Membrane bioreactors in industrial wastewatertreatment – European experiences, examples andtrends

P. Cornel and S. Krause

Technische Universitat Darmstadt, Institut WAR, Petersenstrasse 13, 64287 Darmstadt, Germany

(E-mail: p.cornel@iwar.tu-darmstadt.de)

Abstract In wastewater treatment, micro- and ultra-filtration membranes are used for the separation of the

activated sludge (biomass) from the treated water. This offers the advantages of a complete removal of

solids and bacteria, as well as most of the viruses, namely those attached to the suspended solids.

Compared to the conventional activated sludge process (CAS) this technology allows a much higher

biomass concentration (MLSS) whereby the reactor volume and the footprint decreases. With increasing

MLSS, the viscosity of the sludge increases, which leads to reduced oxygen transfer rates. Depending on

the type of membrane and membrane module, the pre-treatment has to be more sophisticated to prevent

clogging and sludging of the modules.

Due to fouling and scaling, the flux through the membranes will decrease with time. The decrease

depends on the water quality as well as on the measurements taken to minimize fouling. Mainly, three

strategies are available: lowering the flux, increasing the “crossflow” and cleaning of the membranes.

Different strategies including backwash and chemical cleaning “in situ”, “on air” and “ex situ” can be

applied. It has been proven more effective to apply preventive regular cleaning.

Besides the energy demand for oxygen supply – which is typically in the range of 0.3 kWh/m3 for

municipal wastewater – the energy for fouling prevention is substantial. Immersed membranes need

approximately 0.4 to 1 kWh/m3 for the coarse bubble aeration, whereas tubular modules require 1 to

4 kWh/m3 pump energy.

For proper design of industrial wastewater treatment, the verification of applicability and the development

of adequate cleaning strategies, it is a precondition to run pilot tests for a sufficient period of time with the

wastewater to be treated. More than 100 industrial wastewater treatment membrane bioreactors (MBR) are

in operation in Europe. Data of three case studies for a sewage sludge dewatering plant in UK

(12,000 m3/d), a plant for the treatment of pharmaceutical wastewater in Germany (3,600 m3/d), as well for

revamping of an chemical WWTP .2,000 m3/d in Italy, are given.

MBRs will be used in future wherever high quality effluent is required, because of a sensitive receiving

water body or due to the fact of water reuse as process water. MBRs are a perfect pre-treatment in

industrial applications when further treatment with nanofiltration or reverse osmosis is considered.

The technique is advanced and can be applied both in municipal and industrial wastewater treatment.

Higher operational costs must be balanced by superior effluent quality.

Keywords MBR; design; pre-treatment; oxygen transfer; industrial applications

Introduction

Membrane bioreactors (MBRs) are a further development of the conventional activated

sludge process (ASP), where the secondary clarifier is replaced by a membrane filtration.

Membranes with pore sizes of 0.1–0.4mm are typically used. One advantage of this pro-

cess is the complete removal of suspended solids (SS), including bacteria. However, the

pore sizes of the membranes are too large to separate single molecules or ions. As the

MLSS (mixed liquor suspended solids) is independent of the sedimentation behaviour of

the sludge, it can be increased significantly. Typical MLSS values are in the range

of 12–15 g/L at immersed MBRs up to 30 g/L in tubular systems. By the loss of the

Water

Science

&T

echnolo

gyVol53No3pp37–44Q

IWA

Pub

lishing2

00

6

37doi: 10.2166/wst.2006.074

secondary clarifier and the increase of MLSS in the aeration tank, the footprint of the

treatment plant is significantly smaller.

Two main configurations of MBRs can be established. At the so called immersed or

submerged membranes, the fouling control is achieved by an air scour at the membrane

surface by an additional coarse bubble “crossflow” aeration. The necessary shear velocity

is caused by the movement of the bubbles close to the membrane surface. This kind of

membrane is applied in municipal and industrial wastewater treatment and can be placed

either inside the aeration tank or in an external filtration tank. At tubular side stream

membranes, the fouling control is achieved by a high water velocity across the filtration

channel. This configuration is used for industrial wastewater treatment only. All three

applied configurations are shown in Figure 1.

