Real-Time Crash Risk Reduction on Freeways Using Coordinated and Uncoordinated Ramp Metering Approaches

Abstract

This study was conducted to examine the ability of two ramp metering strategies at reducing the real-time crash risk along a typical urban freeway. The strategies were tested using a simulated freeway network of Interstate-4 in Orlando, Fla. created in PARAMICS microsimulation. Measurements of the crash risk along the freeway were estimated using models created by the first writer which calculate rear-end and lane-change risks along the freeway using offline geometric characteristics and real-time loop detector data. The ramp metering strategies tested were the uncoordinated ALINEA algorithm and the coordinated Zone ramp metering algorithm. Additionally, two implementation methods of these algorithms were examined: the traffic-cycle realization which allows vehicles to enter the network in platoons and the one-car-per-cycle (OCPC) realization which allows a single vehicle to enter the freeway per traffic cycle. This study shows that the ALINEA and Zone algorithms both successfully reduce the real-time crash risk along the freeway. Comparing the two implementation methods, the traffic-cycle realization provides better safety and operational benefits than the OCPC realization. However, the OCPC realization provides better safety benefits when applied with the ALINEA algorithm. In general, the ALINEA algorithm performs better using shorter cycle lengths while the Zone algorithm performs best using longer cycle lengths. Comparing the two metering algorithms, the ALINEA algorithm proves to be superior to the Zone algorithm in providing the best overall safety benefits since it is more restrictive and generally allows fewer vehicles onto the network. However, at the 90% loading scenario, evidence of crash risk migration (the lowering of crash risk at one location combined with the increase at another) appears when the ALINEA algorithm is applied. This crash risk migration is a function of the specific geometry and demand patterns of the network used and is reduced by applying the Zone algorithm. This shows that while the ALINEA algorithm provides better overall results, there are cases where a less restrictive algorithm would perform better from a safety perspective.