ratio difficulties in ob-taining adequate denitrification are always greater withsettled than unsettled wastewater; however, because

primary settling removes COD, then for the same flow,the mass of volatile solids and the oxygen demand inthe biological process wiil be lower for settled than forraw wastewater thereby giving rise to savings in plantvolume and oxygenation costs.

6. DESIGN EXAMPLE

Design of a nitrification process without denitrificationwill be considered. When denitrification is required con-straints additional to those imposed by nitrification mustbe applied and these cannot be incorporated in thedesign until the theory of denitrification is dealt with.

For the purpose of constructive comparison thenitrifying activated sludge plant will be designed for thesame waste flows and characteristics as that acceptedfor the design of a process for carbonaceous materialremoval (see Chapter 4, Section 12). The wastewatercharacteristics, for raw and settled wastewaters arelisted in Table 4.3 and the additional characteristicsneeded for nitrification are listed in Table 5.2.

6.1 Effect of nitrification on mixed liquor pH

An important initial consideration is the possible effectof mixed liquor pH on the nurr#Q value. In Section 4.3above it was stated that nitrification consumes alkalinity- 7,14 mg/jf as CaC03 for every 1 mgN/f ammoniaconverted to nitrate - and if there is insufficientalkalinity in the influent, the mixed liquor pH is depress-ed below 7,0 causing a reduction in the jjnf11?0 value (seeEq5.17>.

The influent TKN/COD ratio of the raw wastewateris 0,08 mgN/mgCOD (see Tables 4.3 and 5.?). With arelatively low /jnrn2o value of 0,36/d, the sludge ageneeds to be 10 days or longer to ensure nitrification

TABLE 5 2 RAW AND SETTLED WASTEWATER CHARACTERISTICS FOR

Parameter

Influent TKN concentration

Influent TKN/COD ratio

Ammonia fraction of TKN

Unbiodegradable soluble organic N fraction

Nitrogen fraction of volatile solids in influent

pH of wastewater

Total alkalinity

Maximum specific growth rate of nitrrfiers

Influent flow

Symbol

-

*nu

fr,

-

aik

Q

Raw

48

NITRIFICATION (SEE ALSO TABLES 4.3 AND 6.1)

ValueSettled

41

0,080 0,114

0,75

0,03

0,10

7,5

200

0.36

13,3

'See Chapter 3, Section 8. Note thai settled wastewater characteristics mustsedimentation tank behaviour and the raw wastewater characteristics.

0,83*

0,04*

0,10

7,5

200

0,36

13,3

be selected so that they are

Units

mgN//

mgN/mgCOD

mgN/mgN

mgN/mgN

mgN/mgVSS

-

mg// as CaCOj

/ d .

Mf/d

consistent with primary

5-13

(St = 1,3) at a minimum temperature of 14CC in a purelyaerobic process (fK^ = 0,0) (see Eq. 5-24). At this sludgeage the nitrification capacity is about 0,05 mgN/mgCODfor a TKN, COD ratio of 0,08 mgN/mgCOD. (see Figure5.8a). Hence the nitrate concentration produced (per /influent) should be about = 0,05.600 = 30 mgN/f. Thiswill cause an alkalinity reduction of 7,14.30= 214 mg/ ias CaCO3. Because the influent alkalinity is only200 mg/ / as CaCOj, the mixed liquor alkalinity will dropbelow 40 mg/i as CaCO3 causing the mixed liquor pHvalue to drop below 7 (see Figure 5.6). The low mixed li-quor pH value will cause unstable and incompletenitrification (see Section 4.3 above) and produce an ag-gressive water (which over a few years can cause con-siderable damage to the concrete surfaces of the treat-ment works) and can be the cause of bulking sludgesand high solids content in the settling tank effluent dueto denhrification action in the secondary settling tank.

The simple approximate calculation above,therefore, can make the designer aware, at an earlystage of possible adverse consequences to be expectedfrom a proposed design. In the instance above thedesigner should give consideration to a nitrification-denitrification process, to recover some of the alkalinityloss due to nitrification, thereby to maintain the pH,reduce the possibility of bulking and eliminate (usually)rising sludge problems in the settling tank. (See Chapter6, Section 1.4). In general a high TKN/COD ratio coupl-ed with low alkalinity in the influent are reliable in-dicators warning of problems to be expected in aerobicnitrifying processes.

6.2 Minimum sludge age for nitrification

For the purposes of demonstrating nitrification underpurely aerobic conditions, it will be accepted that the in-fluent alkalinity is sufficiently high to maintain an ef-fluent Alkalinity above 40 mg/ i as CaCO3. Hence noadjustment to nrT,2o a nd Kn20 for pH needs to be madeas the pH is assumed to remain above 7,2. The adjust-ment of the kinetic constants of nitrification fortemperature is given in Table 5.3.

For a completely aerobic process i.e. f I f T 1 -0, withMnrr o = 0,36. d and St = 1,3, the minimum sludge age fornitrification (RsrT1) is found from Eq. 5.24 i.e.

- 3,2 days at 22°C (2,5 days with St - 0,0)

= 8,9 days at 14CC (7,5 days with Sf = 0,0)

Clearly, to ensure nitrification throughout the year forthe relatively low ^nrTl20 value of 0,36/d, the sludge ageof a purely aerobic process should be about 10 to 12days.

6.3 Raw wastewater ^

The influent TKN concentration of the raw wastewateris 48 mgN/f (see Table 5.2 for wastewater character-istics). Accepting an ammonia fraction of the influentTKN (fnfi) of 0,75* and an unbiodegradable solubleorganic nitrogen fraction (fnu) of 0,03* for the rawsewage (from Table 5.2), gives the influent ammoniaconcentration (Nai) (from Eq. 2.11) as

Na, - 0,75-48 = 36 mgN/f

and the unbiodegradable soluble organic nitrogen con-centration (NUI), (from Eq. 2.12) as

NUI = 0,03.48 - 1,50 mgN/f.

Accepting the nitrogen content of the volatile solids inthe influent (fn) as 0,10 mgN/mgVSS (see Table 5.2),then (from Eq. 2.13) the nitrogen concentrationassociated with the unbiodegradable participate COD(Npr) is

NPI - 0,10 (0,13.600)/1,48 - 5,3 mgN/f.

Hence (from Eq. 2.14), the biodegradable organicnitrogen concentration in the influent which can be con-verted to ammonia (NOi) is

NOI = 48 (1 -0,75-0,03)--5,3 - 5,3 mgN/1.

TABLE 53 ADJUSTMENT OF

Constant

Max. specific growth rate

Saturation coefficient

Endogenous respiration rate

O'ganic N conversion rate

KINETIC CONSTANTS OF

Symbol

Mt>T,20

Kr.2C

bn20

Kf

20° C

0,36

1,0

0,04

0.015

NITRIFICATION

6

1,123

1.123

1,029

1.029

FOR TEMPERATURE

14L"C

0,180

0,50

0,034

0.013

22- C

0,454

1,26

0.042

0,016

Eg. No.

5.15a

5.15b

5.16

5 31

*The actual choice of these values makes relatively little difference to the final results of the nitrification calculations: in the case of the former(fr-.a). most of the organic nitrogen is converted to ammonia in the process and in the case of the latter (fnu), this value merely increases the ef-fluent TKN concentration by a fixed value.

5-14

6.4 Settled wastewater

Following the above procedure for settled wastewater,i.e. fri8 = 0,83, fnu = 0,04 (see Table 5.2) yields thefollowing:

N,t - 41,0

Nai = 0,83.41,0

N,., - 0.04.41.6

= 41,0 mgN/l

= 34,0 mgN/l

= 1,50 mgN/i

Npt = 0,10 (0,04.3601/1,48 = 1,0 mgN/l

Noi = 41,0-34,0- 1,5- 1,0 - 4,5 mgN/i

6.5 Nitrification process behaviour

For any sludge age longer than about 2 days The effluentbiodegradable organic nitrogen concentration INoe) isgiven (from Eq. 5.26)

Noe- = NOl/(1 + Kr7XaRhn| (mgN/i)

where from Eq. 4.9a, Rhr,Xa - MfXJ Q. Hence

Noe = No,/(1 + Kr lM(XJ.Q) (mgN/f) (5.31)

From Eq 5.27, the unbiodegradable organic nitrogen inthe effluent is

N u e= Nui= 1,50 mgN/i for raw and settled waste-

water (5.32)

The nitrogen required for sludge production (Ns) isgiven by Eq 4.23 i.e.

V" (5.33)

The influent TKN concentration available for nitrifica-tion (NJ,1 is the difference between the total TKN con-centration in the influent (N.u) and that required forsludge production (N,) (from Eq. 5.19)

K = (5.34)

If the sludge age of the process is longer than theminimum required for nitrification (RE > RsrT1 forS« = 1,00, Eq. 5.14), the TKN concentration available fornitrification will be partially or almost wholly nitrified tonitrate and the effluent nitrate concentration (Nnb) is thedifference between the nitrogen available for nitrifica-tion (N,'1( Eq. 5.34) and the effluent TKN concentration(NIe). If Rs < Rsm, no nitrification takes place; the ef-fluent nitrate concentration (Nne) is zero and the ef-fluent ammonia concentration (Nae) is equal to thenitrogen available for nitrification (NT(, Eq. 5.34) minusthe sum of the unbiodegradable and biodegradable TKNconcentrations in the effluent (Nue+NO(J. For bothRs < RSm and RE > RSrrw the effluent TKN concentration(N,e) is the sum of effluent unbiodegradable andbiodegradable organic nitrogen concentrations (Nue andN,,e respectively) and the effluent ammonia concentra-tion (NaJ.

For Rs > RsnJ

From Eq 5.13, the effluent ammonia concentrationis given by

NM = KnT(bnT+- 1/Rs)/{|inmT-(bnT + 1/RS)} <5.35a)

Hence the effluent TKN concentration (Nte) is

Nte = Nae+Nioe+Nue (mgN/7) (5.36a)

and the effluent nitrate concentration (N,lfe) is

Nrie = N;,-N l e (mgN/i I

= N t l -N s - Nle (from Eq. 5.34) !5.37a)

Analogous to the concentration of activeheterotrophic organisms (see Eq. 4.10), the nitrifierorganism mass is given by

MfXJ - M(Nne> YnRs/(1+bnTRJ fmgVSS)(5.38a)

where

M(Nnfc) = mass of nitrate generated per day

where NMH is given by Eq. (5.37a).

The oxygen demand for nitrification is simply4,57 mgO/mgN times the mass of nitrate producedper day, i.e.

M(0n) = 4,57 M(Nne) (mgO/d) 15.39a)

For Rs < Rsm:

For Rs < Rsm, no nitrification takes place. Hencethe effluent nitrate concentration is zero i.e.

Nne - 0,00 (mgN/f) (5.37b)

The effluent ammonia concentration is

^aft = K i ~ Noe~Nue (mgN/f)

= N t i - N 5 - N o e -N u e (from Eq. 5.34)

The effluent TKN concentration is(5.35b)

(5.36b)

The nitrifier sludge mass and the nitrificationoxygen demand are zero, i.e.

M(XJ - 0,0 (mgVSS)

M(On) = 0,0 (mgO/d)

(5.38b)

(5.39b)

Substituting the respective influent nitrogen con-centrations for raw and settled wastewaters and thevalues of the kinetic constants at 14°C into Eqs, 5.31 to5.39 and plotting the results at different sludge agesyields Figure 5.9. In Figure 5.9a, the different concen-trations of nitrogen in the process versus sludge age forraw and settled wastewater at 14°C are shown. InFigure 5.9b the nitrifier sludge mass and nitrificationoxygen demand for raw and settled wastewater at 14CCare shown. Also shown in Figure 5.9b are the car-bonaceous and total oxygen demands for raw and settl-ed wastewater at 14°C. The calculations were repeatedfor 22°C and are shown in Figure 5.10a and b.

"The subscript e used here and later refers to the effluent. However,assumed.

= Nae, etc. because completely mixed conditions are

5-15

TEMPERATURE = I4"C . pH=7,5

TEMPERATURE = I4°C_, p H = 7,5_

' 4 0 -

*x*30

o

2 0 -

10-

i i i i

RAW WASTEWATERSETTLED WASTEWATER

— - I N F L U E N T TKh N , f

NU 1» l , 5 m g N / l

N | . " N o t • N o , + « l # < -

5 10 15 20 25

SLUDGE AGE (d)

FIGURE 5.9a

30

BOO

010 15 20 25 30

S L U D G E AGE ( d )

FIGURE 5.9b

FIGURE 5.9: Effluent ammonia (NHel, organic nitrogen (NOP),and nitrate (Nnp) concentrations and nitrogen required forsludge production (N5) (Figure 5.9a), carbonaceous (Of),nitrification (0n), and total (0,) oxygen demand and the nitrifiersludge mass <XJ (Figure 5.9b) versus sludge age for raw and

settled wastewater at 14°C

TEMPERATURE * 22°C ; pH=7,5

10 15 20 25 30

SLUDGE AGE £d }

FIGURE 5.10a

8CG0

7000

TEMPERATURE" 22°C -, pH-7 ,5800

5 tO 15 ?0 25 38

SLUDGE AGE (d)

FIGURE 5.10b

FIGURE 5.10; Effluent ammonia (Nac), organic nitrogen [Noel,and nitrate (Nnel concentrations, nitrogen required for sludgeproduction (Ns) (figure 5.10ai carbonaceous (0c), nitrification(0n), and total (O,) oxygen demand and the nitrifier sludgemass (XrJ (Figure 5.10b} versus sludge age for raw and settled

wastewater at 22°C.

5-16

Figures 5.9 and 5.10 show that once the sludge ageis approximately 25 per cent longer than the minimumrequired for nitrification, nitrification is complete andcomparing the results for raw and settled wastewater,there is little difference between the nitrification oxygendemand and the concentrations of ammonia, nitrateand TKN in the effluent. The reasons for this similarbehaviour are: (11 the primary settling tank removesonly a small fraction of the influent nitrogen*; and (2)settled wastewater results in lower sludge production,so that the nitrogen available for nitrification in raw andsettled wastewater are approximately equal. Tempera-ture has little effect on the different concentrations ofnitrogen. However, a change in temperature causes asignificant change in the minimum sludge age fornitrification.

Considering Figures 5.9a and 5.10a, the effluentorganic nitrogen concentration (Nofc) is less than2 mgN/f for all sludge ages greater than about 3 days.For sludge ages about 25 per cent longer than theminimum for nitrification, the effluent ammonia con-centration (Nae) is less than 1 mgN/f so that forRs > 1,25 Rsm the effluent TKN concentration can betaken to be approximately 4 mgN/f. The increase innitrate concentration (Nne) with increase in sludge age isprincipally attributable to the reduction in the nitrogenrequired for sludge production (INI,,). This is importantfor nutrient removal plants as it indicates that increasingthe sludge age of a process, increases the nitrificationcapacity (see section 5.2 above) even though the ef-fluent TKN concentration remains constant at about4 mgN/f (see Figure 5.8). For Rs < Rsm, the effluentammonia concentration (Nae) and hence the effluentTKN concentration tNIe}, increases with increasingsludge age up to Rsrn because N5 reduces for increasesin rV

Figures 5.9b and 5.10b show that the nitrificationoxygen demand increases rapidly once Rs > R5f71 but ifRs > 1,25 Rsm, the increase is marginal irrespective ofthe temperature or wastewater type, i.e. betweensludge ages 10 and 30 days about 2000 kgO/d are re-quired for nitrification. This nitrification oxygen demandrepresents an increase of 40% and 60% respectively*"above the carbonaceous oxygen demand for the rawand settled wastewater. However, the total oxygen de-mand for treating settled wastewater is only 75% of thatfor treating raw wastewater.

In order that nitrification can proceed without in-hibition by oxygen limitation, it is important that theaeration equipment is adequately designed to supplythe total oxygen demand; generally heterotrophicorganism growth takes precedence over nitrifier growthwhen oxygen supply becomes insufficient. This isbecause heterotrophic organisms can grow adequatelywith dissolved oxygen concentrations of 0,5 to1,0 mgO/i whereas nitnfiers require a minimum con-centration of 1 to 2 mgO/jf.

The nitrifiers sludge mass given in Figures 5.9b and5.10b shows a rapid increase once Rs > Rsm and is ap-

proximately the same for raw or settled sewage, but isslightly higher at 14°C than at 22°C. Comparing thenitrifier sludge mass to the heterotrophic sludge mass,even at high TKN/COD ratios, the nitrifier sludge masscomprises only about 2 per cent of total volatile massand may be neglected in the determination of thevolatile solids concentration in the activated sludge pro-cess treating domestic wastewaters.

It is worth repeating that primary sedimentationremoves only a minor fraction of the TKN but a signifi-cant fraction of COD. Even though the settled waste-water has a lower TKN than the raw wastewater theprocess effluent nitrate concentration will not reflectthis difference because the nitrogen removed for sludgeproduction is lower for settled than for raw wastewater.Consequently the nitrate generated is approximately thesame for both waste flows. In contrast, the maximumdenitrification possible (denitrification potential) in theprocess is dependent principally on the influent CODconcentration and this concentration, we have seen, issignificantly reduced by primary sedimentation. Thismay result in a situation where it may be possible to ob-tain complete nitrate removal when treating raw waste-water but highly unlikely when treating settled waste-water. This difference in removal can have a profoundeffect on the process design for biological excessphosphorus removal.

9. REFERENCES

ARKLEY, M. and MARAIS, G.v.R. (1982) Effect of the anoxic sludgemass fraction on the activated sludge process. Research ReportNo W 38, Dept. of Civil Eng., Univ. of Cape Town.

DOLD, P.L., EKAMA, G.A. and MARAIS, G.v.R. 11980) A generalmodel for the activated sludge process. Prog. Wat. Tech., 12,47-77.

DOWNING, A.L., PAINTER, H.A and KNOWLES, G. (1964)Nitrification in the Activated Sludge Process, J. Proc. Inst.SewPurif., &4: 2, 130-158.

EKAMA, G.A., VAN HAANDEL, A.C. and MARAIS G.v.R. (1979)The present status of research on nitrogen removal; a model forthe modified activated sludge process. Presented at the Sym-posium on nutrient removal, SA Branch of the IWPC, Pretoria,May 1979. Proc. published by Water Research Commission ofSouth Africa.

HERBERT, D. (19581 Recent progress in microbiology. VII. Interna-tional congress on microbiology. Ed. by Tunevall, Alinquist andWiksell, Stockholm, 1958, 381-396.

L O E W E N T H A L , R.E. and MARAIS, G.v.R. (1977) Carbonatechemistry of aquatic systems: Theory and application. Ann ArborScience, Michigan, USA.

LOVELESS, J.E. and PAINTER, H.A. 11968) The influence of metalion concentration and pH value in the growth of a Nnrosomonasstrain isolated from activated sludge. J. Gen. Micro., 52, 1-14.

MALAN, W.N. and GOUWS, E.P. (1966) Geaktiveerde slyk vir riool-watersuiwering op Beltville. Research Report, Council for Scien-tific and industrial Research, Nov. 1965.

MONOD. J. II950) Technique of continuous culture - Theory andapplication, (Translation from French), Ann. Inst. Pasteur, 79,167.

NEYTZEL-DE WILDE, F.G. (1977) Treatment of effluents from am-monia plants - Part 1. Biological nitrification of an inorganic ef-fluent from a nitrogen-chemicals complex. Water S.A., 3(3)113-122.

'This depends on the state of the sewage; if the sewage is "fresh" a larger fraction of the TKN can still be in paniculate proteinaceousform.

**Roughly for Rs > 1,25 REf-. the increase in the total oxygen demand due to nitrification is given by M(O,) = (1 -r 5fn5)M10c) where fns is theTKN.COD ratio of the wastewater For processes including denitrificstJon see Chapter 6, Section 7.6.2.

5-17

STENSTROM, M.K. and PODUSKA, R.A. 11980) The effect of VAN HAANDEL, A.C. and MARAIS, G.v.R. (1381) Nitrification anddissolved oxygen concentration on nitrification. Wat. Research, denitrification in the activated sludge process. Research Report W14, 6, 643-649. 42, Dept. of Civil Eng., Univ. of Cape Town.

5-18

CHAPTER SIX

BIOLOGICAL NITROGEN REMOVAL

by

G A. Ekama and G.v.R. Marais

1. PROCESS FUNDAMENTALS

In this chapter the kinetics of the single sludge nitrifica-tion-denitrification process will be considered in detail.It will be shown that for municipal wastewaters, signifi-cant or even complete, removal of nitrate is possiblewith this process.

There are two main biological activities wherebynitngen is removed from the wastewater: (1) sludgeproduction and (2) respiration for energy abstractionfrom organic compounds. These two reactions differfundamentally in their biological objectives.

1.1 Nitrogen removal by sludge production

The nitrogen removal attributable to sludge productionwas considered in detail in Chapter 4, Section 9, and itwas concluded that for most municipal wastewatersonly a minor fraction of nitrogen could be removed bysludge production (see Figure 4.5).

1.2 Biological denitrification

This biological reaction is known as dissimilative reduc-tion of nitrogen or denitrification and involves thereduction of nitrate or nitrite present in the wastewaterto gaseous nitrogen which escapes to the atmosphere.Removal of nitrogen via this process is a consequence'of biological redox reactions, to obtain energy from theorganic material. In these reactions nitrate and nitriteserve the same function as oxygen, i.e. that of ahydrogen ion/electron acceptor (see Chapter 1).

Nitrate readily replaces oxygen as the terminalelectron acceptor because the electron pathway is verysimilar to that of oxygen. The difference is in the elec-tron transfer from the cytochromes where the specificenzyme reductase is replaced by the enzyme nitratereductase which catalyses the final electron transfer tonitrate instead of oxygen (see Chapter 1). Studies ofpure cultures of denitrifying organisms indicate that thepresence of dissolved oxygen prevents the formation ofthe enzyme necessary for the final electron transfer tonitrate (Chang and Morris, 1972). (See Section 2.2below).

From stoichiometry, it can be shown that whennitrate serves as the terminal electron acceptor, theequivalent mass of oxygen, (as 0) if oxygen were theterminal electron acceptor, is

1 mgN03asN =2,86 mgO as 0 (6.1)

Knowing the equivalence between nitrate and oxygen,nitrate consumption for synthesis of new organismmass can be formulated with the equations for oxygenconsumption by only slight modification. It also allows

COD balances to be properly executed on denitrificationprocesses.

1.3 Effect of denitrification on oxygen demand

In Chapter 5, Section 1, it was shown that when 1 mgNammonia or TKN is converted to nitrate 4,57 mgO arerequired. In the section above it was shown that when1 mgN nitrate is denitrified it is equivalent to 2,86 mgObeing supplied. Thus for nitrification, 4,57 mgO'imgNare required, but in denitrification 2,86 mgO/mgN canbe recovered, i.e. with denitrification 2,86/4,57.100^63% of the oxygen demand for nitrification can berecovered, if complete denitrification is achieved.Because the oxygen requirement for nitrification isabout 25 to 35% of the total oxygen requirement, a sav-ing of 15 to 20% of the total oxygen requirement can bemade by including biological denitrification. This isdemonstrated in Figure 6.1 where the oxygen demandfor (i) carbonaceous, (ii) total including nitrification and(iii) total including nitrification and denitrification (perkgCOD load) is plotted versus sludge age for a typicalraw wastewater (see Tables 4.3, 5.2 and 6.1).

Other advantages of processes including denitrifi-cation are discussed in Chapter 4, Section 11.2; see alsoChapter 5, Section 4.3 and Chapter 6, Section 1.4.

RAW WASTEWATER

0,8

I?2 aQ °

a

2 0.6-

0,4

0,2

0,0o -d. Z

O r-

- z

"1 KO -

a. r

CARBONACEOUS

0 10SLUDGE

20A G E ( d )

FIGURE 6.1: Carbonaceous, total including nitrification andtotal including nitrification and denitrification oxygen demand

per unit COD load versus sludge age for raw wastewater.

6-1

1.4 Effect of denitrification on Alkalinity

Biological denitrification in which nitrate is reduced tonitrogen gas is accompanied by an increase in Alkalini-ty. From stoichiometry, for 1 mgN nitrate denitrified,3,57 mg as CaCOj Alkalinity is produced. In Chapter 5,Section 4.3 it was shown that in the nitrification pro-cess, 7,14 mg/ / as CaCOj alkalinity was lost for every1 mgN/ i ammonia nitrified. Consequently, throughdenitrification, half the decrease in alkalinity due tonitrification can be recovered.

The recovery of alkalinity by denitrification is im-portant when treating wastewaters with low alkalinityvalues - less than 200 mg/jf as CaCO3. For suchwastewaters the production of only 15 mgN/f nitratewithout denitrification will reduce the mixed liquor pHvalue to below 5. In this event, because nitrification isextremely sensitive to pH values below 7, only partialnitrification will be obtained, even at long sludge ages.However, should denitrificatton be incorporated in theprocess, the recovery of alkalinity by denitrificationusually will cause the mixed liquor pH value to be main-tained near 7 even for low alkalinity wastewaters. Thisfavourable effect, of maintaining the mixed liquor pHvalue near 7, makes a nitrification - denitrification pro-cess virtually mandatory for activated sludge processestreating wastewaters with low alkalinity values. Thisaspect is discussed in detail in Chapter 5, Section 4.3and Chapter 4, Section 11.2.

2. REQUIREMENTS FOR DENITRIFICATION

For denitrification to occur, the following conditions arenecessary:(1) Presence of nitrate (or nitrite)(2) Absence of dissolved oxygen(3) A facultative bacterial mass(4} Presence of a suitable electron donor (energy

source).

2.1 Presence of nitrate

Normally the presence of nitrate implies nitrification as aprerequisite. The conditions for adequate nitrification inan activated sludge process have been discussed indetail in Chapter 5, Section 4. The prerequisite fornitrification involves the estimation of the maximumsludge mass fraction which may be left unaerated butstill ensuring nitrification.

2 2 AhsfinrR of dissolved oxygen

The inhibitory effect of dissolved oxygen on denitrifica-tion has been extensively reported in the literature. In acomparative test series, Carlson (1972) found that atzero dissolved oxygen level the nitrate removed was TOOper cent, while at 0,2 mg/ i dissolved oxygen (2 to 10per cent oxygen saturation), no significant denitrifica-tion was obtained. Furthermore, it should beremembered that in a completely mixed anoxic reactor,even if no measurable dissolved oxygen concentration(DO) is present, any oxygen entering such a reactor willbe utilized preferentially by the organism mass, thereby

reducing (in stoichiometric proportion) the mass ofnitrate the reactor can denitrify. Possible sources of DOare: (1) high DO concentrations in the recycles from theaerobic zones to the anoxic zones - the DO concentra-tion should be controlled to be between 1 and 2 mgO • ii.e. sufficiently high so as not to inhibit nitrification inthe aerobic reactor but not too high to reduce signific-antly the denitrification in the anoxic reactor; (2) en-trainment into the anoxic reactor via the air-water inter-face due to unnecessarily high mixing intensities; (3)Archimedian screw pumps for recycle pumping; (4)hydraulic jumps, cascades and shooting flow conditionsin interconnecting channels - closed channels and tran-quil flow conditions entrain less oxygen. (See alsoChapter 10 on operation of nutrient removal plants).

2.3 Facultative bacterial mass

The propensity to denitrify is widespread amongbacteria. Dissimilative denitrification with end productsH2, NO and NO2 has been established in numerouscases. The bulk of the bacterial mass in wastewatertreatment is facultative and a significant fraction iscapable of dissimilative denitrification. Studies ondenitrif ication tend to support the view that there is littledifference between the bacterial masses in processeswhere nitrification only or nitrification-denitrificationtakes place. There is little merit therefore in attemptingto analyse the bacterial composition of the sludge indetail. A sludge generated under aerobic conditions,when subjected to the appropriate environmental condi-tions, will show a denitrifying capability immediatelyand will continue to do so, subsequently without ap-parent change in reactivity.

2.4 Electron donor (Energy source)

The oxidation of carbonaceous compounds with nitrateas electron acceptor i.e. denitrification, provides theenergy required by the faculative heterotrophs for syn-thesis of new mass and endogenous respiration. Theelectron donor constitutes the source of the energy.The axis about which all biological denitrification in-vestigations revolve is the energy source which servesas the electron donor in the denitrification process.Denitrification has been investigated for a variety ofsubstances as carbonaceous energy sources. These canbe categorised as follows:

(a) Energy sources not present in the wastewater, i.e.an external carbonaceous energy source which isadded at the denitrification stage of the process.Compounds used as an external energy source in-clude methanol, methane, ethanol, acetone andacetic acid.

(b) Energy sources present in the influent wastewater,i.e. an internal carbonaceous energy source whichenters the system with the wastewater.

1c) Energy sources which are self- genera ted within thesystem by the release of nutrient by the organismsin the death phase.

6-2

The type of energy source utilised has led to dif-ferent process configurations for denazification studies.Usually in municipal wastewater treatment only energysources (b) and (c) above are of interest as these aresources of energy available in the single sludge ac-tivated sludge process. However, in some nutrientremoval plants a readily biodegradable COD source maybe added to the anaerobic reactor to improve thebiological P removal (see Chapter 4, Section 8 andChapter 7, Section 3.3). This practice not only improvesP removal but also the removal of nitrate by denitrifica-tion as it constitutes the addition of an external energysource (group a). Sources of readily biodegradable CODare acid digester supernatant or some form of organicwastewater with a large readily biodegradable CODfraction such as malting industry wastewater, fruit andvegetable canning factory wastewaters, juicing factorywastewaters, etc. The addition of these wastes in-creases the nitrate removal potential by denitrificationbecause the additional readily biodegradable CODallows the initial rapid rate of denitrification to operatefor a longer period of time (see Section 4.1.1 below).

3. DENITRIFICATION PROCESSCONFIGURATION

Two basic activated sludge process configurations havebeen developed for single sludge biological denitrifica-tion: (1) the Wuhrmann configuration and (2) theLudzack-Ettinger configuration.

3.1 The Wuhrmann configuration

The single sludge nitrification denitrification system inwhich endogenous energy release provides the energysource for denitrification was first proposed byWuhrmann (1964). A schematic presentation of the pro-cess is shown in Figure 6.2. It consists of two reactors inseries, the first aerobic and the second anoxic. The in-fluent is discharged to the first reactor where aerobicgrowth of both the heterotrophic and nitrifyingorganisms takes place. Provided the sludge age is suffi-ciently long and the aerobic fraction of the system is

adequately large, nitrification will be complete in thefirst reactor.

The mixed liquor from the aerobic reactor passes tothe anoxic reactor, also called the post-denitrificationreactor or secondary anoxic reactor where it is keptcompletely mixed by stirring, but with no aeration. Theoutflow from the anoxic reactor passes through a settl-ing tank and the underflow is recycled back to theaerobic reactor.

Energy release by the sludge mass due to the deathof organisms provides the energy source for denitrifica-tion in the anoxic reactor. However, the rate of releaseof energy is low, so that the rate of denitrification is alsolow. Consequently, in order to obtain a meaningfulreduction of the nitrate concentration in the anoxicreactor, the anoxic fraction of the system must be largeand this may cause a breakdown of the nitrification pro-cess. Thus, although theoretically the system has thepotential to remove all the nitrate, from a practical pointthis usually is not possible as the anoxic volume fractionwill need to be so large that the conditions for nitrifica-tion cannot be satisfied particularly if the temperaturesare low, below 15CC. Furthermore, in the anoxic reac-tor, organic nitrogen and ammonia are released throughorganism death, some of which passes out with theeffluent thereby reducing the total nitrogen removal ofthe system. To minimise the ammonia content of theeffluent, a flash or reaeration reactor may be placed be-tween the anoxic reactor and the settling tank. In thisreactor the ammonia is then nitrified to nitrate therebyreducing overall efficiency of the nitrate reductioncapability of the process.

3.2 The Ludzack-Ettinger configuration

Ludzack and Ettinger (1962) were the first to propose asingle sludge nitrification-denitrification process utiliz-ing the biodegradable material in the influent as anenergy source for denitrification. A schematic presenta-tion of this system is shown in Figure 6.3. It consists oftwo reactors in series, partially separated from eachother. The influent is discharged to the first reactorwhich is maintained in an anoxic state by stirringwithout aeration. The second reactor is aerated and

NFLUENT/

AEROBICREACTOR

ANOXICREACTOR

\ & \

WASTE FUCW

T SETTLERT—7 h• EFFUCNT

SLUDGE RECYCLE I

ANOXICREACTOR

I N F L U E N T / ^ X

AEROBICREACTOR

i

vWASTE FLOW

SETTLERJO ^

" EFFLUENT

SLUDGE RECYCLE

FIGURE 6.2: The Wutirmann process for nitrogen removal. FIGURE 6.3: The Ludzack-Ettinger process for nitrogenremoval.

6-3

nitrification takes place. As there is only partial separa-tion, the mixed liquor in the first reactor is in com-munication with that in the second reactor. Due to themixing action in both reactors, an interchange of thenitrified and anoxic liquors is induced, and the nitrateentering the anoxic reactor is reduced to nitrogen gas.Ludzack and Ettinger reported that the process gavevariable denitrification results, probably due to the lackof control of the interchange of the contents of the tworeactors.

In 1973, Barnard in developing the Bardenpho pro-cess proposed an improvement of the Ludzack-Ettingerprocess, by completely separating the anoxic andaerobic reactors, recycling the underflow from thesettler to the anoxic reactor, and providing an additionalrecycle from the aerobic to the anoxic reactor (Figure6.4).

AEROBICREACTOR


ANOXICREACTOR

WASTE FLOW

SETTLERINFLUENT

FIGURE 6 4. The modified Lud2ack-Ettinger process fornitrogen removal.

These modifications allow a significant improve-ment in control over the process performance. The highinfluent energy source discharged to the anoxtc reactoralso called the pre-denitrification reactor or primaryanoxic reactor, gives rise to a high rate of denitrificationand a substantially higher reduction of nitrate than inthe Wuhrmann process, even when the pre-denitrifica-tion reactor of this process is substantially smaller thanthe post-denitrification reactor of the Wuhrmann pro-cess. For the purposes of identification, this process willbe called the modified Ludzack-Ettinger process.

With the modified Ludzack-Ettinger process, com-plete denitrification cannot be achieved because a partof the tota! flow from the aerobic reactor is not recycledto the anoxic reactor but is discharged directly with theeffluent.

3-3 The Bardenpho conf igurat ion

In order to overcome the deficiency of incompletedenitrification in the modified Ludzack-Ettinger process,Barnard (1973) proposed combining this process withthat of Wuhrmann and called it the Bardenpho process.A schematic presentation of the Bardenpho process isshown in Figure 6.5. Barnard considered that the lowconcentration of nitrate discharged from the aerobic

PRIMARY AEROBIC SECONDARYANOXIC REACTOR ANOXICREACTOR R5ACT0R

MtXED LIQUOR RECYC, £

WASTE FLOW

SETTLER

EFFLUENT

FIGURE 6 5: The Bardenpho process for nitrogen removal.

reactor to the secondary anoxic reactor will be denitri-fied to produce a relatively nitrate-free effluent. To stripthe nitrogen bubbles generated in the secondary anoxicreactor attached to the sludge floes, he introduced aflash aeration reactor between the secondary anoxicand the final settling tank. The flash aeration reactorwas also considered necessary to nitrify the ammoniareleased during the sludge residence time in the second-ary anoxic reactor. In order to reduce the possibility offlotation of sludge in the settler due to denitrification ofresidual nitrate, the sludge accumulation in the settlerwas to be kept to a minimum. This was achieved byhaving a high recycle rate from the settler, approximate-ly equal to the mean influent flow.

Although in concept the Bardenpho process hasthe potential for complete removal of nitrate, in practicethis is not always possible, an aspect that will bediscussed in detail in Section 6 below.

4. EXPERIMENTAL BASIS FOR DENITRIFICATIONKINETICS

Carlson (1971) and Christensen and Harremoes (1977)suggested that the kinetic reaction for denitrification byactivated sludge mixed liquor can be expressed by

dlSL/dt - KX (6.2)

where

dNn,'dt = denitrificationunit time}

rate (mgN nitrate.-litre/

Nn = nitrate concentration (mgN/i)

t = time (hours or days)

X = volatile solids concentration (mgVSS/i)

K - specific denitrification constant (mgN/mgVSS/time).

This expression for the denitrification rate wasdeveloped from batch tests. It indicates that the nitrate-time relationship is linear and independent of the nitrateconcentration, i.e. the rate is zero order with respect tonitrate concentration, and a function only of the volatilesludge concentration.

In order to evaluate Eq. 6.2 as a basis for describingdenitrification behaviour in the single activated sludgeprocess, Marais and a number of co-workers (Stern,1974; Wilson, 1976; Marsden, 1977), undertook an ex-

6-4

tensive investigation into the kinetic behaviour of themodified Ludzack-Ettinger and Wuhrmann configura-tions using plug flow mixing regimes for the respectiveprimary and secondary anoxic reactors. By measuringthe nitrate concentration along the length of the plugflow anoxic reactors, the deniirification kineticbehaviour was determined by plotting nitrateconcentration-time profiles. Experiments were carriedout using only raw and settled wastewaters at differenttemperatures, sludge ages, sludge concentrations, in-fluent COD concentrations and recycle ratios.

The investigation was divided into two parts: (1)constant flow and load conditions, and (2) cyclic flowand load conditions.

4.1 Constant flow and load conditions

4. 7. 1 Experimental work with plug flow reactors

Typical nitrate concentration-time profiles for second-ary and primary anoxic plug flow reactors are shown inFigures 6.6 and 6.7 respectively. The secondary anoxicprofiles all exhibited a single linear phase that appearedto be independent of the nitrate concentration i.e. theprofiles exhibited a behavioural pattern not at variancewith Eq. 6.2. The primary anoxic profiles, however, ex-hibited profiles that all indicated a two-stage denitrifica-tion response, i.e. (1) a fast primary phase of short dura-tion (1 to 10 mintesl, and thereafter (2) a slow second-

(. = ] , S - OOO mfc C O D / *

tic.,- '•

t>O

O

<

NU. - N

A

•

•

• j L'U

; ( I O O

JIOC

tiOO

• ' 00

FIGURE 6.6. Nitrate concentration profiles observed in a plug-f low secondary anoxic reactor at 20°C [see Figure 6.2. After

Stern and Marais, 1974(,

6-5

K, ACTUAL RETENTION TIME (MINL>

FIGURE 6 7: Nitrate concentration profiles observed in a plug-f low primary anoxic reactor at 20QC (see Figure 6.4. After Stern

and Marais, 1974)

ary phase at about one seventh the rate of that of theprimary phase.

In order to fit the experimental data into the func-tional form expressed by Eq. 6.2, it is necessary todiscuss the following aspects: (i) sludge concentration,(ii) denitrification rate constants, (iii) recycle ratios, andliv) sludge age.

(i) Sludge concentration: As the denitrification reac-tion is mediated by the active Organisms, Eq 6.2should be expressed in terms of the active massconcentration Xa, i.e.

ANn = KXaAt (6.3)

In all the investigations by Marais and his co-workers it was found that the sludge production inanoxic-aerobic systems was not distinguishably dif-ferent from a comparable aerobic system, providedthat the anoxic volume fraction was not too great(i.e. less than about 50 per cent). Consequently,the active mass concentration Xa in the systemcould be expressed by Eq. 4.10 (see Chapter 4,Section 4), (Arkley and Marais, 1982).

(ii) Denitrification rate constants, K: If the denitrifica-tion rate constant, K, maintains the same valuethroughout the actual residence time in the plug-flow reactor, Ra, then the difference between theinfluent and effluent nitrate concentrations, Nni andNn0 respectively, can be written in terms of Eq. 6.3

= KXaRE (6.4)where

Ra — actual retention time in plug flow reactor (h ord)

Eq. 6.4 is valid only if the effluent nitrate from thereactor is greater than zero, see Eq. 6.17 below.The difference (Nm - NniJ is the actual removal ofnitrate concentration in the reactor. However, asthe rate of flow through the reactor is Q (1 +a)(where Q= units of influent flow to the plant perunit of time and a - recycle ratio), the flow passes(a + D/1 times through the reactor. This gives thesystem removal, Nns, i.e.

ANns= <a+1HNni-Nn0) (6.5)

The nominal retention time of the plug-flow reactoris defined by

Rn = Vr/Q (6.6)

where

Rr, = nominal retention time

V, = volume of the reactor (f)

Q = daily average influent flow to process.

Hence the relationship between the actual andnominal retention times is

Ra = Rn/(1+a) (6.7)

6-6

Substituting Eq. 6.4 for Nri, - Nn0 and Eq 6.7 for Ra

into Eq. 6.5, the systen nitrate removal is given by

ANri5= KX,Rn (6.8)

Comparing Eqs. 6.4 and 6.8, note that K is in-dependent of the recycle ratio a. Hence K can bedetermined directly from Eq. 6.4 i.e. from the ac-tual observed reduction in the nitrate concentrationin the anoxic reactor. With these basic equationsthe nilrate reduction in the primary and secondaryanoxic reactors now can be formulated.

• Secondary anoxic reactor:Referring to the experimental secondary anoxicreactor profiles, Figure 6.6, only one constantlinear nilrate time reduction phase was observedand the analyses above aliow the denitrificationrate constant in this reactor, K, to be determineddirectly from Eq. 6.4 i.e.

AN n s s - K3X;, Rns (6.9)

where the additional subscript s refers to thesecondary anoxic reactor.

By applying Eq. 6.9 to all the profiles observedexperimentally, a set of K values was obtained at14°C and 20°C respectively and the mean valuedetermined by a graphical statistical analysis. Thestatistical plot of the K3 values obtained at 14°C isshown in Figure 6.8. From the mean K3 values at14°C and 20°C the effect of temperature on K3wasformulated as

K3T = 0,072 U,03)T20(mgNOrIM/mgVASS/d)(6.10)

0,0015 0,002 0,0025 0,003 0,0035 0,004

• Primary anoxic reactor:Referring to the primary anoxic reactor profiles,two linear reduction phases were observed. Thefirst phase persisted only for a limited period, thesecond phase persisted for the remaining retentiontime in the flug flow reactor, see Figure 6.7. Todetermine the denitrification rate constants of thetwo phases, it was presumed that the secondphase was also operative during the first phase i.e.the slope of the nitrate-time profile of ihe firstphase is the sum of the slopes of the two phases;the slope of the first phase is governed by (Kx + K2).Later it will be shown that there is good reason tobelieve that this is indeed s o - see Section 4.1.2below. The separation of the phases and theirassociated K values are best appreciated bygraphically setting out the nitrate-time profile asshown in Figure 6.9. Hence the system reduction inthe pre-denitrification reactor can be written:

(6.11

where

t, =

FtGURE 6.8: Statistical plot of the observed denitrification rateconstants in a secondary anoxic reactor IK3).

duration of the first denitrification phase;subscripts 1 and 2 refer to the first and secondphases of denitrification; subscript p refers tothe primary anoxic or pre-denitrification reac-tor; subscripts a and n on the R parameterrefer to actual and nominal retention timesrespectively.

From the many experimental profiles the values ofKj and K2 were calculated using the constructionset out in Figure 6,9, and Eq. 6.11. The statisticalplots of the Kj and K2 values at 14°C and 20°C areshown in Figure 6.10. From the mean values atthese two temperatures, the following temperaturedependency equations were obtained,

K:T - 0,720 (1,20)IT-201 (mgNOj-N/mgVASS/d)(6.12)

K2T = 0,1008 (1,08)(T20t (mgrM0rN/mgVASS/d)(6.13)

for T > 13°C

K2T = K3I for 13°C.

The temperature sensitivity of the denitrificationrate constants given above and in Eq. 6.10 can beconsidered reliable only in the range 12CC to 24°C.

From the experimental data as analysedabove, Marais and co-workers came to the conclu-sion that the system reduction of nitrate in the firstphase of denitrification, ANnl,, was approximatelyproportional to the influent biodegradable CODconcentration, SQl (see Eq. 2.7) i.e.

ANr,s = oSb l (6.14)

Experimentally, for raw or settled wastewater, themean value of a was found to be 0,028 mgN/mgbiodegradable COD, and this value of o appearedto be independent of temperature. Taking account

6-7

00

PROBABILITY EQUAL OR LESS (PERCENT)

"5 s-a - •

s s.

A no 01

o

Q

3

O

Q

O

o o o o o o o o

N03-N CONCENTRATION

2m

~11st

| 2

nd

Phase

^\

x / ^

o X / ><Q / a

of these characteristics, the system nitrate reduc-tion in the primary anoxic reactor can be written as

ANnps = aSbi+K2X,Hnp (6.15}

and the duration of the first phase of denitrificationcan be found from Eq. 6.11 as

t , = • • (6.16)

From Eq. 6.16, the duration of the first phase ofdenitrification in the primary anoxic reactor, t3,depends on the recycle ratio and the active massconcentration. Usually tj lies in ihe range 2 to 10minutes.

Recycle ratios: The total system nitrate reductionof the primary and secondary anoxic reactors is thesum of Eqs. 6.9 and 6.15 i.e.

(6.17)

where

ANnts = total system nitrate removal.

An important point to note is that Eq. 6.17 is validonly if the nitrate is not reduced to zero in any oneof the anoxic reactors. Where this condition issatisfied, then from Eq. 6.17 the magnitude of therecycle ratio does not affect the system reduction.However, there must be a minimum recycle; this isdetermined by the recycle value that just causesthe nitrate to be zero at the end of a plug flow reac-tor. If the recycle is less, the nitrate will becomezero before the end of the reactor is reached - therecycle can be increased (with concomitant in-crease in the nitrate removal) to the point wherenitrate just starts to appear in the effluent. Oncenitrate appears in the effluent from all the anoxicreactors, a further increase in the recycle will haveno effect on the system nitrate reduction.

(iv) Sludge age: Mara is and co-workers found ex-perimentally that at a specific temperature thevalue of the respective denttrification constants ap-peared to remain unchanged irrespective of thesludge age for sludge ages between 10 and 25days.

4.1.2 Nitrate removal and substrate utilization

It has been established in the biological growth kinetictheory that for the utilization of 1 mgCOD under aerobicconditions, an amount of (1 -fCuYh) mg oxygen is re-quired. Numerically with fcv= 1,48 mgCOD/mgVSS andYh= 0,45 mgVSS/mgCOD, the oxygen consumptionper substrate utilization is 0,33 mg/mgCOD (Chapter 1,Section 4). In an anoxic reactor with substrate utilisa-tion, nitrate is reduced instead of oxygen. As theoxygen equivalent of nitrate is 2,86 mgO/mgN (fromEq. 6.1), the nitrate consumption per mgCOD utilised is(1 -fcvYh}/2,86 = 0,116 mgN/mgCOD or conversely 8,6mgCOD are required to reduce 1 mgN nitrate. From Eq.6.14 the nitrate concentration removed during the firstphase of denitrification is oSbi where a = 0,028 andSDl - biodegradable influent COD concentration. Thusthe COD reduction during the first phase of denitrifica-

tion is 0,028.8,6. Sbl = 0,24 Sbl. From measurements ofthe readily biodegradable COD in the influent feed, itwas found from Eq. 2.8 that the readily biodegradableCOD fraction fbs of raw wastewater was 0,24. It wastherefore hypothesized that the rapid first phase ofdenitrification is associated with the utilization of thereadily biodegradable COD of the influent. It was alsohypothesized that the second slower denitrification ratein the primary anoxic reactor and the rate in the secon-dary anoxic reactor are associated with the utilization ofthe slowly biodegradable particulate COD. From asimulation study of the denitrification behaviour inprimary and secondary plug flow anoxic reactors, it wasconcluded that both these hypotheses appear to bevalid. Details of this work (see van Haande! and Marais,1981) are not important here, however, the followingconclusions are; li) the utilization of readilybiodegradable COD and slowly biodegradable COD takeplace simultaneously (see Section 4.1.1 and Figure 6.9)and (ii) the second rate in the piirnary anoxic reactor(Kz) and the rate in the secondary anoxic reactor (K3)both arise from the utilization of adsorbed slowlybiodegradable COD but the K2 is more rapid than K3

because the concentration of adsorbed slowly bio-degradable COD in the primary anoxic reactor is higherthan that in the secondary anoxic reactor - in theprimary anoxic reactor the adsorbed slowly biodegrad-able COD arises from that in the influent and organismdeath whereas that in the secondary anoxic reactorarises from organism death only.

4. 1.3 Completely mixed anoxic reactors

Up to this stage, only plug flow reactors have been con-sidered. Although this flow regime is important from aresearch point of view, it has little practical applicationin full scale nitrification-denitrification plants. Becausethe denitrification reaction in an anoxic reactor is ap-proximately zero order with respect to nitrate, thereduction achieved in a reactor is virtually independentof the flow regime. Consequently, the equationdeveloped for the nitrate reduction, Eq. 6.17 should bevalid also for completely mixed reactors. Where theplug flow reactors were replaced by completely mixedreactors of the same volume, experimental data indicatethat the reductions are substantially the same.

Generally, completely mixed reactors are not usefulto obtain basic time behaviour characteristics as it ispossible only to measure influent and effluent concen-trations. However, if the reduction is a single zero orderreaction (as in the case of the secondary anoxicreactor), the correct K value can be calculated from theinfluent and effluent nitrate concentrations (providedthe latter is not zero) but if the reduction is due to twozero order reactions (as in the case of the primary anox-ic reactor), the correct K values cannot be separatedout; the combined effect of the two K values is to giverise to an apparent K value that varies with the retentiontime. This point is mentioned because in many in-stances the literature gives primary anoxic denitrifica-tion rate constants calculated from influent and effluentnitrate concentrations from completely mixed primaryanoxic reactors. A further cause of error is that thedenitrification rate constants in the literature often are

6-9

given in terms of mg nitrate removed per mg volatilesuspended solids per unit time. Constants reported interms of total VSS should be used with caution becausethe denitrification rates are first order with respect tothe active volatile suspended solids <Xa} (see Eq. 6.3),not first order with respect to the total volatile suspend-ed solids (X..)-the inert fraction of the VSS variessignificantly with sludge age and wastewater type (seeFigure 4.2), so that the denitrification rate constants interms of total VSS cannot be applied to wastewatersand sludge ages different to those on which the rateswere measured. However, if the wastewatercharacteristics, sludge age and temperature are knownthen, via the activated sludge theory given in Chapter 4,the denitrification rate constant, if given in terms ofVSS, may be corrected to be in terms of the activevolatile solids concentration.

4.2 Cyclic f low and load conditions

When denitrification experiments were run under cyclicloading conditions, the approach outlined above forcalculating the extent of denitrification no longer ap-peared to be applicable - the total nitrate concentrationremoved was less than observed under steady stateloading conditions, the removal of nitrate decreased asthe severity of the cyclic load variation increased, andthe TKN in the effluent showed a marked cyclicresponse, being high during the high loading phase andlow during the low loading phase, whereas the nitrate

behaved in exactly the opposite fashion. It was evidentthat to establish a general denitrification model, analternative approach to that established for analysingthe steady state behaviour was needed. This wassought in the general model for the activated sludgeprocess developed by Ekama and Marais (1978). Thedetails of this research work are not essential for usingthis monograph, but it can be mentioned that a largemeasure of success was achieved in predicting thebehaviour of denitrification under dynamic loading con-ditions once a number of modifications to the model ofEkama and Marais (1978) were accepted. Interestedreaders are referred to the following publications: Dold,Ekama and Marais (1980); Ekama, van Haandel andMarais (1979); Siebritz, Ekama and Marais (1980); vanHaande! and Marais (1981); van Haandel, Ekama andMarais (1981) and van Haandel, Dold and Marais (1982).

The outcome of the research work described in thereferences above, was a general dynamic model of theactivated sludge process including nitrification anddenitrification. The theoretical predictions of thegeneral model correlated wef! with the experimentaldata observed under both constant and cyclic loadingconditions at laboratory and pilot scale.

The general dynamic model, which describes theactivated sludge process in basic biologicalmechanisms, was utilised to study the validity of theempirical denitrification constants K. A comparison ofthe theoretically predicted and experimentally observedK values is given in Figures 6.11 and 6.12. These figures

0,05

0,0 5

0,00112 14 16 18 20

TEMPERATURE (°CJ

22

FIGURE 6.1V Simulated and experimental denitrification rateconstants Ki, K2 and K3 versus temperature for 10 and 20 days

sludge age.

6-10

0-005

0 004 -~ ^ ^ = 0.0042

0.003 -

0.002 -

0.001 -

1 1

1 1

1 1

2O'C

14'C

EXP.SIM

I !

0.0027

0.005

0,004 —

(IT

X0-003 -

0.002 | -

0.001 -

1 1

1 1

1 1

20°C

14'C

tXP.SIM .

K 3

J 1

0,0033

00027

b 10 15, 20 10 15 0 25

FIGURE 6.12: Simulated and experimental denitriiication rateconstants K; and Kj versus sludge age at 14°C and 20<1C.

show that for sludge ages between 10 and 30 days andtemperatures between 12 and 22°C, good correlationbetween simulated and experimental K values is obtain-ed. The good correlation indicates that although the Kvalues are empirical and have no fundamental biologicalkinetic basis (of the kind on which the general dynamicmodel is based), they nevertheless appear to have suchconsistency that they can be applied to the design ofanoxic reactors for constant loading conditions.

5. DENITRIFICATION POTENTIAL

From the empirical denitrification equation of theprimary and secondary anoxic reactors, Eqs. 6.15 and6.9, it is clear that there is a maximum concentration ofnitrate that can be removed by the anoxic reactors. Interms of these equations, this maximum for the primaryanoxic reactor depends on the infiuent biodegradableCOD, the active mass concentration and the retentiontime, and 1or the secondary anoxic reactor only on theactive mass concentration and the retention time. Themaximum concentration of nitrate that can be removedby an anoxic reactor is known as the denitrificationpotential of the reactor,

5.1 Denitrification potential of the primary anoxicreactor

From Eq. 6.15, the denitrification potential of theprimary anoxic reactor, Dpl is given by

Dpl - QSbl+K2XaR, (6.18)'pi " "bi ' ' V a 1 'np

Now XaRnp = XaVa/Q (where Va = volume of the primaryanoxic reactor and Q - daily average influent flow). Theterm XaVa represents the mass of sludge in the primaryanoxic reactor, which can be expressed as a fraction of

the total mass of sludge in the process. Let fx1 be themass of sludge in the primary anoxic reactor as a frac-tion of the total mass of sludge in the process i.e. theprimary anoxic sludge mass fraction, then from Eq. 4.10

XaVa = fxl M(Xa)

= fxt YhRsGSbi/(1+bhRs)

Dividing though by Q

XaVa /Q=f y lYhR sSb l / (1+bhR s ) (6.19)

Substituting the RHS of Eq. 6.19 for XaRtl)) into Eq 6.18yields

= Sbl{«+K2f fc lYhRs/(1+bhRs}} (6.20)

where

Dp l •= denitrification potential of the primary anox-ic reactor (mgN/Z influent)

Stll - biodegradable COD concentration of the in-fluent

a - fraction of nitrate removed by the initialrapid phase of denitrification

= f b sn- f c vY h ) /2 ,86(see Section 4.1.2 above)

where

fbs = readily biodegradable fraction of the influentbiodegradable COD

fx1 = primary anoxic sludge mass fraction.

In Eq. 6.20, it is assumed that the initial rapid rateof denitrification is always complete, i.e. the actualretention time in the primary anoxic reactor is alwayslonger than the time required to utilize all the readilybiodegradable COD in the influent, given by t1 in Eq.6.16. If t : is taken to be the minimum actual hydraulicretention time, it can be shown by substituting Eq. 4.10

6-11

_ 0,10RAW WASTEWATER

o

^0,08 -

5 1 0,060 °a. *

-oa

1 10,04 L

o— occ £ 0 ,02 •

Q O

2 Q00

PR;MARY A

20°14°

<0

- '

——-

,—-

cc

X i C

<>i

-——-

REACTOR= 0^0- -

-OS-™

0,20

1 L

SECONDARY

>

20°C

ANOXIC

. 1

REACTOF

-

0,50 •

0,35—. -

0L20 .

•

10SLUDGE AOE

20(d)

300 10SLUDGE

20A6E (d)

FIGURE 6.13: Denitrrfication potential per kgCOD applied tobiological process versus sludge age for primary (left) andsecondary Irighi] anoxic reactors of different unaerated sludgemass fractions (fxiand f.^l for raw wastewater at 140C and

20° C.

for Xd and Eq. 6.16 for ti, lhat the minimum primaryanoxic sludge mass fraction fxln,in to deplete the readilybiodegradable COD is

f.J1~fCvYh)(1 + bhR5)xlmin

_ 'bs<

2,86(6.21

Substituting the values of the kinetic constants into Eq.6.21, f i p m i n< 0,08 for F\ > 15 days at 14CC. This value ismuch lower than most practical primary anoxic reactorsso that Eq. 6.20 will be valid in most cases.

A diagrammatic representation of Eq. 6.20 is givenin Figure 6.13 for different primary anoxic sludge massfractions (fx1) and temperatures. The diagram showsthat the denitrification potential of a primary anoxicreactor increases as sludge age, anoxic sludge massfraction and temperature increase, but that thedenitrification potential is most sensitive to the anoxicstudge mass fraction (f,.).

5.2 Denitrification potential of the secondaryanoxic reactor

The denitrification potential of the secondary anoxicreactor is found in a similar fashion as that for theprimary anoxic reactor and is found to be given by

DD3=Sblfx3K3YhR&/(1-t-bhTRs) (6.22)

where

Dp3 = denitrification potential of the secondaryanoxic reactor (mgN/f influent)

fx3 = anoxic sludge mass fraction of the secon-dary anoxic reactor.

A diagrammatic representation of Eq. 6.22 is given inFigure 6.13 for different anoxic sludge mass fractionsand temperatures. The denitrification potential in-creases as the sludge ages, temperature and anoxicsludge mass fraction (fx3) increase, but is most sensitiveto the anoxic sludge mass fraction lfx3).

6. DESIGN PROCEDURE FOR N REMOVALPROCESSES

In the preceding sections, four important parameterswere defined and discussed. These are (1) maximumunaerated sludge mass fraction <fxrTl, Chapter 5, Section4.4), (2) nitrification capacity <NC, Chapter 5, Section5.3), (3) denitrification potential (Section 5 above) andthe TKN/COD ratio (Chapter 3, Section 3,3). Theseparameters greatly facilitate the design of a nitrogenremoval process. The salient aspects of these fourparameters are briefly set out below.

(/') Max/mum unaerated sludge mass fraction

As nitrification is a prerequisite for denitrificationthe fraction of the sludge mass under aerobic con-ditions must be sufficiently large to allow completenitrification. At a particular sludge age, this sets alimit on the fraction of sludge that is not aeratedi.e. sets a maximum unaerated sludge mass frac-tion (fxm). This maximum unaerated sludge massfraction is dependent principally on the maximumspecific growth rate of the nitrifiers at 20°C < nm2o>

6-12

attainable with the waste water, the minimumaverage temperature of the wastewater, (Tnml) andthe sludge age of the process (Rs). In order to en-sure nitrification, it is recommended that fxrT1 islimited at 0,50 to 0,60 and that the minimumaerobic sludge mass fraction is increased by a fac-tor of safety of 1,25 To 1,35 (see Chapter 5, Section4.4).

Hi) Nitrification capacity of process

The nitrification capacity is the concentration ofnitrate per unit of influent flow that the processproduces by nitrification of the influent TKN con-centration. If the unaerated sludge mass fractionsatisfies the conditions set out above, completenitrification is likely to take place and the effluentTKN concentration will be between 2 and3 mgN/7. Under these conditions the nitrificationcapacity is independent of the process configura-tion and is the difference between the influent TKNconcentration and the sum of the nitrogen requiredfor sludge production and effluent TKN concentra-tion (Chapter 5, Section 5.3).

iiii) Denitrification potential of process

The denitrification potential of a process is themaximum mass of nitrate per unit influent flow thatthe process as designed can denitrify. It is directlyproportional to the influent biodegradable CODconcentration and the position and sizes of theanoxic reactors because denitrification in theprimary anoxic reactor is faster than in the secon-dary anoxic reactor.

{ivj The influent TKN/COD ratio

Because the nitrification capacity and denitrifica-tion potential are approximately proportional to theinfluent TKN and COD concentrations respectively,the influent TKN/COD ratio is a relative measure ofthe mass of nitrate generated in the process bynitrification and the mass of nitrate that can beremoved in the process by denitrification.

6.1 Principles of the design procedure

From the wastewater characteristics i.e. influent TKNand COD concentrations (N,, and Sti), maximumspecific growth rate of the nitrifiers at 20°C f/inrT,2o*. t n e

readily biodegradable COD fraction (fbs) and theminimum average wastewater temperature ( T ^ l , themaximum unaerated sludge mass fraction [1Krn) and thenitrification capacity (Nc) can be calculated for aselected sludge age (Rs). The fxm then can be subdivid-ed into primary and secondary anoxic sludge mass frac-tions (fxl and fx3) and this division fixes the denitrifica-tion potential of these two reactors and hence aiso ofthe process. When considering the subdivision of f,^, itshould be remembered thai the denitrification potentialof the primary anoxic reactor (Dpl) is greater than that

of the secondary anoxic reactor <Dp3) for equal anoxicsludge mass fractions, but that sufficient nitrate mustbe recycled to the primary reactor to fully utilize itsdenitrification potential.

The relative sizes of the primary and secondaryanoxic sludge mass fractions (fx1 and fx3), as well aswhether or not a secondary anoxic reactor is to be in-cluded in the process configuration, depend on the in-fluent TKN/COD ratio. If the objective is to remove asmuch nitrogen as possible, the denitrification potentialof the process (which may or may not include a secon-dary anoxic reactor} must be as close as possible to thenitrification capacity. Complete denitrification is possi-ble only when the process configuration incorporates asecondary anoxic reactor and if the sum of thedenitrification potentials of the primary and secondaryanoxic reactors is greater than the nitrification capacity,i.e. Dpl 4 DP3 > Nc (provided Dp, is fully utilized by hav-ing a sufficiently high mixed liquor a-recycle).

If complete denitrification is not possible due to ahigh TKN/COD ratio, nitrate will be present in the ef-fluent. If the effluent nitrate concentration is greaterthan about 5 to 6 mgN/f, the incorporation of a secon-dary anoxic reactor becomes an inefficient utilization ofthe unaerated sludge mass fraction; an improved nitrateremoval will be obtained if the secondary anoxic reactor(fx3) is incorporated into the primary anoxic reactor (f^).Removal of the secondary anoxic reactor from theBardenpho process (Figures 6.5) reduces the configura-tion to a Modified Ludzack-Ettinger (MLE) process(Figure 6.4).

The above discussion indicates that in order toselect a nitrogen removal process configuration, it isnecesary to check whether or not complete denitrifica-tion can be achieved.

6.2 Can complete denitrification be achieved?

In the nitrogen removal processes, the maximum anoxicsludge mass fraction available for denitrification, fKdm, isequal to the maximum unaerated sludge mass fractionf,!r, i.e.

fxdm = f«m (6.23)

where fxm is given by Eq, b.24 for selected R5, ^nrTlT and' min-

To check whether or not complete denitrificationcan be obtained with the maximum anoxic sludge massfraction, the following reasoning is applied: In theBardenpho process (which has the potential for com-plete denitrification), for a fixed underflow s-recycleratio,* the mixed liquor a-recycle ratio governs thedistribution of the nitrate from the aerobic reactor be-tween the primary and secondary anoxic reactors - thehigher the a-recycle ratio with respect to the s-recycleratio, the greater the proportion of nitrate recycled tothe primary anoxic reactor. For the selected a- ands-recycle ratio, the best denitrification performance isobtained when the primary anoxic reactor is just loadedto its denitrification potential. !f the primary anoxic reac-tor is loaded to its denitrification potential, the nitrateconcentration in the outflow of the reactor is zero and

'Recycle ratios specified with respect to the daily average influent flaw rate.

