triennial activity report 2000 - 2002 s-l. jämsä...

Helsinki University of Technology Laboratory of Process Control and Automation

Helsinki 2002

TRIENNIAL ACTIVITY REPORT 2000 - 2002

S-L. Jämsä-Jounela M-L. Viherlaakso J. Kämpe T. Jokinen

TRIENNIAL ACTIVITY REPORT 2000 - 2002

Helsinki University of Technology publicationsHelsinki 2002

S-L. Jämsä-Jounela M-L. Viherlaakso J. Kämpe T.Jokinen

DEPARTMENT OF CHEMICAL TECHNOLOGY LABORATORY OF PROCESS CONTROL AND AUTOMATION4

Distribution:

Helsinki University of Technology

Laboratory of Process Control and Automation

P.O. Box 6100

FIN - 02015 HUT

Phone: +358-9-451 3852

Fax: +358-9-451 3854

Email: [email protected]

© Sirkka-Liisa Jämsä-Jounela

ISBN 951 - 22 - 6067 - 0

ISSN 1455 - 4046

Printing: Erweko Oy

Helsinki 2002

Cover pictures courtesy of Stora Enso and Rautaruukki

DEPARTMENT OF CHEMICAL TECHNOLOGY LABORATORY OF PROCESS CONTROL AND AUTOMATION 5

CONTENTS

PREFACE ........................................................................................................................7

1 INTRODUCTION ...........................................................................................................8

2 PERSONNEL ................................................................................................................92.1 Teaching staff .........................................................................................................92.2 Administrative staff ................................................................................................92.3 Research staff ........................................................................................................ 9

3 RESEARCH ACTIVITIES ........................................................................................... 10

4 STRUCTURE OF THE MASTER’S DEGREE ............................................................... 19

5 TEACHING ACTIVITIES ............................................................................................ 205.1 Undergraduate studies .......................................................................................... 205.2 Post-graduate courses .......................................................................................... 23

6 HARDWARE AND SOFTWARE FACILITIES ............................................................. 256.1 Laboratory Equipment ......................................................................................... 256.2 Automation Systems............................................................................................. 256.3 Computers and related hardware .......................................................................... 266.4 Software .............................................................................................................. 26

7 THESIS....................................................................................................................... 287.1 Thesis abstracts for the Licentiate In Technology................................................... 307.2 Thesis abstracts for the Master of ScienceIn Technology ....................................... 31

8 INTERNATIONAL ACTIVITIES ................................................................................. 458.1 Invited lectures, seminars and technical visits by the academic staff ...................... 498.2 Other activities ..................................................................................................... 52

9 PUBLICATIONS ......................................................................................................... 539.1 Books .................................................................................................................. 539.2 Articles in Scientific Journals ............................................................................... 539.3 Articles in Conferecne proceedings (refereed) ....................................................... 539.4 Other publications ................................................................................................ 55

10 MAJOR STUDENTS.................................................................................................. 58

11 CONTACT INFORMATION ........................................................................................ 59

6 DEPARTMENT OF CHEMICAL TECHNOLOGY LABORATORY OF PROCESS CONTROL AND AUTOMATION

7DEPARTMENT OF CHEMICAL TECHNOLOGY LABORATORY OF PROCESS CONTROL AND AUTOMATION

PREFACE

During the past twenty years the field of process control has undergone aconsiderable transformation. Two decades ago, the typical control-engineeringgraduate had a course in feedback control theory, and those interested in acareer in this field secured a position in the process industries. During thisperiod, however, the number of new control engineering positions in theprocess industry has declined while, at the same time, increasing emphasis hasbeen placed on positions requiring a multidisciplinary knowledge of processengineering, automation and information technology. As a result, controltheory has played a very minor role.

Due to this multidisciplinary requirement for knowledge and the latest developments in the differentfields, especially in Information Technology, the need has also arisen for a change in the educationand courses offered at universities. In order to fulfil the demands of industry, the students will have tobe broadly educated to cope with cross-disciplinary applications and the rapidly changing technology.

The Laboratory of Process Control and Automation at the Helsinki University of Technology (HUT)provides education in process automation mainly in the fields of Chemical Technology, Forest ProductsTechnology and Materials Science and Rock Engineering. Process automation is a special program,crossing departmental lines, that enables students in different areas such as chemical, forest producttechnology and material science engineering to obtain the interdisciplinary training necessary forwork on process automation and its applications.

In this activity report the structure of the master’s degree is presented, followed by the laboratory-industry interaction, and the new applications of teaching and education tools that the laboratory isutilizing to meet the challenges set by these new requirements. Finally, the results of the programevaluations are reported and briefly discussed. An evaluation was conducted to assure the quality ofcourses and outcomes of the programme in summer 2002.Web-based questionnaires were distributedto 100 M.Sc. engineers that graduated from the laboratory.

Generally the feedback from the students has been both positive and encouraging. The educationof students has benefited through the involvement of industry’s in the course design, thus ensuring itsrelevance to future careers. Employment rate was 100%. Based on these results and evaluations by theexternal experts Finland is experiencing a shortage of IT-skilled process engineers. Five years’experience of operating this programme has indicated that the laboratory strategy has been rightselected.

Finally, I would like to thank all our co-operation partners and supporters. Special thanks are dueto the industry, TEKES-Technology Agency of Finland and the Academy of Finland. The outsidesupport has helped us to achieve these successful records in both the education and research activities.

Otaniemi 9th December, 2002

Sirkka-Liisa Jämsä-Jounela


Helsinki University of Technology (founded 1908) is the largest and oldest institute of engineeringeducation in Finland. The University campus is situated about 10 km west of Helsinki in the Otaniemiarea, in the township of Espoo. The number of undergraduates is about 12 000 and that of postgraduatesabout 2500. The basic degree offered is the master’s degree in engineering or architecture (diplomaengineer or architect, minimum 4.5 a). Postgraduate degrees offered are Licenciate of Technology (minimum 2,5 a ) and Doctor of Technology ( minimum 4 a). The University has fourteen departments.The Department of Chemical Technology comprises eight laboratories ( = eight full professorships):Biochemistry and Microbiology, Bioprocess Technology, Chemical Engineering and Plant Design,Inorganic and Analytical Chemistry, Physical Chemistry and Electrochemistry, Industrial Chemistry,Polymer Technology and Process Control and Automation.

The Laboratory of Process control and Automation at the Helsinki University of Technology hasbeen founded in 1987. The laboratory is part of the Department of Chemical Technology.The aim of the laboratory is to serve all fields of process technology -chemical, metallurgical andforest products -in terms of process automation.

The laboratory gives courses in process modelling, simulation, control, optimisation and automationas well as in production control. The undergraduate students typically have a few years background ofprocess engineering studies, after which a one-to-two -year portion of automation studies.

This report describes briefly the staff, facilities and research activities of the laboratory of processControl and Automation during the years 2000 - 2002.

1 INTRODUCTION

The employment of the students after graduation.


2 PERSONNEL

2.1 Teaching staff:

Jämsä- Jounela Sirkka-Liisa D.Sc (Tech.), Professor,Head of the Laboratory

Anna Soffia Hauksdóttir Docent, ProfessorMikko Vermasvuori M.Sc (Tech), Senior assistantTimo Malmi M.Sc (Tech), Assistant

2.2 Administrative staff

Kämpe Jerri M.Sc (Tech), Laboratory ManagerViherlaakso Marja-Leena SecretaryHolmqvist Iris Project Secretary

2.3 Research staff (1.12.2002):

Järvensivu MikaKomulainen TiinaPoikonen RistoSuhonen JoriJokinen TiinaKesti Anna RiikkaLiikala TeemuPikkusaari-Saikkonen JonnaRyynänen TimoSuontaka VilleVatanski NikolaiAdhikari Yuba Raj

(2000-2002):

Bergman SamuliDietrich MaijaEndén PetriHalmevaara KalleHarri IlkkaKinnunen MarjutKuitunen TatuKämpjärvi PetteriLaitinen JariLipiäinen JouniMatinaho SamiMelin KristianNevalainen SusannaRemes AnttiTiili OtsoTimperi Juha


3 RESEARCH ACTIVITIES

The main areas of research in the laboratory areprocess modelling and control, supervisory &artificial control systems, and fault diagnosis. Thehead of the research projects is Professor Sirkka-LiisaJämsä-Jounela.

INTELLIGENT CONTROL OF THE LIMEKILN PROCESS WITH RESPECT TOENVIRONMENTAL REQUIREMENTSJärvensivu Mika

Further reducing environmental impacts such asthe reduced-sulfur emissions will be among themajor challenges facing the pulp and paperindustry in the near future. It will not be easy tofurther decrease the emissions at modern pulpmills because all the major emission sources havealready been eliminated. New strategies, such asthe prevention of emissions at their source, e.g.by means of improved control of the subsequentprocesses, will therefore undoubtedly be requiredin order to conform with the present and alsofuture environmental requirements. An increasein authorities and also public attention andawareness on environmental issues together withintensifying interest in the artificial intelligence(AI) and intelligent systems were also primemotivator for this thesis work.

