Modelling Day-to-Day Variability of Intersection Performance using Micro-Simulation

The performance of signalized intersections in the field exhibit significant variation. Even when considering a specific time period (i.e. PM peak period) on a weekday, the average delay experienced by vehicles varies from day to day. This variation arises from a number of sources including variations in peak period traffic demands. The use of micro simulation tools to analyze and/or predict the performance of intersections, and more generally road sub-networks, is now common practice among traffic engineers. Popular micro simulation tools, such as Paramics, VISSIM, Integration, Aimsum, and NetSim are considered “stochastic” models in that they use pseudo random numbers to control random processes within the simulation, such as lane changing decisions, desired speeds, etc. As a result, traffic engineers and simulation model users typically carry out several model runs, each with a different random number seed, for each set of traffic and control conditions. Often, the results from the replications are averaged in the hope that the mean of the multiple runs is a reliable predictor of the average conditions that would occur in the field. In this study we seek to address two key questions, namely: (1) Does the current practice of conducting multiple simulation runs, each with a different random seed but with the same traffic demands, adequately replicate the day-to-day variability typically observed in the field? (2) What is the best method by which to use simulation models to reflect day-to-day variability in intersection performance? Three methods of modelling day-to-day variability of intersection performance (in terms of delay) are examined. Method 1, reflecting the typical current practice, introduces variability through the use of different random number seeds but traffic demands are held constant. Method 2 consisted of using only a single random number seed but traffic demands are randomly selected from a distribution fit to field data. Method 3 consisted of using different random seeds and randomly selecting traffic demands from the field calibrated distribution. For each method, eleven traffic demand scenarios were developed encompassing intersection volume to capacity (v/c) ratios ranging from 0.6 to 1.10. The results show that the current method of incorporation variability in simulation runs (i.e. Method 1) does not adequately capture the day-to-day variation observed in field peak hour approach volumes. More specifically, 1. The use of Method 1 results in intersection peak hour approach volumes that exhibit only about half the variation observed in the field. Methods 2 and 3 exhibited variations in peak hour approach volumes that were very similar to those observed in the field. 2. The average intersection delays obtained from the three methods differ by as much as 47% suggesting that the method used to simulate day-to-day variability in intersection delays has a significant influence on the results. 3. The differences in average intersection delays are sufficiently small at v/c less than about 0.8 that they are not likely to be of practical significance.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.