Stochastic forestry planning under market and growth uncertainty

Abstract

The forest planning problem with road construction consists of managing the timber production of a forest divided into harvest cells for a given planning horizon. Subject to uncertainty, it becomes a complex large-scale multi-stage stochastic problem expressed through scenarios. A suitable algorithm for these problems is progressive hedging (PH), which decomposes the problem by scenarios. A two-phase solving approach, in which PH is used as a heuristic method to obtain a directly optimized restricted model with fixed variables, is implemented. Multiple adjustments to improve the performance of the method are adopted and tested in a tactical case study. The performance of the proposed method is compared with those of traditional approaches. Thanks to these enhancements, we solved a real original problem including all the complexities of a practical problem not addressed in previous studies. Comprehensive computational results indicate the advantages of the method, including its ability to efficiently solve instances of up to 1000 scenarios by exploiting its parallel implementation.