There are two different capacity models, the gap-acceptance model and the regression model. The gap-acceptance model is based on driver behaviour. The regression model is based on a simple on mostly linear relationship between the major flow and the capacity. Some 10 roundabout capacity models were analysed and compared. An interesting observation is the great variation in geometric and traffic flow parameters used as input to the models. The aim of the modelling work was to develop a gap acceptance and a regression model. A maximum likelihood method developed by Miller and Pretty (1968) was used for the estimating of critical gaps. A special computer program for obtaining the estimates was written. Through regression analysis it was found that the critical gaps were dependent upon the size of the roundabout. About 57% of the variance in the critical gaps was explained by use of the mathematical model. It predicts somewhat longer critical gaps than CAPCAL does. CAPCAL is the Swedish capacity manual, normally used for the calculation of the performance measures. It was found that heavy vehicles had a critical gap about 1 second greater than private cars. No influence of the intensity of the major flow on the critical gaps was found. The follow-up times were not found to be correlated with any geometric or traffic flow parameter and were therefore assumed to be constant, at about 2.4 s. A M3-distribution provided a good fit observed headways, both for on-lane and for two-lane roundabouts. The minimum headway between major vehicles was found to be between 1.5 and 1.8 s. The proportion of free vehicles was linearly dependent upon the intensity of the major flow. A new model based on the empirical results was implemented in a computer program for a comparison with CAPCAL. The new model gives lower capacity. This is explained in terms of greater critical gaps, a different major headway model and a different definition of major flow. The capacity estimates lie between the estimates of the German and of the Australian model. The new model can be used for Swedish roundabouts with a radius greater than 15 meters. (Author/publisher)
Abstract