Capcal for small roundabouts : Current status and improvements

Sammanfattning: Questions have been raised regarding how good Capcal works for roundabouts with a weaving length smaller than 25 meters and if Capcal would work for weaving lengths lower than the 16 meters that is the lower limit in the current model. In this thesis, two possible improvements has been developed and tested on five roundabouts with a weaving length less than 25 meters. Work has been done to build alternative gap acceptance models ranging from removal of the 16-meter limit through re-estimation of the current model to a completely new model. In this thesis a new sub model has been introduced that take into account the uncertainty of if circulating vehicles will exit or continue to circulate. These sub models have been combined in 29 different ways including the current Capcal model configuration and have been tested with five different roundabouts. The conflict delay, which is caused by interaction with other vehicles, has been used to compare the model results with delay observations from video captures of the roundabouts.   Out of all model configurations, the original Capcal 3.3 model had the lowest difference in average between the model result and the observations. If the 16-meter weaving length restriction in the original model is removed, almost the same results are obtained as with the restriction. While none of the model configurations improved the model fit, no configuration could be statistically be rejected. In the results of individual roundabouts, there were differences between the model and observations of up to 50%. This shows that while this thesis could not find any improvements of Capcal, there is a potential for future work to develop improvements of the roundabout model.