6-13

the nitrate concentration in the aerobic reactor is givenby Nc/(a + s-*-1) i.e. the nitrification capacity diluted in-to the total flow entering the aerobic reactor. Knowingthe nitrate concentration in the aerobic reactor and in-cluding the effect of the dissolved oxygen concentra-tion (DO) in the aerobic reactor and underflows-recycle, the equivalent nitrate load imposed on theprimary and secondary anoxic reactors by the selecteda- and s-recycles can be calculated. Now a usefulanalytical device is to increase the equivalent nitrateload on the secondary anoxic reactor by a factorK2T''K3T so that the reactor acts like a primary anoxicreactor except that it receives no readily biodegradableCOD. Now adding the equivalent nitrate load on theprimary anoxic reactor to the adjusted equivalent nitrateload on the secondary anoxic reactor, a total equivalentnitrate load is obtained. Because the secondary anoxicreactor has been transformed to act like a primaryanoxic reactor, their respective contributions to totaldenitrification potential of the process need not beknown, and the maximum anoxic sludge mass fraction(f,din) can be assumed to be in the form of a primaryanoxic reactor. Consequently the process denitrificationpotential Dpp can be calculated by means of Eq 6.20with fx1 = fxdm. Now if the process denitrification poten-tial Dpp of the maximum anoxic sludge mass fractionfxdnv 's greater or equal to the total equivalent nitrateload, complete denitrification is possible for the selecteda- and s-recycle ratios. In contrast, if the D_p of f>t)m isless than the total equivalent nitrate load, completedenitrification is not possible for the selected a- ands-recycle ratios, and nitrate will be present in the ef-fluent.

Following the above reasoning and assuming theDO concentrations in the a- and s-recycles are Ort andOv mgO/jf respectively, the following equation for theeffluent nitrate concentration Nne for the Bardenphoprocess can be developed:

D,

(6.24)

where

Also, the optimum primary and secondary anoxicsludge mass fractions i^ and 1x2, and the total f>d, aregiven by

Mre = effluent nitrate concentration (mgN/f)Nc = nitrification capacity (mgN/f Ia,s ~ mixed liquor and sludge underflow recycle

ratios repsectivety2T,KTT = second denitrification rate in primary anoxic

reactor and denitrification rate in secondaryanoxic reactor respectively (mgN/mgVASS/d)

0a (0 s = dissolved oxygen concentration in mixedliquor a and underflow s recycle respectively(mgO/f)

DPD = denitrification potential at temperature T°Con the assumption that the maximumanoxic sludge mass fraction fydrl found fromEq 6.23 is all in the form of a primary anoxicreactor i.e. DDp is given by Eq. 6.20 withf o ~ W 9 n d Kz adjusted to T°C with Eq.6.13. ' '6.25)

M

= f.

(6.26)

(6.27)

(6.28)

where

fxi< f*3. f»m = primary, secondary and total anoxicsludge mass fractions respectively.

For a selected sludge age (R6) and wastewatercharacteristics (Sti, NT|, fbs, Hr.m2o. Tmin, fup, fus) - fromwhich V , (Eq. 5.24), fxdm (Eq. 6.23}, K2T/K3T {Eqs. 6.13and 6.10), Nc (Eq. 5.29) and D.,P (Eq. 6.25) can becalculated - and selected DO concentrations in thea-and s-recycle ratios (Os and OE), the only unknowns inEqs. 6.24 to 6.28 are the a- and s-recycle ratios. Thes-recycfe ratio usually is specified (at say 1) to obtainsatisfactory settling tank performance. Hence Mne, fxl

and fx3 all depend on the a-recycle ratio only, and hencethese three parameters can be calculated for variouschoices of the a-recycle ratio. If Nn(, ^ 0,0 mgN/jf forsome specified s- and selected a-recycle ratios, thencomplete denitrification is possible at these recycleratios.

In the solution procedure, if Eq. 6.24 yields N re < 0,Nne must be set equal to zero before substitution intoFqs 6.26 and 6.27. When Nne has to be set equal tozero, it will be noticed that fxc!, obtained from Eq. 6.28 isless than the fxdm assumed to calculate Dpp. The reasonfor this is that when Nrie < 0, DPP > Nc. By insertingN n e -0 in Eqs. 6.26 and 6-27, these equations give fxl

and f 3 that vj\\\/ust produce complete denitrification forthe selected a- and specified s-recycle ratios, so that fxd.turns out to be less than f>rirn.

For the specified s-recycle ratio, Eqs. 6.24 to 6.28are valid only for a-recycle ratios falling between a lowerand an upper limit;

(i) Lower limit: the a-recycle must not be lower thanthat which gives fxl as calculated from Eq. 6.26 lessthan fximin as calculated from Eq 6.21 - iffxl < f> lm jn, the readily biodegradable COD is notcompletely utilized in the primary anoxic reactorleading to inefficient denitrification and also mak-ing Eq. 6.20 for Dp1 invalid.

(iil Upper limit: the a-recycle ratio must not be increas-ed above that which gives fx3 from Eq. 6.27 asmaller value than that required to remove only theDO discharged to the secondary anoxic reactor,i.e. Dp3> (1+s) O_a/2,86 where Dp3 is given by Eq.6.22. Hence the minimum secondary anoxic sludgemass fraction is given by

6-14

f.2,86SbhYhR6K3

(6.29)

Only the a-recycle ratios falling between the lower andupper limits specified above are valid a-recycle ratios.

By analysing the process with a specified s-recycleratio for increasing values of valid a-recycle ratios, then,if Nne > 0, IMr (. and the optimal subdivision of flOrn intofxl and fx3, or, if Nne ^ 0, the total anoxic sludge massfraction to just obtain r\lne. = 0, f>dl, and its optimal sub-division into fxl and fx3 can be plotted versus thea-recycle ratio.

For design purposes where the process is tooperate between temperatures Trr.,n and TrnH,., thefollowing sequence of calculations needs to be made tocheck whether complete denitrification can be achievedand if so, to select the optimal process configuration:

Step 1: Select wastewater characteristics; ST|, Ntl, fbs,'up- 'u&' Mnm20- ' rr.6xi ' mm-

Step 2: Select R, and S,.

Step 3: Calculate firn for Jmn from Eq. 5.24.

Step 4: With f,rn and R£ calculate S( for Tmax from Eq.5.24.

Step 5:

Step 6:

Estimate N,e for Tr and Tr

Step

Step

Step

Step

Step

Step

Step

Step

Step

Step

Step

7:

8:

9:

10:

11:

12:

13:

14:

15:

16:

17:

With f.jp and fUL, calculate Ns for Tmax and Tmin

for selected R., from Eq. 4.23.

Calculate Nt from Eq. 5.29.

Calculate Dpp from Eq. 6.25 for Tr1iax and forTmin with fxdm given by Eq. 6.23.

Select s, Oa and Os.

Calculate fxlmJrt from Eq. 6.21 for Tmax and for' mm •

Calculate fv3n,ir, from Eq. 6.29 for TrJWX and for

Select the a-recycle ratio and calculate Nnt forT fninfrom Eq. 6.24.

If N re < 0, set N,.e = 0.

With Nne calculate fx1, fx3 and fxd1 from Eqs.6.26 to 6.28.

Check that fxl ^ fKl^n and fxs fx3mjn. l f n o t -

discard selected a-recycle ratio as invalid.

Repeat steps 12 to 15 for different a-recycleratios.

Repeat steps 12 to 16 for Tmax.

xm

in a-recyThe fxl value cor-

max fixes the primary

The analysis as to whether or not completedenitrification can be achieved must be undertaken atthe lowest expected temperature (Tmirj, because thedenitrification potential D,J(, decreases with decrease intemperature. If complete denitrification can be obtainedat Tmin (i.e. Nne < 0 from Eq. 6.24 for some valida-recycle ratio at the specified s-recycle ratio), then f,. isfixed at the highest expected temperature (Tmax). Inrepeating the calculation at T,,,^ it will be found that fat Tma> is less than fxa. at T ^ , and for a certain a-recycleratio, there will be a minimum fxm*responding to the minimum fxd! at Tanoxic sludge mass fraction for both Tmax and Tmin.Now fx3 is fixed at the minimum expected temperature(Tf.ljrJ at the fx3 value corresponding to the fixed fKl

value and the total anoxic sludge mass fraction fvc,, forboth Jn,ax and Trrmi is given by fy1 + fx3. The fx3 is fixed atTmin because if it were fixed at TmdX, then fxdl will be in-sufficient to achieve complete denitrification at Tm;nbecause for fixed fkI, fxdt at Tmax is less than ftcjl at Tmin.With fj,! and f,3 fixed for both TmaK and TrTlin, the op-timum performance of the process is achieved when it isoperated at the a-recycle ratio corresponding to the fix-ed fxl value, this a-recycle ratio being the optimum ao. Itshould be noted that with fx1 and f,3 fixed,

(i) a0 decreases with decrease in temperature becauseDpl decreases with decrease in temperature, and

(ii) the use of any a-recyc!e ratio other than a0 resultsin a poorer denitrification performance; if a < a0,Dpl is not fully utilized and if a > a0, unnecessarilylarge quantities of DO are discharged to theprimary anoxic reactor.

In the design of a Bardenpho process for a selectedsludge age, if it is found that the total anoxic sludgemass fraction for complete denitrification fxd, is lessthan the maximum allowed fxrJrri, the following optionsare open:

(1) fx1 and fk3 can be increased to give an fxdl equal tof)Cm. This introduces a factor of safety indenitrification by making an allowance for a reduc-tion in the denitrification rates K at 20°C, a reduc-tion in the influent readily biodegradable COD frac-tion fbE or minimum temperature (Tm,n) or an in-crease in influent TKN/COD ratio.

(2) The sludge age (Rs) can be reduced so that fxdrn

becomes equal to fxdt. The lower R£ will allow a sav-ing in process volume (see Figure 4.3). The lowerRs can be estimated as follows: Set fx0^ equal tothe f,dl required for complete denitrification at Tmjn;determine fxm from Eq. 6.23; with fxm calculate Rs

from Eq. 5.24; to check, repeat step by step pro-cedure with the new Rs.

Option (2) is not as good an alternative as option(1) because it exludes a factor of safety on the nitrogen

*lf Trnan is less than about 18°C, a clear minimum cannot be discerned (see Figure 6-15 for 14°C), In this event any reasonable f, i can beselected wiih the proviso that complete utilization of ihe readily biodegradable COD can be assured even wiih a significant decrease in K120(see Eqs. 6.12 and 6.21). The existence of a minimum, and the temperatures above which it becomes discernible, depends much on the choiceof the DO concentration in the recycles 0B and OE.

6-15

removal. However, exclusion of a factor of safety ondenitrification is not critical for N removal in the Barden-pho process because it achieves efficient N removalwith respect to the other N removal process (i.e. theMLE process) up to an effluent nitrate concentration ofabout 5-6 mgN ' f. However, the exclusion of a factor ofsafety on denitrification for the Phoredox process mayhave serious consequences because the discharge ofany nitrate to the anaerobic reactor severely reduces theP removal obtainable in the process. The effect ofdenitrification efficiency on excess biological P removalis discussed in detail in Chapter 7.

6.3 When complete denitrification cannot beachieved

In the Bardenpho process, if it is found that completedenitrification cannot be obtained, (i.e. Nne > 0 fromEq. 6.24 for all valid a-recycle ratios at the specifieds-recycle ratios) nitrate will be present in the effluentand f,,„ will be equal to f,orri. Generally, if the objective isN removal only, the Bardenpho process will be ap-propriate only if Nne is less than 5 to 7 mgN/ i ; whenNne > 5 to 7 mgN/! , - which will usually be the case ifthe TKN/COD ratio > 0,10 mgN/mgCOD - the MLEprocess (Figure 6.4) will yield a better nitrogen removalefficiency for the particular wastewater.

For the selected sludge age (R&), factor of safety onnitrification (Sf) and wastewater characteristics (Stl, Ntl,fbs- Tmir. Tmax, fup, flJS), the firn (Eq. 5.24), Nc (Eq. 5.29),

r, (Eq. 6.23) and Dpp (Eq. 6.25 with for theMLE process are equal to those of the Bardenpho pro-cess and are found by the identical procedure (i.e. steps1 to 8 given above). For the MLE process, the bestdenitrification performance is obtained when the anoxicreactor is just loaded with nitrate to its denitrificationpotential. For a specified s-recycle ratio, the mixed li-quor a-recycle ratio which loads the anoxic reactor to itsdenitrification potential is the optimum a-recycle ratio,ao, and will yield the minimum effluent nitrate concen-tration. When the nitrate load imposed on the anoxicreactor by the a- and s-recycies (including DO concen-trations in the a- and s-recycles of Oa and O& mgO/jfrespectively), is less than or equal to the denitrificationpotential (D,,P), the nitrate concentration in the outflowof the anoxic reactor is zero and that in the aerobic reac-tor and effluent is Nc/(a + s+ 1) i.e. the nitrificationcapacity "diluted" into the total flow entering theaerobic reactor. Equating the nitrate load on the anoxicreactor imposed by the a- and s-recycles (including theDO in the recycles) to the denitrification potential Dpp

and solving for the a-recycle ratio yields a0;

4AC)}/I2A) (6.30)

wnere

A - Oa/2,86

B = Mc-D^ D + {<s + 1)Oa + s.O£}/2,86

C = (s+1}(D p p - sOs/2,86) - slMc

The effluent nitrate concentration Nne for any

a-recycle ratio less than or equal to a0 is given by Eq.6.31 but will be a minimum when a = ao, i.e.

Nne = Nc (a + s + 1) mgN/ ! (6.31)

In the design of an MLE process, it will be foundthe lower Nc is with respect to Dpp (i.e. the lower the in-fluent TKN/COD ratio) the lower the Nneand the higherthe a(j. In practice, to operate an MLE process with anac-recycle ratio greater than 6:1 will be uneconomical:the increase in nitrate removal by increasing a0 from 6 to10 is only 5% and, therefore, will not be cost-effectivewith respect to the increased pumping costs. Hence,the N removal of the MLE process is restricted by amaximum a-recycle ratio of say 6, which results in aminimum Nne of 5 to 7 mgN/! . This restriction on theMLE process will not be a problem in practice becausehigh values for ao are only obtained for wastewaterswith low TKN/COD ratios (< 0,10 mgN/mgCOD} andthese wastewaters are more effectively treated in aBardenpho process (see section above). Alternatively, ifnear-complete or complete N removal is not essential,the sludge age can be reduced; for a fixed TKN/CODratio this reduces Dpp (because fKfT1 is reduced) withrespect to Nc and hence reduces a0. The optimum Rs forthe fixed TKN/COD ratio can be found using the pro-cedure set out above with different choices of Rs or,alternatively, the method of van Haandel, Dold andMarais (1982) can be used.

The design procedure for N removal processes setout above will be demonstrated in the next section withthe aid of worked examples.

7. DESIGN EXAMPLES

The design of biological N removal plant by means ofthe procedure set out above is demonstrated by conti-nuing the numerical example given in Chapter 4, Sec-tion 12 (in which the design procedure for organicmaterial removal was demonstrated) and Chapter 5,Section 6 (in which the behaviour of nitrification wasdemonstrated).

7.1 Variability of wastewater characteristics

7. /. 1 Variability of nitrification rate

The nitrification rate, as expressed by the maximumspecific growth rate of the nitrifiers at 20°C (Mnm2o) 'Sstrongly dependent on the nature and source of thewastewater, so much so that it is considered a waste-water characteristic. The value can range from 0,65/d inpurely domestic wastewaters down to 0,20/ d for waste-waters with a large industrial' contribution. Owing touncertainty in the ^nfT,2o value for a particularwastewater, in the absence of any data on the Hnm2ovalue, a low value, say between 0,3 to 0,4/d, needs tobe selected to provide reasonable assurance that a safedesign with respect to nitrification will be obtained (seeChapter 5, Section 4.1). In this design example, a valueof 0,36/d is selected to demonstrate the situation in a

"The type of industry referred to here is that which produces effluents which inhibit biological activity, e.g. metal finishing industry, platingpaint production plants eic

6-16

design with a relatively lowand 6.2).

value (see Tables 6.1

7. 1.2 Variability in denitrification rates

The denitrification rates K at 20°C are relatively insensitive to the nature and source of the wastewater.However, different values have been observed with dif-ferent wastewaiers - in instances where the waste-water contains an appreciable industrial* contribution,reduced denitrification rates may be encountered. Thedenitrification rates used in this design example are themean values found in the experimental work at theUniversity of Cape Town, i.e. K120=0,720, K23D = 0,101and KHO = 0,072 mgNOj-N/lmgVASS.dl (see Table6.2). These values were found on Strandfontein waste-water, which is an approximately average wastewaterwith a relative smal! industrial contribution. For mostnormal municipal wastewaters encountered in SouthAfrica the denitrification rates given above are likely toresult in a sufficiently accurate estimate of the denitrifi-cation potential of the process. Where there is doubtregarding the applicability of these rates, it is recom-

mended that they are reduced by a factor of safety ondenitrification, i.e. by about 1,1 to 1,2 times. Themagnitude of the factor of safety should reflect theuncertainty in the applicability of the average denitrifica-tion rates given in this monograph. •

7. 1.3 Variability in readily biodegradable COD fraction

The readily biodegradable COD fraction of a wastewatercan vary considerably depending on the compositionand origin of the various contributions that make up thewastewater. It has been found that in approximatelynormal principally domestic raw wastewater, approx-imately 20% of the total (or 25% of the biodegradable)COD is readily biodegradable. However, where in-dustrial wastewaters containing a high biodegradableorganic material content are discharged to themunicipal sewerage system, significantly higher readilybiodegradable COD fractions can be encountered. Forexample, yeast factories, breweries, malting processesand fruit and vegetable juicing and canning industriesproduce wastewaters with high readily biodegradableCOD fractions. Municipal wastewaters containing COD-

T A B L E 6.1 RAW AMD SETTLED WASTEWATER CHARACTERISTICS IMPORTANT FOR DENITRIFICATION (FOR OTHERCHARACTERISTICS SEE TABLES 4.3 AND 5.2

Parameter

Maximum specific growth rale of nitnfiers at 20°C

Denitrtficaiion tales at 20^C

Readily biodegradable COD fraction

Symbol

iK2

K3

ValueRaw Settled

0,36

0,7200,1010.072

0.24

0,36

0,7200,1010,072

0.33-

Units

16

mgNOj-N/mgVASS/d.

"Note increase on f^ fraction in settled wastewater (see Section 7.1.3) -- based on a 40% total COD decrease and a 10% readily bio-degradable COD decrease in the primary settling tanks. (See Chapter 4, Seclion 12.1 and Table 4.3).

TABLE 6.2 TEMPERATURE ADJUSTMENT Of THE DENfTRIFICATlON RATES K-, K; and K3

Constant Symbol 20"C 0

1st Denit. rate in Prim. Anoxic K12o* 0,720 1,200

2nd Denit. rate in Prim. Anoxic K220* 0,101 1,080

Denit. rate in Second, Anoxtc K320* 0,072 1,029

Max. Spec, growth rate of Nitnfiers Hnm2Q** 0.36 1,123

Endog. Resp. rate tor Nitrifiers bn20*" 0,04 1,029

•Units: mgNO3-N/(mgVASS dl"Units- per day

14'C

0,241

0,0636

0,0607

0,18

0,034

22°C

1,036

0,1178

0,0762

0,45

0,042

Eq. No.

6.12

6.13

6.10

5.15

5,16

"Industries producing biodegradable wastes e.g. malting factories, breweries, fruit and vegetable processing factories, etc. do not fall in thisgroup as these may promote biological N and P removal, isee Section 7.1.3).

6-17

tributions from these sources can have relatively highreadily biodegradable COD fractions depending on themagnitude of the industrial discharges.

Many factors can reduce the readily biodegradableCOD fraction of a wastewater. For example thedischarge of waste biological sludge to the seweragesystem for sludge handling at treatment works lowerdown the sewer line can deplete a wastewater of all thereadily biodegradable COD, particularly if the combinedflow is lifted by Archimedian screw pumps. Further ex-amples are cited in Chapter 7, Section 9,

The above examples demonstrate that the readilybiodegradable COD fraction can differ significantly be-tween different wastewaters. Consequently, it is recom-mended that whenever possible, the fraction should bemeasured. The importance of this fraction in biologicalnitrogen (and phosphorus) removal is such that it isalways worth the effort of measuring it.*

It should be noted that the readily biodegradableCOD fraction (as defined by Eqs. 2.8 and 2.9) is affectedby primary sedimentation. This is because primarysedimentation removes a large fraction of the par-ticulate COD but very little soluble COD. Consequently,the readily biodegradable COD concentrations (SbSl) areapproximately the same before and after primarysedimentation but, because the total or biodegradableCOD (Stl, Sbi) has been significantly reduced in theprimary settlers, the readily biodegradable COD fraction^ts' W °f t n e settled wastewater is increased (seeChapter 2, Section 2.1).

For the design example it will be assumed that thereadily biodegradable COD fraction with respect to thebiodegradable COD fbi for the raw wastewater is 0,24,i.e. the value found for approximately normal principallydomestic raw wastewaters. The readily biodegradableCOD fraction (fbs) of the settled wastewater, assumingthat primary sedimentation removes 40% of the totalCOD but only 10% of the readily biodegradable COD is0,33. (see Table 6.1, Chapter 4, Section 12.1 andChapter 2, Section 2).

7.2 Effect of temperature

From the kinetics of denitriftcation discussed above, it isclear that the denitrification rates K1( K: and K3 at!decrease with decreasing temperature. This, combinedwith the fact that the maximum specific growth rate ofthe nitrifiers at 20°C (Hnm2o' a ' s 0 decreases withdecreasing temperature has the result that the criticalcondition for the nitrogen removal process is theaverage minimum wastewater temperature. The designcompleted for the average minimum wastewatertemperature (Tmin) is checked at the average maximumwastewater temperature <Trr,sx) because the oxygen de-mand per unit COD load increases as the temperatureincreases.

To facilitate the calculations of the design exampleat Tmjn = 14°C and Tmax = 22°C, the denitrification rates« i , K2 and K3 are adjusted for temperature in Table 6.2.Temperature adjustment of ^nrr2o ar |d b-, is given inTable 5.3, but for convenience is repeated in Table 6.2.

7.3 Maximum unaerated sludge mass fraction

This aspect is discussed in detail in Chapter 5, Section4.4 but it is briefly repeated here because sizing of themaximum unaerated sludge mass fraction is one of themost important decisions in the design of a nutrientremoval process.

For nitrogen removal processes, the larger theunaerated sludge mass fraction (fxT} the greater thenitrogen removal and the lower the effluent nitrate con-centration. Generally at a minimum temperature of14°C, to obtain an fx, of 0,50 for favourable ^nrn2C, values(0,40/d), a sludge age of 15 to 20 days is required to en-sure nitrification (Sf=1,25> (see Eq. 5.24); for lessfavourable j jnrn20 values (between 0,3 and 0,4/d), the re-quired sludge age is 20 to 30 days (see Figure 6.14).

0,80 TEMPERATURE = I 4 • C FACTOR OF SAFET *f = l,25

(?) 0,60

<* z< 2 0,40z t-

5w 0,20x vi< <

RECOMMENDEDMAXIMUM

o.oo1

FIGURE 6.14: Maximum unaerated sludge mass fract ion ver-sus sludge age for different maximum specifir growth rates ofthe nitrifiers at 2QC'C (^^m2d) at 14°C and factor of safety (Sf) of

1,25. (From Eq. 5.24).

Now because the volume of the process is directly pro-portional to the mass of sludge in it (see Chapter 4, Sec-tion 6) a 30-day sludge age process will require a pro-cess volume about 1 ;3 larger than a 20 day one. Conse-quently, for unfavourable nm2Q values it may bepreferable to accept a lower fxn, and a higher effluentnitrate concentration - a reduction of f,.n from 0,50 atRs^30 d lo ^ = 0,40 at R5-22 d results in a 30% sav-ing in process volume and an increase in effluent nitrateconcentration of 6 mgN/i (for an influent COD of600 mg/ i and TKN/COD ratio of 0,10). The aboveargument applies also to favourable fjrrn2o valuesbecause, depending on the effluent quality re-quirements, it may be acceptable to design a 15-daysludge age process with an effluent nitrate concentra-tion of 5 mg!\i/i rather than a 20-day process which canachieve complete denitrification {depending on theTKN/COD ratio). Also, for instances where both the

' The readily biodegradable COD fraction f,s (or fbs) or the actual concentration SMI cannot be measured directly, its measurement requires theoperation and monitoring of a single completely mixed activated sludge process at a very short sludge age under daily cyclic square waveloading conditions (see Appendix 2).

6-18

Mnm20 a n d TKN/COD ratio are unfavourable, a 20-daysludge age process with external energy source addi-tion, to augment denitrification, may be a moreeconomical process than a 35-day process. Clearly, thedecision on the unaerated sludge mass fraction andsludge age of the process should not be taken too tight-ly, because with these two parameters fixed, the Nremoval of the process with the influent and self-generated energy sources is fixed, so that with fixed in-fluent wastewater characteristics, the effluent qualitywill be fixed /see Chapter 3).

For the wastewater characteristics of the designexample i.e. T= 14DC and / i n m 2 0 = 0,36/d (see Tables 4.3and 5.2) and accepting a factor of safety on nitrificationS, = 1,25, a 25-day sludge age process is required forfxm = 0,50 (see Chapter 5, Section 4.4 and Figure 6.14).As this sludge age is not excessively long for fxm = 0,50,a 25-day sludge age process will be accepted for apreliminary design analysis.

7.4 Nitrif ication

Because a factor of safety of 1,25 was imposed in thedesign of the unaerated siudge mass fraction andsludge age, the minimum sludge age for nitrification issufficiently below the actual sludge age to assure thatcomplete nitrification will take place i.e. the effluentTKN concentration can be assumed to be 3 mgN/f.(see Chapter 5, Section 4,4), This can be checked byrepeating the nitrification calculations given in Chapter5 for 25 days sludge age.

For the raw wastewater characteristics (i.e.fup=0,13 mgCOD/mgCOD, fu s-0,05 mgCOD/mgCOD,

N£ - 600.0,10 [

Tmin=14°C, S,,-600 mgCOD/f - see Tables 4.3 and5.2) and 25 days sludge age, and accepting the nitrogencontent of the volatile solids fn to be 0,10 mgN/rngVSS,the nitrogen required for sludge production is given byEq. 4.23 i.e.

0,45(1-0,05-0,13)(1+0,20.25}

11 +0,2.0,20.25) •+ ^

- 600.0,021 - 12,6 mgN/f.

The nitrification capacity (Nc) is found from Eq. (5.29)i.e. for the raw wastewater at 14°C

Nc - 4 8 - 12,6-3 - 32,4 mgN/ i .

The nitrification oxygen demand is found from Eq. 5.39i.e.

M(On> = 4,57.N, ;.Q

- 4,57.32,4. 13,33. 106 mgO/d

= 1973 kgO/d.

The above calculations for Ns, Nt; and M(OM), in-cluding those for the raw wastewater at 22°C and thesettled wastewater at 14°C and 22°C, are tabulated inTable 6.3.

In the design, because it is intended to reduce thenitrate concentration as much as possible, the Alkalinitychange in the wastewater will be minimized: Assumingthat 80% of the nitrate formed is denitrified, the alkalini-ty change AAlk = 7,14. N r -3 ,57. (Nitrate denitrified) =-7,14.32,4 + 3,57.0,80.32,4= -139 mg/ i as CaCO3.

TABLE 6.3 SUMMARY OF DESIGN CALCULATIONS FOROR SETTLED WASTEWATER AT

Parameter

Temperature

Safely factorMax. unaerated sludge massEffluent TKNN for sludge productionNitrification capacityNitrification oxygen demandMax. anoxic sludge massMax denitrification potentialDO in a-recydeDO in s-recycleUnderflow recycle ratioMin. Primary anoxicMin. Secondary anoxicOptimum a-recycle ratioEffluent nitratePrimary anoxicSecondary anoxicTotal anoxic

THE BARDENPHO AND MLE PROCESSES TREATING25 DAYS SLUDGE AGE. WASTEWATER CHARACTERISTICS GIVEN IN TABLES

Symbol

T

SftxmN»N t

Nc

M!0n)>xdmDpP

o,os5

**imin

fii3min

a0

No.fxi'«3fxttt

AND 6.1

Units

°C

mgN/imgN/imgN/ikgO/d

mgN/imgO/imgN/i

mgN/i

Raw WastewaterBardenpho MLE

14

1,250.503,012,632,419730,5043,12,01,01,00,0620,0243,30,00,160,240,40

22

2,70,502,012,034,020710,5058,72,01,01,00,0170,0235,00,00,160,190,35

'These are the theoretical results from Eqs. 6.30 and 6.31. Owing to an economicatheoretical a0 values should not be mplemented in practice, 'see text!.

14

1,250.503,012,632,419730,5043,12,01,01,0——17,2-1.7*0,50-0,50

22

2,70,502,012,034,020710,5058.72,01,01,0——36,0'0,9*0,50—0,50

RAW4.3. 5.2

Settled WastewaterBardenpho MLE

14

1,250,503,05,732,319610,5031,12,01.01,00,0850,0392,24,90,300,200,50

22

2,70.502,05,433,620460,5041,22,01,01,00,0290,0365.01,70.300,200,50

14

1,250,503,05,732.319610,5031,12.01,01.0—_4,74,80,50—0,50

22

2,70,502,05,433,620460,5041,22,01,01,0——13,4'2,2*0,50—0,50

uppe' limit to the mixed liquor a-recycie ratio, these

6-19

With an influent Alk of 200 mg/ l as CaC03 (seeTable 5.2) the effluent Alk = 200- 139-61 mg/ i asCaCOj, which, from Figure 5.6, will maintain a pHabove 7, (see Chapter 5, Section 4.3).

7.5 Denitrif ication

Accepting a sludge age of 25 days, which allows a maxi-mum uneerated sludge mass fraction fKrT, of 0,50, thedenitrification behaviour of the Bardenpho and MLEprocesses will be demonstrated for the typical raw andsettled wastewaters at 14°C and 22°C.

In the calculations it will be assumed that there is adissolved oxygen concentration of 2 mgO-'i in the inter-connecting flows from the aerobic to the anoxic reac-tors (i.e. Os = 2 mgO/f) and 1 mgO/f in ihe underflowrecycle from the secondary settling tank (i.e. 0b =1 mgO/f). The underflow recycle ratio Is) is assumed tobe 1:1.*

For nitrogen removal processes, the maximumanoxic sludge mass fraction (fldrri) is equal to the maxi-mum unaerated sludge mass fraction (fkm) i.e.fydn-,~0,50 (see Eq. 6.23). The denitrification potentialof the primary anoxic reactor at 14r'C with fxl =fXdm =0,50 for the raw wastewater is found from Eq. 6.20 (seeEq. 6.25}

Dr)P - (1 -0,05-0,13) 600{0,24 (1-0,45.1,48)

2,860,0636 . 0,50 . 0,45 . 25/{1 + 0,20 . 25)}

= 492(0,028 + 0,0596)

= 13,8-29,3

= 43,1 mgN/ i .

The Dop values for the raw wastewater at 22°C andthose for the settled wastewater at 14°C and 22°C arelisted in Table 6.3.

The minimum primary and secondary anoxicsludge mass fractions (fxlmiP and fx3m;n} are found fromEq. 6.21 and Eq. 6.29 respectively. For example for theraw wastewater at 14°C,

0,24(1 -0 ,45. 1,481(1 +0,20.25)rlmin 2,86.0,241 .0,45.25

= 0,062 and(1 + 1)2,0(1 0,20.25)

x3min (1 - 0,05 - 0,13)600 . 0,45 .25. 0,0607 . 2,86

= 0,025

The values for the raw wastewater at 22°C andthose for the settled wastewater at 14CC and 22°C aregiven in Table 6.3.

At this stage of the design procedure, we have col-

lected all the information necessary to check whether ornot complete denitrification can be achieved i.e. steps 1to 11 of the step by step procedure have been com-pleted.

Continuing the procedure with step 12, a value forthe a-recycle ratio is selected and the effluent nitrateconcentration is calculated from Eq. 6.24. Selectinga = 3,3 then for the raw wastewater at 14°C,

32,4 1

Nr,e =2,86* 0,0607 2,86

-43,1

0,0636 0,0636{0,0607+ 0,0607

= -5,47 mgN/'f

Because Nne < 0,0, N re must be set equal to zero.Setting Nne = 0 and substituting into Eqs. 6.26 to 6.28yields the optimum primary (fx1), secondary {f^} andtotal (fxC1) anoxic sludge mass fractions just to achievecomplete denitrification with an a-recycle ratio of 3,3i.e.

f.-. =

32,4Q

12,86M

0,24.0,334.4922,86

492 0,45,25(1+0,20.25)

0,0636

- 0,16

32,4 2_ _ ,0,0636l3,3+1 + 1 2,86 J0,06070]

492

- 0,24

0,45.251 + 0,20.25

0,0636

= 0,40

The above calculation is repeated for differentvalues of a. The results of Nne, fxl, fx3 and fxd1 are shownplotted versus a in Figure 6.15a. Repeating the designprocedure at 22°C (steps 12 to 15) for the rawwastewater produces the results shown in Figure 6.15b(see also Table 6.3}.

Figure 6.15 shows that, for the raw wastewater,complete denitrification can be achieved at 74°C for anya-recycie ratio > 1,8 (Figure 6.15a). At 22GC (Figure6.15b), a minimum fxdt** for complete denitrification of0,35 occurs at a = 5,0. The fxl corresponding to a = 5,0is 0,16 and this fixes the design value of fxl for both14°C and 22°C. Now if fx3 is fixed at 22CC at trie valuecorresponding to f^ i.e. fx3 = 0,19 (Figure 6,15b), thenfxd, will be insufficient to achieve complete denitrifica-tion at 14°C; for a fixed f,1( fxdt at 14CC is greater thanixdl at 22CC (compare Figures 6.15a and b). Hence fx3 isfixed at l ^ C ax the value corresponding to fx1 = 0,16 i.e.^=0 ,24 (Figury 6.15a). Hence 1 ^ = 0,40 for completedenitrification at 14°C and 22°C.

With fBl and fl3 fixed for both 14°C and 22°C, theoptimum performance of the process is achieved whenit is operated at the a-recycie ratio corresponding to the

"This is usually fixed at a value such that satisfactory settling tank operation is obtained. Secondary settling tank behaviour is described inChapter 8.

**tf Tr.iax is less than 18°C, a clear minimum cannot be discerned (see Figure 6.15a at 14°C). In this event, any reasonable fKi can be selectedwith the proviso thai complete utilization or the readily biodegradable COD can be assured even with a significant decrease in K 2o (see Eqs.6.12 and 6.21 i. The existence of a minimum, and the temperature above which it becomes discernible, depends much on the cho-ice of theDO concentrations in the recycles, Oa and 0s.

6-20

RAW

0,50MAXIMUM AKOXIC SLUDQEMAEI ALLOWED t f 1

RAW WASTEWATER

0,5O

UKKIC »LUD«£ MAtflFOR C O H P L E T E DEHITWIfICATIO*

MIXED LIQUOR RECYCLE RATIO (oj

0 2 4 6 9 K)

MIXED LIQUOR RtCYCLE RATIO to)

FIGURE 6.15: Primary (fx1), secondary (f,,;i). total (f*dt! andmaximum allowable lf,drn| anoxic sludge mass fractions andeffluent nitrate concentrations fNne) versus mixed liquora-recycie ratio for the Bardenpho process treating raw waste-water at 25 days sludge age at 14°C (left. Figure 6.15a) and 22°C

(right. Figure 6.15b).

SETTLED EWATER

O

<rt-XUJ

6S

z2 UJ

0,50

0,40

0 3 0

0.20

0,10

0,00

MAXIMUM ANOX IC SLUDGEMASS ALLOWED (f ,dm)

N I X E D L I Q U O R R E C Y C L E R A T I O ( a )

0 2 4 6MIXED LIQUOR RECYCLE

S 10RATIO (o)

FIGURE 6.76: Primary (f*i), secondary [fx3), total (f.dt) andmaximum allowable If.dm) anoxic sludge mass fractions andeffluent nitrate concentration (Nne) for the Bardenpho processtreating settled wastewater at 25 days sludge age at 14°C (left,

Figure 6.16a) and 22CC (right, Figure 6.15b).

6-21

fixed fKl value, this a-recycle ratio being the optimumao; at 14°C, ao = 3,3 (Figure 6.15a) and for 22"C,80 = 5,0 (Figure 6.15b}. This demonstrates that as thetemperature increases, the a-recycle ratio needs to beincreased to maintain optimum performance. This isbecause with increase in temperature, the denitrifica-tion potential of the primary anoxic reactor (Dp1) in-creases and consequently more nitrate must be recycledto it to fully utilize its DP,. It should be noted that withfxl and f,3 fixed, the use of any a-recycle ratio otherthan a0 results in poorer denitrification performance: ifa < ao, Dp1 is not fully utilized and if a > a0, unnecesari-ly large quantities of DO are discharged to the primaryanoxic reactor.

The design parameters calculated above are sum-marized in Table 6.3.

The design procedure for the settled waste water at74°C and 22°C was repeated and the results are shownplotted in Figures 6.16a and b respectively. At 22°C(Figure 6.16b), a minimum effluent nitrate concentra-tion* of 1,7 mgN/i is obtained with a = 5,0. The ^cor -responding to a = 5,0 is 0,30 and this value fixes f(1.Now because the maximum anoxic sludge mass frac-tion is fully utilized i.e. f>d, =1*^, fX3 is fixed at the dif-ference between 1>dm and f(1 i.e. fx3 = 0,20. Hence fxl

and fy3 are fixed at 0,30 and 0,20 respectively for 14°Cand 22°C. The optimum a-recycle ratio is 5,0 at 22°Cyielding an effluent nitrate concentration of 1,7 mgN//and that at 14°C is 2,2 (i.e. the value corresponding tofKl = 0,30, Figure 6,16a) yielding N, ie-4,9 mgN/jf. Thedesign results are listed in Table 6.3.

The design calculations above show that with theraw wastewater, complete denitrification can be obtain-ed and is achieved with a total anoxic sludge mass frac-tion (fxdl) of 0,40 whereas 0,50 is allowed (fxdm}. As Nremoval only is important in this example, the Barden-pho process is appropriate and (i) if a factor of safety forcomplete denitrification is required, option 1 (see Sec-tion 6.2 above) can be taken, i.e. fxl and f>3 are increas-ed such that f^i - fx3 ~ fxdl = fxm = 0,50; or (ii) if no factorof safety for complete denitrification is required, option2 can be taken, i.e. the sludge age can be reduced suchthat fkm = fxOl = 0,40.

For the settled wastewater, complete denitrifica-tion cannot be achieved at both 14°C and 22°C.Because the maximum anoxic sludge mass fraction isfully utilized (i.e. fyc!. = fxdr,., = 0,50), the denitrification ef-ficiency can be improved only by increasing the sludgeage of the process. However, the increase In sludge ageis at the expense of an increased process volume. Asluge age of 40 d, which allows f*m = fxdm = 0,60,** is re-quired to produce complete denitrification at 14CC(found by trial and error using design procedure). FromEq. 4.21 an increase in sludge age from 25 to 40 daysrepresents an increase in process volume of about 40%.

Such a large increase in volume may not be merited foran improvement in effluent nitrate concentration from 5to 0 mgN/jf. Consequently a sludge age of 25 to 30 dis considered the maximum economical sludge agebeyond which there is little cost effective improvementin denitrification efficiency. If complete denitrificationmust be achieved, it may be preferable to treat rawwastewater at 20-day sludge age which allows anfxdrr, = f»m = 0,40 (see Figure 6.15a). Although this in-creases the process volume by 80% it removes the pro-blem of primary sludge disposal. Alternatively, whentreating settled wastewater, and complete denitrifica-tion must be achieved, a 20-day sludge age processwith readily biodegradable COD addition to the primaryor secondary anoxic reactor may be aconomically viable- i f 60*** mgCOD'I influent readily biodegradableCOD is added to the plant, complete denitrification willbe obtained with settled wastewater at 14°C and 20-daysludge age. For the design flow of 13,3 M i /d , the CODaddition is 60.13,3 = 800 kg/d i.e. 10% of the rawwastewater COD load.

If complete denitrification is not essential, theBardenpho process as designed for the settled waste-water i.e. sludge age = 25 d, is appropriate. Alternative-ly, the feasibility of an MLE proces can be checked tosee whether or not this process may yield a moreeconomical design. The design procedure for the MLEprocess is demonstrated below.

The design procedure for the MLE process is iden-tical to that for the Bardenpho process up to step 9 ofthe step by step procedure. For the same design condi-tions of the Bardenpho process (i.e. Rb = 25 d etc), theresults for the MLE process are listed in Table 6.3 andcan be seen to be identical to those of the Bardenphoprocess. Substituting the values for the nitrificationcapacity Nc and denitrification potential Dpp into Eqs.6.30 and 6.31, the optimum mixed liquor a-recycle ratioao and minimum effluent nitrate concentration Nne areobtained; e.g. for the settled wastewater at 14°C

A = 2/2,86 - 0,70

B - 32,3-31,1 + 1 1.1}/2,86 = +2,95

C = (1 + 1)(31,1 - 1.1/2,86) - 1.32,3 = +29,2

Hence

aQ = {-2,95 + VT9E 4.0,70.29,2}/{2.0,70)

- 4,7 and

Nne = 32,3/(4,7 + 1 + 1) - 4 ,8mgN/ l .

The above resutts, as well as those for the settledwastewater at 22°C and those for the raw wastewaterat 14°C are listed in Table 6.3.

•If Tn,av is less than 18CC, a clear minimum cannot be discerned (see Figure 6.16b at *4CC). In this event, any reasonable fxi can be selectedwith the proviso thai complete utilization of the readily biodegradable COD can be assured even with a significant decrease in K- o 'see Eqs.6.12 and 6.21). The existence of a minimum, and the temperature above which it becomes discernible, depends much on the choice of theDO concentrations in the recycles, Oa and Os.

"This is the recommended upper limit for the maximum unaerated sludge mass fraction (see Chapter 5, Section 4.4).

•**Assuming that Nn« = 7,0 mgN// at RE = 20 d (i.e. 2 mgN.'/ greater than at R£= 25 d), the COD consumption for its removal is 7,0 8,6 = 60mgCOD I influent.

6-22

The results show that for all cases except one (i.e.for settled wastewater at 14°C), the optimum mixedliquor a-recycle ratio a0 exceeds 5. Although theseresults include the discharge of dissolved oxygen to theanoxic reactor, recycle ratios above 4 to 5 are not costeffective - the small decrease in effluent nitrate con-centration (Nne) which an increase in a-recycle ratiofrom 5 to 10 produces does not warrant Ihe additionalpumping costs (e.g. for N, = 33, Nr,e = 4,7 for a - 5 and2,7 for a =10). Consequently, the a-recycle ratio islimited at 5, which results in a minimum Nne from theMLE process of about 5 to 6 mgN/i for most municipalwastewaters. Hence for N removal processes, if effluentnitrate concentrations lower than 5 to 6 rng, N/i are re-quired, the Bardenpho process is the indicated processbecause this process can achieve low Nnt with lowa-recycle ratios.

Comparing the Bardenpho and MLE processes,(see Table 6,3) if the effluent quality standards requirethat complete or near complete denitrification must beproduced, then the Bardenpho process is the moresuperior process for both the raw and settled waste-waters. However, the process needs to be operated at asludge age of about 20 days for the raw wastewater(yielding complete denitrificationl and at least 30 daysfor the settled wastewater (yielding an effluent nitrateconcentration of about 3 mgN/i) - the settled waste-water process wiil have a smaller volume than the rawwastewater process but will have the additional problemof primary sludge disposal. Alternatively, if completedenitrification is not necessary and an effluent nitrateconcentration of 5 to 6 mgN/i is acceptable, the MLEprocess will be an acceptable process. For the rawwastewater, a sludge age of only 15 days (yieldingKm= f*dm~ 0,32) is sufficient to produce an effluentnitrate concentration of about 5 mgN/f with the MLEprocess and for the settled wastewater, a sludge age of25 days will yield an effluent nitrate concentration ofabout 5 mgN/i .

7.6 Process volume and oxygen demand

7.6. 1 Process volume

Having determined the subdivision of the sludge massinto anoxic and aerobic fractions to achieve the requirednitrogen removal, the actual sludge mass in the processneeds to be calculated to determine the volumes of thedifferent reactors. The mass of total IMLSS) or volatile(MLVSS) sludge in the activated sludge process for aselected sludge age and wastewater characteristics isgiven by Eq. 4.21 (see Chapter 4, Section 61 i.e. for thegiven raw and settled wastewater characteristics, theMLSS sludge mass in the process at 25-day sludge ageand 14°C is 54 200 and 21 850 kgTSS respectively.Selecting an MLSS concentration in the process of4 000 mg/ i * 'i.e. 4 kg, m3), the volume of the processtreating raw wastewater is 13 550 m3 and that treatingsettled wastewater is 5 460 m3. Now, because the

sludge mass in the nitrogen removal processes isuniformly distributed in the process, i.e. each reactorhas the same MLSS concentration, the volume frac-tions of the reactors are equal to the sludge mass frac-tions. Accepting the Bardenpho process design for theraw and settled wastewaters (see Table 6.3) the volumeof the reactors are found directly from the sludge massfraction and total process volumes; reactor retentiontimes, nominal and actual are found from the reactorvolumes and the nominal and total flows passingthrough them - see Table 6.4. Note that the reactornominal retention time is a consequence of the mass ofsludge generated from the influent COD load, theselected MLSS concentration and the sludge mass frac-tion - the retention time per se has no significance inkinetics of nitrification and denitrification (see Chapter4, Sections 6 and 12).

TABLE 6.4 REACTOR VOLUMES. NOMINAL ANDACTUAL RETENTION TIMES FOR THE

PROCESS TREATING THE TYPICAL RAV\WASTEWATER AT 25-DAY SLUGE AGE.

PERFORMANCE SEE TABLE

Process Parameter Symbol

Sludge mass M X,MLSS concentration X,Total volume Vp

Total flow 0Total nominal retentiontime Rnt

Mixed liquor recycle(14CC) aUnd. recycle sPrimary anoxic

Mass fraction \f)VolumeNom. ret. timeActual ret. time

Main aerationMass fraction*VolumeNom. ret. timeActual ret. time

Secondary anoxicMass fraction f,3VolumeNom. ret. timeActual ret. time

ReaerationMass fraction*VolumeNom. ret. timeActual ret. time

•The reaeration reactor is usually

RawWaste-water

54 2004 000

13 55013 330

24,4

3,31,0

0,202 710

4,91,13

0,456 100

17,02,6

0,304 060

7,33,7

0,056801.20,6

BARDENPHOAND SETTLED

FOR PROCESS6.3

SettledWaste-water

21 85040005 460

13 330

9,8

2,21,0

0,301 640

2,90,92

0,452 460

4,41,4

0,201 090

2,01,0

0,05270

0,500,25

Units

kgTSSmg, /m3

mJ/d

h

m3

hh

m3

hh

m3

hh

m3

hh

taken to have a sludqe massfraction of around 0,05 to 0,07 and this fraction is deducted fromthe aerobic sludge mass fraction;mass fraction constitutes the main

the remaining aerobic sludgeaeration sludue mass fraction.

'A method lor estimating the reactor MLSS concentration for design conditions is as follows: fora selected X, calculate the process volume re-quirements. For accepted settling characteristics of the sludge, calculate the required settler surface area for the selected Xt (see Chapter 8).Estimate the combined cost of the process and sertling tanks. It will be found that as X, increases the cost of the process decreases but thecost of the settling tanks increases. The X! that gives the minimum combined cost is the indicated MLSS concentration (usually between 3 000and 6 000 mgTSS/ with settled wastewater being lower than raw wastewater).

6-23

7.6.2 Daily average total oxygen demand

The total oxygen demand in a nitrogen removal processis the sum of that required for carbonaceous material(COD) degradation and nitrification, less than recoveredby denitrification. The daily average oxygen demand for(i) carbonaceous material removal (M0c) is given by Eq.4.15 (see Chapter 4, Section 7), and (ii) nitrification isgiven by Eq. 5.39 (see Section 7.4 above and Chapter 5,Section 6.5) and these oxygen demands in the Barden-pho process at 25-day sludge age for the typical raw andsettled wastewaters at 14°C and 22CC are listed in Table6.5.

TABLE 6.5 DAILY AVERAGE OXYGEN DEMAND (INkgO-'d) IN THE 25 DAY SLUDGE AGE BARDENPHOPROCESS FOR THE TYPICAL RAW AND SETTLED

WASTEWATERS AT 14CC AND 22°C

Oxygen DemandkgO/d

1. Carbonaceous*2. Nitrification"3. DeniTrification"*4. Total

Raw Settled14"C 22

M(OC) -5105 -5205 -3287 -3352M(O,,I +1973 + 2071 -1961 +2046M(Ob) -1235 -1296 -1048 -1216M(O.) + 5843 + 5980 + 4200 *4185

-See Chapter 4, Section 12.3, Eq. 4.27"See Section 7.4 above

- • * From Eq. 6.32 and Tabie 6.3

The oxygen recovered by denitrification (M0d) isgiven by 2,86 times the nitrate mass denitrified (seeSection 1.3) where nitrate mass denitrified is the pro-duct of the daily average influent flow Q and the nitrateconcentration denitrified. The nitrate concentrationdenitrified is given by the difference in the nitrificationcapacity Nr and the effluent nitrate concentration:Hence

M(OH) - 2,86.(Nr-NLf.)Q (6.32)

From the denitrification performance data of theBardenpho process in Table 6.3, the oxygen recoveredby denitrification for the typical raw and settledwastewaters are listed in Table 6,5.

For the raw wastewater, Table 6.5 shows that (i)the nitrification oxygen demand (MOn) is about 40%that required for COD removal (MOC), (ii) about 60% ofMOn can be recovered by incorporating denitrification,(iii) the additional oxygen demand by incorporatingnitrification and denitrification is only 15% of that re-quired for COD removal only and liv) the effect oftemperature on the total oxygen demand is marginal -less than 3% (see also Figure 6.1).

For the settled wastewater, Table 6.5 shows that (i)the nitrification oxygen demand is about 60% of that re-quired for COD removal, (ii! about 50% of the nitrifica-tion oxygen demand can be recovered by denitrifica-tion, {iii) the additional oxygen demand by incorporatingnitrification and denitrification is about 30% of that re-quired for COD removal only and (iv) the effect of

temperature on the total oxygen demand is marginal -less than 3% more at the lower temperature.

Comparing the oxygen demand for the raw andsettled wastewaters, it can be seen that the totaloxygen demand for the latter is about 30% less thanthat of the former. This saving is possible becauseprimary sedimentation removes 35 to 45% of the rawwastewater COD (see Chapter 4, Section 12 andChapter 5, Section 6.5). Furthermore, for the settledwastewater, the nitrification oxygen demand is agreater proportion of the total, and also, less of thenitrification oxygen demand can be recovered bydenitrification compared to the raw wastewater. Theseeffects are due to the higher TKN/COD ratio of the set-tled wastewater.

Generally for design, preliminary estimates of thetotal daily average oxygen demand can be roughly ap-proximated with the following:

(i) including complete nitrification;

M(O,) = {1 + 5fns}M(Oc) (6.33)

(ii) including complete nitrification, and denitrificationto the degree possible depending on the TKN/CODratio (see Figure 6.17)

M(O,) = {1+2fns}M(Oc) (6.34a)

for fns < 0,09 (i.e. effluent nitrate< 2 mgN/i)

- {1 + 2fns [1 + 10(fns-0,08)]}M(Oc) (6.34b)

for fns > 0,09 (i.e. effluent nitrate> 2 mgN/i)

where fns = influent TKN/COD ratio.

These equations, in conjunction with the carbonaceousoxygen demand M(OC) equation in terms of the influentCOD load (MS,,) (see Chapter 4, Section 7) will allowthe calculation of preliminary estimates of the dailyaverage total oxygen demand in nitrogen removal pro-cesses between sludge ages of 15 to 30 d for raw andsettled wastewaters for known influent COD loads andTKN/COD ratios.

Knowing the average daily total oxygen demand,the peak total oxygen demand can be roughly estimatedby means of a simple design rule. From a large numberof simulations with the genera! dynamic model, it wasfound that provided the factor of safety on nitrification(Sf) is greater than 1,25 to 1,35, the relative amplitude{i.e. (Peak-Average)/Average} of the total oxygen de-mand is a fraction 0,50 of the relative amplitude of thetotal oxygen demand potential of the influent COD andTKN load {i.e. Q{S1i + 4,57.Nli)f. For example, with theraw wastewater design, if the peak influent total oxygendemand potential is obtained at a time of day when theinfluent flow rate, COD and TKN concentrations are23,1 Mf /d , 863 mgCOD-'I and 65 mgN/f respectively- i . e . 23,1 (863 + 4,57.65) = 26 800 kgO/d - and theaverage total oxygen demand potential is 13,33(600+4,57.48= 10 920 kgO.'d (see Table 4.3), therelative amplitude of the total influent oxygen demandpotential is (26 800-10 9201/10 920=1,45; hence therelative amplitude of the total oxygen demand is ap-proximately 0,50.1,45-0,72; from Table 6.5 the

6-24

average daily total oxygen demand is 5 980 kgO/d (at22°CJ and hence the peak oxygen demand is1,72.5 980=10 300 kgO/d. As with all simplified designrules, the above rule should be used with discretion andcaution, and where possible, the peak total oxygen de-mand is best estimated by means of the generaldynamic activated sludge model.

7.7 Closure

In the above design example, the raw wastewaterTKN/COD ratio is 0,08 mgN/mgCOD. If the calcula-tions are repeated for the raw wastewater but assumingdifferent TKN/COD ratios and sludge ages, it will befound that (i) for a fixed sludge age, once theTKN/COD ratio is above that with which completedenitrification can be achieved, the effluent nitrate con-centration increases as the TKN/COD ratio increasesand (2) at a fixed TKfsi/COD ratio above that with whichcomplete denitrification can be achieved, as the sludgeage increases, the effluent nitrate concentrationdecreases. This implies lhat at a fixed sludge age, the Nremoval potential of the process is fixed so that the ef-fluent quality depends on the magnitude of the influentTKN/COD ratio; however, the N removal potential ofthe process may be improved by increasing the sludgeage.

Based on 14°C and the raw wastewatercharacteristics of the design example (except theTKN/COD ratio) the behaviour described above isshown graphically in Figure 6.17, in which the influentTKN/COD ratio is plotted versus the effluent nitrateconcentration that can be achived at different sludgeages. The process volume required at the differentsludge ages (as a fraction of that required at 25 d) is alsogiven on Figure 6.17. The areas for which the Barden-pho and MLE process achieve the optimum N removalare demarcated and generally where the effluent nitrateconcentration is greater than 5 to 6 mgN/\f, the MLEprocess will yield an improved efficient nitrate concen-tration at an economical mixed liquor a recycle ratio i.e.< 4:1. It should be noted that at a fixed TKN/COD ratio,the total average daily oxygen demand does not varysignificantly between different sludge ages so that aera-tion costs do not significantly affect the conclusionsthat can be drawn from Figure 6.17 (see Figure 6.1).

Figure 6.17 shows that for a TKN/COD ratio of say0,095 mgN, mgCOD, complete denitrification can beachieved if the sludge age is 30 days in a Bardenphoprocess. However, if an effluent quality of 7 mgN/f isacceptable, then this can be achieved in an MLE pro-cess at a sluge age of 20 days. The process volume sav-ing that can be made by reducing the effluent nitratestandard from 0 to 6 mgN/f is 1 -0,83/1,16-0,28 i.e.

Oou

zE

T =0,14 ( f

0,12

O.IO

troo

2

ui

O.O8

0,06

0,04

0,02

0,00

-, fbs1 r£5-

0 . 2 4 i Pnmzo = 0 , 3 6 / d , Sf =1,25—i—-i 1—^—i i 1—n r~

- Conditions for whichBardenpho processis indicated

as o fraction ofthat required at25 days sludgeog«)

Conditions for whichMLE process isindicated

0 5 10 15EFFLUENT NITRATE CONCENTRATION {mgN/ l )

FIGURE 6.17: Effluent nitrate concentration attainable for dif-ferent influent TKN/COD ratios for N removal processesoperating at different sludge ages for T=14CC, readilybiodegradable COD fraction |fbb) = 0,24 and maximum specific

growth rate of the nitrifiers at 20c'C (^nm2o) = 0.36/d.

6-25

the MLE process at 20-day sludge age with 7 mglSI/f isabout 30% smaller than the Bardenpho process at30-day sludge age with complete denitrification. Figure6.17 shows also that for low ^nn-20 values, 14lJC and25-day sludge age (to avoid excessively large processvolumes), the Bardenpho process can achieve (i) com-plete denitrification for TKN/COD ratios less than 0,090mgN/mgCOD; (ii} near complete denitrification (i.e. ef-fluent nitrate concentration between 1 and 5 mgN/i)for TKN/COD ratios between 0,09 and 0,10 mgN/mgCOD and (iiii incomplete denitrification (i.e. effluentnitrate concentration above 5 mgN/i) for TKN/CODratios above 0,10 mgN/mgCOD; hence the MLE pro-cess is the indicated process for TKN/COD ratios above0,10. The above limits are relatively sensitive to theHnm2G value and readily biodegradable COD fraction; (i)the limits are given for a low ^nrri2o value (around 0,36/d)because this will be the usual case in design unless suffi-cient information is available justifying the use of ahigher one (see Appendix 3) - if iJnrn20 's greater than0,40/d, the results will be similar to those shown inFigure 6.17 except that they can be achieved at shortersludge ages; (ii) the limits given are for an approximatelynormal raw wastewater readily biodegradable COD frac-tion fbs of 0,24 - there are many factors which canaffect the fbs fraction of a wastewater (see Section 7.1.3above) so that at the design stage, the value of fbs alsowill be uncertain [unless the value has been measured(see Appendix 2)] and it is likely that a value of 0,24 willbe used initially for want of a better estimate; if theactual fb6 value is greater than 0,24, lower effluentnitrate concentrations will be obtained than those givenin Figure 6.17 and if fbs is lower than 0,24, higher ef-fluent nitrate concentrations will be obtained.

Although the limits given in Figure 6.17 should notbe taken as general because of their dependence on thewastewater characteristics, in particular ^nrri2o and fbs,they are useful for setting approximate guidelines for Nremoval process selection and the sludge age at which itis to be operated to achieve a certain effluent nitrateconcentration for a given influent TKN/COD ratio. Theguidelines in Figure 6.17 are not intended to obviate do-ing detailed design calculations but serve the purpose ofdemonstrating the behaviour of N removal processes,indeed it is recommended that in process design, thewastewater characteristics are changed to encompass arange of possible values for the particular wastewaterand to undertake design calculation for combinations ofdifferent wastewater characteristics. This will allow theselection of a process design that can best deal with an-ticipated changes in wastewater characteristics and will

expose weaknesses in the design so that controlmeasures can be incorporated to deal with fluctuationsin wastewater characteristics during the life of the plant.

8. REFERENCES

ARKLEY, M. and MARAIS, G.v.R. (1982) Effect of the anoxic sludgemass fraction on the activated sludge process. Research ReportNo. W 38, Dept. of Civil Eng., Univ. of Cape Town.

BARNARD, J.L. (1973) Biological denitrification, Wat. Polfut. Con-trol. 72, 6, 705-720.

CARLSON D.E. (1971) Nitrogen removal and identification for waterquality control, OWRR. Project No. 4040, Dept. of Crvil Eng.,Univ. of Washington.

CHANG, J.P. and MORRIS, J.E. (1962) Studies of the utilization ofnitrate by Miaococcus Denitrificans, J. Gen. Mtcrobia/., 29, 301.

CHR1STENSEN, M.H. and HARREMOES P. (1977! Biologicaldenitrification of sewage: A literature review. Prop. Wat. Tech.. 8,415. 509-555.

DOLD, P.L., EKAMA G.A. and MARAIS G.v.R. (1980) A generalmodel for the activated sludge process. Prog. Wat. Tech.. 12,47-77.

EKAMA, G.A. and MARAIS, G.v.R. [1978) The dynamic behaviour ofthe activated sludge process. Research Report No. W 27, Dept. ofCivil Eng., Univ. of Cape Town.

EKAMA, G.A., VAN HAANDEL, A.C. and MARAIS, G.v.R. (1979)The present status of research on nitrogen removal: a model forthe modified activated siudge process. Presented at the Sym-posium on Nutrient Removal, SA Branch of IWPC Pretoria, May1979. Proceedings published by Water Research Commission ofSouth Africa.

LUDZACK, F.J.andETTINGER, M.B. (1962) Controlling operation tominimize activated sludge effluent nitrogen JWPCF. 34. 920-931.

MARSDEN, M.G. and MARAIS, G.v.R. (1977) The role of theprimary anoxic reaior in denitrification and biological phosphorusremoval. Research Report No. W 19, Dept. of Civil Eng., Univ. ofCape Town.

SIEBRITZ, I.P.. EKAMA, G.A. and MARAIS, G.v.R. (1980) Excessbiological phosphorus removal in the activated sludge process atwarm temperature climates. Proc. Waste Treatment and Utiliza-tion, Vol. 2, 233-251. Eds. C.W. Robinson. M. Moo-Young andG.J. Farquhar, Pergamon Press, Toronto.

STERN, L.B. and MARAIS, G.v.R. (1974) Sewage as tha electrondonor in biological denitrification. Research Report Nlo. W 7,Dept. of Civil Eng., Univ. of Cape Town.

VAN HAANDEL, AC . EKAMA. G.A. and MARAIS, G.v.R. (1981)The activated sludge process part 3 - Single sludge denitrifica-tion. Water Research, 15, 1135-1152.

VAN HAANDEL, A.C. and MARAIS G.v.R. (1981) Nitrification anddenitrification kinetics in the activated sludge process. ResearchReport No, W 39, Dept. of Civil Eng., Univ. of Cape Town.

VAN HAANDEL, A .C , DOLD, P.L. and MARAIS, G.v.R. (1982) Op-timization of nitiogen removal in the single sludge activated sludgeprocess. Wat. Sci. Tech., 14, 443.

WILSON, D.E. and MARAIS. G.v.R. (1976} Adsorption phase inbiological denitrification. Research Report No. W 11, Dept. of CivilEng., Univ. of Cape Town.

WUHRMANN, K. 11964) Nitrogen removaf in sewage treatment pro-cesses. Vehr. Int. Vcr. Theor. Angew. Limnol, 15'in German).

