Please use this identifier to cite or link to this item:
|Title:||Investigating the impacts on cyclist casualty severity at give way roundabouts with mixed traffic|
|Abstract:||Roundabouts are designed to improve the safety for all vehicles by decreasing the number of conflict points at intersections and reducing entry and circulating speed. Previous research suggests that the design does not provide similar safety benefits for vulnerable users, particularly cyclists. Local road authorities usually have very limited budgets for improving cycling facilities however, recently in the UK, there is more emphasis on policies to promote cycling particularly in urban areas. This means that cyclists increasingly are using give way roundabouts in mixed traffic, and therefore, there is a need for a fundamental understanding of which design parameters influence cyclist safety and what are the behaviour related contributory factors. The global aim of this research is to investigate statistically significant variables, considering geometric design parameters, sociodemographic descriptors of cyclist, meteorological conditions, traffic characteristics and driver/rider behaviour related contributory factors that have impact on cyclist casualty severity at give way roundabouts with mixed traffic. The first analysis explored the significant geometric design parameters, socio-demographic characteristics of cyclist, meteorological conditions and speed limit on casualty severity. Two components namely Approach Capacity (number of lanes on approach, half width on approach, number of flare lanes on approach and entry path radius) and Size of Roundabouts (number of arms, type of roundabout and number of circulating lanes) emerged from the Principal Component Analysis. The Multiple Logistic Regression suggested that a unit increase in number of lanes on approach, entry path radius and speed limit increase the probability of serious casualty occurrence with odd ratios 4.97, 1.04 and 1.02, respectively and a higher Approach Capacity increases the probability of serious casualty occurrence by 86% (odds ratio 1.86). Linear Regression suggested that if the entry path radius was more than 80 metres, the casualty severity was more likely to be serious. The second analysis explored the impact of driver/rider behaviour related contributory factors on cyclist casualty severity. One-unit increase in cyclist age group, junction restart, failed to look properly and failed to judge other person’s path or speed, increased the probability of Killed or Seriously Injured (KSI) casualty occurrence with odds ratio 1.15, 2.09, 2.82 and 1.64, respectively. Multilevel Logistic Regression showed that the regional variance between cities in England was not statistically significant at the 95% confidence level. In the final analysis, a comparison was made between roundabouts in England and Belgium using threeway chi square test of independence, Multiple and Multilevel Logistic Regressions. The results showed that older cyclists were more likely to be involved in KSI than slight casualties in both countries. Cyclist’s non respect of the priority to drivers increased the probability of KSI casualty. Speed limit emerged as a significant contributory factor in KSI casualties in England (tangential design); however, it did not show any significance in Belgium (radial design). In addition, country residual was statistically significant in the multilevel modelling. This research has demonstrated that speed has a dominant impact on cyclist casualty severity but the novelty rests with identifying that it is the approach capacity, and more specifically entry path radius and number of lanes, that most influences vehicle speed. This has enabled generic predictors for the probability of severity, which are valid for specific countries and regions with similar design approach.|
|Appears in Collections:||School of Engineering|
Files in This Item:
|Akgun N 2019.pdf||4.46 MB||Adobe PDF||View/Open|
|dspacelicence.pdf||43.82 kB||Adobe PDF||View/Open|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.