Agent-based micro simulation to assess the performance of roundabouts considering different variables and performance indicators

Abstract

Traffic congestion problems in intersections are usually solved by building infrastructures such as roundabouts. Several variables influence its performance, e.g. geometry, size and driving behaviour. Thus, it becomes necessary to compare these variables. This paper proposes a simulation model, developed to compare the performance of roundabouts, employing the object and agent modelling paradigms of Simio, to model the individual behaviour of vehicles. The results indicate the optimum size of roundabouts is around 40 meters of diameter and that the driving style has a greater influence on the performance of the roundabout than its unbalancing. In addition, it was found that roundabouts considering unbalancing and human behaviour decreased: The flow of vehicles in 8%, the waiting time per vehicle in 3 minutes, the queue size in 90%, the number of stops per vehicle in 88% and vehicles spent three times more fuel, than the roundabouts that did not consider these variables.