6-26

CHAPTER SEVEN

BIOLOGICAL EXCESS PHOSPHORUS REMOVAL

by

G.A. Ekama, G.v.R. Marais and I.P. Siebritz

1. INTRODUCTION

From the first publication reporting phosphorus removalin excess of normal metabolic requirements in some ac-tivated sludge plants there has been controversy as tothe mechanism whereby the excess P removal is ac-complished, whether the mechanism is a precipitationof Inorganic compounds, albeit biologically mediated,or biological through metabolic formation and ac-cumulation of phosphorus compounds in or on theorganisms. The objective of this chapter is not todiscuss the evidence that supports the excess biologicalP removal hypothesis,* but to describe briefly thetheory of biological excess P removal as understood bythe authors and to demonstrate how this theory can beused as an aid for the design of biological P removal ac-tivated sludge processes. This does not imply thatprecipitation of inorganic phosphorus salts due tochemical changes resulting from biological action e.g.alkalinity, acidity gnd pH does not take place:** Suchinorganic precipitation certainly can take place, but itwould appear that in the treatment of municipalwastewaters by an appropriately designed activatedsludge process, within the normal ranges of pH, alkalini-ty, acidity and calcium concentrations in the influent,excess P removal is principally mediated by a biologicalmechanism.

2. BACKGROUND

2.1 Earlier developments

From the reports of the earlier investigators into excessbiological P removal (e.g. Funs and Min Chen, 1975 andBarnard, 1976), one conclusion emerged which now ap-pears to be generally accepted: excess P removal isstimulated by stressing the organisms by withholdingthe oxygen supply. Quantification of this stress,however, presented major problems and still does,because the parameters in terms of which the stress isto be formulated are not yet explicitly identified.

Barnard (1976) from pilot scale studies (Barnard,1975a, b) on the Bardenpho process (Figure 7.1)reported that excess biological P removal is induced if atsome point in the process configuration the organismmass is stressed by subjecting it to an "anaerobic"*'"*state (i.e. a state in which neither oxygen nor nitrate ispresent) such that phosphorus is released by the sludgemass to the bulk liquid. He proposed to produce thisstage efficiently by including an anaerobic reactorahead of the primary anoxic reactor in the Bardenphoprocess, this reactor receiving the influent flow and theunderflow recycle from the secondary settling tanks,(Figure 7.2). This configuration has become known inSouth Africa as the 5-stage Phoredox or simply thePhoredox process, and in the United States as theModified Bardenpho process.

To explain the excess removal phenomenon, Bar-nard in 1976 hypothesized that it is not the release persethat stimulates the excess uptake mechanism but thatthe release indicates that a certain low redox potentialhas been established, i.e. that the low redox potentialtriggers off the release and thereby stimulates excess Puptake. In terms of this hypothesis nitrate recycled viathe underflow to the anaerobic reactor will restrain, insome degree, the level to which the redox potential canbe lowered and consequently, nitrate can be expectedto influence excess P uptake adversely. Mo data onredox potential was reported. Barnard apparently ac-cepted that the Bardenpho section of the plant shouldreduce the nitrate sufficiently that any nitrate in theunderflow would not prevent the attainment of the lowredox potential necessary for P release in the anaerobicreactor. In any event he considered that nitrate enteringthe anaerobic reactor could be countered by increasingthe retention time of this reactor. For design of theanaerobic reactor he suggested a nominal retention timeof one hour, Barnard (1976).

Barnard's work on the nitrification-denitrificationexcess P removal process stimulated extensive researchinto this process, to gain experience on its behaviour, todelineate more precisely the factors influencing excessremoval and to develop criteria for design. Nicholls

'For a review of the observations that support the biological excess P removal hypothesis, see Marais, Loewenthal and Siebritz (1982).

'Points of view with regard to biological mediated precipitation of phosphorus are presented in the Post Conference Seminar on PhosphateRemoval in Biological Treatment Process, IAWPR, Pretoria 1982. Published in Water Science and Technology Vo! 15 No 3/4 1983.

'Anoxic; anaerobic: The meaning we will attach to these two terms, m nitrification denitrification processes, follows that originally used byBarnard, i.e. Anoxic: a slate in which nitrate is present but no oxygen; Anaerobic: a state in which neither nitrate nor oxygen is present. Theinadequacies of these definitions are apparent when anempting to compare the state of two reactors of the same size in a completely mixedand a plug flow reactor respective1*/• A completely mixed anaerobic reactor, for example, will have no nitrate in the reactor and effluent; theequivalent plug flow reactor however may contain nitrate for a considerable portion of the reactor length i.e. be partly "anoxic", partly"anaerobic" - the inadequacy arises in that no indication is given as to the intensity of the state.

7-1

PRIMARY AEROBIC SECONDARYANOXIC REACTOR ANOXICREACTOR RfACTOR

MIXED LIQUOR

ANOXICREACTOR

AEROBICREACTOR

WASTE FLOW

SETTLER

EFFL UENT


INFLUENT

WASTE FLOW

SETTLER

SLUDGE RECYCLE *

FIGURE 7.1: The Bardenpho process for biological nitrogenremoval-

FIGURE 7.4: The modified Ludzack-Ettinger process forbiological nitrogen removal.


MIXED UQUOR RECYCLE

ANAEROBICREACTOR

INFLUENT

AEROBIC SECONDARYREACTOR ANOXIC

REACTORREAERATIONREACTOR

0 • WASTE FLOW

ANAEROBIC ANOXIC AEROBICREACTOR REACTOR REACTOR

RECYCLE MIXED LIQUOR

SETTLER

SLUDGE RECYCLE •

INFLUEN

WASTE FLOW

SETTLER' C C ClEFFLUENT

FIGURE 7.2 The Photedox process for biological nitrogen andphosphorus removal, aiso called the Modified Bardenpho pro-

cess.

FIGURE 7.5: The UCT process (or biological nitrogen andphosphorus removal.

ANAEROBICREACTOR

ANOXIC AEROBICREACTOR REACTOR

MIXED LIQUOR

WASTE FLOW

ANAEROBIC ANOXIC AEROBICREACTOR REACTORS REACTOR

MIXED LIQUOR RECYCLES

EFFLUENTINFLUENT

WASTE FLOWSETTLER

EFFTJJENT

FIGURE 7.3: The 3 stage Phoredox process for biologicalnitrogen and phosphorus removal.

FIGURE 7.6: The modified UCT process for biological nitrogenand phosphorus removal.

7-2

(1975) at full scale, and McLaren and Wood (1976) andSimpkins and McLaren (1978) at laboratory and pilotscale applied the Phoredox process at approximately20°C to treat effluents from Johannesburg and Pretoria,and were successful in attaining excess P removal. Theywere also successful in obtaining excess P removal inthe 3-stage Phoredox (Figure 7.3) i.e. a Phoredox pro-cess without the secondary anoxic and reaeration reac-tors. Simpkins and McLaren found that (1) theanaerobic reactor was necessary for excess P removal,(2) nitrate in the recycle adversely affected P removaland (3) increasing the volume of the anaerobic reactorincreased the excess P removal. These findings were inconformity with Barnard's redox potential hypothesisalthough again no redox potential measurements werereported.

None of the investigations above provided areliable model to predict the magnitude of thedenitrification to be expected even though it was evi-dent that for design, evaluation of the nitrate in therecycle could be crucial in assessing the success of aprocess both in stimulation of the P release andmagnitude of the P uptake.

Marais and his group (Stern 1974, Martin 1975,Marsden 1976, Wilson 1976 with Marais) recognized theimportance of quantifying the nitrate removal. To ob-tain information on the magnitude and kinetics ofdenitrification they replaced the completely mixed reac-tors in the Bardenpho process by plug flow reactors andmeasured the nitrate along the reactor axes under con-stant flow and load conditions. At the same time theyalso monitored the phosphate behaviour in the variousreactors making up the processes. Their findings ondenitrification kinetics are reviewed in Chapter 6. Withregard to the phosphorus behaviour, with the rawsewage they used as influent, difficulties were ex-perienced in obtaining nitrification, even at long sludgeages of 10 to 20 days, if the unaerated mass fraction ofthe sludge exceeded 30 to 40%. When the plants wereoperated at unaerated mass fractions that allowednitrification the denitrification obtained was insufficientand the effluent nitrate was high. Yet these plants gavevery good phophorus removal. However when thesewage source was changed the good P removal pre-viously obtained declined virtually to zero.

In a subsequent study of the Phoredox process,Rabinowitz and Marais ! 1980} selected the 3-stagePhoredox (Figure 7.3) as the basic configuration inpreference to the 5-stage for the following reason: Thewastewater source did not allow an unaerated massfraction of greater than 40 per cent at 14°C for a sludgeage (Rs) of 20 days if efficient nitrification was to bemaintained; taking account of the fact that in thePhoredox process the anaerobic reactor cannot con-tribute to the process denitrification potential, the5-stage process could not reduce the nitrate to zero forthe measured TKN/COD ratio of the waste flow. Con-sequently (as discussed earlier in Chapter 6) the second-ary anoxic reactor volume was added to the primaryanoxic to obtain the maximum nitrate removal andhence the minimum nitrate concentration in the under-flow recycle. The findings from this investigation can besummarized as follows:(1) When the nitrate concentration in the effluent (and

underflow recycle) was low usually P release andexcess uptake were observed. In general there wasa tendency for the excess uptake to decrease quitedisproportionately as the nitrate in the recycle in-creased, a behaviour also noted by Simpkins andMcLaren, and Barnard.

(2) With different batches of wastewater having thesame nitrate concentration in the recycle, onebatch may give high P release and excess removalwhereas the next may give no (or little) release andlittle excess removal. No apparent reason for thisbehaviour could be discovered.

The overall P removal performance was disappoint-ing; not only did the plant not remove P in excess overlong periods of time but the removal was erratic due tothe effects of (1) and/or (2) above. Increasing theanaerobic mass fraction during periods of low P removalwas found to be counter-productive as this could bedone only at the expense of the anoxic mass fractionwhich in turn gave rise to increased nitrate in the re-cycle. It was finally concluded that for the waste flowsused in the experimental investigation, treatment by thePhoredox type process was not suitable for excess Premoval; this did not imply that the process might notbe suitable for other waste flows but the investigationdid bring to light that there were constraints, not ade-quate^ recognized before, that may prevent high Premovals:

(1) For any selected sludge age and minimumtemperature the requirement for efficient nitrifica-tion imposes an upper limit on the unaerated massfraction.

(2) The limitation on the unaerated mass fraction cor-respondingly limits the concentration of nitrate thatcan be removed. If the nitrate generated is higherthan the denitrification achievable, nitrate will ap-pear in the effluent and, in the Phoredox system,the P removal will be adversely affected.

2.2 Later developments

2.2. 1 Positive exclusion of nitrate from the anaerobicreactor

From the findings on the Phoredox process the Maraisgroup concluded that irrespective of other factors thatmay affect the excess P uptake, a major factor influenc-ing the uptake was the presence of nitrate in the under-flow recycle. If the nitrate content in the underflow tothe reactor could be kept at low concentration, thenthere was a high expectation that excess P removalwould be obtained. The principal obstacle to attainingthis desirable end in the Phoredox process appeared tobe that the nitrate discharged to the anaerobic reactor islinked directly to the concentration in the effluent. If forany reason the nitrate concentration increased while theCOD remained constant, i.e. if the influent TKN/CODratio increased, the process appeared to offer little op-tion to reduce this by operational means. The onlyoperational means available was to reduce the

7-3

magnitude of the underflow recycle but This was a risky (i)option as the settleability of the mixed liquor in theplants tended to be poorer than in pure aerobicsystems. Evidently a process configuration was neededthat made the anaerobic reactor independent of the ef-fluent nitrate concentration. Towards this end, after aseries of attempts, the configuration shown in Figure7.5 was devised called the University of Cape Town pro-cess (UCT process).

In the UCT process, the settling tank underflows-recycle as well as the mixed liquor a-recycle aredischarged to the anoxic reactor and an additional mix-ed liquor r-recycle from the anoxic to the anaerobicreactor is introduced. The nitrete recycled to the anoxicreactor can be controlled by appropriately adjusting themixed liquor a-recycle such that the nitrate concentra-tion in the outflow of the anoxic reactor remains ap-proximately zero. Consequently the mixed liquorr-recycle from the anoxic to the anaerobic reactor willcontain very little or no nitrate and the anaerobic condi-tion in the anaerobic reactor will be optimal. Thus, inthe UCT process by application of an appropriate opera-Tiona/ control strategy, the anaerobic reactor can bemaintained independent of the nitrate in the effluenteven if the influent TKN/COD ratio to the plant varies.

Laboratory scale tests on the UCT process usingwaste flows from Cape Town showed improved excessP removal in both the concentration removed and con-sistency in removal over those obtained in the Phoredoxprocess. But perhaps the most important achievementfrom a research point of view, was that with the UCTprocess it was possible to eliminate the confounding ef-fect of the nitrate in the recycle flow to the anaerobicreactor on excess P removal so that other factors in-fluencing the excess removal could be investigated withgreater ease. In the experimental response data the ef-fects of these other factors became clearly evident:

(1) For the same influent COD, one batch of sewagegave high removals, another gave low, an observa-tion previously surmised but not explicitly identifieddue to the difficulty of isolating this effect whennitrate was recycled to the anaerobic reactor.

(2) Generally as the storage time of a batch of sewage (ii(at 5°C) increased, both the excess P removal andthe nitrate removal declined. Evidently on storage aprogressive change was taking place in the con-stitution of the sewage which acted adversely onthe two phenomena; direct evidence of changewas indicated by the gradual reduction in COD ofthe batch during storage.

2.2.2 Phosphorus removal process analyses

Owing to the importance attached to the nitrate con-centration discharged to the anaerobic reactor in excessbioiogical P removal, the nitrogen removal theory wasapplied to the P removal processes to investigate theirpropensity for removing nitrogen.

The Phoredox Process: In this process, a certainfraction (fxa) of the total unaerated sludge massfraction (f(T) is "set aside" as an anaerobic reactorto establish the prerequisites for excess phosphorusremoval. Only the remaining unaerated sludgemass fraction (f,,d1 = \%x - fKS} is available as primaryand secondary anoxic reactors for nitrogen re-moval. If no nitrate is to be recycled to theanaerobic reactor (to ensure the most "intense"anaerobic condition) complete denitrification mustbe achieved in the anoxic sludge mass fraction(fxdl). The denitrification behaviour of the anoxicreactors in the Phoredox process for completedenitrification is the same as those in the Barden-pho process for complete denitrification. However,due to the presence of the anaerobic reactor (fxa),the anoxic sludge mass fraction (fxd,) for thePhoredox process <f>:ci, = f,, — fxaJ is less than thatfor the Bardenpho process (fxdT = fK,). Hence, theupper limit of the TKN/COD ratio for completedenitrification in the Phoredox process is lowerthan that for complete denitrification in the Barden-pho process, by about 0,005, for approximatelynormal municipal wasiewaters with fKa = 0,15 at14°C. Consequently, for an unsettled "normal"municipal wastewater, at 25 days sludge age and14°C, the 5 stage Phoredox process will achievecomplete denitrification only if the TKN/COD ratiois less than about 0,085.* Should the TKN/CODratio exceed this limit, complete denitrification isunlikely resulting in nitrate in the effluent andunderflow s-recycle. From a design point of view,to provide a factor of safety, the TKN/COD ratiolimit should not exceed 0,07* to 0,08' at 14°C, forsludge age 20 to 30 days, to have reasonableassurance that complete denitrification will be at-tained. The safe upper limit, of a 0,07 to 0,08TKN/COD ratio, for successful implementation ofthe Phoredox process, restricts application of theprocess in the treatment of municipal wastewatersbecause the TKN/COD ratio of raw wastewaterranges between 0,07* and 0,09* and that of settledwastewater generally is above 0,10*.

The UCT Process: The analysis of the UCT process(Figure 7.5) showed that for this process, there isalso an upper TKN/COD ratio above which excessP removal is unlikely to be attained. For a normalmunicipal wastewater at a TKN/COD of 0,14*(at 14°C and 25 days sludge age) the nitrate con-centration in the effluent and hence en theunderflow s-recycle is so high that this recycle byitself (al s= 1) fully loads the anoxic reactor to itsdenitrification potential i.e. the mixed liquora-recycle needs to be reduced to zero. Hence, forTKN'COD ratios > 0,14, nitrate will be present inthe primary anoxic reactor and a discharge ofnitrate to the anaerobic reactor cannot be avoidedleading to a decline in excess P removal. Fordesign, to introduce a factor of safety on the

'These limits apply to a readily biodegradable COD fraction f-s = 0,20 (fbs = 0,24). For higher (lower) fu fractions these limits are higher ilower). Itshould be noted that throughout this monograph, where TKN/COD ratio limits are given, these apply for fB = 0,20 <fM = 0,24) unless otherwisestated.

7-4

denitrificaiion, the upper limit for the TKN/CODratio above which excess P removal is unlikely to beattained in the UCT process is 0,12 to 0,14. Thislimit is above that for most settled and rawmunicipal wastewaters.

!iii) The Modified UCT Process: Experience with theUCT process exposed two different types of pro-blem that could significantly affect the successfuloperation of the process, these were (i) processcontrol and (ii) sludge settieability.

(a) The mixed iiquor a-recycle ratio needs to becarefully controlled so that the primary anoxic reac-tor is just underloaded with nitrate to avoid a nitratedischarge to the anaerobic reactor. Under full-scaleoperation such careful control of the a-recycle ratiois not possible due to uncertainty in the TKN/CODratio, particularly under cyclic flow and load condi-tions. To simplify the operation of the UCT processa modification was sought whereby careful controlof a-recycle would not be necessary-

(b) As the TKN/COD ratio increases, the a-recycleratio needs to be decreased to avoid a nitratedischarge to the anaerobic reactor, this in turncauses an increase in the actual anoxic retentiontime. For relatively high influent COD concentra-tions (> 500 mgCOD//), and TKN/COD ratios> 0,11, the actual anoxic retention time exceedsone hour. From experimental observation, in someof the units, when the actual anoxic retention timeexceeded 1 hour, the settieability of the sludgedeclined.* To preserve good settieability of thesludge, a modification to the UCT process wassought whereby the actual anoxic retention timecould be limited at 1 hour.

Both problems above are associated with thea-recycle ratio; both problems can be accommodatedby a modification of the UCT process called theModified UCT process (Figure 7.6). In the ModifiedUCT process, the anoxic reactor is subdivided into tworeactors, the first having a sludge mass fraction ofabout 0,10 and the second having the balance of theanoxic sludge mass fraction available. The first anoxicreactor receives the underflow s-recycle and ther-recycle to the anaerobic reactor is taken from it. Thesecond anoxic reactor receives the a-recycle. Theminimum a-recycle is that which introduces just suffi-cient nitrate to the second anoxic reactor to load it to itsdenitrification potential. Any recycle higher than theminimum will not remove additional nitrate so that athigher recycles more nitrate is introduced than removedin the second anoxic reactor and nitrate will appear inthe effluent from this reactor. This however is im-material insofar as it affects the nitrate in the aerobicreactor which remains constant once a > amtn. Conse-

quently one can raise the a-recycie to any value greaterthan a.Ihin, to give the required actual retention time,without affecting the nitrate recycled to the first anoxicreactor - careful control of the a-recycle is no longernecessary. This improvement however is obtained at acost: The maximum TKN/COD ratio to give a zeronitrate discharge to the anaerobic reactor is reducedfrom 0,14 in the UCT process to 0,11 in the ModifiedUCT process. However a TKN/COD ratio of 0,11 mgN/mgCOD includes most settled and raw municipalwastewaters. Furthermore, by making provision thatthe r-recycle can be taken from either the first or secondanoxic reactor, the process can be operated either as amodified UCT or a UCT process, as may be required.

The developments described so far were all guidedby the hypothesis that excess P removal is stimulated,and is best achieved, by having an anaerobic reactorand that it is optimized by preventing nitrate from enter-ing the reactor. No information was available to quan-tify firstly, the conditions in the anaerobic reactor thatcause the stimulation and, secondly, the magnitude ofthe excess P removal to be expected. Accordingly thesetwo aspects were further investigated.

3. EVOLUTION OF EXCESS PHOSPHORUSREMOVAL MODEL

3.1 Prerequisites for excess phosphorus removal

The improved understanding of the nitrogen removalprocess behaviour led to a new approach in investigat-ing the prerequisites in the anaerobic reactor for excessphosphorus removal. Ekama, van Haandel and Marais(1979) hypothesized that if the mass of nitrate enteringan unaerated reactor is less than that reactor'sdenitrification potential, the difference defines an"anaerobic capacity" in the reactor. They speculatedthat this anaerobic capacity could substitute for theredox potential level (suggested by Barnard, 1976) as ameasure for predicting when phosphorus release wouldtake place in the anaerobic reactor. Siebritz, Ekama andMarais (1980) intensively investigated this hypothesis.They plotted the P release in the anaerobic reactor, theP uptake in the aerobic reactor and the system Premoval versus the anaerobic capacity IAC) for datafrom UCT and modified Phoredox processes (Figure7.7). This plot shows that (i) if Ac < 10 mgNOj-N/l, noP release in the anaerobic reactor is obtained; in fact, Puptake is noted; (ii) when Ac > 10 mgNO3-f\l/7, Prelease is obtained; and (iii) as Ac increases above10 mgNOj-N/i so the P release in the anaerobic reac-tor, the P uptake in the aerobic reactor and the system Premoval increase."*

The data in Figure 7.7 appeared to support theanaerobic capacity hypothesis. Consequently if theanaerobic hypothesis was correct, then provided thiscapacity could be induced in any system other than the

* For reasons not yet understood the anaerobic retention time, both nominal and actual does not 3ppear to affect the settle ability of the sludge.

"From Figure 7.7 (and Figure 7.8) it can be seen that having stimulated P release in the anaerobic reactor, the P removal increased from A toonly 6 mgP 7. The reason for this low increase in excess P removal is that the experimental units turned Out to be poorly designed to achievehigh P removals because the factors affecting the magnitude of excess P removal were not understood; once these factors were understood,it became possible to design processes capable of achieving high levels of excess P removal Isee Figures 7.10 and 7.11).

7-5

£L

e

zUiozoo

V)

Ia.moXa.

u

8

6

4

2

n

2

4

6

8

0

12

!4

16

8

*

4

-

-

—i 1 1 1 1 1 1 I

i i ( t t \ i'" "T T t f T * 1 ^

* * •

Minimum anasrobiccopacity to inducephosphoruB relecs* —

PHOREDOX UCT

RELEASE • OSYSTEM UPTAKE A ASYSTEM REMOVAL 4*

i . , . . i .

i l l• ;

ii

•

i

•4i

•i

—*t i

A

ji

i : .

—i—r-i

o

#f

9 iT T

• oA.

A

1 i

7* O 3 O*

*

*

° ° a•• o1 1 r

G

i i • •

A

A

. A•*• A .

\

\

A. :

0 2 4 6 8 10 !2 14

ANAEROBIC C A P A C I T Y ( m g N / l i n f l u e n t )

FIGURE 7.7: P release in the anaerobic reactor, process P up-take and process P removal versus the anaerobic capacity ofthe anaerobic reactor (Ac) for the modified Phoredox and UCT

processes.

a 420

ifr

moX

Minimumto induct

photpr>oruf

iC, RELEASE •I&TSYSTEM UPTAKE *

SYSTEM REMOVAL

OB K) 16 20 25 30 35 40 46 50 5B 60S0LU1LE READILY BlODEQRADABLE SUBSTRATE. S_ I

* of a

FIGURE 7.8: P release in the anaerobic reactor, process P up-take and process P removal versus readily biodegradable CODconcentration in the anaerobic reactor (Sbaa) for the modified

Phoredox and UCT processes.

7-6

UCT and Phoredox processes, P release and uptakeshould be observed.

To test the above conclusion experimentally, threeModified Ludzack-Ettinger (MLE) processes (Figure 7.4)and a modified UCT process were set up and fed fromthe same wastewater source. The three MLE units weregiven unaerated sludge mass fractions of 40, 55 and 70per cent respectively and ihe mixed liquor a-recycleratios were set such that the anaerobic capacities rang-ed from 6 to 35 mgNOj-N/ / in the anoxic reactors. Overtwo months of operation no P release nor excess Premoval were observed in any of the MLE units. In con-trast, the modified UCT process with a 0,10 anaerobicsludge mass fraction, consistently gave P release andexcess P removal. It was concluded from these resultsthat the anaerobic capacity hypothesis is not sustained.

In seeking an explanation for the different P releasebehavioural patterns in the modified UCT and MLE pro-cesses, it was noted that the only evident difference layin the concentration of readily biodegradable COD sur-rounding the organisms in the anaerobic reactor (Sbsa).In the modified UCT process the readily biodegradableCOD concentration in the anaerobic reactor (S[lWJ is themaximum possible as no nitrate is recycled to theanaerobic reactor; in contrast, in the MLE process suffi-cient nitrate is recycled to the anoxic reactor to utilize allthe readily biodegradable COD i.e. Sbsa^0,0. Thereforethe different behavioural patterns of the processeswould be consistently described if it is assumed that theconcentration of readily biodegradable COD in theanaerobic reactor (Sbsa) surrounding the organisms isthe key parameter determining whether or not P releaseand excess P uptake takes place.

To test the assumption above, it was necessary todevelop a general equation by means of which Sl)Sa canbe calculated:

Consider the anaerobic reactor in the UCT process,

Sbsa = (ft* Sbl-ASbs)/(1 + r) (mgCOD//) (7.1a)

and that in the Phoredox process

St,H = tfbsSb(-ASbsl/(l -t-s) (mgCOD/i) (7.1b)

The term ASbE is the concentration of readilybiodegradable COD utilized for synthesis of cell materialwith nitrate (introduced via the r-recycle) and dissolvedoxygen (introduced via the influent) serving as terminalelectron acceptors. Because 8,6 mgCOD are utilized forthe removal of 1 mgN03-N and 3,0 mgCOD with1 mgO, (see Chapter 1), ASDs can be expressed in termsof the nitrate in the r-recycle, Nhr, and the DO in the in-fluent (0,) and r-recycle (O,), i.e.

Nr, and O refer to the nitrate and DO concentrationsrespectively. Symbols r and s and subscripts r,sgnd i refer respectively to the r- and s-recycles andthe influent.

The calculation of Sb6a fortunately was possiblebecause once the importance of the influent readilybiodegradable COD fraction (ffaj was recognized fromthe work on denitrification, it became standard ex-perimental procedure to measure or determine almostdaily, fbs. fys< *uD. Stj, Nnr, Or and 0,. Consequently Eqs.7.1 and 7.2 were applied to the data used to plot Figure7.7 and replotted in Figure 7.8 with Sbsa on the horizon-tal axis instead of the anaerobic capacity. From Figure7.8 the following conclusions were drawn:

for the UCT process

ASCS - r(8,6.Nm-r3,0.Or) -t 3,0

and for the Phoredox process

ASbE - s(8,6Nns + 3,0.Os) + 3,0

where

(1) The minimum readily biodegradable COD concen-tration in the anaerobic reactor (Sbsa) to stimulatephosphorus release in the reactor is about25 mgCOD/f.

(2) The degree of P release appears to increase as Sbsa

increases above 25 mgCOD//, i.e. P release in-creases as (Sbsa-25) increases.

(3) Excess phosphorus uptake is obtained only whenphosphorus release takes place, and tends to in-crease with (SbE,d-25).*

The plots in Figures 7.7 and 7.8 show a strikingsimilarity but on closer inspection, this similarity is moreapparent than real. The similarity is there only for theUCT and Phredox data: for these processes, the twosets of data fall in approximately the same positionrelative to the coordinate axes. This is so because inthese processes, the principal contributor to Ac is thereadily biodegradable COD in the influent, Sbsi - thecontribution by the slowly biodegradable influent CODSbpi is very small owing to the relatively small anaerobicsludge mass fractions (fxa) - so that Ac and Sbsa areessentially equivalent parameters. In contrast, in theMLE process, the influent readily biodegradable CODSbsi will have been completely utilized by the nitrateentering the anoxic reactor, i.e. Sbsa = 0. Neverthelesshigh Ac values are obtained from the SbDI contributionbecause the unaerated sludge mass fractions are large.Consequently for the MLE process, there is no relation-ship between SDsa and Ac. Hence in Figure 7.7, the MLEdata fall in the right hand side of the plot (i.e.Ac > 10 mgN03-N/i) whereas in Figure 7.8 they wil! fallin the left hand side of the plot (i.e. S K M - 0 ) . Conse-quently, the MLE behaviour is incorrectly predicted bythe Ac hypothesis (Figure 7.7) but correctly predicted bythe Sb£a hypothesis (Figure 7.8). These results allowed

17 2a) ^ e * o t i o w i n 9 statement to be made: A prerequisite for Prelease in the anaerobic reactor is that the concentrationof readily biodegradable COD surrounding the

(7.2b) organisms in the anaerobic reactor must exceed approx-imately 25 mgCOD/f.

1 From Figure 7.7 (and Figure 7.8) it can be seen that having stimulated P release in the anaerobic reactor, the P removal increased from 4 to only6 mgP/j[. The reason for this low increase in excess P removal is thai the experimental units turned out lo be poorly designed to achieve high Premovals because the factors affecting the magnitude of excess P removal were not understood; once these factors were understood, itbecame possible TO design processes capable of achieving high levels of excess P removal (see Figures 7.10 and 7.11).

7-7

3.2 Magnitude of excess phosphorus removal

A prerequisite for P release was stated above; however,the magnitudes of the P release and subsequent uptakeand removal were not explicitly defined, once the prere-quisite was satisfied. For this purpose an extensive andintensive investigation was inaugurated utilizing theUCT and modified UCT configurations. Series of testswere done under constant flow and load conditions: (1)at 14°C and 20°C, <2> using raw wastewater influentswith COD's of 800, 500 and 300 mg / i , and (3) varyingthe r-recycle and anaerobic mass fraction. In all the in-vestigations the influent wastewater characteristics, i.e.COD, TKN, total P and readily biodegradable COD frac-tion fbs, and filtered TKN, NH«, NOj and total P in theprocess reactors were measured daily. In addition, allthe normal parameters such as MLVSS, oxygen utiliza-tion rate, stirred settling velocity, pH and alkalinity weremeasured at regular intervals.

An analysts of these data* indicated that providedSbM > 25 mgCOD/l,

(1) for fixed r-recycie ratio and influent COD, increas-ing the volume of the anaerobic reactor (at the ex-pense of the other reactor volumes) tended to in-crease P release and the P removal;

(2) for fixed r-recycle ratio and anaerobic reactorvolume, an increase in influent COD concentrationtended to increase the P release and the P removal.

Although the trends stated above were evident itwas not possible to identify a consistent behaviouralpattern and it was concluded that the basic parametershad not been properly identified or isolated. In part, thiswas due to the interactive effects between the differentparameters; for example, increasing the r-recycle ratio,(i) increases the mass of sludge passing through theanaerobic reactor, (ii) decreases the time the sfudge isretained in the anaerobic reactor, (iii) increases the con-centration of sludge in the reactor, and (iv) decreasesthe concentration of Sbsa in the anaerobic reactor (evenif no nitrate is present in the r-recyclet. With so manyassociated changes it was difficult to see directly whichparameters are the significant ones. It was decided,therefore, to investigate the system theoretically on thebasis of the perceived and hypothesized behaviour.

Considering Figure 7.8, the P release, uptake andremoval appeared to be linearly related and it was decid-ed to utilize the removal as the criterion against which tojudge the effect on P by any modification. As a conse-quence, P release per se was not incorporated into thehypothesis proposed to describe the P removal. Thefollowing behavioural pattern was hypothesized on theP removal based on experimental observations:

(1) Excess P removal is obtained only when Sb&a > 25mgC0D/ i . #*

(2) As Sbsa increases above 25 mgCOD/i, so the Premoval increases.

(3) The longer the actual anaerobic retention time(Ran), the higher the P removal.

(4) The larger the mass of sludge recycled through theanaerobic reactor, the higher the P removal;

or. Transformed to a more useful equivalent:

The greater the mass of sludge recycled throughthe anaerobic reactor each day expressed as a frac-tion of the mass of sludge in the process, n, thehigher the P removal.

The statements (2) and (41 above do not give an ex-plicit quantitative expression of the expected behaviour.To obtain such an expression an hypothesis on the ex-pected behavioural pattern of (2) to (4) has to be made.A hypothesis consistent with the observations is:

"When any one of the factors (Sili3-25), Rart, or n,is zero, excess phosphorus removal will be zero."

From this hypothesis, the simplest form of an equa-tion expressing the tendency of a process to achieve ex-cess P removal is

<= (Sbsa-25).Ran.n (7.3)

where

= excess phosphorus removal propensity factor.

If Pf as defined above describes the observedbehaviour of the three parameters, then the P removaldue to excess uptake in the sludge (Ps) should be afunction of P,, i.e.

Ps= f(P,l (7.4)

Now R.ri. n in Eq. 7.3 can be expressed in terms ofthe process configuration and operational parameters,and for both the UCT and Phoredox processes this termcan be shown*** to be equal to fty i.e. Ran.n = fxa wherefKS is the anaerobic sludge mass fraction. Substitutingfxa for R3n-n in Eq. 7.3 yields

Pf = (Sbsa-25)fxs; when Sbsa < 25, Pf = 0,0 (7.5)

Hence for the UCT and Phoredox process, it wasconcluded that the propensity to give excess P removalunder constant flow and load conditions is a functiononly of the magnitude of readily biodegradable CODconcentration in the anaerobic reactor above25 mgCOD/f and the magnitude of the anaerobicsludge mass fraction.

3.3 Quantitative model for excess phosphorusremoval

As excess P removal is accepted to be a biological ac-tivity, it can be assumed that, (i) only the active fractionof the sludge can take up phosphorus in excess, (ii) thephosphorus removal will increase as the proportion ofphosphorus in the active mass increases, and (iii} theproportion of phosphorus in the inert fractions of thesludge will remain unchanged. Accepting theseassumptions, Martin and Marais (1975) developed thefollowing equation for phosphorus removal based on

•Detailed description of the experimental investigation is given in a report by Siebritz, Ekama and Marais (1983).

"Sbsa is the concentration after allowing for removal of Sbs due to nitrate and oxygen entering the anaerobic reactor.

• " S e e Siebritz, Ekama and Marais (1982).

7-8

the steady state activated sludge process modelpresented by Marais and Ekama (19761 (see Chapter 4,Section 9) i.e.

I us ' 'up' ' i U

(7.6)

where

f,. =

y =

P removal from the wastewater by incorpora-tion in the sludge (mgP/f)P content of the endogenous and inert frac-tions of the volatile mass (mgP/mgVSS)coefficient of excess P removal i.e. the P con-tent of the active mass ImgP'mgVASS)

With the aid of Eqs. 7.1, 7.2, 7.5and7.6, the coef-ficient of excess P removal y and the P removal propen-sity factor P, were calculated for daily sets of ex-perimental data* measured over a period of 18 months.In calculating the daily y and P. data pairs, those datafor which there were known causes that may haveadversely or favourably influenced the processbehaviour were discarded.1"1 The remaining daily y andP, data pairs are shown plotted in Figure 7.9.

The data in Figure 7.9 show considerable scatteralthough a definite trend is discernible. Some of thescatter can be attributed to the selection of the data

values in the calculation of Pt; The fLs, nitrate and Premoval values used in the calculations were allmeasured on the same day, i.e. the assumption wasmade that the response of y to Pf is instantaneous. This,in fact, is not so; due to the long hold up time in thesystem it was repeatedly observed that the P removalresponse is out of phase with the input by some fractionof a day. (This is apparent in Figure 7.10 which recordsthe observed and predicted P removal in one set of ex-periments).

Accepting the relatively large scatter, the form ofthe relationship and the constraints on the y ~ P, rela-tionship were derived as follows: From Figure 7.9 ap-parently

(i) there is an upper limit to y, of about 0,35 at veryhigh propensity factors,

(ii) there is a lower limit to y of about 0,06 at a propen-sity factor of zero,

(iii) for propensity factors ranging from infinity to zero,the maximum change in y is (0,35-0,061=0,29and the decrease in y with decrease in P, appears tobe of an exponential form.

On the basis of the above the following expression

30,35

X

>0,20

0,15

o

9 op

Sop

CODin-900 mQ^ •

fx a =0,IOlT=20°C)

10 ^A Uo =0,20(7-20^e f , 0 -0,(6 [ 1. )

o Ua =O,IO( - ) •

• *xo •O.I5(T-J4°C)

PHOSPHORUS REMOVAL PROPENSITY FACTOR, p f-<sb i e-25)f,10

FIGURE 7.9: Coefficient of excess P removal (7) versus Premoval propensity factor (P|l observed in the modifiedPhoredox, UCT and modified UCT processes at differentanaerobic sludge mass fractions lf.11). influent COD concentra-

tions and temperatures.

'The readily biodegradable COD fraction ^ in the influent was measured in accordance with the procedure developed by Do!d, Ekama andMarats (1980). Details of this procedure are given in Appendix 2.

'it W3s found that after a disruption of the process due to a mechanical breakdown, or a change in influent sewage characteristics or processparameters, the process required approximately 2 to 3 days to restabiliie.

7-9

iO 20 30

DAY OF OPERATION DAY OF OPERATION

FIGURE 7 10: ieff, daily experimental data of influent COD,TKN and readily biodegradable COD concentrations and nitrateconcentrations in the effluent and the recycle to the anaerobicreactor and right, the measured and predicted phosphorusremoval for a modified UCT process with the following sludgemass fractions; anaerobic 0.10, first anoxic 0,08, second anoxic0,22, aerobic 0,60. (Data from Siebritz, Ekama and Marais,

1983}.

was chosen to represent the trend in the plotted data

Y - 0,35- 0,29 exp(-C.P f)

The value of C was found by a feast square analysisof the differences between the measured and predictedphosphorus removals and found to be equal to - 0,242,i.e.

y - 0,35-0,29.exp( -0,242 P,| (mgP/mgVASS)(7.7)

To check the predictive capacity of Eq. 7.7, it wasapplied to the daily P, data in all the five sets of ex-periments on the modified UCT, UCT and Phoredoxprocesses in which Sbsi and the nitrate in the recycleswere measured. (As stated above these sets of datacovered influent COD values from 300 to 800 mg/ i at14° and 20°C under a variety of TKN.'COD ratios,anaerobic mass fractions from 0,10 to 0,20 and sludgeages 12' to 25 days). As an illustration two sets of dataare shown in Figures 7.10 and 7.11 together with thepredicted P removal values. Evidently the model gives agood prediction of the observed P removal. In Figure

7.10, the pronounced favourable effect of an increase inthe readily biodegradable COD concentration (5bsi) onthe P removal, and in Figure 7.11 the depressing effectof nitrate in the recycle to the anaerobic reactor, areclearly illustrated.

The theory was tested further by feeding additionalreadily biodegradable COD (acetate and glucose) to theanaerobic reactor. Taking due account of the increasedCOD, again the predictions of P removal were in satis-factory correlation with those observed.

While doing a biodegradability investigation of thewaste flow from the town of Caledon in South Africa,unexpected verification of the theory was obtained Theprocess configuration was a Modified Ludzack-Ettingerprocess (Figure 7.4) with an anoxic sludge mass fractionof 0,30 operated at 20-day sludge age, at 20cC, andwith a mixed liquor (a) and an underflow (s) recycle ratioof 4:1 and 1:1 respectively. The wasteflow contained anindustrial fraction (from a malting factory) constitutingapproximately 2/3 of the total COD load. The influentCOD and TKN strengths were 1300 mg/B and70 mgfsj/i respectively. It was found that the malting

•The units in the short sludge age range were non nitnfication-denitrification processes of ihe MLE configuration (Figure 7.4) with the first reac-tor anaerobic and the second aerobic.

7-10

Fffluent rsiUoleEfflueni TKN

:OD

20 40DAY OF OPERATION

Predicted removal

Measured removal

- 3r

O

o :o^

Nitrate recycled1o cnoerobiC if reoctor^ I

• ^

0 20 40 60DAY OF OPERATION

60

300

Ou

200 yCD<Q

I5O<CDUJo

I00 OCO

>-:

50

JO

iOO

FIGURE 7.11: Response of a laboratory scale UCT ptocess (at14°C; anaerobic mass fraction = 0,15; anoxic mass traction 0,30;aerobic mass fraction 0.55; sludge age 20 days;) treating un-settled f low from Mitchell's Piain. (Data from Siebritz, Ekama

and Marais, 1382b).

waste was virtually completely readily biodegradableresulting in a very high readily biodegradable COD frac-tion (fbs) and concentration (Sb£j), 0,70 and980 mgCOD// respectively. Taking due account of thediluting effect of the total recycle flow of 5 and thereduction of the readily biodegradable COD in the"anoxic" reactor by the nitrate discharged in the recycleflows (nitrification was complete, effluent TKN =

4mgN' f ) , the readily biodegradable COD concentra-tion in the "anoxic" reactor iSbsa) was estimated at101 mgCOD/i, the propensity factor IPf) at 22,9 (Eq.7.5), the y coefficient at 0,35 (Eq. 7.7) to give anestimated P removal of 31 mgP/1 influent (Eq. 7.6 withfus = 0,05 and fup-0,06). The observed P removal was29mgP/ i . The resufts of this experiment should becontrasted with those discussed earlier on the Modified

7-11

Ludzack-Ettinger processes. In those processes, no ex-cess P removal was observed or predicted, even thoughthe anoxic mass fraction was as high as 0,70, principallybecause the readily biodegradable COD concentrationin the influent <S,S) was low and completely utilized inthe anoxic reactor by the nitrate recycled to this reactor.Clearly if sufficient Scsi is available and the TKN not ex-cessive, it is possible to obtain high P removals eventhough the nitrate recycled to the anaerobic reactor ishigh, without having to iesort to Phoredox or UCT con-figurations, provided the conditions for P release aresatisfied as set out in Section 3.1 above.

At ful! scale Nicholis, Osbcrn and Marais (1982)and Nicholis (1982) tested the predictive power of thegeneral nitrification-denitrification model on theGoudkoppies 5-stage Phoredox plant receiving an ef-fluent that had passed through primary settling andthen through an equalization tank. The P removalbehaviour of the plant was also monitored by measuringP conditions in the anaerobic and aerobic reactors andin the effluent. To evaluate both the model and Premoval theory, inter alia the readily biodegradable CODin the influent discharged to the process (Sw.) wasmeasured, as described by Dold, Ekama and Marais(1980). The SL&I was found to be very low due to itsutilization by an anaerobic sludge layer that formed onthe bottom of the unstirred equalization tank. In termsof the theory set out here the SbSi was so low that withthe 1:1 underflow recycle from the secondary settlingtank to the anaerobic reactor, even if no nitrate was pre-sent in the recycle, Sssa would be less than 25 mg/I andconsequently little excess P removal was to be ex-pected, and, virtually no excess removal was observedfor the plant. Measurements of SLs) on the influent tothe equalization tank however indicated that the Sb&,was sufficiently high that, if the equalization tank wasbypassed, excess P removal was theoretically possibleprovided the nitrate in the underflow recycle to theanaerobic reactor could be maintained at less than5 mgN/ i . Now at this point a computer-based equaliza-tion control strategy was implemented on the equalization tank (Dold, Buhr and Marais, 1982) and operated insuch a fashion that any settled material was flushed dai-ly from the tank. The flushing action prevented build-upof a sludge layer and effectively eliminated the loss ofSi,,,; in the tank. The process recycles in the plant werechanged in accordance with those predicted by thegeneral model to achieve maximum nitrate removal. Asthis removal was still insufficient to reduce the nitrate inthe underflow recycle to less than 5mgN/f , theaerators next to the discharge points from the primaryanoxic reactor were switched off, thereby effectively in-creasing the anoxic zone and further reducing thenitrate in the effluent. (Sufficient aerator capacity wasavailable in the rest of the aeration basin not to affectthe nitrification efficiency). In this fashion the nitratewas kept between 3 and 5 mgN/i in the effluent andthe underflow recycle. Under these conditions thepredicted P removal was 4 to 5 mgP i influent; observ-ed mean removal was 5 mgP/ /. The results from this in-vestigation are particularly instructive because theydemonstrate that little or no P removal is achievable ifSbs, is too low, or, where SHSI is sufficiently high,although the potential for removal is there, it will not be

possible to realise this potential if nitrate in the recycleto the anaerobic reactor is so high that it reduces theSuss concentration to very low levels.

When investigating P removal behaviour inlaboratory scale processes, batches of sewage usuallywere obtained from the outfall sewer and stored at 5°Cfor subsequent feed to the laboratory process units.Often it was noticed that with time the magnitudes ofdenitrification and P removal declined while feedingfrom the same raw sewage batch. Measurement of thereadily biodegradable COD fraction of the daily feedtaken from the cold storage batch indicated that nor-mally there was a slow continuous loss of this fraction,giving rise to an associated reduction in denitrificationand P removal. Hence, batches should not be storedlonger than about one to one and a half weeks and theSu , in the feed should be measured daily. Furthermore,by monitoring the SLb. on a routine basis, changes in St,s;between batches are picked up that explain the "er-ratic" P removal behaviour noted earlier between dif-ferent batches of sewage. Also, it is vitally importantthat the tanker transporting the batch to the laboratory,and the cold storage tank, are thoroughly cleaned be-tween batches otherwise hydrogen sulphide generationand rapid loss of Sbsl will be encountered. Highhydrogen sulphide concentration in the feed to a pro-cess can have adverse effects on the process response.

3.4 Comments on the parametric model andrecent research developments

The parametric model described above was developedfrom observed data on experimental processes operatedover a wide range of conditions. The conditions rang-ed as follows:

(i) Influent COD concentration : 250- 800 mgCOD/f

iii) Readily biodegradable COD: 70-220 rngCOD//i.e. fraction f.£ : 0,12-0,27

(iii) TKN/COD ratio

(iv) Siudge age

(v) Temperature

: 0,09-0,14

: 13 and 25 days

: 12°C and 20°C

Development of the model required the identifica-tion of the principal factors and conditions that ap-peared to influence the excess P removal and integra-tion of these into a consistent quantitative expressionfor estimating the excess P removal. Summarizing, themain factors identified are:

(il As the concentration of readily biodegradable CODin the influent increased, so the excess P removaldid likewise.

(ii) As the anaerobic sludge mass fraction increased sothe excess P removal did likewise.

(iii) For identical conditions for (i) and (ii) above, as thesludge age decreased, so the excess P removal in-creased.

7-12

5 10 15 20 25MEASURED PHOSPHORUS REMOVAL (mgp/l)

FIGURE 7.12. Comparison of measured P removals with thosepredicted by the parametric model for all the data observed

under the different operating conditions.

(iv) Temperature reduction appeared to have a slightpositive effect on excess P removal.

The above tendencies formed the basis for struc-turing the formulations for estimating the excess Premoval and the equations so derived were calibratedagainst the observed data (see Section 3.3 above). Acomparison of the theoretically predicted and ex-perimentally measured process P removal data for theconditions set out above is shown in Figure 7.12: clearlya good correlation is obtained.

// is of the greatest importance to note that themodel is semi-empirical - its use must be limitedstrictly to within the ranges of process para-meters and wastewater characteristics fistedabove. In particular, the model should not be usedfor design of plants with sludge ages less thanabout 14 days: in South Africa no research hasbeen done on the process at short sludge ages andthere is no certitude that the model predicts thebehaviour adequately.

Despite the good correlation, the parametric modeits unsatisfactory from a scientific point of view becauseit is independent of any formal hypothesis on thebiological mechanisms driving the process. Conse-quently the model should serve as an interim measureonly for design and other practical purposes. What is re-quired is a model based on fundamental biochemical

and physical rules that will give rise to similarbehavioural characteristics that formed the basis of theparametric model.

Historically important contributions towardsdeveloping such a model were made by Fuhs and MinChen (1975) and the Johannesburg group (Nicholls,1978). The latter group in particular, inaugurated a newapproach which has influenced more recent hypothesiz-ed mechanisms proposed by Rensink (1981) and Maraiset al. (1982). Preliminary indications are that a modelbased on the behaviour proposed by Rensink andMarais ef al. has the potential to explain most of thebehavioural characteristics of excess P removal andshould in time allow a kinetic description to be incor-porated into the genera! activated sludge processtheory.

In essence, the basic concepts governing theirmodel are as follows: Poly-P organisms, such asAcinetobacter, store phosphate internally as polyphosphate chains under aerobic conditions - somestrains can do so also under anoxic conditions withnitrate serving as the electron acceptor. When poly-Porganisms are discharged to an anaerobic reactor whichreceives the wastewater influent flow, the organismssurvive the anaerobic conditions by breaking downsome of the accumulated po!y-P chains and the energythereby released is utilized to complex the lower fattyacids surrounding the organisms to higher forms, suchas aceto-acetate or polyhydroxybutyrate and storethese compounds internally. When the organismsthereafter enter an aerobic (or anoxic)* zone, theyutilize the complexed compounds with oxygen (ornitrate)* as the electron acceptor for growth andreplenishment of the poly-P stores.

If lower fatty acids were to be discharged to theaerobic (or anoxic) zones, the poly-P organisms wouldhave to compete with the heterotrophs (or facultativeheterotrophs) for this substrate because now it isavailable to these other organisms due to the presenceof oxygen (or nitrate) as electron acceptor. However,because the poly-P organisms are relatively slow grow-ing compared to the other organisms, they will obtainonlv a minor fraction of the lower fatty acids in theaerobic reactor and consequently their growth will bevery small. (This explains the reason why poly-Porganisms are rarely found in considerable numbers inpurely aerobic or anoxic-aerobic systems). Hence, byincorporating an anaerobic zone (which receives nonitrate from return recycle and to which the influentflow is discharged), the growth of the poly-P organismsis favoured, by allowing the organisms the opportunityto sequester the lower fatty acids by using their poly-Pstores. In the anaerobic zone the other organisms are in-operative due to the unavailability of external electronacceptors.

With regard to the source of lower fatty acids forgrowth of the poly-P organisms this is present in the in-fluent in the readily biodegradable COD fraction.However, not all the readily biodegradable COD is likelyto be in a lower fatty acid form. Under these cir-cumstances the behaviour of the facultative

'Florentz and Granger (1982). using Pn nudear magnetic resonance techniques, present evidence showing polyphosphate accumulation underanoxic conditions with nitrate serving as the electron acceptor.

7-13

hetetotrophs may well be advantageous to the poly-Porganisms. In the anaerobic reactor the facultativeorganisms can derive a small amount of energy (prin-cipally for survival) by generating internal to itself anelectron acceptor to break down the higher forms ofreadily biodegradable material (like glucose) to towerfatty acids, via the Embden-Meyerhof pathway.

Under anaerobic conditions these acids cannotenter the Krebs cycle because of the lack of an externalelectron acceptor, and the lower fatty acids formed bythe facultative heterotrophs are released to the bulkliquid. In this way the poly-P organisms still cansequester lower fatty acids even though none might beavailable in the influent. In this fashion a slow growingorganism like Acinetobacter can flourish in ananaerobic-snoxic-aerobic mixed culture system eventhough it may be an obligate aerobe. Apparently withinthe time scale of normal anaerobic zones, i.e. actualretention times up to 2 hours, a significant reduction ofslowly biodegradable paniculate COD to lower fattyacids is unlikely so that where lower fatty acid produc-tion does take piace in the anaerobic reactor, this isprincipally from the readily biodegradable COD fraction.

In terms of the biochemicai mode! the P released inthe anaerobic zone is due to the utilization of poly-Preserves to provide energy for complexing the lower fat-ty acids. From stoichiometry (see Marais et a/. 1982) forevery P released there is an approximately constantmass of lower fatty acid complexed - the mass of po!y-P organisms that will be formed depends on the mass oflower fatty acid complexed. The excess P removal at-tainable depends on the mass of poly-P organismsformed* which in turn depends on the proportion ofmass of lower fatty acids and readily biodegradableCOD in the influent that is complexed by the poly Porganisms.

Research into P release under batch anaerobic con-ditions using acetate as lower fatty acid (Law, Dold andMarais, 1982] has supported the above conclusions thatthere is an approximate proportionality between themass of P released and the mass of acetate complexed-about 1 mgP/f phosphate to 2 mgCOD/i acetate.The rate of acetate complexing is very fast. Now in anormal municipal wastewater the lower fatty acid frac-tion of the readily biodegradable COD is likely to be low.Here the presence of the facultative heterotrophs maybe important for the generation of the lower acids.However, in this case the po!y-P organisms can com-plex the lower fatty acids only as fast as these are pro-duced by the facultative heterotrophs: batch test ex-periments where P release was stimulated usingmunicipal wastewater, indicated that the rate of lowerfatty acid compiexing, or equtvalently the rate of Prelease, is about 5 times slower than that with acetate.Fu, ..hermore, these tests support the earlier conclusionthat only the readily biodegradable COD is reduced tolower fatty acids for complexing, by the consistent wayin which the mass of P released follows the mass ofreadily biodegradable COD input to the batch test even

with a relatively prolonged anaerobic period (about 3hours).

In order to develop a greater understanding of thekinetics of readily biodegradable COD reduction tolower fatty acids and P release, a Modified UCT processwith a plug flow anaerobic reactor treating municipalwastewater was operated. It was found that the mass ofphosphorus released per unit elemental length of plugflow reactor decreased exponentially along the length ofthe reactor. This means that the rate of readilybiodegradable COD reduction and complexing is firstorder with respect to the readily biodegradable CODconcentration. Using this first order approach, anelementary steady state kinetic model has beendeveloped. This model indicates that because the readi-ly biodegradable COD is reduced and complexed accor-ding to a first order rate, greater efficiency would bederived by subdividing the anaerobic reactor into a

50

40

za.0)OT.Q.

1 0 -

0

fx0 C.042Reactor 1

TV-op » 20'C3 t i , » 150- f50

0,084 0,1262 3

0,1684

0 50 100 ISOACTUAL HNAEROBIC RETENTION TIME

( min)

FIGURE 7.13a: Phosphorus concentration profile in a 4 in-seriesanaerobic reactor in a modified UCT process. S7i = 500mgCOD'f raw wastewater, Sbsi = 150 mg COD''i. sludge age 15days, r-recycle ratio 1:1, temperature 20rC, MLVSS in

anaerobic reactions 1450 mgVSS/f.

"At steady state, the nett number of poly-P organisms thai form is equal to the number removed in ihe daily waste sludge. In essence excessbiological P removal is ihe continuous formation of poly-P organisms and their removal from the process at a stage when they are filled withpoly-P.

7-14

0 50 100ACTUAL ANAER0E1C RETENTION TIME

{rr<; n}

FIGURE 7.13b: Phosphorus release per reactor in 4 inseriesanaerobic reacior. Data from Figure 7.13a.

number of compartments and operated in series. Ex-periments with continuous Modified UCT processes inwhich the anaerobic reactor was subdivided into 4 equalcompartments, corroborated the first order approach(Figure 7.13)-for the same anaerobic sludge massfraction and readily biodegradable COD in the influent,ihe mass of P release in the anaerobic reactor wasgreater for the series reactor anaerobic system than forthe single reactor anaerobic system; the P removal toowas proportionally higher in the former than in the lat-ter.

On the strength of the results discussed above itwould appear that the anaerobic reactor should be sub-divided into 2 or more compartments in series toachieve improved excess P removals for a fixedanaerobic sludge mass fraction. This subdivision alsoallows greater process flexibility in that not only can theanaerobic sludge mass fraction be changed but alsoallows different nutrient removal processes to be incor-porated on the same plant (see Section 5 below).

Comparing the results of the experiments underbatch conditions and under continuous conditions withplug flow and multi-compartment anaerobic reactorswith those predicted by the parametric model, it ap-peared that there is an inconsistency. The results show-ed that P release is obtained for acetate and readilybiodegradable COD concentrations below 25 mgCOD/fwhereas in the continuous experiments on which theparametric model is based this was apparently not so.The causes for the difference can be ascribed to (1)measurement procedure and (2) kinetic behaviour.

(i) Because the P concentration in the influent con-sists of two fractions i.e. a large soluble (80%) anda small particulate 1^20%), an error is made whencalculating the magnitude of the P release in theanaerobic reactor, In calculating the magnitude ofthe P release a mass balance on P across the

anaerobic reactor is made and this is done byassuming that all The P entering and leaving thereactor is in a soluble form, even that of the in-fluent. The limitation of this method of measuringthe P release was realized and is probably the prin-cipal reason why m the parametric model it wasfound that the P content of the active mass (y)once anaerobic conditions were imposed is 0,06 in-stead of the more usual value of 0,03 for purelyaerobic processes i.e P release probably did takeplace for S[JS,a < 25 mgCOD/f and this stimulatedthe additional P uptake from 0,03 to 0..06.

iii) The parametric model indicates that apparently aminimum of 25 mgCOD/i readily biodegradableCOD in the anaerobic reactor is needed to induceexcess P removal. A reason for this behaviour nowcan be advanced in terms of the kinetics describingthe rate of P release observed on the seriesanaerobic reactors. A mode! has been constructedincorporating the kinetic P release hypothesis andall the data observed to date processed. The modelpredicts P uptake that is not as good as that obtain-ed by the parametric model; evidently it requiresfurther development. In consequence the modelwill not be described here. However, on one aspectof the model predictions deserve mention. Evident-ly for sludge ages of 20 days and anaerobic massfraction of 0,15 influent COD of raw sewage500 mg/f and Sbs,= 100 mg/i (the "normal" con-ditions tested in the experimental investigation) inthe anaerobic reactor only about 70 - 80 per cent ofthe readily biodegradable COD is sequestered i.e. ±20 to 30 mg/X COD is discharged from theanaerobic reactor - this would explain why(Sbhi-25) appears in the parametric model. Ifhowever the anaerobic reactor is subdivided into 2or more reactors the SL.S in the effluent from such aseries reduces to 5 to 10 mg/ i . The effluent Sbs

furthermore depends on the influent Sbb so that itwould seem the factor (Shs-25) is only an approx-imation.

The discrepancies discussed above are not crucialto the predictive power of the parametric model - aseries of calculations using the kinetic and parametricapproaches show that significant differences arise onlyfor low strength COD wastewaters, and here theparametric model tends to predict lower removals thanthe kinetic model. However, the kinetic model is not yetverified or developed sufficiently to replace theparametric modei.

The above discussion is not intended to cast doubton the parametric model and its usefulness; that canclearly be seen in Figures 7.10, 7.11 and 7.12. Theparametric model evolved out of an intensive researchprogramme to develop an understanding of and designcriteria for biological nutrient removal processes with lit-tle emphasis on kinetics. In the future, it is very likelythat it will be superseded by a more elegant kineticmodel without necessarily leading to very much betterpredictions. However, for the present, the parametricmodel is probably the best and most exhaustively testedmodel available and can be of great value for analysing

7-15

biological P removal process behaviour as it is under-stood at present. The remaining part of This Chapter willdemonstrate the use of the parametric model fordesign.

4. IMPLICATIONS OF THE MODEL ON DESIGN

The parametric model for excess biological P removalconsists essentially of two concepts i.e. (i) the coeffi-cient of excess P removal (y) and <ii) the excess Premoval propensity factor <P,), which are linked by anempirical relationship, Eq. 7.7.

4 1 Coefficient of excess P removal (y)

When excess P removal lakes place, this will bereflected in the magnitude of the coefficient of excess Premoval y in Eq. 7.6, the value being greater than 0,02 to0,03 mgP/mgVASS usually found for purely aerobicprocesses. With excess P uptake found in processes in-corporating anaerobic reactors, y can range from aminimum of about 0,06 to a maximum of about 0,35when the appropriate anaerobic conditions are present.

A plot of Eq. 7.6 (divided through by S,,) versussludge age {RJ is given in Figure 7.14 for raw and settl-ed wastewater (see Tables 4.3, 5.2 and 6.1) at 14°C and22°C assuming a constant y value of 0,18 mgP/mgVASS, a value readily attained in appropriatelydesigned P removal plants. Figure 7.14 shows that themass of P removed per mgCOD in the influent, P£/St!,decreases sharpiy with sludge age so that to remove themaximum P, the sludge age should be kept as low aspossible. The P removal per unit COD load is approx-imately the same for raw and settled sewage. Hence,

because primary sedimentation removes as much as40% of the COD load, P removal with settledwastewater is about 40% less than with raw waste-water. For a constant y the effect of temperature onremoval is small because the nett mass of sludge pro-duced does not change significantly with temperature.

4.2 Excess P removal propensity factor IPt)

The magnitude of the coefficient of excess P removal yis influenced by a number of conditions in the anaerobicreactor: (1| the magnitude of the readily biodegradableCOD fS b J in excess of 25 mgCOD//, (ii) the actualretention time and (iii) the mass of sludge (expressed asa fraction of the mass of sludge in the process) passingthrough the anaerobic reactor each day. These condi-tions were quantified for the UCT and Phoredox pro-cesses and formalized into the parameter called the ex-cess P removal propensity factor (P.) given by Eq. 7.5.The readily biodegradable COD concentration in theanaerobic reactor (S tsJ for the UCT and Phoredox pro-cesses can be calculated from Eqs. 7.1 and 7.2. Thesepredictions are based on the model derived from obser-vations on processes with sludge ages between 13 and25 days. Where the sludge ages are outside the limits, inparticular where the sludge ages are less than 13 days, itwould be unwise to extrapolate the model to these con-ditions.

4.3 Excess P removal model

The coefficient of excess P removal (y) and the propen-sity factor (Pf) described above were linked with an em-pirical relationship i.e. Eq. 7.7 found from data given inFigure 7.9. Equation 7.7 shows that the higher the P1

LLJoocra.

u o

< °o oS m

* ICO Q=> O

0,04

0,03

0,02

0,01

0,00

RAW AMD SETTLED WASTEWATER

Temperature

0

y = 0,!8 mg P / mg VASS

10SLUDGE AGE

20Id)

FIGURE 7 14 Phosphorus removal per kg COD load onbiological process versus sludge age for an excess P removalcoefficient (,] of 0.18 mgP/mgVASS at 14"C and 20cC for raw

and settled wastewaters.

7-16

Q.

RAW WASTE WATER

C 5 10NITRATE CONCENTRATION DISCHARGED

TO ANAEROBIC REACTOR (mgN/l)

FIGURE 7.15: Propensity factor (Pf), rapidly biodegradableCOD concentration in anaerobic reactor (Sbsal and coefficientof excess P removal if) versus the nitrate concentrationdischarged IO the anaerobic reactor for an anaerobic sludge

mass fraction (fxa) of 0,10 for raw wastewater.

factor, the greater the y coefficient. Consequently, at afixed sludge age, the higher the P, factor in a process,the greater the P removal per unit COD load (Ps/Stl).For a fixed fia, high P. factors are obtained with (i) highinfluent COD concentrations (for normal fbs fractions inthe wastewater of about 20%), (ii) high readily bio-degradable COD fractions <fbs), and (iii) a zero nitrateconcentration in the recycles to the anaerobic reactor(Nnr and Nns = 0). The sensitivity of excess P removal tonitrate discharged to the anaerobic reactor isdemonstrated in Figure 7.15. in Figure 7.15, the Sbsa

concentration, P< factor and y coefficient calculatedfrom Eqs. 7.1, 7.2, 7.5 to 7.7 for the raw wastewatercharacteristics given in Tables 4.3, 5.2 and 6.1 are plot-ted versus the nitrate concentration discharged TO theanaerobic reactor for an anaerobic sludge mass fractionf»a — 0,10, underflow recycle ratio s - 1, DO in s-recycle0.,= 1,0 mgO't and DO in influent 0,-0,0 mgO/f. itcan be seen that for nitrate concentrations > 7 mg/f,the y coefficient ia reduced to one third of the value fora zero nitrate concentration; at 25-day sludge age and14°C, this reduction in y from 0,22 to 0,07 mgP/mgVASS, results in a decline in P removal from 9,3 to3,5 mgP/I. This example demonstrates that for good Premoval, nitrate discharge to the anaerobic reactorshould be avoided.

