Airport Rail Services: Forecasting Hourly Demand



Airport Rail Services: Forecasting Hourly Demand

Authors

K Kenny, Sinclair Knight Merz, UK

Description

Assisting the development of the Stansted Express rail strategy, understanding the distribution of demand across the year.

Abstract

Airport rail is often seen as a ?premium? service, and it is understood that the typical air passenger on rail has specific requirements in regards to level of service. In particular, with additional requirements such as luggage provision, the importance of crowding is stronger than for regular commuting services. There is an expectation that demand would never exceed capacity.

Forecasts of rail demand have typically focussed on annual or busy day patronage, with the latter often reflecting a typical day in a ?busy? month. For general regional or commuting demand this is considered a robust approach in determining capacity requirements, with smaller daily or seasonal fluctuations in demand. However, in the case of airport surface access, where the majority of demand for rail is not for regular regional or commuter traffic, it is even more important to understand the variability in rail demand across the entire year when determining the appropriate rail strategy.

In preparing for future surface access planning at its London airports, BAA (the operators of Heathrow, Gatwick and Stansted airports) commissioned the development of a time period model capable of predicting demand by mode by time of day. Initially this model has been applied in forecasting for the typical ?busy day? at both Heathrow and Stansted airports, but has now been supplemented, in the case of Stansted airport, to predict the variability of the forecast demand across the year, as a distribution of hourly demand.

This paper explores the philosophies underlying this model and variability analysis, and describes some of the principal challenges in its development.

The process of producing the final distributions of demand involve an initial ?busy day? forecast of demand by direction by hour. This forecast is undertaken using the London Airport Surface Access Model (LASAM), comprised of two components; the surface access mode choice model, and a time period model, allocating the runway passenger demand to surface access modes by time of day. The former model has been estimated on recent CAA air passenger data, drawing on the track record of HSAM, developed in support of the Terminal 5 inquiry in the mid 90?s. This mode share model is not described in detail in this paper, however the key elements of the time period model are presented.

The time period model was developed to interface with a surface access mode share model for air passengers. In providing forecasts of the time profile of surface access demands by mode of transport, the modelling system reflects the effects of the following factors:

- the time profiles of flights into/out of the airport;
- the mix of passengers by time of day;
- the impacts of peak period congestion on surface access mode shares;
- the unavailability of certain transport modes at certain times of the day; and
- the time air passengers take between their surface access trip and their flight departure/arrival time.

A significant part of the modelling concept involved reconciling different times of day, relating to the air passenger?s journey through the airport and the times recorded in the various data sets used in the model. The lag/lead times used in the model measure the time interval experienced by air passengers between the airport surface access transport node (rail station, coach station, airport car park etc) and the flight departure (or arrival).

Whilst forecasts of the profile of flight into/out of the airport are available for the busy day and were used in the application of the time period model, these do not enable the understanding of the distribution across the entire year, essential in deriving an annual distribution of rail demand. However, analysis of base year flight data, available across every hour of the entire year, has been undertaken to establish statistical distributions of runway demand by hour. These have then been supplemented by analysis of the CAA data to establish statistical distributions of passenger segmentation across the year.

In conjunction with the busy day forecasts from LASAM, the application of these distributions provide an estimate of the annual distribution of rail demand. When superimposed onto the underlying regional demand on these services (which in the majority of airport rail services is small) this enables the extraction of a number of key statistics, including; the proportion of rail services for which demand exceeds capacity and the proportion of air passengers who wish to travel on rail during these periods. These statistics are vital in determining an appropriate rail strategy, including decisions on service frequencies and train lengths.

Publisher

Association for European Transport