Urban Road User Charging Online Knowledge Base
What Are The Research Gaps?
In addition to problems already raised, the DISTILLATE project has carried out a wide-ranging review to identify knowledge gaps within the context of modelling and prediction of the impacts of road pricing schemes (Shepherd et al, 2006). It is worth emphasising that models are abstractions of complex reality and while recognised to be useful for informed decision making, there are still drawbacks with the current state of the art. These gaps are summarised below:
The first four issues relate to the difficulty in modelling of schemes themselves…
• Modelling of different schemes. In general distance based charging and cordon schemes can generally be modelled reasonably accurately. However, because conventional modelling makes a per trip assumption, it is difficult to model area-based charging schemes. There are now models that provide tour-based representations of travel patterns but these have not yet been adopted widely. The recent work of Maruyama and Sumalee (2007) uses a simple trip-chain model to attempt to compare area-based and cordon charging to address this issue.
• Modelling of exemptions or discounts. Within the conventional modelling framework as cited above, it is difficult, if not impossible to match data to trips in the assignment. The conventional approach would be to apply some factors outside the main modelling process. This could reduce the robustness of forecasts without further thorough sensitivity testing. In Edinburgh, the congestion charging proposals involved limiting the maximum charge to £2 a day and if trips crossed the first cordon to enter the innermost cordon, they would not be charged twice (Saunders, 2005). Unfortunately, attempting to model this form of exemption within most tactical level models is not feasible at present.
• Models used for road pricing prediction tend to be static and they are quite unable to handle dynamic changes e.g. predicting the influx of traffic into the charge area at the end of the congestion charge period.
• Modelling the possibility of purchasing monthly tickets/passes. If users can purchase a fixed price monthly pass, this would affect their timing, frequency and mode choice, since once that pass is purchased, the marginal cost of using that pass is zero.
The remaining issues relate to the modelling of changes in travel behaviour as a result of the introduction of road user charging scheme:
• The possibility of trip chains being formed as a result of a congestion charging scheme is difficult to represent unless sophisticated modelling techniques are used. Even if the models are available, the data collection and the effort required to do these will be extremely demanding.
• The possibility of changes in vehicle occupancy. Vehicle occupancies are normally assumed to be fixed in most models, yet road user charging would give an incentive to increase vehicle occupancy (Glaister and Graham, 2004). Recent research (Hensher, 2008) suggests that this could be detrimental in estimates of the value of time which therefore affect the prediction of traffic impacts of schemes.
• Park and ride could possibly be important and it is still quite difficult to model this with much robustness.
• Modelling of land use changes. It is uncertain whether cordons for entry into the city centre may lead to extensive development of out of town urban development sites which can be accessed by un-priced routes and what implications these outcomes might have on future urban form.
These gaps imply that there might be problems using outputs generated from predictive models to support the appraisal process. Other relevant areas where doubts exist over the abilities of the prediction process to produce accurate outputs include:
• parking policies
• equity and accessibility (and their interactions)
• distinctions between private and business traffic
• departure time changes (as experienced in Stockholm)
• destination changes
• impacts of using revenues outside the transport sector.
In addition, it is difficult to predict the acceptability of a scheme with conventional modelling tools. At present, our understanding of what constitutes an acceptable scheme is limited. More discussion of this theme is contained in Chapter 11.