Forecasting and Appraising Travel Time Variability in Urban Areas: A Link-based Approach
I Black, Indepedent Consultant, UK; J Fearon, John Fearon Consulting, UK; C Gilliam, Hyder, UK
This paper will describe a "link-based" approach to modelling urban travel time variability (TTV), as an alternative to the "distance-based" modelling approach described in a paper by the same authors at the 2008 ETC.
Recent years have seen a growing interest in monitoring travel time variability (TTV) on the highway network and developing methods to forecast TTV. This paper describes work undertaken for the UK Department for Transport to develop methods for calculating and forecasting TTV in urban areas using travel time data collected from Global Positioning System (GPS) tracker-equipped vehicles.
A paper by the same authors at the 2008 ETC examined an approach to forecasting TTV for an urban journey that used the length of the journey and an average congestion index on that journey as independent variables. This paper will describe an alternative approach to explaining TTV in urban areas, that was developed using the same GPS data. This approach uses a model formulation that predicts TTV on a journey as a function of the TTVs on the individual links that make up the journey - i.e. a link-based approach compared to the distance-based approach described in our 2008 paper. TTV on an individual link is related to the average delay on that link in a particular timeslot. The summing of individual link TTVs requires that the issue of adjacent link delay correlation is addressed. The paper will provide new evidence on this critical issue.
The paper will present the link-based approach to forecasting TTV in urban areas. It will describe the analysis undertaken to understand the impact on individual link TTV of such factors as link and/or junction characteristics. This will be followed by an examination of the adjacent link correlation, its size and the factors that influence it. Finally the paper will examine the relative merits of the link-based approach vis-à-vis the distance-based approach, thus demonstrating the accuracy that can be expected from such models of TTV in urban areas.
Association for European Transport