Metaheuristics applied for storage yards allocation in an Amazonian sustainable forest management area

Abstract

In the sustainable management of Amazonian forests, it is essential to carry out the optimal planning of logging infrastructures to reduce costs and environmental impacts. However, there is a high degree of complexity due to the number of variables involved. Among these infrastructures, wood storage yards are of utmost importance as they directly influence the opening of forest roads and trails. The objective of this research was to evaluate the allocation of wood storage yards through exact solution and metaheuristics in a forest management area. The study area was a native forest under sustainable forest management regime located in the Brazilian Amazon. Three instances were formulated involving 5947 trees and 3172 wood storage yards facilities. We used a binary integer programming model solved by CPLEX and the metaheuristics Greedy Randomized Adaptive Search Procedure (GRASP), Tabu Search (TS), Variable Neighborhood Search (VNS) and Simulated Annealing (SA). GAP values increased as a function of instances. Although all metaheuristics obtained significant solutions with shorter processing times, only SA obtained feasible solutions in all executions for all three instances. In general, the metaheuristics were efficient in obtaining feasible solutions faster than CPLEX, which represents the feasibility of the planning of allocation storage large areas, and without significant losses of best-known solution. The SA presented the best performance in the three evaluated instances. Contribution of this study can be highlighted: evaluation of alternative computational methods for planning the allocation of wooden storage yards; evidence was obtained of effectiveness and efficiency of assessed metaheuristics and, the applicability of approximate methods in this problem was evaluated.