5. DESIGN PROCEDURE"

When selecting a Phoredox or UCT process for excessP removal, it is necessary to establish whether or notnear complete denitrification can be achieved for theselected sludge age (Rs) and maximum specific growthrate of the nitrifiers at 20°C ( ^ ^ K if near completedenitrification can be achieved for sludge ages less thanabout 25 days and total anoxic sludge mass fractions(fxai) less than about 0,40 at the minimum temperature{Tnitri), then a Phoredox process is indicated; if it cannotbe achieved, then a UCT process is indicated.

For the P removal processes, an anaerobic reactoris required for the stimulation of excess P removal, sothat the maximum anoxic sludge mass fraction availabiefor denitrification (fxdm), is the difference between themaximum unaerated sludge mass fraction {fx r j and theselected anaerobic siudge mass fraction (fxa) i.e.

f. = f - f (7.8)

where fK^ is given by Eq. 5.24 for a selected Rs, nFTl2o. Sf

and Tmin.The denitrification behaviour of the Phoredox pro-

cess is identical to that of the Bardenpho process, i.e.Eqs. 6.24 to 6.29 apply. Note however, that fl(1:r, for thePhoredox process is given by Eq. 7.8 whereas f,.,,,. for

'The design procedure described in this section has been published in a shorter form by Ekama, Siebriiz and Marais (1982),

7-17

The Bardenphc process is given by Eq. 6.23. Hence forthe same unaerated sludge mass fraction f,_, thePhoredox process has a lower fxdrT. than the Bardenphoprocess, by an amount equal to fx5. Now frcm thebehaviour of the Bardenpho process 'see Chapter 6,Section 7), it was shown that when the wastewatercharacteristics are favourable, in particular, a high readi-ly biodegradable COD fraction if^J, a high fjnrT.i?0 valueand a lowTKN. COD ratio, near complete denitrificationcan be achieved at sludge ages around 20 to 25 days(see Figure 6.17). Consequently for favourable waste-.vater characteristics near complete denitrification, andhence excess P removal can be achieved in thePhoredo.; process 'because the effluent nitrate concen-tration is low and hence little nitrate is discharged to theanaerobic reactor}. The excess P removal attainable, P,j(is calculated with the aid of the parametric modeldescribed above and kncwmg P5, the effluent P concen-tration P,£ given by the difference between the influentP concentration P,, and P., i.e.

'ft) (7.9)

In the design of a Phoredox process for a selectedsludge age, if it is found that the total ancxic sludgemass fraction for complete denitrification fMJl is lessthan the maximum allowed f,,are open:

(1) f,-, and fx3 can be increased to give a total anoxicsludge mass fraction ?qua! to fKr1m which introducesa factor of safety for denitrification, the magnitudeof which is equal to f,dr-./f,(jI.

!2? The sludge age can be reduced so that f>dm

becomes equal to f.,.;. The lower sludge age willallow a saving in process volume (see Figure 4.3}.This lower sludge age Rs, can be estimated asfollows: set f,,,,m equal to fv;:. required for completedenitrification at T^,.,; determine f,m from Eq. 7.8;with fxn, calculate R5 from Eq. 5.24; to check repeatstep by step procedure for the new R,.

(3) For the Phoredox process, f>e can be increased toimprove excess P removal, so that fvdrn = f,a! fromEq. 7.8.

Options (2} and (3) above are not recommended asthese exc!ude a factor of safety in denitrification: Exclu-sion of a factor of safety on denitrification for thePhoredox process may have serious consequencesbecause the discharge of any nitrate to the anaerobicreactor reduces the P removal attainable in the process(Figure 7.15).

For design purposes where the Phoredox process isto operate between temperatures L r i and Tm5., thefollowing sequence of calculations needs to be made tocheck whether or not complete denitrification can beachieved in a Phoredox process and, if so, to select theoptimal process configuration for N and P removal:

Step 1: Select wastewater characteristics; S.t, Ntj, Pt(,

Siep

Step

Step

Step

Step

Step

Step

Step

Step

J :

4:

5:

6:

7:

8:

9:

10

11

Step 12:

the following options Step 13:

Calculate f,T for T_,. from Eq. 5.24.

With f, , and Rc calculate Sf for T^.s, from Eq.5.24.

Estimate N... for T ^ , and T^!rv

With fup and fu£, calcuiate Nb for T ^ and T ^for selected Rs from Eq 4-23.

Calculate Nc from Eq. 5.29

Select f/a.

Calculate Dz, from Eq. 6.25 for T,,ax and forTn.>, with fld,n given by Eq- 7.8.

Select s, 0^ and Os.

Calculate f7l.»,m from Eq. 6.21 for Tmax. and forTV.,,

Calculate fj3ni;n from Eq. 6.29 for T-1B11 and for' mm-

Select the a-recyc!e ratio and calculate Nne forTmJnfrom Eq. 6.24.

Step 14: If IM,,e<0J set N,,e = 0.

Step 15: With N.,,, calculate t,6.26 to 6.28.

f,3 and f^r, from Eqs.

-.iX' T,.,;

Step 2: Select Rs and Sf.

Step 16: Check that f^ > fxlmm and f>3 > fl3n.;r,. If not,discard selected a-recycle ratio as invalid.

Step 17: Repeat steps 12 to 15 for different a-recyc!eratios.

Step 18: Repeat steps 12 to 16 for T , ^ .

Step 19: Fix f,, and f,3 by procedure given in Chapter 6,Section 6.2. This fixes Nnfc and hence Nns.

Step 20: Select O.

Step 21: Calculate ASbs? from Eq. 7.2b.

Step 22: Calculate Sl/sa from Eq. 7.1b.

Step 23: Calculate P, from Eq. 7,5.

Step 24: Calculate y from Eq. 7-7

Step 25: Calculate Ps from Eq. 7.6.

Step 26: Calculate P,e from Eq. 7.9.

Step 27: Is P-e sufficiently low? If not, decrease Rs orincrease f,a at expense of f ,^; however, im-provement will only be achieved if fxdt < fKdm

i.e. Nne < 0,0; if fxdl = f,dm i.e. Nne > 0,changes wili increase P;e.

7-18

Step 28: Repeat calculations for TV,-,.

Step 29: Can required PIfc be achieved with thePhoredox process? If not, select a UCT pro-

Once fta and Rs are selected, the denitrification canbe calculated by applying the design equations for Nremoval in the MLE process, except that f dm is given bythe difference between f>r_, and fia (see Eq. 7.8).

For design purposes, where the process is tooperate between temperatures Jmm and TV,^, thefollowing sequence of calculations is recommended tofacilitate selection of the optimal process configurationfor N and P removal in the UCT process:

cess.

When the required P removal cannot be attained inthe Phoredox process due to incomplete denitrification,The UCT or Modified UCT processes are the indicatedprocesses." The configuration of the UCT processes issuch that the conditions in the anaerobic reactor are in-dependent of the denitrification behaviour forTKN/COD ratios up to about 0,13. Hence in theirdesign, the P removal can be estimated from the condi-tions in the anaerobic reactor without the need to knowthe extent of denitrification i.e. N and P removal can bedealt with separately in the design of these processes.

In the design for excess P removal in the UCT pro-cess, the readily biodegradable COD concentration inthe anaerobic reactor (ShS5) can be found directly fromEq. 7.1a and 7.2a without the need for selecting Rs orMnn-,2d: Sij!>i3 depends only on the influent COD strength(SJ, the readily biodegradable COD fraction (fbs) andthe r-recycle ratio from the anoxic to the anaerobic reac-tor. The r-recycle ratio usually is selected at 1:1; higherratios lead to greater dilution of S,.,. with a concomitantreduction in P removal, and lower values lead to inor-dinately large anaerobic reactor volume fractions withrespect to the anaerobic sludge mass fractions (see Eqs.7.10 to 7.11). Selecting r = l and the nitrate (Nn,) andDO (O,) concentrations in the r-recycle of say 1 mgN//and 1 mgO/i to include a factor of safety for excess Premoval, then Shs,a can be calculated from Eqs. 7.1a and7.2a and will remain unchanged irrespective of thetemperature. Knowing Sb&a, the propensity factor P, canbe calculated from Eq. 7.5 for a selected anaerobicsludge mass fraction f,c], and from the Pf value, the ycoefficient can be calculated from Eq. 7.7. Knowing y,the P removal from the particular wastewater can becalculated from Eq. 7.6 for a selected sludge age (Rs).The P removal should be determined at Jmb> because,for a constant S,,sa or y, the P removal at T,,ias is less Step 13: Estimate N,e for T ^ and Tmin.

Step 1: Select wastewater characteristics; S(1, N,,, Pll#

't>&< t j p . T u s , ^nm20< ' rnyx- • min-

Step 2: Select r (usually 1,0) and Nnr and O, (say1 mg/ i each).

Step 3: Calculate S l l t i l from Eqs. 7.1a and 7.2a.

Step 4: Select fxa (between 0,10 and 0,25).

Step 5: Calculate P, from Eq. 7.5 and y from Eq. 7.7.

Step 6: Select R&.

Step 7: Calculate Ps from Eq. 7.6 for T..irtl, and for Tmjn.

Step 8: Calculate P.t. from Eq. 7.9 for T.,,., and for' fTiilT

Step 9: Is P;e adequate? If not, repeat steps 4 to 8 fordifferent fxa and/or Ra {see Design ChartFigure 7.18).

Step 10: Select S,.

Step 11: Calculate f(n, for Tn,,n from Eq. 5.24.

Stop 12: With fM11 and Rs calculate S, for Tniil, from Eq.5.24.

than at TrT1(ri (Figure 7.14). Hence, for various values off,a, the P removal at different sludge ages can becalculated, and estimates of fxa and Rs giving thedesired or optimal P removal can be found. In selectingfvi, it is recommended that f,a should not be less than0,10 and not greater than 0,25: for f,.d < 0,10, the Premoval usually will be inadequate even at relativelyshort sludge ages for normal wastewaters, particularlyfor settled wastewaters; for f,0 > 0,25 no UCT ex-perimental response data ate available so that theparametric model has not been calibrated to include theprocess behaviour for f>a > 0,25." At fxa = 0,25, theprocess has been found to operate satisfactorily (Sie-britz et at. 1983). With regard to selecting R£,cognizance must be taken of its effect on the maximumunaerated sludge mass fraction fk,n (Eq. 5.24) because,with f a fixed, fxn, fixes the maximum anoxic sludgemass fraction f<t, . (Eq. 7.8) which, in turn, governs thedegree of denitrification that can be achieved.

Step 14: With fljp and f^, calculate Ns for Tn.s>. and Tmin

for selected Rs from Eq. 4.23.

Step 15: Calculate N, from Eq. 5.29.

Step 16: Calculate DPJ| from Eq. 6.25 for TniS, and forT,:,,r, with f,d.,. given by Eq. 7.8.

Step 17: Select s, Oa and Os.

Step 18: Calculate ao from Eq. 6.30 and Nnf. from Eq.6.31.

Step 19: Are Ple and Nne adequate? If P1(. is too high, in-crease f,a or reduce Rs - these changes willincrease Nne; if PIt is too low, decrease fxa orincrease Rc - these changes will decreaseNne.

"It is possible IO design a Modified Phoredox.'Modified UCT, UCT process combination (see iater in this section).

"Laboratory and pilot scale data on process behaviour for f,,a > 0,25 are available from the NIWR (see Interim Design Guide by ISIIWR).

7-19

Step 20: Repeat from steps 4 to 17 until required or op-timal P,e and Nr.e is obtained.

Having completed the design for the UCT process,if the influent TKN/COD ratio is less than about 0,11 to0,12 mgN/mgCOD, the process may be converted to amodified UCT process (Figure 7.6). The conversion isdone by subdividing the anoxic sludge mass fraction*f*orr* i m o two sub-fractions; the first is usuallyallocated a siudge mass fraction i,c- of about 0,10, andthe second, f,d2, having the remaining sludge mass frac-tion i.e. fy02

= Utm ~~fxdi• Having made this subdivision,the possibility of operating the modified UCT process(Figure 7.6) as a UCT process (Figure 7.5) is not exclud-ed; by making provision that the r-recycle flow to Theanaerobic reactor can be taken from either the first orthe second anoxic reactors, the process can beoperated as either a modified UCT or a UCT process'Figure 7.16). The advantages of operating the processas a modified UCT process were set out in Section 2.2.3above (see also Chapter 3, Section 3.3.3).

It is necessary to check whether or not themodified UCT/UCT process combination can beoperated as a modified UCT process. For successfuloperation of the modified process, the denitrificationpotential of the first anoxic reactor must be greaterthan, or at least equal to the equivalent nitrate load onthe reactor at the minimum expected temperature(Tmm). The denitrification potential is calculated bysubstituting f,d1 = 0,10 for fxl in Eq. 6.20. The equivalentnitrate load is found from the effluent nitrate concentra-tion (Nne) from the UCT process, the s-recyc!e ratio andthe DO concentration in the s-recycle (Os) i.e. theequivalent nitrate load is s(Nne + Os/2,86). ForTKN/COD ratios < 0,12 mgN'mgCOD, it"will be foundthat the denitrification potential is sufficient to denitrifythe equivalent nitrate load with a reasonable factor ofsafety (S1d) of say 1,3, where Sfd is the ratio of thedenitrification potential and the equivalent nitrate loadon the first anoxic reactor.

By setting fxdl = 0,10, usually it will be found thatthe reactor is sufficiently large for complete utilization ofthe influent readily biodegradable COD (S t,J. This canbe checked by means of Eq. 6.21. If Tn.,in is less than14°C, it may be found from Eq. 6.21 that 0,10 is insuffi-cient for complete utilization of the Sbw. In this event,f,d1 can be increased slightly to a maximum of 0,12.

ANOXIC AEROBICREACTOR REACTORS RLACTUR

MIXED L15U0R RECYCLES

WASTE FLOWSETTLER

INFLUENT GS ^ "•GY/7 \ j T ErFLUENT

OPTIONAL RECYCLE

FIGURE 7.16. The modified UCT. UCT process combination forbiological P and N removal.

Values of f>a, > 0,12 are not recommended because fortemperatures > T~,n, the reactor will be over-designedleading to a reduced nitrogen removal efficiency of theprocess. By designing the first anoxic reactor (ffcCi) toallow the possibility of complete or near completeutilization of the SbSi, the process can achieve excess Premoval up to the TKN. COD ratio at which St,s, is com-pletely utilized. This TKN/COD ratio can be calculatedby trial and error by testing increasing TKN'COD ratioson the process.

Generally it will be found that the higher theTKN/COD ratio, the lower the factor of safety Std forcomplete denitrification of the equivalent nitrate load onthe first anoxic reactor. When Sfd < 1,1, the prospectof operating the modified UCT/UCT process combina-tion (Figure 7.16) as a modified UCT process has littlemerit. This limit is strongly dependent on the influentreadily biodegradable COD fraction fbs: For normal fbs

ratios in raw wastewaters of 0,20 to 0,25, it will befound that Sfa is less than 1,1 for TKN/COD ratiosabove 0,11 to 0,12 mgN. mgCOD at a minimum tempera-ture of 14CC. When the frjs fraction and TKN/COD ratioof the influent are such that operation of the modifiedUCT process is not possible then it must be operated asa UCT process. In the UCT process, if the a-recycleratio is too low to produce an actual anoxic retentiontime lower than 1,5 hours, poor sludge settlingcharacteristics may be obtained, and should be takeninto account in the design of the secondary settlingtanks (see Chapter 8).

When the influent readi!y biodegradable COD frac-tion and TKN/COD ratio are such that operation of amodified UCT process is possible, the mixed liquora-recycle ratio is set at the greater value of two lowerlimits i.e. (1) that which loads the second anoxic reactorto its denitrification potential or (2) that which will pro-duce an actual anoxic retention time of less than 1 hour.The first limit is calculated with the aid of Eq. 6.30 ex-cept that DPP is replaced by the denitrification potentialof the second anoxic reactor, which can be calculatedby substituting the anoxic sludge mass fraction of thesecond anoxic reactor fxd2 for fxl in Eq 6.20. The secondlimit is found by calculating the volume of the secondanoxic reactor as set out below (see Section 7.5 above)and with the specified s-recycle ratio and influent flowrate, finding the a-recycle ratio which would produce anactual retention time of 1 hour.

A problem with the design of a nutrient removalprocess is that at the design stage the wastewatercharacteristics, in particular the TKN/COD ratio andmaximum specific growth rate of the nitrifiers at 20°CVntr.2G*' a r e n o t sufficiently accurately known to designan optimal nutrient removal process. With regard to theHnm2o value little can be done (besides measuring it, vanHaandel (1981) and Sehayek (1981) with Mara is) otherthan selecting a conservative value. However, withregard to the TKN/COD ratio virtually complete processflexibility can be provided for, so that depending on theTKN/COD ratio, the process can be operated as aModified Phoredox, Modified UCT or UCT process. Aschematic diagram of a process with this flexibility isgiven in Figure 7.17 (incorporation of the 5 stagePhoredox is difficult due to the problem of having toconvert the secondary anoxic reactor to an aerobic

7-20

ANAEROBICREACTORS

ANOXICREACTORS

AEROBICREACTOfl

INFLUENT

MIXED LIQUORRECYCLE

1 1

;. .

MIXED LIQUORRECYCLE

SLUDGE

A

1—.JRECYCLE

WASTE^FLOW

SETTLER

"V7

s

r wEFFL

*

OPTIONAL RECYCLES

MODIFIED UCT PROCESS

UCT PROCESS

MODIFIED PHOREDOX PROCESS

FIGURE 717: Schematic process configuration layout showingadditional and optimal recycles and flows allowing operation

as a Modified Phoredox, Modified UCT or UCT process.

reactor). Note in Figure 7.17 that the anaerobic sludgemass fraction is subdivided into 4 equal sized compart-ments in series. The reason for this is twofold:

(1} From a practical point of view, from the sameanaerobic reactor volume, the UCT process has ananaerobic sludge mass fraction half that for theModified Phoredox process. Consequently in thedesign, the anaerobic volume must be such thatwhen the process is operated as a UCT process,the required anaerobic sludge mass fraction is ob-tained. If then the process is required to beoperated as a Modified Phoredox, the first and se-cond compartments can be bypassed therebyallowing the same anaerobic sludge mass fractionto be used for the Modified Phoredox process. Itshould be noted that if the anaerobic volume is notreduced when converting to a Modified Phoredox,the aerobic sludge mass fraction will be reduced,and this reduction may be detrimental for achievingnitrification, particularly at the lower temperatures.The equations for calculating the relationship be-tween the volume fractions and sludge mass frac-tions for the different processes are given in Sec-tion 7.5 below.

(2) From a theoretical point of view, the kinetics ofbiological excess P removal are such that for thesame total anaerobic sludge mass fraction subdivi-sion into a number of equal sized completely mixedcompartments in series is more efficient than asingle completely mixed reactor (see Section 3.4abovel. Theoretically the greater the number ofcompartments in series the better, but practicallyvery little advantage is gained with more than 3. Inthe proposed composite configuration (Figure7.17), the anaerobic zone comprises 4 compart-ments when operated as a UCT process and 2when operated as a Modified Phoredox. With theModified Phoredox consideration could be given tcutilizing three of the four anaerobic compartmentsbut caution should be exercised in not increasingthe anaerobic sludge mass fraction too much

because its increase is at the expense of the aerobicsludge mass fraction with the result that nitrifica-tion and/or phosphorus uptake may be adverselyaffected (see Chapter 10).

The design calculations for the Phoredox and UCTprocesses set out above will be demonstrated with theaid of worked examples after the presentation of adesign chart for graphically estimating the excess Premoval attainable.

6. DESIGN CHART

The excess biological P removal theory described inSection 3 above can be summarized in a design chart(see Figure 7.18). For a raw or settled wastewater, thechart is constructed as follows:

(i) In the bottom right hand quadrant, a graph of totalinfluent COD concentration (S,,) versus readilybiodegradable COD concentration (Shsi) for variousreadily biodegradable COD fractions (fbs or f.J isplotted from Eqs. 2.8a or 2.9a.

(ii) In the top right hand quadrant the propensity factorPf versus Sbsl is plotted for different anaerobicsludge mass fraction (fxa) and fixed recycle ratio tothe anaerobic (r or s) assuming say no nitrate ordissolved oxygen is discharged to the anaerobicreactor from Eqs. 7.1 or 7.2 and 7.5.

(iii) The left hand side of the P, axis is rescaled in termsof y by means of Eq. 7.7.

(iv) In the upper left hand quadrant a graph of Premoval per unit COD (P£/S,,) versus y is plotted fordifferent sludge ages for fixed temperature fromEq. 7.6.

(v) In the bottom left hand quadrant, a graph of Ps ver-sus Sh is plotted for fixed removals of P in mg/jf.

7-21

The diagram constructed by the above procedure isvalid for the selected wastewater characteristics fb, fuS

and tempereture. An example of such a diagram isgiven in Figure 7.18 for the typical raw wastewatercharacteristics given in Table 4.3. In the construction ofthe diagram, it is assumed that no nitrate (or dissolvedoxygen) is discharged to the anaerobic reactor. Conse-quently the chatt is valid for (i) the Phoredox process,provided that complete deniuification can be achievedso that the nitrate concentration in the under flow recy-cle is zero; <\\) ihe UCT process, provided the mixed

liquor a-recycle ratio is carefully controlled so that itdoes not overload the primary anoxic reactor withnitrate; and (iii) the modified UCT process, provided theeffluent nitrate concentration is sufficiently low so thatthe underflow recycle does not overload the first anoxicreactor with nitrate.

The design chart is used as follows: For the rawwastewater characteristics given in Table 4.3 (i.e.f._.-0,13, f., =0,05) and taking a temperature of 20cC,draw a horizontal line at the influent COD concentrationS, of 600 mg COD// (line AB). Now the readily biode-

ANAEROBIC SLUDGEMASS FRACTION

RECYCLE RATIOTO ANAEROBICREACTOR ( s or r

,0

- P REMOVAL / mo 1NFL. COD

90 !2O 160 200 24OREADILY BIODEGRADABLE

COD CONCENTRATION( mg COD/I)

1000

FIGURE 7 18: Design chart for graphically estimating the ex-cess P removal attainable in nutrient removal processes a: 14rCtreating raw wastewater provided no nitrate is discharged to

the anaerobic reactor.

7-22

gradabie COD fraction with respect to total COD (fts) is0,20 (i.e. ft.s=0,24, the fraction with respect to thebiodegradable COD).

Moving lo the right: where line AB cuts thef^= 0,20 line, draw a vertical line (line CD). Moving up-wards along line CD; where line CD cuts the selectedanaerobic s'udge mass fraction line (say 0,10), draw ahorizontal line (line EF). Moving left along line ER theintercepts with the P, and y axes give the values of theseparameters. Where line EF cuts the selected sludge ageline (say 25 daysi, draw a vertical line time GH). Movingdownwards along line GH; the intercept with the Ps/Stl

axis gives the value of this parameter, and moving fur-ther down, the P removal in mgP/i is given at the inter-cept of lines GH and AB i.e. 8,8 mgP/jf. Hence the Premoval for a raw wastewater with S,.^600, f.s = 0,20(f^ = 0r24), fK6 = 0,10 and Rs = 2 5 d ' a t T = 20GC is8,8 mgP/'I. Clearly, with the aid of the diagram, the ef-fect of changing Rt, fxa and ftt (fbs) on P removal canreadily be ascertained (subject to the provisos mention-ed above for the different processes}.

7. DESIGN EXAMPLES

The design of excess P removal in nutrient removal pro-cesses by means of the procedure set out above isdemonstrated by continuing the numerical examplegiven in Chapter 4, Section 12 (in which the design pro-cedure for organic material removal was demonstrated),Chapter 5, Section 6 (in which the behaviour of nitrifica-tion was demonstrated) and Chapter 6, Section 7 [inwhich nitrogen removal was demonstrated). The waste-water characteristics used in the design example, whichare those of a typical raw and settled municipal waste-water, have all been defined and are given in Tables 4.3,5.2 and 6.1. The causes and effecis of variability insome of these characteristics, such as temperature,denitrification rates, readily biodegradable COD fractionetc. have been discussed in detail in Chapter 6, Sections7.1 and 7.2.

Many of the decisions related to the design of ex-cess P removal in a nutrient removal process are con-tingent upon the nitrogen removal behaviour of the pro-cess. Consequently, in the design example continuousreference to the design examples for nitrogen removalwill be made, i.e. Chapter 6, Section 7 and a thoroughunderstanding of this section, and nitrogen removal ingeneral, will be presumed.

7.1 The Phoredox process

In the calculations for nitrogen removal in the Barden-pho process treating the raw wastewater at 25-daysludge age and 14°C, it was found that a total anoxicsludge mass fraction (f,d,) of 0,40 was required toachieve complete denitrification (see Table 6.3). Nowthe total unaerated sludge mass fraction (fyrr) wasfound to be 0,50. Hence the largest anaerobic sludgemass fraction (f,a) to effect excess P removal in aPhoredox process, (which requires near completedenitrification to achieve good excess P removal) (seeChapter 3, Section 3) is 0,50-0,40 = 0,10 (from Eq.7.81. Because complete denitrification can be achieved

with f iat = 0,40, the nitrate concentration in theunderflow recycle (Nnb) is zero, and, taking the DO con-centration in the underflow recycle (Os) as 1 mgO/f,the readily biodegradable COD in the anaerobic reactor(SUs) for an underflow recycle ratio (s) of 1 is foundfrom Eqs. 7.1b and 7.2b, i.e.

AS^ = 1(8,6.0-3,0.1) + 0

= 3

Sbsa - {0,24(1-0,13-0,05) 600-3}/{1-r1)

= 57,5mgCOD/f.

The propensity factor P( is found from Eq. 7.5 i.e.

P, = (57.5-25).0,10 = 3,25 mgCOD, f

and the coefficient of excess P removal y is found fromEq. 7.7 i.e.

y = 0,35-0,29exp (-0,242.3,25)

- 0,218 mgP/mgVASS.

Hence the P removal at 25-day sludge age and 14&C isfound from Eq. 7.6 i.e.

,20.0,20.25)

+ 0,015 0,13/1,48}

= 600 {0,0615 (O,218 + O,O15KO,OO13|

- 600.0,0156

- 9,4 mgP/jP.

The influent P concentration is 10 mgP/i so that with9,4 mgP/f removal, the effluent P concentration Plf, is10,0-9,4 = 0,6 mgP/i (see Eq. 7.9). Similarly, the Premoval at 22°C is found to be 8,0 mgP/f. Note that SbM,P, and y remain unchanged at different temperatures butPs is decreased because b(l in Eq. 7.6 is increased to 0,25/dat 22°C (see Table 4.1). The important design results arelisted in Table 7.1.

In the design of the Phoredox process above, f 3 wasset at the maximum value leaving a just sufficiently largefxdl to attain complete denitrification. This leaves thePhoredox process in a critical state because to achieve theestimated P removals, complete denitrification should beachieved (see Figure 7.15). However, this cannot beguaranteed because there is no factor of safety on thedenitrification, To introduce a factor of safety on thedenitrification two options are open; (i) the primary anoxicsludge mass fraction (fyl) can be increased, but this willhave to be at the expense of a reduction in fxa (because fjmis fixed at 0,50 to achieve nitrification) but a reduction in fxa

will cause a reduction in P removal - from the designchart at 20cC (Figure 7.18) a reduction in fxa from 0,10 to0,07 reduces the P removal from 8,8 to 7,7 mgP/f, (ii) thesludge age can be increased allowing a larger unaeratedsludge mass fraction (f,m), and hence, with fX3 fixed at0,10, allowing a larger anoxic sludge mass fraction (fxom).However, the sludge age not only requires a larger process

7-23

TABLE 7.1 SUMMARY OF DESIGN CALCULATIONS FORSETTLED WASTEWATER AT 25-DAY SLUDGE

Parameter

Temperature

Safety factorMax. unaergted sludge mass frscjonEffluent TKNN for sludge productionNitrification capacityAnaerobic sludge mass fractionMax. anoxic sludge mass fractionMax. de nitrification potentialDO in a-recydeDO in s-recycle

Underflow recycle ratioPrim, anoxic sludge mass fractionSec. anoxic sludge mass fractionTotal ancxic sludge mass fractionOptimum a-recycie ratioEffluent nitrateNitrate recycled 10 anaerobic reactor

Symbol

T

s.f.-r

Ns

K1

f*rtm

ow0,

°\0-1cf . i

Uvao

KAr\L. I

Readily biodegradable COD in anaerobicreactor

Propensity factoiy coefficient

P removalFffluent P concentration

"These are not the calculated optimum

Sbsa

P.

PS

Pw

values but a

THE PHOREDOX AND UCT PROCESSESAGE (WASTEWATER CHARACTERISTICS

Units

°C

mgN.'imfiN/fmg.N/J

mgN/fmgO'f

mgCM

mgW/i

mgN/i

mgCOD/imgCOD'f

mgP/mgVASS

mgP/imgP//

;e values at

6 D

Raw WastewaterPhoredox UCT

14

1,250,503,012,632,40,100,4037,22,0

1,0

1.00.160,0240,403,30.0

0,0

573,3

0,229.40,6

22

2,70.502,012.034,00,100,4049,72.0

1,0

1.00,160,0230.404,20,0

0,0

573.3

0,228,02,0

14

1.250.503,012,632.40,100,4037,22,0

1,0

1,00,40--0,405,0*4,6

0.0

573,3

0,229.40,6

the practical and economics

22

2.70,502,012.034,00,100,4049,72,0

1.0

1,00,40-0,405,0-4,8

0,0

573.3

0,223,02.0

limit

GIVEN INTREATING RAW ORTABLES 4 3, 5.

Settled WastewaterPhoredox UCT

14

1,250.503,05.732,30,100,4027,22.0

1.0

1,00.260,0140,401,38,0

8.0

16,50,00

0,061.96,6

22

2,70,502.05.433,60.100,4035,42.0

1,0

1,00,260,0140,403,04,8

4,8

30,00,50

0,0932,36.2

14

1.250,503,05,732.30,100.4027.22.0

1,0

1,00,40—0,402,57,2

0,0

50,82,57

0.1955,13,4

see Chapter 6, Section 7.5).

2 and

22

2.70,502.05,433.60.100.4035,42,0

1.0

1.00,40—0,40s.o-4.8

0,0

50,82.57

0.1954,34,2

volume, but also reduces the P removal - from the designchart (Figure 7.18), an increase in RE from 25 to 30 days(which increases f,:j. from 0,40 to 0,43) reduces the Premoval from 8,8 mgp/i to 7,6 mgP/I. Clearly the rawwastewater influent TKN/COD ratio (i.e. 0,08) is just onthe limit for successful implementation of the Phoredoxprocess - any increase in TKN/COD ratio will result insignificant concentrations of nitrate in the effluent andunderflow recycle (> 3 mg!M/i} which, when dischargedto the anaerobic reactor, will cause a deterioration in Premoval (Figure 7.15).

From the above, it is clear that the TKN/COD ratioof the settled wastewater (i.e. 0,114} is too high to at-tain excess P removal in the Phoredox process. This isdemonstrated in the calculations below: For this pro-cess operating at 25 day* sludge age with ^ = 0,10'and hence fxdm = 0,40, from the denitrification theory anNne of 4,8 mgN/ i is obtained at 22°C with fyl = 0,26,fx3 = 0,14, ao = 3,0 and Nne = 8,0 mgN/f at 14°C withao = 1,3 (see Table 7.1). At 14"C, with 8 mgN/i nitratein the effluent and underflow s-recycle, and taking thes-recycle ratio to be specified at 1:1, and containing aDO concentration (Os) of 1 mgO/i , then from Eqs 7.1band 7.2b, the readily biodegradable COD concentrationin the anaerobic reactor (Sbsa) is only 16,5 mgCOD/f,which is too low for stimulating excess P removal.Hence from Eq. 7.5 the propensity factor (P«) is 0,0 andfrom Eq. 7.7 the / coefficient is at its minimum of

0,06 mgP/mgVASS. From Eq. 7.6, with y - 0,06, the Premoval (Ps) at RE = 25 and T=14°C is 1,9 mgP// ,which, from Eq. 7.9, with PTi = 8,5 mgP// gives an ef-fluent P concentration (Ple) of 6,6 mgP/jf, (see Table7.1). A similarly poor P removal is obtained also at 22°C(see Table 7.1 J. Clearly when the wastewater character-istics are such that near complete denitrification cannotbe obtained, the Phoredox process will not function asan efficient biological P removal facility.

7.2 The UCT processes

In the UCT processes, the complicating effect of thepossibility of a discharge of nitrate to the anaerobicreactor is absent. Consequently, in their design, Thenitrogen and phosphorus removal aspects can be dealtwith separately and the P removal in the processes canbe calculated without first having to determine the ef-fluent nitrate concentration.

In the design of the UCT or modified UCT pro-cesses, first the UCT process (Figure 7.5) is designedand once this is complete, a check is made as towhether or not this configuration can be converted to amodified process (Figure 7.6). It is recommended thatwhere it is possible To convert the UCT process to amodified one, the option of operating the modified pro-cess as a UCT process should always be included (seeFigure 7.16). Consider also the option of incorporating

'Selected the same as for the raw wastewater case so that a direct comparison can be made between the settled and raw wastewaters.

7-24

the Modified Phoredox/Modified UCT, UCT combina-tion (see Figure 7.17).

At the same sludge age, anaerobic sludge massfraction and temperature, the P removal attainable inthe UCT process is the same as that attainable in ihePhoredox process with complete denitrification i.e.when no nitrate is discharged to the anaerobic reactorsof both processes. Consequently, with the UCT processoperating at 25-day* sludge age with fxa = 0,10,* the Premoval from the raw wastewater is 9,4 mgP•'t at 14-Cand 8,0 mgP/f at 22CC Note also that for the sameprocess parameters Rs, f^, fxOm and temperature, thenitrification capacity Nc and denitrification potential DPP

for the UCT process are identical to those for thePhoredox process. The nitrogen removal behaviour ofthe UCT process is identical to that of the MLE processand hence the effluent nitrate concentration Nr,e andoptimum mixed liquor a-recycfe ratio a0 are found fromNcand 0ap with the aid of Eqs. 6,30 and 6.31; from Eq.6.30, a0 at both 14CC and 22°C is greater than the prac-tical limit of 5 and hence a0 is limited at 5 which givesNMfi at 14°C and 22°C as 4,6 mgNl/f and 4,8 mgN/£respectively from Eq. 6.31. The important design detailsemeiging from the above calculations are given in Table7.1. A comparison of the Phoredox and UCT processwill be given once the design of the UCT processtreating the settled wastewater has been demonstrated.

With the UCT process treating the settled waste-water at 25-day sludge age and an anaerobic sludgemass fraction of 0,10, the P removal is calculated fromthe theory (note that now, the P removal is not thesame as that in the Phoredox process because with thesettled wosiewater in this process, nitrate is dischargedto the anaerobic reactor). Taking the nitrate recycled tothe anaerobic reactor in the UCT process as 0,0 and anr-recycle ratio of 1:1 with a DO concentration (Or) of1 mgO/jf, the readily biodegradable COD concentrationin the anaerobic reactor (Sbsa) is found from Eqs 7.1aand 7 2a i.e.

AShs = 1 (8,6.0 + 3,0.11 + 3,0.0

= 3mg COD/i

Sbsa =|0,33.(1-0,08-0,04) 360-3}/(1 + 1)

= 50,8 mgCOD/l.

The propensity factor P. is found from Eq. 7.5 i.e.

P. = (50,8-25).0,10 = 2,58 mgCOD/7

and the y coefficient is found from Eq. 7,7 i.e.

Y = 0,35-0,29exp (-0,242.2,58)

- 0,195 mgP/mgVASS.

Hence the P removal P& at 25 days sludge age and 14CCis found from Eq. 7.6 i.e.

= 360 \~'(1 +-0,20.25)

= 5,1 mgP/f.

(0,195 + 0,015.0,20.0,20.25)

+ 0,015.0,04/1,48}

The effluent P concentration Ptfc is given by Eq. 7.9 i.e.

P,c = 8,5- 5,1 - 3,4 mgP/i.

Similarly, Ps and P1(i at 22DC are 4,3 and 4,2 respectively.

The nitrogen removal behaviour of the UCT processis found from Eqs. 6.30 and 6.31 - note that the nitrifica-tion capacity N(. and denitrification potential D^, withfA(lm = 0,40 are identical to those of the Phoredox process.At 14°C the effluent nitrate concentration Nnj;. and op-timum a-recycle ratio at, are 7,2 mgN// and 2,5 and at22CC, the calculated a0 is greater than the practical limit of5 and with a = 5, Nne = 4,8 mgN / /. Important designdetails of the N and P removal design are given in Table7.1.

7.3 Comparison of Phoredox and UCT processes

Comparing the Phoredox and UCT processes at 25-daysludge age and anaerobic sludge mass fraction of 0,10(see Table 7.1), it can be seen that for the raw wastewater,the Phoredox and UCT processes achieve the same ex-cess P removals, but the N removal is better in thePhoredox than in the UCT process; the former achievescomplete denitrification whereas the latter produces an ef-fluent nitrate concentration of 5 mgN/i1. However, thePhoredox process is in a critical state in that an unex-pected increase in TKN/COD ratio above 0,08 will result ina deterioration in excess P removal. In contrast the UCTprocess can accommodate unexpected increases inTKN/COD ratio up to about 0,13 (0,11 for the modifiedprocess), while maintaining its calculated excess Premoval of 9,4 mgP// at 14"C; this flexibility is at the costof producing a minimum effluent nitrate concentration of5mgN/f.

From the above results, the following general conclu-sion can be made: For wastewaters with TKN/COD ratiosand readily biodegradable COD fractions such that thePhoredox process can achieve good excess P removals, itseffluent nitrate concentration will be near zero. For thesame conditions the UCT process will also achieve goodexcess P removal but its effluent nitrate concentration willbe 5 to 6 mgN// .

In the Phoredox process, the good effluent quality isat the cost of process flexibility to accommodate unex-pected TKN/COD ratio increases; this flexibility is incor-porated in the UCT process but it is at the cost of a pooreffluent quality. In these situations (which usually arisewith wastewater of TKN/COD ratio less than 0,08 with 20to 2b% readily biodegradable COD fraction, fTJ, the pro-cess to select would depend on the variability inTKN/COD ratio of the wastewater.

Comparing the Phoredox and UCT processes treatingthe settled wastewater with a TKN/COD ratio of 0,114, itcan be seen that both processes produce approximatelysimilar effluent nitrate concentrations (see Table 7.1).However, with respect to excess P removal, the Phoredoxprocess can be seen to have a poorer P removal facilitythan the UCT process; the difference arises from the factthat in the Phoredox process, the exclusion of nitrate fromthe anaerobic reactor becomes impossible once near com-

'Selected the same so that a direct comparison between The Phoredox and UCT process can be made. A design with different Rs and fx6 to ef-fect better P removal (at the cost of poorer N removal) wilt be presented below.

7-25

ptete denitrification cannot be achieved, whereas in theUCT process, this is possible up to TKN. COD ratios ofabout 0,13 {depending on the fM (f,s) fraction).

7 4 Redesign of the UCT process

Although in the UCT piocess the exclusion of nitratefrom the anaerobic reactor is possible, thereby allowingthe greatest potential for P removal, the actual Piemovai attained in the process for the settied waste-water is insufficient to reduce the effluent P concentra-tion (P.€i below 1 mgP//. This is because the anaerobicsludge mass fraction (fxil; and siudge age (R,J have beenincorrectly selected to attain P.e < 1 mgP/f. To im-prove the P removal, Rs needs to be reduced and f,a in-creased. Ftom the design chart for raw wastewater'Figure 7.18). if f.a is increased from 0,1C to 0,15and thesiudge age decreased from 25 to 20 days, the estimatedP removal (at 20cC with no nitrate and DO recycled tothe anaerobic reactor) is 11,5 rngP/ .

Accepting an anaerobic sludge mass fraction of0,15 and the sludge age of 20 days, the design of theUCT process for raw and settled wastewater at 14°Cand 22°C was undertaken and the important design

results are listed in Table 7.2. By following the step bystep procedure, the catenations for the raw wastewaterat 14°C and 22°C is demonstrated: Accepting anr-recycle ratio to the anaerobic reactor (r) of 1, with1 mgr\)/£ nitrate {Nr,j and 1 mgO / dissolved oxygen(Of) (to introduce a factor of safety), the readilybiodegradable COD concentration in the anaerobicreactor (S,^) is found from Eqs 7.1a and 7.2a

AS:,S - 1(6,6.1^3,0.0)^3,0.0 = 11,6mgC0D7

S^.a = {0.24(1 -0,13-0,05) 600 -11,6}. <"; -t-11

- 53,2 mgCQD/f.

With ~ 0,15, the propensity factor P, is found fromEq- 7.5 i.e.

Pf - (53,3-251.0,15 = 4,25 mgCOD/i

The >• coefficier.t is found frcm Eq. 7.7 i.e.

y = 0,35-0,29exp (-0,242.4,25)

- 0,246.

With Rs = 20 d, the P removal P at 14CC and 22fC isfound from Eq. 7.6 i.e.

TABLE 7.2 SUMMARY OF DESIGN CALCULATIONS FOR THE UCT PROCESS OPERATING AT 20 DAY SLUDGE AGEWITH AN ANAEROBIC SLUDGE MASS FRACTION OF 0,15 TREATING RAW AND SETTLED WASTEWATERS

(WASTEWATER

Parameter

Temperature

Nitrate in r-recydfeDO in t-recycleInfluent biodegradable CODRap'dly biodegradable COD tractionr-recycle ratioRapidly biodegradable COD m anaerobicAnaerobic sludge mass fractionPropensity factor>• coefficier.tP removalEffluent PFacior of safetyr/&x. unaerstea sludaf mess fractionEffluent TKNTJ for sludge productionNitrification capacityMax. anoxic sludge mass fractionDenitrification potentialUnderflow recycle mileDO in a-recycleDO in s-recycleOpt. a-recycle ratioEffluent nitrate

UCT process modification1st Prim AnoxicOenitrification potentialEquiv. nitrate loedFactor of safetyModification successful

"The calculaied a- valuft n.e 9,3' is ureati--f

CHARACTERISTICS GIVEN IN TABLES 4.3, 5

Symbol

T

OrStji%•>

r

PfyPs

s,fxr-:N,eNs

N t

UamDpp5

O3

osBoN™

f,d1

DPdi

Sid

Units

•C

mgN/fmgO/fmgCOD'f——mgCOD/X

mgCOD /mgP/mgVASSmgP/fmgP/f

—mgN/ /mgN/fmgN'/—mgN/

mgO/£mgO. /—mgN//

—mgN-'fmgN'f—

than the practical limit of 5. Hence

2 AND 6.1]

RawWastewater

14

1.01,0

4920,241.0

53,30,154,240,246

12,20,01,250,423,0

13.231,80,27

29,01.02,01,03,65,7

0,1019,66,13,2

Yes

:• \s SL't f-Uual I

22

1,01.0

4920.241,0

53,30,154,240,246

10,40.02,90 /22,0

12,633,40,27

37,31.02,01,05,0*4,8

0,1021,84,94,5

o thr practica

SettledWastewater

14

1.01.0

3170,331,0

46,50,153,220,2176.71.81,250,423,06,1

31,90,27

22.01.02,01,00,93

10,9

0,1016,011.31,4

22

1,01.0

3170,331,0

46,50,153,220.2175,62,92,90,422,05,7

33,30,27

27,31,02,01,02,257,9

0\1018,08.32,2

Yes

lirnil of R

7-26

(1+0,20.20)

= 12,2mgP/X.

and at 22°C

(0,2464 0,015.0,20.0,20.201

•+ 0,015.0,13/1,48}

(l-r 0,015.0,20.0,25.20)

+ 0,015.0,13/1,48}= 10,4 mgP/f.

As both these removals are greater than the influent Pconcentration, the effluent P concentration can betaken to be zero.

Accepting a nitrification safety factor (Sf) of 1,25the maximum unaerated sludge mass fraction at 20-daysludge age, 14°C and ^nril2o = O,36/d is found from Eq.5.24. (Temperature dependencies of nitrification con-stants are given in Table 6.2], i.e.

f,m = 1-1,25 (0,034+1/201/0,18

- 0,42

With fvn = 0,42, SH at 22yC is found from Eq. 5.24, i.e.

S, = (1-0,42).0,454/(0,042+1/20)

- 2,9.

Because a reasonable S, at 14°C has beenselected, complete nitrification can be assumed to takeplace and hence effluent TKN concentration of3,0 mgN/f at 14°C and 2,0 mgN/ i at 22°C are accept-able. The nitrogen requirement for sludge production at14°C and 22°C is found from Eq 4.23 i.e. 13,2 at 14°Cand 12,6 at 22GC. Hence the nitrification capacity Nc isfound from Eq. 5.29 i.e. 31,8 at 14CC and 33,4 at 22CC.

With f^, and fva fixed at 0,42 and 0,15 respectively,the maximum anoxic sludge mass fraction (fldm) is givenby Eq. 7.8 i.e. fxdm = 0,42 -0,15 = 0,27. With f>dm = 0,27,the denitrification potential of the primary anoxic reac-tor with fxl = fkbm = 0,27 is found from Eq. 6.20 i.e.29,0 mgN/i at 14°C and 37.3 at 22°C. With Nc and Dpp

known and selecting the underflow s-recycle ratio as 1and the DO in the a- and s-recycles (Oa and Os respec-tively) as 2 and 1 mgO/f respectively, the nitrogen per-formance is found from Eqs. 6.30 and 6.31 i.e. the op-timum a-recycle ratio (acl} and effluent nitrate concen-tration (N,,e) at 14CC are 3,6and 5,7 mgN/f respectivelyand at 22CC the calculated ao is greater than the prac-tical limit of 5 (i.e. 9,3) so that ao is set at 5 givingNne = 4,8 mgN/f (see Table 7.2). For design details ofthe above procedure see Chapter 6, Section 7.5.

The results in Table 7.2 show that for the rawwastewater at 14°C and 22CC, more P can be removedthan is present in the influent so that an effluent P con-centration of less than 1 mgP/i can be obtained with afactor of safety. The effluent nitrate concentrations are5,7 mgN/i at 14cCand4,8 mgN.'f at 22CC. Comparingthese results with the Phoredox process ax 25-daysludge age (Table 7.1), it can be seen that Phoredoxprocess produces a better effluent quality. However,the UCT process incorporates a factor of safety on the Premoval whereas the P removal in the Phoredox processdepends on the attainment of near complete denitrifica-

tion without a factor of safety. Consequently, althoughthe UCT process yields a poorer N removal, the Premoval can be attained with greater surety in the eventof unexpected increases in the TKN/COD ratio. The up-per limit of ihe TKN, COD ratio can be calculated bytesting higher TKN COD ratios on the design and forapproximately normal wastewater is about 0,13.

For the settled wastewater, the P removal (P,.} is6.7 mgP' i at 14°C and 5,6mgP/f at 22DC, leaving1.8 mgP/i and 2,9 mgP/i in the effluent at 14°C and22°C respectively. The effluent nitrate concentrations(Nrj are 10,9 mgN/7 and 7,9 mgN/X at 14°C and 22°Crespectively. !f Ps is insufficient, then fx6 can be increas-ed to its maximum of 0,25, but this will result in a poorerN removal i.e. with fya-0,20, Ps = 7,5mgP// andNne=12,2 mgN/i at 14°C and Pj.^6,4 mgP/i andNne = 9,7 mgN/ i at 22CC. If the N removal withf,;. - 0,20 is insufficient, then the sludge age can be in-creased to improve the N removal but the increase in Rs

will reduce the P removal, (see design chart). Clearlythere are limits to which biological nutrient removal canbe achieved even in an optimally designed process.These limits, and hence the effluent quality, are depen-dent on the influent wastewater characteristics, in par-ticular the readily biodegradable COD fraction, theTKN/COD ratio and the P/COD ratio i.e. the higher thefirst and the lower the other two the better the effluentN and P quality. However, it should be noted thatbiological N and P removal can be improved by addingreadily biodegradable COD to the process (see Chapter3, Section 3). The effect of this external COD additionon the N and P removal in a process can be checked byappropriately adjusting the readily biodegradable CODfraction and influent COD concentration.

Having completed the design for the UCT process,it is necessary to check whether or not the process canbe operated as a modified UCT process (Figure 7.6).The conversion to the modified UCT process is done bysubdividing the anoxic sludge mass fraction in two sub-fractions; the first is usually allocated a sludge massfraction of f>d1 of about 0,10, and the second, f*d2,having the remaining sludge mass fraction i.e., fxd2

= fxc!rr.-f id i. For successful operation of the modifiedUCT process, the denitrification potential of f,,.2 (D[ld-t)must be greater than its equivalent nitrate load imposedby the underflow s-recycle at the minimum expectedtemperature Tniin. The equivalent nitrate load iscalculated from the effluent nitrate concentration forthe UCT process before subdivision and is given bys(N,lfc- Os.'2,86). The factor by which Dpal is greaterthan s(Nne + Os/2,86) is called the factor of safety (Sfa)and it is recommended that this factor be greater than1,2. (see Section 5 above).

The calculations for the modification of the UCTprocess treating the raw and settled wastewaters at 20days sludge age with an anaerobic sludge mass fractionof 0,10 are given in Table 7.2. For both the raw andsettled wastewaters the modification is acceptable — thefactor of safety S)d at 14°C for the raw wastewater is 3,2and for the settled wastewater is 1,4. Generally it will befound that the modification is acceptable for TKN/COD< 0,12 with a rpadily biodegradable COD fraction (fbs)around 0,25. The TKN/COD ratio limit for successfulapplication of the modification is strongly dependent on

7-27

The readily biodegradable fraction fbs - the lower thefraction the lower the TKN/COD ratio limit (see Section5 above). It should be noted also that the magnitude ofihe anaerobic sludge mass fraction fxa, plays a role as towhether or not a UCT process modification will be suc-cessful. For example, at 20 days sludge age withfxa = 0,20, the effluent nitrate concentration treatingsettled wastewater at 14CC is 12,2 mgN/i (as previous-ly calculated), giving an equivalent nitrate load of 12,6m g N i ; with Dpcl = 16,0 mgN/i fsee Table 1.2), thefactor of safety Si6~ 1,26, which is lower than that ob-tained with fxa = 0,15 (see Table 7.2), and near the limitwhere the modification would be acceptable.

When checking whether or not the modified UCTprocess can be successfully applied, it is necessary tocheck that the first anoxic reactor is sufficiently large toallow complete utilization of the readily biodegradableCOD. This is checked by means of Eq. 6.21 (see Section5 above). From Eq. 6.21, the minimum anoxic sludgemass fraction to completely utilize the readilybiodegradable COD at 14CC is 0,062 and 0,085 for theraw and settled wastewaters respectively (see Table6.3). Hence f(dl =0,10 is adequate. By designing f>dl toallow the possibility of complete utilization of the readilybiodegradable COD, the modified process can achieveexcess P removal up to the TKN/COD ratio at which theeffluent nitrate concentration is so high that all thereadily biodegradable COD is utilized. This TKN/CODratio can be calculated by trial and error by testing in-creasing TKN/COD ratios on the process.

Having made the modification to the UCT process,the possibility of operating it as a UCT process is not ex-cluded; by making provision that the r-recycle flow canbe taken from either the first or the second anoxic reac-tor (see Figure 7.15) the process can be operated eitheras a modified UCT or a UCT process. The advantages ofdesigning the process as a modified UCT process wereset out in detail in Section 2.2.3 above. Furthermorecomplete flexibility of process operation can be incor-porated into the plant by providing optional recyclesallowing the plant to be operated as a ModifiedPhoredox, Modified UCT or UCT process (see Figure7.17). Also, irrespective of the type of process or com-bination of processes that are selected for a particulardesign, it is strongly recommended that the anaerobicreactor is subdivided into three or four compartmentsoperated in series. The advantage of this series systemanaerobic reactor is given in Section 3.4 above.

7.5 Process volume

The volumes of the reactors of the Phoredox processare calculated by means of the method described inChapter 6, Section 7.6.1 and for the same sludge age,temperature and COD load, the total volume of aPhoredox process will be the same as that of theBardenpho process, i.e. subdivision of the processvolume between the different reactors is proportional tothe sludge mass fractions of the different reactors. Thereactor volumes and retention times for the Phoredoxprocess at 25-day sludge age (see data in Table 7.1) areset out in Table 7.3.

With the UCT process, and its modification theMLSS for MLVSS) is not uniform throughout the pro-

TABLE 7 3 REACTOR VOLUMES, NOMINAL ANDACTUAL RETENTION TIMES FOR THE

PROCESS TREATING RAW AND SETTLECAT 25-DAY SLUDGE AGE

SEEFOR PROCESSTABLE 7 1

PHOREDOXWASTEWATER

PERFORMANCE

Process Parameter Symbol Wastewater U n J t ERaw Settled

Sludge massAverage MLSS cone.Total volumeTotal flowTotal nominal retentionlimeMixed liquor recycleM4°C)Underflow recycleAnaerobic

Mass fractionVolumeNominal retention timeActual retention time

Primary anoxicMass fractionVolumeNom. retention timeActual retention time

Main aeration*Mass fractionVolumeNom. retention timeActual retention time

Secondary anoxicMass fractionVolumeNom. retention timeActual retention time

Reaeration1

Mass fractionVolumeNom. retention timeActual retention time

"The reaeration reactor is

MX, 54 200X, 4 000Vp 13 5500 13 330

Rhi 24,4

a 3,35 1.0

f,a 0,101 355

2.41,2

f . i 0,162 158

3,90,74

0,456 098

11,02,1

fx3 0,243 252

5,92,9

0,056781,20.6

21 850 kgTSS4 000 mg' /5 460 m3

13 330 m ' ' d

9,8 h

1,3 -1,0 -

0,10546 nv1,0 h0,5 h

0,261 420 m ;

2,6 h0792 h

0,452 457 m1

4,4 h1,34 h

0.14764 m !

1.4 h0,7 h

0.05273 m3

0.50 h0,25 h

usually taken to have a sludge massfraction of around 0,05 to 0,07 and this fraction is deducted fromthe aerobic sludge mass f 'action; the remaining aerobic sludgemass fraction constitutes the main aeration sludge mass fraction.

cess with the result that the reactor volume fractions arenot equal to the sludge mass fractions. There are twomethods for determining the volume fractions (orvolumes) cf the reactors:

(0 By calculating the reactor volume fractions fromthe mass fractions for a fixed total process volume;the anaerobic volume fraction is given by

a W } (7.10)

and the anoxic and aerobic volume fractions aregiven by

Wfxd - Wf«b = 1 - { W ( r + fM)} (7.11)

where

first subscript v and x refer to volume and massfractions respectively and second subscripts a, dand b refer to anaerobic, anoxic and aerobic reac-tors respectively.

7-28

Knowing the volume fractions, the volumes ofthe reactors are calculated from the total volume ofthe process for a selected mean process MLSSconcentration X. (see Eq. 4,16), For the selectedmean'MLSS concentration Xlt The MLSS concen-trations in the reactors are given by the followingequations; in the anaerobic reactor 'X.a),

X,. = X..f,,/f H (7.12)

in the anoxic (X,d) and aerobic (X.J reactors

(7,13}

From Eq. 7.13, it can be seen that the sludge con-centration in the anoxic (XId) and aerobic (Xtb)reactors is greater than the selected mean concen-tration X.. For r= 1 and f,a-0,15, XId and XIb are15% higher than Xt, This higher than mean aerobicreactor sludge concentration must be taken intoaccount in the design of the secondary settlingtanks, which receive the outflow from the aerobicreactor.

(ii) By calculating the reactor volumes from the sludgemass fractions for a fixed aerobic reactor sludgeconcentration Xlb; if the anaerobic sludge massfraction is fxa, then the remaining sludge(1 - f^lfVHXJ, is diluted into the anoxic and aerobicreactors i.e. the combined volumes of the anoxicand aerobic reactors is given by (1 - f*a)M(X;)/Xlb.The subdivision of this volume is easily found fromtheir sludge mass fractions because the sludgeconcentrations in these reactors ate the same. Nowthe sludge concentration in the anaerobic reactorXld is a fraction r/(1+r) of the concentration inthe aerobic reactor i.e. X ! a -X ; b r/(1 -r r). Knowingthe mass of sludge in the anaerobic reactor i.e.f** M(X;), and the concentration Xla the volume isgiven by f^ M(X,)/Xia. This method will yield alarger process volume than method (i).

Comparing the volumes of the UCT process attain-ed with the above two methods with volume of thePhoredox process, method (i) will produce a processvolume equal to that of the Phoredox for the samemean process concentration X, but the aerobic reactorsludge concentration in the UCT process will be greaterthan that in the Phoredox {by a factor (1 + fxa)}, andmethod (ii) will produce a process volume larger thanthat of the Phoredox {by a factor (1 +fsa)} for the sameaerobic reactor sludge concentration. Reactor volumes,retention times for the UCT process as found frommethod (ii» are give in Table 7.4.

7.6 Oxygen demand

The average daily oxygen demand for a nitrogen andphosphorus removal process is calculated in the iden-tical fashion as that for a nitrogen removal process. Thisis discussed in detail in Chapter 6, Section 7.6.2.

A simple design rule by means of which theaverage peak daily oxygen demand can be estimatedfrom the influent COD and TKN load variation over theday is also given in that section.

TABLE 7.4 REACTORACTUAL RETENTION

VOLUMES ANDTIMES FOR THE

PROCESS TREATING RAW AND

NOMINAL ANDMODIFIED UCT

SETTLED WASTEWATERAT 20 DAY SLUDGE AGE. FOR PROCESS PERFORMANCE

SEE TABLE

Process Parameter

Sludye massAerobic reactor MLSSconcentrationAerobic mass fractionAerobic sludge massAnoxic + Aerobic sludgemassAnoxic + Aerobic volumeTotal anoxic mass fraction

First primary anoxicSecond primary anoxic

Aerobic mass fraction1st Anoxic volume2nd Anoxic volumeAerobic volumer-recycle ratioAnaerobic MLSS cone.Anaerobic volumeTotal volumeAverage process cone.Opt. a-recycie ratio

Underflow recycle ratioInfluent flowRetention timesAnaerobic; Nominal

Actual1st Anoxic: Nominal

Actual2nd Anoxic: Nominal

Actual

Aerobic: NominalActual

* Based on the calculate!

Symbol

MlX:]

X;t,

f*dmf«Hf*d2

u

I

X,e

X,a0

sQ

optimum'* Based on a mixed iiquor recycle

retention time of aobut 1 hour.

7.2

WastewaterRaw Settled

54 200

4 0000,15

8 130

46 07011 518

0.270,100,170.58

1 3552 3047860

120004065

155833480

3,6

1.013 330

7,33,62,40,84,10,9

14,23,1

21 8bO

40000,15

3 278

18 57246430,270,100,170,58546

1 4753 168

120001 6396 2823480

0,93'/

' i .o13 330

3,01.51.00,32,7

1,4*/1 ,1 "

5,7

3,0'/

Units

kgTSS

mgTSS /—kgTSS

kgTSSm3

m3

m>m 1

mgTSS/tmJ

m>mgTSS/1

m'/d

hhhhh

hh

h

mixed liquor recycle ratio ao.ratio to yield an actual anoxic

8. GUIDELINES FOR PROCESS SELECTION

From the theory of biological excess P removal and thenitrogen removal behaviour of the various processes,the following broad guidelines can be given to assist theselection of a process:

8.1 Process selection

(1) If the readily biodegradable COD concentration inthe influent (Sbsj is less than 60 mgCOD/i (ir-respective of the influent COD concentration),significant excess P removal is unlikely to be ob-tained in any of the processes.

(2) If Sbs, > 60 mgCOD/f, excess P removal can beachieved provided nitrate can be excluded from theanaerobic reactor, the removal increasing as S.)S, in-creases. Whether or not nitrate can be excluded

7-29

from the anaerobic reactor depends on the influentTKN/COD ratio and the process type. The limits*given below are for approximately normal waste-water, (eg. with readily biodegradable COD frac-tion ft£ around 15 to 20%) and where completenitrification is obligatory"

(i) If the TKN COD < C.08 mgN-'mgCOD, com-plete nitrate removal is possible and thePhoredox process is indicated.

<ii) If 0,08 < TKN/COD < 0,11 complete nitrateremoval no longer is possible, but nitrate canbe excluded from the anaerobic reactor by us-ing the modified UCT process.

(iii) If 0,11 < TKN/COD < 0,14, the modified UCTprocess no longer can exclude nitrate from theanaerobic reactor and the UCT process is in-dicated provided the 3-recycle ratio is carefullycontrolled. However, poor settling sludgesmay be encountered.

(ivJ If TKN/COD > 0,14 it is unlikely that biologicalexcess P removal will be achieved with normalmunicipal wasteflows.

8-2 Anaerobic sludge mass fraction selection

The readily biodegradable COD fraction (fts) is approx-imately 0,20 for raw municipal wastewaters (i.e. ffc,Eabout 0,25), and approximately 0,30 for settledmunicipal wastewaters (i.e. ftlS about 0,34) - see Table6.2 and Chapter 3, Section 3. Consequently, for approx-imately normal municipal wastewaters, the lower theCOD strength (Stl) the lower the SbS(; when Sti for rawwaste-water is less than 250 rrigCOD/l, Sbs, is less than50 mgCOD/7 and the attainment of excess P removalfor r or s-recycle ratios of 1 ;1 is unlikely or sporadic evenif the process is correctly selected and designed. Con-versely the higher the STI, the greater the StJ5( and theeasier it is to establish the conditions for excess Premoval and obtain high P contents in the active massof the sludge y. As y is refated to both (Sbsa-25) andthe anaerobic sludge mass fraction fvs, the adverse ef-fect of low (Sj,^.,-25) at low Stl can be countered to adegree by increasing fx£ to its maximum* of about 0,25.

The following guidelines based on raw CODstrengths are not unreasonable for an initial estimate offxa to obtain good excess P removals or approximatelynormal municipal wastewaters:

(i) Slt < 400 mgCOD'f, fva - 0,20-0,25

<ii! 400 < S-, < 700, fXc = 0,15-0,20

(iii) S., > 700 mgCOD/f, f,a - 0,10-0,15

The above initial estimates must be checked andmodified depending on the actual P/COD and readilybiodegradable COD fraction.

8.3 Subdivision of anaerobic reactor

Recent research has indicated that it is advantageous tosubdivide the anaerobic reactor into a number of com-partments operated in series. This is because the ab-sorption of readily biodegradable COD by the poly-P ac-cumulating organisms appears to be a first order reac-tion (see Section 3.4 above) so that, for the sameanaerobic sludge mass fraction, a series of completelymixed reactors operates more effectively than a singlecompletely mixed reactor. Theoretically, the greater thenumber of equal volumed compartments the better, butpractically very little advantage is gained by havingmore than 3 or 4. Four compartments would appear tobe the most suitable because then, by providing the re-quired additional recycle flows, the process can beoperated as a Modified Phoredox or UCT type processeach with approximately the same anaerobic sludgemass fraction (see Figure 7.17).

9. CLOSURE AND CONCLUSIONS

The above calculations demonstrated that:

(1) The average wastewater characteristics govern thedesign of and the effluent quality attainable in anactivated sludge process for biological N and Premoval. The principal wastewater characteristicsare:

(i) the COD concentration (Sti)(ii) the readily biodegradable COD fraction (f,E)(iji) the TKN/COD ratio(iv) the maximum specific growth rate of the

nitrifiers at 20°C (^nnW(v) the maximum and minimum temperatures

<Tma, and Tm,n)(vi) the P/COD ratio.

