Alternative Gravity Modelling Approaches for Trip Matrix Synthesis

Alternative Gravity Modelling Approaches for Trip Matrix Synthesis


O Feldman, Transport for London; J Forero-Martinez, Halcrow; D Coombe, The Denvil Coombe Practice, UK


The purpose of this paper is to report on the research programme undertaken by Transport for London on the process of calibrating a gravity model.


An important stage in the development of highway assignment models is the development of a matrix of the number of trips going from each origin to each destination. A gravity model is one of the approaches for synthesising trip matrices, which uses a concept of the deterrence function, namely, it assumes that the interaction between two sites declines with increasing distances, cost and time of travel but is positively correlated with the amount of activity at each site. We report on a research programme on the process of calibrating a gravity model using different deterrence functions and constraints and propose to use a deterrence function based on a log-normal distribution as an alternative to the commonly used Tanner, exponential or power functions. We also discuss the application of sector level movements as constraints in a three-dimensional Furness which allows the synthetic matrices to be controlled to the partial trip matrices derived from roadside interview survey data. This approach is currently being implemented in the London Sub-regional highway assignment models.
This paper is concerned with use of a gravity model for the synthesis of base year trip matrices which are subsequently subject to various adjustments to match counts in order to achieve a satisfactorily validated highway assignment model. The models discussed in this paper are not intended to be used for forecasting future year demands or demand responses of highway interventions.
The opinions expressed here are those of the authors and not necessarily those of the Transport for London or Halcrow Group Limited.

Keywords: Gravity modelling, Trip distribution modelling, Deterrence function, Log-normal distribution, OD matrix


Association for European Transport