A column generation algorithm for tactical timber transportation planning

Abstract

We present a tactical wood flow model that appears in the context of the Canadian forestry industry, and describe the implementation of a decision support system created for use by an industrial partner. In this problem, mill demands and harvested volumes of a heterogeneous set of log types are given over a multi-period planning horizon. Wood can be stored at the forest roadside before delivery at a financial cost. Rather than solve this as a network linear programme on the basis of out-and-back deliveries, we choose to model this problem as a generalization of a log-truck scheduling problem. By routing and scheduling the trucks in the resolution, this allows us to both anticipate potential backhaul opportunities for cost and fuel savings, and also minimize queuing times at log-loaders, management of which is a major concern in the industry. We model this problem as a mixed integer linear programme and solve it via column generation. The methodology is tested on several case studies.

Keywords:

• forestry
• vehicle routing
• decision support system
• integer programming
• column generation

Acknowledgements

This work falls in line with Theme 4 of the NSERC Strategic Research Network on Value Chain Optimization (VCO). The authors would like to thank the VCO Network for their collaborative efforts, and Mitacs and the FQRNT for their financial support. In addition, we would like to thank Dave Lepage from FPInnovations for his contributions in development and testing.