A computational study of two-period relaxations for lot-sizing problems with big-bucket capacities

The multi-item Capacitated Lot-sizing problem with Setup Times (CLST) is an important problem from both a theoretical and a practical perspective. This talk is part of a research stream that studies two-period relaxations of CLST. We present computational experiments that investigate the strength of valid inequalities that are derived from two-period relaxations. Four families of valid inequalities are considered, all of which are generalisations of cover inequalities, as described in Padberg et al. We present a numerical study in which we compare the strength of these inequalities with the (l, S) inequalities of Barany et al. We find that, for certain instances, some families are very efficient, and are able to improve the lower bound by a great margin.