Optimal Parameter Settings for Solving Harvest Scheduling Models with Adjacency Constraints

Abstract

Optimal parameter settings to improve the efficiency of solving harvest scheduling models with adjacency constraints were examined using Ilog’s Cplex® 11.2 optimizer tuning tool.  A total of 160 randomly generated hypothetical forests were created with either 50 or 100 stands and four age-class distributions.  Mixed integer programming problems were formulated in Model I form with four different adjacency constraint types, two Unit Restriction Model (URM) adjacency constraints (Pairwise and Maximal Clique) and two Area Restriction Model (ARM) formulations (Path and Generalized Management Unit).  A total of 640 problem sets-where a set is a common forest size, age-class distribution, and adjacency constraint type-were tuned to determine optimal parameter settings and then were solved at both the default and optimal settings.  In general, mean solution time was less for a given problem set using the optimal parameters compared to the default parameters.  The results discussed provide a simple approach to decrease the solution time of solving mixed integer forest planning problems with adjacency constraints.