Travel Time Reliability during Incident Duration Time

In order to describe the stochasticity in road network performance under the influence of an incident, the travel time reliability of the road network was defined as the probability that the mean travel time during incident is smaller than a prespecified threshold. The incident duration was assumed to be a stochastic variable with normal distribution, and is divided into several equal sub-sections. Then, a Monte-Carlo based simulation methodology was put forward to compute the travel time reliability, in which the logit principle and link transmission model are incorporated. A test network was used to illustrate the methodology, in which the mean value of incident duration varies between 8 and 20 min, and variances between 0.5 and 5.0 min, the travel demand of each sub-sections is 4.0 and 4.5 vehicles, and the threshold is 2.0 and 2.2 times the travel time before incident. The results show that the network travel time reliability decreases with the mean incident duration in all cases. In addition, when the travel demand of each sub-section is 4.5 vehicles and the threshold is 2.0 times the travel time before incident, the network travel time reliability increases with the incident duration variance; however, when the demand is smaller than 4.5 vehicles and the threshold is larger than 2.2, the network travel time reliability decreases with the incident duration variance.