(2) Once the sludge age of a selected process is fixedthe removals of N and P attainable are fixed and theN and P removal efficiencies then depend on theTKN/COD and P'COD ratios of the wastewater.Because for approximately normal wastewaters(fls = 0,20) complete denitrtfication can only beachieved for TKN/COD ratios less than 0,08 (with-out the addition of an external energy source), thehigher the ratio above 0,08, the lower N removal ef-ficiency.

(3) Attainment of excess P removal is crucially depen-dent on the readily biodegradable COD concentra-

* Generally, the higher the f.s ratio above 0,20, the higher these limits and the lower the fis ratio below 0,20 the lower these limits. These dif-ferent limits can be calculated with the theory presented in this monograph using different fls ratios.

" I n South Africa, complete nitrification is obligatory. Wherp complete nitrification is noi obligatory, nitrate discharge to the anaerobic reactorcan be avoided by limiting nitrification by e.g. limiting the oxygen supply to the process, but in South Africa, such a strategy falls in linemore with an emergency action than a design criterion.

'"Laboratory and pilot scale data on process behaviour for f,a > 0,20 is available from the NIWR (see Interim Design Guide by NIWR).

7-30

tion in the influent SLji.. For recycle ratios to theanaerobic reactor of 1:1, if Sb&l < 50 mcCOD/i, itis highly unlikely that excess P removal can beachieved in any process; if Sbs. > 50 mgCOD ' I ,excess P removal can be achieved provided nitratecan be excluded from the anaerobic reactor, inwhich event, the degree of excess P removaldepends on the magnitude of S._,LI above50 mgCOD/i and the influent COD strength S...The sensitivity of excess P removal to the S ts i con-centration is such that any practice that reducesSbS! should be avoided (see Chapter 3, Section 3!and conversely, any practice which increases Sbs,should be encouraged (see Chapter 3, Section 3and Chapter 6, Section 7.1.31.

(4) In general, the establishment of a readily biode-gradable COD concentration in the anaerobic reac-tor IS.,M) in excess of 25 mgCOD/f, to stimulateexcess P removal, becomes increasingly more dif-ficult as the raw influent COD strength of thewastewater (S,() decreases: For average readilybiodegradable COD fractions ranging between 0,20to 0,25, the minimum S,t for which St,SiT is greaterthan 25 mgCOD/f is about 250 mgCOD/f; as S.;increases above 250 mgCOD.'f, the establishmentof SbS3 greater than 25 mgCOD/f becomes in-creasingly easier and the excess P removal per unitinfluent COD increases and the attainment of ex-cess P removal becomes less sensitive to externalfactors, e.g. DO control in the process.

(5) Cyclic flow and load conditions do not appear to af-fect the mean daily excess P removal, in fact, theremoval often is improved, particularly under lowaverage influent COD conditions. The reason forthis improvement is that under the cyclic condi-tions (with constant recycle flows) during the peakperiods the peak COD is higher than the daily meancausing a higher (Sl>ba-25} condition and hencehigher P removals during the associated high flowperiod. However, under cyclic flow conditions theDO concentration in the aerobic reactor must beclosely controlled; too high DO concentrationreduces the nitrate removal whereas too low con-centration may inhibit nitrification, inhibit excessuptake of P and adversely affect settling. The practical difficulties associated with oxygen control aresuch that serious consideration should be given toequalization tanks operated under a controlstrategy that equalizes both flow and load. Such astrategy is now available (Dold, Buhr and fvlarais,1982) tsee Chapter 10 on the operation of nutrientremoval processes).

(6) For successful excess P removal in the Phoredoxprocess, near complete denitrification must beachieved to avoid excessive nitrate discharge to theanaerobic reactor. Complete denitrification can beachieved only for TKN^COD ratios < 0,08* mglSI/mgCOD for readily biodegradable COD fractions

(fIS) of 0,20 at 14fC. For TKN/COD ratios > 0,08*mgN/mgCOD, complete denitrification is unlikelyto be achieved and the indicated processes for ob-taining excess P removal are the UCT type pro-cesses.

(7) In the UCT type processes, by appropiate controlof the mixed liquor a-recycle, the anaerobic reactorcan be protected against nitrate discharge forTKN/COD ratios up to 0,13 mgN/mgCOD eventhough nitrate will be present in the effluent. ForTKN/COD ratios < 0,11, the modified UCT processcan achieve excess P removal and has the advan-tage over the UCT process of less need for operatorintervention and is more likely to maintainreasonable good sludge settling characteristics. ForTKN/COD ratios > 0,11 mgN/mgCOD, the onlyprocess that can achieve excess P removal is theUCT process, but it is possible that rather poorsludge settling characteristics will be obtained if theinfluent COD is high.

(8) For TKN/COD ratios > 0,14 mgN/mgCOD, it isunlikely that biological excess P removal can be ob-tained when complete nitrification is obligatory dueto the inability to achieve sufficient denitrificationwithout the addition of an external energy source.However, such high TKN/COD ratios are unlikelyto be encountered in raw or settled municipalwastewaters.

(9) Primary sedimentation is unfavourable for achiev-ing high removals of both N and P because itsignificantly increases the TKN/COD and P'CODratios and significantly reduces the COD strengthof the wastewater although the readily biodegrad-able COD concentration is only marginally af-fected. However, primary sedimentation signific-antly reduces the process volume requirements andtotal oxygen demand.

(10) For design of nutrient removal plants the influentwastewater characteristics (or their ranges) need tobe known with much greater surety than for con-ventional activated sludge plants otherwise adesign may be produced that fails totally in its Premoval objectives. Vital influent characteristics arelisted in (1) above. Useful data are the TotalAlkalinity and the unbiodegradable COD fractionsfus and fup. In the event that the vital characteristicsare known only approximately, values should beselected that lead to conservative designs, usuallyat the cost of reduced N and P removal.

(11) The selected process must have factors of safetybuilt into it or have flexibility in operation, so that ifthe influent characteristics are more adverse thanaccepted for design, it will be possible to accom-modate them. For example, if a Phoredox processis selected it is recommended that normally themaximum TKN ' COD ratio should not exceed 0,07*

'See ' on pages 7-30 and 7-32.

7-31

mgN/mgCOD and the process should have a built-in safety factor by designing it for 0,08* mgN/mgCOD. For TKN/COD ratios > 0,07* mgN/mgCOD, the Phoredox/modified UCT/UCT pro-cess should be selected; with this process com-bination, by appropriate regulation of the recycles,excess P removal can be induced (subject to SbS)).Theoretically, at low TKN/COD ratios (= 0,08*),the Phoredox process will be the optimal one, butpractically its inflexibility to accommodateTKN/COD ratio variations which exceed 0,08* is adisadvantage; with the modified UCT'UCT pro-cess excess P removal can be obtained (subject toSbsl) at the expense of N removal. By providing thevarious optional recycles, the best results can beobtained when required during the life of The plant.

(12) The selected process should require the leastamount of decision making from the operator toobtain consistent results. In this regard themodified UCT UCT process appears to lend itselfmost readily to setting operational procedures toabsorb adverse conditions when these arise. Alsooperational adjustment to mitigate the regularchanges in wastewater flow and characteristicsobserved over different days of The week can bepreprogrammed.

10. REFERENCES and BIBLIOGRAPHY

BARNARD, J.L. (1975a) Biological nutrient removal without The addi-tion of chemicals. Water Research, 9, 485-490.

BARNARD, J.L. i1975b) Nutrient removal in biological systems. Wat.Pol/ut. Control, 74, 2, 143-154.

BARNARD, J.L. 11976) A review of biological phosphorus removal inthe activated sludge process. Water SA.. 2, 13) 136-144.

DOLD, P.L., EKAMA, G.A. and MARAIS G.v.R. 119801 A generalmodel for the activated sludge process. Prog. Wat. Tech... 12,47-77.

DOLD. P.L., BUHR, H.O. and MARAIS, G.v.R. (1982) Design andcontrol of equalization tanks. Research Report No. W 42, Dept.ofCivil Eng., University of Cape Town.

EKAMA, G.A , SIEBRITZ, I P. and MARAIS, G.v.R. (1982) Con-siderations in the process design of nutrient removal activatedsludge processes. Wa:. Sci. Tech., 15, 3'4, 283-318.

EKAMA, G.A., VAN HAANDEL. A.C. and MARAIS, G.v.R. |1S79iThe present status of research on nitrogen removal: a mode! forthe modified activated sludge process. Presented at the Sym-posium on nutrient Removal, SA Branch of IWPC Pretoria, May1979. Proceedings published by Water Resea'ch Commission ofSouth Africa.

FUHS, G.W. and MIN CHEN (19751 Microbiological basis ofphosphate removal in the activated sludge process for the treat-ment of wsstewaier. Microbial Ecology, 2, 119-138.

FLORENTZ, M. and GRANGER, P. (1982) Phosphorus 31 nuclearmagnetic resonance of sctivsted sludge use for the study of the

biological removal of phosphates from wastewater. Submitted forpublication to Biochemical and Biophysical Research Communica-tions .

HOFFMAN, R. and MARAIS, G.v.R. (1977) Phosphorus removal inthe modified activated sludge process. Research report W. 22.Dept. of Civil Eng., Univ. of Cape Town.

LAW, S.M, DOLD, P.L. and MARAIS, G.v.R. (1982/ Phosphaterelease and uptake by activated sludge in batch experiments.Research Report VV 44 Dept. of Civil Eng., University of CapeTown.

MARAIS, G.v.R. and EKAMA, G.A. (1976) The activated sludge pro-cess Pan I - steady state behaviour. Water S.A., 2(4) 163-200.

MARAIS, G.v.R.. LOEWENTHAL. R.E. and SIEBRITZ I.P. (19821Review: Observations supporting phosphorus removal bybiological uptake. Wat. Set. Tech., 15, 3/4, 15-41.

MARSDEN, M.G. and MARAIS, G.v.R. (1977) The role of theprimary anox'C reactor in dfenitrification and biological phosphorusremoval. Research Report W 19, Dept. of Civil Eng., University ofCape Town.

MARTIN, K A C and MARAIS, G.v.R. (1975) Kinetics of enhancedphosphorus removal in me activated sludge process. ResearchReport W 14, Dept. of Civil Eng., University of Cape Town.

McLAREN, A.R. and WOOD. R.J. (1976) Effective phosphorusremoval from sewage by biological means. Water SA., 2( 1) 47-50.

NICHOLLS, H.A. (1975) Full scale experimentation on the newJohannesburg extended aeration plant. Water SA., 1(31 121-132.

NICHOLLS, H.A. (1978) Kinetics of phosphorus transformations inaerobic and anaerobic environments. Presented ai 9th IAWPRpost conf. seminar, Copenhagen, Prog. Wat. Tech. 10,

NICHOLLS, H.A. (1982) Application of the Marais-Ekama activatedsludge model to large plants. Wat. Sci. Tech.. 14, 581-598.

NICHOLLS, H.A., OSBORN, D.W. and MARA1S, G.v.R. (1982) Per-formance of large scale nutrient removal activated sludge plants.Research Report W 43, Dept. of Civit Eng., University of CapeTown.

RABINOWITZ, B. and MARAIS, G.v.R. (1980) Chemical andbiological phosphorus removal in the activated sludge process.Research Report W 32, Dept. of Civil Eng., University of CapeTown.

RENSINK, J.H. (1981) Biologische defosfatering en pro-ce5sbepalende faktoren. in Symp. Book - Defosfatering nieweontwikkelingen en praktijkervaringen in Nederland and Zweden.NVA Symp. (in Dutch).

SIEBRITZ, I.P., EKAMA, G.A. and MARAIS, G.v.R. (19801 Excessbiological phosphorus removal in the activated sludge process atwarm temperature climates. Proc. Waste Treatment and Utiliza-tion, Vol. 2, 233-251, Cds. C.W. Robinson, M. Moo Young andG.J. Farquhar, Pergamon Press, Toronto.

SIEBRITZ, I.P., EKAMA, G.A. and MARAIS, G.v.R. (1983) Biologicalphosphorus removal in the activated sludge process. ResearchReport W 46, Dept. of Civil Erg., University of Cape Town.

SIEBRITZ, I.P., EKAMA, G.A. and MARAIS, G.v.R. (1982) Aparametric mode! for biological excess phosphorus removal. Wat.Sci. Tech.. 15, 3/4, 127-152.

SIMPK1NS. M.JN.P. and GERBER, A, 11981) Technical guide onbiological nutrient removal: The Phoredox Process. Confidentialreport. SAIDCOR, Pretoria.

SIMPKINS, M J . and McLAREN, A.R. 11978) Consistent biologicalphosphate and nitrate removal in an activated sludge plant. Prog.Wat. Tech.. 10, (5 CJ 433-442.

STERN, L.B. and MARAIS, G.v.R. M974J Sewage as the electrondonor in biological denitrification. Research Report W 7, Dept. ofCivil Eng., University of Cape Town.

WILSON, D.E. and MARAIS, G.v.R. (1976! Adsorption phase inbiological denitrification. Research Report W 11, Dept. of CivilEng., University of Cape Town.

'These values are valid for a readily biodegradable COD fraction (ft,sl of 0,24. Generally the higher (lower) the fraction, the higher (lower) theselimits.

7-32

CHAPTER EIGHT

SECONDARY SETTLING TANKS

by

G.A- Ekarna, A.R. Pitman, M. Smollen and G.v.R. Marais

1. INTRODUCTION

In the activated sludge process, it is necessary toseparate the treated wastewater from the biologicalsludge mass thereby producing a clear final effluent.This solid/liquid separation phase is traditionally achiev-ed by gravity sedimentation tanks although flotation hasalso been shown to be feasible.* Only solid/liquidseparation by gravity sedimention in secondary settlingtanks wilt be considered in this chapter.

The secondary settling tank is a vital part of the ac-tivated process. It combines the function of a clarifier(producing a clarified final effluent) and a thickener(producing a continuous underflow of thickened sludgefor return to the biological reactor) (Figure 8.1). Shouldthe settling tank fail in either of these two functions,sludge will be carried over the effluent weirs and escapewith the effluent. Besides delivering an effluent of poorquality, loss of sludge could affect the behaviour of thebiological process by uncontrolled reduction of thesludge age to values below that required for properplant performance; for example, if the sludge age isreduced to below that required for nitrification, nitrogenremoval by denitrification will cease.

The conditions in the biological reactor affect thesettling and clarification characteristics of the sludge.For example, under-aeration reduces the settle-ability ofthe sludge; over-aeration may lead to pin-point Hoc for-mation and poor clarification even though the sludgemaintains good settling characteristics and often in pro-cesses where large proportions of the sludge are inten-tionally unaerated as in nutrient removal processes,poor settling characteristics are encountered. Thus, thefunctions of the biological process and that of thesecondary settling lank interact upon each other and

the design of the one cannot be undertaken in-dependently of the other - failure in either of these twounit processes causes failure of design objectives.

The settling characteristics of the sludges in ac-tivated sludge processes differ widely between differentprocess types and also vary with time in the same pro-cess depending on the composition of the influent (seeSection 7 below). For the various types of wastewatersand sludges to be treated, the design procedures for thesecondary settling tanks have tended to be of an ad hocnature - for each type of sludge, design rules were setdown that experience has shown will provide generallyfor adequate functioning of the settling tank over therange of sludge settling behaviour that was found couldarise. These rules often do not directly incorporate thesettling characteristics of the sludge - these are lumpedtogether with the hydraulic characteristics, as illustratedby the design rules taken from the IWPC designers'handbook:** (1) the overflow rate must not exceed1 m3/m2/h at Peak Wet Weather Flow (PWWF); (2) theretention time must not be less than 1,5 h at Peak DryWeather Flow (PDWF); (3) maximum discharge per unitlength of effluent weir is 8,3 m3/h/m. These designrules implicity accept that the sludge settling character-istics will not fall below a certain minimum quality, aquality unknown but in conformity with past experienceof similar sludges. Where processes are developed thatgenerally tend to affect the settling characteristicsadversely these rules may result in inadequate designsresulting in failure of the settler and the process.

Considerable research has been undertaken inunravelling the behaviour of secondary settling tanksand to develop a reliable design procedure thatrecognizes the settling characteristics of the sludge.However, the settling process is complex and the fac-

Influentflow

w

Wasteflow

Biologicalreactor

Sludge return

i L

F"

flow

Settlingtank

Clarifiedeffluent

w

FIGURE 8 1: The activated sludge process incorporating thesecondary settling tank for liquid-solid separation.

1 A design guide lor solid liquid separation and waste sludge thickening by flotation is available (Bratby and Marais, 1976, 1977).

""Guide lo the design of sewage purification works" prepared by the IWPC.

J-1

tors affecting settling so many that general proceduresfor design have not emerged. This is principally the casewith activated sludge where constituents in the influentwastewater or conditions imposed on the process mayradically change the settling characteristics of thesludge, and ad hoc rules specific only to certaincategories of sludge have tended to be the usualmethod.

Lack of understanding of settling behaviour in settl-ing tanks, and of its interaction with the process in bothdesign and operation has been the source oi con-siderable difficulties in the operation of activated sludgeprocesses in South Africa. The first indication of failurehas nearly always been a icss of sludge over the effluentweirs of the settling tank. This loss of sludge does notnecessarily imply that the settling tank is at fault; theconditions in the biological reactor may have caused adeterioration in the settling characteristics of thesludge.

The design of the secondary settling tank based onthe settling characteristics of the sludge and the interac-tion between the secondary settling tank and biologicalprocess are briefly discussed in this chapter.

2. SECONDARY SETTLING TANK DESIGN

The settling tank performance is not only governed bythe overflow rate (clarification function) but also by thetransmission of the applied solids to the bottom of thetank (thickening function). These two functions operatesimultaneously but usually either one or the other will becritical during the daily cyclic operation of the settlingtank. Conventional design procedures specifying onlyan overflow rale in effect tend to provide mainly for theclarification function, in contrast although research hasbeen devoted to thickening, design criteria for thicken-ing have not been developed to a large degree.

Earlier research into thickening has been mainlyundertaken with non flocculant suspensions (e.g. Coeand Clevenger, 1916 and Kynch, 1952). These investiga-tions have established the importance of the mass fluxapproach. Although activated sludge is a flocculantsuspension, it has been shown that the mass-flux con-cept also can be applied to activated sludge (Dick, 1970and 1972). Application of the mass flux concept to thesettling tank design has been particularly assisted by thegraphical method of Yoshioka ef a/. (1957), whichallows estimation of the settling tank area for thethickening function (see Section 5.1 below).

The mass flux approach requires the measurementof the change in settling velocity of the sludge withchange in sludge concentration as the sludge settlesand thickens in the bottom of the settling tank. Amethod of assessing the settling behaviour of thesludge in the bottom of the tank is by means of the stir-red batch settling test.

3 STIRRED BATCH SETTLING TEST

Mixed liquor from activated sludge plants usually showa strong flocculating tendency and even at low concen-trations (lOOOmg MLSS/f) exhibit a zone settling

mode. In the zone settling mode, the concentration ofsludge is such that the particles irrespective of size, alltend to settle at the same rate throughout the zonedepth. The rate of settling is controlled by the rate atwhich the water passes upward through the mass andthis is inversely related to the concentration. Althoughthe mass settles as an entity leaving a clear liquid-mixedliquor interface, the particles are supported by the waterbetween the particles, i.e. upper particles are notmechanically supported by lower ones.

Compression is observed at the bottom of thebatch settling test. In the compression region each layercf particles provides mechanical support to the layersabove it. Particle movement no longer is governed onlyby hydraulic fractional forces and the links between theparticle. Forces transmitted by particle contact via thecompressible properties of the sludge and the interstitialpressure caused by the squeezing out of water from thecompressing particles govern the settling behaviour inthe compression stage.

Consider a batch settling test tube filled withsludge of concentration such that zone settling com-mences immediately (see Figure 8.2). initially the con-centration is uniform throughout and region B occupiesthe total depth (Figure 8.2a). Immediately after settlingcommences, a solid-liquid interface develops and aregion of clarified liquid A is formed (Figure 8.2b). Inregion B the particles still settle at uniform velocityunder zone settling conditions, and throughout region Bthe concentration is constant. The interface settlingvelocity is a constant value and is a function of the concentration of region B, which is the same as the MLSTconcentration at the start of the test. Simultaneouslywith the formation of region A, regions C and D areformed. Region D is one in which compression occursand region C is the transition stage from zone settling to

A

B

C

1• •

FIGURE 8.2: Chronological progress of a stirred batch settlingtest.

8-2

2irLU

O

0

10 -

2 0 -

3 0 -

40-

5 0 -

60

\

c

I 1 I 1

^ - ^ _ SettlingX/^V^Iocity

\ \ \ -\C, N . Solids\ ^ Concentration

1 1 . _L _l

0 20TIME"

60min)

80 100 120

FIGURE 8.3: Solid-liquid interface height versus time observedin batch settling tests ut different initial soiids concentrations.

compression. Region C is still a zone settling region butthe concentration increases down the zone. In a par-ticular test, the Transition zone depth generally is con-sidered to remain constant until the solid-liquid interfacereaches it.

A plot of the solid-liquid interface height versustime is given in Figure 8.3. The interface settling velocityis slow initially (A-B) because the activated sludge re-quires some time ro reflocculate from ihe disturbance ofthe sludge floes during the filling of the test cylinder. Tominimize this effect, it is necessary to fill the testcylinder from the bottom as slowly as practically possi-ble so as to cause minimal disruption of the sludgefloes. After a time a constant interface velocity isobserved (B-C) until region B sinks into region C.Thereafter the interface velocity decreases as denserand denser concentrations appear at the interface dueto the subsidence of the less dense layers into thedenser ones, until the compression region D is reached.Thereafter subsidence continues but now the velocity isgoverned by the compression behaviour of the sludge.

The interface velocity of the constant concentra-tion zone settling region is given by the slope of line B-Cin Figure 8.3. This velocity is defined as the settlingvelocity (Vs) of the sludge at the concentration equal tothat of the uniform concentration in the test cylinder atthe start of the test (X). The settling velocity Vs

decreases as the sludge concentration increases (XI(Figure 8.3).

150

5 IO 15SLUDGE CONCENTRATION (kg/m3)

FIGURE 8.4; The gravity solids flux versus sludge concentra-tion curve, i.e. downwards sludge transport due to gravity

sedimentation.

4. SOLIDS FLUX

Solids flux is defined as the product of the settlingvelocity Vs and the concentration of ihe sludge, i.e.

G6 = VSX (8.1)

Gs = solids flux (kg/m2/d)

Vs = settling velocity (m/d)

X = sludge concentration (kgMLSS/m3)By conducting a number of batch settling tests, the

flux at different sludge concentrations can be determin-ed. A typical flux curve for activated sludge is shown inFigure 8.4 in which soiids flux (G5) is plotted versussludge concentration (XI. At low X, although VB is highthe product Gs is low. At high X. Vs is low also giving alow Gs. At some intermediate X (about 3 to 5 kg 'm3 foractivated sludge), Gs is a maximum.

5. GRAPHICAL APPLICATION OF FLUX THEORY

5.1 Graphical procedure

In continuous settling tank operation (shown schemati-cally in Figure 8.5), the sludge entering the settler istransferred to the bottom by two flux components (1)the gravity flux (Gs), and (2) the flux caused by thedowndraft flow generated by the sludge abstractionflow from the bottom of the settler, called the bulk flux.The gravity flux is given by Eq. 8.1. The bulk flux (Gb) is

8-3

Sludgewasteflow

Influentflow Qi Operating

caxjmtiution

A = Surfocearea

Underflowconcentration X r

Thickened sludgereturn flow Qr

Clarifiedeffluent

• Q;

OverflowrateQj /A«U,

Underflowrate

FIGURE 8.5: Idealized schematic representation of the secon-dary settling tank in the activated sludge process.

the product of the sludge concentration X and theunderflow velocity (Uu) where the underflow velocity isthe downdraft velocity caused by the underflow sludgeabstraction i.e.

and

X

= Q,/A

(8.2)

(8.3)

where

Uu = underflow velocity (m/d)Qr - underflow rate (m3.'d!A - surface area of settling tank

150

0 5 10 (5SLUDGE CONCENTRATION (kg/m3)

FIGURE 8.6: The bulk flux curve versus sludge concentrationfor fixed underflow rate (Uu) i.e. downward sludge transport

due to the underflow sludge withdrawal.

For a fixed underflow rate, the bulk flux is propor-tional to the sludge concentration. This is depictedgraphically in Figure 8.6.

The total flux (G,) to the bottom of the settler is thesum of the two flux components i.e.

G, = Gs

- XIV, + u ) (8.4)

For a selected value of the underflow velocity Uu

(i.e. Q, and A), the two flux components can be addedgraphically 'see Figure 8.7). For the particular choice ofUU( the total flux Gt attains a minimum value G1 at aconcentration Xlr and this minimum limits the rate atwhich sludge can reach the bottom of the settling tank.Hence to ensure that all the sludge reaches the bottom,the sludge mass applied to the settling tank per unit sur-face area i.e. the applied flux (Gapf must be equal to orless than the limiting flux (G^. The applied flux on thesettling tank is given by the product of the sludge con-centration in the reactor (Xc) and the combined recycleand influent flow (Q, + 0,1 per unit area (see Figure 8.5).Hence, for safe operation of the settling tank

where

Gap = XO(Q, + a ) / A = XO(UO+ Uu)

where

(8.5)

(8.S)

Xo = operating MLSS sludge concentration in bio-logical reactor (kg/m3)

Uo — overflow rate

= 0,/A (see Figure 8.5) (8.7)

Q, = influent wastewater flow to process (rnVd}

Gj - limiting flux (kg/mVd)

8-4

2 0 0

SLUDGE5 10

CONCENTRATION (kg/m15

3)FIGURE 8.7: Total flux in settling tank versus sludge concen-tration i.e. downward sludge transport due to gravity sedimen-

tation and underflow sludge withdrawal.

Q. = Influent flowQ = Recycle flowA= Surface area

-OperatingConcentration (X ) ~

Overflow rate

State point

Underflow -concentration .

B

0 5 10SLUDGE CONCENTRATION

FIGURE 8.8: The gravity flux curve onto which is superimpos-ed the settling tank behaviour; the state point defines theloading slate of the tank and the underflow rate (Uu). overflowrate (Uo) and the operating solids concentration (Xo) lines in-

tersect at the state point

From a mass balance around the settling tank, if nosludge is lost over the effluent weirs, the mass of sludgeapplied is equal to the mass of sludge abstracted via thesludge underflow i.e.

QrX, = X0(Q,

where

Qr) (8.8)

X. = MLSS concentration in underflow recycle(kg.'m3)

For specified reactor MLSS concentration Xy andinfluent flow Q; and selected underflow recycle Gr, thearea A is found by trial and error using the graphicalconstruction shown in Figure 8.7. The correct value ofA is that value which makes Eq. 8.5 an equality; valuesof A which make the LHS > RHS are too small whichmeans settling tank failure in thickening will occur andvalues of A which make the LHS < RHS are too largewhich means the settling tank operates successfully butthe area can be reduced.

The value of A found by the above procedure isvalid only for the specified Xo and Q, and selected Qr. Byrepeating the procedure for different values of Q,, dif-ferent values of A will be obtained. For the proposedplant, the maximum area for the anticipated range of Q,is the required design area.

The clarification criterion is satisfied if the overflowvelocity (Uo) is less or equal to the settling velocity ofsludge at the concentration at which it is discharged tothe settling tank (Xo) I.e.

^ V . a t X r (8.9)

Clearly the above procedure is extremely tediousbecause for every choice of Qr, different total fluxcurves need to be constructed.

The above graphical design procedure has beenconsiderably streamlined by Yoshioka et al. (1957), (seealso Dick and Young, 1972; Pitman, 1980; and White,1975) by introducing the concept of the state point. Onthe gravity flux plot (see Figure 8.8), the operatingsludge concentration (Xo) is represented by a verticalline at the specified Xo value. The overflow rate(Uo = Q/A) is represented by a line (called the overflowline) from the origin with the slope equal to the overflowrate. The intersection point of the overflow line and theoperating concentration line is called the state point.The underflow rate (Uu = 0,/A) is represented by a line(called the underflow line) with the slope equal to theunderflow rate passing through the state point (seeFigure 8.8). The intersection of the underflow line withthe horizontal axis gives the underflow concentration(XJ (as given by Eq. 8.8) and that with the vertical axisgives the applied flux Gao (as given by Eq. 8.6).

The state point defines the operating state of thesettling tank and operating changes are reflected on theflux curve as follows: (i) a decrease in the underflowrate rotates the underflow line about the state point inan anticlockwise direction while the state point remainsm position (see Figure 8.9a); tii) an increase in overflowrate rotates the overflow line about the origin in an anti-clockwise direction, moves the state point upwardsalong the operating sludge concentration line while theunderflow line slope remains constant but continues tointersect the state point (Figure 8.9b), (iii) an increase in

8-5

THICKENING FAILURE THICKENING FAILURE, . j • . , ,

Safe condition Safe conditionCrrticat condition -Failure condition -

Constantoverflow r a t e

Constantoverflow rate '

Constantunderflow

Decreasingunderflow rate

!_|Jn creasingconcentration



FIGURE 8.9a. Graphical representation of the settling tankoperation conditions on the gravity flux curve showingdecrease in underflow rate (Uu); HI safe operating conditions,(2) critically loaded conditions and (3) overloaded conditions.

FIGURE 8.9c: Graphical representation of the settling tankbehaviour on the gravity flux curve showing increase inoperating solids concentration 1XO); 111 safe operating condi-tions, 12) critically loaded conditions and (3) overloaded condi-

tions.

150THICKENING FAILURE

I t I I I j T , I T

Safe condition— Critical corxiition

Failure condition

Increasing

rate

Constantunderflow rate

Corrstontconcentration

0 5 10 15SLUDGE CONCENTRATION (kg/m3)

FIGURE 8.9b: Graphical representation of the settling tankoperation conditions on the gravity flux curve showing in-crease in overflow rate IUJ; (1) safe operating conditions, (2)

critically loaded conditions and (3] overloaded conditions.

operating solids concentration moves the operatingsolids line to the right and because the slopes of theunderflow and overflow lines remain unchanged, thestate point moves along the overflow line while theunderflow line continues to pass through the state point(Figure 8,9c).

The state of the settling tank, i.e. whether or not itis in an under- or overloaded state, and in the overload-ed state, whether failure wili occur in its thickening orclarification function, can be obtained from the statepoint on the gravity flux curve. Referring to Figure 8.9a,b and c:

f1 When the state point is within the envelope of theflux curve, the clarification criterion is met andthickening governs the behaviour of the settlingtank. When the underflow linefi)

(ii)

cuts the flux curve at one point only (i.e. nearA), safe operating conditions prevail;

12)

cuts the fiux curve at one point only (i.e. nearAl and is tangential to the flux curve (i.e. nearBl, critically loaded conditions prevail,

(iii) cuts the flux curve at 3 points (i.e. once near Aand twice near B) overloaded conditionsprevail.

When the state point is on the flux curve criticalconditions with respect to clarification prevail.

3-6

Under these conditions, thickening will also becritical if the underflow line is tangential to the fluxcurve on the state point; any underflow raie lessthan this will cause thickening failure.

(3) When the state point is outside the envelope of theflux curve, clarification failure conditions prevail(and thickening also because conditions 1u) andl(ii) above cannot be mei).

(4} When the slope of the underflow line is too steep tomake a tangent to the flux curve, but the statepoint is within the envelope of the flux curve, thethickening criterion no longer needs to be met andclarification only governs the design.

The above conditions can be readily applied todesign. For given Q. and XLI, select A which gives Uc,(Eq. 8.7) and with Xo, fixes the state point. The area Amust of course be selected such that the state point fallson or within (he flux curve. On the state point, draw theunderflow line such that it makes a tangent to the fluxcurve. From the slope of the underflow line and selectedA, determine Q, (Eq. 8.3). Notice that in this procedureEqs. 8.5 to 8.9 are being solved graphically: the in-tercept of the underflow line with the vertical is the ap-plied flux Gyp and when the underflow line makes atangent to the flux curve, the applied flux equals thelimiting flux Gj (see Eq. 8.5 and Figure 8.8).

Although the above procedure is still a trial and er-

ror one in that a repeated selection of A is made and thecorresponding Q, for safe operation calculated, it is farsimpler than the total flux curve method because eachselected A can be analysed on the same flux curve, thusobviating the need for constructing different total fluxcurves (Figure 8.7). Note that the two methods giveidentical results; the condition described by the horizon-tal line from Q1 in Figure 8.7 is identical to that describedby the underflow line tangential to the flux curve inFigure 8.8.

5.2 Concentration profiles

The gravity flux curve can also be used to plot idealizedsludge concentration profiles in Ihe settling tank (seeAlkema, 1971; Pitman, 1978; White, 1975; Vesilind,1968 and Dick and co-workers, 1967, 1970, 1972). Theconcentration profiles in the settling tank representedby the safe, critical and overloaded conditions in Figs.8.9a, b and c are shown in Figure 8.10. The stepchanges in concentration shown in Figure 8.10 areidealized; in practice the changes are more gradual.*

At critically loaded conditions, three concentrationzones occur in the settling tank (Figure 8.10a) (1) theunderflow concentration zone in the bottom of concen-tration Xr where X, is given by the intercept of theunderflow line with the horizontal axis, (ii) the zonesettling region where the concentration is equal to thelimiting concentration Xj which causes the limiting fluxGj, where Xl is given by the concentration at the point

E

Xt -Q-LJQ

u

1 -

2 -

3 -

Overflow

Feed point

c 5^ O «

Jjnti Sy

o'g"h4 O ^

• 1kBottom t

Concentration

CRITICALLY LOADED

( a )

Concentration

UNDERLOADED

( b )

xef x a f x

Concentration

OVERLOADED

( C )FIGURE 8.10: Sludge concentration versus depth profiles in thesettling tank for (i) critically loaded conditions (Figure 8,10a).(ii) underloaded conditions (Figure 8,10b) and overloaded con-

ditions (Figure 8.10c).

"The depth required to allow these changes in concentration to take piace that governs the design of the depth of ihe settling tank. The fluxtheory perse gives only the area reauirement for thickening; the depth must be estimated from experience obtained in practice (see Section 7below) and to allow a cenatn accumulation of sludge in the settling tank 'see Section 5.3 below).

i-7

of tangency of the underflow line of the flux curve; and(iii) a dilute zone settling region above the zone settlingregion of concentration Xd/ where Xdz is given by theconcentration of the intersection point of the underflowline with the flux curve to the left of the turning point(i.e. near point A on Figures 8.9a, b and c. Undercritically loaded conditions the fluxes of (it the upperdilute zone, (ii) the zone settling region i.e. the limitingflux G ] ( and (iii) the underflow withdrawal flux are allequal to the applied flux Gap (Eq. 8.6) i.e.

dL) = X C ( U

U

= xru,

(8.6)

(8.10a)

(8.10b)

(8.10c)

where

Vi.*a* and VSJll is the settling velocity of the sludgeat concentrations Xdz and X-.

The zone settling region of limiting sludge concen-tration is not stable: the layer will contract or expanddepending on whether the conditions are slightlyunderloaded or overloaded respectively.

When conditions change from a critically loaded toan underloaded condition (by reducing the overflowrate or operating sludge concentration or increasing theunderflow rate, see Figure 8.9), the zone settling regionwill contract downwards until it sinks into the underflowconcentration layer. The time taken for the layer todisappear depends on the rate of sludge withdrawalfrom the tank bottom by the underflow recycle. Whensteady state conditions again become established, onlytwo concentration zones wili be present, (see Figure8.10b) i.e. the underflow concentration zone Xr and thedilute zone settling region; these concentrations can befound from Eqs 8.10a and c respectively. When condi-tions change from a critically loaded to an overloadedcondition (by increasing the overflow rate or operatingsludge concentration or reducing the underflow rate,(see Figure 8.9), the limiting concentration zone settlingregion will expand upwards until it reaches the height ofthe feed point (see Figure 8.10c). The solids transfer tothe bottom of the settler is limited by the limiting flux G!and the difference between the applied flux GaD and Cx

is the flux that causes the expansion of the zone settlinglayer i.e. sludge is being accumulated in the settlingtank. The time taken for the zone settling layer to reachthe feed point depends on the magnitude of the dif-ference between Gap and G1( the larger the differencethe faster the upward movement. After the zone settl-ing layer reaches the feed point a sludge layer above thefeed point develops (see Figure 8.10c). The concentra-tion of this layer is different to the zone settling layerbelow the feed point because the direction of waterflow above the feed point is upwards whereas below thefeed point it is downwards (see Figure 8.5). The con-centration of the sludge layer above the feed X2f can befound from the flux moving upwards, i.e.

X d ,<U 0 - Vs,a() = G a p - G, (8.11)

where

Vs*ai = s e u l ' n 9 velocity of sludge at concentrationXa ,

When this layer reaches the effluent overflow, sludgeloss with the effluent will begin. The concentration thatis lost with the effluent Xe,.given by

Xef = ( G a p - G^/Uo 18.121

When sludge is lost from the settling tank via the ef-fluent flow, the operating concentration decreases. Theoperating concentration will continue to decrease until anew critical loaded steady state condition develops (seeFigure 8.9c from overloaded to critically loaded condi-tions}. The limiting flux G3, limiting sludge concentra-tion X] and underflow concentration X, which definethe sludge handling capacity of the settling tank aregiven by the vertical axis intercept, tangent point andhorizontal axis intercept of the new underflow linetangential to the flux curve and parallel to the existingoverload underflow line. Note that this new underflowline does not pass through the existing state point butthat a new state point will develop at a lower operatingconcentration (see Figure 8.9c for overloaded to critical-ly loaded conditions).

The above discussion shows that some timeelapses between the onset of overloaded conditions andthe time of sludge loss with the underflow. The lengthof time that elapses as well as the concentration ofsludge that is lost depend on the severity of theoverload (i.e. Gap - G^, the higher the overload, theshorter the time and the higher the concentration.

it should be noted that with overloaded conditions,the concentrations of the sludge profile are not given bythe state point condition on the flux diagram. This isbecause, for the overflow, underflow and sludgeoperating concentrations lines to intersect at the statepoint a mass balance between sludge applied andsludge withdrawal by the underflow must be obtainedi.e. Eq. 8.8 does not apply because sludge will eventual-ly be lost via the overflow. If the overloaded conditionsare allowed to persist sludge will be lost from the systemthereby reducing the process sludge operating concen-tration. Consequently, for new stable conditions todevelop under a continuous overload, it can be seenthat a considerable mass of sludge must be lost to effectthe required decrease in operating concentration.

5.3 Dynamic conditions

Under normal daily operation of the settling tank, theoperation sludge concentration and underflow rate canbe taken as remaining approximately constant. How-ever, the influent flow and hence the overflow changescyclically over the day (see Figure 8.9b). It may happenthat under peak flow conditions, overloaded conditionsdevelop, in which event the zone settling layer will beginto rise. The overload, depending on its severity, maypersist for a few hours without any sludge loss, bywhich time the peak flow period may have passed andthe zone settling layer again begins to drop. Conse-quently, over the day, the sludge layer will move up anddown and the settler may even experience a temporaryOverload without sludge loss in the effluent ITracy andKeinath, 1973).

The principal factor that prevents sludge loss dur-ing a temporary overload is sludge storage, which, oncethe area has been designed appropriately according tothe flux theory, is directly proportional to the depth ofThe settler. Generally the deeper the settler, the greaterthe temporary overload that can be sustained withoutsludge loss. Unfortunately, no guidance regarding thedesign of the depth can be derived from the flux theoryper se and the depth is best based on practical ex-perience. A depth of 3 to 4 m appears to be adequate(see Section 7 below). Deep tanks also have the advan-tage that allowance is made for the various concentra-tion zones such as the underflow concentration zonewhere sludge compression occurs. Under cyclic loadingconditions the compression zone depth also varies inaccordance with the load, being deeper at the peak flowperiod. Deep tanks are also less prone to disturbance ofthe settling zone sludge layers caused by hydraulic tur-bulence of the discharge and sludge withdrawal flows(see also Section 6.5 befow).

5.4 Discussion

Clearly the graphical approach to the flux theory

described above gives considerable insight into thedesign and operation of secondary settlers taking dueconsideration of the settling characteristics of thesludge. However, despite convenient developments inthe graphical approach, the procedure remains rathertedious as for each sludge concentration, the settlingvelocity versus sludge concentration curve needs to beconsulted to estimate the settling velocity. Perhaps thebiggest drawback is that the differences between thesettling characteristics of different sludges can only bedescribed graphically.

6. ANALYTICAL APPLICATION OF FLUX THEORY

6.1 Empirical equation for settling velocity versussludge concentration

In order to overcome the difficulties associated with thegraphical approach, research has been undertaken tofind an empirical relationship between the settlingvelocity and sludge concentration. Such a relationshipwould aliow direct analytical solutions to the designequations.

200 i 1 1 1 1 1 1 1 I I

Vo =137 m/d

Settling characteristics

VQ = 137 m /d

n = O,37m3/kg

data points

- 'og(96/2),2,3O3(11.5-1,0)

= 0,37 m3/kg

1 1 1 L

0 2 4 6 8 10 12 14SLUDGE CONCENTRATION { kg MLSS/m')

FIGURE 8.11: Determination of settling characteristics Vc andn from multiple batch test settling data.

8-9

Many mathematical expressions linking the settlingvelocity and sludge concentration have been proposedbut the two most widely accepted are the following

Vb = V c X " (8.13)

where

s = Voe nX '8.14)

Vs = settling velocity (m/d)

X - MLSS concentration (kg/m3)

Vc,n = constants describing settling character-istics of sludge.

The first expression can be represented by astraight line on a fog V, versus tog X plot {i.e. log - log) ,the second by a straight line on a log V, versus natural Xplot (i.e. semi-log). Considerable research effort hasbeen expended to establish the applicability of each ex-pression to flux theory - Dick and Young, 1972 and Pit-man, 1978 using Eq. 8.13 and Vesilind, 1968a, 1968band White, 1975 using Eq. 8.14. A comprehensive com-parison is presented by Smollen (1981), who concludedthat Eq. 8.14 is superior to Eq. 8.13 because it (1) gives atheoretically consistent description of the observedgravity flux curve with defined turning and inflexionpoints and (2) gives a closer correlation with datameasured on laboratory, pilot and full scale plants overa number of years (rz > 0,95). Consequently in thischapter, Eq. 8.14 is accepted for describing the inter-face zone settling velocity versus sludge concentrationin the batch settling test.

The settling characteristics Vo and n are readily ob-tained by linear least squares regression analysis ofmultiple batch test settling data over a range of sludgeconcentrations from 1,5 kg/m- to 15 kg/m3 on the ex-pression (see Figure 8.11)

logVs = log V,, - nX/2,303 (8.15)

Substituting Eq. 8.14 for Vb into Eq. 8.1 yields thegravity flux in terms of the concentration and settlingcharacteristics i.e.

G, = X Vr,e"nX (8.16)

Equation 8.16 describes the gravity fiux curveshown in Figure 8.4.

6.2 Mathematical properties of theoretical fluxequation

The slope of the flux curve is given by the derivative ofEq. 8.16 i.e.

nX) (8.17)dGs/dX - VCie- r X( l

which, when set to zero, gives a turning point atX = -t-i/n and a maximum flux of V0/ne, DifferentiatingEq. 8.17 with respect to X and substituting 1/n for Xshows that the turning point with coordinates (X - 1/n;G = V0/ne) is a maximum. Setting the second derivativeto zero shows that there is an inflexion point in the fluxcurve at coordinates (X = 2/n; G = 2Vo,'ne2).

From an inspection of the flux curve, no tangentialunderflow line can be constructed on the ffux curve (see

Glmox • * V<M*>• 2 - G i p

Turning point

SLUDGE CONCENTRATION (kg/m'J

FIGURE 8.12- Mathematical properties of the flux curve basedon the settling velocity-sludge concentration equation Vb = V&

expf-nXi.

Section 5.1 above) with a slope greater than that of theflux curve at the inflexion point. The intercepts of thetangent at the inflexion point with the vertical andhorizontal axes therefore give respectively the maxi-mum limiting flux (Glmax) and the minimum underflowconcentration (Xrrr)in) for thickening to govern thedesign of the settling tank. The slope of this tangent is- V0/e2 (from Eq. 8.17), and hence

= 2/e.Gr (8.18)

X,^ir, = 4/n = 2.XP = 4.X:P (8.19)

where-subscripts IP and TP refer respectively to the in-flexion and turning points.

The above features of the flux curve are shown inFigure 8.12.

6.3 Application of flux equation to settling tanks

Defining s as the underflow recycle ratio, i.e.

s = Qr/Q (8.20)

the maximum overflow rate to satisfy the thickeningcriterion is found from Eqs. 8.3 and 8.5 to 8.8 and isgiven by

Q,/A= G:/-!XfJ(1 + s)} = Gi/IX^s) (8.21)

where

X, = (1 + s)X0/s (8.22)

The equation of the underflow line with a point oftangency to the flux curve is

G - G.(1 - X/X,i (8.23)

and its slope is

dG/dX = - G r ' X (8.24)

At the point of tangency (see Figure 8.8) X - X1(

and both the fluxes and the slopes of the flux curve and

8-10

the underflow tangent are equal i.e. Eq. 8.16 - Eq. 8.23and Eq. 8.17 = Eq. 8.24. From these equations taking Xr

as known, solving for X1 and, ignoring the impracticalsolution, yields

X, = (Xr/2H1 +

Substituting Eq.limiting flux i.e,

(1

n -

8.25 into Eq.

•exp^-nX.d

(8.25)

8.16 gives the

!2)

where

(8.26a}

(8.26b)a = V 1 -4 / (nX ( ,

if a < 0, there is no solution for G,. Setting a = 0yields Xr = 4/n and hence Gj = 4V0/ne2 which cor-responds to Xrn,in and G l rns, respectively as found above(see Figure 8.11). No solution for G, is possible lorX. < 4/n because the curvature of the flux curve is suchthat no valid tangential underflow line can be con-structed within the envelope of the flux curve. Thisresult can be interpreted as the thickening criterion notbeing important in a design when Xr < 4/n, leaving onlythe clarification criterion to be met (see condition 4 inSection 5.1 above).

Substituting Eq. 8.22 for X, into Eq. 8.26 andsubstituting Eq. 8.26 into Eq. 8.21 yields

A S ( 1 - a ) exp - , s)Xo(1

where

a = \r i-4s/Jn(1 (8.27b)

Equation 8.27 relates the overflow rate to the re-cycle ratio, s, for a selected value of the sludge concen-tration in ihe reactor and settling characteristics of thesludge to meet the thickening criterion. If a ~ 0,X r . -4s/(1 + s)n. Hence from Eq. 8.27, if Q = 0

—'A • (e;s) (8.28)

(8.27a)

Consequently, if Q,/A is less than VG/(e2s) thenboth the thickening and clarification criteria have to bemet. If 0,/A is greater than Vu/(e

2s), then thickeningapparently is not a governing criterion and the clarifica-tion criterion only has to be met. This criterion is met ifthe overflow rate is less than the settling velocity of thesludge at the feed concentration i.e. from Eq. 8.9

Q(/A •-= Voe •"*<> (8.29)

6.4 Settling tank design and operating chart

By plotting Q,.'A versus s from Eqs. 8.27 to 8-29 for dif-ferent values of X,,, a steady state design and operatingchart for the settling tank is obtained (Figure 8.13). Thehyperbola on the diagram described by Eq. 8.28distinguishes between the domain in which both thethickening and clarification criteria have to be met (areabetween axes and hyperbola) and in which the clarifica-tion only has to be met (area above hyperbola). The

50

- 4 0

DESIGN AND OPERATING CHARTfor sludge with Vo* l37m/d,n=0,37m*kg

UJ

5Li-CCUJ>o

30

20

10

o

Boundary of region in **iichclarification criterion onlyneeds to be met.

FAILURECONDITIONSREGION 4

SAFECONDITIONSREGION

Clarificationcriterionboundary

Cf» 5 kg/m3"6

0,0 l,0 2,0UNDERFLOW RECYCLE RATIO («»Qr/Qj:

FIGURE 8.13. Design and operating chart for secondarysettling tanks showing the different types of failure that canoccur for different combinations of overflow rate and

underflow recycle ratio.

8-11

thickening criterion is given by Eq. 8.27 which, for dif-ferent Xo values, is a family of curves from the hyper-bola to the origin. These curves show that as underflowrecycle ratio s decreases the allowable overflow ratedecreases. The clarification criterion is given by Eq. 8.29and being independent of s, is a family of horizontallines. In the reqion below the hyperbola, the thickeningand clarification criteria have to be met simultaneously.In some instances in this region, the allowable overflowrate is higher for thickening than for clarification so thatclarification is the governing criterion for the settlingtank. Consequently for a specified reactor concentra-tion Xc,, failure of the settling tank is represented byoverflow rate-underflow recycle ratio data pairs fallingabove the thickening and the clarification lines.

It should be noted That the design and operationchart is aiso valid for daily cyclic conditions. For suc-cessful operation of the settling tank, the conditionsprescribed by the design and operation chart need to bemet at any instant of the day. For example, for a fixedsettling tank area (A) and underflow rate (Q,), as the in-fluent flow to the plant (Q) increases, so the overflowrate (G;/A) increases and the recycle ratio (s) decreases,which moves the intersection point of the overflow rateand recycle ratio upwards to the left in Figure 8.13. Ateach instant of the day, the overflow rate and recycleratio intersection point must subscribe to the thickeningand clarification criterion in the diagram i.e. must at alltimes be below the thickening and clarification lines.

6.5 Design example

Assume that the MLSS concentration of thenutrient removal processes designed in Chapters 4, 5, 6and 7 is 5 kg/m3. For the nutrient removal process, thesettling characteristics are selected to be rather poor i.e.Vo - 137 m/d or 5,7 m/h and n - 0,37 mVkg givingV0/n = 15,4 kg/m2/h (see Table 8.1}. The average dryweather flow (ADWF) is 13,3 Mi/6 and assume thatover the day the flow varies in a cyclic pattern with aminimum (MDWF) of 0,6. ADWF and a peak(PDWF) of2,0. ADWF, and the peak wet weather flow {PWWF) isestimated to be 3,0. ADWF.

For the settling characteristics of the sludge thedesign and operating chart is constructed as set outabove (see Figure 8.13). Assume that at PWWF,clarification must be achieved and that the thickeningcriterion will be met by adjusting the underflow recycle.The PWWF is 3.13,3 = 40 Mi /d. From Figure 8,13 themaximum overflow rate at X0 = 5kg/m3 is 21 m/d(given by horizontal clarification criterion line). Hencethe settling tank area is 40.103'21 = 1905 m2. In orderthat thickening failure does not occur under PWWFconditions, the recycle ratio at PWWF must be greaterthan 0,82 (given by the s value of the intersection pointof the clarification line and the hyperbola, see point Don Figure 8.13). Hence at PWWF of 40 M i / d theunderflow Q, should be 0,82.40 = 32,8 M i /d or 2,46times ADWF otherwise thickening failure will occur (i.e.for s < 0,82, the Q,/A and s intersection point will fallabove the thickening criterion line in Figure 8.13).

Under dry weather conditions, the overflow rate atPDWF is 2,0.13,3.103' 1905 = 14 m/d. In order that

thickening failure does not take place at PDWF, theunderflow recycle ratio s must be at least 0,52 (given bythe s value of the intersection point of the horizontalQ,-A- 14 m/d line and the thickening criterion line forXo = 5 kg/m3 see point A on Figure 8.13). Hence theunderflow should be at least 0,52 26,6= 13,9 Mi /d orapproximately 1:1 based on ADWF. Accepting the settl-ing tank area of 1905 m2 and fixing the recycle flow at aconstant 14 Ml ,'d, the overflow rate and recycle ratio atADWF are 7 m/d and 1,05:1, at MDWF are 4,2 m/dand 1,75:1 and the PWWF are 14 m/d and 0,52:1 re-spectively. Hence in the design and operating chart, thelocus of the points defining the operating conditionsover the day under cyclic influent flow moves betweenpoints of coordinates Qj /A=14m/d and s = O,52 atPDWF (see point A in Figure 8.13) and Q;/A = 4,2 m/dand s= 1,75 at MDWF (see point B in Figure 8.13)through Q,/A = 7,0m/d and s=1,05 at ADWF (seepoint C in Figure 8.13). From the chart, all these pointsrepresent safe operating conditions with respect tothickening and clarification and hence under dryweather conditions, the settling tank should operatesatisfactorily. Provision for wet weather flow was madeup to a PWWF of 3. ADWF but the underflow recycleneeds to be increased to 32,8 M i /d to accommodatethis. For lower PWWF, the underflow need not be in-creased so high. The required underflow can be deter-mined from the design and operating chart.

The above design is based on the accepted settlingcharacteristics V o=137m/d and n = 0,37 m3 'kg andshould the actual characteristics be better than thesethe settling tanks will still operate satisfactorily. How-ever, if the settling characteristics are poorer than theaccepted ones, difficulty in operating the settling tankmay be encountered. To a decree, a deterioration in set-tling characteristics can be accommodated under dryweather conditions by increasing the underflow recyclebut this strategy depends on how poor the character-istics are below the accepted ones; however, to in-crease the underflow recycle beyond the value wherethe thickening criterion no longer needs to be met willnot improve the situation and will be wasted on pump-ing costs. Under peak wet weather conditions, sludgewtli probably be lost depending on the duration of thepeak wet weather flow.

High peak wet weather flows can be accom-modated by the settling tank by allowing them tobypass the main biological process and dischargingthem to the effluent from the last aerobic reactor.Although the overflow rate in the settling tank still in-creases, the applied flux to the settling tank does not in-crease due to the dilution of the mixed liquor concentra-tion by the influent. The effect of this is to discharge tothe settling tank a significantly reduced operatingsludge concentration. In contrast, if the storm flow isdischarged to the head of the biological process, boththe overflow rate and applied flux would increase and alarge mass of sludge would be displaced from thebiological process to the settling tank. The emergencystrategy of bypassing storm flows is permissiblebecause usually storm flows are very low in COD, TKNand P concentrations and a short contact period withthe mixed liquor in the settling tank will significantlyreduce the COD and TKN; however, a higher P concen-

8-12

tration may have To be accepted for the short durationof the storm flow.

ditions in the return sludge, these clarifiers should incor-porate the following features:

7. PRACTICAL ASPECTS OF FINAL CLARIFIERDESIGN

The basic purposes of acivated sludge final clarifiers areto produce a clear effluent free of suspended solids anda thickened sludge for recycle to the inlet of the pro-cess. However, in nutrient removal versions, two addi-tional requirements are desired. The quality of theunderflow sludge should be such that a large mass ofnitrate is not recycled to the anaerobic zone. Also, if theanaerobic conditioning of the activated sludge can beinitiated in the bottom of the clarifier it will benefitphosphate removal.

Full-scale experience on sloping bottom scrapedclarifiers has shown that the solids in the sludge layermove in semi-plug flow and despite their endogenousnature, can denitrify quite well. For example, up to5 mgN/I nitrate can be lost on passing through thislayer and if denitrification is complete, anaerobic condi-tions are initiated. The phosphate released under theseanaerobic conditions will not cause problems as it isusually trapped in the sfudge layer and moves out withthe return sludge before diffusing into the effluent.

If the sludge can be encouraged to thicken well inthe clarifier, denitrification will be favoured even more.For example, if the sludge is thickened up to give arecycle concentration of 20 000 mg/ l instead of10 000 mg / i , half the mass of nitrate will be returnedeven if no denitrification occurred at all. However, at20 000 mg/ l solids the ratio of biomass to nitrate is verylarge and despite the endogenous nature of the bacteriapresent, very good denitrification will occur in mostcases using up all the available nitrate. Under this situa-tion there will be no nitrate feedback to the anaerobiczone. Of course, the ability to thicken up to10 000 mg/ / solids will depend on the sludge propertiesand will usually only happen with low SVI sludges (i.e.less than 100 mi/g). In this respect operators should beencouraged to ensure that such sludges are produced inthis process.

The denitrification and anaerobic conditions thatcan be encouraged in final clarifiers can only be ofbenefit to the process and designers should bear this inmind in clarifier design.

As mentioned earlier, good denitrification has beennoticed on full-scale sloping bottom scraped clarifiers aswell as in conical bottom (60° cone) settling tanks.However, these effects are less pronounced on flat-bottomed suction scraped clarifiers, as these aredesigned to lift settled sludge off the floor before it hasthe opportunity to denitrify. It has also been found thatsuction type clarifiers tend to return a relatively weaksludge and if attempts are made to thicken the sludgeand so obtain better denitrification, the sludge willrefuse to pass up the withdrawal tubes due to its in-creased viscosity. Also, the rotating suction arm tendsto cause hydraulic instability in the tank at high recyclerates.

Therefore with the view to encouraging denitrifica-tion in final clarifiers and the initiation of anaerobic con-

(i) they should preferably be circular and a slopingbottom with mechanised scrapers to move sludgeto a central bottom withdrawal point,

(ii) mixed liquor to be introduced near the surface atthe centre with the inlet system being well baffledto destroy kinetic energy of the influent,

(iii) the side water depth should be at least 4 m,

(iv) the sludge withdrawal system should be designedto handle sludges with a higher viscosity thanwater.

The clarifier surface area has a very important in-fluence on performance. It is recommended that thesolids flux theory (described in Sections 5 and 6 above)be used to determine this. The Vesilind equation;

V - Voe- n*

which describes the settling of activated sludge is an im-portant aspect of this theory with the two factors Vo

and n varying with sludge settling properties. Ex-perimentally derived values of these factors can show alarge variation depending on the degree of filamentousgrowth in the sludge.

Typical results obtained are shown in Table 8.1 inorder of deteriorating settling properties.

It was found that the ratio Vo/n is a very good in-dicator of settling properties which correlates with morewellknown parameters such as SVI. Thus, it can beseen in Table 8.1 that Plant D has a poor settling sludge;the sludge in Plant A settles fairly well. High values ofthis ratio (i.e. 30 kg/m2/h) indicate good settling pro-perties, while low values (i.e. 10 kg/m2/h) are foundwith poor settling sludges. Designers must decide whattypes of sludges are likely to occur in their plant duringits lifetime. If they are sure that poor settling sludges willnever be encountered, a V0/n ratio of 30 or above canbe used but if poor settling sludges are encountered,values of less than 15 must be chosen. However, thisdecision must be taken carefully as a value of 15 willlead to a clarifier design with twice the surface area aswould be obtained using Vo/n = 30. It should also benoted that experience to date has shown that biologicalnutrient removal plants can be prone to the encourage-ment of filamentous growths which give rise to poorsettling sludges so that a conservative approach toclarifier design might be prudent.

Plant operation can also have a dramatic effect onactivated sludge settling properties and hence VG/nvalues. An important variable is the dissolved oxygenconcentration in aerobic zones. It has been found bothin South Africa and overseas that plants run at lowdissolved oxygen concentrations (below 1,0 mg/ / andin particular, below 0,5 mg/i) are prone to the growthof filamentous bacteria which are the cause of poor set-tling sludges. Designers should therefore bear in mindthe quality of operation that will be available, in par-ticular, skills for the control of dissolved oxygen inputand process overloads.

Another factor to be remembered is that the solids-

8-13

A.

ai

C.

D.

TABLE 8 1 SETTLING CHARACTERISTICS

Plant

NORTHERN WORKS MODULE 2mean valuebest result^vo'st result

NORTHERN WORKS MODULE 1mean valuebest resultworst result

GOUDKOPPIES MODULE 2mean valuebest resultworst result

GOUDKOPPIES MODULE 1mean valuebest resultworsi result

: convert to units per day, multiply by 24.

OBSERVED AT FOURJOHANNESBURG

Vo

(m'h)»

7,909,496,07

7,949,497,02

6.689.843,84

5,958.965,20

DIFFERENT

nImVkg)

0,290,280,35

0.360.270.40

0.400,310.37

0,430,410,55

ACTIVATED SLUDGE

V0/nIkg /m ' /h r

27,2-335,8517.34

22,0635,1617.55

16,7031.710,4

13,821,99,5

PLANTS IN

SVI<m//g)

10065

180

12070

170

21070

300

250130300

flux theory covered in this chapter assumes an idealisedsettling tank which will not occur in practice andalthough it has been found Ihat some full-scale clarifiersobey the flux model within 20%, designers should makeuse of safety factors either in the initial choice of Vo andn or applied to the final ctarifier area calculated from thismodel.

8. REFERENCES

ALKEMA, K.L. (1971) The effect of settler dynamics on the activatedsludge process. MSc thesis, Depi. of Chemical Engineering. Univ.of Colorado.

BRATBY, J R. and MARAlS, G.v.R. (1976) A guide for the design ofdissolved-air (pressure! flotation systems for activated sludge pro-cesses Water SA, 2 (2) 87-100.

BRATBY, J.R. and MARAlS, G.v.R. 11977) Aspects of sludgethickening by dissolved-air flotation, Wat. Pollut. Control, 77. 13)421-432.

COE. H.S. and CLEVENGER. G.H. (1916) Methods for determiningthe capacities of slime settling tanks. Trans. Am. Inst. MiningEng., 55, 356.

DICK, R.I. (1967) Evaluation of activated sludge thickening theories, -J. Sanir, Eng. Div.. ASCE, 93, SA4.

DICK, R.I. (1970) Role of activated sludge final settling tanks, J.Sanir, Eng. Div.. ASCE, 96, 423.

DICK, R.I. and YOUNG K.W. (1972) Analysis of thickening perfor-mance of final settling tanks. Prod. 27th Ind. Waste Conf.. PurdueUniv., Series No. 161, Lafayette, Indiana.

KYNCH, G.J. (1952) A theory of sedimentation. Tr&ns. Faraday Soc,48, 166.

PITMAN, A.R. (19801 Settling properties of extended aerationsludges, JWFCF, 52, 3.

SWOLLEN. M. (1981) Behaviour of secondary settling tanks in ac-tivated sludge processes, MSc thesis, Dept. of Civil Eng., Univ. ofCape Town.

TRACY, K.D. and KEINATH, T.M. (1973) Dynamic model for thicken-ing of activated sludge. Water 1973, 70. 136. Chem. Eng. Pro-gress, Symp. Series AlCHE.

VESILIND, P.A. (1968a) Design of prototype thickeners from batchsettling tests, Wat. and Sew. Works, 115, 302.

VESILIND, P.A. (1968b) Discussion of evaluation of activated sludgethickening theories, J. Sanit. Eng. Div., ASCE, 94, SA1, 185-190.

WHITE, M.J.D- (1975) Settling of activated sludge. Technical ReportTR11, Water Research Centre, Medenham, England.

YOSHIOKA, N, HOTTA, Y., TANAKA. S.. NAITO, S. andTSUGAMI, S. (1957). Continuous thickening of homogenous floc-culated suspensions, Chem. Eng., Tokyo, 21, 66.

8-14

CHAPTER NINE

MICROBIOLOGICAL ASPECTS

by

L Buchan

1. EUKARYOTIC AND PROKARYOTIC CELLS

The basic units of living organisms are cells. Organismsare divided into two major sub-groups on the basis ofthe internal structure of their cells. Two distinct types ofcells exist, i.e. eukaryotic and prokaryotic cells.Eukaryotic cells are the more complex and althoughtheir complexity may vary in different organisms, it isthe basic structural unit of plants, animals, rotifera, protozoa, fungi and algae. Protozoa, algae and fungi arerelatively simple organisms and have been placed in asub-group of Ihe eukaryotes, that is, the protists. This isdone to distinguish them from the more complex plantsand animals. The prokaryotic cell is a less complexstructure and is the structural unit of bacteria.

2. PROTISTS IMPORTANT IN ACTIVATEDSLUDGE

Fungi and algae are not considered to be important inactivated sludge.

