The current drive toward ecosustainability has provided a strong impetus to implement waste minimization within the chemical industries. However, conducting a waste minimization analysis is expensive, time-consuming, laborious, and knowledge-intensive. In part 1 of this series [Halim, I.; Srinivasan, R. Systematic Waste Minimization in Chemical Processes: Part I. Methodology. Ind. Eng. Chem. Res. 2002, 41 ( 2), 196], we described a systematic methodology based on the Environmental Optimization (ENVOP) technique for waste minimization analysis of chemical plants. In part 2 [Halim, I.; Srinivasan, R. Systematic Waste Minimization in Chemical Processes: Part II. Intelligent Decision Support System. Ind. Eng. Chem. Res. 2002, 41 (2), 208], we reported on ENVOPExpert, an expert system that implements the methodology. While the focus in parts 1 and 2 were on continuous processes, in part 3 we extend the waste minimization methodology to batch process plants and describe its implementation as an expert system called BATCH-ENVOPExpert (BEE). In contrast to continuous processes, during batch processing different wastes may be generated at different times even from the same process unit. The process flow diagram that serves as the backbone for ENVOP analysis therefore does not completely represent the waste generation mechanisms, and the process recipe has to be analyzed to both diagnose the source of waste generation and identify process modifications that will eliminate it. In this paper, we extend the process graph based waste diagnosis approach described in parts 1 and 2 for continuous processes to batch processes by incorporating knowledge and operating procedures of the process recipe, which can be modeled using Grafcets. A process graph (P-graph) model of the operation can then be generated by considering the flow diagram and the Grafcet of the process. This P-graph can be analyzed to diagnose the waste sources and heuristic-based waste minimization solutions can be derived at a high level. Specific variable-level solutions can also be implied by combining cause-and-effect knowledge and functional models of the process. We illustrate the approach by performing waste minimization analysis on an industrial herbicide manufacturing case study and compare our results with the available experts' solutions.