The MULTITUDE is a joint research project funded by the European Union COST programme (COST Action TU0903). The main objective was to develop, implement and promote the use of methods and procedures for supporting the use of traffic simulation models, especially regarding model calibration and validation, to ensure their proper use and the validity of the results and the decision made on them. It consisted of four working groups: updated review of traffic simulation practice and research, highway modelling, network modelling, as well as synthesis, dissemination and training.
In the past three years, the SVT group has been involved in this research project, especially in the working groups concerning network modelling, and synthesis, dissemination and training. Our tasks included understanding the role and impact of parameters on model outputs; developing techniques for network model calibration, validation and O/D matrix refining; and disseminating guidelines and best practices for model calibration and validation.
One of our main contributions was the development of the quasi-Optimized Trajectory based Elementary Effects method (quasi-OTEE). The quasi-OTEE is a general sensitivity analysis (SA) method based on the Elementary Effects (EE) method. It is applied to screen the most influential parameters of a complex simulator with very low computational cost. This method was validated with many case studies using different simulators such as VISSIM and Aimsun. The results showed that it can properly identify the most influential parameters from a computationally expensive model, for which other quantitative SA techniques are not feasible to be applied from the beginning.
In addition, under the scope of MULTITUDE, we conducted a research named Calibration Study of VISSIM (CSV) together with the Modelling and Simulation (MuS) group within the Traffic Division of Transport for the City of Zurich. This research helped the MuS group to optimize the calibration process of VISSIM, tailoring their specific needs and requirements. The research was also selected as one pilot project of the new National Programme projects within MULTITUDE, and the experience gained is being disseminated as part of the work on best practices for model calibration through the MULTITUDE network.