Application of Variable Speed Limits and Ramp Metering to Improve Safety and Efficiency of Freeways

Research recently conducted at the University of Central Florida involving crashes on Interstate-4 (I-4) freeway in Orlando, Florida led to the creation of new statistical and neural networks models that are capable of determining rear-end and lane-change crash risks along the freeway in real-time. For determining the rear-end crash risk, it was found that rear-end crashes occur within one of two distinct traffic regimes: regime 1 (congested conditions) or regime 2 (uncongested conditions). The lane-change crash risk was mainly based on a single neural network model. Based on these models, the safety and travel time effects of two Intelligent Transportation Systems (ITS) strategies, ramp metering and variable speed limits (VSL), for reducing rear-end and lane-change crash risks along I-4, were evaluated. Using the PARAMICS micro-simulation package, a 36.25 miles of I-4 was simulated. For applying VSL, the idea of homogeneous speed zones was introduced to determine the distance over which VSL should be implemented from the station of interest. Two ramp metering strategies were tested on this simulated network, the uncoordinated ALINEA algorithm and the coordinated Zone algorithm. Two implementation methods of these algorithms were examined, the traffic-cycle (TC) realization and the one-car-per-cycle (OCPC) realization. Moreover, the potential of combining both ITS strategies (ramp metering and VSL) along the same simulated network was also tested. There were two combination methods used, implementing ramp metering everywhere (through the whole network) in conjunction with VSL everywhere, as well as implementing ramp metering downtown (in downtown areas only) in conjunction with VSL everywhere. It was concluded that VSL is an effective crash prevention strategy when the freeway is operating at uncongested conditions. At free-flow conditions (60 percent loading scenario) and conditions approaching congestion (80 percent loading scenario), VSL can be used to reduce crash risk and prevent crash occurrence, in addition to reducing travel time, especially at free-flow conditions. The highest decrease in travel time is found to be around 0.8%. Moreover, VSL was not found to effectively reduce crash risk at congested situations (90 percent loading scenario). The study indicated that both ramp metering algorithms successfully reduce realtime crash risk along I-4. The traffic-cycle realization provides better safety and operational benefits when applied with the ALINEA algorithm. The ALINEA algorithm works best with shorter cycle lengths, while the Zone algorithm performs best with longer cycle lengths. The ALINEA algorithm proves to be superior to the Zone algorithm at reducing the crash risk as it is more restrictive. In spite of the fact that the ALINEA Abdel-Aty, Haleem, Cunningham and Gayah 2 algorithm provides better overall results, the Zone algorithm is much better in terms of the overall network travel time (measure of efficiency), where the Zone algorithm yielded smaller travel times than the ALINEA algorithm. For examining the combination of both ITS strategies, applying ramp metering alone was more beneficial at congested situations, while applying VSL alone was more beneficial at free-flow conditions. At conditions approaching congestion, the combination of ramp metering and VSL produced the best benefits. These results illustrate the significant potential of ITS strategies to improve the safety and efficiency of urban freeways.