A Heuristic System to Solve Mixed Integér Forest Planning Models

Abstract

Implementing forest plans involves decisions regarding the geographic arrangement and timing of management activities, and the design of efficient transportation networks to provide access for these activities. These planning problems have been modeled successfully as mixed integer linear programs. However, these models are difficult to solve with commercial software due to their relatively large number of binary variables. We present a heuristic approach that has solved these problems well. The heuristic rules and a linear programming package interact within a closed system requiring no special user intervention. We describe several applications of this system for decision making in the USDA Forest Service.