Extending the OTM to Predict Time Period Choice for Car Drivers



Extending the OTM to Predict Time Period Choice for Car Drivers

Authors

J Fox, B Patruni, ADaly, RAND Europe, UK; H Paag, TetraPlan, DK

Description

RAND Europe, working in collaboration with TetraPlan and Significance, were commissioned by the Danish Road Directorate to extend the existing ├śrestad Traffic Model (OTM) so that it is able to predict time of day choice for car drivers.

Abstract

RAND Europe, working in collaboration with TetraPlan and Significance, were commissioned by the Danish Road Directorate to extend the existing ├śrestad Traffic Model (OTM) for the Greater Copenhagen region so that it is able to predict time of day choice for car drivers. Congestion charging policies are currently high on the political agenda in Copenhagen, and there was a need to extend the OTM to be able to assess the effectiveness of policies where charging varies by time period in order to reduce congestion levels at particular times of day.

Historically, time period choice models have been developed from stated preference data. However, this was not possible within the timing and financial constraints for this project, and so instead the possibilities of developing models from purely revealed preference data were investigated. The results to be presented in the paper show that this approach can work for some travel purposes.

The starting point for the model estimations was existing models of mode and destination choice developed in 2006 that were extended to model time period choice. A total of nine time periods were defined, including two one-hour periods for the morning peak, and three one-hour periods for the evening peak. The OTM uses tour-based models, and so the time period choice alternatives are combinations of outward and return time period. The paper will describe how the time period alternatives were defined, taking account of the volumes of data available for each possible time period combination.

The model estimation was undertaken using household interview data collected in 2003 and 2005. The estimation strategy adopted was to seek to estimate the relative sensitivities of mode, destination and time period combination alternatives from the household interview data alone. Different structures for these three choices were tested, and assessed by reviewing the overall fit to the data, model elasticities, and fit to observed trip length distributions. For some travel purposes, it was possible to estimate relative sensitivities for mode and time period combination alternatives from the household interview data. These estimates were validated against evidence from stated preference studies (Hess et al. 2007). For other travel purposes, the relative sensitivity of mode and time period combination alternatives was imported directly from the stated preference evidence.

The paper will report on sensitivity tests made at the model estimation stage by making changes to travel times in a morning peak period, and assessing the impact on the predicted outward and return time period distributions. The paper will also report on testing of particular charging policies that will be made over the coming months when the new models are implemented within the OTM model forecasting system.


References

Hess, S., Daly, A., Rohr, C. and G. Hyman (2007) On the development of time period and mode choice models for use in large scale modelling forecasting systems, Transportation Research Part A: Policy and Practice, 41, pp. 802-926, November.

Publisher

Association for European Transport