The aim of the paper is to share the European experiences with the MBR process, to

elaborate the differences to the conventional activated sludge process and to present a

possible technical solution for the design of the MBR process. Three case studies chosen

from more than 100 industrial applications show the broad operative range. Perspectives

and further needs round off the paper.

Design of MBRs

As the MBR is an activated sludge process, the same generally accepted design regu-

lations for the conventional ASP can be applied for MBR design. The F/M-ratio is one,

or the key, design parameter and as high MLSS can be achieved the resulting tank

volumes are smaller. The aeration equipment has to be adapted to the resulting high

specific volumetric oxygen rates.

The hydraulic load and the achievable flux are the key parameters for the design of

the membrane surface, whereby the membranes have to permeate the maximal flow. As

membranes are still a relevant expense factor, strategies to avoid or cut peak flows might

pay off.

For the design of the configuration, maintenance and membrane cleaning facilities, it

is important to specify the type of membrane and membrane module as early as possible.

Overall, in MBRs the automation level is higher compared to conventional wastewater

treatment plants due to back flush, cleaning procedures, etc.

Pre-treatment

The wastewater needs to be carefully pre-treated before entering the MBR plant and

there must not be any emergency bypass, so that only pre treated wastewater will enter

Figure 1 Options of membrane configurations

P.C

orneland

S.K

rause

38

the MBR. It is advisable to remove abrasive or sharp edged materials which can damage

the membranes, as well as fibers or hairs which can clog the membrane (modules), and

lead to a dramatic and rapid decrease of the flux. Screens, or even better sieves, with

mesh sizes ,0.5mm have proved suitable (ATV-DVWK, 2002). Further, a grease trap

should be installed, because oil and grease may influence the flux of the membranes

negatively. The hydraulic equalisation is of importance, because the costly membrane

surface has to be designed according to the maximum inflow. Thus, a storage, and if

necessary an equalisation tank to cut hydraulic peaks, can be advisable.

Aeration tank

General. The use of membranes to separate the biomass leads to some changes in the

design and operation of the plant. As already mentioned, the MLSS in MBRs can usually

be raised to approximately 10 to 15 g/L for immersed membranes and up to 30 g/L for

tubular side stream membranes.

The aeration tank design is (as in the conventional ASP process) based on the load

(F/M-ratio). Higher MLSS and similar F/M as in conventional activated sludge plants

result in smaller aeration tank volumes. However, a minimum hydraulic retention time

(HRT) should be observed, at least 4 h in municipal MBR according to Engelhardt

(2003).

Recirculation. The filtration leads, locally, to an increase of the sludge concentration

whereby the sludge needs to be returned to the aeration tank. In case of configuring the

membranes inside the aeration tank, the mixing due to the oxygen input and the fouling

control by coarse bubbles is usually sufficient. If the membranes are located outside of

the aeration tank, in general, a recycle rate of 4 to 5 is sufficient.

Sludge production. Our experiences with a municipal MBR show that the surplus

sludge production is in the same range as conventional ASP plants. Furthermore, it can

be shown that MBR can be simulated by the activated sludge model ASM 1 (Henze et al.,

1987) and ASM 3 (Gujer et al., 1999), both by IWA. Thus, in principal, the

biodegradation of organic compounds does not differ from conventional plants.

Sludge characteristics and oxygen transfer. The sludge characteristic differs from

conventional activated sludge, mainly due to the higher MLSS. The sludge viscosity

increases with increasing MLSS. The viscosity of the MBR sludge is non-Newtonian, i.e.

it decreases at higher shear stress (Cornel et al., 2003). The higher viscosity may lead to

a lower a-value ( ¼ ratio of the aeration coefficient kLa under process condition to the

clean water aeration coefficient) which is approximately 0.5 ^ 0.1 at MLSS content of

approximately 12 g/L. In Figure 2, a-values of two municipal full scale MBRs are

shown. It is obvious that the a-value decreases at increasing MLSS.

