Predicting travel time to limit congestion at a highway bottleneck

Abstract

A new method is proposed to predict the travel time on a highway route with a bottleneck caused by an on-ramp. The method takes advantage of the slow variation of the bottleneck throughput when congestion exists. The predicted travel time for a vehicle leaving the origin is given by the current number of vehicles on the route divided by the estimated throughput. The latter is an average of N / T recorded as each vehicle reaches the destination where N is the number of vehicles at the start of the trip and T is the time to complete the trip. Drivers divert to an off-ramp when the predicted travel time exceeds a target value. The target could be historical average travel times of alternative routes or chosen to limit the amount of congestion. Simulations employing three-phase traffic theory show that the travel time converges to the target value and remains close to or below it with the proposed prediction strategy. Strong oscillations in travel time obtained when other strategies are used for diversion do not develop with the new method because the inherent delay is effectively removed.