2.1 Protozoa

Protozoa exhibit a wide range of form and way of life.Three classes are important in activated sludge. TheRh/zopoda engulf food particles within the pseudopodiaby which they move, i.e. phagotrophic ingestion; solu-ble foods may, however, also be absorbed through theceil membrane, i.e. osmotrophic ingestion. TheMastigophora move by means of whiplike flagella andthese organisms may also be phagotrophic and orosmotrophic. The ciliophora are motile by means ofnumerous short hairlike projections which are termedcilia. The Ciliophora in activated sludge are mostlyphagotrophic. The Ciliophora are divided into free-swimming types, for example, Paramecium, crawlingtypes, e.g. Aspidisca and stalked types, e.g. Vorticella.

2. 1. 7 Metabolism and growth

Metabolism is the process whereby an organism obtainsthe necessary energy and cell building blocks to surviveand proliferate. Protozoa respire aerobically, i.e. requireoxygen for their metabolism. Organisms that ingestsoluble matter are the primary feeders and agents ofpurification; they thus form what is termed the basictrophic level. Phagotrophic animals occupy higher

levels in the system in that they prey on the primaryfeeders, i.e. they form the second trophic level.

A definite succession of protozoa occurs when ac-tivated sludge forms and as it becomes more efficient.The food of the primary feeders is the organic matter inthe sewage, thus bacteria (osmotrophic feeders) andosmotrophic protozoa, mostly rhizopods andflagellates, are in direct competition for their foodsupply. The osmotrophic protozoa cannot competewith the more efficient bacteria and as the latter in-crease in number, the osmotrophic protozoa areeliminated. Generally then, the osmotrophic Rhizopodaand flagellates occur in systems just starting up, in in-efficient systems or when organic concentrations arehigh, The phagotrophic flagellates then appear beforethe ciliates because they have relatively lower energy re-quirements than the citiates. However, because of theless efficient feeding mechanism of the flagellates theciliates start predominating. Within the ciliates there isalso a succession of species. Free-swimming ciliatespredominate when there are a large number of free-swimming bacteria and their presence thus points topoor flocculation. The attached and crawling ciliateshave lower energy requirements and thus replace thefree-swimming forms as their food becomes limited.The attached and crawling forms, together with higheranimals such as rotifers, indicate a well-flocculatedsludge and a low COD effluent, i.e. they indicate thatthe number of free-swimming bacteria have been reduc-ed below the demands of the free-swimming ciliates.

3. PROKARYOTES

The prokaryotes (bacteria) are small osmotrophicunicellular organisms. Some of these organisms canform multicellular filamentous forms of growth (e.g.Sphaerotilus) and some can form a mycelial (fungal-like) structure {Actinomycetes). Although a certainamount of morphological variation exists among theseorganisms, their differences are primarily expressed interms of metabolic behaviour, i.e. the organisms areclassified in terms of what they do rather than on theirappearance.

The average size of the prokaryotes is considerablysmaller than that of the protists. Whereas the volume ofa structural unit of the eukaryotes is 5 to 150 000 000jim3, that of the prokaryotes is 0,01 - 5 000 fjm3.Because the rate of metabolism is inversely proportionalto the size of the cell, the small size of bacteria confers aselective advantage where bacteria must compete withprotists for available nutrients,

9-1

3 1 Nutritional requirements

Bacteria are extremely diverse in their requirements fornutrients and this property forms a basis for theirclassification.

3. 1. 1 Carbon

Most organisms obtain their carbon from organicnutrients for satisfying their biosynthetic and energeticrequirements. Such organisms are termed hetero-trophic. Some organisms can obtain their energy fromthe oxidation of inorganic compounds such as ammonia[Nitrosomonas and Nitrobacter). These organisms ob-tain their carbon for cell synthesis from COj and theselatter organisms are termed autotrophic.

Organisms such as Pseudomonas and Acineto-bacter can use many different compounds as carbonand energy sources, whereas others are extremelylimited.

3.1.2 Nitrogen and sulphur

Nitrogen and sulphur occur in the building blocks ofcells. Many cells can obtain these elements fromnitrates and sulphates, others must obtain the elementsIn a reduced form, e.g. as ammonium salts. Theseelements may also be present in the organic materialmetabolised by the bacteria and thus become availableto them,

3.1.3 Growth factors

Organic compounds which cells require, but cannotsynthesize, ate known as growth factors, i.e. thesecompounds must be supplied to the organisms.Organisms vary greatly in their growth factor require-ments and this property exerts a considerable selectivityeffect in any given system. This is one of the reasonswhy organisms such as Pseudomonas and Acineto-bacter, which have very low or no growth factor re-quirements, proliferate in activated sludge, whereasfastidious organisms like the enteric bacteria, rapidly dieoff,

3.2 Bacterial energetics

The basis for the diversity of chemical activities amongbacteria and for the extensive array of extreme en-vironments in which they are to be found, lies in themethods they have available for obtaining the energy todrive metabolism. These organisms are thus capable ofusing various electron acceptors as alternatives tooxygen, viz photosynthesis, anaerobic respiration andfermentation.

3.2.1 Aerobic respiration

Organisms that require molecular oxygen (02) for

respiration, i.e. molecular oxygen functions as the ter-minal electron acceptor (oxidising agent), are termedobligate aerobes.

3.2.2 Anaerobiosis

Two general mechanisms exist which permit bacteria togrow in the absence of oxygen, viz. fermentation andanaerobic respiration. While certain species areclassified as facultative anaerobes in that they are ableto adapt their metabolism to grow either in the presenceof oxygen or its absence, others are obligate anaerobesand can only grow in an oxygen-free environment.

(0 Anaerobic respiration

The principal electron acceptors which can serve asan alternative to oxygen are nitrate, nitrite,sulphate and carbon dioxide.

iiit Nitrate and nitrite as terminal electron acceptors

Some facultative anaerobes, in the absence ofoxygen can make use of nitrate as the terminalelectron acceptor where the oxygen in the nitrateion is used instead of molecular oxygen. Nitraterespiration is the term used for the reduction ofnitrate to nitrite. It is a widespread reaction and canbe carried out by many genera includingPseudomonas, Bacillus, Clostridium and variousEnterobacteriaceae. Denitrification is the term usedfor reduction through nitrite to nitrogen. This pro-cess is confined to a limited number of species,e.g. Para coccus denitrificans and Thiobacillusdenitrificans. Both these organisms can use nitrateas the terminal electron acceptor and CO2 as car-bon source.

(iiil Sulphate as terminal electron acceptor

These suphate reducing organisms are strictanaerobes and their respiratory mechanism canonly be coupled with sulphate or CO2. Their nutri-tion is restricted in that only a limited range ofsubstrates can serve as electron donors. There arebasically only two genera, i.e. Desulfovibrio andDesulfotomaculum. These organisms thus coupletheir substrate oxidation to the reduction ofsulphate to sulphide. Their electron donors arelimited to the oxidation of hydrogen, pyruvate, lac-tate, ethanoi, formate or maiate, which are con-verted to acetate and C02. These organisms arethus only found in sulphate-rich anaerobic en-vironments in association with other organismswhich excrete products such as lactate.

(iv) Fermentation

This is the process where oxidation of one sub-strate is coupled to the reduction of another.Fermentation occurs in a number of bacterialgroups and genera, i.e. the lactic acid and pro-pionic acid bacteria, the Enterobacteriaceae,

9-2

Staphylococcus, Bacillus and Qostridtum. AllCiosthdia are obligate anaerobes and extremelysensitive to even small quantities of oxygen. Thepropionic acid bacteria are also strict anaerobes butdo have some oxygen tolerance. The otherorganisms are facultative anaerobes and are thuscapable of aerobic respiration in the presence ofoxygen. Fermentation processes have a low yieldof energy and hence growth for the amount ofnutrient consumed; these processes lead to the ex-cretion into the surrounding medium of fermenta-tion products such as short chain falty acids andalcohols.

3.3 Summary of nutritional categories amongbacteria of importance in activated sludge

The most simple classification is that based on thenature of the energy and principle carbon source.Organisms that are dependent on a chemical energysource (this is applicable to all organisms in activatedsludge) are termed chemotrophs. Phototrophs areorganisms that can use light as the energy source.Organisms able to use CO2 as principle carbon sourceare termed autotrophs and organisms dependent on anorganic carbon source are termed heterotrophs.

3.3.1 Chemoautotrophs

These organisms use a chemical energy source and CO;as the principal carbon source. Energy is obtained bythe oxidation of reduced inorganic compounds such asNH3, N02, reduced forms of sulphur (H2S, S) and fer-rous iron. Ammonia is oxidised to nitnte by organismsbelonging to the genera Nitrosomonas, Nitrosospira,Nitrosococcus and Nitrosolobus. Nitrite is then oxidisedto nitrate by Nitrobacter, Nitrospira and Nttrococcusspp.

Reduced suiphur compounds are oxidised by strict-ly aerobic organisms and also by the organismThiobacillus denitrificans which can use nitrate as theterminal electron acceptor. The Thiobcicillus spp are themost widespread sulphur oxidising organisms. Thefilamentous Beggiatoa-Thiothrix group also oxidisesulphide. Ferrous iron is oxidised to ferric iron byorganisms such as Thiobocillus ferro-oxidans, Fer-robaciilus ferro-oxidans and the stalked bacteria of thegenus Gallionella. The filamentous Sphaerotilus andLeptothrix spp can also oxidise Fe2* but these two latterspp are heterotrophs and their importance in Fe2' oxida-tion is unclear.

3.3.2 Chemoheterotrophs

The chemoheterotrophs include rotifers, protozoa,fungi and most bacteria. These organisms use achemical energy source and an organic substrate as thecarbon source. Because these organisms obtain theirenergy from the oxidation of organic compounds theyare often termed chemo-organotrophs. The bulk of thebacteria in activated sludge are considered to be gram-

negative, chemoheterotrophic (chemo-organotrophic)rod-shaped bacteria.

4. SEWAGE AS A SUBSTRATE FOR MICRO-ORGANISMS

All micro-organisms require certain basic nutrientelements. Apart from the carbon required for energyand synthesis, adequate supplies of various otherelements must be present in assimilable form. All micro-organisms require nitrogen, phosphorus, sulphur,potassium, calcium and magnesium. Trace elementssuch as iron, copper, zinc and cobalt are also requiredby many species and more fastiduous organism? alsorequire growth factors such as vitamins. Carbon,nitrogen and phosphorus are required by micro-organisms in balanced amounts, the ratios can differbut a reasonable estimate for the ratio is COD:N:P of100:6:1,5. In the activated sludge process however, thisratio can be mostly diffeient. Experience with full scalenutrient removal plants has shown up ratios more in theregion of 100 total COD:3 total N to 0,5 total P. This isbecause only a fraction of the incoming constituents areinvolved in bacterial synthesis. Large quantities of inertor non-biodegradable material can be present which areremoved by processes other than metabolism {i.e. ad-sorption onto floe particles). Nitrogen is only consideredfully available when present as ammonia andphosphorus as soluble phosphate, although other formsof these elements can be converted to ammonia andphosphate and thus become available.

Domestic sewage usually provides a nutritionallybalanced food with the necessary elements andvitamins for bacterial activity. However, for the pro-cesses required for denitrification and enhancedphosphorus removal, the organic carbon source is ofteninsufficient.

5. ORGANISMS PRESENT IN SEWAGE ANDTHEIR FATE IN ACTIVATED SLUDGE

The predominant bacteria in sewage are those found inhuman intestines. However, most intestinal bacteria arestrict anaerobes and die rapidly on leaving their host.The coli-aerogenes group is probably the principalgroup of organisms entering a plant; these intestinalbacteria function optimally at 37 °C, and as activatedsludge plants operate at 15 to 15 °C, these organismsdo not function effectively. Furthermore, the dilutenutrient concentration in an activated sludge plant alsodoes not favour the proliferation of the intestinalbacteria. Thus, most of the organisms entering theplant with the sewage, die rather than grow. Theorganisms that do proliferate in a plant enter by wind orinfiltration into sewers. Such organisms include speciesof Pseudomonas, Alcaligenes and Acinetobacter. Thetemperatures and low nutrient concentrations in ac-

9-3

tivated sludge plants are natural conditions for thesaprophytic organisms.

6. SELECTIVE PRESSURES AND THESUCCESSION OF ORGANISMS IN THEACTIVATED SLUDGE ENVIRONMENT

6.1 General

In activated sludge the relative number of each specieswill be determined by their growth rates, availability ofsuitable food and predation. Furthermore, physical con-ditions prevailing in the plant will have a different effecton the rate of proliferation of various species. Such fac-tors include temperature, availability of oxygen, pH andmode of agitation. Of the many species entering theplant only a relatively small number wil! find the environ-ment suitable and these species will dominate thepopulation to differing degrees. However, because ofsuccessive different conditions in the plant and influent,it is impossible for specific species to totallypredominate. The organic waste to be treated forms thefood of the primary feeders, i.e. the bacteria andosmotrophic protozoa. Phagotropnic protozoa androtifers prey pn these osmotrophic forms. Theautotrophic bacteria require iess complex energysources and are thus not in competition with theheterolrophic organisms. The numbers of predators areinfluenced by the number of prey and this is also in-fluenced by the extent to which the prey can seekrefuge from the predator. The degree of flocculation issignificant in this respect, as also is the extent to whichthe bacteria can clump or form clusters too large for thepredators to ingest, Obviously organisms which clusterwill also be retained in the system whereas free-swimming forms will be washed out.

The dominant organisms in terms of activity are notnecessarily those occurring in the largest numbers butrather those which constitute the bulk of the biomassand/or those that make the largest contribution to thetotal metabolic activity.

The dominant bacteria will be those capable ofmost effectively utilising the organic waste and of form-ing floes to ensure their retention in the system. Thenature of the bacteria in the plant is thus chiefly deter-mined by the nature of the organic waste and secondlyby conditions in the plant, chiefly the degree of aerationand particularly anoxic, anaerobic and aerobic fractionsof the total solids retention time and the magnitude ofthe latter in nutrient removing activated sludge pro-cesses.

6.2 Selection of organisms in an aerobicenvironment

The dominant aerobic bacteria present in such systemsappear to be gram-negative rods belonging to thegenera Pseudomonas, Achromobacter, Alcaligenes andFiavobacterium. Pseudomonas and Acinetobacter

species are versatile chemoheterotrophs and can use awide range of common substrates as sole energy andcarbon sources, Acinetobacter spp can generally onlyutilise short chain carbon compounds as carbon andenergy sources. If these compounds are present in asewage Pseudomonas spp will generally predominatewith low agitation but in a well-agitated mediumAcinetobacter spp will often predominate, The reasonfor this phenomenon is that whereas the Pseudomonasspp are motile organisms 2nd thus capable of migratingtowards food and oxygen, the Acinetobacter spp arenon-motile. The latter organisms, although metabolical-ly very active and competitive, thus need to be broughtinto contact with their substrates and oxygen.

6.3 Selection of organisms in a nitrogen removinganoxic'aerobic system

In a totally aerobic system certain aerobic heterotrophsand autotrophs will predominate, depending principallyon the nature of the sewage.

The inclusion of a zone devoid of oxygen but con-taining nitrates as electron acceptor will causeorganisms capable of nitrate respiration and denitrifica-tion to proliferate. These organisms have a preferencefor short chain organics as substrates and will thusdeplete these compounds present in the influentsewage. This in turn will affect the nature of the popula-tions established in the aerobic zone, as aerobicorganisms which can only utilise short chain organicswill not be able to grow unless short chain organics canbe produced extracellularly from adsorbed longer chainorganics.

6.4 Selection of organisms in an anaerobic/anoxic/aerobic system

As substrate and mode of aeration are the principal fac-tors in the establishment of microbial populations, thecombination of an anaerobic/anoxic/aerobic sequencein an activated sludge process will result in a restructur-ing of populations in comparison to an aerobic or anox-ic/aerobic system.

In the present context, anaerobic refers to theabsence of oxygen and nitrates, and anoxic refers to thepresence of nitrates and the absence of dissolvedoxygen. The anaerobic zone will lead to the enrichmentof organisms capable of fermentation, i.e. organic com-pounds serve as electron donors and electron accep-tors. Organisms normally living in soil and water andcapable of fermentation, are inter alia, species ofEnterobacter, Serratia, Proteus and Aeromonas. Theseorganisms are also capable of aerobic respiration andwould thus be capable of surviving the alternatinganaerobic 'aerobic conditions prevailing in the plant. Asfermentation is an inefficient mode of energy produc-tion, the cell yield per unit mass of substrate utilised isconsiderably less than under aerobic conditions and thepopulation increase of the fermentating organisms willthus be limited.

The fermentative organisms mentioned are alsocapable of growing under aerobic conditions. However,

9-4

the transition from a fermentative metabolism to arespiratory metabolism does not occur rapidly and con-sequently, by the time these organisms have switchedtheir metabolism to a respiratory mode in the aerobiczone, available substrate will probably have been utilis-ed by the obligate aerobes. This will thus limit thepopulation increase of these organisms under aerobicconditions. The conversion from a respiratory mode to afermentative mode of metabolism is rapid and theorganisms have the necessary pathways operational forfermentation before the DO has reached zero, i.e.fermentation can commence virtually immediately whenthe organisms enter the anaerobic zone.

During fermentation compounds such as lactic,succinic, propionic, butyric, acetic acids and ethanolare produced, and as these compounds cannot beutilised further under anaerobic conditions, they are ex-creted into the medium. These highly biodegradablecompounds can then be used in the anoxic zone fordenitrification and the balance thereof will pass onto theaerobic zone where aerobic organisms, particularlyAcinetobacter sp, which can store PHB in the anaerobiczone, can then proliferate. It has also been suggestedthat certain aerobic organisms can accumulate theorganics excreted in the anaerobic zone by the fermen-tative organisms and store these compounds in theform of internal carbon and energy reserves, probablyas poly-/3-hydroxybutyrate (PHB). As Acinetobacter sppare metabolically very active the availability of theirpreferred and prestored intracellular substrates in thepresence of oxygen could thus lead to their proliferationin the aerobic zone. It has been shown at the ContraCosta Phostrip plant in California that up to 10 hours ofanaerobiosis has no effect on respiratory rates whenaerobic activated sludge organisms are once again ex-posed to oxygen.

Anaerobic conditions thus probably do notadversely affect the aerobic populations established inthe aerobic zone.

It has been shown that certain Acinetobacter spppredominate in activated sludges removing excessamounts of phosphorus and these organisms containlarge inclusions of polyphosphates. These organismshave also been observed to release some of this ac-cumulated phosphorus under anaerobic conditions,thus providing a probable explanation for the increase inortho-phosphate often observed in the anaerobic zone.Chemical adsorption,-desorption effects could also berelevant to the nett release and uptake observed inanaerobic zones. In enhanced-phosphorus-removingsludges the ceils are mainly present in large clusterscontaining hundreds of cells. Intracellular polyphos-phate inclusions can be observed in activated sludge bystaining dried smears of mixed liquor on microscopeslides for 1 min with an aqueous solution of 1 %methylene blue plus 1 % borax. The slide is rinsed inwater, allowed to dry and observed by light microscopy.Poly phosphate inclusions stain a purplish colour,whereas the cells and sludge matrix stain a blue colour.The property of a dye to change its colour when itreacts with certain compounds, is termed meta-chromacy, i.e. polyphosphates stain metachromaticallywith methylene blue. That these purplish inclusions arerich in phosphorus has been confirmed by electron

microscopy combined with microprobe analysis. By thelatter method is has been estimated that thepofyphosphate granules contain in excess of 25 % P(m/m) and in excess of 8 % Ca (m/m). It has also beenobserved that these inclusions can occupy the bulk ofthe cell volume.

Although a controversy stiil exists as to whetherAcinetobacter spp are the principal organisms responsi-ble for enhanced phosphorus uptake, the dominance ofthese organisms has been shown in many enhanced-phosphorus-removing plants. The assay of theseorganisms has been performed by plating, biochemicaland immunofluorescent techniques. The abundance ofthese organisms has also been indicated by the observa-tion that protozoa in such sludges contain mainly thecharacteristic Acinetobacter ceils.

Acinetobacter possesses a relatively characteristicmorphology, i.e. short, plump rods 1,0 -- 1,5 ^m by1,5-2,5 ^m in logarithmic phase approaching coccusshape in stationary phase and predominantly in pairsand short chains. The cells are very large compared tomost other bacteria. Acinetobacters are non-motile,gram negative and oxidase-negative.

An Acinetobacter species isolated from enhanced-phosphorus-removing sludge has been acknowledgedas a distinct species, the organism is designated Aphosphodevorans (i.e. devourer of phosphoius).

If these organisms are the principal agents ofenhanced-phosphorus uptake, the question still to beanswered is how these organisms can be enriched andmaintained within a system, i.e. whet causes theseorganisms to have a selective advantage within the ac-tivated sludge environment.

The fact that many plants performing enhancedphosphorus removal receive septic sewage, points tothe substrate nature as being important in enrichingthese organisms. The augmentation of the feed to the50 M i /d plant with the supernatant of an acid digesterhas been performed with great success, whereas theaugmentation of the same plant with a similar additionalCOD but not of acidic nature, did not succeed in induc-ing enhanced P-uptake. Whether other factors such asthe exposure of the organisms to an anaerobic condi-tion is necessary stiil needs to be elucidated.

It thus appears as if the anaerobic zone providessuitable substrates for the proliferation of the aerobicphosphorus-accumulating bacteria. This is supportedby the observation that the mixed liquor of phosphorus-removing sludges is often very black in colour. Thesulphate-reducing bacteria can only use a restrictedrange of compounds, e.g. lactate or malate as theirsource of carbon and energy. Such compounds are rarein fresh sewage and the sulphate reducers musttherefore utilise the metabolic products of fermentationprocesses which in turn utilise more complex organiccompounds, for example, cellulose. The sulphide form-ed by these bacteria reacts with ferrous iron to formblack ferrous sulphides. The production of sulphidethus indicates that fermentative processes do occur inplants with black MLSS, The presence of sulphides alsoencourages the growth of Beggiotoa and TNothrix sppof filamentous organisms often observed in phos-phorus-removing sludges. The filamentous Sphaeroti-lus Leptothrix group are also encouraged to grow in

9-5

such anaerobic systems as their preferred substrates are and it may be for this reason thgt their presence in thealso formed during fermentation. anaerobic basin inhibits overall phosphorus removal by

Nitrates are inhibitory to fermentation processes the plant.

For Photomicrographs see Appendix 5, page A5.1.

96

CHAPTER TEN

PRACTICAL DESIGN CONSIDERATIONS

by

G.F.P. Keay

1. INTRODUCTION

This section is not intended as a comprehensive anddetailed guide to the design of biological treatmentplants but as a guide to emphasize those aspects of thedesign of biological nutrient removal plants which re-quire particular attention. As such only the broad designprinciples which should be applied in the design of theseplants are outlined. As each particular plant is in manyrespects unique, the detailed implementation of theprinciples is left to the Design Engineer.

2. PLANT CONTROL

There are a number of design parameters which can bedetermined and fixed at the design stage and otherswhich can be selected subsequently during plant opera-tion. In the case of some parameters, selection is com-pletely controlled by the designer whereas in the case ofother parameters, the designer exercises limited controlonly. The parameters chosen relate to process input,process output and factors internal to process control.

2.1 Process input parameters

Inputs to the process over which the designer may exer-cise full control are:

(a) Influent flows including peak flows and recycleflows.

The extent to which the process must be designedto cater for peak flows and diurnal flow variationscan be modified by the use of flow balancing orequalization. Where excess flows are likely to havea detrimental effect on the process, these flowsmay be diverted around sensitive stages in the pro-cess scheme. Thus provision could be made in thedesign for peak storm flows and certain recycleflows to be diverted to the aerobic zone, bypassingthe anaerobic and anoxic zones and avoiding thedetrimental effect resulting from dilution andreduced retention times in these zones. Similarly,flows resulting from severe storm peaks could bebypassed to the end of the aeration zone. The needfor bypass facilities depends on the magnitude ofthe peak flows in the system and the assessed sen-sitivity of the system to these flows.

(b) Influent load.

Variations in influent load (as distinct from influent

flow) can be reduced by the introduction of a loadbalancing step to the process scheme. Methods bywhich this can be achieved will be considered at alater stage.

(c) Dissolved oxygen levels.

The desired residual dissolved oxygen level isselected at the design stage and the design of theaeration system oxygenation capacity is based onthis level. In selecting the residual dissolved oxygenlevel, the minimum desirable level necessary forreliable nitrification and the production of a sludgeof good settleability should be provided. This isgenerally accepted as a dissolved oxygen concen-tration of not less than 2 mg/f.

There are some parameters over which the designer canexercise only limited control. These include:

{a) Influent sewage characteristics which can to someextent be predetermined by an effective industrialeffluent control policy, and

(b> by the incorporation of selected additional stages inthe process scheme such as:

(i} Primary sedimentation to reduce the organicload on the biological treatment stage in thecase of high strength wastewater with afavourable TKN:COD ratio; or

(ii) high rate primary sedimentation to partiallyreduce the organic load on the biological treat-ment stage while maintaining a favourableTKN:COD ratio; or

(iii) load balancing achieved by subjecting a pro-portion of the influent to primary sedimenta-tion and using the primary sludge so derived toincrease the organic load of the influent duringperiods when lower than average loads are ex-perienced.

The parameters that are dependent on the pro-cess are sludge age and effluent quality.

2.2 Process output parameters

The designer can select the design sludge age or adesign range for the sludge age within normally ac-cepted limits.

The standard of effluent to be produced by thetreatment process will, in most cases, be prescribed bythe regulatory authority.

10-1

2.3 Other process considerations

Based on the influent sewage characteristics and the re-quired effluent quality, sludge and mixed liquor recyclerates can be selected, and 3 suitable biological reactorconfiguration chosen. In selecting the biological reactorconfiguration the possibility of increasing or decreasingzone detention times to cater for the initial flow condi-tions and possible future changes in the character of thesewage should be considered and provided for ifnecessary. This can be achieved by compartmentaliza-tion of the anaerobic and anoxic zones, providing forstep feed of sewage to the various zones of the reactorand the installation of appropriate mechanical equip-ment, such as a combination of mechanical mixers andassociated air spargers in all zones.

3. INSTRUMENTATION

The successful monitoring and control of the biologicalnutrient removal process depends on robust, reliableand easily maintained instrumentation. In the selectionof instrumentation, the initial cost, the availability oflocal agents and their capability to service and repair theequipment must be taken into account.

In areas with a high incidence of lightning, thepossibility of lightning damage lo electronic instrumentsmust be considered. Lightning protection devicesshould be incorporated and special consideration mustbe given to the siting of instruments and the location ofsignal cables to minimize the likelihood of lightningdamage.

3.1 Flow metering

To facilitate process control, the following processstreams must be metered. The flow meters should in allcases incorporate instantaneous rate of flow indicationend flow totalization.

(1) Influent to the plant.(2) The flow to each treatment module.(3) The flow of waste sludge.(4) The studge recycle flows.(5) The mixed liquor recycle flows.

In small plants where the installation of flow meterson the waste sludge, sludge recycle and mixed liquorrecycle flows cannot be economically justified, Venturiflumes should be constructed for these flows and theflow rates checked at regular intervals using a dip stickand flume calibration chart.

3.2 Oxygenation control

Oxygenation control is important from two aspects:

(1) it enables energy usage to be optimized, and(2} it enables the dissolved oxygen concentration to be

maintained at levels necessary for nitrification andgood sludge settleabiiity.

Where the character of the sewage is such that

biological nutrient removal becomes marginal, the abili-ty to closely control the oxygen levels in the biologicalreactor becomes extremely important. Dissolved oxy-gen levels should be controlled so that the levels arehigh enough to ensure good nitrification and sludgesettleabiiity but not so high that denitrification orphosphate removal are adversely affected.

Oxygen control methods include:

(1) Dissolved oxygen monitoring equipment:

The most common system in use today operateselectronically by means of a dissolved oxygen sen-sor which is submerged in the mixed liquor of thebiological reactor and which provides a -directreading of dissolved oxygen level at the pointwhere it is situated. The output signal can be usedto automatically control the oxygen input to theprocess. The major advantage of the system is thatthe equipment is readily available from a variety ofmanufacturers. The major disadvantage is that thesensors are sensitive to membrane fouiing andregular cleaning of the membrane {at least onceweekly) is required to ensure satisfactory opera-tion.

Portable dissolved oxygen meters withrecorders can be used to determine the temporalvariation of oxygen demand. This demand can thenbe matched by setting up time controls to switchthe aerators on and off according to the predeter-mined demand pattern.

(2) Methods other than the direct monitoring ofdissolved oxygen concentration which are present-ly in the developmental stage are:

(a) On-line monitoring of the ammonia concentra-tion of the aerobic zone. A low ammonia con-centration indicates that sufficient oxygen isbeing provided to ensure full nitrification. Thismethod depends on the presence of a smallammonia residual in the mixed liquor since ifzero ammonia is detected, the possibility existsthat excess oxygenation is occurring.

(b) By measuring respiration rate of the activatedsludge, the oxygen demand of the micro-organisms can be determined and the oxygeninput and oxygen demand matched.

fc) Total oxygen demand measurements. Bymeasuring the total oxygen demand of the in-fluent, sufficient oxygen can be provided tomeet the measured demand.

3 3 Temperatures

Thermometers must be provided to measure both am-bient and liquor temperatures. As biological reactionrates are temperature dependent, knowledge of theliquor temperature is important for setting process con-trol parameters such as sludge age and recycle rates.

10-2

3.4 Energy consumption

To check the continuing efficiency of the process andequipment, the total energy consumed in the processand in particular the energy used by the aeration systemshould be recorded on kilowatt-hour meters.

3-5 Sludge blanket sensors

Process operation can be further optimized by monitor-ing the level of sludge in the clarifiers. In the case ofprimary sedimentation tanks, monitoring the sludgelevel in the tanks will provide data on the inventory ofsludge available for addition to the process or whichmust be treated by anaerobic digestion.

Where poorly settling sludges could occur themonitoring of sludge levels in the final clarifiers will pro-vide an early warning of imminent failure of theclarifiers. Appropriate counter-measures such as reduc-ing the influent to the plant, increasing the sludgewasting rate or reducing the sludge recycle rate can betaken timeousiy to prevent carry-over of sludge into thefinal effluent.

A system in which the sludge recycle rate is variedin direct proportion to the influent flow rate can alsoassist in greater process stability.

3.6 Sampling

Regular sampling of ihe process streams is essential formonitoring and control of the process. Sampling pointsshould be identified at the design stage and necessaryfacilities provided. Sampling points must be locatedwhere conditions are such that a representative sampleis obtained. The sampling point must also be readily ac-cessible.

When the sampling frequency at each point hasbeen determined, the method of sampling must bedecided on. Where sampling at frequent intervals is re-quired, on-line monitoring should be considered. Suchintensive sampling should only be carried out where it isessential for hour to hour process control. Sampling atless regular intervals can be carried out using automaticsampling machines.

In both cases a source of electric power should beprovided at the sampling point to power the samplingequipment.

3.7 Running hour meters

The implementation of a routine and preventativemaintenance programme for mechanical plant andmachinery on a works is facilitated if running hourmeters are provided on all plant and machinery.

4. DISSOLVED OXYGEN CONTROL

One key aspect in the design of biological nutrientremoval plants is the control of dissolved oxygen levelsin order to ensure true anaerobic and anoxic states inthe reactors. These states are essential for conditioningof the sludge for phosphate removal and for denitrifica-tion respectively.

Direct or indirect transfer of oxygen to the liquidcan occur through the aeration system or as a result ofturbulent hydraulic conditions.

4.1 Indirect oxygen transfer

Indirect transfer of oxygen to the liquid can occur as aresult of:

• Steep site topography which gives rise to hydraulicdrops in the distribution channels, pipework and intothe biological reactor. (The same aeration effect canalso occur in the sewer reticulation system).

• Hydraulic turbulence at venturi flumes and sharpbends in channels and pipework

• The lack of adequate internal baffling in the reactorwhich allows backmixing of aerated liquor into theanoxic and anaerobic zones.

• The use of weirs for hydraulic control; especially onprimary sedimentation tanks, dividing tanks and ad-justable weirs for controlling the level of dissolvedoxygen in the reactor.

• Hydraulic turbulence caused by vortexing, inducedby mechanical mixers in the anaerobic and anoxiczone.

.• Hydraulic turbulence resulting from the poor selec-tion of recycle pumps.

It has been shown that the dissolved oxygen con-centration in mixed liquor can be substantially increasedby its passage through Archimedian screw pumps. Ifthese pumps are used they should be located remotefrom the anaerobic zone in order to allow bacterialmetabolism to reduce the DO before the mixed liquorenters this zone. Air lift type pumps should not be usedfor obvious reasons. Mixed liquor recycle pumps shouldbe selected whose operation will result in minimum tur-bulence to the mixed liquor so that excess entrainmentof air is avoided.

Where the readily biodegradable fraction of thesewage is low, the use of an aerated grit chamber forthe removal of grit from the sewage should be avoidedsince the aeration may further deplete the readilybiodegrable fraction.

The following precautions can assist in eliminatingthe indirect entrainment of oxygen:

• Pipes should be used in preference to channels andwhere channels are used, the chosen channel cross-section should minimize the liquid surface area.

• By suitable positioning of baffle walls and directionof rotation of surface aerators, backmixing in thebiological reactor between the aerobic and anoxiczones can be utilized to eliminate the need for mixedliquor recycle pumps.

• Where possible, weirs should be designed assubmerged weirs.

10-3

• The plant site should be selected and the plant layoutarranged so that the drop in elevation {the slope ofthe hydraulic grade line) through the plant is minimiz-ed.

4 2 Aeration systems

It is important that adequate aeration capacity is provid-ed in the aeration zone where phosphorus uptake oc-curs. In selecting an aeration system, initial capital cost,oxygen transfer efficiency, operational flexibility,system reliability, maintenance costs, operating costsand ease of maintenance must be considered. Systemsin use include mechanical surface aeration, diffusedaeration, turbine aeration and pure oxygen systems.When determining the peak aeration demand, thepossible load balancing effect within the biological reac-tor and reduced peak oxygen demand should be takeninto account. The load balancing effect within the reac-tor will depend on the type of aeration system, the con-figuration of the aerobic zone, denitrification effectsand the magnitude of the internal recycle flows.

4.2. 7 Mechanical surface aerators

A design incorporating numerous small aerators favoursoperational flexibility, system reliability, efficiency anduniformity of dissolved oxygen concentration in theaeration zone. Against this must be weighed the in-creased capital and maintenance costs of such a design.As surface aerators operate most efficiently at maxi-mum immersion and electric motors operate most effi-ciently at maximum load, optimum efficiency is obtain-ed with a dissolved oxygen control strategy that mat-ches oxygen input and oxygen demand by stopping andstarting aerators or by using aerators with two-speedmotors rather than a system in which the depth of im-mersion of the aerator is varied.

The ability to effectively control dissolved oxygenconcentrations in the biological reactor depends on theuniformity of the dissolved oxygen concentration levelsthroughout the aeration zone. Surface aerators withgood mixing characteristics should promote uniformdissolved oxygen concentration levels, As a rule,dissolved oxygen levels in mechanically aerated reactorsare not uniform. The measured dissolved oxygen level isaffected by the proximity of the sensor to the aerator,whether nearby aerators are operating or not, and theproximity to the basin inlet. This non-uniformity ofdissolved oxygen levels must be considered whendeciding on the number and position of dissolved oxy-gen sensors to be provided in each aeration zone. Adissolved oxygen sensor located where it is likely to beaffected by any of the abovementioned phenomena isunsuitable for control purposes. In general a number ofdissolved oxygen sensors or sockets for installingdissolved oxygen sensors should be provided in theaeration zone. A suitable control sensor can be selectedfrom these depending on The operating conditions andexperience with the system. Because of the non-uniform dissolved oxygen levels in the reactor only ac-tual experience will enable the selection of a suitablecontrol sensor and suitable control limits for that sen-sor.

4.2.2 Fine bubble diffused aeration

A number of different systems utilizing this principle arecommercially available and each should be individuallyevaluated to assess its suitability for a particular applica-tion. Important in a fine bubble diffusion system is its"turn down" capability or the range of air flows overwhich the system is capable of operating. This capabili-ty may be of importance for example if the plant is to bevery underloaded at start up or if it is to be subjected tolarge variations in load. Systems using ceramic domesto produce fine air bubbles require a minimum air flowthrough the domes to prevent both fouling of thedomes and the ingress of water into the air pipes. Thislimits their "turn down" capability and thus their opera-tional flexibility. Systems in which the air flow can bestopped completely provide greater flexibility. Withsuch systems air flow to sections of the aerobic zonecan be stopped, if necessary, to provide a larger anoxicvolume in the reactor. Dissolved oxygen control is nor-mally accomplished by controlling the volume of airdelivered to the reactor by means of inlet vane controlon centrifugal type blowers. On smaller plants positivedisplacements blowers with variable speed drives arecommon. Where there is more than one reactor, controlof dissolved oxygen levels in each reactor should be in-dependent of the other reactors.

4.2.3 Turbine aeration system

In this system mechanical mixers are combined with airspargers to produce an aeration system which has anefficiency somewhere between that of a fine bubble anda coarse bubble aeration system. Although less efficientin energy usage than a fine bubble system ormechanical aeration system, it is attractive because ofits operational flexibility. By stopping the air flow to thesparger the aerator can be converted to a mixer. Depen-ding on the reactor design this system would allow virtually unlimited operational flexibility in determining theposition and volume fractions of the anaerobic, anoxicand aerobic zones.

4.2.4 Pure oxygen systems

There is little experience with the use of pure oxygen inbiological nutrient removal processes although onesuch system has been patented. The use of oxygen foraeration is most likely to find an application in hybridsystems designed either to correct existing process defi-ciencies or to upgrade existing works. The system alsohas applications where strong wastes are encounteredwhich cannot be handled with a conventional aerationsystem.

5. HYDRAULICS

In the case of the biological nutrient removal processes,particular attention should be given to the followingaspects of hydraulic design:

10-4

(1) To facilitate process control the inlet must bedesigned to ensure equal distribution of flow to thevarious treatment modules.

(2) Where flow balancing is included in the processscheme, the design of the balancing tank shouldattempt to eliminate or at least minimize thedeposition of solids in the tank. A means of remov-ing the deposited solids which does not cause highstrength sludge loads to be scoured from theequilization tank should be provided.

(3) In general the hydraulic design should attempt toeliminate the unintentional entrainment of oxygen.This can be done by eliminating cascades and otherunnecessary hydraulic turbulence. See section 4.1.

6. "BACK UP" FACILITIES

In situations where conditions for reliable biologicalnutrient removal are marginal or where, from time totime, the process is not able to produce an effluentcomplying with the prescribed standard, "back-up"facilities to bring the effluent into compliance with thestandard should be considered.

Possible methods of accomplishing this include:

(1) Chemical addition to the main flow stream or to asidestream of the process to chemically precipitateresidual phosphorus not removed by the biologicalnutrient removal process. The object is to utilizethe more expensive chemical precipitation processonly when conditions are unfavourable forbiological phosphorus removal. The use ofchemicals should be avoided unless absolutelynecessary to achieve the prescribed standard.

(2) When sub-standard effluents arise, arrangementsother than discharge to streams can be made todeal with them. Examples of strategies to preventthe sub-standard effluent from entering the receiv-ing water are:

(a) Storage of the sub-standard effluents whichcan then be returned to the process for retreat-ment when conditions are more favourable.

(b) Using sub-standard effluents as low grade in-dustrial water or, where sufficient land isavailable, for irrigation.

One of the main reasons why the biological nutrientremoval process fails to produce effluents comply-ing with the phosphorus standard Is the occurrencefrom time to time, of an unacceptably highTKN:COD ratio in the influent sewage. This pro-blem can be overcome by supplementing the in-fluent COD during periods of COD deficiency withCOD removed from the influent sewage duringperiods of excess organic load and stored for use insuch a contingency.

7. SCUM HANDLING AND CONTROL

Under certain conditions the biological nutrient removalprocess is characterised by the growth of filamentousorganisms which results in the formation of filamentousscums on both the biological reactor and the finalclarifiers. This can give rise to bad odours and high ef-fluent suspended solids concentrations. These scumsare known to occur in plants both with and withoutprimary sedimentation. The cause of these filamentousscums is not known but they are believed to result fromthe occurrence of one or mere of the following:

(a) suppressed dissolved oxygen levels and excessivelylong anoxic retention times;

(b) the occurrence of temperatures and/or substrateswhich favour the growth of these organisms inpreference to other micro-organisms.

As the cause of these scums is unknown and thereis no known method of preventing or eliminating thescums, both the biological reactor and the final clarifiersmust be provided with adequate descumming facilities.Descumming troughs on the final clarifiers should bepositioned taking into account the prevailing wind direc-tion.

Methods which have proved successful in amelio-rating the problem are:

(1) spraying the scum layer with water to break it upand entrain the scum In the liquid;

(2) spraying the sum with chlorine solutions or theperiodic dosing of chlorine to the aeration basin.

Both methods appear to control the scum by sup-pressing the growth of the filamentous organisms.

8. TREATMENT OF WASTE ACTIVATED SLUDGE

The principles applied in the treatment of waste ac-tivated sludge from conventional biological processescan be applied in the case of sludge from biologicalnutrient removal processes but, in the case of biologicalnutrient removal sludges, special consideration must begiven to the excess phosphorus associated with thesludge which will be released into solution underanaerobic conditions. The sludge must be either main-tained in an aerobic condition or, if the sludge is allowedto go anaerobic, provision must be made for the treat-ment and disposal of phosphorus enriched liquors aris-ing in the treatment process.

The following principles should be applied:

(1) Mixed liquor should be wasted from the aerobiczone of the biological reactor where the dissolvedoxygen concentration is high in preference towasting from the clarifier underflow where dissolv-ed oxygen concentrations are lower and release ofphosphorus from the sludge could already be oc-curring.

10-5

(21 The sludge treatment facilities should be located inclose proximity to the biological reactor so thatlong conveyance of the sludge is avoided. Whenpumping the waste sludge is unavoidable, theretention periods in the pump station sump and inthe rising main should be kept as short as possible.

(3) The retention lime of the sludge in the thickeningand dewatering processes should be as short aspossible.

(4) Preference should be given to thickening anddewatering processes which will maintain thesludge in an aerobic condition.

In the case of sludge thickening, dissolved air flota-tion (DAF) and cenuifugation are the most suitable pro-cesses currently available for rapid dewatering.Thickening of waste activated siudge can be carried outby centrifugation or by means of filter belt presses. Inthe case of both gravity thickeners and filter presses,the sludge is retained in the process for a considerablelength of time during which it goes anaerobic and thephosphorus is released into solution.

Wherever possible the underflows, centrates,filtrates and wash waters from the thickening anddewatering process should be segregated so that anysidestream containing a high phosphorus concentrationas a result of phosphorus release from the sludge canreceive further treatment. Such treatment can take theform of chemical addition or land irrigation to im-mobilize the phosphate ion. Where the phosphorus-richwaste activated sludge is stabilized by anaerobic oraerobic digestion, the excess phosphorus in the sludgewill be released into solution and the supernatant liquorsfrom these processes must receive suitable treatment.

Dissolved air flotation thickening of waste ac-tivated sludge produces a sludge which is saturatedwith air. To avoid the possibility of air binding in thepumps, positive displacement pumps should be usedwith this sludge in preference to centrifugal pumps.

The sludge also has the rheological properties of anon-Newtonian fluid. The pumping system andpipelines must be designed accordingly.

9. SECONDARY CLARIFIERS

The secondary clarifiers are expected to produce a finaleffluent containing a minimum of suspended solids aswell as a highly concentrated underflow for recycling tothe biological reactor. In the light of the known poorsettling properties of these siudges, ciarifier designshould be based on the "solids flux theory" approach.The total ciarifier area must be sufficient to pass thedesign sludge mass at the predetermined flux and theciarifier must be deep enough to permit the requiredthickening of the underflow. Laboratory studies shouldbe carried out on a representative siudge to determinethese parameters.

To avoid anaerobic conditions developing in theciarifier with the subsequent release of phosphorus intosolution, the ciarifier should be designed so as tominimize the length of time the recycled sludge is retain-

ed in the ciarifier. In this regard the activity of the sludgeas measured by the respiration rate is important. Veryactive sludges will quickly utilize the available dissolvedoxygen and nitrates in the sludge liquid and anaerobicconditions wiil develop rapidly. Provided the releasedphosphorus is retained within the underflow liquid, thephosphorus will not be released into the effluent. Thelikelihood that the phosphorus will escape into the ef-fluent will depend on the depth of the ciarifier, theheight of the sludge blanket and the method by whichsludge is withdrawn from the ciarifier.

Circular ciarifiers with conical bottom and centredraw off of sludge under hydrostatic head, providedwith mechanical scraper mechanism and a depth notless than three metres but preferably four metres areconsidered to be most suitable. Tanks with thesecharacteristics are most likely to have hydraulic regimesapproaching the ideal "plug flow" condition.

Circular clarifiers with suction-type sludge removalmechanisms enable siudge to be removed rapidly fromthe ciarifier but this type of ciarifier has a number ofdisadvantages;

(1) They require much attention from the operator toensure that the syphoning action is maintained.

(2) The design should enable each suction pipe to bemonitored visually and controlled independently toensure that each is removing sludge with highsolids concentration.

(3) in essence the syphon method is not a "fail safe"design and if failure of the syphoning action is notdetected, the tank can fill with sludge and sludgeoverflows into the effluent can result.

The suction type mechanism produces an inherent-ly unstable hydraulic regime in the ciarifier which is like-ly to cause mixing and greater deviation from ideal"plug flow" conditions. It has also been observed thatexcessive rotational speed of the scraper mechanismcan cause turbulence in the sludge blanket. If it is pro-posed to use this type of ciarifier, the design parametersshould be checked against known successful installa-tions.

The number and size of clarifiers to be provided willdepend on considerations of overall cost, operationalfactors and the possibility that under initial underloadedconditions there will be increased retention time in thetanks. When the clarifiers are sized allowance must bemade for any flow variations which may occur includingpossible surges in flow resulting from rapid lowering ofthe adjustable weirs used to automatically controldissolved oxygen input to the reactor. Because of theprevalence of filamentous scums it is essential that ade-quate means of descumming is provided.

10. BIOLOGICAL REACTOR

The biological reactor consists of a number of different2ones. Each has a specific function to perform and thedesign of each zone must provide the optimum condi-tions for the performance of these functions.

10-6

10.1 Anaerobic zone

The retention time of the influent sewage in this zoneplus the development of authentic anaerobic conditionsare of extreme importance. Conditions which are likelyto affect either, such as return flows with high dissolvedoxygen or nitrate concentrations, must be avoided. Toensure an adequate retention time and no dilution of thereadily biodegrable COD concentration in this zone,provision should be made to divert peak storm flowspast this zone directly into the aeration zone.

Enclosure of this zone to exclude atmosphericoxygen is not necessary provided excessive surfaceturbulence as a result of mixing does not occur. To limitTurbulence, the energy imparted to the liquid by Themixers should be kepi to a minimum. It has been foundthat an energy input of about 4 W.'m3 provides ade-quate mixing without causing excessive turbulence. Ac-tual power requirements for mixing will depend on theshape and size of the reactor and the type of mixer in-stalled. Mixer manufacturers should be consulted regar-ding the most suitable configuration. Where feasible,the use of the kinetic energy in the influent flow or recy-cle streams to achieve hydraulic mixing in both theanoxic and/or anaerobic zone should be considered.

The concentration of readily biodegradable COD inthe anaerobic zone should be as high as possible to pro-mote biological phosphorus removal. This can best beachieved in a plug flow reactor or in a series of smallcompletely mixed reactors rather than in a single largecompletely mixed reactor. Baffles between the zonesshould be designed to prevent backmixing but need notbe completely watertight.

10.2 Anoxic zone

The introduction of extraneous dissolved oxygen intothese zones must be prevented. The size of the primaryanoxic zone must be based on denitrification ratesdetermined by experiment using the sewage to betreated. The second anoxic zone is not always needed,it is most effective when treating high strength sewageparticularly when the sewage has a high biodegradablesolids content. Under these conditions paniculateorganic matter is absorbed onto the sludge and is car-ried through to the second anoxic zone on the sludge.Metabolism of ihe absorbed paniculate COD in the se-cond anoxic zone enables higher denitrification rates tobe achieved.

With low strength sewages, for example settledsewages, and with sewages containing a large propor-tion of soluble and colloidal biodegradable matter whichare readily removed in the primary anoxic 2one, there islikely to be a deficiency of organic matter in the secondanoxic zone and only minimal quantities of nitrate maybe removed.

10.3 Aerobic zone

The main aerobic zone must be sized for completenitrification at the minimum sewage temperature, at theselected residual dissolved oxygen concentration andsludge age. The choice of aeration system and the

measurement and control of dissolved oxygen levelshas been discussed in Section 4.2.

The final aerobic zone is required:(1) to boost the dissolved oxygen concentration in the

mixed liquor before it enters the clarifiers; and(2) to polish the final effluent by removal of additional

phosphorus and the oxidization of residual am-monia.Excess aeration in this zone must be avoided as it

will result in the conversion of organically boundnitrogen to nitratephosphorus.

and the slow aerobic release of

11. ECONOMICS

The main objective of wastewater treatment is to pro-duce an effluent meeting the required standard at thelowest possible cost. To accomplish this the trade-offsbetween the cost components of the various waste-water treatment processes must be determined in orderto establish the optimum scheme. For example, whencomparing the cost of biological filtration with chemicalphosphorus removal and the cost of biologicalphosphorus removal in the activated sludge process,the major trade-off is cost of chemicals as opposed tothe cost of electricity. Thus for an existing biologicalfilter plant where little additional capital expenditureneed be incurred, the most cost effective solution isprobably chemical precipitation of phosphorus. Thealternative is some modification of the activated sludgeprocess which will incur additional capital expenditureand will involve the use of electricity.

11.1 Typical treatment costs for a biologicalnutrient removal plant

The treatment costs of a first generation (Phoredox)150 M//d biological nutrient removal plant are givenbelow. The major cost components are:(ij Interest on and redemption of capital,(ii) The cost of electrical energy.

Treatment Costs - 150 M i / d 'Phoredox' Plant:Year Ending June 1981

Flow (m3 per annum)

Costs (cents/m3)

Actual Flow Design Flow

31 587 x 103 54 750 x TO3

OperatingMaintenanceGeneral overheadsElectricityLaboratory ServicesInterest and Redemption

Total Costs

0,430,380,770,770,584,67

7,60

0,250,220,441,710,342,69

5.65

10-7

NOTES:

1. Cost of electricity taken at 2,03 cents per kWh, in-clusive of service charge, demand charge andenergy charge in accordance with JohannesburgElectricity Department tariffs.

2. In addition to the abovementioned costs, the costof treatment and disposal of waste activated andprimary sludge must be added.

3. Based on the latest models of the biologicalnutrient removal process, for process configurationwhich optimizes the operating costs by minimizingelectricity consumption, the electricity cost will beof the order of 0,5 cents/mJ.

11.2 Selection of process configuration

if energy usage is to be minimized, a process configura-tion should be selected which can operate at a highTKN:COD ratio and will minimize load to be oxidized. Inaddition, a smaller biological reactor, smaller finalciarifiers, reduced oxygenation capacity with resultingenergy saving will be required. These savings will be off-set to some extent by the additional cost of primarysedimentation and the cost of the treatment anddisposal of raw sludge.

Provided that the TKN:COD ratio after primarysedimentation is such that the process remains viable,the benefits of incorporating primary sedimentation intothe process scheme are:

(i) lower loadings on the biological reactor;(ii) reduced energy consumption;(iii) smaller civil and correspondingly reduced land re-

quirements;(iv) smaller mass of waste sludge to be treated.

The disadvantages are:

(i) the need for anaerobic digestion to treat primarysludge;

(ii) the additional costs associated with the primarysedimentation stage in the process scheme, in-cluding the cost of anaerobic digestion.

11.3 Effect of equalization {flow balancing) on thesizing of the units and associated costimplications.

The most notable benefits of flow equalization are:

(i) Reduced clarifier size as a result of reduced peakflow rates downstream of the equalization tank.

(ii) A levelling of aeration power demand resulting inreduced installed aeration capacity and lower ex-penditure on energy.

(iii) The biological treatment process benefits fromreduced fluctuation in loading as well as morestable retention periods and biomass concentra-tions in the aeration basin.

(iv) Reduced reactor volume.(v) Simplified manua! and automated control of flow

dependent operations such as chiorination andfiltration.

(vi) Savings realised on both capital and operating costresulting from smaller ciarifiers and aeration basin,fewer or smaller aerators and smaller pipes andchannels.

In the following analysis an attempt is made toevaluate the cost savings resulting from smallerciarifiers, smaller aeration basins and fewer or smalleraerators for an existing works with a nominal treatmentcapacity of 150 M i /d .

The additional cost of the balancing tank must beoffset against ihese savings. In the analysis it is assum-ed that the additional ground required for the balancingtank is offset by the additional land which would be re-quired for secondary ciarifiers if flow equalization is notpractised.

The parameters used in the original design of theworks have been used to estimate the requirementswithout flow equalization.

7 7.3.7 Aeration

The original design was based on:Maximum 2 hour load without balancing 4 286 kg BODMaximum 2 hour load with balancing 3 000 kg BODMinimum 2 hour load without balancing 570 kg BODMinimum 2 hour load with balancing 900 kg BODMaximum; Minimum 2 hour load ratio

without balancing 7,5with balancing 3,3

2 hour peak BOD: Average daily BODwithout balancing 2,3with balancing 1,6

The total installed oxygenation capacity (mainaerators only) with flow equalization had beencalculated to be 3 244 kg O2 based on an oxygen inputper aerator of 90,1 kg/h (see Annexure 1).

The number of additional aerators required withoutflow equalisation:

4 296 - 3 24490,1

= 11,58 (say 12 aerators).

or four additional aerators per module, which could beaccommodated adequately in the existing aerationvolume with little increase in energy density.

The cost of the additional aerators (excluding pro-vision of civil support structures) is estimated atR413 000.

11.3.2 Biological reactor

In a conventional plant the vulume of the aeration basinis designed for the average flow/load and a safety fac-tor is then applied to account for variations in flow andload when there is no flow equalisation.

Factors ranging from 1,3 - 2,5 have been proposedto account for influent variability where treatmentcapacity is estimated using a steady state model. Fur-thermore, reactions in the anaerobic and anoxic zonesare rate-dependent and a minimum detention time mustbe provided- Where no flow equalization is providedthese zones should be designed to provide the minimumdetention time at peak flow.

10-8

Under conditions of no equalisation the volume ofthese zones would have to be increased to maintain theexisting detention times.

ZONEINCREASED ADDITIONAL

VOLUME COST(per module) m3 Rand

Anaerobic ZonePrimary Anoxic ZoneMain Aeration Zone

9152 1104 410

41 00093 000

195 000

329 000

In this analysis the secondary anoxic zone has beenomitted as experience has shown That it is ineffectiveand should possible be excluded in future designs.

/1.3.3 Secondary sedimentation

Based on the original design parameters six ciarifierswould be required for each module for non-equalizedflow instead of the four ciarifiers provided with equalisa-tion (see Annexure 3).

The cost of these additional ciarifiers wouldamount to R299 000 for mechanical equipment andR263 000 for civil structures, a total of R492 000. Thecost could be reduced to some extent by using a smallernumber of larger ciarifiers but it is felt that the size of theclarifier in use is approaching the limit for the design.

17.3.4 Pipework, channels, etc.

Flow conveying elements of the treatment works suchas pipes, channels etc. must be designed to convey thebalanced flow. Sizing of these elements is very sitespecific; greater costs are likely to be incurred on flatsites than on sites with adequate falls. No attempttherefore has been made to quantify the expected costdifferences.

11.3.5 Downstream processes

Other downstream processes also benefit from flowequalisation since these processes can be sized to treatthe average flow rather than the expected peak flow.Chlorination equipment, chlorine control tanks andfiltration units will be smaller in size. Because of theconstant flow, these processes will also be easier tooperate.

11.3.6 Cost implications

Details of the estimated increase in capital and electrici-ty cost where no equalisation is provided are given inAnnexures 2 and 4.

The estimated increase in capital cost for a plantwith a nominal treatment capacity of 150 Mjf/d if noprovisions is made for flow balancing is approximatelyR1,0 million.

The estimated annual savings resulting from the in-corporation of flow balancing, based on savings on in-terest and redemption of capital and electricity costs, isR180 000 per annum or a saving of 0,33 c/nV of sewagetreated.

Both primary sedimentation and flow balancingcan adversely affect the TKN:C0D ratio of sewage. Ifcertain minimum TKN:COD ratios are necessary for theoperation of a particular biological nutrient removal pro-cess, the use of primary sedimentation and possibleflow balancing might be counter-productive and lead tounstable performance.

The implications of the TKN:COD ratio in processdesign and selection of a reactor configuration are dealtwith fully in Chapter 7.

11.4 Effect of load balancing

A similar exercise to the above can be carried out whenit is intended to incorporate load instead of flow balanc-ing. Assuming that the total influent load eventually allpasses to the biological reactor, the advantages of loadbalancing are:

(i) A more stable process in that when periods ofadverse TKN:C0D ratios occur in the sewage thesecan be balanced out.

(ii) A reduction in maximum/minimum power demandcommensurate with the degree of load balancingapplied to the reactor. The installed aerationcapacity can be reduced because of the reductionin peak loads. A corresponding reduction in peakdemand electricity charges also results. As it isassumed that the total load is not reduced, the totalaeration demand and hence energy consumptionwill remain unchanged.

(iii) if all the primary sludge produced in the ioad balan-cing stage is returned to the process stream, no ad-ditional costs will be incurred in the treatment anddisposal of primary sludge, but an increased massof waste activated sludge will have to be treatedand disposed of.

The disadvantages are:

(i) As peak flows will not be attenuated by the loadbalancing operation, the final ciarifiers, biologicalreactor, pipework and channels must be designedto cater for these peak flows.

(ii) Additional capital expenditure will be incurred forthe provision of a sludge storage tank and primarysedimentation tanks; although the latter could besmaller and less efficient than the normal design.

11.5 Full and partial nitrification and denitrification

Depending on the effluent standards with respect toammonia and nitrates, the plant can be designed and; oroperated to achieve full and 'or partial nitrification anddenitrification.

Full nitrification without full denitrification is lessefficient in respect of energy usage as the energy storedas bound oxygen in the nitrates is only partiallyrecovered. In a process configuration suitable forsewage with a high TKN:COD ratio, less energy is con-sumed by the oxidation of the carbonaceous substratebut there would be a lower energy recovery as a resultof only partial denitrification and higher nitrate concen-trations would be discharged in the effluent.

10-9

Low dissolved oxygen concentrations in thebiological reactor can achieve reduced energy usage bysuppressing nitrification. This procedure can howeverresult in operational problems as a poorly settlingfilamentous sludge may be produced.

12. STAFFING

There is no doubt that the operation of a biologicalnutrient removal piant requires greater understanding ofthe theory behind the process than is normally requiredfrom sewage treatment plant operators in South Africa.This is particularly true when the conditions forbiologicsl phosphorus removal are marginal. The effecton the process of variability in the influent maynecessitate regular process adjustments and may re-quire the full-time attendance of a competent cer-tificated operator who is fully conversant with thebiological nutrient removal activated sludge process.

This has the following implications:

i\) Higher salaries will be necessary to attract a betterquality, properly qualified operator who must becapable of understanding the operational require-ments and making the correct process adjustmentswhen necessary.

(ii) Provision must be made for adequate staff trainingto ensure that the operator is competent andcapable of operating the plant.

(iii) Full-time attendance of a trained operator is re-quired. It is most unlikely that such a plant will pro-duce results consistently within standard if it is pro-vided with only limited operator attendance.The biological nutrient removal process places

greater reliance on instrumentation for success thanother biological processes. The operator should becapable of minor routine maintenance on the in-struments such as checking zero adjustment andrenewing DO probe membranes. There is an extremeshortage of qualified instrument technicians in SouthAfrica, and provision should be made for maintenancecontracts for instrumentation.

ANNEXURE1

CALCULATED AERATOR PERFORMANCE

Size of electric motorsNumber of electric motors

Full load efficiency of motorsEfficiency of gearboxesGuaranteed oxygen transfer efficiency

= 110 kW= 12 per module= 36 total= 94%= 93%= 1,81 kgO2/kWh

Oxygenation capacity of the aerators must be adjusted to allow for:

(a) Altitude (1 500 m above sea level).(b) Sewage a? opposed io clean water {a = 0,85).(c) Temperature (13°C).

Oxygenation capacity = 1,81 .0,94.0,93 = 0,98kg/kWh

Power at aerator drive output shaft at maximum immersion = 92 kW

Oxygen input per aerator = 92.0,98 = 90,10 kg/h

TOTAL OXYGEN INPUT 13 MODULES = 36 AERATORS) = 3 244 kgO2/h

10-10

ANNEXURE2

CALCULATION OF ELECTRICITY COSTS

Aeration is the major consumer of energy and thus has a predominating effect on the costs of power. (Mixing,pumping and secondary clarifiers constitute only a small proportion of the total power demand and can beneglected in a comparison of power costs.

The analysis is based on the following electricity tariff:

Service charge 22c per dayDemand charge 11,66c per kVA per dayConsumption charge 1,39c per kWh

The power required depends on the oxygen demand of the incoming load (COD and ammonia). The averagedaily load to be treated is the same whether flow equalization is provided or not. Only the peak load will vary.This must be catered for by the provision of additional aerators where no flow equalization is provided. This willresult in higher peak electricity demands and a higher demand charge only.

With flow equalization Without flow equalization

Number of aerators per module 12 Number of aerators per module 16Size of aerator motors IkW) 110 Size of aerator motors (kW) 110Power factor (with PF correction) 0,96 Power factor (with PF correction) 0,96kVA per aerator 114,6 kVA per aerator 114,6Maximum kVA per module 1 375 Maximum kVA per module 1 834Demand charge per module per day R159,52 Demand charge per module per day R212,70Demand charges per annum R174 678,00 Demand charges per annum R232 904,00

ADDITIONAL COST OF ELECTRICITY WITHOUT FLOW EQUALIZATION: R58 226,00 per annum.

ANNEXURE 3

SECONDARY SEDIMENTATION REQUIREMENTS WITHOUT FLOW EQUALIZATION

Peak (2 hour) dry weather flow without balancing = 8 967 m3/hper module - 2 989m2/h

Solids loading (flux): In winter = 2,5kg/m2hIn summer = 3,5 kg/'mzh

Mi>.ed liquor suspended solids: In winter - 5,1 kg/m3

In summer = 3,5 kg/m3

Winter Conditions

Solids load on clarifiers = 2 989 m3, h • 5,1 kg/m3 = 15224 kg/h

15 224Area of clarifiers required ~ —=--?— — 6 097 mz

2,b

Summer Conditions

Peak wet weather flow without balancing - 13 078 mJ;h

per module = 4 359 m2/ h

Solids loads on clsrifiers - 4 359 A 3,5 = 15 258 kg/h

Area of clarifiers required = --a £-• = 4 359 m^

Winter conditions are critical and 6 - 36,2 m diameter clarifiers would be required for each module.

