A mixed integer linear programming approach to forest utilisation management problems
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
The research reported here concerns the use of mathematical programming techniques to model resource flows in a system comprising industrial forests and subsequent wood processing and marketing activities. A review is made initially of methods to delineate management alternatives in industrial forests through formulating and solving linear programs. Linear programming techniques used to represent the associated Forest Management Problem (FMP) are discussed and solution methods analysed. The use of linear programming and mixed integer linear programming to represent resource flows in a problem where, in addition to forest management activities, the utilisation and marketing of wood based resources are also considered is explored in considerably more depth. Previous research on this latter class of problem, termed a Forest Utilisation Management Problem (FUMP), has been limited. Forest utilisation management problems may be characterised by the joint occurrence of standard Operations Research problems such as those of location, resource allocation, budget measures, and fixed charge specification. Mixed integer linear programming techniques appeared to provide a viable means to resolving FUMPs that pose non-convex programming problems. The possibility of redundancy in FUMPs is considered, and a technique to assist in preventing it is presented. The implications of redundancy to problem formulation are made apparent. Discussion also covers representational difficulties anticipated in certain components of FUMPs. A small test problem is discussed in relation to data requirements, matrix generation, and report writing. Recommendations concerning formulation and solution of FUMPs are made. Conclusions drawn relate to the feasibility of representing FUMPs as a class of mathematical program.