Membranes. The design of the membranes surface area is important for economic

efficiency. The flux depends on the membranes, the modules, the transmembrane

pressure, the wastewater composition and on fouling/scaling. For the design, the net flux

is the important parameter which characterises the overall flow rate including breaks and

back flushes. For industrial wastewater in general, pilot tests have to be performed. The

resulting flux is often as low as 8–15 (20) L/(m2 h) for immersed membranes and up to

120L/(m2 h) for tubular membranes.

The design has to consider that the initial flux will not stay constant. Typically, the

flux will decrease with time at constant pressure, respectively, the pressure difference has

P.C

orneland

S.K

rause

39

to be increased to keep the flux at a constant level. These phenomena are a result of foul-

ing or scaling, hence the accumulation of organics, colloids, particles and precipitates on

and in the membranes.

Fouling can be controlled by reducing the flux, which results in higher membrane sur-

face area and higher energy demand, by increasing the crossflow resulting in higher

energy demand again and/or regular cleaning of the membranes (Judd, 2004).

Membrane cleaning

The membranes require regular cleaning to remove membrane fouling and keep the flux/-

pressure (or permeability ¼ flux divided by the pressure in L/(m2 h bar)) loss in a given

range. Overall, different cleaning procedures have been established and will be described

in the following (see also Figure 3):

† backwash with permeate: “in situ”, automatically every few minutes mainly; hollow

fibre membranes only

† chemical enhanced backwash: “in situ” or “on air”; chemicals such as acids or oxidiz-

ing agents; automatic control mode, e.g. daily

† maintenance cleaning: “in situ”, “ex situ” or “on air”; chemicals such as NaOCl, citric

acid (0.5%) etc.; e.g. weekly

† intensive cleaning: “ex situ”, e.g. different chemicals, 35 8C, 1–2 times per year

Typical chemicals used for cleaning are acids such as nitric, sulphuric, hydrochloric

acids or weak acids such as citric acid to remove scaling, and oxidizing agents such as

Figure 3 Cleaning specific terms for immersed membranes

Figure 2 a-values in dependence of MLSS for municipal full-scale MBRs

P.C

orneland

S.K

rause

40

sodium hypochlorite or peroxide, eventually in combination with caustic soda, to adjust

the pH. Frequency, as well as type and concentration of chemicals, depend strongly on

wastewater composition, membrane and module type and are not standardised so far. In

contrast, cleaning strategies are a focus of research with regard to avoiding the use of

chlorinated products (AOX-formation) and to reduce the so-called “aging” of membranes

caused by the use of oxidizing chemicals which corrode the membranes. Some commer-

cial agents with unknown formula are also offered.

Energy demand

The energy demand of MBRs is higher compared to conventional AS systems, mainly

due to the additional energy demand for fouling control. Immersed membranes require

approximately 0.4–1 kWh/m3 for coarse bubble aeration, the energy demand for tubular

side stream membranes ranges from 1 to 4 kWh/m3 (ATV-DVWK, 2002). Air cycling

(Zenon), or stacking of the membrane modules (Kubota and others), are attempts to use

the coarse bubble energy more efficiently (Figure 4). The energy consumption for main-

taining the underpressure is almost neglectable with values around 0.01 kWh/m3 depend-

ing on the trans membrane pressure. The energy demand for oxygen supply depends on

the COD load and on the MLSS, as the a-value is affected (see above). Comparing

an MBR with MLSS of 12 g/L and thus a a of 0.5 with a conventional system with an

a-value of 0.7, the aeration energy is by the factor of 1.4 (0.7/0.5 ¼ 1.4) higher. For

municipal stabilization plants with a typical specific aeration energy demand of 0.3 to

0.4 kWh/m3 thus by 0.12 to 0.16 kWh/m3 higher. Due to high salt content and a wide span

of COD concentrations, a value for industrial wastewater applications is not possible.

