Modelling roundabout capacities

There has been extensive research and development into the capacity of modern offside priority roundabouts since the 1970’s. Despite this, there remains a major gap in existing knowledge with regards to the factors and variables which affect roundabout entry capacity. This is reflected in the differences and inconsistencies in inputs and methodologies between existing state-of-the-art models. Evaluations with recent data collected from 35 roundabout entry lanes in the field have shown that this limits the accuracy of state-of-the-art models, particularly in their ability to explain site-to-site variation in entry capacities. New empirical models have thus been developed for lane capacity using regression, and benchmarking against neural networks showed that they performed well with the shortlisted explanatory variables. These regression models were based on exponential-in-Qc and linear-in-Qc forms, and outperformed existing state-of-theart models. In the new models, entry-exit separation distance and exiting flows on the same arm were found to be more useful predictor variables (when used in conjunction with other variables) compared to others used in more-established models (e.g. entry radius and entry angle). To investigate the effects of separation distance and exiting flows through microscopic simulation, stochasticity in separation distances was modelled through a novel approach in Vissim involving multiple exit connectors. This was significant as the variability of separation distances had not been explored before, whether through analytical or simulation approaches. The separation distance was found to have a piecewise linear relationship with capacity, while exiting flows had a linear positive relationship which becomes negative as the inhibitory effect increased at low separation distances. The two main mechanisms explaining these effects of exiting flows were the inhibitory mechanism (caused by drivers unable to distinguish between circulating and exiting vehicles), and changes in circulating headways. A revised empirical model incorporating this piecewise relationship performed as well as the exponential-in- Qc and linear-in-Qc models, suggesting that the impacts of exiting flows were modelled reasonably well. By improving our understanding of the impacts of these two variables on capacity, this is an important step towards the improved modelling of roundabout entry capacity.