Multi-criteria Pareto optimised planning method for distillation plants with non-sharp splits

Kausch C., Preuss M., Henrich F.

Abstract
The investigation of diverse structural alternatives and modes of operation of a distillation plant results in highly complex search topologies including multi-modalities and non-convexities, making very flexible (e.g. evolutionary) optimisation algorithms necessary to maintain diversity in the Pareto sets, i.e. to assure that there are enough different solutions with clearly differing objective function values. On the other hand, virtually identical solution sets can be found which stem from utterly different configurations, hence it is possible to take into account additional decision factors besides the pair of objective functions already optimised. Thus even greater flexibility is possible in the design and planning phase. With the method described in the paper (multiple runs of a standard EMOA combined with a rigorous simulation) we analyse different combinations of conflicting objective functions and based on the respective sets of optimal solutions, different configurations and operation modes are discussed with respect to the variables of the separation task. In a further step we investigate a distillation with regard to key cost factors by means of scenarios and analyse the impact on the solution space and on the attained Pareto sets and discuss to what extent a Pareto set with a suitable choice of objective functions already contains information regarding the sensitivity to key cost factors. We thus show the potential of the described approach in the initial phase of such a project and underline the necessity for optimisation based decision aids for the planning engineer and specifically the benefit gained by the multi-criteria Pareto approach.

Keywords
Distillation; Evolutionary algorithms; Exergy; Non-sharp splits; Pareto optimisation