The primary objective of the research, whichhas been carried out as a co-operative effortbetween academic and industrial parties, hasbeen to lower of the total reduced-sulfur (TRS)emissions from a pulp mill by means of intelligentcontrol techniques. The research was focused onthe lime reburning process, which is one of themain sources of the TRS emissions at modernpulp mills. In addition, the environmentalrequirements for lime kilns have become tighterand even at well-managed mills, the emissionstend periodically to exceed the limits set by theauthorities. It has also been widely recognizedthat control of the rotary kiln used for limecalcination is, in many respects, a demandingtask. So far, most of the kilns have therefore beenoperated without supervisory-level control system.However, there are outstanding economical andthe environmental improvement potentialsassociated with improved control. Hence,supervisory-level control of the lime reburning

process is undoubtedly a prospective applicationfor intelligent control techniques.

In the first phase of the research, acomprehensive study of the operation of the limereburning process was carried out at one of themajor Finnish pulp mills, with special attentionpaid to the factors affecting the TRS emissions.The results showed that, in addition to theconsiderable enhancement potential in theperformance of the kiln process operation,improved kiln control is also a feasible means toreduce emissions. An overall supervisory-levelcontrol schema that takes into account both theenvironmental and operational requirements, wasthen designed on the basis of the results of thestudy. The supervisory-level control system,embedded with a certain degree of intelligence,was then incrementally developed andimplemented at the pulp mill. The controlstructure combines both feedforward (FF) controlmodels and supervisory-level feedback (FB)controllers that are based on the linguisticequation (LE) approach, strengthened withcertain capabilities for adaptation and constrainthandling. Advanced capabilities and highlydeveloped functionality of the control systemwere achieved by combining information fromdifferent knowledge sources, and by usingappropriate techniques to solve each of therecognized problems. On the other hand, thecomplexity of the lime reburning process washandled by implementing a modular systemstructure and by t aking advant age of anincremental system development approach.


The results obtained during extended testingperiods of the system demonstrate that theproposed control schema can be successfullyrealized in an industrial environment, and thatit also provides quantifiable benefits in both theeconomical and ecological respect. The majorbenefit from the ecological point of the view wasan almost 30 % decrease in the mean of the TRSemissions and a considerable reduction, about 90%, in the proportion of high emissions periods.The main verified economical benefits were anincrease of about 5 % in the long-term productioncapacity. Improvements in reburned lime qualityand enhancements in the energy efficiency werealso obtained compared to the situation duringmanual operation.

USE OF HYBRID AND DYNAMICSIMULATION MODELS THROUGH THELIFE-CYCLE OF A PAPER MACHINELaukkanen Ismo

In this project a mathematical model of basepaper production process was developed. Thedynamics of stock and water systems of a modernpaper mill was studied using a mechanistic hybridsimulation model. The combined response of theprocess equipment and the control system underthe transient operational conditions wereinvestigated; both major process equipment andpaper mill control structures are included in thesimulation model. The effects of the transientoperational situations on the water usage of thepaper mill were examined.

The simulator was developed during the papermill design project. The simulator was used forthe verification of process and control design,operator training and process analysis studies. Allapplications were performed before start-up of theplant, which was possible because mechanistichigh-fidelity simulation models were used. Themodels were incorporated into a simulationprogram called APMS, which can be used toanalyze the dynamics of a papermaking process.

On a basis of the research the extended life-cycle concept for using hybrid dynamic simulatorthrough the life-cycle of paper machine wasproposed. Optimally the life cycle of dynamicsimulator starts at the design phase of the papermill and follows the life cycle of the mill. Thesame model can be used in the followingapplication areas: process and engineering,operator training, research and development, andpaper mill.

Dynamic process simulation, based on highfidelity mechanistic models, turned out to be apowerful tool for verification of process andautomation design as well as for operator training.In future, it provides a strong basis for the virtualpaper mill and product developmentenvironments as well as for advanced multimedia-training systems.

FAULT DIAGNOSIS SYSTEM FOR THEOUTOKUMPU FLASH SMELTINGFROCESSVapaavuori Eija, Salmi Tomi, Grönbärj Marko,Haavisto Sasa, Endén Petri, Vermasvuori Mikko

The aim of this project was to develop a faultdiagnosis program for the Outokumpu copperflash smelting process using Kohonen Self-Organizing Maps in conjunction with heuristicsrules.The purpose of the fault diagnosis system was todetect abnormal process states and inform theplant operator accordingly. A process deviationfrom the normal operating range is often causedby equipment malfunctions or an incorrectcontrol strategy. Early detection of undesirableprocess states enables the execution of correctingactions, thus minimising the damage caused byprocess malfunctions.

The implemented fault diagnosis systemconsists of a process interface, data preprocessing,data controllers, and symptom generation objectsand rules. The inferencing results are stored in adatabase, which can be used to perform astatistical analysis of the faults. The structure ofthe system is presented in the following picture.


The monitoring and classif ication areperformed by using Self-Organizing Maps(SOM). The SOM is a neural network algorithmwhich is used to form a neural network model ofan unknown system based only on the datareceived from the system. The nature of thephenomena monitored with SOMs differ, someof them being closely linked with process controland process disturbances, while the rest monitorthe states of the process. One aim was to train aSOM for the quality of the feed material, whichis the most important phenomenon that affectsprocess control in copper flash smelting process.

The overall state of the process was monitoredin order to detect undesired process statesbecause a deviation from the desired state cannotalways be considered as a direct processdisturbance. From an economic point –of –view,it is just as important to detect a decrease in thequality of the product as to detect an equipmentfailure; both lead to a reduction in productivityand profitability. The monitored process stateswere viscosity of the slag and the matte,temperature of the waste heat boiler, and thesooting, i.e. cleaning, of the boiler or thegooseneck.

The following actual process disturbanceswere searched for: flooding of the feed material,aggregation of feed material in the concentrateburner, formation of dust aggregations inside theboiler or the gooseneck, and malfunctions of thefields of the electrostatic precipitator and gasblower. The most important part of a rule-basedfault diagnosis system is the rules that are applied.With well functioning rules a very effectivesystem can be built while, on the other hand, aprogram cannot predict phenomena correctly ifits rules are false or feeble.

In the system the rules are always connectedto a specific piece of equipment, and can beformed using only those variables that are knownfor that equipment. This prevents the formationof illogical rules, that may be the reason whysystems show unexpected behaviour. Rules thatare based on equipment also make it possible toknow which rules are affected if certain changesare made in the process equipment.

The fault diagnosis system has been givingpromising results in both process monitoring anddetection of the condition of process equipmentfailures.

AN INTELLIGENT, INTEGRATEDCONTROL SYSTEM FOR A PRESSUREFILTERKämpe Jerri , Vermasvuori Mikko, RyynänenTimo, Kesti Anna

Artificial intelligence methods like expert systemsand self-organizing maps have proved to beexcellent tools for the control of mineralprocesses. This technology is currently beingembedded directly into process equipment likedewatering filters. AI methods can be seen mostapplied in industrial applications since 1991compared to other methods. They are used innearly 40 % of all applications reviewed, andtherefore represent the most important methodsapplied in the control and monitoring of MMprocesses.

Separation of solids from a liquid by a porousmedium or screen, which retains solids whileallowing the liquid to pass, is called filtration. Thepressure f i l ter can be divided into eightsubsystems: slurry feeding system, water pressingsystem, air drying system, cake washing system,cloth washing system, hydraulic system,discharging system and control system. Since infiltration the slurry volumes are extremely large,any downtime of the process is expensive.

An intelligent control system has beendeveloped for a pressure filter. The applicationhas been programmed with platform independentobject oriented programming language, Java.

The remote operating service system is a partof the fault diagnosis module. In addition to theimplemented system the plant site needs adatabase server (SQL server connected to a Laroxpressure filter), an IIS server, workstations, and a


local area network. At the client end there areworkstations, and both ends are connected toeach other via the Internet. Data are transportedusing a browser and HTTP as the interface. Dueto security reasons, all data transmissions on theInternet between the plant and the client areencrypted with a DES algorithm, and access tothe local network is permitted only through thefirewalls. The encryption key is pre-shared, andencryption/decryption of the data limits thecapacity to approximately 8 Mb/s. The databasesinclude measurements and the learning part.

The aim of the learning database is to suggestpossible optimal control parameters to theoperator. Suggestions are based on data fromprevious operation cycles.

The structure of the system is modular inorder to keep it easily expandable andmaintainable. The overall system consists of themodeling, classi f ication, economic, faultdiagnosis, control and database modules. Themodeling module contains models of thedifferent operating stages of the filter.

The aim of the optimisation is to maximisethe capacity and to decrease cycle t ime.Economic aspects are followed on the basis ofoperating costs calculated from on-linemeasurements. The determination of themaximum capacity is based on onlineclassification of feed type, predicting models offiltration behaviour and a self-learning database.

