Identification of the most effective calibration parameters in safety microsimulation models (Case study: Mashhad-Sabzevar-Tehran suburban highway)

Traffic accidents, as one of the most important threats to human beings, have shown themselves as a dangerous phenomenon. The impact of various factors on the incidence of accidents with different contributions has made it difficult to reduce the consequences of this unpleasant event in the safety analysis. Calibrated simulation models by taking into account the interaction effects of factors such as various behavioral characteristics of users in interaction with road have the potential to assess the safety performance of traffic facilities in different situations. Identifying the effective calibration parameters in simulation models due to their different impacts according to the purpose of the study as well as computational constraints is one of the essential steps in the calibration process. This study aims to provide a solution to identify the effective calibration parameters in the safety microscopic simulation model. For this purpose, 4 detectors located at the Mashhad-Sabzevar-Tehran route and vice versa with 23 accidents and 306 traffic records were selected as case studies and by deriving geometric information from available databases, the simulation model of the study area was built in the SUMO microscopic simulation software. By logically executing the simulation model on selected accidents, the effective calibration parameters in the car-following model, as well as the lane-changing model, were identified and by adjusting them in multiple scenarios, the simulation model in the case study routes was calibrated and validated with the use of the safety effective calibration target parameters such as Surrogate Safety Measures (SSMs). Among all the parameters of the simulation model, the global parameters of the normal distribution of permissible speed coefficient index with a mean and standard deviation of 0.95 and 0.10, respectively, and also local parameters of the cooperative, assertive and keep right lane-changing indexes were identified as effective.