Investigation of a ‘proof-of-concept’ operational level forest biomass supply chain decision support system

The objective of this doctoral thesis was to develop a “proof-of-concept” operational level decision support system (DSS) to assist forest planners to supply their customers’ energy requirements for the least delivered cost using forest biomass (particularly logging residue) stored at roadside.

Worldwide interest in use of forest biomass as an energy source has increased recently as a potential means to reduce global greenhouse gas emissions. Australia is one of a group of countries with sufficient forest resources for use of forest biomass for energy production to make a substantial contribution to reducing its GHG emissions, but, unlike some other countries in this group, Australia has made little use of this resource. Wood processing waste (offcuts, sawdust and shavings) is typically the cheapest forest biomass source as it has already been transported from the forest and been comminuted to some degree. As a consequence, the majority of this material is already being used for energy production or other high value uses and is not considered further in this thesis. The focus of this thesis is the use of logging residue for fuel, as it is the major underutilised forest biomass source both in Australia and globally. Logs and whole trees are also considered as potential energy sources as sizeable quantities may be available in Australia during periods of reduced demand for particular wood products and from stands damaged during natural disasters.

Forest biomass is characterised by low energy, bulk and spatial densities and high spatial and temporal variability in availability, which increases its delivered costs relative to other fuel sources. The characteristics of forest biomass, combined with the need to minimise its delivered costs, have resulted in a plethora of published models exploring aspects of forest biomass supply chains. Forest planning is generally divided into a hierarchical structure from strategic to tactical then operational planning with increasing detail and spatial precision with progression along the hierarchy. Most forest biomass supply chain models have been produced overseas for different species, weather conditions, and seasonal energy requirements than in Australia and are targeted at strategic and tactical levels. A search of the published literature did not find any operational level forest biomass supply chain DSS based on Australian species and conditions.

As operational level planning involves converting high-level plans to on-the-ground activities, decisions at this level can have a profound impact on delivered costs of forest biomass. Transport costs are usually a major contributor to forest biomass delivered costs which has resulted in many cost reduction strategies focussing on this area. Although drying forest biomass prior to transport from the forest to the customer or intermediate depot can reduce its transport costs through weight reduction and increases in its net calorific value, no operational level DSS incorporating forest biomass drying prior to transport were found in the published literature.

Development of the DSS in this thesis required a number of knowledge gaps to be filled, in particular, forest biomass drying models for Australian tree species and weather conditions and the interactions between forest biomass drying and costs associated with its storage, transport and chipping. The methodology used to fill these knowledge gaps involved synthesising the results of two field experiments and two desktop simulation studies conducted using a tactical forest planning DSS based on a linear programming solver to develop the models and relationships underpinning the DSS.

The first field experiment compared two plantation harvesting techniques to examine their impact on the cost of transporting logging residues to roadside. Piling residues during forest harvest (the fuel-adapted harvesting method) was found to reduce logging residue primary transport costs by 28% and increase the proportion extracted by 26%, but resulted in a 15% increase in log harvest and primary transport costs.

Many published forest biomass natural drying models have been developed from portable weather station data, however, the cost of deploying and maintaining multiple weather stations across a forest estate would be prohibitive. The second field experiment developed drying models relating weather variables from multiple sources with the weight loss of a pile of Eucalyptus nitens logs. While weather data sourced from portable weather stations produced the most accurate drying model, the model produced using weather data from the nearest Bureau of Meteorology weather station was comparable in accuracy to that from the portable weather station data with the advantage that the data are available for free online.

The desktop studies quantified the delivered cost savings achievable through drying logging residues (≥28% cost savings) or logs (≥22% cost savings) stored at roadside prior to transport by truck to the customer. The reduction in bulk density resulting from drying required the use of trucks with a higher volumetric capacity than commonly used in the Australian forest industry to gain the maximum returns from roadside drying, as lower capacity trucks reached their volumetric capacity well before reaching their weight capacity. The studies also concluded that storage costs and losses of material during storage were minimal over typical operational forest planning timelines (2-3 months).

The development of the operational level DSS compared four mathematical models to develop potential least-cost solutions (Generalised Reduced Gradient non-linear (GRG), Evolutionary, Greedy and Linear Programming (LP)). The models were compared in terms of their performance using measures commonly applied to forest supply chain mathematical model comparisons (solution speed, delivered cost and, for the heuristic methods, the percentage difference between the optimum solution cost and the heuristic solution cost) and in terms of whether solutions could be practically implemented. The LP and Greedy methods gave the best overall results. The GRG method solutions were not implementable. In terms of practical implementation, the Evolutionary method solutions were similar to those from the LP and Greedy methods but had a significantly higher delivered cost (25%) resulting from a higher delivered MC.

The addition of a cost penalty ($3000 per chipper move) to discourage movement of chippers between harvest sites improved the Greedy method solutions by reducing the number of chippers required and chipper moves, attributes valued by operational forest planners.

The development of the operational forest biomass DSS showed the potential for savings of ≥14% in forest biomass delivered costs that could be achieved through the use of the DSS to assist forest planners to schedule deliveries of forest biomass to customers. Use of the DSS would also aid the development of a new industry for Australia, utilising what is currently a waste product. Undoubtedly obstacles remain that will limit the savings made in practice, such as the quality of the data describing the roadside forest biomass piles and the need for linkages to existing forest management systems to facilitate data flows to and from the DSS. These obstacles and the research required to continue development of the DSS and similar forest biomass planning tools and models are discussed in the thesis.