Application of Existing Demand Models: Theory Versus Application, an Unhappy Marriage?



Application of Existing Demand Models: Theory Versus Application, an Unhappy Marriage?

Authors

M Cusack , P Mijjer 4cast BV, NL

Description

Increasing demands are being placed on the capability of demand models and in particular the outputs of the models and their subsequent use in the planning process. The key issue being the need to design and apply models that are fit for purpose

Abstract

Currently within the European transportation field ever increasing demands are being placed on the capability of transportation demand models and in particular the outputs of the models and their subsequent use in the planning process. This tension is particularly apparent within the economic and environmental evaluation phases of infrastructure projects. The transportation demand modelling profession as a whole is being confronted with a number of credibility issues. One of these issues is the need to deliver and apply models that are demonstrably ?fit-for-purpose? with some form of guarantee that the applied methodology and the results are plausible and more importantly reliable and explainable.

Existing models need to provide answers to a wide range of questions, ranging from National Policy measures such as Infrastructure investment programs to Highway access issues for a local authority. Both of these problems deserve equal attention and require the appropriate tools and maybe much more importantly need to be applied in the correct manner.

The paper will highlight the fact that there is a growing need to ensure that the right modelling technique is chosen for the right problem and more importantly that the model is applied in the way that the model developer had in mind. A good example here is how O-D demand matrices that have been derived from a Regional model application are transferred across to a Dynamic model simulation study and how the subsequent model results are used with economic evaluation studies. The question being is it theoretically correct to transfer such O-D matrices and how robust is the methodology to ?dynamise? the O-D matrices. Other aspects that need to be considered are the age old issues relating to model validation, calibration and unreliability.

The modelling profession needs to protect itself to ensure that the work carried out by practitioners is credible, being not only theoretically ?state of the art? but at the same time applicable to the problem in question and applied in such a manner that not only the methodology but the application process can be professionally defended as and when required, be in it in the law courts or to assist financial institutions in investment decisions.

In addition the paper will aim to highlight a number of aspects relating to the question: Do we really know how the model works and are we using the model in the correct manner? Do we clearly understand and know what the key drivers are behind the generated forecasts. Such issues as the effect of scenario related issues as compared to other model inputs, data sets supporting model estimation, the role and an importance of factors such as values-of-time, assignment techniques and model convergence etc.

Based on our experience both here in the Netherlands and elsewhere we will address all of the issues above and provide a number of recommendations as to the way forward.

Publisher

Association for European Transport