Primal Heuristics for Branch-and-Price Algorithms

Abstract

We study a cement transportation problemwith asset-management (equip- ment selection) for strategic planning scenarios involving multiple forms of trans- portation and network expansion. A deterministic mixed integer programming ap- proach is difficult due to the underlying capacitated multi-commodity network flow model and the need to consider a time-space network to ensure operational feasi- bility. Considering uncertain aspects is also difficult but important. In this paper we propose three approaches: solve a deterministic mixed integer program optimally; solve stochastic programs to obtain robust bounds on the solution; and, study alter- native solutions to obtain their optimal cost vectors using inverse programming.