Experience with MBR

In Europe, more than 30 municipal MBRs are in operation, most of them in the UK (20

MBRs) and Germany (8) (Cornel and Krause, 2003). The market is dominated by two

suppliers Kubota (Japan, flat plate membrane) and Zenon (Canada, hollowfiber mem-

brane). The largest operated plants are in Brescia (I) (38,000m3/d) and in Kaarst (D)

(80,000 p.e./ ,45,000m3/d). There is little reliable information about the lifetime of

membranes. Although some have been in operation for 5 to 6 years without failures with

annual replacements of less than 3% (Churchhouse and Brindle, 2003), others had to be

replaced after 2 to 3 years because of serious fouling or even mechanical destruction.

Table 1 lists industrial MBRs in Europe resulting from a questionnaire dated February

2002. Meanwhile, although some plants could be added, the table gives an overview of

different applications. The leachate MBRs are dominated by the tubular system, whereas

all other branches are operating mainly immersed MBRs. The membrane flux range form

Figure 4 Air-cycling process for immersed membranes (Zenon)

P.C

orneland

S.K

rause

41

as low as 8 L/(m2 h) with submerged hollow fiber modules in the chemical industry up to

100 L/(m2 h) with tubular systems for leachate treatment. Therefore, it is advisable to

pilot industrial MBRs.

Case studies of industrial wastewater treatment

Some data may illustrate the status and degree of development industrial MBRs have

reached.

Case A

Treatment of wastewater of a central sewage sludge dewatering plant in the UK (effluent

of sludge dewatering centrifuges and sludge dryers).

Design parameters:

† four parallel operated units (denitrification volume 2,300m3; nitrification tank

9,400m3) with eight separate tanks to operate Kubota flat membrane units (immersed

inside aeration tank)

† Flow: 12,000m3/d

† installed membrane surface: 20,500m2 (Kubota flat membranes)

† max. flux: 24L/(m2 h)

† MLSS is 20,000mg/L

† chemical cleaning: twice per year

Table 2 contains the average values of some key parameters.

Case B

Treatment of chemical/pharmaceutical wastes; high effluent requirements for suspended

solids and P because of a very small receiving water body

Table 1 MBRs in industrial wastewater treatment (ATV-DVWK, 2002)

Industry sector Numbers Flux (m3/d)

Automobile 1 225Chemical industry 15 70–1,360Leachate 48 10–18,000Food industry 9 100–1,840Tannery 5 40–800Composting 2 40–50Cosmetics 3 120–680Malthouse 2 100Paper 1 900Pharmaceutical industry 15 50–1,500Ships/cruisers 15 10–740Tank cleaning 3 200Textile industry 5 10–1,440Rendering plants 4 40–960

Table 2 Average values of case A (Klegraf, 2004)

Design in Actual value in Actual value out Value guaranteed

Flow, m3/d 12,000COD, mg/L 2,500 3,000 45BOD, mg/L 1,300 – ,3 10NH4-N, mg/L 167 210 0.12 3PO4-P, mg/L 30 – 0.5 2Susp. solids, mg/L 300 670 – 10

P.C

orneland

S.K

rause

42

Design parameters:

† three activated sludge units followed by four Zenon hollow fibre membrane filtration

units (immersed external)

† Installed membrane surface 15,840m2

† average flux 12.5 L/(m2 h), max. peak 17 L/(m2 h)

† MLSS is 12,000mg/L

Table 3 lists the average values of some key parameters

Case C

† Revamping of an existing wastewater treatment plant in Italy

† Low footprint requirements; expansion of capacity on given space

† Non/bad flocculating sludge

Design parameters (Brockmann, 2004)

† three parallel operated units with Zenon hollow fibre membranes (8,500m2 in total) in

external filtration units

† aeration with pure oxygen but crossflow generation with compressed air

† Flow 3,000m3/d at max. of 100m3/h

† COD 6,000mg/L

† Vaeration tank 2,000m3

† MLSS 12,000mg/L

† DCOD .97%

Performance

† Requirements of 160mg/l COD, 40mg/l BOD and 15mg/l N fulfilled

† Foaming reduced by anti foaming chemicals

† Scaling leads to pressure drop increase; regular maintenance cleaning once per weak

with citric acid

Perspectives of MBRs

The MBR technology is ready to take a major role in wastewater treatment. The experi-

ence in recent years shows that this technique is “adult” and can be applied both in

municipal and industrial wastewater treatment. The performance of a successful design

and operation is possible.

