Developing a New Transport Demand Model for the London Area

Traffic on the streets of London

Photo by Garry Knight/CC BY 2.0

With rising populations in London leading to higher demand on road and transport networks, Transport for London asked researchers to develop a new strategic travel model for London, which will be used to develop strategic land use and transport policy scenarios to assist with future investment in London’s transport systems.


London’s population has grown significantly over recent decades, resulting in significant pressure on the existing road and public transport networks. Transport for London require the best possible tools to assess this demand and to allow them to test policies aimed to relieve some of this pressure. Crossrail 2, for example, aims to relieve pressure on the Underground.


RAND Europe, working together with Jacobs, SYSTRA and Mott MacDonald, worked on behalf of Transport for London to develop a new strategic travel model for the London area. The model will be used to assess a range of transport policy options, including schemes to improve the public transport services across the region, and to assess demand for cycling.

The aim of RAND Europe’s contribution to the project was to develop individual-level models capable of predicting levels of transport demand in London in response to a range of different policy options and in the light of demographic changes.


The approach that RAND Europe followed in this study was to develop individual-level discrete choice models using the London Travel Diary Survey (LTDS) data. This approach allowed the model to capture variations in individuals’ preferences across different socio-economic groups, including car availability, income, working status, occupation, age and gender. To assess current and future demand for cycling, a cycling propensity approach has also been developed. This links propensity to cycle to a number of socio-economic characteristics observed in the LTDS data.


Model Design

The models were developed using a tour-based approach, the key unit of analysis being home-based (HB) tours, a series of linked trips starting and ending at the travellers’ home. Two types of non-home-based (NHB) tours were also represented.

A number of different travel purposes were identified—seven for HB and six for NHB—along with seven modes of transport and four separate time periods. Of the total 1,729 travel zones across Great Britain represented in Transport for London’s New Demand Model (NDM) system, 1,295 were within the study area inside the M25 boundary.

Data Sources

Data were gathered from a variety of sources:

  • The London Travel Diary Survey (LTDS) provided weekday data on observed mode-destination choices alongside person and household information, which was required for the specification of segmentation variables.
  • Highway level-of-service (LOS) data were supplied from TfL's London Highway Assignment Model (LoHAM), which was supplemented using toll information for the Congestion Charge and Dartford Crossings, alongside parking information.
  • Public transport (PT) LOS data were taken from TfL’s Railplan assignment model, representing a 2011 base year.
  • Cycle and walk LOS data came from TfL’s CYNEMON assignment model, representing a 2014 base year.
  • Population, employment and education enrolment data were supplied by TfL and used as attraction variables in the mode-destination modelling.

Model Specification

  • Attraction variables were used to represent the attractiveness of competing destination alternatives.
  • The (monetary) cost formulations used in the mode-destination models incorporate cost damping.
  • For non-business purposes, car cost sharing between car drivers and car passengers is represented so that predicted occupancy levels are sensitive to changes in car costs.
  • Variation in cost sensitivity with income was investigated by interacting the cost sensitivity parameters with income bands.
  • For walking and cycling, the model specification tested whether the overall preference for cycle and the disutility of generalised cycle distance varied between High Cycle Propensity (HCP) and Low Cycle Propensity (LCP) groups.
  • A range of person and household segmentations were tested in the model development work, including the key segmentation of car availability, which was specified as a function of individual licence holding, household licence holding and household car ownership.
  • Structural tests were undertaken to determine the relative sensitivity of main mode and destination choice, and to test for evidence of sub-mode nests for active modes.

Information on the nested logit results are summarised in the report.