The main aim of the classification module forthe pressure filter control system is to train a SOMusing information about different feedcompositions, and to determine for each neuronthe correct values of the variables used in control.Measurements for feed type classification infiltering depend on the installation and could befor example the particle size distribution and thedensity of the slurry.

The preliminary testing of the modellingmodule has been performed in the pilot plant.Testing period of the whole integrated system inthe industrial scale has been started.

INTEGRATED MONITORING ANDFAULT DIAGNOSIS SYSTEM FORPROCESS EQUIPMENT AND UNITPROCESSESBergman Samuli, Komulainen Tiina, KämpjärviPetteri, Nevalainen Susanna

The aim of this project is to develop a modularsystem for process monitoring and fault diagnosis.The goal is a software prototype for monitoringthe process st ates and process equipmentespecially in chemical unit processes.

In the first part of the project controlperformance indices were developed and tested.The model predictive controller wasimplemented to control the quality of theproduct. Control performance calculations wereimplemented in the controller. A performanceindex that accounts for restrictions of controlledand manipulated variables was also developed.Indices were also developed for supervision oferror signal and control moves. Using predictionerror variance to evaluate control performancegave promising results. With prediction errorvariance, device faults were detected. Theperformance indiced applied to contol signalswere able to detect aggressive control action.Indices developed for supervising processconstraints were able to detect situations wherethe value of a controlled or manipulated variablewas at its boundary.

The second part of the project consisted of astudy of different fault detection and processmonitoring methods applied in the chemicalprocess industry and implementation of someapplicable methods to monitor thedearomatization process. The methods were firsttested with simulator data and later online withreal process data from Fortum oil refinery.Artificial neural networks, fuzzy logic andstatistical multivariable methods includingprincipal component analysis and partial leastsquares were the selected methods to be used witha simulator.

Both self-organizing maps and projection ontolatent structure models gave promising results.Principal component pre-processing clearlyimproved the SOMs’ ability to classify the processstates. The drawback of the PLS model was itsinability to identify the different abnormal states.In practice, the use of SOMs for processmonitoring might be easier for a process operator,since the map with coloured neurons is veryvisual.


The results showed that variables have asignificant effect on performance of the models.Monitored process variables should be selectedbased on both principal component analysis andprocess knowledge. Constructing computationalvariables that describe the monitored processphenomena is of great importance. The secondpart of the project also included an online faultdiagnosis and model updating system for ethyleneplant. The aim was to detect faults in analyzersand reconstruct missing measurement values.

The project was done in collaboration withNeste Engineering and the National TechnologyAgency of Finland, Tekes.

INTELLIGENT PILOT FLOTATION CELLWITH FIELD DISTRIBUTED CONTROLSTimperi Juha, Poikonen Risto, Kämpe Jerri

The aim of this project was to build a fielddistributed control (FDC) based automationsystem for a pilot flotation tank in laboratory ofprocess control in HUT and to research thepossibili t ies in util izing digit al f ieldbustechnology in device diagnostics. The goal wasto build an automation system similar tocommercial Proscon system. The main differenceto Proscon system was utilization of digitalfieldbus instead of analog 4-20 mA signaling.Fieldbus technology was chosen to beFoundation Fieldbus because it fits well to beused in device diagnostics.

The flotation cell process in our laboratory isa slightly modified version of the OutokumpuTC-3 TankCell. An additional 3.5 m3 tank is usedto provide circulation.

The process is equipped with state of the artFoundation Fieldbus instruments. Three digitalNeles Automation valve positioners are used tocontrol the cell output valves and air supply valve.

Level, flow and pressure measurements areimplemented using Smar, Rosemount andMiltronics technology. An Ilmeco flotation airblower and feed circulation pump are controlledby Vacon frequency converters. The fieldbusuniversal bridge (DFI302) that acts as a linkbetween foundation and ethernet network wasalso from Smar.

Foundation Fieldbus offers a possibility to usefield-distributed control. This means that all thecalculation and control actions are made by fielddevices, not by PLC. This enables a “Plug’n Play”type installation via a standard Ethernet cableinto a Outokumpu PROSCON 2100NT Process

Management Station without any wiring to thePLC. When system was implemented, anarchitecture that would be scaleable and modularas possible was aimed. Therefore use ofproprietary interfaces was not appropriate. Defacto interfaces standarts like OPC and ODBCwas used in system integration. System wasconfigured using Syscon 302 from Smar and theuser interface software that was used to controlthe process was GE Fanuc’s Cimplicity. Both ofthese softwares works via OPC server.

WIRELESS AUTOMATION IN PROCESSINDUSTRIESSuhonen Jori

The aim of this project was to find out the currentstatus and future trends of wireless automationespecially in process industries. The project wascarried out in co-operation with Nokia.

The project concentrated on the three mainparts. In the first part several case studies fromthe industry were studied and especially thesuitability of these technologies for processautomation purposes was of great interest. Thetechnologies included all the relevant networktechnologies, connection protocols as well asapplication development technologies usednowadays and in the near future. The case studies


included many interesting and innovativeapplications using wireless automationtechnologies ranging from simple small-scalesystems to entire communication systems of afactory. The most interesting ones were wirelessuser interface based on a WLAN technology anda sensor ball application where a sensor-ball goesalong with the process liquid collecting data fromthe process on the way.

In the second part a comprehensivequestionnaire study was carried out. Altogether220 professionals of the automation industryanswered the questionnaire. The questionscovered wireless automation from several aspectsand the results from the questionnaire study toldus how the respondents saw the current status andfuture trends of wireless automation. We canconclude that the respondents strongly believethat in the future wireless technologies will beused more and more extensively for processautomation purposes. In particular, the use ofwireless automation in process monitoring andprocess control applications seems to beincreasing rapidly.

In the last part a wireless test environment wasbuilt in the Laboratory of Process Control andAutomation using Bluetooth and commercialGSM-network. The aim of the environment wasto simulate a communications structure of a tankfar away from the backbone system. Bluetooth wasused for short-range communications betweenmeters and an SMS gateway. SMS was then againused for long-range communications between thegateway and the backbone system.

Experiences gained from the test environmentwere encouraging. First of all the system seemedto work very well for the job and secondly thesystem was relatively easy to implement and toconfigure. The only small problem with thesystem was the use of SMS, which cannot be usedin real-time meter reading applications. Using e.g.GPRS instead would solve this problem. In thefuture the environment will be used foreducational and scientific purposes.

SOFTWARE FOR THE LEVELCONTROLLER TUNING OF FLOTATIONCELLS IN SERIESDietrich Maija, Halmevaara Kalle, Tiili Otso

The main goal of this project was to developtuning software for level control of slurry indifferent Outokumpu flot ation cells .Mathematical models were constructed for thefirst flotation cell, intervening cells and the lastcell in the series.

A number of models were developed:The properties of an ideal tank were assumed andclassical PI controllers tested, physical propertiesof the cell (shape) were included, the feedforwardcontrol strategy was implemented, dynamics ofthe Larox pinch valves and Outokumpu Dart-valves were included in the models and the valveswere resized, viscosity effect on valve sizing wasconsidered

Mathematical models for the double cellswere constructed in a similar manner. Theprinciple dif ferences compared with themathematic models of single cells in series arethat, in a double cell, both pulp levels arecontrolled by manipulating a control valve in thesecond cell outflow.

The simulation of the configurations of threeand six TC-50 cells in series resulted inparameters for the PI controllers. All thedisturbances introduced into the system could becompensated using both valve sizes.

Different valve sizes and different cell typeshad an effect on the proportional gain of the PIcontrol: the proportional gains were larger whenusing valves sized according to the ISA standardthan oversized valves. Similarly, the proportionalgains were larger in the case of ideal tanks thanwith cells. Higher proportional gains reduce thesettling time of the system in PI control. TheOpening speeds (from 0% - 100 %) of the valveshave been estimated to be 30 seconds regardlessof the valve size, i.e. the bigger the valve, theshorter the rise time.

The influence of valve sizing and additionalfeed forward control were also studied in a pilotplant. The test configuration consisted of aflotation cell, pulp circulation cell, pulp pump,air feed device and two different control valves.The inflow was measured with magnetic flowmeter and the pulp level with a float. The testswere performed with flotation cell with similargeometries (as described above). The resultsachieved were analogous to those obtained in thesimulations. The control performance was


noticeably improved by adding the feed forwardcontrol. It was also found that an adaptive feedforward control gaincould result in even greaterimprovements in control performance.

The geometry of the cell had the greatesteffect on the performance of the level control. Adifference in performance was clearly evidentwhen the same configurations with ideal cells andwith different geometry cells were compared. Inevery case the results were clearly better.

Feed forward control implemented with PIcontrol improved the compensation of thedisturbances in every case. An adaptive feedforward control gain could result in even biggerimprovements in control performance. In futurethe research extends to determine the effects ofrecycling flows on the control strategy.

