Calibration of traffic models in SIDRA

Sammanfattning: This thesis studies calibration of SIDRA Intersection roundabout models. The calibration has been performed with three different methods; manual calibration of gap-acceptance parameters, manual calibration of the parameter environment factor and automatic calibration of the environment factor based on optimization. The main aim has been to evaluate and compare 3 the three methods. One part of the purpose was also to test the possibility of finding a general value of environment factor by testing a few different types of models. SIDRA Intersection is a micro-analytical model used for capacity and performance estimations. The capacity model is based on gap-acceptance theory where follow-up headway and critical gap have a great impact. An application for automatic calibration was developed in Microsoft Excel/VBA. The calibration process is based Differential evolution and the cost function uses root mean square percent error where capacity and average delay are used as performance measures. The application has been tested with a test model at first to make sure the calibration is correct. The application has then been tested on data from a site in Farsta, Stockholm. For this site all three calibration methods, the two manual and the automatic, has been tested. The results show that all three methods performed equally well, with only small differences in precision. The most effective method considering results and time required for process was the manual calibration of environment factor.In the last part of this project the method for manual calibration of environment factor was tested with three more models with different characteristics in order to investigate the possibility of finding one value for environment factor suitable for all approaches. However the tests show difference in environment factor between small and large roundabouts and difficulties to set one value for environment factor for two-lane approaches.