Modelling the Effects of Policies to Reduce the Environmental Impact of Freight Traffic in Great Britain
A Fowkes, D Johnson, A Whiteing, ITS, University of Leeds, UK
This paper reports the theoretical developments underlying version 4 of the LEeds Freight Transport model (LEFT), built as part of the EPSRC funded GREEN LOGISTICS project.
There is now increased interest in national level freight demand models. Partly, this arises from a concern to manage the balance of modes used, for example to avoid costly road building programmes or to reduce carbon emissions. Two elements are key to forecasting the effect of policy changes on the freight market. The first is the size of the total market, which will decline as freight generalised costs rise, but in a complex way. The second is the split of market by mode in the new situation, which requires us to know the length of haul distribution (since each mode will be relatively stronger at particular distances) and take account of road feeder movements to other modes.
This paper reports the theoretical developments underlying version 4 of the LEeds Freight Transport model (LEFT), built as part of the EPSRC funded GREEN LOGISTICS project. It does not look at growth of traffic over time, but purely at the effect of a policy (scenario) at a point in time. Everything is driven by generalised costs, and the paper describes how those for LEFT4 were derived. The first step is to strip out road collection and delivery traffic which is part of rail movements. The second step is to predict the market size effect, ie. the effect on the total traffic by all modes. This is done by stripping out mode shift effects from generalised cost elasticities derived from the literature. These are applied to base data (by mode, commodity and distance band) for both Tonnes and Tonne-Km. The third step is to sum figures by mode, and reassign the resulting totals over distance bands so as to give the forecast Tonne-Km figures from the forecast Tonnes figures (effectively using the implicit average length of haul to determine the new spread over distance bands). The fourth step is to split this traffic by mode, using a binary logit model calibrated for each commodity and distance band cell. The fifth step is to compute the new road collection and delivery trips associated with the forecast rail movements, and add them back in.
The paper will give detail and discuss various complications in carrying out the above, and report elasticities used etc. Usage of the developed model is discussed, including methods for converting first to vehicle kilometres and then to emissions. Hopefully, this work will be of interest and use to other freight modellers.
Association for European Transport