(286e) Combining Expert System with Process Simulation for Waste Minimization in Batch Plant Operation

The issue of environmental sustainability has prompted the batch chemical industries to switch from end-of-pipe treatment to waste minimization as their top priority in combating pollution. However, the multi-scale nature of batch operation has posed challenges in implementing waste minimization. First, batch process often involves chemistry that is complex and poorly understood. As a result, waste minimization options through byproduct suppression or recycling of valuable components may be overlooked. Second, as batch operation delivers its product in discrete amounts, the waste flowrate, composition and other properties could vary considerably with the execution of each operation within the cycle time. This needs to be accounted for when formulating a strategy for minimizing waste. Third, the substantial involvement of process operators would make batch operation highly susceptible to unplanned waste discharges caused by human errors. All these factors have motivated us to develop a methodology, which is robust and efficient in identifying waste minimization opportunities in the batch plants.

Previously, we have described a methodology for waste minimization in continuous processes [1]. An automated waste minimization support tool, called ENVOPExpert, has also been implemented based on the methodology [2, 3]. ENVOPExpert has been tested on a number of industrial-scale case studies including a hydrocarbon separation process, a chemical intermediate manufacturing process, a hydrodealkylation process of toluene to benzene and an acrylic acid production process. In all case studies, ENVOPExpert has been found to identify almost all the waste minimization suggestions identified by the human expert and some others not mentioned in the human's results.

In this paper, we extend the waste minimization methodology to batch processing environment. The fundamental difference between the batch and continuous process operation necessitates new developments of the underlying knowledge representation and inference schemes that are already implemented in the ENVOPExpert. The new methodology called BATCH-ENVOPExpert has been developed by interfacing between G2 (Gensym) expert system shell and HYSYS simulator. BATCH-ENVOPExpert comprises of heuristics methods and rules to trace the origins of each waste material and highlight the qualitative waste solutions. Whereas HYSYS simulator provides modeling and simulation capabilities at the process variable level to derive most practical as well as cost effective solutions. To derive qualitative solutions, we implement P-graph as representation of the waste-material flow in the process. Through P-graph, the operating procedures, streams and unit operations that produce wastes will be tracked down. This is achieved by tracing backwards starting from the waste streams and upstream to the whole network of the batch operating procedure, to find materials which contribute to the waste streams. Once the waste origins are identified, top-level alternatives to eliminate them are then proposed using design heuristics. Subsequently, detailed alternatives are proposed through quantitative simulation and this is done by interfacing between BATCH-ENVOPExpert and 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 BATCH-ENVOPExpert platform. As a result, any synergies and trade-offs between different waste minimization alternatives can be highlighted. We test our methodology on a case study involving a biodiesel production. The findings show that the framework is able to provide guidance to the non-expert in conducting waste minimization analysis to the plant.

References:

[1] I. Halim, and R. Srinivasan, Systematic Waste Minimization in Chemical Processes: Part I. Methodology, Industrial Engineering and Chemistry Research 41(2), 196-207, 2002 [2] I. Halim, and R. Srinivasan, Systematic Waste Minimization in Chemical Processes: Part II. Intelligent Decision Support System, Industrial Engineering and Chemistry Research 41(2), 208-219, 2002 [3] I. Halim, and R. Srinivasan, An Integrated Decision Support System for Waste Minimization Analysis in Chemical Processes, Environmental Science and Technology 36(7), 1640-1648, 2002