10-11

ANNEXURE 4

COST IMPLICATIONS OF FLOW EQUALIZATION

WITH

UNITPrimary sedimentation tanksBalancing tankAeraiion basinsSecondary clarifiers

MECHANICALSPrimary sedimentation tanksBalancing tankAerators onlySecondary clarifiers

"Actual final cost(1976)* * Primary sedimentation

capacity.

rounded off to nearest R1 000.retained as exclusion would require

COST (R)

EQUALIZATION* WITHOUT EQUALIZATION

221 000236 000

1 720 000525 000

152 000

1 240 000458 000

R4 552 000

221 000**

2 049 000788 000

152 000

1 653 000687 000

R5 550 000 .

larger aeration basin volume and greater aeration

Based on a redemption period of 15 years at 10% p.a. interest for mechanical equipment and 30 years at 10%p.a. for civil construction work, the annual cost of the alternatives is:

SCHEME

With equalizationCivilsMechanicals

Without equalization:CiviisMechanicals

ADDITIONAL ANNUAL

CAPITAL (Rl

2 702 0001 850 000

TOTAL R4 552 000

3 058 0002 492 000

TOTAL R5 550 000

INTEREST ANDREDEMPTION

(R p.a.]

287 000 .243 000

R530 000

324 000328 000

R652 0O0

COST OF SCHEME WITHOUT EQUALIZATION AMOUNTS

ADDITIONAL COSTOF ELECTRICITY

(R p.a.)

—

58 000

R58 000

TO R180 000.

10-12

CHAPTER ELEVEN

OPERATION OF BIOLOGICAL NUTRIENT REMOVAL PLANTS

by

A.R. Pitman

1. BACKGROUND

This chapter will highlight operating experiences on full-scale biological nutrient removal activated sludgeplants. Most of the discussion will reflect informationgained on the five-stage Phoredox process layout butshould easily be adapted to other processes such as thethree stage Phoredox or UCT systems. The basic prin-ciples involved should be the same for all biologicalnutrient removal plants.

Even when research provides sufficient under-standing of phosphorus removal mechanisms to permitdesigners to provide more optimum plants, the suc-cessful operation of any wastewater treatment plant wiltdepend on the quality of its operators. In addition, theextra complexity of nutrient removal processes requiresthat they be controlled by a better calibre of operatorthan one who operates a trickling filter plant. In thisrespect, suitably trained and experienced operators aremost important. This is not a peculiarity of biologicalplants as good quality operation would also be requiredon chemical nutrient removal plants.

Suitable laboratory facilities manned by properlytrained, experienced staff would also be required onnutrient removal plants in order to ensure that regularanalyses are carried out. Owner authorities must there-fore ensure that adequate staff will be available beforeembarking on the planning and design of these plants.

2. INFLUENT SEWAGE

2.1 Characteristics

In general, biological nutrient removal is more suited tostrong sewages than weak ones. Experience has shownthat for successful operation the feed COD should be atleast 300 mg f and preferably above 400 mg/f. Also,the TKN/COD ratio of the feed is important; this shouldpreferably be below 0,10. The sewage should also havea relatively high readily biodegradable COD content.The above requirements will ensure that denitrificationis virtually complete and that phosphate release in theanaerobic zone will readily occur.

Where readily biodegradable COD is available as awaste product from industry, its discharge to the sewershould be encouraged. Primary sedimentation tankscan have a marked detrimental effect on the desiredsewage characteristics and thus lead to inferiorphosphorus removal. The COD is reduced in primarysedimentation tanks and the TKN/COD ratio can in-crease adversely as COD is usually removed to a greater

extent than TKN. The reduction in COD load and con-centration over weekends, holiday periods and afterheavy rainfalls, can also be a problem. This can causephosphorus removal to deteriorate for a period and onlyrecover once the normal load has been restored.

The combination of primary sedimentation and lowstrength weekend flows could result in the completeloss of phosphorus removal with phosphate peaks oc-curring in the effluent.

As with many other continupus processes,biological nutrient removal can benefit from a constantquality of feed. In this respect the use of flow balancingtanks can be an advantage. They must however, becontrolled to ensure that effluent flowrates are constantand that solids deposition on the balancing tank floorsare minimised.

2.2 Storm flows

During periods of rain, sewage flows tend to be ofgreater magnitude and dilute strength due to the infiltra-tion of rainwater into the sewer system. These factorscan produce some unique problems to the biologicalnutrient removal process.

The weaker sewage could result in inadequateCOD being available to:

(i) achieve the usual degree of denitrification; and

(ii) ensure that the anaerobic zone is properlyanaerobic. Also, if adequate DO control is not prac-tised, DO levels will rise as the respiration ratedrops and excessive nitrification of organicnitrogen can occur.

The higher flow would result in the retention timesin the zones being reduced to the extent that the desiredaction does not take place in them. This is particularlyso in the case of the anaerobic zone.

For the latter reason the possibility of bypassingpeak storm flows should be considered. These couldflow to holding tanks to be pumped back during lowflows or go to oxidation ponds for stabilisation. If noneof these facilities are available, peak flows could bebypassed to the final ciarifiers or preferably to the end ofthe biological reactors. This will increase the upwardvelocity in the ciarifiers but not increase the solid fluxload to the same extent because of the dilution effect onthe MLSS. The advantage of flowing through the lastzone or the end of the reactor is that some contact withbiomass is achieved, which will absorb COD and notresult in a severe deterioration in effluent quality.

11-1

In the case of weaker sewages the dilution of thestrength caused by storm water can be a serious pro-blem leading to ;he failure of phosphorus removal.

2.3 Improving feed characteristics

If the quality of the sewage fed to the process is inade-quate, various options can be tried to improve the situa-tion. Possibilities are:

• Overloading of the primary clarifiers to increasesolids carry-over and hence the feed COD.

• Continuously feeding back of primary sludge intofeed to biological reactors.

• The addition of volatile fatty acids produced by highrate anaerobic digestion of primary sludge. Thiswould be the ideal substrate for Acinetobacter andwould also be a readily biodegradable substrate pro-ducing rapid denitrification. These acids wouid alsoincrease the degree of release in the anaerobic zoneand produce a much more (biologically} active mixedliquor. Either an acid supernatant liquor can be addedor the whole acid sludge containing sludge solids canbe used. The tatter will however, increase the MLSSand power demands far more than the former. Theacid material should be bled back into the sewagefeed to the anaerobic zone.

3. BASIC CONCEPTS IN BIOLOGICALPHOSPHORUS REMOVAL

In the conventional activated sludge process which isnot designed or operated to achieve biological excessphosphorus removal, the bacteria only use phosphorusin quantities which satisfy their basic metabolic re-quirements. Because of the usual nutrient imbalance insewage only a limited quantity of the feed phosphoruswill be removed in such plants. For example, from afeed sewage with 10 mg P// total phosphorus, an ef-fluent containing 7 - 8 m g / i total phosphorus will beproduced. The waste activated sfudge from such a pro-cess will contain relatively little phosphorus in a typicalcase, 2 % on a dry solids mass basis.

Bacteria which accumulate excess quantities ofphosphorus will be absent or only present in lownumbers in such plants and the conditions which induceexcess phosphorus uptake by the biomass will notoccur.

In plants designed for biological phosphorusremoval however, the environment is created for theproliferation of a biomass which will accumulatephosphorus in excess of normal metabolic require-ments. Certain bacteria growing in such plants can ac-cumulate and store within their cell walls large quan-tities of phosphate as a relatively insoluble storage pro-duct. Furthermore, the possibility exists that certainchemical precipitates containing phosphorus can be ad-sorbed to the surfaces of organic matter in activatedsludge. Such plants can therefore produce an effluentcontaining less than 1 mg/1 phosphorus from a feedsewage containing 10 mg P/l. The waste activated

sludge will contain enhanced quantities of phosphorus,typically 4 - 5 % (as P) in the case of most local plants.

The sludge from a plant that is removing phos-phorus well, will have the following properties:

3.1 Ability to release phosphate

Under anaerobic conditions, that is, in any environmentfree of nitrate and dissolved oxygen an activated sludgethat has accumulated excess phosphorus will tend torelease phosphate from storage areas out into surround-ing liquid. The rate of release will depend on the activityof the sludge, but over a few weeks of anaerobiosis alarge proportion of the accumulated phosphate will bereleased. However, in the short retention period in theanaerobic zone of a plant only 10-20 % of the totalphosphorus in the sludge will be released. Whennecessary, anaerobic release can be prevented by keep-ing the sludge under aeration or by maintaining a certainconcentration of nitrate in the sludge.

Anaerobic release of phosphate can be stimulatedby the presence of biodegradable COD- When aphosphate rich activated sludge is fed with readilybiodegradable soluble COD, the rate of release underanaerobic conditions is speeded up. In fact, volatile fat-ty acids such as acetic acid can produce very rapidphosphate releases.

Anaerobic release is an important prerequisite forbiological phosphorus removal. For a plant to removephosphate well, release must always occur, either in theanaerobic zone or elsewhere in the process. Phosphaterelease indicates that the sludge has been conditionedto remove further phosphorus. Anaerobic release condi-tions are usually achieved by passing activated sludgecontinually through a zone which is free of nitrates andmolecular oxygen so that at any one time, a certain frac-tion of the total activated sludge inventory is being sub-jected to the anaerobic environment. In certain plants,on/off operation can be used whereby all the sludge inthe process is anaerobically conditioned at once, usuallyby switching off all the aerators for a period of time.This 'off' period is repeated daily or after a number ofdays and the plant is operated in its normal aerobic statefor the rest of the period. If release in the anaerobiczone is lost, for example, due to the presence ofnitrates, phosphate removal will fail. In most plants thisusuafly manifests itself as an increase in effluentphosphate concentration a day or two after release waslost, i.e. there is a time delay between the loss of releaseand the failure of phosphorus removal. Where all thesludge has been anaerobicatly conditioned on a batch oron/off basis, the conditioned sludge will continue totake up phosphate under aerobic conditions for a num-ber of days or even weeks before the effect of the con-ditioning is lost. Sufficient aeration capacity should beavailable to ensure rapid uptake of released phosphate.If this is not on hand, temporary chemical precipitationshould be considered. Plants in Johannesburg, Pine-town and Brits have been successfully operated in thismode without addition of chemicals.

A phosphate rich sludge can also release solublephosphate under aerobic conditions. This is a far slowerprocess than anaerobic release and takes place whenthe sludge is aerated at relatively high DO levels in the

11-2

absence of substrate, i.e. endogenous conditions. Ex-amples of where such release can occur are, at the endof a plug flow aeration basin and during aerobic sludgedigestion. However, it must be emphasized that thistype of release is not the same as the anaerobic releaserequired to induce phosphorus removal.

Phosphate release under aerobic conditions willalso occur if the pH of the sludge is decreased (say to5 - 6> by the additional acids or by excess nitrification ina poorly buffered water. The latter can typically occurduring aerobic digestion of the sludge.

Although anaerobic release must occur beforebiological phosphorus removal is possible, release ofphosphate can cause problems in certain cir-cumstances. For example, excessive release underanaerobic conditions can result in inadequatephosphate uptake in the subsequent aerobic zone.Also, if a phosphate rich sludge is anaerobicallydigested, phosphate release wilt occur and liquors andfiltrates from the dewatering of such a sludge will have ahigh phosphate content. Furthermore, when waste ac-tivated sludge is dewatered on drying beds, The bedunderflow can have an elevated phosphate content.The recycle of these liquors to the nutrient removal ac-tivated sludge reactors can impair phosphorus removal.If This is a problem artificial precipitation of phosphatefrom these liquors by chemical addition should be con-sidered.

3.2 Phosphorus uptake ability

Under aerobic conditions the sludge will removephosphate from the liquid phase and accumulate it inquantities which are in excess of normal metabolic re-quirements. However, the sludge will only do this if ithas been previously subjected to anaerobic release con-ditions, i.e., the sludge will only take up phosphate if ithas previously released phosphate. During aerationsuch a conditioned sludge will accumulate phosphatefairly rapidly, often leaving only trace quantities in solu-tion. Sludges which do not perform this way have eithernot been anaerobically conditioned properly or the rateof aeration is inadequate. In particular, sludges whichhave released excessive quantities of phosphate mightneed a longer aeration time to reduce solutionphosphate concentrations to low levels.

4. OPERATING CONDITIONS IN VARIOUSZONES AND CLARIFIERS

This section will describe typical operating conditions inthe zones of a five-zone plant and point out the condi-tions which have been found to be necessary for effi-cient nutrient removal.

4.1 Anaerobic zone

This zone is essential for phosphorus removal as thebacteria in the activated sludge passing through it arepreconditioned to take in excess phosphate. The releaseof a certain quantity of phosphate from the biomass intosolution indicates that the bacteria have been suitablyconditioned. To achieve this the feed sewage and ac-

tivated sludge are retained in the absence of nitrate anddissolved oxygen, for a period of anything between 30minutes and 4 hours. The actual retention time isdependent on the volume of this reactor and the flowrates of the feed sewage and recycled activated sludge.The strength of the sewage can also affect the retentiontime needed for successful preconditioning. For exam-ple, a high COD feed, particularly one containing a largereadily biodegiadabie fraction will cause a rapidphosphate release and hence conditioning of thesludge, while a low COD feed will produce a far slowerrate of release with a concomitantly lower excess Premoval.

Bacteria in the mixed liquor are stressed or condi-tioned in this zone to the extent that they wilt take upexcess phosphate under aeration in the aerobic zones.The level of phosphate release in the anaerobic zone is ameasure of the degree of anaerobic stress, i.e., thegreater the release the greater the stress achieved.

Conditions in the anaerobic zone should be suchthat phosphate is released into solution to levels abovethe feed sewage phosphate concentration. For exam-ple, for good phosphorus removal to occur in a planttreating a sewage containing 8 mg/ / total P with aMLSS concentration of 3 000 mg/ i , the ortho-phosphate concentration in solution in the anaerobiczone should average about 20 mg P/£ and never dropbelow 10 mg/ i . Although release is necessary forphosphorus removal, excessive release should be avoid-ed as it might prove difficult to take up all of the releas-ed phosphate in the aeration time available in theaerobic zone. In the example above, a release of30 mg PI i may be counter-productive as 20 mg P/imight be perfectly adequate.

Nitrate and dissolved oxygen discharged to thiszone should be zero or as near as possible to zero at alltimes. If nitrates or DO are detected anaerobic condi-tions will not be possible, phosphate release will not oc-cur and biological excess phosphorus removal will notoperate, that is, even a nitrate concentration of 0,5 mgN/X will block phosphate release and hence uptake. Thesituation might even occur where nitrates and DO arenot detected in the anaerobic zone and adequatephosphate release will still not occur in this case andnitrate feedback via the return sludge might just be suf-ficient to prevent anaerobic conditions occurring butthe amount returned would be denitrified to the extentthat no measurable quantities of nitrate were present.The loss of release in the anaerobic zone indicates thatphosphate removal is about to fail. This usually occurseither the following day or the day thereafter, for exam-pie, low phosphorus release patterns over weekendscan result in effluent phosphate concentrations increas-ing on Tuesdays.

4. 7.7 Nitrate feedback

In plants which have an anaerobic zone at the inlet end,the problem of nitrate feedback in this zone via thereturn sludge cannot be over-emphasized. When nitrateconcentrations in the return sludge are high, an ex-cessive denitrification load can be placed on this zoneand the anaerobic conditions required for phosphateremoval are lost. As soon as nitrates are detected in the

11-3

anaerobic zone, even at low concentrations (0,5-1,0mg N/f) impending failure of phosphorus removal is in-dicated. The ideal condition is that the return sludgeshould contain no nitrates but this is not always possiblein practice. The degree of nitrate feedback that can betolerated depends on the strength of the sewage feed tothe anaerobic zone and in particular, its readilybiodegradable COD concentration. For a high strengthsewage (COD = 600 mg/f) with a good readily bio-degradable content, a return siudge nitrate level of 7 mgN/f might be tolerable. Alternatively, for a low strengthfeed of COD - 200 m g / i , a return sludge nitrate level ot1,5 mg Kl/jf might be undesirable. However, it shouldbe borne in mind that the greater the recycle of nitrate,the lower the degree of P removal obtained in the pro-cess (see Figure 7.131. If nitrates cannot be detected inthe anaerobic zone and a good phosphate release is oc-curring, then phosphate uptake is possible. If nitratescannot be detected in the anaerobic zone andphosphate release is still at low levels, then the nitratefeedback in the return sludge might be just sufficient toprevent proper release. Although the key to obtaininglow return sludge nitrates is efficient denitrification inthe biological reactor, any additional denitrificationwhich can be achieved in the sludge layer in the finalciarifier can be a great help. Although the design of theclarifier is important in this regard, adaptations inctarifer operation can also help. For example, reducingthe clarifier underflow rate (recycle ratio) will mean thatthe sludge will stay a longer time in the clarifier, thickenup a bit more and thus consume a larger amount ofresidual nitrates. Care should be taken however, inreducing clarifier underflows to not reduce the solidshandling capacity of the clarifier (see Chapter 8) and risksolids carryover in the effluent. Also, if solids are retain-ed too long in the clarifier, all the residual nitrates can beconsumed, the sludge will go anaerobic and aphosphate leach-out into the effluent will occur. In par-ticular, this could take place in a very active sludge hav-ing a relatively low nitrate content or where the sludgesurface level is standing fairly high in the clarifier.

The easiest way of ensuring that the return sludgenitrates are low is to ensure that the effluent from thebiological reactor has a low nitrate concentration. Thisdepends on the denitrification efficiency which dependson many other factors described elsewhere in this docu-ment.

Nitrate feedback to the anaerobic zone can also oc-cur via backmixing from adjacent anoxic zones. For ex-ample, when the baffle separating the anaerobic andanoxic zones has ports in it to allow transmission offlow, the mixers in the anoxic zone can easily backmixmixed liquor containing nitrate into the anaerobic zone.In mufti-mixer plants, the mixer nearest the baffle couldbe turned off to minimise ihrs effect. (This should onlybe done if the remaining mixer/s can still keep the solidsin suspension).

4-1.2 Oxygen input to anaerobic zone

Experience has shown that molecular oxygen can enterthe anaerobic zone from a number of sources and if dif-ficulties are experienced in obtaining suitable phosphate

release, the operator should investigate and try andeliminate these unwanted oxygen inputs.

Excessive turbulence in the streams feeding intothe anaerobic zone and in the zone itself can introduceunnecessary oxygen. For instance, cascades, screwpumps and aerated grit channels can oxygenate thefeed sewage while screw pumps in the sludge recycleline can have the same effect. Excessive mixing in theanaerobic zone could also be detrimental.

Where this occurs and more than one mixer isavailable, consideration should be given to operatingfewer mixers. In addition, by alternating the operatingmixers, better utilisation of mechanical plant will be ob-tained.

Although many of the sources of oxygen input inthe anaerobic zones of existing plants will be due to in-herent design faults which the operator will not be ableto correct easily, knowledge of the extent of this pro-blem can be of great use in plant operation.

4. 1.3 Retention time

The basic actual retention time of the liquid in theanaerobic zone will have been fixed by the designer andthe operator will have little leeway to vary it. In caseswhere suitable anaerobic conditions are difficult toachieve consideration should be given to increasing theretention time in this zone by reducing the sludge recy-cle rate. This will also reduce the mass of nitrate recycl-ed to the anaerobic zone. Before reducing the sludgerecycle however, the operation of the final clarifiersshould be considered. When the underflow recyclefrom the clarifiers is decreased, the solids handlingcapacity of the clarifiers is reduced as well, so that if theunderflow is reduced too much, a solids flux failure willoccur in the clarifiers and activated sludge solids will becarried into the effluent. Thus, there will be a limit tohow much the sludge recycle can be reduced, andoperators should be aware of this limit befoie consider-ing ways of increasing the anaerobic retention time. Ingeneral, plants should be operated with as low anunderflow recycle as possible while maintaining a safetyfactor in the clarifier solids handling capacity for peakload conditions. Large feed fiowrates above the designlevel can also drastically reduce the anaerobic retentiontime. These can occur during storm-flow conditions sothat consideration should also be given to bypassingstorm peaks to the anoxic or aerobic zones.

The anaerobic retention time can also be increasedby making the first anoxic zone anaerobic as well. Thiscan be done by cutting the recycle of mixed liquor tothis anoxic zone. If this is done however, the operatormust be sure that good denitrificatinn can be achievedelsewhere in the plant, that is, by creating anoxic zonesin the aerobic zone. This option is usually only availablewith surface aerators when certain aerators (usually atthe inlet end of the aerobic zone) can be switched off tocreate an anoxic zone. If this cannot be done, i.e., indiffused air plants, and plants with limited aerationcapacity, then the effluent nitrate levels will increasewhen the mixed liquor recycle is cut, nitrate feedback tothe anaerobic zone will increase and any beneficial ef-fect of the bigger anaerobic zone wilf be negated. Also,it should be remembered that aerators should not be

11-4

switched off at the expense of achieving adequatedissolved oxygen levels in aerobic zones. On certainplants treating a high strength sewage having a largereadily biodegradable content, phosphate release in theanaerobic zone could be excessive and the retentiontime in the aerobic zone would then be inadequate forgood phosphate uptake, in this case the retention timein the anaerobic zone can be decreased by increasingthe sludge recycle. The sludge recycle should be in-creased only until the level of phosphate release dropsto the optimum level (say 20 mg P/f in the case of theearlier example.

4.2 Anoxic zone or primary anoxic zone

This zone is the main denitrification reactor in the pro-cess. It is fed by the effluent from the anaerobic zoneand the mixed liquor recycle from the aerobic zone.Nitrates can be present in solution but oxygen inputsshould be restricted.

The size of this zone will have been fixed by thedesigner using denitrification kinetics and the mass ofnitrate recycled to it can be varied by changing the mix-ed liquor recycle rate, (see Chapter 6}.

If desired, the size of this zone could be increasedon surface aerator plants by switching off the first fewaerators in the adjacent aerobic zone. The first sectionof the aerobic zone will be anoxic and the operatingaerators will backmix nitrates into the zone, furtherassisting denitrification. This practice can only be donein surface aerator plants with excess aeration capacity.This advantage afforded by surface aerator plants canbe further exploited in multi-aerator zones where anoxicareas can be created in the aerobic zone by the judiciousswitching off of certain aerators.

The introduction of oxygen into the anoxic zonecan be detrimental so that knowledge of its points of in-gress is important. High mixing intensities can changethe liquid surface rapidly and bring oxygen from the at-mosphere. In plants where this might occur, considera-tion should be given to not operating some of themixers in this zone. Also, oxygen can enter via themixed liquor recycle from the aerobic zone, particularlywhen high intensities of aeration occur near the mixedliquor abstraction point, tn this case, aeration can betapered off near the end of the aerobic zone (provided itis not completely mixed), to give lower dissolved oxy-gen levels at the abstraction point. Oxygen can alsoenter the anoxic zone via backmixing through gaps inthe baffle separating it from the aerobic zone. Surfaceaerators can easily backmix oxygen through holes inthis baffle and a possible remedy is to modify the baffleto create one-way flow only, or to switch off theaerators nearest the baffle.

The mixed liquor recycle rate from the aerobic tothe anoxic reactor is a variable that can be controlled.An upper operating limit for the flowrate would be sixtimes the feed flowrate, i.e. 6:1 and this can be reducedas low zero, as mentioned earlier. A good operatingfigure in most cases would be 4:1. The optimum wouldbe to control the mixed liquor recycle so that nitrateconcentrations are not very high in this anoxic zone.Due to variations in the strength and volume of the feedsewage and the usual lack of on-line nitrate monitoring,

this ideal is not attainable and usually a fixed ratio ofsay, 4:1 is set and is not changed very often. Even withthe mixed liquor pumps not operating, denitrificationcan still occur in certain plants. For example, surfaceaerators on some plants backmixed nitrates to the anox-ic zone through holes in the baffles and in highly loadedplug flow diffused air plants the oxygen demand at theinlet end of the aerobic zone could be so great that theaerators cannot cope, oxygen levels drop to zero and ananoxic zone is created. If the mixed liquor recycle to theanoxic zone is zero and other backmixing does not oc-cur, then this zone becomes a further anaerobic zoneand additional release of phosphate can occur. This canbe beneficial to phosphate removal when the mainanaerobic zone is not functioning properly. When goodrelease of phosphate occurs in the anaerobic zonehowever, a further release in the anoxic zone could bedetrimental.

In the cati. j f a plant treating a high strength feedwith a low TKINI/COD ratio and having a high readilybiodegradable fraction, a good release will occur in theanaerobic zone and nitrates will be readily consumed inthe anoxic zone causing it to go anaerobic despite amixed liquor recycle ratio of say 4:1 being used. A fur-ther phosphate release could then occur in the anoxiczone. The degree of release could then be so high thatthe phosphate will not be taken up by the biomass againin the time allowed in ihe aerobic zone. Thus, despitethere being a good phosphate release, phosphorusremoval will be bad due to poor uptake in the aerobiczone. The remedy for this would be to reduce the reten-tion time in the anoxic zone by increasing the mixedliquor recycle to 6:1 or even 8:1 provided pumpingcapacity for this is available.

At the opposite end of the scale, plants which treata low COD waste with a small readily biodegradablefraction, will have relatively high nitrate concentrationsin the primary anoxic zone (e.g. 5 mg N/i) when themixed liquor recycle is 4:1. In this case the mixed liquorrecycle rate is excessive and could be reduced to pre-vent overloading the zone with nitrates. If however,nitrates cannot be reduced by lower recycles then in-adequate COD is available in the feed sewage of dissolv-ed oxygen levels in the returned mixed liquor are toohigh.

In the typical nutrient removal plant phosphate andammonia concentrations in mixed liquor solution willdrop between the anaerobic and anoxic zones. This isdue to the dilution effect of the recycled mixed liquor. Ifhowever, they decrease to a greater extent than can beaccounted for by dilution, particularly in the case of am-monia, then the oxygen input to this zone is too great,nitrification is occurring in the anoxic zone and the pro-cess is probably heading for a failure. Ways should thenbe found to restrict this oxygen input as describedearlier.

4.3 Aerobic zone

The function of this zone is to oxidise organic matter inthe sewage, to oxidize ammonia to nitrate and provideand environment in which the biomass can take up allthe phosphate released in the anaerobic zone, plus allthe phosphate which enters the process in the feedsewage. The reactor size and dissolved oxygen levels

11-5

should be such that all these processes can operate ef-fectively.

Although the size of this zone will have been fixedin the design so that nitrification will occur in the pro-cess, the DO levels in this zone and the aerobic sludgeage cr overall sludge age should be such that nitrifica-tion will occur at all operating temperatures. This shouldbe borne in mind particularly when aerators are switch-ed off or aeration is trimmed to create anoxic zoneswithin the aerobic zone.

As mentioned earlier, the primary anoxic zone sizecan be increased by switching off aerators at the inlet ofthe aerobic zone, but aerators can also be cut at the endof this zone to increase the size of the second anoxiczone and/or to reduce DO levels in the mixed liquorrecycle to the primary anoxic zone. However, it shouldbe remembered that increasing the sizo of the first anox-ic zone will be far more profitable than increasing thesize of the second as the denitrificatinn rp'e in the latteris very stow. (See 4.4). if possible therefore, theoperating aerobic sludge age should be 1,5 times theminimum required by nitrification kinetics at theoperating temperature. For example, if the minimum re-quired in winter is 8 days then an aerobic studge age of12 days should be used. This would probably give inmany plants a total sludge age of about 20-25 days.

In most aerobic zones where completely mixedconditions do not occur the oxygen demand of the mix-ed liquor is higher at the inlet end. Therefore, if possi-ble, more air should be introduced here than at theoutlet end. That is, if possible, tapered aeration shouldbe used. However, the latter is not easily achieved withequally spaced surface aerators. Where it is possible tovary aerator depths, then the aerators near the inletcould be set lower in order to introduce more oxygen.With diffused air plants, diffuser spacing might be suchthat tapered aeration has been buiit into the plant or theoperator can vary air supplies to various points in thezone to give the correct aeration pattern.

Phosphate is taken from solution into storage bybacteria in the aerobic zone. Various factors can affectthe rate of phosphate uptake by the bacteria and ideallyuptake should be complete at the end of the aerobiczone so that only trace quantities 'i.e. < 0,2 mg P/f) aredetected in solution at this point. Aeration rate or aera-tion intensity can affect phosphate uptake so that if DOlevels ar& too low, phosphate uptake will be suppressedand relatively high phosphate concentrations will be insolution at the end of aeration (i.e. 2,0 mg P/jf). There-fore, if effluent phosphate concentrations are highdespite there being a good release of phosphate in theanaerobic zone, lack of aeration in the aerobic zone isoften the cause. (Sometimes however, an increase inoxygen levels can cause a slight increase in effluentphosphate concentration instead of a decrease asdescribed in 3.1 above).

Aeration should therefore be adequate to

(i) promote rapid uptake of released plus feedphosphorus,

lii) ensure oxidation of carbon compounds and am-monia, and

(iii) suppress the growth of filamentous micro-organisms which produce poor settling sludges.

The aeration rate should be well matched to the in-coming load variations and be just adequate to ensurethe above, for if excess aeration is used, then:

(i) Further nitrification can occur when some of theorganic nitrogen bound up with the sludge is con-verted to nitrate. This increases the quantity ofnitrates to be removed and can result in higher ef-fluent nitrate levels, a possible greater nitrate loadon the anaerobic zone and the eventual loss ofphosphorus removal. This is however, not a criticalfactor but emphasizes that good dissolved oxygencontrol is important in these plants.

(ii) The slow aerobic release of absorbed phosphatewill be encouraged resulting in unnecessarily higheffluent phosphate levels. This is much slower thananaerobic release.

(iii! Wastage of energy will occur.

(ivl High DO feedback to anoxic zones will occur thusreducing their efficiency.

(v) Floe break-up can occur leading to pinpoint floe ineffluent.

It is difficult to give a recommended DO concentra-tion which will suffice in all operating conditions. Ex-perience has shown however, that for the above criteriato be met, the average DO in the aerobic zone should beabout 2 - 3 mg/f and should not be allowed to dropmuch below 1 mg/ i at any one point. A DO profile star-ting at say 1 mg/I near inlet and ending at 3 mg/ / ,would be very suitable. DO levels in aerobic zonesshould never be allowed to drop below 0,5 mg/f as thiswill encourage filamentous growths and suppressnitrification. Excessively high DO levels should be avoid-ed, i.e. 4 - 5 mg/f, as these can cause floe break-up topinpoint floe and wastage of power. By trial and erroran optimum DO level for each plant can be found. But,it should be remembered that if filamentous growths areallowed to proliferate when DO levels are dropped, itcan take many months of prolonged aeration to restoregood settling sludge. Often, the monitoring of ammoniaand nitrate levels in filtered mixed liquor samples fromthe end of the aerobic zone can give a good indicationof aeration levels. For example, high ammonia levelsand low nitrate levels could indicate inadequate aerationor vice versa.

On some plants which experience low strengthsewages or variable strength with say, low COD levelsover weekends, inadequate COD might be available forgood denitrification. This would mean that when thefeed COD was low, effluent nitrate would be high andphosphorus removal could be lost. At such times, itmight be advantageous to suppress nitrification byreducing DO levels and allowing some ammonia outwith the effluent. The General Standard would still beadhered to but the effluent ammonia levels would beallowed to increase from say, 1 mg N/f to 7 mg N / i .

11 6

This could however, encourage the growth of filamen-tous organisms which produce very poor settlingsludges and thus be counter-productive in the long-term. For these reasons the suppression of nitrificationshould only be considered as a short-term emergencyprocedure.

Operation under these conditions is difficult as onetends to have an 'all or nothing at all' situation, i.e. if theDO levels are up, nitrification will be complete. If theyare too low, nitrification will tend to stop completelyand effluent ammonia levels will exceed Standard re-quirements. However, it is possible by judicious aera-tion control, to allow some ammonia to leave with theeffluent and thus keep nitrate concentrations low. DOlevels for this operation will usually be below 1 mg/ iand possibly below 0,5 mg/f so that the risk of en-couraging filamentous growths should be kept in mind.

During periods of low load it is also desirable toreduce aeration in order to match it to the load and ob-viate over-aeration and consequent over-nitrification.To do this successfully can create problems, particularlyin the early life of new plants when aeration capacity in-stalled to handie future loads remains unused.

On diffused air* plants, aeration might not be ableto be reduced enough without going below theminimum airflow requirements for diffusers, i.e. theremight not be sufficient turn-down capacity.

In the case of surface aerators, particularly thosehaving small numbers of aerators, so many aeratorsmight have to be turned off that mixing efficiency is lostand solids settle on the floor of the aeration tank and goanaerobic. This will cause a release of phosphate intosolution so that when this sludge is resuspended, thereleased phosphate coutd be washed out with the ef-fluent before it can be taken up by solids again. Theanaerobic conditioning of the sludge on the floor can bebeneficial to the process as it will stimulate the sludge toaccumulate excess phosphate but the problem ofphosphate washout cannot always be obviated.

With both types of aeration care must be taken notto reduce oxygen input to levels where phosphate up-take in the aerobic zone is prejudiced. It should also beborne in mind that filamentous organisms which cancause the activated sludge to settle poorly are en-couraged by large unaerated zones and low dissolvedoxygen levels in the aerobic zone. Operators shouldtherefore take care when reducing aeration, not to pro-duce an environment that will encourage thesenuisance growths.

For good control of oxygen input the number andposition of surface aerators is important as is thedistribution of air in diffused air plants. In surface aera-tion it is better to have a number of smaller aeratorsthan a few large aerators. This gives a greater flexibility.

The placement and maintenance of dissolvedoxygen probes is an important aspect of aeration con-trol. If possible three or four probes should be placed atstrategic points in the aerobic zone. In the case ofsurface aerator plants, the probe should not be placedtoo near the operating aerators or they will read non-

* Another disadvantage of diffused air plants is That one is not able tocieate anoxic zones in the aerobic zone as is possible with surfaceaerator plants.

representative high values. Also they should not beplaced in areas where aerators are to be switched off forlong periods. The variation in dissolved oxygen levelswith tank depth should also be borne in mind and themeasurement of dissolved oxygen profiles with a port-able meter can give a good idea of the best location forthe probes. In fact, a good, reliable, portable dissolvedoxygen meter should always be available for calibratingthe in situ probes and checking dissolved oxygen levelselsewhere in the process. Continuous recording of themeasured dissolved oxygen values can also be veryuseful in plant operation, Dissolved oxygen probesshould also be cleaned out and calibrated weekly andregularly serviced.

Although the control of aeration via the readingfrom dissolved oxygen probes will suffice in most cases,it is possible to control oxygen input by making use ofammonia and ortho-phosphate concentrations in themixed liquor at the end of the aerobic zone. Experiencehas shown that provided an adequate release occurs inthe anaerobic zone, when aeration is decreased in theaerobic zone, phosphate and ammonia levels will be in-creased and when aeration is increased they willdecrease. When both parameters reach approximatelyzero concentration, the optimum aeration level hasbeen reached. Thus, by monitoring trends and changesin ammonia and phosphate levels in the effluent or ifpossible, at the end of the aerobic zone and phosphatesin the anaerobic zone, it is possible to control the aera-tion to optimum levels. Such a system however, will de-pend on the use of on-line analysers and is thus onlysuitable for large plants. Small plant operators couldalso approximate the above system if quick on-sitelaboratory analyses were available.

4.4 Second anoxic zone

Further denitrification takes place in this zone. Ex-perience on many plants has shown that because of thevery slow denitrification rate in this zone, the quantity ofnitrate removed by it is small, i.e. 1-2 mg HI I. There-fore, this zone is only really needed when very low ef-fluent nitrate levels are desired.

In the case of low strength sewages, particularlythose containing small amounts of biodegradable solids(settled sewage), this zone will remove minimalamounts of nitrate. This is because most of the sub-strate will have to be metabolised in the aerobic zone,resulting in a sludge with a very low activity entering thesecond anoxic zone. For high strength sewages con-taining larger amounts of biodegradable solids, thesesolids might not all be consumed in the aerobic zoneand thus create a higher activity and hence denitrifica-tion rate in this zone.

Sometimes, when treating a high strength sewagecontaining a fair proportion of biodegradable solids, ac-tivity in this zone might be so great that denitrification iscomplete; the zone goes anaerobic and phosphate isreleased again. This could be detrimental as thisphosphate may not be picked up quickly enough in thefinal aerobic zone and effluent phosphate levels will beincreased. A method of avoiding phosphate release inthis zone is to allow DO or some nitrate to bleed into itfrom the main aerobic zone. Quantities should be just

11-7

sufficient to keep this zone from going anaerobic andnot in excess, particularly in the case of DO.Sometimes, phosphates will continue to be picked upby the solids in this zone. This will usually take placewhen DO and nitrate enter it in sufficient quantities andcan be useful when uptake tends to be slow in the mainaerobic zone.

Although the feeding in of untreated feed sewageinto this zone would be beneficial in speeding updenitrification, it is not really advisable as it addsphosphate which might not be removed in time, andammonia which will tend to be nitrified in the finalaerobic zone thus increasing effluent and return sludgenitrates. The latter would increase the nitrate load onthe anaerobic zone unnecessarily.

4.5 Final aerobic zone

This zone has a short retention lime and is used to in-crease the dissolved oxygen in the mixed liquor before itenters the final clarifier. Residual ammonia in the mixedliquor will continue to be nitrified in this zone and ifphosphate had not been completely taken up in theprevious zones it will also continue to be taken up in thiszone.

Aeration should be sufficient to promote phos-phate uptake and maintain good aerobic conditions butexcess aeration should be avoided as it will encouragethe conversion of organically bound nitrogen to nitrateand also cause the slow aerobic release of phosphatefrom the solids.

Mixing in this zone should always ensurehomogeneous conditions. In the case of multiple finalclarifiers this will ensure the equal distribution of solidsto the clarifiers and when waste sludge is removed fromthis zone, will ensure that the correct quantity of wastesludge is withdrawn. For example, if such a zone has anumber of surface aerators and those near the wastesludge take off are switched off to prevent over-aerationthen a liquor lower in suspended solids than the mixedliquor could be wasted if settlement of solids occursnear the take-off point.

4.6 Final clarifiers

These should be operated to give the minimumsuspended solids in the final effluent and a thick sludgefor recycle to the inlet of the biological reactors. The useof the 'solids flux' approach will give the best procedurefor optimising the design and operation of theseclarifiers.

Effluents from biological nutrient removal plantsusually contain quantities of nitrate so that the greaterthe suspended solids content in the return sludge, theless the mass of nitrate returned to the anaerobic zone.High return sludge solids concentrations are achievedby lowering sludge recycle rates or ratios, but in lower-ing these, care should be taken not to exceed the solidshandling capacity of the ciarifier. For a fixed feed-sewage flowrate, the lower the sludge recycle rate, thegreater the retention time in the anaerobic zone so thatin general, low recycle rates should encouragephosphorus removal. These recycle rates shouldhowever, still be above the minimum aNowed_by the

solids flux theory in addition to allowing a safety factorand the handling of peak sewage flows.

Phosphate release into solution can taken place infinal clarifiers particularly in the case of high activitysludges (i.e. sludges having a high respiration rate).With these sludges, DO and residual nitrate are quicklyconsumed and anaerobic conditions ensue in the sludgelayer. Anaerobic phosphate release occurs. If thisreleased phosphate remains in the liquid interstices inthe sludge blanket, problems will not occur. In fact, theinitiation of anaerobic conditions in the sludge in theclarifier might benefit the overall process. However, ifihis released phosphate diffuses into or is mixed into theupper layers of the clarifier liquid, it can get into the ef-fluent and negate some of the phosphorus removalachieved in the biological reactor.

In this respect, the depth of the clarifier, the heightof the sludge blanket and the distribution and retentiontime of solids in the clarifier, are of great importance.Therefore, in design, sloping bottom ciarifiers which en-courage a 'plug-flow' of sludge solids in the sludge layerare recommended. Mechanical scrapers would thencompliment this system.

Although the operation of the final clarifier cangreatly affect its performance, sludge settling propertieshave an overriding effect with the clarifier operationshaving to be adapted to the sludge settling character-istics. The latter are largely influenced by the operatingconditions in the biological reactor, in particular, the DOlevels in the aerobic zones and the fraction of the plantthat is unaerated.

If phosphate release in the clarifier becomes a pro-blem and contamination of the effluent occurs, this canbe removed by bleeding a little more nitrate or DO fromthe biological reactor. This should not be overdone, as itwill increase effluent nitrate levels and run a greater riskof nitrate feedback to the anaerobic zone.

Because this process appears to run the risk of en-couraging the proliferation of organisms which cause afrothy growth on the biological reactor, final clarifiersshould be equipped with good scum baffles and scumremoval facilities. Scum withdrawal points at the end ofthe biological reactor would also be useful. Once thisscum has been removed, it should not be recycled backto the plant inlet. Its disposal with the waste activatedsludge is thus recommended.

The growth of algae in effluent launders is also aproblem and facilities should be available to minimisethese growths or to clean them off on a regular basis.

5. SLUDGE AGE CONTROL

As was shown in Chapter 4, the parameter sludge age isuseful in the design of activated sludge plants. It canalso be of great benefit in the operation of such plants.

It is desirable therefore to use the concept ofsludge age control in the operation of nutrient removalplants in preference to the control of MLSS concentra-tion. In sludge age control the aim is to operate at fixedpredetermined sludge age (or solids retention time). TheMLSS concentration will then vary freely in response toincoming COD load changes. Procedures for controlling

11-8

sludge age are adequately covered in Chapter 4 and willthus not be repeated here.

6. REMEDIAL MEASURES WHEN NUTRIENTREMOVAL FAILS

Occasionally certain biological nutrient removal plantswill stop removing phosphorus while others might bebadly planned and designed, giving no excessphosphorus removal at all. When the usual operationaladjustments (mentioned elsewhere in the text) do notbring results, the following options can be considered:

6.1 Supplementing COD

In many cases the failure of phosphorus removal will bedue lo the lack of COD, particularly readily bio-degradable COD in the sewage feed to the reactors.This situation can sometimes be remedied by the addi-tion of extra COD to the feed from sources within theworks or from external supplies. It can be added con-tinually or only during low load periods, i.e. weekendsand holidays.

Primary Sludge. On works having primary sedimenta-tion tanks, and where primary sludge is not needed forother purposes, (i.e. gas production). The sludgeshould only be added in quantities, which boost thefeed COD into the range where phosphorus removaloccurs as any extra might overtax the aeration systemand waste energy. Addition should preferably be con-tinuous but often batch dosing of the plant can be suc-cessful. One problem which might occur is the concur-rent addition of rags with the sludge. On some plantsthese can tangle and foul up mixers and clarifier sludgewithdrawal systems. In this case, precautions should betaken to remove these rags, i.e. additional screens onthe sludge supply line.

Acid Digested Primary Sludge. When sludge is sub-jected to high rate fermentation some of the solidmaterial is converted by bacteria to volatile fatty acidsand other soluble organic compounds. The use of thismaterial for the increase in COD toad is very beneficialas the soluble organics will be a readily biodegradablecarbon source to intensify denitrification and stimulateanaerobic zone release while the fatty acids would bethe ideal substrate for certain phosphate accumulatingbacteria. Either the whole fermented sludge could beadded or the sludge couid be partially settled and anacid supernatant liquor added. Acid sludge can be pro-duced by the high rate digestion of primary sludges in aseparate reactor or by allowing primary sludge to ac-cumulate in sedimentation tanks or in any other holdingtank.

Anaerobically Digested Sludge or Supernatant Liquor.Sludge from a conventional methane producinganaerobic digester can also be used as a supplementaryCOD source, although it is not as suitable as the abovetwo sludges. Such a sludge will contain greater quan-tities of inert solids which do not benefit the process butincrease the MLSS concentration. The relatively high

concentrations of ammonia and phosphate couid causeproblems. The ammonia would have to be oxidized tonitrate which might not be denitrified completely. Also,the process might not be able to remove the additionalphosphorus. The latter problem could be overcome bypre-treatment of say, the supernatant liquor toprecipitate phosphorus and remove ammonia.

Waste Materials from Industry. Some industries pro-duce waste biodegradable organic materials. These areoften used in boosting the COD, particularly if they aresoluble. Examples are certain food and chemical in-dustries. Such materials could be discharged to thesewer feeding the works or tankered out and stored onsite. The discharge of these wastes could be timed sothat they arrive at the plant at the desired time.

Purchased Organic Compounds. When the above op-tions are not available, the purchase of organic carbonsources such as methanol or acetic acid might benecessary, but in most instances this would prove pro-hibitively expensive.

6.2 Chemical addition

This practice provides a useful alternative to CODboosting with various carbon sources. Chemicals canbe added to the activated sludge during periods wheneffluent phosphate peaks are likely to occur or con-tinuously to remove any residual phosphate which thebiological mechanism cannot remove, i.e. in the case ofsewages having high phosphorus contents. Lime or fer-ric chloride are useful chemicals for this application withferric chloride addition to the end of the aeration basinbeing a well-tried process. Any residual unused iron willthen be absorbed onto the floe and remain in the pro-cess to be available for further phosphate removal. Ironand aluminium salts are usually added to the main-stream, i.e. in to the biological reactor, while lime isused to treat phosphate with side-streams such as the li-quid phase separated from anaerobic zones. Either in-dustrial chemicals can be bought in bulk or suitablechemical by-products can be obtained from metal ormining industries. In general, the addition of chemicalsshould only be considered as a last resort and the readeris referred to the WRC publication 'Guidelines for theChemical Removal of Phosphates from MunicipalWastewaters' for details of methods that can be used.

7. UNWANTED MICROBIOLOGICAL GROWTHS

Biological removal plants are susceptible to two types ofunwanted biological growths which can adversely af-fect their performance.

7.1 Nuisance scums

Experience has shown that biological nutrient removalplants tend to favour the growth of scums which ac-cumulate on the surface of the biological reactor andfinal clarifiers. They are usually caused by the excessivegrowth of filamentous micro-organisms such as Nocar-dia and Microthrix which produce a surface active agent

11-9

enabling them to be frothed up and float as a surfacemat. These scums are an unsightly nuisance, candecompose, producing offensive odours and when theyescape into the effluent, can significantly increase itsCOD and suspended solids content.

The exact cause of these growths is not known atpresent but it is possible that they are favoured by cer-tain substrates in the feed sewage, or the environmentalconditions in the plant (DO levels; temperatures). Oncertain plants these growths are seasonal occurring onlyin the winter months but on others they can occur atany time of the year.

Successful methods of controlling these scums arenot readily available at present but possible remedies tobe tried are:

• Surface spraying with heavily chlorinated water (i.e.surface chlorination).

• Complete physical withdrawal of all the scum on theplent and the scrubbing down of all surfaces toremove residual contamination which might reseedthe process. In this respect, very good scum removalfacilities are needed which are not available on manyplants.

• Surface spraying with effluent or pumped mixed li-quor.

• Changes in aeration.

• Drastic decrease in sludge age. A combination of anyof the above could be more effective. Once the scumsolids ere removed from the plant they should bekept out and not recycled to the inlet of the works.Recycling can mean that the process is re-innoculated with a seed of these organisms.

7.2 Hlamentous growths

Filamentous growths which result in poor settling ac-tivated sludges can also occur. Large quantities of theseorganisms can cause problems with final clarifier opera-tion and result in the washout of activated sludge solidsinto the effluent.

Experience has shown that these growths appear10 be favoured by long anoxic retention limes. If the ac-tual anoxic retention time in a plant is over 1,5 h,filamentous growths could be encouraged. Also if thedissolved oxygen values in the aerobic zones are allow-ed to drop too low (i.e. below 0,5mg/ i ) , these un-wanted growths will be encouraged.

Therefore, in order to suppress filamentousgrowths, actual anoxic times should be reduced asmuch as possible (without adversely affectingdenitrification) and dissolved oxygen levels at any pointin the aerobic zones should not be allowed to dropbelow 0,5 mg ' f l . Further research is, however, neededto find better long-term solutions to these problems. Aneffective short-term remedy for filamentous growths is

the wellknown practice of chlorinating the return ac-tivated sludge line. This can produce a dramatic im-provement but is costly and requires careful control.

8. MONITORING

In general, because of their greater sensitivity,biological nutrient removal plants need more intensivemonitoring than the usual sewage treatment processes.Although some authorities will have better laboratoryfacilities than others it is essential that these plants bemonitored regularly to ensure proper operation.

For the Typical 5-stage plant the following would bea recommended daily monitoring schedule with allsamples being 24 h composites:

Feed sewage (24 h composite): COD, TKN, NHit TotalP, o-PO,

Anaerobic zone* (24 h composite of filtered samples):N03, o-PO,

First anoxic zone* (24 h composite of filtered sannples):NO3 O-PO.

Main aerobic zone* (24 h composite of filteredsamples}: N03, o-PO4, NH4

Second anoxic zone* (24 h composite of filteredsamples}: NO3, o-PO4

Final aerobic zone* (24 h composite of filteredsamples): N03, o-POA, NH4, SS, SVi (or equivalent)

Return sludge* (24 h composite of filtered samples):N03, o-P04, SS

Final effluent (24 h composite sample unfiltered): NO3,o-PO,,, NH,, SS

Feed, return sludge and waste activated sludgeflowrate should be monitored continuously as shouldthe DO levels in the unaerated zones. Feed and returnDO concentrations should be checked occasionally witha portable DO meter tu deteul unnecessary oxygen in-put to these zones. A portable DO meter is also usefulto check and calibrate permanently installed DO meters.

The analysis of filtered mixed liquor samples fromthe various zones can provide very useful informationabout the operation of a plant. These samples should betaken and filtered immediately. Either a snap sample canbe taken or a 24 h composite whereby the zone issampled every hour, the sample is filtered immediatelyand the filtrates composited into one sample, it is veryimportant to filter the mixed liquor samples quickly asthe sample will change if stored in the presence of solids(i.e. nitrates will be consumed and phosphates will bereleased into solution).

The feed and effluent samples may be analysed ona weekly basis for other parameters such as BOD, pH,alkalinity, OA, etc., and weekly microbiological ex-amination of the mixed liquor is also recommended.

Occasionally a 24 hour run can be done on theplant in which the feed, effluent and possibly some ofthe zones would be sampled hourly and each sampletested to determine the variation in plant performance.

' If a 24 h composite oi filtered samples is not possible to obtain, as many snap samples as possible per day should be taken and filtered im-mediately after being obtained.

11-10

This procedure is very useful in picking up operationalproblems such as the appearance of unnecessaryphosphate or ammonia concentration peaks in the ef-fluent. If such peaks can be detected, operationalchanges coutd be made to eliminate them and improveoverall performance.

!n addition, the feed sewage couid occasionally bemonitored to determine the concentration of certainmetals such as Ca, Mg, Fe, Al, Zn and Mn.

On-line chemical monitoring systems usingautomatic analysers or specific ion electrodes can bevery useful in detecting plant performance variationsbut their use will probably only be economically justifiedon larger installations. These units are also quite expen-sive and require trained personnel to maintain them.

Although the above sampling and analysis pro-gramme may seem ambitious to certain smallerauthorities, it is a model which should be aimed at. Theabsolute minimum requirement to achieve reasonablygood operation would be the once weekly analysis of24 h composite samples for o-PO4, NH4, NO3 andsuspended solids, with continuous monitoring of flowsand DO levels. Also, snap samples of filtered mixedliquor and effluent should be analysed for NH«, NO3 ando-PO4 as often as possible.

8.1 Typical results of chemical monitoring

Table 11.1 shows examples of typical results whichcould be expected from a five-stage plant under variousmodes of operation.

Example A: Shows probably the ideal situation withgood nitrogen and phosphorus removal. A substantialphosphate release has occurred in the anaerobic zonewith food uptake in the aerobic zone leading to very loweffluent phosphate levels. Nitrification and denitrifica-tion are good while the additional denitrification in thefinal clarifier has resulted in very low levels of nitrate inthe return sludge. The nitrate load on the anaerobiczone was thus minimal, contributing greatly to theachievement of suitable anaerobic conditions andrelease in the first zone.

Example B: Shows a plant where no excess phosphorusremoval occurred. Feed and effluent ortho-phosphateconcentrations are much the same with no release inZone 1. Virtual complete nitrification has occurred butdenitrification is only partial probably due to the lack ofCOD in the feed sewage. Effluent nitrate levels are high,reflecting in high return sludge nitrates and a largenitrate load on the anaerobic zone manifested by the1 mg N/f nitrate in this zone and the lack of release.

Example C: In this case, denitrification is adequate, thenitrate returned to the first zone is low and anaerobicphosphate release is good. However, effluent phos-phate levels are relatively high due to a poor uptake inthe aerobic zones. This slow rate of uptake is probablydue to a lack of aeration in Zone 3 which is also shownby the poor nitrification in this zone. In this case theaeration could have been decreased too much. In cer-

TABLE 11.V TYPICAL CONCENTRATION PROFILES THAT CAN BE FOUND IN 5 - ZONE PHOREDOX PLANTS

o-PO«,PNH,/NNOj.' N

o-PO«/PNH4/NNOj'N

o-P(VPNH4/NNO, N

o-PCVPNH4'N

(1) (2) (3) (4) (5) (6) 17)

2. Anaerobic 3.Primary 4. Aerobic 5. Secondary 6.Reaerotionanoxic anoxic

Mixed liquor recycle

Influent .^"I 5 Xs

1•5 —

Sludgewaste flow

8 Sludge recycle

(8)

6300

630-

630—

630_

20140

6151.0

20250

30150

640,3

649

8132

1550,1

0,40,62

60,615

2,583

3,50,20,6

0,20,52

60,515

2,883

2,20,20,6

0,20,52

60,416

2,073

1,80,20,6

0,20,52

60,416

2,073

1,80,20,6

——0,5

0,515

—1

—0,2

7Effluent

11-11

tain cases the ammonia levels could be much lower thanindicated in this exampie and the aeration could stil beinadequate for complete phosphate uptake.

Example D: Here the phosphate release in the anaerobiczone is very high and is followed by a further release inthe primary anoxic zone (a mass balance using a mixedliquor recycle of 4:1 and a return sludge recycle of 1:1will confirm this). This excessive release has meant thatphosphate uptake in the aerobic zone is incomplete,resulting in phosphate washout into the effluent. It canbe seen that phosphate uptake under aeration waspoor, despite there being enough oxygen to achievecomplete nitrification.

9. SCHEME FOR PROBLEM SOLVING INMUTRIENT REMOVAL PLANTS

Figure 11.1 shows an algorithm for the identificationand solving of problems in Phoredox plants and incor-porates many of the concepts expounded in earlier sec-tions. It should be studied by operating personnel andused as a means of identifying problem areas and sug-gesting solutions.

In this algorithmc^H represents an action. For ex-ample, the first action at the start is "measure effluentphosphate concentration". The symbol r. "77".representsa question and decision, i.e. one must decide 'yes' or'no' whether the effluent phosphate concentration isbelow 1 mg Pit or not. The symbol Q means thatone ceases action as the process should be performingas required. One then merely follows the arrowsthrough the algorithm until a solution to the problemhas been found or the effluent phosphate concentrationis below 1 rng P/ i .

10. TREATMENT OF WASTE ACTIVATED SLUDGE

When the sludge from a biological nutrient removalplant is allowed to go anaerobic, it releases ac-cumulated phosphate. This fact should be borne inmind when considering methods for the treatment ofwaste activated sludges.

• if phosphate release into the liquid phase is to beminimised, treatment should take place as soon aspossible after removal from the process, while thesludge is still in a fresh 'aerobic' state. The timeavailable before significant release of sludge occurs,depends on the activity or respiration rate of thesludge.

• In thickening and riewatering, liquid low inphosphate concentration should be removed quicklyfrom the sludge, particularly if these liquids are to berecycled back to the process.

• When large quantities of phosphate have beenreleased as in anaerobic digestion, supernatant li-quors and filtrates from sludge dewatering shouldnot be recycled to the main process. These will con-tain high concentrations of phosphate and will result

in a phosphate build-up and a consequent loss ofphosphorus removal.

• Phosphate rich liquors should either be chemicallytreated to remove phosphate or disposed of on land.

Gravity thickening of waste activated sludgeshould be approached with care. Although it is arelatively cheap operation, the risk of anaerobicphosphate release is real. If gravity thickening is donequickly, release can be minimised. Therefore, sludgeswith good thickening properties (tow SVI), which arewell stabilised (low respiration rate), couid be gravitythickened without large releases of phosphate. As withfinal clarifiers, phosphate will be released mostly in thesludge layer at the bottom of the thickener where it willdo no harm provided it is not allowed to diffuse or bemixed into the upper layers. If this thickened sludge issubsequently dewatered, the filtrate or centrate willthen contain the released phosphate. Gravity thickeningwill usually give thickened sludges of 2 - 3 % solids.

Dissolved air flotation thickening is an alternativeoperation of phosphate rich sludges. It thickens farquicker than gravity thickening and beause air is in-jected, the sludge remains aerobic for longer periods.This operation will therefore give a liquid phase of lowphosphate content and a thicker sludge than gravitythickening. Float solids concentrations vary from 3,0 to6 %, depending on sludge properties. It is obviouslymore costly than gravity thickening.

Dewatering. Continuous dewatering equipmentsuch as belt-presses or centrifuges provide a rapidmeans of activated sludge dewatering, thus minimisingphosphate release during this operation. Experience hasshown that bolt-presses give drier cakes than cen-trifuges. Even drier sludge is produced by the plate filterpress but because of the relatively longer pressing time,phosphate release will be greater in this type of equip-ment. When dewatering on drying beds, the recycle ofbed underflow to the biological reactor should be avoid-ed as it will usually be rich in phosphate due toanaerobic conditions in the sludge layer.

Digestion, anaerobic digestion provides a conve-nient way of stabilising waste activated sludge, par-ticularly if it is carried out in conjunction with primarysludge. However, anaerobic digestion does cause thesolubilisation of accumulated phosphate so that thedisposal of digested sfudge or digester liquors must becarefully considered. If anaerobic liquors are to berecycled to the liquid phase they must be chemicallytreated to remove phosphates. Also, the presence ofactivated sludge in an anaerobic digester candeteriorate the thickening properties of the digestedsludge.

• Aerobic digestion will also stabilise the sludge andcause the release of phosphates into solution.Aerobic phosphate release is generally slower thananaerobic release and is via a different mechanism,that of cell lysis. Also, if the pH in the aerobicdigester is reduced by nitrification of organic

11-12

IN. . , "1

! l

, .

> J £ I

i . , .

I

<

FIGURE 11.1: Scheme for problem solving in biologicalphosphorus removal plants

11-13

nitrogen in the sludge, phosphate release can bedramatically increased. This could be partially avoid-ed by incorporating a denitrification step in the diges-tion process. The anoxic step will preserve thealkalinity of the system end mean that chemicalssuch as lime will not have to be added to maintain thepH.

11. SLUGE DISPOSAL

Because the sludge from this type of process will behigher in phosphorus content than the other sludges,agricultural disposal could be beneficial. However, careshould be teken to ensure that the sludge Is properlystabilised either anaerobically or aerobically before it isallowed out into The environment.

• Activated sludge cannot be really stabilised in theseration tank so that once away from an aerationsource it will putrefy readily, causing significantodour problems. Thickened and dewatered activatedsludges are even more prone to causing odour pro*blems arid should be stabilised or disposed of as soonas possible.

• In dry climatic areas sludge may be disposed of onland without problems, provided it is spread thinlyand not allowed to accumulate in pools (in the caseof liquid sludge) or heaps (in the case of dewateredor dried sludge).

12. DISPOSAL OF LIQUORS

Liquids separated during the thickening and dewateringof sludges can be recycled to the liquid phase, providingthe returned niuogen and phosphorus can be removedin the main process, if this is not possible, intermediatechemical treatment for their removal will be necessaryor the liquors can be disposed of elsewhere. For exam-ple, the irrigation of liquor onto agricultural land couldbe possible. Agricultural disposal has attractivepossibilities as chemical precipitation can often be cost-ly.

13. COMPLIANCE WITH STANDARDS

In designing nutrient removal plants, the designer mustbe aware of the effluent standards requirements and theability of the proposed plant to meet desired effluentquality limits. In this respect, the reliability of the pro-cess and the effluent quality variability must be con-sidered.

In most of the "sensitive" areas in South Africa,the modified General Standards will have to be com-plied with. These are summarised in Table 11.2

Regulations in terms of the 1956 Water Act requirethat these limits be adhered to at all times. However,because of the variability of any real process, 100 %compliance will not be possible. There will always beoccasions when the limit is exceeded even in the mostoptimum designs. Therefore, in this chapter we will

TABLE 11.2- SUMMARY

ParameTei

Chemical Oxygen DemandAmmoniaSuspended Sciidso-Phosphate

OF GENERAL

CODN

P

STANDARDS

Limit mg / /

Less than 75Less lhan 10Less than 25Less than 1,0

consider the case of say, 95 % compliance with theStandards with the aim being to design reliable, stableplants which will achieve this goal.

Statistical analyses of the data from sewage treat-ment plants has shown that the performance of theseplants on an annual basis is quite variable. Also, thedata is not normally distributed, with a log-normaldistribution giving the best fit of observed data. Thismeans that when plotted on log-probability graphpaper, the data gives a straight line with deviations fromlinearity at either end.

Figure 11.2 shows a log-probability plot of datafrom the Goudkoppies plant in Johannesburg. Whenthis plant was underloaded {top graph}, it could onlycomply with the phosphate limit for just over 60 % ofthe time. However, when a high strength sewage wastreated (lower graph), a 94 % compliance was possible.

Figures 11.3; 11.4 and 11.5 show similar graphs forammonia, COD and nitrate. Although there is at presentno nitrate standard it is interesting to note how improv-ed phosphate removal coincides with better denitrifica-tion.

As can be seen the ammonia limit of 10 rng N/ iwas very easy to meet, in fact, better than 99 % com-pliance was possible for both sets of data shown. In thecase of the COD, better than 95% compliance with the75 mg/ / limit was possible with the weak sewage butonly 45 % when treating the strong sewage due to theletter's industrial waste component.

Tables 11.3 and 11.4 show further results of thestatistical handling of these two sets of data. As can beseen, the geometric means (obtained by averaging thelogarithms of the results), were usually much lower thanthe arithmetic means. This indicates that the data is"skewed", i.e., that there are a few results much higherthan the bulk of the results. It is these results which pre-vent compliance with the standard 100% of the time.

Also shown on the bottom lines are the meandesign values which should be used (bearing in mindthe variability of the data) to achieve the General Stan-dards for 95 % of the time. A nitrate limit of 10 rng N/ ihas also been included for interest sake. It is interestingto note that to achieve the phosphate limit of 7 mg P/lfor 95 % of the time, one must use a design value of= 0,4 mg P/L Similarly, design values of =3 mg N/i toachieve an ammonia of 10 mg/t and =45 mg Oil forCOD of 75mgfft must be used. (It should be remem-bered however, that the effluent COD is a function ofthe effluent suspended solids and the influent soluble

11-14

^ o3 °"

o -c

to o

a n7T O

•a to

PERCENTAGE PROBABILITY EQUAL OR LESS3 tn

BI X PERCENTAGE PROBABILITY EQUAL OR LESS

3

m

cm

o2

3 *•

<" o °

o 3a r*

"D

m

cm

0

oi

"D

oCO"0JO>

m o

^^3«a -o- • o

-, a

CO ^ T

c -<" * T I

OnO 3

PERCENTAGE PROBABILITY EQUAL OR LESS

mg

m

6

Hm

o-i (ji -gO O O O)

1 1 I I I

1 1

11

1 1

1 1

-

: • 1 1 1 1

"V. ' '

\

feedonly

i ( i

"Si 1

s/ w

ithm

ainsew

ef

i i

i ' i

\

\\

i i i i

! I

1 1

1 1

n>

ft>

a>

, (

*

i i

i '

• < i 1

main and

-

1 -

ooso"0

rnCO

Soocr~m

3

O

in

3

oc-?•g

•°. s

S §'2.

