CORMSIS seminar by Alistair Clark Event

Production Lot Sizing and Scheduling with Non-Triangular Sequence-Dependent Setup Times

Many production lot sizing and scheduling problems feature non-triangular sequence-dependent setups. When setup times are triangular then there is always an optimal lot sequence with at most one lot (AM1L) per product per period. However, non-triangular setup times imply that it may be optimal for a "shortcut" product to be produced in more than one lot within the same period, thus breaking the AMIL assumption present in much research. This paper formulates a new suitable model that also excludes all prohibited subtours a priori using a polynomial number of constraints. Computational tests demonstrate the effectiveness of the ML model.