Integrated solvent selection and solvent recycling under uncertainty.

摘要

The recovery of waste solvents is of great economic and environmental importance. Among different solvent recovery strategies, separation (batch or continuous) is widely adopted, where solvent selection, process synthesis and process design are three major steps in the set up of the recovery systems. This dissertation focuses on the integrated optimal design of these steps. The problem is formulated as a multi-objective optimization problem, the objectives of which are to decrease pollutant emissions and operating cost as well as increase process operability.; In order to solve this complex multi-objective problem, a hierarchical improvement in genetic algorithm is proposed here. The computational efficiency of genetic algorithms (GA) is improved by capitalizing on the uniformity property of Hammersley sequence sampling (HSS) technique, resulting in efficient genetic algorithm (EGA). Moreover, another variant of GA, stochastic genetic algorithm (SGA) is developed by using the HSS technique to reduce the number of samples for sampling error, leading to the Hammersley stochastic genetic algorithm (HSGA). To deal with the integrated solvent selection and recovery process design, multi-objective efficient genetic algorithm (MOEGA) is developed based on the HSS incorporated GA. Recovery process simulation is implemented in the chemical process simulation software, Aspen Plus 12.1. In order to solve real world case studies the optimization framework is set up to use the P-graph and residue curve maps (RCMs) techniques to generate various process synthesis and design alternatives. The coupled MOEGA-ASPEN framework is then applied to a real world case study---recycle acetic acid from its aqueous solution.; The results reveal that the newly developed algorithms improve on both efficiency and robustness. New solvents have been found and the alternative waste recovery process brings more flexible process operations. The Pareto set calculated from the multi-objective framework provides more representative choices among conflicting objectives for decision makers.