(626d) An Integrated Simulation-Optimization Framework for Waste Minimization Analysis in Chemical Plant

Growing environmental problems have initiated technological advances to reduce the pollution impact from chemical processing plants. These include the use of process simulators such as AspenPlus, HYSYS, ProII, SuperPro and ChemCAD to design a process that is environmentally benign. While these simulators offer a convenient way of simulating the plant's performance without extensive experimentation or disruption of plant operations, currently they have limited innate capability to support waste minimization studies. A workshop jointly sponsored by the US Environmental Protection Agency, the Department of Energy, and AIChE's Center for Waste Reduction Technologies (CWRT) had indeed highlighted this limitation and suggested the coupling of expert systems with process simulators as a decision support for waste synthesis in chemical plant (Hilaly and Sikdar, 1996). However, to date no such development has been reported.

In this paper, we present an integrated framework that answers the above challenges. Our objective is to create waste minimization analysis tool by accessing the already available physical property and flowsheet information from the process simulator. This framework called ENVOPExpert has been developed using Gensym's G2 expert shell and consists of two elements. First, qualitative analysis is performed to identify the sources of wastes and to propose top-level alternatives for eliminating or minimizing them. These alternatives include reduction of byproducts formation and effective separation of useful components from the waste stream. Next, quantitative simulation are performed based on the qualitative results through interfacing between ENVOPExpert and HYSYS. This is done by capitalizing on the CAPE-OPEN capability of HYSYS process simulator. In this case, ENVOPExpert provides waste minimization decision support and uses XML data exchange and Microsoft Excel-VBA to access physical property, process chemistry, and flowsheet information from the HYSYS simulator. In this way, the effects of variable changes (pressure, temperature, heat input and feed composition) to the amount of waste generation in the plant can be investigated in HYSYS platform. At the same time, interpretation of the simulation results and the direction of the process variable changes (increase or decrease) can be searched automatically in ENVOPExpert platform with objectives of minimizing the environmental impact and the operating expenditure. Any synergies and trade-offs between different waste minimization alternatives can then be highlighted. At the same time, the resulting multiobjective optimization problem is also solved using meta-heuristic based simulated annealing method. We illustrate our integrated framework using an industrial case study involving methyl chloride production (Dantus and High, 1996). Comparisons of the results with the waste minimization suggestions identified by experts prove the potential of ENVOPExpert as the ?clean design? decision support tool to any CAPE-OPEN based process simulators.

Reference: Dantus, M.M. and K.A. High, Ind. Eng. Chem. Res. 1996, 35(12), 4566. Hilaly, A.K. and S.K. Sikdar, Chem. Eng. 1996 (Feb), 98.