The TRANSYT simulation, and optimization program incorporates platoon dispersion by applying a form of geometric smoothing to the incoming traffic on each link, in order to predict the arrival pattern of vehicles at the next intersection. The calculation of stops and delay by the TRANSYT program, and the effectiveness of the signal timings resulting from the optimization procedure, depend on the platoon dispersion model to predict traffic flow patterns from one signal to another. In much of the research to date, it has been suggested that timing plans are sensitive to platoon dispersion, and that platoon dispersion is a function of roadway, and traffic conditions. This suggests that the platoon dispersion algorithm used in the TRANSYT model should be calibrated for specific conditions. On the other hand, if the optimal timing plans generated by the TRANSYT program are not particularly sensitive to platoon dispersion, then calibration of the algorithm to local site conditions may not be required. The importance of accurate modelling of platoon dispersion in developing signal timing plans depends upon this sensitivity. The purpose of this paper is to determine the sensitivity of optimized signal timing plans to changes in the variables within the TRANSYT model that describes platoon dispersion, and to quantitatively show what effect inappropriate dispersion values have on the resulting timing plans. The results of this sensitivity analysis show that accurate modelling of platoon dispersion in developing traffic signal timing plans is indeed important to the effective and efficient implementation of those plans. In a field where improvements of two or three percent have a major impact on fuel consumption, the amount and type of emissions, as well as traffic accidents, the differences shown in the above analysis indicate that the platoon dispersion algorithm should be calibrated for local site conditions. (A)
Abstract