Behavioural Impacts of Highway Journey Time Reliability Effects in a Multimodal Urban Model
N Raha, MVA Consultancy, UK
An analysis of the impacts of highway journey time reliability measures in a major multimodal urban transport model, contrasting both off-model and within-model tests and alternative approaches to capturing non-modelled behavioural effects.
The Transport for London (TfL) London Transportation Studies (LTS) model is a major urban four-stage multimodal transport model. It forecasts future year travel demand for London and the surrounding area based on exogenous socio-demographic, economic and transport supply inputs. LTS incorporates responses based on changes to a wide range of cost components within all main transport modes and their interactions and has recently been enhanced to include improved representations of parking costs and other previously unmodelled cost responses.
Nevertheless, comparisons of LTS forecasts with recent trend data have shown that the model tends to over-predict highway traffic growth. It is likely that standard assumptions regarding future fuel efficiencies and network performance issues (related to perceived travel times in congested areas) are over-optimistic, and that further non-modelled factors are influencing behavioural responses towards lower than anticipated car use.
One key element which is not currently included within the LTS model as standard is to account for the impact of responses to changes in the unreliability of highway journey times in congested urban conditions. This paper reports on the results of tests carried out on behalf of TfL to apply an approach to the estimation of the perceived changes generalised cost due to changes in urban journey time reliability as set out in the UK Department for Transport (DfT) Transport Analysis Guidance (WebTAG) Unit 3.5.7.
The WebTAG approach uses a relationship for change in journey time reliability (expressed as a change in the standard deviation of travel time) derived from an equation relating the Coefficient of Variation to the Congestion Index and distance. The functional form and coefficients were derived from empirical data and analyses from a number of urban routes in England.
The current study adjusted the WebTAG Unit 3.5.7 coefficients to account for an alternative average speed appropriate to the London / LTS model context (and alternative model units) and applied the resulting formula to account for the difference in journey time reliability between a 2031 forecast year and a 2007 base year.
The results from the application of unreliability effects were contrasted with those from a control run without any additional cost components, and an alternative approach to the representation of non-modelled highway responses in 2031 through the inclusion of a simple global distance-based highway cost adjustment. Additionally, the impacts of both approaches were estimated firstly using a simple off-model matrix based calculation taking time and distance skims from the control test run, and secondly incorporated as part of the full model supply-demand equilibrium feedback loop.
Conclusions are drawn as to the efficacy of each approach in the context of the observed data trends in highway traffic growth in London and findings in other studies. We also comment on the spatial variation of the two approaches and potential downstream impacts of using each approach in the wider modelling contexts of forecasting accessibility changes to feed into Land Use Transport Interaction (LUTI) models and Wider Impact appraisal.
Association for European Transport