Integrated optimization of operational and tactical planning for log production
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Log merchandising, which involves scheduling of stand harvests, planning stem conversion strategies, and market planning for logs, are some of most important decisions made in the management of a resource. Much research attention has focused on developing modelling methodologies for components of log merchandising decision making. However, many planning approaches either do not have the capability to solve these decisions in an integrated manner, or are suitable for making integrated decisions, but only for single time period. Non integrated and single period models cannot be used for evaluating trade-offs between timing of stand harvest and stem conversion strategies in a flexible manner. As a Consequence, such systems can fail to identify merchandising strategies which offer the greatest opportunity benefits. There are two separate considerations in the integration of such plans. The first is the integration of planning across all temporal levels of decision making, while the other is integration of log production planning for all the stands in a forest estate. In this study, a planning framework was adopted which enables the log merchandising decisions to be solved in an integrated manner. The adopted modelling framework comprises three hierarchical levels: a medium term tactical model, a short term tactical model, and an operational scheduling model. The medium term tactical model is similar to the many harvest scheduling models in forestry literature, and addresses issues of sustainability of production, regulatory requirements, and preliminary scheduling of log production over a medium term horizon. The short term tactical model dissaggregates the medium term plan into detailed production plans, with stands, group of logging settings, or logging settings as the production units. The model also addresses issues of harvesting system availability, and the optimal location of landings. This model was not developed in the study, but a modelling approach recommended for solving this level is a mixed integer model based on facility location theory, interfaced to a Geographical Information System (GIS) for generating feasible logging layouts and for presentation of reports. The operational allocation model solves for optimal scheduling of harvesting systems to production units (settings), and also develops detailed log mix production and allocation plans. The problem is formulated as a mixed integer model, with the integer variables representing possible crew allocations, and is solved with commercial mathematical programming packages. The models are integrated with the hierarchical planning paradigm, in which separate models are developed for each decision level with higher level plans forming part of constraints to plans at the lower levels. A yield description system was developed, which enables the evaluation of log production plans to be conducted in an integrated fashion. The method involves estimating the yield potential of a stand by a few aggregated stem quality classes with a bucking model. The values of the log classes used to specify product preferences in the evaluation, are defined as those which yield maximum volume of a class, without leading to a reduction in the yield of higher quality classes. Planning of outturn from each stand is then determined by dissaggregating the log quality class into market log types with similar quality requirements, the formulation for which is built into optimization models for the various decision levels. In the determination of log mix to be produced, downgrading is employed as a marketing strategy and for balancing demand and supply by log assortments, a capability that is built into the models. A method was developed for controlling outturn during field implementation. The method involves the iterative revision of log specification variables for each stand, to determine new ones that lead to a satisfactory realization of log production targets. When used in field implementation, the new specifications should result in improved achievement of log production targets, and in reduced variability in the quality of material of each log assortment in a shipment. The modelling system was demonstrated and tested with an indicative case study. The results demonstrate the major advantages of the integrated planning system, which are the simplicity of model structure, and the feasibility of plans developed at the different planning levels. The yield description and outturn optimization system, when used with the outturn control method, provides plans that are feasible, and which can be used to manipulate the quality of log shipments in an easy and flexible manner. This capability has not been evident in planning approaches reported in the literature. The integrated log merchandising methodology developed in this study can be easily adapted for use with existing planning systems. There is need for further work on the development of the short term tactical model, in order for the framework to provide fully integrated plans. The method, however, requires further testing and comparison with alternative planning techniques, in order to fully document its potential.