Developing a Model of Travel Time Variability for Congested Motorways

Developing a Model of Travel Time Variability for Congested Motorways


C Drury, UK Highways Agency, UK; J Bates, I Black, J Fearon, C Gilliam, Hyder Consulting, UK


This paper describes a study for the UK Highways Agency to develop a model of travel time variability on congested motorways, which includes a representation of ?flow breakdown?. The work builds on a rich set of data collected on the M6 motorway.


The topic of Travel Time Variability (TTV) is receiving increasing attention in a number of countries. However, while some progress has been made on the valuation of TTV, significant problems remain in predicting the levels of TTV will occur in given circumstances. If the TTV benefits of improvements are to be incorporated in the appraisal process, it will be essential to be able to reflect the impact of policies on TTV.

Motorways are particularly susceptible to TTV under congested conditions. Whereas under normal operating conditions they offer much shorter travel times than the standard road network, they are more vulnerable to capacity problems because of their limited entry and exit points. Information systems can alleviate these problems to some extent, but more understanding of the contributing factors to TTV is required.

One of the key elements relates to the occurrence of ?flow breakdown?, a term which is quite widely used, but not usually clearly defined! This appears to be a partly stochastic phenomenon, in that essentially similar levels of demand do not necessarily lead to the same outcomes. In order to conduct a thorough investigation, a large amount of high quality data is needed.

Most of the current authors were engaged on a previous project which specified a careful data collection process on a 12 Km highly congested section of the M6 Motorway near Birmingham in 2003. Under the auspices of the UK Highways Agency, they have now re-grouped with the intention of analysing the data. The aim is, firstly, to develop appropriate supply models which can reflect what happens, on average, for a given profile of demand during the course of the day, and, subsequently, to use this model to allow for stochastic effects, both in demand and supply, in order to model TTV.

Two different approaches are being taken to the supply model ? one is an aggregate level model at the link level, while the other involves the use of microsimulation, using PARAMICS software. Both approaches are dynamic in concept, and allow for variation in demand levels in 5-minute intervals.

The paper will explain the approaches taken, and evaluate their success, both in relation to the ability to replicate the observed travel times, and their ability to identify the different components of TTV.


Association for European Transport