In Sweden a lot of roundabouts have been built in the last years to improve traffic safety. According to before and after studies, intersections that have been rebuilt into roundabouts display an overall increased traffic safety. This thesis consists of a brief literature study and an empirical study. The purpose of the thesis is to examine what kind of design that is the most favourable for the safety of cyclists in roundabouts and to try to explain why conflicts arise in the different roundabouts. The literature study showed that the different before and after studies have varied results regarding the safety of cyclists and pedestrians in the rebuilt intersections. A study from Vaxjo showed large improvements, a Dutch study showed small improvements and a Danish study showed no improvements at all. The diversity of the results is probably due to different methods of study, different yielding rules in the different countries and the choice of design for cyclists in the roundabouts. The empirical study of this thesis illustrates the impact of the design on the safety of cyclists in roundabouts. There are three different types of designs for bicyclists in roundabouts. Two of them are studied in this thesis: integrated without a cycle lane and separated with a cycle path. The design with integrated cyclists on a cycle lane was excluded from this study because this design is quite rare in Sweden. Two different pairs of roundabouts were studied (Eslov - Hassleholm and Vaxjo - Landskrona). The roundabouts within the pair (one integrated and one separated) were similar regarding geometrical size, car flow and bicycle flow. The comparisons within each pair were done using speed measurements, studies of yielding behaviour, the Swedish conflict technique, route choice studies and flow measurements. The speed measurements in the first pair showed that the motorised vehicles in roundabouts with integrated cyclists had the lowest average speeds, but the measurements from the second pair showed that the lowest average speeds were in the separated roundabout. This study did not clarify the influence of the design on the average speed. The location of the roundabout and the flow of pedestrians and cyclists seem to have an influence on the vehicles' average speeds. Motorised vehicles in the more central roundabouts had the lower average speeds. The study of yielding behaviour of incoming vehicles indicated that the design with integrated cyclists had a higher rate of vehicles giving way to cyclists than the separated design. This result is probably due to the choice of design for cyclists. An incoming driver's attention is directed towards the left since threatening vehicles are coming from that direction. This is probably the reason why drivers more easily detect integrated cyclists and can give way to them. In the separated design, drivers have to split their attention in two directions and therefore they do not always discover the cyclists as soon as they would in the integrated design. The conflict study showed that there were fewer conflict situations and only half as many serious conflicts in the integrated roundabouts as there were in the separated ones. However, this divergence is not statistically significant since the number of registered conflicts was low. The degree of seriousness of the conflict situations was in the overall view also lower in the roundabouts with integrated cyclists. The route choice studies showed that totally there were more cyclists who made accurate route choices in the integrated design. The results from the different pairs are not unambiguous. The first pair showed a better result for the separated design while the second pair showed a better result for the integrated design. Details in the design seem to have great influence on all the results from the studies. Details like the connection between the cycle path and the road, if the cycle paths are one-way or two-way and other details are important for the results of the study. The overall results from the empirical studies indicate that roundabouts with integrated cyclists are safer for cyclists than the roundabouts with separated cyclists. This result is in accordance with the study from Vaxjo, but the results disagree with the studies from Holland, Denmark and the Swedish VTI-study. This empirical study contains only four roundabouts. Further research about the safety of cyclists in roundabouts is needed to ensure that the results are statistically significant. (Author/publisher)
Abstract