LEVEL CONTROL STRATEGIES FORFLOTATION CELLSKämpjärvi Petteri

Flotation is a difficult process to run efficiently.One way to make flotation performance better isto improve cell level control. However,controlling pulp levels in flotation cells is acomplex control t ask because of stronginteractions between the levels in the flotationcells. Therefore advanced controllers are neededto give good level control. This research dealtwith a model of six flotation cells in series.Simulations were performed to compare differentcontrol strategies. Four control strategies wereconsidered: one SISO controller and threedifferent MIMO controllers. It was shown thatlevel control performances of the MIMOcontrollers are significantly better that of theclassical SISOcontroller.

The difference between different MIMOsystems were somewhat smaller. All thecontrollers performed robustly to disturbances inpulp feed and to set point changes. Thedecoupling controller had the best IAE and IDEindices. However, the decoupling controller issensitive to model uncertainties. This also meansthat process changes can strongly degrade thecontrol performance.

DESIGN AND IMPLEMENTATION OF ACONTROL PERFORMANCEMONITORING SOFTWAREPoikonen Risto, Georgiev Zdravko, ZuehlkeUrsula

The number of control loops used in industry isgrowing continuously and there are problems inkeeping them well tuned. In order to ensurehighest product quality it is essential to maintainthe control systems in an adequate manner. Animproved control performance has a considerableeffect on variations in end product quality, andthus on the productivity of the plant. Additionalbenefits are low consumption of raw materials andenergy, as well as a longer life span of theinstruments.

During the last decades considerable effort hasbeen placed on developing suitable indices forevaluating control performance. The evaluationmethods can be divided into two categories:stochastic and deterministic methods. The mostwidely studied stochastic indices are those basedon using of MVC (minimum variance controller)calculation as a benchmark. Deterministicindicators are more informative in the case of asudden load disturbance or a set point change.Various dimensionless indices for setpointchanges have been proposed in the literature.Different kinds of indices were tested using aflotation cell process model that was programmedin Matlab Simulink toolbox. Indices thatfunctioned well in the simulations, and whichwere applicable to on-line monitoring, werechosen for further development.The indices were implemented using the GEFanuc CIMPLICITY HMI Plant Edition and itsscript language. Cimplicity is a user interfaceprogram for process control, and is widely usedin flotation processes. Outokumpu pilot Tankcelland the above-mentioned programs were to testthe indices in practice. The cell instruments wereconnected to the Foundation Fieldbus ancontrolled by Smar DFI 302. The controller unitwas connected via Ethernet to theOPC(OLE(Object Linking and Embedding) forprocess Control ) server, and Cimplicity used theOPC port to control the flotation cell.Simulations and tests performed with the pilotflotation cell proved that the indices provided thenecessary information about the controlperformance. This monitoring tool could be usedin plants to monitor the key controllers forimproving process control and product quality.More attention will be paid to the economicaleffects in the future research.


INTELLIGENT MODELLING ANDSIMULATION APPROACHES TOHYDROMETALLURGICAL PROCESSESCziprian Zoltan, Grau Rodrigo, Kojo Tero

The use of computer simulation is now a basictool for process design and optimization asevidenced by its increasing use of practicingengineers through commercially availablesimulator software products. Engineers, in chargeof designing equipment for thehydrometallurgical plants have to be able to selectthe most suitable equipment for each applicationand to predict the performances under differentprocess conditions. In order to allow the processengineers to concentrate on their task, thesimulation software has to offer a user-friendlyinterface combined with quick and accuratecalculations.

A new simulator is under development, whichoffers a fully graphical user-interface, opendatabase connectivity and application orientedunit model library. The purpose of the developedsoftware is to integrate the already existentcalculation algorithms, database files and processunit models into a completely new user-friendlyand open-architecture system. The abovementioned knowledge exists in universities andcompanies under different representation forms.In the field of hydrometallurgy the simulationfacilities offered by the available commercialsoftware products are very limited and the unitmodels usually are fixed. Our aim is to implementspecific unit models and to adapt them to theresearch and/or industrial application in study.At this st age the focus is on thehydrometallurgical process, however in future theresearch might concentrate on other process unitsdepending on the industrial demand.

The most important feature of this software isto gather all the following proprieties:- graphical user interface- specific unit models for hydrometallurgicalapplications- chemical database with open connectivity- dynamic simulation- steady-state material and energy balancecalculations- model development possibilities

MODEL-BASED CONTROL OF COPPERSOLVENT EXTRACTION -ELECTROWINNINGVille Suontaka

During the last two decades copper leaching,solvent extraction and electrowinning (LX/SX/EW) process has become an important processoption for producing copper from a low-gradeoxidized ore. LX/SX/EW-copper production hasgrown very fast during the last decade and furthergrowth can be expected. At the moment there isabout 50 plants in the world producing copperwith LX/SX/EW-process. In year 2001 world LX/SX/EW-copper production was about 2.8 milliontons of which about a half was produced in Chile.

The process is relatively young andintrinsically quite stable and therefore previouslydevelopment has mainly focused on the processequipment. This has led to a low level ofautomation compared to other mineral processingoperations. Usually only a low-level regulatorycontrol, where controller set points are setmanually, is implemented. Improvements in plantautomation would make it possible to keepprocess parameters near their optimal valueswhich would increase amount of copperproduced, reduce amount of chemicals andenergy used and improve copper quality. Theseimprovements would lead to remarkableeconomical benefits.

The aim of this project is to study possibilitiesto use model-based control in copper LX/SX/EW-process and to find out the possibilities to improveprocess performance and economics with suchcontrol strategy.

First current status of modeling, simulationand control of the LX/SX/EW-process was studied.It was found out that research on advanced controlof the process has not been published widely.


Next a process monitoring application for copperSX/EW-process is developed. The application isbased on steady-state mass balances of the process.Flow measurements and on-line analyzer assaysare used to calculate amount of copper transferredin different stages of the process. These valuesare compared to the calculated theoretical values.The application assists process operators inprocess control and functions as a basis for controlapplication to be developed in the future.

The project is done in a co-operation betweenOutokumpu Technology Oyj.

KNOWPAP - MULTIMEDIA LEARNINGENVIRONMENT FOR PAPERTECHNOLOGY AND PAPERMILLAUTOMATIONMäenpää Tom

KnowPap is a new generation learningenvironment covering paper technology, processcontrol, paper mill automation and maintenance.It consists of multimedia training material andsimulation models that are accessed by a standardweb browser. KnowPap is used for self-studyingand as supplementary material for training withinpaper manufgactoring companies, suppliers andconsulting companies, universities and othereducational institutes.

The development of KnowPap was started in1997. The Project was managed by VTTAutomation and the system was completed in thebeginning of 2001. KnowPap was then alreadyin use in the participating companies, HelsinkiUniversity of Technology, Technical ResearchCentre of Finland (VTT) and in more than 20other educational institutes.


4 STRUCTURE OF THE MASTER’S DEGREE

The extent of master’s degree in technology is 180 credits. The degree consists of two parts and it’s nominalduration has been confirmed as 5 years. Part I of the degree consists of approximately 70 credits and includesmathematical and scientific subjects, such as mathematics, physics, information science and general studiesas well as foreign languages. In part II of the degree the student chooses process automation as his/her majorand minor which can be chosen outside the degree programme or even from another university of technology.Part II of the degree in process automation also contains studies in computer science and related topics. Thetopic of the M.Sc. thesis is chosen to fit within the major subject.

Optional studies by different fields

Major students of process control are able to include several courses in computer science and engineeringin both major and option studies. Students can choose courses from Computer and Information science,Information Processing Science, Software Technique, Telecommunications Software and Applications,Interactive Digital Media, Contents Production and Usability Research.

The questionnaire study from graduated students indicated that a good knowledge in computer sciencehas been an important competence for our students working in the Finnish process industry.


5.1 Undergraduate studies

Kem-90.131 Optimisation (2 cr)

Lecturer: M.Sc.(Tech.) Tatu KuitunenContents: Applications of optimisation methods in process industry.Laboratory exercises with

the MATLAB optimisation toolbox.Literature: T.F. Edgar and D.M. Himmelblau, Optimisation of Chemical Processes, McGraw-

Hill 1988. Course Notes.

Kem-90.139 Process Industry Measurements (2 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: Process Industry Measurements. Guest speakers from the industry.Literature: O.Aumala, Teollisuusprosessien mittaukset, Pressus Oy, Tampere, 1996 Course

NotesKnowpap – Multimedia learning environment for paper technology and papermill automation.

Kem-90.147 Basic Cource in Process Automation (2 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-Jounela,M.Sc.(Tech) Mika Järvensivu

Contents: Automation in process industry. Instrumentation and special measurementsat the plant. Dynamic modelling and simulation of the prosesses. Feedbackcontrol. PID- controller and it’s tuning. Structure of the automation system.Design and planning of the automation project. Course is a composed of seriesof lectures given by experts in process automation.