<O

Q

?lodul

tnc3

• m

CO

Log-pro

crC

• <

o"

cre

3ooo

rcmzH

oOa

^ ro

PERCENTAGE PROBABILITY EQUAL OR LESS

m

ro

Q o•a 3

TABLE 11.3: RESULTS

Number o f resuiisArithmetic meanStandard deviationCoefficient of variationGeometric meanMean design value for 95

OF A STATISTICAL ANALYSIS OF THE PERFORMANCE OF GOUDKOPPIES MODULETREATING THE MAIN SEWER FLOW

\A hourly samples taken between 14 June and 31 December

TotalPhosphorus

9611,110.720,650,93

% reliability —

Ortho-Phosphate

1 1230.890.740,820,480,40

1981)

TKN Ammonia

9882,711,840,682,23

1 0501.332.101.580,133,0

COD

1 12343,4312,590,29

41,4149

2 WHEN

Nitrate

1 1094,512,940,652,91

4.5

TABLE 11 1: RESULTS OF A STATISTICAL ANALYSIS OF

Number of resultsArithmetic meanStandard deviationCoefficient of variationGeometric meanMean design value for

"Based on an effluent

THE PERFORMANCE OF GOUDKOPPIES MODULETREATING THE MAIN AND RELIEF SEWER FLOW

(4 hourly samples taken between 10 January

TotalPhosphorus

1 0640.660,420.650,58

95 % reliability' —

Ortho-Phosphate

1 0770,360,350,910,270,45

Standard of COD <7. mg / ; Ammonia <10 mg N// o-PO

and 11 July 19821

TKN Ammonia

9062,780,930,332.63

—

t <1 mg P'J

1 0710,390,972,520,01

2,7

and NO,

COD

1 07077,1825,68

0,3372,97

46

<10 mg N/f

2 WHEN

Nitrate

1 0651,602,160,350,54

3,1

non-taiodegradable COD fraction and this cannot reallybe designed for}.

Thus designers should use these values in theirdesigns rather than the values in tho Genera! Standardsin order to conceive plants which will comply with theStandard in a more reliable manner.

14. REFERENCE

NIKU, S., SCHROEDER, ED. , TCHOBANOGLOUS, G-, ANDSAMANEIGO, F.J. (1981) Performance of Activated Sludge Pro-cess: Reliability, Stability and Variability. USEPA Publication,December 1981.

11-16

CHAPTER TWELVE

SUMMARY OF CASE STUDIES ON BIOLOGICAL NUTRIENT REMOVAL

AT

I. JOHANNESBURG'S SEWAGE PURIFICATION WORKS by A.R. Pitman

II. THE NtWR DASPOORT PILOT PLANT by A. Gerber

1. INTRODUCTION

Biological nutrient removal is today practised widely inSouth Africa, as well as in isolated instances in SouthWest Africa, Zimbabwe, the United States, Canada andEurope. At the IAWPR post conference seminar onphosphate removal in biological treatment processes,held in Pretoria during April 1982, a number of casestudies on biological nutrient removal were presented,amongst others, by Paepcke (1982), Pitman era/(1982),Kerdachi and Roberts (1982) and Barnard (1982).

In this chapter two case studies carried out undercontract to the Water Research Commission are brieflydealt with. In the first case study operating experiencewith nutrient removal activated sludge plants in Johan-nesburg is described. For more detailed information thereader is referred to the joint publication by City Council

of Johannesburg and the Water Research Commissionentitled "Optimisation of the Modified Activated SludgeProcess for Nutrient Removal" which can be obtainedfrom the Water Research Commission, P O Box 824,Pretoria, Republic of South Africa. In the second casestudy information is presented of pilot-scale applicationof the Phoredox and UCT processes at the DaspoortSewage Treatment Works in Pretoria. Greater details ofthis study are given in the joint National Institute forWater Research and Water Research Commissionpublications entitled "Technical Guide on BiologicalNutrient Removal: The Phoredox Process" and"Biological Nutrient Removal from Wastewater Ef-fluents: Performance Evaluation of the Phoredox andUCT Processes". Both these publications are obtain-able from the Water Research Commission at theabovementioned address.

2, CASE STUDY NO. I

OPERATING EXPERIENCE WITH NUTRIENT REMOVAL ACTIVATED SLUDGE PLANTS -

GOUDKOPPIES AND NORTHERN PURIFICATION WORKS, JOHANNESBURG

The Johannesburg City Council operates two largebiological nutrient removal activated sludge plants atGoudkoppies and the Northern Works. Each has adesign capacity of 150 Mf/d which is split into three5 0 M / / d modules incorporating the five-stagePhoredox layout (See Figure 12.1). Both plants haveprimary sedimentation tanks and settled sewage is fedto the biological reactors via in-line flow balancingtanks. These plants which were designed in 1973 and1974 respectively, represent the initial phase of the pro-cess design evolution of biological nutrient removalplants. Each has anaerobic zones of approximately 1hour nominal retention time which by 1983 technologyis rather short. The performance of these plants shouldthus be viewed with the knowledge that their designdoes not represent the present state of the art of pro-cess design.

Experience at Goudkoppies covers two distinctperiods; first when the plant was underloaded, receiving50 per cent of design load and second, when the wasteflow to the plant included a high strength flow contain-ing considerable quantities of yeast industry effluentand the plant operated at approximately its designorganic loading.

•

When the plant was underloaded, and the influentwastewater total phosphate was between 7 to 8 mg/f(as P), an effluent with a mean ortho-phosphate con-centration of just under 1 mg P/i could be produced(see Table 12.1), but performance was erratic with fre-quent effluent phosphate peaks usually occurring after

TABLE 12

Wastewater

1: AVERAGE PERFORMANCE

: low andcharacteristics

Mean daily flowSuspended solidsCODOAlSABSiTDSNH,asNNO3asNTotal PasPP0<3SPpHConstruction CostCommissioning Date

im> » 103'd>(mg/f)(mg/i)(mg//)(mg/i)(mg//)(mg/i)(mg//)(mg//l

July 1980Influent

88210640

20

7,5277,3R11

- June 1981Effluent

1648

7550

4,73,51.40,77,6

000 OOO1978

12-1

15

1 SCREENING2 GRIT REMOVAL3 PRIMARY SEDIMENTATION4 FLOW BALANCING5 FLOW CONTROL6 ANAEROBIC ZONE7 FIRST ANOXIC ZONE8 MAIN AERATION ZONE

9 SECOND ANOXIC ZONE10 REAERATION ZONE11 MIXED LIQUOR RECYCLE12 WASTE ACTIVATED SLUDGE13 CLARIFICATION AND SLUDGE

THICKENING14 ACTIVATED SLUDGE RETURN15 EFFLUENT

FIGURE 121: Goudkoppies Purification Works - Plant layout

500COD

OPO./P

TKN/N

TotP

(All asmg.'il

J

Jan Mar 1982

FIGURE 12.2 Performance (Based on 4-hourly sampling)^ Feed . a ^ . = Effluent

12-2

weekends and holiday periods. Phosphate removal wasvery dependent on denitrification. Feed sewages havinghigh TKN/COD ratios (0,15) and/or low COD concen-trations (300mg/i) produced effluents containingrelatively high nitrate levels (i.e. 5-10 mg N/i). This wasreflected in high return sludge nitrates (4-7 mg N/i]from the clarifiers and excessively high nitrate loads onthe anaerobic zone with the consequent loss of thedesired anaerobic conditions. Phosphate release intosolution in the anaerobic zone was found to benecessary for good phosphate removal with ortho-phosphate concentrations in the range 15-25 mg P/ibeing ideal. The recycle of nitrate to the anaerobic zonetended to inhibit this phosphate release. For this reasonit was essential to maintain an effluent with a low nitrateconcentration (less than 2 mg N/i). When adequatedenitrification could not be achieved the operation waschanged to limit the oxygen input to the main aerobiczone. This action suppressed nitrification and ensuredthat the nitrate in the recycle remained below 2 mg N/fbut the ammonia concentration in the effluent increasedlo approximately 5 mg/jf. However, this mode of opera-tion proved difficult to optimise due to the varying quali-ty of the feed sewage.

When the higher strength sewage containing theyeast waste was treated the performance of the Goud-koppies Works improved dramatically. Denitrificationwas more efficient and effluent total phosphate concen-trations were decreased from 7-8 mg/ i (as P) to below1 rng/f (as orthophosphate P) for 95 per cent of thetime and often betow 0,2 mg P/ £. Nutrient removal wasfar more stable and less erratic (see Figure 12.2), whilethe plant was easier to control. A mean effluent ortho-phosphate concentration of 0,36 mg P/f and a totalnitrogen (TKN + NOj) level of 4,3 mg N/jf was obtainedduring this period.

Underloading and unfavourable characteristics* ofthe settled sewage also led to poor nutrient removal atthe Northern Works. Both the effluent nitrate andphosphate concentrations exceeded 5 mg/jf and therewas no phosphate release in the anaerobic zone duemainly to the large recycle of nitrates via the clarifierunderflow and the low concentration of readily biode-gradable COD. !n an experiment however, when aciddigested primary sludge was fed to the anaerobic zoneof one module, to increase the readily biodegradableCOD concentration effluent phosphate concentrationsdropped to below 0,5 mg P/7 and nitrates to below2 mg N/f. The higher COD and more favourablecharacteristics of the feed to the plant imbued by theacid sludge, greatly improved nutrient removal.

Studies also showed that these plants (particularlyGoudkoppies) were prone to the growth of filamentousmicro-organisms which resulted in poor settling sludgesand nuisance scums. Poor sludge settling created fre-quent operational problems, particularly when SVIvalues in excess of 300 m i ' g were encountered. Sludgesettling properties could change very rapidly, often forno obviously apparent reason, but once filamentousorganisms predominated in the activated sludge theywere extremely difficult to eradicate. Factors which ap-

*High TKN. COD raiio of influent; low readily biodegradable COD-

peared to encourage these nuisance growths wereoperation of modules with large unaerateri mass frac-tions and the suppression of aerobic zone dissolvedoxygen levels to below 1 mg/ I .

Microscopic observations during periods of goodphosphate removal showed activated sludges contain-ing large quantities of clustered bacteria which contain-ed significant amounts of polyphosphate storagematerial. These bacteria had the appearance ofAcinetobacter spp and seemed to be a dominantspecies in the sludges. Poor settling of activatedsludges was found to be due to the prolific growth offilamentous organisms.

Application of the Marais-Ekama activated sludgemodel (described in previous chapters} to data fromJohannesburg's nutrient removal works demonstratedthat the model accurately predicated the COD andnitrogen removal performance of both the Goudkoppiesand Northern Works as well as their oxygen consump-tion rates and mixed liquor suspended solids concentra-tions. The model also gave a good indication of whyphosphate removal was not optimum and whatmeasures would be necessary to improve performance.These predictions were verified when the addition ofacid sludge at Northern Works and yeast waste atGoudkoppies greatly improved phosphate removal.

Another interesting experiment at Goudkoppieswas the investigation of an on/off operating mode. Thisstudy was undertaken during the initial period when theplant was underloaded. Because of inadequate sewageflow and COD concentration, Module 3 was operatedon a reduced feed of approximately 20 Mi/day, whichrepresented about 25 per cent of its design COD load.Under these circumstances, phosphate removal was im-possible with the normal continuous operating mode.However, when an on/off operating mode wasadopted, phosphate removal improved. In this modesewage flow to Module 3 was stopped completely be-tween 07h00 Saturdays and 08h00 on Mondays and allmechanical plant was switched off for this period. Ac-tivated sludge settled to the floor of the reactor andwent anaerobic releasing phosphate into solution. OnMonday mornings feed to the module was graduallyreinstated and when aerators became sufficiently im-mersed again, mechanical plant was switched on andthe module run at its usual setting until the nextweekend. Performance was characterised by a high ef-fluent phosphate peak soon after start up on Mondayswhich dropped rapidly on Tuesdays to become wellbelow 1 mg Pit for the rest of the week. On Thursdaysand Fridays only trace quantities of phosphate werefound in the effluent and overall there was a net 50 percent improvement in phosphate removal over previousoperating modes.

An important finding during the operation of theseplants was the inefficient performance of the secondanoxic zone. On an average this zone removed approx-imately 1 mg N/f of nitrate and contributed very little tothe overall denitrification achieved in these plants.

The detailed study of the performance and op-timisation of these plants was the topic of a WaterResearch Commission contract with the JohannesburgCity Council. The work was undertaken in collaborationwith the University of Cape Town and the National tn-

12-3

stitute for Water Research. For further details of theCouncil's contribution to this project, the reader is refer-red to the detailed final report entitled "Optimisation ofthe Modified Activated Sludge Process for Nutrient

Removal" which is available from the Water ResearchCommission, P 0 Box 824, Pretoria, 0001, Republic ofSouth Africa. References to other publications produc-ed during this study can also be found in this report.

3. CASE STUDY NO II

BIOLOGICAL NUTRIENT REMOVAL AT THE DASPOORT SEWAGE WORKS PILOT PLANT,

PRETORIA

The National Institute for Water Research of the Coun-cil for Scientific and Industrial Research investigatedbiological nutrient removal under simulated practicalconditions over a period of five years.

For this purpose a 100 mVd pilot plant, situated atthe Daspoort Sewage Treatment Works in Pretoria, wasoperated in the Phoredox and UCT modes at sludgeages varying between 15 and 20 days. A multiple basindesign was used, with the number of reactors in theprimary anaerobic and anoxic stages varying from 2 to 3and in the aerated zone between 3 and 5. When includ-ed, the secondary anoxic and aerated stages consistedof 4 and 1 basins respectively. Each basin had a nominalhydraulic retention time of approximately 1,2 h. Themixed liquor (at and sludge (s) recycle ratios usuallywere maintained at values of 2:1 and 1:1 respectively forthe Phoredox configurations but a number of mixedliquor return rates were used when studying the UCTmode. In the Phoredox configuration the mixed liquorrecycle ratio of 2:1 was selected after several pilot planttests indicated this ratio was near optimal for the ex-perimental conditions and plant layout used.

Settled sewage was fed during the study, with theexception of a six month period during which screenedraw sewage was used in an effort to determine the ef-fect of increased influent COD concentrations onphosphate removal efficiency. Median values for feedCOD (as mg/f), total phosphate (as mg P//) and theTKN/COD ratio (mg N/mg 0) during the five-year in-vestigation were respectively 320, 6,2 and 0,10 while thecentral 80 per cent of the individual values lay in therespective ranges 210 to 410, 4,2 to 8,6 and 0,14 to 0,08.

Several additional studies were performed duringthis period apart from evaluating the long termbehaviour of the Phoredox and UCT configurations perse. These included the effect of variable a-recycle rateson process performance, the elimination of peak ef-fluent phosphate concentration by the addition ofrelatively high strength sewage collected beforehand,the influence of elevated DO concentrations in thes-recycle and the influence of low DO concentrations inthe aerated stage.

The median effluent concentrations observed forthe five-stage Phoredox process during a 14-month in-vestigation between March 1977 to May 1978 were 22;0,8; 0,02 and 5,3 mg/f for COD, orthophosphate (as P),ammonia (as N) and nitrate (as N) respectively. Thecentral 80 per cent of all observations lay in the respec-tive ranges 14 to 37; 0,1 to 3,1; 0,0 to 0,2 and 2,3 to 8,0mg/jf. These results were achieved at mean influent

COD and total phosphate (as P} values of 277 and 5,5mg/f respectively, and a TKN/COD ratio of 0,11. Dur-ing the period October 1973 to February 1979, when themean influent total phosphate concentration was 6,0mg/ i and the COD and TKN/COD values slightly morefavourable for enhanced phosphate accumulation,namely 307 mg/ i and 0,10 respectively, the processproduced effluents of which the median COD, ortho-phosphate, ammonia and nitate concentrations were20; 0,4; 0,01 and 3,8 mg/ ! respectively. The central 80per cent of all observations during this phase of thestudy lay in the respective ranges 18 to 23; 0,1 to 2,5;0,0 to 0,15 and 1,7 to 6,7 mg/ i . When the mean in-fluent COD value was artiftcally increased to 330 mg / i ,by collecting peak strength sewage on Fridays and blen-ding this product in a 1:1 ratio with the relatively lowstrength feed normal for weekends, the plant producedmedian effluent COD, orthophosphate, ammonia andnitrate concentrations of 22; 0,4; 0,1 and 5,9 m g / i . Thiswas achieved despite a relatively high mean influenttotal phosphate concentration of 7,2 mg/f and an un-favourable TKN/COD ratio of 0,11. The correspondingranges in which the central 80% of the aforementionedparameters lay were 18 to 25; 0,1 to 1,5; 0,0 to 0,1 and3,5 to 9,5 mg/f. While not necessarily considered to bea practical procedure the blend resulted in a sharp im-provement in effluent and an almost complete elimina-tion of the peak values characteristically occurring onSundays or Mondays.

The phosphate removal capability of the pilot plantdeteriorated upon changing the feed from settled toscreened raw sewage, despite relatively favourablemean influent COD and total phosphate values of 369and 5,3 mg/ i respectively and a favourable TKN/CODratio of 0,09. During the period May to October 1978,when this mode of operation was evaluated, the medianeffluent concentrations were 23; 1,3; 0,03 and 4,0 mg/ ifor COD, orthophosphate, ammonia and nitrate respec-tively, with the central 80 per cent of all values in thecorresponding ranges 16 to 30; 0,3 to 3,2; 0,0 to 0,4;and 0,4 to 6,9 mg/f. The relatively poor performance ofthe plant with respect to phosphate removal during thisperiod led to the conclusion that primary sedimentationhad a beneficial effect on the subsequent phosphate up-take capabilities of Phoredox plants under similar ex-perimental conditions to those of this study.

The UCT process was evaluated during threedistinct operational periods. The best of these was thatfrom November 1980 to December 1981, during whichthe mean influent COD and total phosphate values were

12-4

324 and 7,5 mg/f respectively and the mean TKN/CODratio 0,10. The median effluent concentrations for COD,orthophosphate, ammonia and nitrate were 26; 0,8;0,14 and 8,0 m g l respectively, with the central 80 percent of all observations in the corresponding ranges 19lo 35; 0,1 to 3,1; 0,0 to 0,7; and 4,0 to 12,0 mg/f. Thisbest performance of the UCT process was achieved on-ly after the anaerobic stage was extended from two tothree basins.

The study showed that the processes of carbonoxidation, nitrification and denitrification should notpresent the designer with significant problems inestablishing an appropriate nutrient removal plant.Such a facility should prove capable of producing a vir-tually completely nitrified effluent with average CODand total nitrogen values in the respective ranges 20 to30 and 4 to 7 rug!t for situations with wastewater andplant characteristics similar to or more favourable thanthose pertaining to the Daspoort plant (refer to Chapter7). Effluent phosphate concentrations from thePhoredox process can be expected to exhibit peak

values in sympathy with reduced COD input overweekends but otherwise should be sufficiently low toachieve overall average levels in the range 0,5 to 1,0mg/ i (as P) when settled sewage is fed to the plant. It isnoteworthy that this degree of phosphate removal canbe achieved in the Phoredox process despite averagenitrate concentrations around 5mg/ f (as N) beingreturned to the anaerobic stage. Nonetheless, it shouldbe remembered that effluent concentration is a functionof influent phosphate concentration and achievableremoval, and this should be considered for each sewageand process. The UCT process is highly successful insuppressing effluent peak phosphate values but overallis not expected to prove advantageous over thePhoredox process for conditions similar to those underwhich this case study was performed. Detailed resultspertaining to the performance of the pilot plant undervarious modes of operation may be found in the reportsby Simpkins and Gerber (1981) and Gerber and co-workers (1982}.

4. REFERENCES

BARNARD, J.L. (1982) Design consideration regarding phosphateremoval m activated sludge plants. 1AWPR Post ConferenceSeminar on Phosphate Removal in Biological Treatment Process,CSIR Conference Centre, Pretoria, 5-6 April 1982.

GERBER. A., SIMPKINS, M.J., WINTER, C.T. and SCHEEPERS,J.A. (1382) Biological nutrient removal from wastewater effluents:Performance evaluation of the Phoredox and UCT processes.Contract report C WAT 52 prepared for the Water Research Com-mission by 'he National Institute for Water Research (Obtainablefrom the Water Research Commission, P O Box 824, Pretoria,Republic of South Africa 0001)

KERDACHI, D.A. and ROBERTS. M R. (1982) Full scale phosphateremoval experiences in the Umhlatuzana works at different sludgeages. IAWPR Post Conference Seminar on Phosphate Removal inBiological Treatment Processes, CSIR Conference Centre,Pretoria. 5-6 April 1982.

PAEPCKE, B.H. (1982) Performance and operational aspects ofbiological phosphate removal plants in South Africa. IAWPR PostConference Seminar on Phosphate Removal m Biological Treat-ment processes, CSIR Conference Centre, Pretoria, 5-6 April1982.

PITMAN, A.R., VENTER, S.L.V. and NICHOLLS, H.A. (1982) Prac-tical experience with biological phosphorus removal plants inJohannesburg. IAWPR Post Conference Seminar on PhosphateRemoval in Biological Treatment Processes, CSiR ConferenceCentre, Pretoria, 5-6 April 1982.

PITMAN A.R. (1983) Optimisation of the Modified Activated SludgeProcess for Nutrient Removal. Joint publication by the City Coun-cil of Johannesburg and the Water Research Commission. (Ob-tainable from th Water Research Commission, P O Box 824,Pretoria, Republic of South Africa 0001).

SIMPKINS, M.J. and GERBER, A. (1981) Technical guide onbiological nutrient removal: The Phoredox process. Confidentialreport prepared for the Water Research Commission by the Na-tional Institute for Water Research, 1981. (Obtainable by arrange-ment with Messrs SAIDCOR, P O Box 395, Pretoria, Republic ofSouth Africa 0001).

12-5

APPENDIX 1

LIST OF SYMBOLS

a = mixed liquor recycle ratio from the aerobic to the anoxic reactors.

a0 Subscript o denotes opt imum.

A = surface area of secondary settling tank

bh T ~ endogenous mass loss rate for heterotrophic organisms at T°C (/d)

br,T = endogenous respiration rate for nitrifying bacteria at TC'C (/d)

- b r i20(1,029) lT-201

t>h» = the rate at 20°C = 0,24/d

bn20 = the rate at 20°C - 0,04/d

Dp = denitrif ication potential ( m g N / / influent)

Dpi, Dp3 Subscripts 1 and 3 refer respectively to the primary and secondary anoxic reactors

Dpoi, DPd2 Subscripts d1 and d2 refer respectively to the first and second anoxic reactor of a Modif ied UCT pro-cess

Dpp = denitrif ication potential of the process when the maximum anoxic sludge mass fraction is all in theform of a primary anoxic reactor

f = unbiodegradable fraction of active mass = 0,20 mgVSS/mgVSS

Uv> fa i = active fraction of the sludge mass wi th respect to the volatile and total solids concentrations respec-

tively

f,)S ~ readily biodegradable COD fraction of the influent wi th respect to the biodegradable COD concentra-

tion

fcv - COD to VSS ratio of the volatile sludge mass = 1,43 mgCOD/mgVSS

fi = MLVSS to MLSS concentration ratio of the mixed liquor

fn = nitrogen fraction of the MLVSS (mgN/mgVSS) = 0,10 mgN/mgVSS

fna - amonia fraction of the influent TKN

fnu = soluble unbiodegradabie organic nitrogen fraction of the influent TKN

fns = TKN/COD concentration ratio of the influent

fp = phosphorus fraction of the inert MLVSS and endogenous residue MLVSS = 0,015 mgP/mgVSS

ffps = fraction of raw total COD removed by primary sedimentation

fu = unbiodegradable COD fractions in the influent (mgCOD/mgCODI. Additional subscripts pand s referrespectively to particulate and soluble fractions

fv* = general parameter for volume fractions

f," = general parameter for sludge mass fractions'Additional subscripts a, b, d, 1, 3 and m refer respectively to anaerobic, total aerobic, total anoxic,primary anoxic, secondary anoxic and maximum unaerated allowable sludge mass fraction

AM

f^dl Additional subscripts t and m following the d subscript refer respectively to the total and maximumfxdm allowable sludge mass fractions

fximin Subscript min following subscripts 1 and 3 refer to the minimum primary and secondary anoxicf*3™ sludge mass fractions

f >d1 Subscripts 1 and 2 following the d subscript refers respectively to the first and second anoxic reactors

f,d2 of a Modified UCT process

F/M = Food to Microorganism ratio

G = General parameter denoting flux (kg/mVdh Subscripts ap, b, ! and s refer respectively to appliedflux, bulk flux, limiting flux and gravity settling flux

K = general parameter for denitrification rate (mgNO3/mgVASS/d). Subscripts 1 and 2 refer respectivelyK1( K2 to the 1st and 2nd rates in the primary anoxic and 3 to the rate in the secondary anoxic. AdditionalK3 subscripts T and 20 refer to T°C and 20 cC respectively

K, = conversion rate of biodegradable organic nitrogen t o free and saline ammonia by active mass =0,015 X / m g V A S S / d . Subscripts T and 20 refer to rates at T°C and 20°C

Kn = half saturat ion coefficient in Monod equat ion for nitr if ication = m g N / i .Subscripts T and 20 refer respectively to T°C and 20°C.Subscripts pH and 7,2 refer respectively to Kn at different pH values and that at pH 7,2

LF = load factor

M = prefix denot ing mass as opposed to concentrat ion of a variable

N = general parameter denot ing nitrogen concentrat ion (mgN/j f )

N a , N n - Subscripts a, n, o, t and u refer respectively to ammonia, nitrate, biodegradable organic n i t rogen ,N o , N, total TK.1M and soluble unbiodegradable organic nitrogen concentrat ions. Additional subscripts e, i, r,Nj s and a refer respectively to the concentrations in the effluent, influent, r-, s- and a-recycle flows

Nr. - nitrification capacity (mglM/I influent)

Np i = unbiodegradable paniculate organic nitrogen concentration in influent (mgN/ i )

Ns = nitrogen required for sludge production (mgN/ f influent)

NJi = nitrogen available for nitrification

= N t ( - N s ( m g N / i influent)

O = general parameter for oxygen

Oc, On Subscripts c, n, d and t refer respectively to the oxygen demands for carbonaceous material degrada-Od, O, t ion, nitrification, that recovered by denitrification and total oxygen demand0,, Os, 05 Subscripts i, a snd s refer respectively to the dissolved oxygen concentrations in the influent a- and

s-recycles

OUR = oxygen utilization rate (mgO/ f / h )

P{ = excess P removal propensity factor (mgCOD/ i )

Ps = phosphorus in daily sludge wastage per i influent f low ( m g P / i ) i.e. the phosphorus removal f r o m t h e

wastewater

P( = total phosphorus concentrat ion ( m g P / i ) .Addit ional subscripts i and e refer respectively to influent and effluent

q = f l ow rate for waste sludge f rom the process reactor ( i / d )

A1-2

Rs

s

s,

s

sb,

s b s .s b l ,

s U f s

s b s i

Q = daily mean influent f low rate ( f / d )

r = mixed liquor recycle ratio from the anoxic to the anaerobic reactor in the UCT type processes

Rhn = process total average nominal hydraulic retention time (d)

= general parameters for actual and nominal hydraulic retention times. Subscripts p and s refer toprimary and secondary anoxic reactors

= sludge age (d)

= minimum sludge age for nitrification

= underflow recycle ratio

= factor of safety with respect to nitrification

= factor of safety with respect to conversion of a UCT process to a Modified UCT process

= general parameter denoting COD concentration

Subscripts b, u and t refer respectively to biodegradable, unbiodegradable and total COD concentra-tions

Additional subscripts i, e, s and p refer respectively to concentrations in the influent and effluent, andreadily biodegradable and particulate COD

S| 1 5 a = readily biodegradable COD concentration in the anaerobic reactor

Scoo = substrate concentration with respect to COD

SROD = substrate concentration with respect to BOD

SUR = substrate utilization rateSubscripts a, v and t refer respectively to the rate with respect to active, volatile and total sludge con-centrationsAdditional subscripts COD and BOD refer to the rate with respect to the substrate measurement interms of COD or BOD

t = time

t, = duration of first phase of denitrification in the primary anoxic reactor

Uo, U,j — overflow and underflow velocity in the secondary settling tank (m/d)

v = volume of waste sludge abstracted from process reactor per day

V = general parameter denoting volumeSubscripts p and r refer respectively to the total process and reactor

Vs = gravity settling velocity of sludge (m/d)Additional subscripts denote velocity at specified sludge concentrations

X = general parameter denoting sludge mass concentrationSubscripts a, e, i, v, t and n refer respectively to active, endogenous, inert, volatile, total and nitrifiersludge concentrationsAdditional subscripts f and i, and a, d and b refer respectively to concentrations in effluent and in-fluent and those in the anaerobic, anoxic and aerobic reactors

X, = process average total sludge concentration (mgTSS/i)

X = for settling tank theory denotes total MLSS sludge concentration (kgMLSS/m3}Subscripts af, dz, ef, I, o and r refer respectively to the siudge concentration of the layer above thefeed point, the dilute zone below the feed point, the effluent, the limiting zone, in the biological reac-tor and the underflow sludge return

Al-3

Yh = heterotrophic organism yield coefficient = 0,45 mgVSS/mgCOD

Yr, - nitrifier organism yield coefficient = 0,10 mgVSS/mgN

a = denitrification attributable to the readily biodegradable COD (mgN03-N/mg biodegradable influentCOD)

y = coefficient of excess phosphorus removal {mgP/mgVASS) i.e. the proportion of phosphorus in the

active mass

A = prefix denoting the change in the parameter following

<j>,1s = pH sensitivity parameter for nitrification in the simplified Arrhenius equation = 2,35j j n = specific growth rate of the nitrifiers (/d)

Subscript m denotes the maximum rateAdditional subscripts T and 20 refer respectively to the rate at T°C and 20cCAdditional subscripts pH and 7,2 refer respectively to the rate at different pH values and at pH 7,2

2,86 = oxygen equivalent of nitrate i.e. 2,86 mg oxygen can accept as many electrons as 1 mgN03-N nitrate

8,6 = mg mass of COD utilized per mgNOj-N nitrate denitrified

4,57 = mg mass of oxygen required for nitrifying 1 mgN ammonia to 1 mgN nitrate

= decimal place

= multiplication symbol

Al-4

APPENDIX 2

BRIEF DESCRIPTION OF THE DETERMINATION OF THE READILYBIODEGRADABLE COD FRACTION OF A MUNICIPAL WASTEWATER

1. INTRODUCTION

In the experimental investigation into the dynamicbehaviour of the activated sludge process (Doid, Ekamaand Marais, 1980}, it was found that in the aerobic com-pletely mixed process at short sludge ages (1,5 to 3days) under daily cyclic square wave loading conditions(feed 12 hours on, 12 hours off}, there is a precipitousdecrease in oxygen consumption rate immediately afterthe termination of the feed period (see Figure A2.1). Inmodelling the activated sludge process, it was foundthat this behaviour was due to the readily biodegradableCOD in the influent; at feed termination, the addition ofreadily biodegradable COD stops so that the utilizationof the readily biodegradable COD stops, causing a stepchange in the oxygen consumption rate. Hence amethod for determining the readily biodegradable CODin the influent is the measurement of the step change inoxygen consumption rate at feed termination in a dailycyclic square wave loaded completely mixed processoperated at a very short sludge age.

2. APPARATUS

The test is performed using a single reactor completelymixed aerobic activated sludge unit operated at a sludgeage of about 1,5 to 3 days depending on thetemperature; the higher the temperature the shorter thesludge age. The data shown in Figure A2.1 are from a2,5 day sludge age unit at 20°C. A schematic represen-tation of the process is shown in Figure A2.2.

The following points warrant notice regarding thesetting up and operation of the process:

(It Aeration. The most convenient method of supply-ing aeration is to use an ordinary fish tank air pumpwith a fine bubble sparger stone attached to aperspex tube. By adjusting the depth of immersionof the sparger the DO level can be controlled to thedesired value of about 2 mgO/f.

(2t Inlet and outlet from aeration reactor. All streamsto and from the reactor should enter the reactorbelow the liquid surface and not be dripped into thereactor; dripping feed into the reactor may result inincorrect readings by exposing the feed to air andallowing DO entrapment via the liquid/air inter-face.

(3) Mixing. The contents of the feed tank should bestirred slowly and gently in order to keep par-ticulate matter in suspension, but to avoid ex-cessive DO entrapment into the reactor contents.

Covering the reactor contents by floating on it aSagex disc is a very effective method for stoppingDO entrainment into the reactor caused by mixing.

(4) Sludge wastage. In order to maintain the correctsludge age, the correct volume of mixed liquormust be withdrawn from the reactor. Say the reac-tor volume is 10 i and the sludge age is to be 2,5days, then 10. 2,5 = 4,0 t of mixed liquor needs tobe wasted daily. Because this volume constitutes alarge proportion of the volume of the reactor, (thiswould be the case up to about 10 days sludge age)the sludge is best wasted with the aid of a pumpwhich is activated by a timer. The wastage arrange-ment should be such that about 0,3 to 0,5 f isabstracted over a few minutes and repeated everyfew hours. The high rate of abstraction isnecessary to avoid settlement of the sludge fromthe abstraction tube back into the reactor; if thisoccurs the correct sludge age would not beestablished. In setting up the sludge wastagesystem, it is recommended that about 0,5 / lessthan the required volume is withdrawn so that theexact volume can be made up by hand. Also notethat if sludge is abstracted when the influent feed isstopped, the reactor volume will be reduced. Con-sequently it is best not to waste sludge during thenon-feed period.

3. METHOD

3.1 Unit start-up

From start up, operate the unit under constant flow andload conditions for at least 3 sludge ages allowing theunit to reach steady state - steady state is achievedwhen the daily oxygen consumption rate and reactorMLVSS concentration show approximately steadyvalues. When steady state is achieved switch the unitover to daily cyclic square wave loading conditions bydischarging the daily volume of feed over say 12 hoursinstead of 24. (Keep the feed volume and COD concen-tration constant; simply increase the flow rate}. Operatethe unit under square wave conditions for about 2sludge ages. Note that now the oxygen consumptionrate will vary over the day but the MLVSS concentra-tion should not change significantly.

3.2 Readily biodegradable COD measurement

(1) At some time at (east 6 hours after commencingthe feed, measure the oxygen utilization rate (OUR)

A2-1

oo

EST NUMBER

STARTFEED

L4

STOPFEED

SLUDGE AGE 2-5 DAYSTEMPERATURE 2Q.0°CPH= 7.00 S T A R T

FEED

CYCLIC *STEADY YMODELCYCLIC TOTAL

STEAOY TOTAL

0,00 4, 00 8.00 12.00 16.00 20.00 24.00 28.00TIME (HOURS)

FIGURE A2.1: Experimental (and theoretical) oxygen con-sumption rate over a 26 hour period in a single completely mix-ed aerobic activated sludge process at 2,5 days sludge age and20cC under daily cyclic square wave loading conditions 112hours on. 2 to 14 hours and 12 hours off, 14 to 26 hours). Noteprecipitous decrease in the measured rate at feed termination(14hOO) but that the oxygen consumption rate for nitrificationdoes not change (theoretical line}. Experimental (andtheoretical) data for the equivalent steady state process (i.e.receiving the same feed volume of 24 hours) are also shown

(from Dold, Ekama and Marais 1980).

AERATIONSETTLING

TANK

EFFLUENT

SLUDGE RECYCLE

FIGURE A2-2. Schematic layout of process configuration.

A2-2

by raising the DO to about 6 mgO/i , switching offthe air supply, and monitoring the rate of decreaseof DO on a recorder. Measure the OUR every 20 to30 minutes for about 2 hours before the feed is ter-minated.

125 Switch off the feed to The reactor and again raisethe DO to about 6 mgO/1. About 1 to 2 minutesafter the feed has been stopped again measure theOUR. Measure the OUR every 20 to 30 minutes forabout 2 hours after the feed has been terminated.

(3) Restart the feed and run plant for about 4 hoursbefore repeating the measurement to allow theOUR to stabilize to the feeding condition again.

(4) Plot OUR for each period of about 2 hours beforeand after the feed was terminated. Estimate theaverage OUR before and after the feed termination.

The following information is required to calculatethe concentration of readily biodegradable COD in theinfluent feed:

Q - feed flow rate { f /d j over the 3 hourperiod before the feed is terminated

VD - reactor volume (/)

OURb = average OUR before feed termination(mgO/f/h)

OUR,., - average OUR after feed termination(mgO/jf/h)

S,, - total influent COD concentration of feed

AOUR = (OURb-OURat (mgO/f/h)

Then the influent readily biodegradable COD concentra-tion

Sbsi = (AOUR. Vp.24)/(Q. 0,334)

where

24 = number of hours per day

0,334 = general conversion factor for COD tooxygen. This factor states that for everyunit of COD utilized by the micro-organisms, 0,334 units of COD areoxygen consumption and the remaining0,666 units of COD are converted to neworganism mass fsee Chapter 1).

The influent readily biodegradable COD fraction withrespect to the total COD (f1s) is given by

f = 9 / S •

and the influent readily biodegradable COD fractionwith respect to the biodegradable COD (fDs) is given by

— c /c ; i _ f _ f )~ ^bsi' °t i* ' 'us 1UD''

3.3 Example calculation

The following data were obtained from an experimentalunit operated on primary settling tank effluent at theGoudkoppies plant

Q = 50 f /d for period prior to feed termination

Vp = 7,5 i

OURti - 24,1 mgO/f/h

OURa - 20,1 mgO/i /h

St; - 370 mgCOD/f

AOUR -24 ,1 20,1 - 4,0mgO/l /h

Sbs, = (4,0.7,5.241/(50.0,334)

- 47,5 mgCOD/i

Sbs'Stl= 47,5/370 - 0,13mgCOD/mgCOD

fbs = 47,5/(370(1-0,09 0,04)]

= 0,15mgCOD/mgCOD.

4. NOTES

Hi Suggested loading: The COD load on the reactorshould be such that the OUR (excluding nitrifica-tion which may or may not occur at the shortsludge age depending on (jnm) is greater than25 mgO/f/h. This can be achieved if the feed rate,COD feed concentration and reactor volume aresuch that the load is greater than 2 400 mgCOD//reactor/day. Thus, if the feed concentration isabout 350 mgCOD/Z and a reactor volume of 10i isused the feed flow rate Q prior to feed terminationshould be greater than 2400.10/350 = 68,5 t/ti.

(2) Nitrification: The degree of nitrification which takesplace in the unit does not affect the readilybiodegradable COD determination (see FigureA2.1). This is because at the short sludge ages atwhich the determination is done, the oxygenutilization for nitrification is the same before andafter feed termination due to the high ammoniaconcentration in the reactor and effluent. 11 the am-monia concentration in the reactor and effluent be-fore and after feed termination is low (<5 mgN/i),the change in the oxygen utilization rate for nitrifi-cation may lead to errors in the readily biodegrad-able COD determination.

5. REFERENCES AND SUGGESTED READING

OOLD, P.L., EKAMA, G.A. and MARAIS, G.v.R. (1980) A generalmodel for the activated studge process. Prog. Wat. Tech. 1247-77. An understanding of the paper is strongly recommended asbackground information.

A2-3

EKAMA, G.A. and MARAlS. G.v.R. 11979). Lener to the Editor. MARAIS, G.v.R. and EKAMA, G,A. (19761 The activated sludge pro-Ware/- SA 5(1) 57-60. A detailed discussion on the method and cess, Part I - Steady state behaviour. Water SA 2(4) 163-200.possible errors thai can be incurred in measuring the oxygen con- This contains further detail on the operation of the experimentalsumption rate- units.

APPENDIX 3

BRIEF DESCRIPTION OF THE DETERMINATION OF THE MAXIMUMSPECIFIC GROWTH RATE OF THE NITRIFIERS

1. INTRODUCTION

The maximum specific growth rate constant of thenitrifiers ( nn-.) defines the maximum rate at which thenitrifying organisms can grow in the activated sludgeprocess. Although ^n m is a kinetic "constant" in thenitrification theory, its magnitude can differ appreciablybetween different municipal wastewaters, and can varyeven between batches of the same wastewater. Thisbehaviour is so marked that ^nn, should be classified as awastewater characteristic.

The magniiudes of j i ) i m that have been observed inmunicipal wastewaters at 20°C range from 0,30 to 0,65per day. The low values appear to be due to inhibitionby some substance(s) in the wastewater. Thesesubstances are more likely to be present in wastewatershaving significant industrial* contributions; in general,the greater the contribution, the lower the ^,l(r, value.The reduction in f jn m is not the result of toxicity becausea high efficiency of nitrification can be achieved provided the sludge age is sufficiently long - toxicity wouldtend to terminate nitrification irrespective of sludge age.

The magnitude of ^n m has a significant effect onthe minimum sludge age for nitrification Rsrii; the lowerVnm. the longer RSfT1. Clearly, due to the link between thewasteflow composition and ^nrril an experimentalestimate of the ^rirn value is important for efficienttiesign of nitrifying plants. A method for estimating ^ n m

is set out below.

where

2. METHOD AND APPARATUS

2.1 General

The test is performed in a single completely mixed reac-tor at about 6 to 10 days sludge age receiving a constantflow and load but with alternating cycles of anoxic andaerobic periods of 2 to 3 hours each. The selection ofthe sludge age depends on two factors, (i) thetemperature of the wastewater and (ii) whether or notthe ^nni value is expected to be high (0,5/d at 20°C) orlow (0,2/d at 20°C). The higher the temperature theshorter the sludge age, and the lower the expected nm

value the longer the sludge age.The minimum sludge age for nitrification R5m at

T°C is given by (see Chapter 5, Section 4.4)

Rcm = S(/{unmT(1 -f .) " bnT} (A3.1)

= maximum specific growth rate at T°C

'7-201 (A3.2)

Hnm20 = r a t e a t 20°C

briT = endogenous respirat ion rate of thenitrifiers at T°C

T

- 0,04(1,029}(T-201

- temperature in deg C.

(A3.3)

Sf = factor of safety to ensure complete nitrifi-cation, usually 1,25.

f, = fraction of the total time that the process isunder anoxic conditions.

With the aid of Eqs. A3.1 to A3.3, the sludge agerequired for nitrification in the process can be estimatedfor an expected ^nm2o value, and generally for finm2oaround 0,40/d will be found to be between 6 to 10 daysat 20"C for a 2-hourly cycle of anoxic and aerobicperiods (f, = 0,50).

The length of the anoxic and aerobic cycles ischosen such that the concentration of ammonia duringthe aerobic period does not decrease below 2 mgN/LThe reason for this is that if the ammonia concentrationis greater than 2 mgN/ f , the nitrification rate proceedsat the maximum possible and production rate of nitratecan be directly linked to Mnm; if the ammonia concentra-tion is below 2 m g N / l , the nitrate production rate nolonger can be directly linked to Mnrr, because now thelimited supply of ammonia reduces the nitrification rateeven though the ^nm value remains unchanged (seeFigure 5.1 in Chapter 5). If it is found that the ammoniaconcentration reduces to below 2 mgN/ f in too short atime interval, the sludge age should be reduced. Thishas the effect of reducing the concentration of nitrifyingorganisms, with the result that less ammonia can benitrified per unit time.

The experimental data shown in Figure A3.1 weremeasured at 6 days sludge age and 20°C with an anoxicand aerobic cycle of 2 hours each. The data aretabulated in Table A3.1

A schematic representation of the experimentalprocess and some points regarding its operation aregiven in Appendix 2.

'The type of industry referred to here are Those which produce effluents which have high metal ion concentrations e.g. metal finishing industry,plating works, production, tannery wastes, etc.

A3-1

! ? •

- SO

c

o£ tZ EC

iK

I.=>

• 30

s

0

ox

o-

BNBX

P '

UPT RPTE

02 UPT. RATE

EQ- 62 UPT. RATE

-

iii

1 ©1, Q 1

i

1 1 11 1 1 1 1

2DQ

ISO

100

.

50

0

CGD PNC VSSEFfLUEM

*.- ceo 6* ez

0= vss

BhCX hhCX f-net> 1 r * t—\

1 1

1 1

1 1

1 1

1 1

* ' i1 A1 11 1

C 1 ( j 1 Q1 1

1 1

1

—i

TKNtFFLUEKT

THEO- EXP-• = F F N 6 - S " - C

. UOOO

09 14 IB 22Tinf fMCJRS I

14 IBT IfiE HBUtSJ

14 18T l iE lMBURSI

NHEFFLUENT

ThCB. E *P .r KH3 4 . 7 3.S

RHOX BHBX

U9 I « IB11 HE I HOURS 1

. . N I T R R T E RLKRL1NITYTHED- E ' P -

: NB3 5 - 9 7 0

BW8X CN8X fiNQ*

300!

100 .

09 14 IST[nEIHBUR£l

FIGURE A3.V Experimental (and theoretical) data of la) theoxygen consumption rate Ibi reactor MLVSS concentrationand filtered COD concentration and filtered reactor, (c) TKN,(d! ammonia, («) nitrate and (f) Alkalinity concentrationsversus time in a 2 hourly alternating anoxic and aerobic processunder constant f low and load conditions at 20^0 and 6 days

sludge age. (From van Haandel and Marais 1981).

14 IB 22TIMEIHflUHS)

2.2 Experimental unit s tar tup

From start-up, operate the unit under constant flow andload conditions at the selected sludge age under purelyaerobic conditions for about 2 to 3 sludge ages by whichlime nitrification should be taking place.* If necessaryoperate for 1 or 2 sludge ages longer to ensure nitrifica-tion is complete and the nitrifiers are well established inthe process. Only when nitrification is complete shouldthe process be switched over to the alternating aerobic-anoxic phase, say two hours aerobic, two hours anoxic,Even for very low values of ^nrr, (0,20 at 20°C), it isunlikely that nitrification will not take place at sludgeages around 10 days at 20°C with 50% anoxic times.Operate the process under alternating conditions for notless than one sludge age to allow the system to stabilizeto the alternating conditions.

During the stabilization period, measure on a dailybasis the following -(i) influent COD and TKN concentration(ii) reactor MLVSS concentration and pH\\\\) effluent4* TKN, ammonia, nitrate concentrations(ivj effluent** COD concentration.

When these data show approximate stability over aperiod of about one sludge age, the system can be con-sidered to be stable and the analyses for the ^ i n m deter-mination can be commenced.

2.3 Analysis requirement for / jn m measurement

Once the process is operating satisfactorily, an Intensivetesting programme of about 14 to 16 hours is performedon it. The data measured during such a programme arethe folowing:

' A simple Qualitative test to check the presence of nitrate is the Merckoquantr paper test, or to leave a 100 mi (or larger) measuring cylinderwith mixed liquor standing overnight; if upon tapping the cylinder, bubbles are released from the sludge layer, or some sludge has risen to thesurface, nitrification is laking place - evidenced by the nitrogen bubbles released by dertitrification of the nitrate in the sludge layer.

"Grab samples collected from the effluent bucket after a full 24 hour period.

A3-2

TABLE A3.1 EXPERIMENTAL DATA MEASUREC2-HOURLY ALTERNATING

AT 6 DAYS

Process ParametersReactor Volume = 6 iInfluent flowSludge ageTemperatureRecycle ratio

Day's average

Influent (unfilt.)Effluent lunfilt.)Effluent (fill.)

TIME OCR

9h30' 6910h30llh30- 6911h45 6712hl5 7212h30 6913h00 7013h30" 7014h3015h30* 7216h30 6917hOO 7117h30- 7019h30' 70

'start of aerobic'start of anoxic

= 20

IN AANOXIC-AEROBIC PROCESS

SLUDGE AGE AND 20°

/ / d= 6 days= 20- 1:1

VSS

2B22

__

2773

—

2832———

2692

periodjeriod

C

COD

4926254

COD

66

70

—

59—67—

6769

TKN

49,07.16,9

TKN

4,69,2

11,5__

7,7-5,68,88.66,3—6,3

11,3

N H ;

33,53,53,5

N H ,

1.14.06,6__

2,8

0,74,06,62.5—

0,76,6

A lk

245114114

Alk

80115147_

102

70118150109—

75153

C

NO

7,15.E

NC

12,6.0,——8-

14507-

150

3

1

540

2

6555

55

pH

7,87.5-

pH

7,17.27.4-—

7.2—

7.17,27.4—--

-

centration does not decrease below about 2 mgN/ i ,this rate is directly linked to ^nm i.e.

Mim = rUr, Yn/Xn (/ 6} (A3.4)

where

rNn = nitrification rate by the nitrifying organismsin rng NO3-N/f reactor volume/day

Yr. = yield coefficient for the nitrifiers= 0,10 mgVSS/mgNO3-N nitrified

Xn = reactor nitritier VSS concentration(mgVSS/iJ.

The nitrifier sludge concentration is calculated fromthe nitrate concentration generated per litre influent.Note that this concentration is not equal to the nitrateconcenfation in the effluent because during the anoxicperiods nitrate is removed from the system bydenitrification. An estimate of the nitrate generated perlitre influent, Nhg, is found from the following equation:

NnQ = N* ~ N,& - Ns (mgN/i) (A3.5)

(a) grab samples from influent and effluent bucketsover 24 hours period during which testing periodoccurs;(i) unfiltered influent: COD, TKN, NH,, Alk and

PH(ii) unfiltered effluent: COD, TKN, NH3, Alk and

PH(iii) filtered effluent: COD, TKN, NH3, Alk, NO3

and pH.(b) on grab samples from the reactor:

(i) samples at hourly intervals, filtered TKN, NH3,NO,, Alk and pH

(ii) samples at 2- to 3-hourly intervals, MLVSS andfiltered COD.

(c) on reactor every | hour during aerobic periods:(i) oxygen consumption rate.

As an example of the data measured during an in-tensive testing programme, the data plotted in FigureA3.1 are listed in Table A3.1.

3. THEORY AND CALCULATION

The ^n;r. value is calculated from the increase in thenitrate concentration during the aerobic stages of thealternating cycle. From increase in the nitrate concen-tration the nitrification rate of the nitrifying organismsrNn can be calculated, and provided the ammonia con-

whereNng = nitrate generated per litre influent (mgN/f}Nti = influent TKN concentration (mgN/f)Nle - effluent TKN concentration (mgN/i}Ns = nitrogen required for sludge production

{mgN/i influent)

(mgN/i) (A3.6)

Vp

Xv =

QRs

- nitrogen content of the VSS= 0,10 mgN/mgVSS- volume of the reactor (I)= average VSS concentration over testing

period (mgVSS/f)= influent flow rate ( i /d )= sludge age (d).

Knowing Nng, the concentration of nitrifier sludgemass in the reactor is given by

X. =Yn(Nng) R5

<1+bnTRs)Rhr(mgVSS/f) (A3.7)

where

Rhri = nominal retention time of the process= VP/Q (d).

The increase in the nitrate concentration in thereactor during aerobic period r:j, is the net effect of tworeactions (i) the increase in nitrate by nitrifyingorganisms which is the r^ rate required for the Mnmdetermination and (ii) the change in nitrate concentra-tion in the reactor caused by nitrate leaving the systemvia the effluent, i.e. hydraulic effects, r!Jh, i.e.

•"Mr, = rNT + <Nh (A3.8)

Taking a time interval of At = t2 — t-, dur ing theaerobic period where ^ and t2are the beginning and endrespectively of the time interval, and the nitrate concen-trations in the reactor at times t- and t2 are Nn i and Nn 2

m g N / i respectively, then the net rate of nitrate increaseper litre reactor volume

r*r = <Nn2 - N n l ) / l t 2 - t-,1 ( m g N / i / d ) (A3.9}

The rate at which nitrate leaves the reactor per litre

A3-3

reactor volume at a time instant is simply the product ofeffluent flow and the concentration of nitrate at the timeinstant divided by the reactor volume i.e. at time t,:rNhl - Q.N,.;/VP - Nrl/Rhr., Similarly at t2: rNr,2 = Nr,2/Rh_,.Now the rate of nitrate leaving the reactor during thetime interval bounded by \-{ and t2 can be approximatedby the average rates at times ^ and t2. Hence theaverage nitrate loss over the interval At due to hydrauliceffects is given by

rNh = (Nn1 + Nn2)/(2Rhn) (mgN/I/d) IA3.10)

The hydraulic effect causes the rate of nitrategeneration to decrease so that this rate must be addedto the rate of nitrate generation measured from thenitrate concentrations in the reactor as given by Eq.A3.8.

It should be noted that although only the nitrateconcentrations during the aerobic cycles are used forthe determination of jjnrTV the other analyses are usefulto check that the experiment has progressed satisfac-torily. These data can be used to determine and/orcheck constants other than ^nrn; the nitrate concentra-tions during the anoxic cycles can be used to determinethe 2nd rate of denitrification in the primary anoxic reac-tor {K-J, but this determination requires a knowledge ofthe readily biodegradable COD concentration of the in-fluent (fbs). Also the Alkalinity values during the aerobicand anoxic periods can be used to check the ^ r n and K2

determinations. For details see van Haandel and Marais(1981).

When all the data are measured, the constants canbe determined and/or checked with the aid of computersimulation (see Figure A3.1). However, computersimulation is not essential because generally it has beenfound that when the manually calculated values of fjnm

'and K7) are used in the simulation, very good cor-respondence with the experimental data is obtained.

4. EXAMPLE

To demonstrate the calculations, the fjnm value will becalculated from the data listed in Table A3.1.

The nitrate concentrations measured at hourly in-tervals during the aerobic periods are listed in TableA3.2. From these data, the reactor nitrate generationrate rNr and the rate of nitrate loss by hydraulic effectsrNh| are calculated from Eqs. A3.9 and A3.10 respective-ly. The sum of ihe two rates gives the rate of nitrategeneration by the nitrifiers rNr) i.e. Eq. A3.8. From the 4periods of one hour, a mean value of i\.n is 8,45 mgN03-N/litre reactor volume/hour is found.

The average MLVSS concentration during the ex-periment is 2780 mgVSS/f. From this, the nitrogen re-quired for sludge production N is determined from Eq.A3.6 i.e.

TABLE A3.2. CALCULATION OF NITRIFICATION RATESFROM NITRATE CONCENTRATION PROFILES DURINGTHE AEROBIC PERIOD OF AN ALTERNATING ANOXIC-

AEROBIC COMPLETELY MIXED PROCESS UNDERCONSTANT FLOW AND LOAD

Time

11h3C12h3012h3015h3016h30T7h30

0,08,214,60,57,5'5.5

•8,2•6,4

•7,0-8,0

-0,57•1,68

•0,56-1,60

Mean

•8,77•7,98

•7,56-9,60

•8.45

•units mgN0)-N7i reactor volume/h.-i—.-&I hydraulic retention time = €'20 - 0.30d

Ns - 0,10 = 13,9 mgN/f influent flow.

The average effluent TKN concentration (unfiltered) is7,1 mgN/i . Hence from Eq. A3.5, the nitrate generatedper litre influent, Nng, is

Nr,g = 49,0 - 7,1 - 13,9 - 28,0 mgN/ / .

At 6 days sludge age and measured temperature20°C, the nitrifier concentration in the reactor is foundfrom Eq. A3.7

Hence from Eq. A3.4, the maximum specific growthrate of the nitrifiers ^nrri at 20°C is

Mnn, = (8,45.24).0,10/45,2= 0,45 16. at20°C

Hence/jnrn2o = 0,45/d.If in this example, the temperature was not 20°C,

the endogenous respiration rate in Eq. A3.7 for Xnneeds to be adjusted with Eq. A3.3 for the giventemperature. The value of ,,,,. found from Eq. A3.4 israte at the given temperature and to calculate /jnm2o- tn i s

value needs to be adjusted to 20°C with Eq. A3.2.

5. REFERENCES

VAN HAANDEL, A.C. and MABAIS. G.v.R. (1981) Nitrification anddenitrification kinetics in the activated sludge process. ResearchReport W 39, Depi. of Civil Eng., Univ. of Cape Tovvrt

SEHAYEK, E. and MARAiS, G.v.R. (1981) Kinetics of biologicalnitrogen removal in the activated sludge process. Research ReportW 41, Dept. of Civil Eng., Univ. of Cape Town.

MARAIS, G.v.R. and EKAMA, G.G. (1976) The activated sludge pro-cess: Part I - Steady state behaviour. Water SA, 2(4) 163-200.

A3-4

APPENDIX 4

LIST OF RSA BIOLOGICAL NUTRIENT REMOVAL PLANTS

1. Driefontein Water Pollution Control Works,City of Roodepoort

City of RoodepoortPrivate Bag X30ROODEPOORT1725

2. Units 53 and 2/53, Sasol II, Secunda

SASOL IIP 0 Box 1SECUNDA2302

3. Meyerton Sewage Purification Works

Municipal ChemistMeyerton Town CouncilP 0 Box 9MEYERTON1960

4. Goudkoppies Sewage Purification Works,City Council of Johannesburg

Assistant Chief Sewerage EngineerWastewater ReclamationCity Engineer's DepartmentP 0 Box 4323JOHANNESBURG2000

5. Unit 3 of Northern Sewage Purification Works,City Council of Johannesburg

City Engineer's DepartmentSewerage BranchP O Box 4323JOHANNESBURG2000

6. Baviaanspoort Sewage Purification WorksCity Council of Pretoria

City Council of PretoriaP O Box 1409PRETORIA0001

7. Rynfield Purification Works,Town Council of Benoni

Town Council of BenoniPrivate Bag X014BENONI1500

8. Randfontein

Consulting ChemistCity of KrugersdorpP O Box 94KRUGERSDORP1740

9. Standerton Sewage Purification Works

Municipality of StandertonP O Box 66STANDERTON2430

10. Klerksdorp Sewage Purification Works

Town Council of KlerksdorpP O Box 9KLERKSDORP2570

11. Stilfontein Sewage Purification Works

Town Council of StilfonteinP O Box 20STILFONTEIN2550

12. Mitchell's Plain

and

13. Cape Flats Sewage Purification Works

Cape Town City CouncilP O Box 1694CAPE TOWN8000

14. Gammans Sewage Works

City Engineer's DepartmentP O Box 59WINDHOEK9000SOUTH WEST AFRICA

15. Rietspruit Sewage Works,Town of Vanderbijlpark

Town Council of VanderbijlparkP O Box 3VANDERBIJLPARK1900

A4-1

16. Hartebeesfontein Sewage Purification Works,City of Kempton Park

Chief ChemistCity of Kempton ParkP 0 Box 13KEMPTON PARK1620

MONDI TIMBERS, Sabie

Chief EngineerMONDI TIMBERSP O Box 69SABIE1260

18. Delmas Sewage Purification Works

Municipality of DelmasP O Box 6DELMAS2210

19. Pietersburg Sewage Purification Works

Municipality of PietersburgP O Box 111PIETERSBURG0700

20. Middelburg Sewage Purification Works

Municipality of Middelburg. TvlP O Box 14MIDDELBURG, Tvi1050

21. Potchefstroom Sewage Purification Works

Municipality of PotchefstroomP 0 Box 113POTCHEFSTROOM2520

22. Bethal Sewage Purification Works

Municipality of Betha!P O Box 3BETHAL2310

23. Umhlatuzana Wastewater Treatment PlantBorough of Pinetown

Borough of PinetownP 0 Box 49PINETOWN3600

24. Bethlehem Sewage Purification Works

City Engineer's DepartmentMunicipality of BethlehemBETHLEHEM9700

25. Bellville Sewage Purification Works

Municipality of BellvilleP 0 Box 2BELLVILLE7530

26. Brits Sewage Purification Works

Municipality of BritsP 0 Box 16BRITS0250

A4-2

APPENDIX 5

PHOTOMICROGRAPHS OF BACTERIA FOUND IN NUTRIENT REMOVALACTIVATED SLUDGE PROCESSES

(al A methylene-blus-stained specimen of enhanced Premoving sludge. Note the purplecolour of the cluster of cells. The arrow indicates a single cell within the cluster.

(b> A cluster of phosphorus-rich cells in an enhanced Premoving sludge as observed In a scan-ning electron microscope. Dotted outline indicates a single cell.

4-i f

(c) A section of a cluster of phosphorus-rich cells as observedin a transmission electron microscope. The black sphereswere polyphosphate granules and the white areas werewhere the polyphosphate granules had been torn from thecells during specimen preparation. Dotted outline indicatessingle cells.

FIGURE A5.1: Photomicrographs relating to Acinetobacter spp

A5-1

(a) The appearance of a sample of enhanced P-removingsludge in a scanning electron microscope. Framed area in-dicates a cluster of cells in the sludge matrix.

(b) Energy dispersive analysis of X-rays area map ofphosphorus for the same area as (a! above. Concentrationof dots indicated that phoaphorus was associated with thecell cluster-

Ic) Polyphosphate-containing cells as observed in a transmis-sion electron microscope. Dotted outlines indicate singlecells: arrows indicate pofyphosphate inclusions.

&¥• t

FIGURE A5.2: Photomicrographs relating to Ac/netobacter spp

A5-2

(d) Energy dispersive X-ray spectrum for a polyphosphategranule indicating elements associated with the granule.Quantitative analysis indicated that the granules contained>2S % P and >8 % Ca.

(e) The appearance of a specimen of sludge taken from theanaerobic zone of an enhanced-phosphorus-removing ac-tivated sludge plant as observed by light microscopy. Thespecimen was stained with Sudan-Black, a stain used forindicating PHB The black granules indicated PHB inclu-sions.

(f) A transmission electron microscope picture of a sludgespecimen taken from the anaerobic zone of an enhanced-phosphorus-removing activated sludge plant. Thepolyphosphate granules had disintegrated (black arrows)and PHB had appeared (white arrows).

FIGURE A5.2 (Continued)

A5-3

(a) Ught microscopic pictures of bulking activated sludge in-dicating the filsmentous nature of organisms within thesludge.

(b) As in (a) above

FIGURE A5.3. Photomicrographs of filamentous activated sludge

A5-4

Ic) A scanning electron microscope picture of a bulkingsludge.

-rat?

*

(d) A transmission electron microscope picture of a filamen-tous organism in bulking activated sludge. The single cells • f(black arrow) are enclosed within a sheath (white arrow). I

FIGURE A5.3 (Continued)

A5-5

(a) A light n . ^ . ^uu^ i prcture 01 inp protozoan. VorticeJfa, in (c) A section of a Vorticef/a as observed in a transmission elec-i d ld t 0 i o s p e The arrows indicate two phosphorusrichactivated sludge.

t r 0 " microscope. The arrows indicate two phosphorus-richcells within the Vorticella.

\

(b) A scanning electron microscopic picture of Vorticella in (d) A section of an amoeba as observed in a transmission elec-activated sludge. Note the bacteria (arrows) on the Vor- tron microscope. The amoeba had ingested a large numberticeffa. of phosphorus-rich cells. Dotted outline indicates amoeba.

Note the three clumps of cells within the amoeba - thearrow indicates a single cell.

FIGURE AS.4 Photomicrogrspha of various protozoa and metazoa

A5-6

TO ORDER: Contact Publications Telephone No. 012 330 0340Fax Number: 012 331 2565E-mail: [email protected]

W a t e r R e s e a r c h C o m m i s s i o nPrivate Bag X03. Gczina, 0031. South Africa

Tel: +27 12 330 0340. Fax; +27 12 331 2?6?Web: littp: www.wrc.org.za

tt-16-84[1]

Documents

ekama mr

buchan mr

lombard mr

mr i p siebritz mr

water resources

j hu5selmann mr w

coordinating research

water bodies