Solving the integrated forest harvest scheduling model using metaheuristic algorithms

In forestry, the highest operational costs arise from the construction of forest roads and the transportation of harvested wood. Hence, optimization models have been used at the tactical level of planning to reduce these costs by integrating decisions on: (1) the allocation of harvest-blocks, (2) the allocation of access roads to these blocks, and (3) the transportation costs that result from the latter two decisions. The integration of these three decisions, in one optimization model, has been referred to as the integrated model. The integrated model, when binary decision variables are used to represent the cut-blocks and roads, is NP-hard and has been solved using two approaches: exact and metaheuristic algorithms. Unlike exact methods, metaheuristic algorithms have thus far not solved the integrated model, but have solved models which either exclude transportation costs from the objective function, or solve the model sequentially. This is a significant gap in prior research because exact solution methods can only be used on smaller forests and metaheuristic algorithms have therefore been used to solve the tactical forest planning problem, without the integration of transportation costs, on large forests. This failure to integrate transportation costs, on a large scale, is the major economic consequence of this gap. The objective of this paper is to present and evaluate a new solution procedure in which all three elements of the integrated tactical planning model are included in the objective function and solved using metaheuristics. The solution procedure was applied to three forests and the attributes and qualities of the solutions were compared to near-optimal solution values generated using an exact solution approach. The results indicate that this metaheuristic procedure generated good quality solutions. We conclude that this research is a useful first step in representing transportation costs in the integrated tactical planning models to be solved using metaheuristics.