Literature: Course Notes.

Kem-90.148 Process Modelling and Simulation I; basic course (2 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: Physico-chemical modelling and simulation of the unit operations in process

industry. Laboratory exercises with the SIMULINK and MATLAB toolboxes.Literature: W.L Luyben, Process Modelling, Simulation and Control for

Chremical Engineers, McGraw-Hill, 1990.Course Notes.

Kem-90.149 Process Modelling and Simulation II (2 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: Practical experiment design. Modelling of discrete systems.

Identification methods. Laboratory exercises with SIMULINKand MATLAB system identification toolbox.

Literature: W.J. Diamond, Practical Experiment Design for Engineers andScientist, Van Nostrand Reinhold, 1981.R.Isermann, Identifikation Dynamisher Systeme, Band I-II,Springer-Verlag, 1988. Course Notes

5 TEACHING ACTIVITIES


Kem-90.156 Process Automation Project Work (3 cr)

Lecturer: Lab. Manager Jerri KämpeContents: Configuration and Implementation of process automation system. Laboratory with

Wonderware FactorySuite 2000 and Proscon 2100 automation system. GuestSpeakers from the industry.

Kem-90.160 Information Technology In Process Control (3 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: Planning methods, tools and interfaces supporting application software of process

automation. A small scale implementations of process automation. The objectiveis to learn trough practice. The content of the course varies every year. The courseis designed for the students who major in process control.

Examination: assignments, final exam

Kem-90.161 Process Control Structures and Applications (4 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: Review of the classic and modern control and theory. Process control ofbasic

functions: level, pressure, volume flow and mass flow, energy, and concentration.Process control of common unit processes and plantwide control systems.Laboratory exercises with MATLAB model predictive control and control systemtoolboxes and SIMULINK.

Literature: Course Notes

Kem-90.162 Intelligent Process Control Methods (3 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: The aim of the course is to provide an overall understanding of the appli cation

of artificial intelligence (AI) in process control. The course will give an insighthow AI techniques can be used to improve control systems.Special emphasiswillbe put on intelligentcontrol-expert control, fuzzy control and the use of neuralnetworks for control. The course will combine conceptual

Examination: Formal lectures, session notes, assignments, final exam.Literature: C-T. Lin and C.S. George Lee, Neural Fuzzy systems, Prentice Hall, 1996.

Course Notes.

Kem-90.163 Production Planning and Control in Process Industry (3 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: Operations Planning and Control: Forecasting for Operations, Inventory Planning

and Control , Operations, Scheduling and Statistical Quality Control Methods.LINEAR PROGRAMMING: Simplex Method; MATHEMATICALPROGRAMMING: Network analysis , Dynamic Programming, IntegerProgramming, Nonlinear Programming: PROBABILISTIC MODELS: QueuingTheory, Inventory Theory, Forecasting. Laboratory exercises with Arena.

Literature: F.S. Hillier and G.J. Lieberman, Introduction to Operations Research 5. Ed.McGraw-Hill, 1990W.D. Kelton, R.P. Sadowski and D.A. Sadowski, Simulation with ARENA,McGraw-Hill, 1998, 547 p.Course Notes


Kem-90.173 Control and Information Systems in Pulp and Paper Industry A (3 cr)

Lecturer. Prof. Sirkka-Liisa Jämsä-JounelaContents: The course gives an overview of the control and information applications in the

pulp and paper industry at processes, department, mill and corporate levels. Thebasic hardware solutions, models, functions and results achieved will be discussed.Further needs and thends in automation and information technology will also bepresented. This course is mainly for linkage program students and for thosestudents that have selected process as their minor subject.

Kem-90.174 Control and Information Systems in Pulp and Paper Industry B (3 cr)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: The course gives a detailed knowledge of existing control and information system

applications in the pulp and paper industry at processes, department, mill andcorprate levels. It provides an overview of classical control and basics in digitalcontrol methods. Also modern control strategies such as multivariable control andpredictive control will be discussed. This course is for major students of processautomation and partly organized together with the courses Kem-90.152 and Kem-90.173.

Kem-90.V Instrumentation and Control ( 1,5 cr)European Mineral Engineering Course (EMEC) ( 2,5cr) (12.3.2001- 16.3.2001)

Lecturer: Prof. Sirkka-Liisa Jämsä-JounelaContents: The aim of the course is to give an overall understanding of process automation:

the basic hardware solutions and functions, instrumentation and specialinstruments, dynamic modeling and simulation of chemical processes, feedbackcontrol, PID -controller and its tuning, feedforward control, ratio and cascadecontrol, control applications in mineral processing industry

Literature: Course notes


5.2 Post-graduate courses

The laboratory is involved in two graduate schools: Graduate school in Chemical Engineering (GSCE)and international Ph.D Programme in Pulp and Paper Science and Technology (PAPSAT). The coursecatalogues are available in http://www.abo.fi/gsce and http://www.hut.fi/Units/Faculties/P/Papsat/.Thefollowing courses organised by the Laboratory of Process Control and Automationwere included inthe course catalogs of these graduate schools.

Kem-90.V Linear Control Systems (5 cr) ( 28.8.2000 - 8.9.2000 )

Lecturer: Prof. Anna Soffía Hauksdóttir, University of Iceland, IcelandProf. Tom McAvoy, University of Maryland, USA

Contents: Solution of the state equations, The matrix exponent, Modal decomposiotion, Observability and controllability, Similarity transformations from a general form over to a special form, Kalman’s decomposiotion,Dual systems, output, Definitions of special formsand their properties, transfer function coefficents, Pole placement and design of observers, Similarity transformations to observer and the controller forms, Ackermann’s formula,A combined observer/controller - the separation principle,Observer/controller in transfer function form, reduced order observers, Optimization, Optimization of zeroes, Alinerar system of equations of the form Ax = y, Multivariable systems, minimum forms,zeros , Control of multivariable systems, decoupling controllers, Balanced realization andmodel

Literature: W.L. Brogan: Modern Control Theory, Prentice Hall, 1991, 736pK. Furuta, A.Sano, D.Atherton: State Variable Methods in Automatic Control, Wiley 1988, 220pT. Kailath: Linear Systems, Prentice Hall, 1980Arnar Gestsson: Multivariable Systems Synthesis - Decoupling Controllerfor a Ferrosilicum Furnace. M.S. -theseis, Faculty of Engineering,1994F. Zhou, J.Doyce, K. Glover: Robust and Optimal Control, Prentice Hall,1995, 596p

Professor Anna Soffia Hauksdottir and her students

Kem-90.V Advanced Topics in Model-Based Process Control (3 cr)

Lecturer: Prof. Frank III Doyle, Department of Chemical Engineering, University ofDelaware, USA

Contents A. Single Loop Internal Model Control (IMC) DesignStructural analysis,PID tuning using IMCExtensions:Unstable systems,RHP zeros,Constraints,Time DelaysB. Model Predictive Control MotivationComponents, Vendors, Models for MPC, Process Identification,Unconstrained MPC, Properties,Constrained MPC,Nonsquare systems,State estimation, Nonlinear MPCC. Grade Transition ControlGeneral problem ,Cost functionsControl strategies:Nonlinear direct synthesis,Gain scheduling,Nonlinear MPCApplication examples:Paper machine,Pulp digester,Polymer reactorD. Process IdentificationModel structures:IR, SPR,State-space,ARMAX ( BJ, OE, BJ, ARX, etc.)Identification methods:Prediction error methods,Correlation methods,Subspace ID,Recursive methodsPractical issues:Input sequence design,Structure selection,ValidationE. Run to Run ControlIntroduction, overview, review of adaptive controlIterative learning control for batch-to-batchOptimization framework for run-to-run controlAdvanced issues -incorporation with MPC

Literature: Ogunnaike & Ray - Process Dynamics, Modeling and Control (Oxford)Morari & Zafirou - Robust Process Control (Prentice)Ljung - System Identification : Theory for the users (Prentice)

Topic B: Ogunnaike & Ray - Chapter 27Topic C: Journal ArticlesTopic D: Ogunnaike & Ray - Chapter 13 (also Ljung)Topic E: Journal Articles

Professor Frank III Doyle with his students in Suomenlinna


6 HARDWARE AND SOFTWARE FACILITIES

6.1 Laboratory Equipment

- fully instrumented and automated heatexchanger system consisting of four pilot heatexchangers- fully instrumented and automated mixing tankprocess.- fully instrumented and automated flotation cellprocess, with Foundation Fieldbus

6.2 Automation Systems

Outokumpu TankCell Automation system- SMAR DFI302 Foundation Fieldbus system- SMAR System 302 configuration Software- Bluegiga wireless process automation system- SAMR OPC-server- Cimplicity Control Software (HMI)- MS SQL 7.0 databaseAutomation system is connected to theOutokumpu tankcell research&teaching process.

