Surrogate Modeling for Continuous-time-formulation in scheduling Fall 2024 UROP

Abstract

Scheduling in multipurpose batch chemical plants is a challenging task due to the complexity of shared resources and interdependence in production. Continuous-time scheduling models provide an effective framework for addressing these challenges, particularly in scenarios involving variable processing times. In this project, a slot-based state-task network (STN) continuous-time scheduling model from Sundaramoorthy and Karimi was implemented. Furthermore, backward propagation tightening methods from Merchan et al. was integrated to improve computational efficiency by introducing demand-based valid inequalities. Using Julia and the Gurobi optimizer, the method was applied to a classic state-task network problem, demonstrating the practical benefits of constraint tightening in reducing computational times for large-scale scheduling problems.