Managing Computational Complexity of Large-Scale Traffic Micro-Simulations

In an R & D project called MASE ("a model-based approach for evaluating the safety and environmental effects of traffic policy measures", sponsored by IWT-Flanders), our team collaborates with research teams specialized in traffic management, environmental and safety studies. The main objective of the project is to provide usable instruments (models as well as tool-support) which support making well-founded assessments on the effects of governmental traffic policies (e.g. charges per kilometre, free public transportation) on safety and environment. The methodology followed in the project consists of the following steps: (1) modelling the effect of traffic policies to individual behaviour (using an activity-based modelling); (2) simulating traffic using these individual behaviour models-based on agent-based micro-simulations; (3) evaluating the simulation results with respect to criteria related to safety and environment. The micro-simulation obviously plays a crucial role in the success of the project. Therefore, we use application- and simulation-specific characteristics for reducing the computational complexity of the micro-simulations. Technological solutions can and will help: parallelizing the simulations and using parallel or distributed computer systems, simulations can speed-up simulations by a factor that is at most linear with respect to the number of processing units. However, more structural solutions are necessary for coping with the scale we envisage in the project.