Modelling delay at roundabouts under mixed traffic flow conditions in Kampala, Uganda

Abstract

Roundabouts are a safer type of junction for motor vehicles and pedestrians than other forms of at-grade intersections. Vehicles however, experience traffic congestion with long delays and queues at roundabout approaches during peak periods which is the case with roundabouts in Kampala the Capital City of Uganda. Significant research has been conducted worldwide to model delay experienced by vehicles at roundabouts, however, much of the work has been inclined to the entry stop line delay without considering the effect of differing roundabout geometry on the time taken by vehicles to negotiate the roundabout. A few studies to estimate the total delay (sometimes referred to as control delay) at roundabouts have been conducted in developed and developing countries but transferability of models between countries is constrained by differences in infrastructure standards, motorists’ behaviour, and vehicle characteristics. This research was undertaken to develop a total delay per vehicle model based on roundabout geometric features and mixed traffic conditions experienced in Kampala. The video recording technique using unmanned aerial vehicles mounted with cameras was employed at selected roundabouts in Kampala to obtain the subject entry and exit traffic, circulating and rightturning traffic volumes. Roundabout geometric features namely, central island diameter, inscribed circle diameter, circulating width, approach entry width and entry radius were obtained using georeferenced aerial imagery in Autodesk AutoCAD Civil3D 2022 and confirmed through field measurement. Multiple linear regression analysis was conducted on the predictor variables namely, the subject approach entry, exit and conflicting traffic volumes, percentage of right-turning traffic in the subject entry, subject entry width and radius, roundabout island diameter and width of the circulating carriageway to obtain the most significant variables in total delay time prediction. This indicated that the entry width, circulating width, central island diameter, entry traffic volumes, conflicting volume, and percentage of right-turning traffic in the subject entry had the greatest significance on the total delay time prediction at 95% confidence level. The entry radius and exit traffic volumes were not significant in total delay time prediction at 95% confidence level. Compared with other total delay models in accessible literature, the developed total delay model by this research study performed better at predicting actual field total delays. The developed model estimates the total delay per vehicle as a function of the roundabout entry width, circulating roadway width, central island diameter, entry traffic volume, circulating/conflicting traffic volume, percentage of right-turning traffic volumes in the subject entry and the ratio of circulating to entry volume. The model shows that total delay per vehicle decreases with the increase in subject approach entry width, however, it increases with an increase in width of circulating roadway, central island diameter, entry volume, circulating volume, percentage of right-turning volume in the subject entry and the ratio of subject entry conflicting to entry traffic volume. The model was then presented in graphical form as a nomograph to provide a quick guide to practitioners for roundabout design. Finally, geometric road designers in Uganda can make use of the total delay model and nomograph developed in this study for estimating total delays when selecting roundabouts as an intersection type on crossroads for cities in Uganda with flat terrain and similar traffic conditions as those used in the model development. Moreover, the total delay model and nomograph can be incorporated in the revised Ministry of Works and Transport Road Design Manual, Geometric Design Manual for estimation of total delays at roundabouts for Ugandan conditions. For future research, the effect of entry radius on total delay per vehicle at roundabouts should be studied. This geometric parameter is frequently used in roundabout design to control vehicle entry speeds and was thought to significantly affect total delay contrary to the findings of this research.