Goal programming in forest management under parametric certainty and uncertainty

BROZ, DIEGO RICARDO; DURAND, GUILLERMO; ROSSIT, DANIEL ALEJANDRO; ROSSIT, DIEGO GABRIEL; FRUTOS, MARIANO; TOHMÉ, FERNANDO ABEL

Durban

Congreso; XIV Congreso Forestal Mundial; 2015

Organización de las Naciones Unidas para la Alimentación y la Agricultura (FAO)

Development processes based on the exploitation of forest plantations are characteristic of several regions in South America. The adequate management of this kind of plantations is thus fundamental for the future of those regions. It involves multiple goals, namely: economic, environmental and social ones. Mathematical programming is a powerful tool for the solution of these problems. Here we present an extended goal programming model (EGP) for a tactical planning (18 years) in which the goals are: 1) the maximization of the present value of production; 2) the maximization of carbon storage in forests; 3) to balance inter annual harvesting; 4) to balance inter annual transport; and 5) to balance inter annual harvesting areas. The feasible plans obey to silvicultural spatial and physical constraints. As a testbed we consider a forest of 40 stands of Pinus taeda L., and an industrial facility which includes a pulp mill, a MDF factory, a sawmill and a plywood factory. Through the simulation of the forest´s growth, the partial volumes of tree bole were generated according to industry requirements. For the optimization, was used GAMS 24.1.3 and CPLEX 12.1 as deterministic solvers. The model showed a better performance in mixed and weighted approaches. We found that the cash flow rate of returns has a significant impact on the harvest schedule and the present value. Since uncertainty affects economic variables in the long term, an EGP stochastic two-stage model was proposed to handle the uncertainty in the cash flow rate. The model was implemented in a two-stage modeling framework (GAMS/EMP 24.1.3) and solved by the commercial software LINDO 8.0.1. The results from both approaches, deterministic and stochastic, did not show significant differences. However, the stochastic approach yields more robust solutions for uncertain futures.