An investigation of lane changing parameters on an urban arterial for different traffic congestion levels

This paper investigated the variations in lane changing parameters including accepted gaps and lane change duration on an arterial under free flow, recurrent, and non-recurrent congested traffic conditions. Descriptive statistics (mean, median, standard deviation, and skewness) and best-fit distributions were obtained for accepted gaps and lane change durations collected through video recordings for three types of traffic conditions. Hypothesis testing using Mann-Whitney U Test showed that the mean values of accepted gaps were statistically different between all three types of traffic condition. Similarly, mean values of lane change durations were statistically different between recurrent and non-recurrent traffic conditions and between free flow and non-recurrent conditions. Further, the results revealed that the mean size of accepted gap was the least during non-recurrent traffic conditions and the value was the highest during free flow traffic condition. The result was quite the opposite for lane changing duration. The mean lane changing durations were the highest during non-recurrent traffic conditions and it was statistically the same for recurrent and free flow traffic conditions. The best fit distributions of accepted gaps and lane change durations complemented the above findings. The results of the study indicate that degree of frustration may be a reason for drivers accepting smaller gaps during congested traffic conditions. The authors believe that the new drivers visiting the study location for game day (non recurrent congestion) exhibit different driver behavioral characteristics when compared to the regular commuters (recurrent congestion) and this could have also contributed to the statistical differences in the lane changing characteristics during non-recurrent congestion. A sensitivity analysis of simulation results (delay) using lane change duration values obtained from the study confirmed that simulation results using default lane changing parameters may not be relied upon and hence appropriate values and specific distributions of the parameters should be chosen. The findings from this study have direct implications on the lane changing parameters used in microscopic traffic simulation and also help transportation planners and managers to understand the driver behavior during free flow, recurrent and non-recurrent congestion in better managing the facilities.