Estimating Passenger Demand for Parkway Stations

Estimating Passenger Demand for Parkway Stations


W F Lythgoe and M Wardman, ITS, University of Leeds, UK



Most inter-city stations are located in central urban areas whilst suburban station stations serve distinct local populations. Interest in a different type of station emerged in the 1980s. A Parkway station does not serve a local population, but instead acts as a convenient out-of-town railhead. Easy road access combined with good parking facilities and an attractive rail service are essential features of the Parkway product.

Considerable interest in Parkway stations remains. Proposals to at least consider new Parkway stationsare a common feature of bids for train operating company franchises in Great Britain and, for example, the Midland Mainline franchise extension deal provides for a major new Parkway linked to the M1 motorway and East Midlands Airport. They can also provide a means by which a train operator can better compete with a rival operator. For example, the study into the now opened Warwick Parkway on the Chiltern Line between Birmingham and London forecast that much of the new traffic would be abstracted from competing Virgin West Coast services. Parkways can provide improved accessibility to the rail network, particularly given congested urban centres, and the improved integration with car will stimulate modal switch. Parkways can therefore help to satisfy the commercial objectives of rail operators and the strategic interests of central and local government.

The Passenger Demand Forecasting Handbook contains a recommended forecasting procedure and set of demand parameters that are widely used by various organisations within the railway industry in Great Britain. It is an incremental approach, based on the application of elasticities to current demand levels.

Such an incremental approach is of no use when there is no current level of rail demand. This is recognised in the Handbook, where a Parkway access model is recommended. There are, however, a number of deficiencies of this model. These relate to the use of all-or-nothing procedures to allocate rail travellers between stations, the failure to allow newly generated rail travel unless the new station is better than an existing station, and the extensive data requirements to apply the model.

The purpose of the study reported here, conducted for the Association of Train Operating Companies, is to develop a Parkway demand forecasting model based upon observed demand levels at existing Parkway stations. Demand levels for 1999/2000 have been obtained from the CAPRI ticket sales system for ten Parkway stations. This yields around 3,000 observations for modelling purposes.

The model is of the direct demand type. Trips from a Parkway are explained in terms of the spatial distribution of population around the station, drive times and/or distance from each population zone to the station, the fare and timetable related service quality from the station to the destination in question, and the same measures for a number of stations which compete with the Parkway. The estimation requires the use of non-linear least squares. This builds upon our previous work using a similar procedure but based solely on ?conventional stations? which obtained a model which has found considerable application in recent practical demand exercises.

UK census data is used to locate populations at Enumeration District (ED) level (or Output Area level in Scotland). These populations are allocated to grids of polygonal zones surrounding the stations, and information on characteristics of population, such as age distributions, car ownership levels, social class and unemployment, is also collected.

A 1:250,000 road network provides the means to determine the drive times and distances from the population weighted zone centroids to the Parkway station and to neighbouring stations that compete with the Parkway station. The models recognise that individuals start their journeys from many specific origins that are served by a given Parkway station.

This paper, in addition to describing the data needs and estimation procedure, will highlight the use of the model to forecast demand levels for possible new Parkway stations.


Association for European Transport