Alternative Strategies to Estimating Large-scale Travel Demand Models: the U.K. National Travel Model
E Petersen, RAND Europe, UK; H Gunn, HGA, UK
This paper will compare two different strategies that have been used to estimate the mode/destination demand model component of the U.K. National Transport Model (NTM).
This paper will compare two different strategies that have been used to estimate the mode/destination demand model component of the U.K. National Transport Model (NTM). One strategy has been based on adjusting an aggregate a-priori model designed to reproduce aggregate elastistities, to match observed aggregate travel demand from a national survey. The other has been to analyse the travel survey at a disaggregate level, retaining the possibility of adjusting parameters should implied elasticities appear indefensible.
The first part of the paper will briefly describe various national travel models and outline some similarities and differences with the NTM. The paper will highlight the most interesting features of the NTM and end with a section on how lessons learned from this re-estimation effort can be used to inform other comparable national transport models.
One key characteristic of the NTM is that it operates in two zonal systems ? a more detailed zone system for highway and transit assignments and a more aggregate zone system, the level at which the mode/destination demand model operates. This two-level system is imposed to a certain extent by the characteristics of the National Travel Survey (NTS), which is the primary data source for the demand component of the NTM. While more disaggregate information is available from the NTS, confidentiality requirements lead to certain restrictions being imposed on the release of this data. The paper will discuss the modelling techniques used to overcome these particular data restrictions. Given that many countries have limited data collection at the national level, or this information is restricted in various ways, there are clearly lessons to be learned from the NTM about modelling in a data-restricted environment.
Another important feature is that NTM was re-estimated with one eye towards testing certain policy measures, such as road pricing. This paper will provide details on how the models were designed to incorporate such policy tests, including how decisions were taken on the trade-offs between different features, including allowing an appropriate sensitivity to policy changes, ensuring the implied values of time were within acceptable ranges and model run time. While the details of the NTM are specific to the U.K. context, developing any national transport model will inevitably require taking decisions on which policies should be tested and which aspects of the model are most critical if different features come into conflict. Thus, the lessons learned from re-estimating the U.K. National Transport Model can be used by other practitioners developing their own national models in other national contexts.
This paper is designed to complement a paper focussing on the new NTM offered at the WCTR conference.
Association for European Transport