Vehicle-vehicle Interactions at Roundabouts and their Implications for the Entry Capacity - A Methodological Study with Applications to Two-lane Roundabouts

Abstract: Problem: The thesis deals with the capacity of two-lane roundabouts and with vehicle-vehicle interactions there. Method: The interactions and the corresponding capacity are modelled by gap-acceptance theory. Various components of this theory are discussed, models for headways in the multi-lane case and for the estimation of critical gaps for the two major streams being developed and being applied to field data. The headway models were derived from renewal theory and were then applied to field data collected in Swedish roundabouts. Parameters of the headway distributions were estimated by maximum likelihood, the method of moments and least squares. The resulting distributions were tested by goodness-of-fit tests, using the D-statistic. Finally, a validation study was carried out. The method for estimating of critical gaps was based on a maximum-likelihood model for a situation involving one major stream only. This model was extended to two major streams and was then tested by descriptive sampling and Monte Carlo simulation. Bootstrap was also used for control of the parameter estimates given by the model. Finally, the model was applied to field data from Swedish roundabouts. Results: It was found that the Cowan M3 distribution could be employed for modelling the major flow in roundabouts, although in the validation test several of the samples were rejected. Minor-stream vehicles were impeded by vehicles in both the major streams. The strength of the impediment was measured by the size of the critical gaps. The critical gaps for a driver were significantly larger for the near major lane than for the far one, irrespective of which of the two minor lanes the driver was driving in. For driver turning to the right, the critical gap in the far lane was so large that it was evident that these drivers were influenced by drivers in the far lane, despite they not physically interacting with them. The correlation between the critical gaps was negative, a result that can be explained by some sort of risk compensation taking place or, what appears more likely, by ecological correlation.