Simplifying Multi-stage Modelling Advice While Retaining the Essentials

Simplifying Multi-stage Modelling Advice While Retaining the Essentials


N Paulley, P Emmerson, TRL Limited; P Bly, Independent Consultant; T van Vuren, C White, Mott MacDonald; S Porter, Department for Transport, UK



The UK Department for Transport (DfT) recognises that models used to assess highway schemes and highway decongestion benefits now need to allow for how demand will respond to changes in road conditions , and to changes in other parts of the transport system, in competing modes for example. Ideally this responsiveness should represent the various demand mechanisms explicitly, which adds a number of additional stages to the modelling routinely used for scheme assessment. However, the DfT has also recognised that the skills to do this need developing to (a) identify which multi-stage modelling approaches are appropriate in particular circumstances, (b) create and estimate these models and (c) assess relevant parameter values for different scheme types and environments.

We have considered the practicality of simplifying techniques or the language of the advice aiming to make it understandable to traffic planners, engineers and assessors who are familiar with assignment packages, but who now need to expand their expertise to cope with predicting changes in demand. The aim has been to provide:
* clear step-by-step guidance on how to incorporate a practical form of variable demand modelling into predicting traffic flows on the road network, and on the techniques which will be satisfactory for the type and extent of scheme proposed;
* advice on how the suggested techniques can be adapted to best represent the local situation;
* illustration of parameter values to be used where appropriate values cannot be calibrated directly from the available local databases, based on examination of a large number of existing models in the UK and abroad.

The advice is intended primarily for highway schemes and highway decongestion assessments and aims to extend the expertise in coupling variable demand modelling with detailed assignment networks. The advice organises the necessary information in a way which leads the user step by step through the whole decision process, and explains it in a user-friendly way.

Where appropriate, the advice provides the relevant technical justification for the procedures it describes in appendices which are not essential to the main advice, so that the user can concentrate on the application of the techniques rather than their derivation. In this way it aims to avoid unnecessary detail, but at the same time it is important that the user has a proper understanding of the uncertainties and limitations of the methods used, and guidance is given on this. It aims to be consistent with a range of currently-available software packages, for both assignment and demand modelling, covering both:
* simple elasticity mechanisms, which are already available in some assignment packages and may suffice in some cases. These mechanisms merely adjust demand as travel costs change according to an overall assumed elasticity.
* explicit demand mechanisms, which provide a more specific representation of people?s travel behaviour and choices. A number of different mechanisms may be applied sequentially to represent the various effects of a change in travel costs on different aspects of behaviour, from trip generation, through trip distribution and mode choice, allocation to the various time slots (a recent addition to the behavioural hierarchy), and finally assignment to a particular route.

Software to help with the modelling has been developed in parallel through the DfT?s DIADEM project. DIADEM provides software which can implement the hierarchy of demand mechanisms, and interface the prediction of demand with an assignment model. Initially this is limited to a small number of existing modelling packages, but it will provide a framework for other package providers to develop and adapt their products to include all desirable aspects of an integrated demand and assignment modelling system.

The advice suggests a minimum set of requirements for testing transport scheme appraisal against the likely response of demand, for application in the wider area of scheme assessment where, until recently, the response of travel demand to a scheme was often considered rather cursorily. It represents a change in the Department for Transport?s expectations of good practice in a number of important areas. Key elements include:

* The advice provides methods for making a preliminary estimate of the likely effects of the different variable demand mechanisms on both road traffic and economic benefit.
* An incremental rather than a synthetic modelling approach should be used unless there are strong reasons for not doing so.
* A presumption that the effects of induced traffic on scheme benefits will be estimated quantitatively unless there is a compelling reason for not doing so.
* A recognition that the amount of detail required in demand modelling will depend upon the particular application, since the effort and cost involved should be commensurate with the investment being assessed. For small schemes a demand model based on simple elasticities may be sufficient.
* Where a multi-level variable demand model is appropriate, it should include a distribution mechanism, and even where mode choice appears unimportant it will generally be desirable to include that or other mechanisms to act as a reservoir of trips which can generate or suppress car trips as congestion reduces or grows.
* The sensitivity parameters in the demand mechanisms should be based on local knowledge where possible (from existing local models, for example). Failing this the Advice provides illustrative values, obtained from a review of current multi-stage demand models. The sequence of the distribution and mode split stages in the calculation hierarchy should depend upon the relative strengths of the sensitivity parameters, but trip generation should always be calculated first and time-period choice, if it is to be included, last with sensitivity parameters increasing along this sequence from first to last.
* It is essential to apply ?realism testing? to ensure that the model responds rationally and with acceptable elasticities. As a minimum, it is desirable to check the elasticity of demand with respect to car journey time and car fuel costs. It is also desirable to apply sensitivity testing to the results of the assessment against variation in those parameters which are uncertain. This is especially true of time-period choice, where the sensitivity parameter should be varied by 50%.

The paper will describe the process of developing the advice, while concentrating on discussing the main issues and the difficulties arising in constructing a multistage model. This points to areas where further research is required and suggests pragmatic approaches for the interim, which will also be discussed. The thrust of the work has been to build on previous and continuing research worldwide to provide the best and most practical advice for successful application of variable dema


Association for European Transport