Outokumpu Proscon 2100 NT automationsystem- GE Fanuc 90-70 Programmable LogicController- SMAR PCI302 Foundation Fieldbus system- Cimplicity Control SoftwareAutomation system is connected to the mixingtank teaching process.

Wireless AutomationA wireless test environment using Bluetooth andSMS technologies has been built in order tosimulate the communication structure of aflotation cell located far away from the backbonesystem.

PC-based automation system- Opto 22 remote I/O unit, with Ethernetconnection- Wonderware Factorysuite 2000 softwareThis automation system is connected to theteaching process containing 4 different heattransfer units.


6.3 Computers and related hardware

Desktop computersProcessor Model Amount

Intel Pentium IV 2 400 MHz 4Intel Pentium IV 2 000 MHz 2Intel Pentium IV 1 500 MHz 2Intel Pentium III 1 000 MHz 3Intel Pentium III 800 MHz 2Intel Pentium III 750 MHz 2Intel Pentium III 700 MHz 1Intel Pentium III 650 MHz 2Intel Pentium III 600 MHz 1Intel Pentium III 550 MHz 1Intel Pentium III 500 MHz 1Intel Pentium II 450 MHz 1Intel Pentium II 400 MHz 2

Laptop computersProcessor Model Amount

Intel Pentium 2 000 MHz 1Intel Pentium 1 133 MHz 1Intel Pentium 700 MHz 2Intel Pentium MMX 233 MHz 1Intel Pentium 133 MHz 1

Primary Network Server Computer- Intel Pentium III Dual 1 133 Mhz- 512 MB RAM- 2 * 18 GB and 2 * 36 GB SCSI hard disks usingRAID 1- Windows 2000 Server Operating system

Secondary Network Server Computer- Intel Pentium II Dual 350 Mhz- 512 MB RAM- 4 GB and 2 * 36 GB SCSI hard disks- Windows 2000 Server Operating system

Network SQL- and Www-Server Computer- Intel Pentium III Dual 933 MHz- 512 MB RAM- 2 * 18 GB and 2* 36 GB SCSI hard disks usingRAID 1- Windows 2000 Server Operating system- Microsoft SQL Server 7.0- Microsoft Internet Information Server 5

Protected Network Server Computer- Intel Pentium 133 MHz- 128 MB RAM- 2 GB and 9 GB SCSI hard disks- Windows NT Server 4.0 Operating system

Printers- HP Laserjet 4100DTN (1200 dpi)- HP Laserjet 4050N (1200 dpi)- HP Laserjet 4000N (1200 dpi) (2 printers)- HP Color Laserjet 4500N (600 dpi)

Firewall- Cisco PIX-506 VPN

UPS- APCC Smart-UPS 3000VA- APCC Smart-UPS 1400VA (2 systems)- APCC UPS 600VA

Bluetooth- Bluegiga Starter Kit- Bluegiga WRAP 1260 Microserver

GSM Terminals- Nokia 30 GSM Connectivity Terminal (2Terminals)

Scanners- HP ScanJet 6200c- HP Scanjet 4c

Data projector- Infocus LP 350- Infocus LitePro 720

Copy Fax machine- HP Laserjet 3330

Digital video camera- Sony Mini DV DCR-PC10E PAL

Digital camera- Canon Digital Ixus 300

Camera- Olympus m-zoom wide 80


The communication architecture of the wireless test environment

6.4 Software

Control Engineering Software- 20 licenses of Wonderware Factory Suite 2000(containing Intouch, Industrial SQL Server etc.)- Matlab 6.1/Simulink www-serverClassroom licenses of:- Wonderware Factorysuite 2000- Rockwell Software Arena 5.0Classroom licenses are installed in the computerclass room of the department of chemicalengineering.

Software and Internet development tools- Borland JBuilder Professional- Microsoft Visual Studio Professional Edition 97- Microsoft Visual Studio 6.0 Enterprice- Microsoft Visual C++ 5.0- Microsoft Visual Basic 5.0- Microsoft FrontPage 2000- CompanyM Multimaker

Office automation software- Microsoft Office 2000- Adobe PhotoShop 5.5 and 4.0- Adobe Premiere 5.0- Adobe Acrobat Writer 5.0- Corel Draw 10- Microsoft Project 98- Microsoft Publisher 98- Omnipage Limited Edition- StatSoft Statistica- Seagate Crystal Reports Developer Edition


7 THESIS 2000-2002

Thesis for the Degree of Licentiate In Technology

Laukkanen, Ismo Studies in using hybrid dynamic simulation trough the life-cycle on of paper mill

Cziprián, Zoltan Simulation software development for material balancecalculations in reacting systems.

Thesis for the Master of Science In Technology

Chemistry:Bergman, Samuli Monitoring of a Dearomatization Process TMP refinerForsell, Marko Instrumentation audit of sugar manufacturing process

and Technical Solutions on Data SecurityGeorgiev, Zdravko Control loops performance monitoring and process assessmentGrönbärj, Marko Inference Methods in Industrial Fault DiagnosticKämpjärvi, Petteri Diagnosis of process data and update of an on-line modelLaiho, Emmi Training Material About Turbocompressors and Their Control

MethodsMustonen, Valtteri Control Strategies of the Continuous Pressurised Three Phase

Separating TankMuurinen, Matti Simulation study on the control of the bed and furnace of a

fluidised bed boilerNevalainen, Susanna Control performance indexes in petrochemical plantsPeltoniemi, Jyrki OPC data transfer in large-scale dynamic process simulationSuhonen, Jori Wireless Automation in Process Industries

Pulp and Paper:Dietrich, Maija Real-time economic model of paper machine operationsHafrén, Lena Measuring the rheological properties of surface sizesHevonoja, Tommi Development and Start-up of Ultrafiltration Process Control

System in a Publishing Paper MillMalmi, Timo Uniformity of Quality at Metsä-Botnia Ab’s Kaskinen PlantMatinaho, Sami Development of BCTMP mill simulation model and

modelling of TMP refinerNäränen, Janne Simulating and Analysing the Effects of Refiner Segments

on TMP-Process Using a Theoretical ModelPetäjä, Janne Applying Remote Diagnostics in Service BusinessRönkkö, Kati Optimising Wet End Chemistry in Wallpaper Base MachineYlä-Jarkko, Olli Modelling the Technical Age of a Pulp Mill


Mining, mineral and metallurgy:Enden, Petri On-line testing of a process monitoring system and analysis

of resultsGrau, Rodrigo Mass Balance Algorithms for Reacting Systems and Strategies

for Computer SimulationHaavisto, Sasa Rule based fault diagnosis system for industrial processesLipiäinen, Jouni Remote Support System for Pressure Filter - Data Aqcuisition

and Technical Solutions on Data SecurityRemes, Antti Effects of the XRF analyzer accuracy and sampling frequency

on the control performance of flotationSihvo, Olli The Intelligent Field Devices - a Part of Asset ManagementTiili, Otso Development and Implementation of Control Strategy for

Pulp Levels in a Series of Flotation CellsTimperi, Juha Utilization of Fieldbus Technology in Control and

Information System of a Process DeviceVermasvuori, Mikko On-line Monitoring of Process Disturbances for the

Outokumpu Copper Smelting Process

Thesis for the Master of Science in Progress

Hankimäki, Janne Dynamic Simulation of Paper Machine, ( VTT Automation)Komulainen, Tiina On-line monitoring and fault diagnosis for the

dearomatization process, ( Fortum Oyj, HUT)Koskelainen,Jussi TBALiimatainen, Tommi Glass model of an on-line multimit calender ( Metso Paper

Oyj)Nevalainen, Seppo Dynamic Modeling and Simulation of Alumina Calcination

Process, ( Lurgi Metallurgie GmbH, Frankfurt)Poikonen, Risto The Design and Implementation of a Control Performance

Monitoring Program, ( Outokumpu Oyj, Tekes, HUT )Rinta-Kokko, Pekka Predicting of Mechanical Pulp Production, Broke Use and

Electricity Costs in a Paper Mill, (KCL Science andConsulting)

Seppälä, Hannu Numerical Simulation For Evaluating the Usability ofSecondary Treated Effluent at Process Water at MechanicalPulp Mill, ( KCL)

Suomalainen, Suvi Improving the Software Engineering Process of the SafetyRelated System, ( Mipro Oy)

Suontaka, Ville Model-based control of copper solvent extraction-electrowinning, ( Outokumpu Oyj, HUT)

Wikström, Tua Advanced Retention Control in Newspaper Machine, ( UPM-Kymmene Oyj, Metso Oyj)


7.1 Thesis abstract for the Licentiate InTechnology

Studies in Using Hybrid Dynamic SimulationTrough the Life-cycle of Paper MillLaukkanen, Ismo

This study focuses on the applications of hybriddynamic simulation in the paper industry. Theprinciples of modelling the paper machineprocess are introduced. A dynamic, hybridsimulator consists of continuous, discrete time,discrete event and static models which are usedto model a thermohydraulic network and papermachine submodels. The simulator developed inthis study was incorporated in a program calledAPMS, which was used as a modellingenvironment in this study.

