Assignment of traffic to a network of signalized city streets

Sie sind hier: Homepage > Suche

Assignment of traffic to a network of signalized city streets

Bentley, R.W. / Lambe, T.A.

Abstract A road network model that allows each vehicle to choose the route that minimizes its travel time is developed and tested using data from a medium-size city. Assumed as given are the street configuration, the travel speed for streets with and without perfect progression of their traffic lights, any intersection with special turning phases, and the origin and destination (but not routes) of all vehicles entering and leaving the city during a typical period of time. The model incorporates the expected delay at intersections due to the congestion resulting from random fluctuations in traffic flow. It also can adjust the length of the green phase at each traffic light to minimize overall travel time. A test with data from the City of Vancouver, B.C., confirmed the basic assumptions of the model. Compared with observed traffic flow, the morning rush hour volumes per street had a median absolute error of 16%. Furthermore, the error decreased steadily to this value as the linear programming iterations reduced the objective to its optimal value. On an IBM 370-148, computer time to optimality was 110 minutes for the test area which included 59 intersections with an average flow per street of 640 vehicles per hour. Subsequent optimization of the red and green phases of the traffic lights slightly altered the flow pattern, but had a negligible effect on the overall travel time. The introduction of anticipated growth in total traffic volume indicated that the system could adjust to 60% more vehicles before the street capacity and configuration would require major alteration.