QUALITATIVE REASONING AND OPTIMIZATION OF CHEMICAL-ENGINEERING PROCESSES

A qualitative model can be developed by two main ways; by abstraction of quantitative equations or by using e.g. a flowsheeting program to evaluate derivatives, from which the qualitative model is generated by using either direct modeling or linear synthesis. Two different modeling shells for qualitative model development are described. An application to a qualitative decision support system is represented. For qualitative model simplification the following methods are proposed: ignoring equations, partial modeling and linearization. The number of solutions from a qualitative model can be restricted by using a qualitative question or a projection. Spurious solutions can be eliminated by noncausal, causal and functional constraints. The principle of qualitative optimization is presented. Qualitative optimization is defined as an activity striving to improve the value of objective function by the methods of qualitative reasoning. Also an extension to qualitative vector optimization is described. An approach to flexible semiqualitative reasoning is described which can accept several types of data. Same equations can be used for both quantitative calculations and qualitative reasoning. Evaluation rules are used to evaluate the type (accuracy) of data after every operation.