A multistage stochastic programming approach for short-term scheduling of batch processes under type II endogenous uncertainty

Abstract

In this study, we present a novel multistage stochastic programming approach for scheduling of batch operations under type II endogenous uncertainty (where time of uncertainty realization is model dependent). The proposed multistage framework follows a node-based formulation and enforces non-anticipativity implicitly. The key novelty of this approach is that it does not require any auxiliary binary variables or explicit non-anticipativity constraints (NACs). The proposed framework is validated using three different case studies: two case studies adapted from the literature and an actual large-scale industrial case study. Computational studies were conducted and VSS (value of stochastic solution) was estimated for all case studies. It was observed that with an increase in number of stages, the VSS also increases ranging from 4.4% to 11% for the industrial case study. We conducted a comparison study of the proposed approach with an approach involving binary variables to define NACs as most studies available in the literature use auxiliary binary variables to define NACs when type II uncertainties are involved. The results shows up to 85% reduction in the computational time while using the proposed node-based approach in comparison to using an approach that requires binary variables to define the NACs.