The SHRP 2 Naturalistic Driving Study (NDS) data were used to evaluate the gap acceptance behaviour of drivers at left-turn lanes with offsets ranging from -29 ft to 6 ft. The study included 3,350 gaps evaluated by 145 NDS drivers and 275 non-NDS drivers (whose turns were visible from the in-vehicle camera of an NDS driver) at 14 two-way stop-controlled intersections and 44 signalised opposing left-turn pairs. Logistic regression was used to model the critical gap length for drivers in each offset category, under conditions both when their sight was blocked by an opposing left-turning driver and when it was not. The analysis found that the critical gap was longer for negative offsets than for zero or positive offsets, and also longer when sight distance was blocked by an opposing left-turning driver than when it was not. These longer gap lengths decrease the number of left-turning vehicles that are able to complete a left-turn manoeuvre in a given amount of time and can reduce intersection operations. Sight distance was much more likely to be restricted by an opposing left-turning driver at negative-offset intersections than at zero- or positive-offset intersections, and drivers at negative-offset intersections were less likely to accept a gap when an opposing left-turn driver was present. An analysis of the shortest post-encroachment times showed that while drivers, on average, wait to accept longer gaps at negative-offset intersections, they are also more likely to leave the shortest amount of time between their turn and the arrival of the next opposing through vehicles. While no specific safety concerns were identified in the study (no near misses and very few avoidance manoeuvres were observed), this finding indicates that the potential for a left-turning angle crash may be higher at negative-offset left-turn lanes. Other factors that may contribute to gap-acceptance behaviour were considered, including: driver age and gender, time spent waiting for a gap, left-turn signal phasing (permissive only vs. permissive/protected), vehicle type of turning driver, weather and light condition, and the presence of a following driver. For many of these considerations, the number of measured accepted gaps was too few to complete an analysis, but descriptive statistics are presented. Recommendations from this study include 1) correcting any negative-offset left-turn lanes to a zero or positive offset where possible, and 2) considering protected-only signal phasing where negative offsets cannot be avoided. Suggestions for future research using the NDS to evaluate left-turn gap-acceptance behaviour are provided. (Author/publisher)
Samenvatting