Capacity of U-Turning Traffic at Uncontrolled Median Openings on Four-Lane Divided Roads: A Gap Acceptance Based Approach

Abstract

The present study estimates the capacity of U-turning traffic at uncontrolled median openings on four-lane divided urban roads under mixed traffic conditions using Harder’s model and the adaptive neuro-fuzzy inference system (ANFIS) technique through a gap acceptance approach. The U-turn capacity estimation requires three crucial parameters: critical gap, conflicting traffic volume, and follow-up time. This study considered spatiotemporal conflicting factors for the accurate estimation of conflicting traffic volume. The critical gap of U-turning vehicles was estimated using five different methods: modified Raff, Ashworth, binary logit, occupancy time, and support vector machine. The average follow-up time was estimated for different classes of U-turning vehicles from the recorded video data. Subsequently, the capacity of the U-turning traffic stream was computed using Harder’s model and the ANFIS technique. The efficacy of both models was checked by comparing the estimated capacity with the field capacity determined using Kyte’s method. The strong correlation and low mean absolute percentage error (MAPE) values confirm the effectiveness of both the models in estimating U-turn capacity across different vehicle classes on four-lane divided urban roads. This study provides significant information to enhance traffic management and alleviate congestion during peak periods at uncontrolled median openings.