A freight distribution exercise

Abstract

Current state-of-the-art freight transport models capture behaviour-level processes driven by decisions on logistics aspects (i.e. choice of transport chain) and micro-level terminal operations (i.e. crane operation). Yet, models with this resolution rely on data resources that are often scarce, as freight commodity surveys are costly and need to overcome confidentiality issues. This paper addresses this limitation and proposes a methodology to disaggregate interregional production-consumption matrices into facility-to-facility matrices as a preliminary process for the construction of an agent based freight model. The method takes the form of a set of transport distribution modelling approaches. This model is applied to wood-based products in New Zealand. We used the inter-regional tables from the latest National Freight Demand Study, information on the demand for logs from different industries, and open source geographic data to generate firm-to-firm OD tables. The method uses three stages. First, OD pairs from the interregional matrices are adjusted through iterative proportional fitting to match totals reported for every sector (Timber, Wood Panel, and Pulp production) and for every stage within the value chain of forestry-based products. Next, adjusted matrices, sector specific statistics and routed distances between facilities are used to calibrate spatial interaction models. New interregional matrices are generated upon the application of gravity models and are used to formulate a linear programming problem that distributes interregional flows amongst relevant facilities without compromising the original pattern of movements. The results provide a geographic representation of movements between forests, sawmills, panel-manufacturing plants, pulp-manufacturing plants, wood product retail stores and ports.