03:30 PM - 03:55 PM
Solving a Multi-Period Log-Truck Scheduling Problem with Column Generation
We present here a multi-period routing and scheduling problem that arises in the Canadian forestry industry. The problem is modeled as a mixed integer linear program and solved via column generation with the the columns, representing log-truck routes, generated through the resolution of resource constrained shortest path problems.
03:55 PM - 04:20 PM
Demand-Driven Harvest Scheduling
We seek harvest teams schedules for a year that will meet request for mills, without losing sight of the other activities in the value chain (transportation, inventory management).We develop/tactical operational planning models to organize these activities for a year. Every formulation can be solved in a day.
04:20 PM - 04:45 PM
Construction and Relocation of Variable Sized Logging Camps in Forestry Industry
Based on logging demands for the next five years, we investigate the optimal number, locations and sizes for new camps in order to host workers involved within logging activities and balance transportation and camp construction costs. The problem includes a very detailed cost structure and is tackled by mathematical programming.
04:45 PM - 05:10 PM
Cut-to-Length Based Wood Procurement Planning Model: A Hybrid Approach
In this paper we develop three MIP models for a wood procurement-planning problem based on cut- to-length (CTL) bucking system, for the Canadian context of operations where harvesters are not equipped with computers. This important problem in forest management is difficult to solve since it integrates the bucking problem and the multi-commodity supply planning problem, while minimizing a combined non linear harvesting cost (the harvesting cost increases non linearly with the number of harvested products) and an aggregated transportation cost.
Each model was used to evaluate a different harvesting scenario: the first scenario is to apply one bucking pattern to each stand, the second is to apply a bucking pattern for each sector (a group of stands predefined by the forest company) and the final one is to apply a bucking pattern for each species. The aim of these scenarios is to explore the effects of the harvesting system structure on the harvesting cost. These scenarios allow investigating the gains and losses that could arise from the use of aggregation in the harvesting structures.
We develop a hybrid approach based on large neighborhood search, tabu search heuristic and linear programming for solving this large-scale mixed integer programming problem.