An integrated simulation and optimization model for freeway work zones

The main objective of this study was to develop and test a methodology for optimizing traffic control procedures at urban freeway work zones. To accomplish this goal, a microscopic traffic simulation and optimization model was developed. The integrated model has the capabilities of realistically representing the urban freeway setting (basic segments with/without lane closures and/or ramps), and of formally optimizing a prespecified traffic system performance measure. The model was thoroughly verified and subsequently validated against field data collected at several sites on the Chicago Area Expressway System. Predicted responses were statistically similar to those observed in the field. It was found from field measurements that the lane capacity at work zones (1900-2000 vph) is higher than that indicated in the literature. Formal optimization of system travel time and probability of stoppage in the closed lane (i.e., lane to be closed) was performed with the model, assuming various profiles of vehicle arrivals. The decision variables were the proportion of drivers in the closed lane(s) that must evacuate it (them) at various points upstream of the taper. The general findings from the study indicated that the optimum merge execution points were uniformly distributed upstream of the construction taper. In other words, while the merge attempt distributions may vary, the merge completion distributions are virtually identical. The study also showed that the optimum merge pattern results in 20% of closed lane traffic completing their merges in the taper area. Regarding the arrival pattern, it was concluded that system performance with cyclic arrivals was generally inferior to that with uniform arrivals, although the differences diminished as volumes decreased. Furthermore, results indicated that average travel time, probability of stoppage and speed gradient can be used interchangeably to optimize system performance in terms of their effect on delays and safety. Finally, the model may be used to investigate macroscopic system behavior on the basis of its microscopic components.