The objectives of this work was to develop and test an algorithm based on Simulated Annealing (SA) metaheuristic to solve problems of forest management with integer constraints. The algorithm SA developed was tested in five problems containing between 93 and 423 decision variables, periodically subject to singularity constraints, minimum and maximum production.The problems had the objective of maximizing the net present value. SA was codified into delphi 5.0 language and the tests were performed in a microcomputer AMD K6II 500 MHZ, with RAM memory of 64 MB and hard disk of 15GB. The SA performance was evaluated according to the efficacy and efficiency measures. The different values or categories for the SA parameters were tested and compared in relation to their effects on the algorithm efficacy. The selection of the parameters' best configuration was performed by using the L&O test at 1% probability and analyses via descriptive statistics. The parameters' best configuration provided for SA average efficacy of 95.36%, minimum value equal to 83.66%, maximum value equal to 100%, with coefficient of variation of 3.18% of the mathematical optimum, obtained by the exact algorithm branch and bound. As for the larger problem, the efficiency of SA was ten times superior to the efficiency of the exact algorithm branch and bound. SA came out as a quite attractive new approach in forest management for solving important problems of difficult solution through the use of the current computational instruments.
Forest management; metaheuristics; Simulated Annealing
