Chemical process optimization applications in energy analysis of flowsheets and scheduling of batch processes.

摘要

Chemical process optimization, in its most general and global form, is a multivariable, multiobjective optimization problem. In practice, subproblems corresponding to simplified forms of this global problem are solved by considering a subset of the variables and objectives and formulating the problem accordingly. Two such process optimization problems have been studied, the first applied to continuous processes and the second applied to batch processes.; The first optimization problem deals with the energy analysis of process flowsheets for the extraction of stream data. This is important for synthesizing optimal heat exchanger networks from both energy and capital cost viewpoints. The stream data represent the complete energy requirements of the process specified in terms of supply temperatures, target temperatures and heat capacity flow rates of different process streams. An intelligent automated software system called INSPECT (INtelligent System for Process flowsheet analysis to ExtraCt stream daTa) has been developed and implemented. For the purpose of stream data extraction by INSPECT, several data extraction rules have been formulated in an if--then form based on data extraction principles and experience gained from analyzing flowsheets. INSPECT has been successfully applied for stream data extraction from some representative process flowsheets.; The second optimization problem considered addresses the optimization of batch processes based on an efficient time utilization of the existing batch equipment. An in-depth study has been performed of the simulated annealing approach to the scheduling of serial multiproduct batch processes under the assumption of a permutation schedule. Four versions of the simulated annealing algorithm have been studied based on two move acceptance criteria, the Metropolis algorithm and the Glauber algorithm, and two annealing schedules, the exponential schedule and the Aarts and van Laarhoven schedule. Of these four versions, the Metropolis algorithm with the Aarts and van Laarhoven annealing schedule is found to give the best results, with all four versions giving significantly better results than the IMS heuristic, the best currently available solution method.