ABSTRACT
Microscopic simulation models are being increasingly used in traffic engineering applications, but various issues concerning the extent to which its outputs reproduce field data still need to be addressed. In this perspective a proper calibration of the model parameters has to be performed so as to obtain a close match between the simulated and the actual traffic measurements. This paper aims to highlight the importance of calibration process, as the adjustment stage of the microsimulation models’ parameters, applied to the analysis of urban, at-level, intersections. After the selection of the case studies (one roundabout and one intersection with traffic signal control), field observations were made, allowing the creation of a database that supported both the models’ development and calibration. The next phase focused on the application of an Aimsun microscopic simulation model to the selected intersections. This involved the development of an optimization based calibration methodology, coupled with a sensitivity analysis. The optimization framework was implemented in Matlab using the pre-defined genetic algorithm in the optimization extension. With the models properly calibrated and validated, the performance indicators were obtained and conclusions about their approximation to reality were drawn. The recommended calibration methodology easily allows the replication of the observed conditions, revealing however poor adaptation to other intersections and a general lack of representativeness. The results obtained by the genetic algorithm are very sensitive to small changes in the initial set of parameters. The calibration methodology allows the replication of the observed conditions, revealing however poor adaptation to other intersections and a general lack of representativeness.
