New effective MILP models for PFSPs arising from real applications

Abstract

Permutation flow shop problems (PFSPs) with makespan minimization that model production lines working in industry often have some special features: they are typically large-scale and the jobs can be sorted into types so that jobs of the same type have equal processing time values at each machine. We define the related R-PFSP, the Permutation with Repetition Flow Shop Problem, which is of less complexity if the number of types is bounded. Moreover, a subproblem set of R-PFSPs, the RL-PFSP is considered too, where only those permutations are in the design space in which subsequent tuples of a certain size contain jobs of the same type. We construct adequate new MILP models for R-PFSPs and RL-PFSPs and investigate their effectiveness experimentally. We demonstrate that via our new MILP models significantly larger problems can be solved than via the classical MILP models.