Two New Continuous-Time Models for the Scheduling of Multistage Batch Plants with Sequence Dependent Changeovers

This paper presents two new multiple-time grid, continuous-time mixed integer linear program (MILP) models
for the short-term scheduling of multistage, multiproduct plants featuring equipment units with sequence
dependent changeovers. Their main difference results from the explicit consideration of changeover tasks as
model variables rather than implicitly through model constraints. The former approach is more versatile in
terms of type of objective function that can be efficiently handled (minimization of total cost, total earliness,
and makespan), and despite generating larger mathematical problems, it is also a better performer in single
stage problems. The latter is better suited for multistage problems, where the former approach has some
difficulties even in finding feasible solutions, particularly as the number of stages increases. The performance
of both formulations is compared to other MILP and constraint programming models. The results show that
multiple-time grid models are better suited for single stage problems or, when minimizing total earliness, that
the constraint programming model is the best approach for makespan minimization and that the continuous-time model with global precedence variables is the best overall performer.