As a result, MBRs are of interest wherever high quality effluent is required, e.g.

because of water reuse as process water in industry. MBRs are a perfect pre-treatment in

industrial applications when further treatment with nanofiltration or reverse osmosis is

considered.

The future of MBRs is also in the expansion of the capacity. Presently, the world’s

biggest municipal wastewater treatment MBR is some 1,900m3/h (MBR Nordkanal,

Erftverband, Germany). The prices for MBRs are decreasing and with it the potential

Table 3 Average values of case B

Design in Actual value in Actual value out Value guaranteed

Flow 3,600 m3/d;200 m3/h

3,500 m3/d

COD, mg/L 4,000 3,500 (2,000–4,500) 300 (250–400) 500 or .90%removal

NH4-N, mg/L 130 100 ,1N total, inorg. mg/L 20–30 50PO4-P, mg/L 38 ,10 (5–30) ,1 2Calcium, mg/L ,1,300 700–900 – –

P.C

orneland

S.K

rause

43

market is growing (e.g. Judd, 2005). The further development and research of membrane

bioreactors will be in:

† the improvement of cleaning strategies, especially in the application of safe chemicals

for the water body

† the optimisation of the energy demand, improvement of the fouling control (maximum

utilisation of the used air)

† membranes

W longer lifetime

W high and stable flux and high permeability

W low cost

W resistant against mechanical forces.

ReferencesATV-DVWK (2002). Aufbereitung von Industrieabwasser und Prozesswasser mit Membranverfahren und

Membranbelebungsverfahren, KA – Wasserwirtschaft, Abwasser, Abfall (49) No. 10 (part 1) and No. 11

(part 2) (also available at www.dwa.de).

Brockmann, M. (2004). Membranbelebungsverfahren zur Abwasserreinigung in der chemischen und

pharmazeutischen Industrie. In ATV-DVWK und DVGW Membrantage, June 2004, Kassel.

Churchhouse, S. and Brindle, K. (2003). Long term operating experiences of membrane bioreactors. In

Proceedings of the International Conference on Membrane Bioreactors MBR4; Cranfield University,

April 2003.

Cornel, P., Wagner, M. and Krause, S. (2003). Investigation of oxygen transfer rates in full scale membrane

bioreactors. Water Science and Technology, 47(11), 313–319.

Cornel, P. and Krause, S. (2003a). State of the art on MBRs in Europe. Proceedings Applications and

Perspectives of MBR in Wastewater Treatment and Reuse, 28–29 April 2003 – Cremona, Italy.

Engelhardt, N. (2003). Membranbelebungsverfahren – eine beherrschbare und erfolgreiche Technik –

Erfahrungen nach vierjahrigem Betrieb; Membrantechnik 5.

Gujer, W., et al. (1999). Activated Sludge Model No. 3. Water Sci. Technol., 39(1), 183–193.

Henze, M., et al. (1987). Activated Sludge Model No 1. IAWPRC, London, Scientific and technical report

No. 1.

Judd, S. (2004). Submerged membrane bioreactors: a matter of control. In IWA Yearbook 2004.

IWA Publishing, London.

Judd, S. (2005). MBR5 Short Course, Cranfield University, 18–19th July.

Klegraf, F. (2004). Membranverfahren zur Behandlung und Wiederverwendung von Industrieabwasser.

In ATV-DVWK und DVGW Membrantage, June 2004, Kassel.

Author’s Note

This paper is based on Cornel, P. and Krause, S. 2004. Membrane bioreactors; design,

experience and perspectives; Proceedings Research and innovation in wastewater treat-

ment, Workshop at Politecnico Milano, Italy, 25/26 Nov. 2004.

P.C

orneland

S.K

rause

44