A mathematical simulation model of the papermachine process was developed during the papermachine design project. The goal of modellingwas to analyse the process dynamics of the newpaper machine, including new process andcontrol design concepts. There were no industrialscale experiences on the controllability of similarprocesses available. The combined response ofthe process equipment and the control systemunder transient operation conditions wasinvestigated; both major process equipment andpaper mill control structures were included inthe simulator. Both normal and transientoperation of the process were verified. As a result,the faulty operation of two control circuits wasidentified. Redesign and fine-tuning could becarried out before the initial start-up. It wasestimated that 0.5 - 1 day troubleshooting timewas saved.

The first application for the simulator wasoperator training, including training in normal,grade change, web break and process disturbancesituations. A real process control system (ValmetDamatic XD) was used to operate the trainingsimulator. As a result of simulator training, thestart-up of the machine was very easy and theoperators responsible for running the machinelearnt how to operate the new process. Thetraining turned out to be too difficult for someoperators. On the basis of these results, a web-based multimedia learning environment calledKnowPap was developed. The usability of

KnowPap, both in classroom training and forindividual study, was investigated. It turned to beout a very good training tool for initial andintermediate training, while the full scopesimulator is a good tool for advanced operatortraining.

The simulator was next used in processanalysis studies. A problem, which has beenreported in many paper mills, concerningconsistency fluctuations of save-all filtrates wasstudied. The reason, a rapid change in the blendchest level was identified, using the simulator.The change causes disturbances in the sweetenerstock composition and in the save-all filteroperation. The short and long term effects of thechange are observed as fluctuating filtrateconsistencies and fluctuations in paper qualityproperties, respectively.

Based on these results, we propose that by theoptimally use of hybrid models, the life cycle ofthe simulator follows the life cycle of the papermachine. The hybrid simulator ismultifunctional, fulfilling the requirements ofprocess and control engineering, operatortraining, research and process analysisapplications. As a result, the quality of the design,training and process operation can be improved.All applications can be performed before the start-up of the machine, which is possible becausemechanistic high-fidelity hybrid models are used.This is a very important characteristic, whichmakes its industrial use both practically andeconomically feasible. The extended life-cycleconcept was tested on the industrial scale casestudy.

Dynamic process simulation, based on hybridmodels, turned out to be a powerful tool forverification of design, for operator training andprocess analysis. In future, it will provide a strongbasis for the virtual paper machine and paperproduct development environments as well as foradvanced multimedia-training systems.


Simulation software development formaterial balance calculations in reactingsystems.Cziprián , Zoltán

The aim of this work was to develop a simulationsoftware tool that can be used as a frameworkfor specific process simulations. The developedsimulator offers a framework for other researchprojects, and it is not intended to be asophisticated chemical simulator. In this way,various chemical, mineral and metallurgicalprocess simulation-related research projectscarried out at the Helsinki University ofTechnology could share and extend the sourcecode of the same framework.

The general structure of chemical processsimulators is reviewed in the beginning of theliterature part. The basic material and energybalance calculation theory is then presented.The two main flowsheet solving methods,namely the sequential-modular and equation-oriented strategies are compared. Thesequential-modular strategy is presented in moredetail because it has been used in the simulatorimplementation. Solution algorithms for steadystate material balance calculations are presentedin more detail, with special emphasis on reactingsystems.

The design and implementation of thesimulation software are presented in theexperimental part. The software architecture andmain parts of the graphical user interface arepresented in detail . The structure andimplementation of the chemical database arealso briefly presented. The implementeddynamic and steady state flowsheet solverroutines are discussed.

All the basic unit models described in thematerial balance section of the literature part areimplemented in the simulator. Asimplementation of the unit models andflowsheet solvers is based exactly on theequations and methods presented in theliterature part, the equations describing the unitmodels are not repeated in the experimentalpart. However, the elementary energy balancecalculation theory is presented in the literaturepart, the simulator does not include energybalancing, which is the subject of laterdevelopments.

The accuracy of the developed simulator isevaluated using three case studies. The copperheap leaching case study is based on real plant

data and illustrates the dynamic simulationcapabilities of the simulator. Two steady-statematerial balance calculation examples illustratethe accuracy of the solution algorithms underseveral different conditions. The steady state casestudies are based on textbooks, and the calculatedresults are compared with those obtained for thesame flowsheets using two commercial simulationsoftware packages.

7.2 Thesis abstract for the Master ofScience

7.2.1 Chemistry

Monitoring of an Dearomatization ProcessBergman, Samuli

Early detection and identification of abnormal andundesired process states are essential requirementsfor safe and reliable processes. This helps to reducethe amount of production losses during abnormalevents. The aim of this thesis was to study differentfault detection and process monitoring methodsapplied in the chemical process industry and tocompare the performance of these methods inmonitoring an industrial dearomatization process.

In the literature survey part of this study,different methods used for monitoring and faultdiagnosis of chemical processes are discussed.Artificial neural networks, fuzzy logic andstatistical multivariate methods are replacing thequantitative causal models that have traditionallybeen used for process monitoring.

In the experimental part of the study, theperformance of Kohonen self-organizing maps andst atist ical multivariate methods, principalcomponent analysis (PCA) and partial least squares(PLS), in monitoring a dearomatization process arecompared.

The data that were used in constructing themonitoring models were collected using a dynamicprocess simulator. The simulator model wasmodified to correspond to an industrialdearomatization process. Both normal state anddifferent fault conditions were simulated. Theprocess phenomena that were selected forsimulation were an internal leakage in a heatexchanger, fouling of a heat exchanger andchanneling of the catalyst.


The best PLS-model had six input and fouroutput variables. The model was able to detect53 percent of leakage cases, six percent of foulingcases and 69 percent of channeling cases asabnormal events. The advantage of the PLS-method is its ability to detect even novel faults.

The self-organizing maps were constructedseparately for each abnormal st ate. Thechanneling of the catalyst was identified mostreliably. The principal component pre-processedmap was able to identify 94.5 percent of thechanneling cases. The pre-processed map trainedfor leakage was able to identify 78 percent of theleakage cases. The detection of fouling wasunreliable, since only 42 percent of the foulingcases were identified. The advantage of the self-organizing maps is that they are easy to use, sincethe map with colored neurons is very visual. Theprincipal component pre-processing improves themaps’ ability to classify process states.

The variables have a significant effect on theperformance of the models. The variables shouldbe selected based on both principal componentanalysis and process knowledge. Constructingcomputational variables that describe themonitored process phenomena is of greatimportance

Instrumentation audit of sugar manufacturingprocessForsell, Marko

In this thesis the generally used measurementtechniques in sugar plants are explored. Also, thephysical and chemical properties of sugars andpolyols are studied. The unit processes of sugarmanufacture are introduced. The objective of thisthesis is to build a comprehensive discourse aboutthe measurement techniques and why thesetechniques can be utilized in sugar plants. Thechemical phenomena that occur in the processescan be understood on the basis of the propertiesof the components.

The literature part of the thesis reviews thecharacteristic properties of saccharose, glucose,fructose, xylose, xylitol and lactitol. The chemicalbackground of the chromatographic separation,the evaporation cryst allation and thehydrogenation as unit processes of sugarmanufacture are examined. Refractometry,polarimetry, pH, density, conductivity, mass flowand volume flow measurement are introducedhere. Of infrared applications, FTIR, NIR andRaman spectrocopies are studied in more detail.

In the applied part of the thesis ,refractometers, a density meter and a radioactivedensity gauge were tested. The tests were carriedout in the chromatographic separation pilot plantin Kantvik (Finland) and in the xylose productionplant in the crystallation process in Lenzing(Austria). In the separation pilot plant, the meterswere tested with different feed solutions anddifferent parameters. The meters are comparedto the reference values measured in thelaboratory. Also, the correlation functions basedon the mathematical least square method isdeveloped.

Based on the results from the tests, decision-making concerning the measurementinstruments for different applications in futureprojects will be easier. In the betaine solutionseparation process, the density-based drysubstance meters do not operate properly. Therefractometer is more stable and operates moreaccurately in betaine separation. With impuremolasses solution, the prism of the refractometercan become dirty and therefore cause inaccuracyin the measurement. In the end of thecryst allation process, the eff iciency andoperabili ty of the refractometer and theradioactive density gauge decrease substantially.

Inference Methods in Industrial FaultDiagnosticGrönbärj, Marko

The aim of this thesis was to develop an inferenceengine for a fault diagnosis system.

Fault diagnostics has recently been a subjectof great interest and development. Knowledgecan be transferred from process engineers to thefault diagnosis system by reasoning methods. Withan expert system, faults can be diagnosed moreaccurately and maintenance is easy.

In the literature part, the basic structure andfunctions of a fault diagnosis system wereintroduced. Also, different rule-based inferencemethods were studied with examples fromindustrial applications.

