On the Marginal Cost of Road Congestion: an Evaluation Method with Application to the Paris Region

On the Marginal Cost of Road Congestion: an Evaluation Method with Application to the Paris Region


F Leurent, V Breteau, Université Paris-Est ? LVMT, FR


The paper analyzes the sensitivity of the marginal congestion cost on a roadway network to the level of aggregation in space and time, from utmost aggregate to utmost disaggregate on the basis of a static network assignment model.


In the cost-benefit analysis of a transport project, the effects on users? travel times must be taken into account since a time gain makes a utility gain to the individual user. In an urban area, a public transport (PT) project may entail a modal shift from the car to public transport, since it improves the PT quality of service. As roads are subject to congestion, the modal shift decreases the road flows and potentially also the road travel time, hence the road congestion.

What congestion relief is associated to a given PT project? Can it be evaluated in a way simpler than by use of a network assignment model? Both issues were tackled by a research project conducted by the Lvmt on behalf of the French Department for Transport. A methodology and practical guidance have been developed, in which a critical issue pertains to the time and space disaggregation of the relief in road congestion. Specific attention was given to the Paris metropolitan area, on the basis of a roadway network traffic assignment.

The purpose of the paper is to demonstrate the effects of the level of aggregation on the evaluation result. The main variable of interest is the marginal congestion cost associated with a unit distance of travel on the road network, with respect to the time and network location of the unit distance.

The paper is organized in the following way:
1. At the most aggregated level, the entire road network is considered as a whole: an aggregate speed-flow relationship is provided, from which a marginal congestion cost results. The sensitivity to the functional specification is analyzed.

2. At the most disaggregated level, all the network links are considered separately; no clear network pattern of the marginal congestion cost could be identified, because of the extreme variability of the link values.

3. A typology of the network links was established to yield ?traffic segments? with respect to the hourly conditions, the type of urban setting and the operating conditions. Then a statistical analysis was performed by traffic segment, yielding mean values as basis for practical recommendation.

4. A spatial aggregation at the level of the traffic assignment zone was conducted, using path skimming by origin-destination pair to track the marginal congestion cost of any trip along the network. This yielded a clear spatial pattern of which zones induce the largest marginal congestion costs.

5. All the previous steps were based on the result of a static traffic assignment model, which ignores the severe congestion phenomena such as queuing and stop-and-go traffic states. A dynamic evaluation of marginal congestion costs was also performed on the basis of a dynamic assignment model, namely the Ladta Toolkit developed by Lvmt. This enabled us to disaggregate the marginal congestion cost with respect to both space and time.


Association for European Transport