Stochastic Multi-Objective Process Optimization by using the Composite Objective Function

摘要

This paper presents a novel computer-aided tool for multi-criteria optimizations of process flow sheets with larger numbers of uncertain parameters. Two ideas have been implemented in order to perform the optimization within a reasonable CPU time. The former consists of converting a multi-criteria into a single-criterion problem by applying a single sustainability objective function composed of the economic, environmental and social criteria, thus avoiding a time-consuming generation of multiple Pareto solutions. The latter relies on a reduced number of scenarios for Monte Carlo optimization of a process flow sheet. Rather than determining the required number of scenarios based on a given confidence limit, relative frequency and cumulative probability functions are generated by gradually increasing the number of scenarios. A new, cumulative distribution difference indicator has been proposed which measures a relative contribution of an additional scenario to the change of the cumulative function. When the value of this indicator approaches 0, a practically reasonable number of Monte Carlo scenarios has been incorporated in the optimization problem and a sufficiently accurate cumulative function has been generated.