Effects of Arrival Type and Degree of Saturation on Queue Length Estimation at Signalized Intersections

TL;DR

This study evaluates how arrival type and saturation level affect queue length estimation accuracy at signalized intersections, revealing that saturation level significantly influences overestimations, while arrival type does not.

Contribution

It provides empirical evidence that degree of saturation impacts queue length overestimations, emphasizing the importance of numerical analysis over relative comparisons.

Findings

Overestimations are significantly related to X values, especially below 0.5.

Arrival type does not significantly influence overestimations.

Queue length correlates strongly with delay and saturation levels.

Abstract

Purpose of this study is evaluation of the relationship between different arrival types and degree of saturation (X) with overestimations of HCM 2010 procedure for estimating the back of queue within a study area. Further analysis is performed to establish the relationship between queue length and delay and also between each of them individually and X in cases with overestimation. The analyses are based on the 50th percentile queue lengths for data collected at four signalized intersections along a corridor in 4 time periods (off peak period and AM, Noon and PM peak periods). Based on the statistical test results, arrival type did not play a role in overestimations. However, there is a significant relationship between the overestimations on minor and major street and different ranges of X. On minor streets, about 59% of the overestimations are at X values less than half; while near 23% of the overestimations are at oversaturation condition with X values greater than 1. The relationship between amount of overestimations and degree of saturation should be established based on the numerical amount of overestimations versus X values rather than the relative amounts; since the statistical comparison between the relative amount of overestimations and X values, resulted in a wrong idea of the real world condition. There was a significant correlation between field queue and delay data of the cases with overestimated queue length in all cases on major and minor streets. Also, field queue is correlated to X, in all cases on minor and major streets.