In the experimental part, the structure of thefault diagnosis system was designed. The structureconsists of a link between a database and theprocess, databases, a Kohonen map application,a symptom generator module and an inferenceengine. In this thesis, the inference engine,database link and symptom generator wereimplemented.

The inference engine developed was tested


with data collected from the steam dryer in thecopper smelting process of Harjavalta Metals, andthe knowledge of the process engineers was used.During testing, the system responded correctlyto faults and normal conditions.

The rule-based inference used in the systemsucceeded in diagnosing faults and it is applicableto a structurally changing process.

Training material about turbocompressorsand their control methodsLaiho, Emmi

Turbocompressors are used in process industryto raise the pressure of gas and to transportcompressed gas. Because response times incompressor control need to be shorter than innormal process control, used control methods aredifferent.

The purpose of this thesis was to preparetraining material about turbocompressors andtheir control methods. The training material wasintended for use of control room operators in thePorvoo refinery of Fortum Oil and Gas. The aimof the training is to improve the operators'independent initiative during abnormal processconditions.

In the literature survey, the construction,characteristics, control methods and operation ofturbomachinery are studied. The principles ofpedagogy are also discussed.

In the experimental part of the work, HTMLand javascript based e-learning material wasimplemented. The material consists of text,pictures and animations. To apply and practisethe subjects studied, there are also exercises atthe end of every part.

The second part of the training is done withsimulator. It focuses on improving the practicalskills in operating the controls of turbomachinery.In this part, important operating commands arepractised and their impacts on the simulator areobserved.

The e-learning material and the simulator partwere tested, and the test results are reported. Theimplemented training material meets therequirements set in advance and it received goodfeedback during the tests. In addition, thetechnology used in e-learning material was newand implementation will be used as a model forfuture training material in Porvoo refinery.

Control Strategies of the ContinuousPressurised Three Phase Separating TankMustonen, Valtteri

The continuous pressurised CTO-production(Crude Tall Oil) process developed in RinteknoOy consists of a CSTR-type (Continuous StirredTank Reactor) acidification reactor and aseparating tank. After the acidification reactionof solvent-bearing soap, the reaction mixture,which consists of three phases, is decanted. Rawtall oil is obtained as a product. The process hasbeen implemented in the Kaukas researchfacilities of Sterol Technologies Oy. It has beennoticed that the functioning of the processcontrol is unsatisfactory. Control loops stabilisethe controllable variables poorly and require theoperator’s constant supervision.

The objective of this thesis is to study thedynamics of the process and to improve thecontrols in such a way that the process is bettercontrolled. The work aims to find basic controlloops and a solution which would reduce theconst ant supervision of the process. Oneobjective was also to assess the instrumentationof the process.

The literature part of the thesis reviews thephysical phenomena of the pressurised CTO-production process and the theory of theseparatory event. In this part the dynamicmodelling and development methods of dynamicmodels of similar processes were examined. Theliterature concerning basic controls and higher-level controls of similar processes was studied.

In the applied part of the thesis, a dynamicmodel of the process was developed. Modellingwas based on step tests. With the help of thedynamic model, optimal control loops andcontrol parameters were selected. In the processcontrol system, higher-level pressure control wasdeveloped to reduce routine setpoint valuechanges made by the process operator. Aninformative expert system was included in thesystem.

It was observed that the basic controls hadalready been selected in the optimal manner. Theobstacle to functionality was process technicaldeficiencies. The higher-level pressure controlreduces the setpoint value changes made by theprocess operators. It gives information to operatethe process at the correct pressure. Thisinformation was previously unavailable. It wasnoticed that the information system contains theright elements to automate the process. Todevelop the information system into a process


supervision system more research is required.Suggestions for addition and change in theinstrumentation of the process are given.

Simulation Study on the Control of theBedand Furnace of a Fluidised Bed BoilerMuurinen, Matti

Fluidised bed combustion has established itselfas a method of solid fuel combustion, especiallyin small-scale plants. In view of processengineering, the good opportunities to usevarious kinds of fuels with a high efficiency canbe regarded as one of the strengths of fluidisedbed boilers. Also their simple structure can beregarded as another strength. Rapid and strongvariations in the quality of fuels, however, putheavy demands on the control systems of thefluidised bed boiler. In order to be able to usefluidised bed boilers with as high efficiency aspossible, also the control methods used in themshall be developed and various needs shall betaken into account, for example, in view of theoptimisation of operating conditions.

The purpose of this study was to develop theautomation of the furnace of a fluidised bed boilerto better take into account the quality variationsof fuel and remote-control aspects. With bettercontrol of the furnace and the bed, thecombustion conditions can be stabilised, thetemperature fluctuations of the furnace can bebalanced, and that way the power output can best abilised. With active control of bedtemperature, the available fuel selection can beexpanded, operational safety improved and theamount of incombustible in the ash diminished.Until now, the operators have controlled the bedtemperature manually with the aid of correctioncoefficients.

In the study, an adaptive control system of bedtemperature based on a physical and chemicalmodel was developed. The development of bedtemperature in respect of control variables is verynon-linear, and one of the sub-entities of thestudy was to compile a model describing thebehaviour of bed temperature on the basis ofliterary research and existing data. Another sectorof the study was the improvement of thefunctioning of the air distribution of the furnaceand the stabilisation of combustion conditions.In the study, a control structure for thedistribution of combustion airs was developed,based on the regulation of air coefficients

prevailing in the furnace, irrespective of theoperating point or power level of the power plant.The calculation of air coefficients is carried outon the basis of steam circuit measurements, inwhich case the delays occurring in the flue gasanalysis can be efficiently avoided. Thestandardisation of air coefficients is deemed toimprove efficiency and diminish nitrogen oxideemissions.

The operation of developed control structuresin fault situations was tested with APROS, whichis a dynamic simulation environment thatsupports control design, developed in co-operation between Fortum and VTT, theTechnical Research Centre of Finland. Apreviously verified model of fluidised bed boilerwas used as the simulation model. On the basisof the simulation results , with the bedtemperature control system, the bed temperaturecan be controlled automatically according to thewishes of the user, despite the strong non-linearityof the process. With the developed controlsystem, the linearisation of the process responsein various operating points was successful. Withthe new combustion air distribution, a faster andmore accurate control result was reached on thebasis of the simulation results.

The developed bed management system canbe duplicated to different power plantsmoderately easily, and it can be added on top ofthe normal automation as a layer of its own. Dueto this capacity, the introduction of the systemdoes not involve great risks. The control structuresdeveloped on the basis of positive simulationresults are tested in the next stage of the projectin plant-scale tests.

Further development of the control conceptwill aim at the development of an algorithm thatoptimises the combustion conditions of a suitablefurnace and testing it at an operational plant.

Control performance indexes inpetrochemical plantsNevalainen, Susanna

The purpose of this thesis was to develop indicesto evaluate the performance of a model predictivecontroller. The aim is to distinguish betweendifferent situations that cause the deviation ofcontrolled variables. In addition to a tuningproblem, the deviation can result from a devicefault, a poorly tuned lower level controller orrestrictions in the process.

In the literature part, the performance indices


of SISO-controllers were studied and differentways of evaluating the performance of a modelpredictive controller were surveyed. Othermethods used in performance evaluation werealso studied.

In the experimental part , the controlperformance indices were developed and tested.Applications developed by Neste Engineeringwere used. The simulated process was adearomatization unit named LARPO. The modelpredictive controller was implemented to controlthe quality of the product. Control performancecalculations were implemented in the controller.In particular the applicability of the predictiontraining and automation design and testingpurposes. Using this kind of connection betweena full-scale DCS and a large simulation modelrequires, however, a large throughput from thedata connection.

OPC (OLE for Process Control) is a widelyused, component-based specification forcommunicating between numerous data sourcesin the automation and process control industries.This thesis will introduce the architectural issuesthat affect the performance of data exchangeusing the OPC standard. Detailed design issuesthat affect the throughput are also discussed.

A new OPC server for the dynamic processsimulator, Apros, has been developed andperformance tests have been carried out. Theperformance metrics of the new server arecompared to those of the existing OPC server.This comparison shows the benefits of the newarchitectural approach. The results show thatcommunication based on the OPC standard canoffer sufficient data-transfer capacity for mostdynamic process-simulation purposes.

Requirements for the optimized OPC clientare given. Finally, new application areas that canbe managed using the new, more effective dataexchange techniques are discussed.

OPC data transfer in large-scale dynamicprocess simulationPeltoniemi, Jyrki

Progress in the area of dynamic process simulationwill bring about a need for an effective and easyway to configure data exchange between asimulator and co-operating applications. Oneimportant example of this need is the task oftransferring data between a distributed controlsystem (DCS) and a process simulator.

An integrated system, in which the DCS andsimulator are connected, can be used for operatortraining and automation design and testingpurposes. Using this kind of connection betweena full-scale DCS and a large simulation modelrequires, however, a

triennial activity report 2000 - 2002 s-l. jämsä...

Documents