On potential of interactive multiobjective optimization in chemical process design

Abstract

Interaktiivinen monitavoiteoptimointi soveltuu hyvin teollisen kemian prosessisuunnitteluun. Jussi Hakanen esittelee väitöskirjassaan menetelmään pohjautuvan uuden työkalun, jollaista ei aiemmin ole ollut tarjolla. Hakanen on soveltanut työkalua onnistuneesti useiden kemiallisten prosessien suunnitteluun huomioimalla uudella tavalla enemmän kuin kaksi prosessin käyttäytymiseen vaikuttavaa tavoitetta.Aikaisemmin kemian prosessien suunnittelussa on otettu huomioon enintään kaksi prosessin toimintaan vaikuttavaa ristiriitaista tavoitetta. Hakasen esittelemä lähestymistapa ei rajoita huomioitavien tavoitteiden määrää, mikä mahdollistaa prosessien aiempaa realistisemman suunnittelun. Väitöstyössä esiteltyjä suunnittelutehtäviä ovat muun muassa paperinvalmistusprosessin vesien kierrättäminen ja lämmön talteenottojärjestelmän suunnittelu sekä sokeriteollisuudessa tärkeän fruktoosin ja glukoosin erotusprosessin suunnittelu.Teollisen kemian prosessisuunnittelu, kuten reaalimaailman optimointiongelmat yleensä, sisältää tyypillisesti useita keskenään ristiriitaisia tavoitteita, jotka halutaan saavuttaa samanaikaisesti. Jotta prosessille saadaan paras mahdollinen suorituskyky, täytyy ristiriitaisten tavoitteiden välille löytää sellainen kompromissi, joka ottaa parhaiten huomioon kyseisen prosessin erityispiirteet.

In this thesis, we study the potential of interactive multiobjective optimization (MOO) in solving chemical process design problems with several conflicting objectives. When designing real-world chemical processes, there are several performance criteria that need to be considered simultaneously, for example, economical aspects, environmental impact and process operability to name a few. These aspects are often conflicting and, therefore, the process considered can not be optimal with respect to each criteria at the same time. This means that some compromise must be made.In this thesis, interactive MOO is used to help the designer in finding the best compromise between the conflicting performance criteria. A compromise in MOO is called a Pareto optimal solution and it means that the performance of the process can not be improved with respect to any criteria without impairment in some other criteria. Previously, chemical process design problems have been usually solved with two performance criteria at most. The approach described in this thesis is not restricted by the number of performance criteria as is often the case in methods used by chemical engineers. Therefore, process design problems can be considered in their truly multiobjective character without unnecessary simplifications.A process design tool based on IND-NIMBUS, an implementation of the interactive MOO method NIMBUS®, has been developed in this thesis. The IND-NIMBUS process design tool developed consists of three parts: a modelling tool, an optimizer and a graphical user interface. The modelling tool produces a numerical model of the process considered. The optimizer consists of NIMBUS® and some suitable single objective optimizer that finally produces new Pareto optimal solutions by solving the single objective subproblems produced by NIMBUS®. Different modelling tools and single objective optimizers can be used depending on the problem to be solved. The graphical user interface enables the designer to input his/her preferences during the interactive solution procedure as well as comparing the Pareto optimal solutions obtained with different types of visualizations. The preference information is used to guide the search of the most preferred solution by generating desirable Pareto optimal solutions.Some preliminary ideas are also introduced to aid the designer in the decision making procedure. In this thesis, trade-off information is studied as a way of supporting the designer. The idea is to guide the designer towards the most preferred solution by showing him/her tradeoff information during the interactive solution procedure. Trade-off information can help the designer in realizing what kind of compromises could be available. This information is also beneficial in convincing the designer that (s)he has found the best compromise solution available.To test these ideas in practice, several industrial process design problems related, for example, to paper making and sugar industries were solved. Previously, these problems were considered in a simplified formulation by including only one or some of the important criteria. Here, these problems were considered in a totally new way with more than two important performance criteria included. The IND-NIMBUS process design tool was applied to these problems and all the problems were solved with the help of real designers who were experts in the specific areas of the applications considered. The results obtained were promising and the designers found the interactive solution procedure easy to understand and use. To summarize, we showed that interactive MOO can be succesfully applied to solving chemical process design problems in their true multiobjective character and our approach turned out to be efficient in practice.