A comparative analysis of MIP formulations for the multi-mode resource-constrained independent task scheduling problem

Abstract

We consider the problem of scheduling independent multi-mode tasks with multiple renewable resource constraints, aiming to minimize the weighted sum of tardinesses. This problem commonly arises in the context of multiproject scheduling. We propose three pseudo-polynomial mixed-integer programming formulations and conduct a theoretical comparative analysis of their respective linear programming relaxations. We present the results of extensive computational experiments that provide evidence that a formulation based on Dantzig-Wolfe reformulation exhibits very good empirical performance consistently solving all instances with fewer than 30 tasks. In addition, it achieved an average success rate of 94.28% for instances with tasks ranging from 40 to 100, compared to other formulations that on average achieved success rates of 64.28% and 58.57%.