Overview
Traffic assignment is the stage of modelling the choice of routes by road users and of calculating link flows and travel times for a fixed traffic demand and a given road network.
Many different rules, algorithms and methods, the most commonly used in assingment models are described and evaluated, based on the results of test applications. The study‘s focus is on congested road networks in urban areas. lt is assumed, that - as is often the case in real life - a given origin - destination (OD) matrix of present or future travel demand has to be assinged without availability of demand models (traffic generation, choice of destination or modal split).
Following steps were planned during the project:
- literature review
- survey
- theoretical assessment of existing models
- practical tests based on case studies
Funding
Results
Route choice models are based on the assumption that for a trip from zone i to zone j road users choose the route with minimal costs. A main component of these costs are travel times. Travel times in a road network depend on traffic flows. As traffic flows increase, speeds on links drop and delays at junctions grow.
Commercialised assignment models can be characterised by their rules and procedures implemented, e.g.
- incremental loading
- deterministic user equilibrium (UE)
- stochastic user equilibrium (SUE) and by the more basic aspects such as
- whether or not junction delays are being modelled, and if yes, in what detail
- whether or not dynamic effects are taken into account
- whether or not demand elasticity is taken into account
Using as an example a fictitious network representing an urban situation, the above mentioned methods are used to assign OD-matrices for moderate, high and very high travel demand. The analysis of the test results leads to the following conclusions and recommendations:
- Equilibrium assignment is the preferred method for congested networks. SUE is more difficult "to apply but is based on more realistic assumptions than UE. However, for congested networks both methods give similar results and the use of UE is recommended in such situations.
- While in situations with low demand the choice of the assignment algorithms might be crucial, for congested aspects such as capacity, speed-flow-relationships, dynamic effects and demand elasticity become much more important.
- In congested urban networks, junction delays become a major component of travel times and have to be taken into account in the assignment stage. This requires detailled modelling of flow dependent delays an queuing.
- With static assigment methods, over capacity flows and unrealistically high delays will result for congested situations. When higher accuracies of assignment results are required, dynamic effects can not be neglected.
- A commonly used method to forecast future demand is to apply growth factors to an OD-matrix. If this "upgraded" od-matrix is assigned to the existing, already congested network, unealistic results will be obtained (much too high flows and delays), if demand elasticity is not taken into account.
Given the situation, as is assumed in this study, where an externally established OD-matrix has