Maximizing the operational efficiency with which the existing roadway capacity is utilized has been the main objective of the Active Traffic Management (ATM), which is considered as the most cost-effective approach by the traffic engineers who need to manage congestion and improve safety with the limited resources. While various ATM strategies have been developed for freeway management, the variable speed limit control has been emerged as one of the most effective tools in improving the safety of the traffic flow under the recurrent and non-recurrent bottleneck conditions. For example, the recent study by Papageorgiou et al. shows that the speed limits have an effect of decreasing the slope of a flow-occupancy diagram, thus reducing flow rates and suppressing shock waves. However, the current state of the practice in VSL control has not reached a point where optimal speed management strategies for potential bottlenecks are automatically determined and implemented in real time in a proactive manner. To be sure, most VSL systems currently in operation focus on harmonizing traffic flows, i.e., reducing speed differences, or providing safe speed limits under the prevailing traffic and environment conditions without explicit consideration of mitigating shock waves caused by downstream bottlenecks. Literature search resulted in relatively few past research efforts in developing dynamic speed limit strategies that were explicitly intended to manage shock waves. Early studies proposed simple threshold-based dynamic speed control laws. A model-based predictive control approach with VSL for suppressing shock waves was proposed by Hegyi et al. Lin, et al. also developed a model-based optimal VSL strategy for work zones to maximize throughput while minimizing delay. Recently Hegyi et al. proposed a VSL algorithm called SPECIALIST, which tries to determine the VSL solutions for the shock waves that are solvable. As reviewed above, the existing VSL strategies found in the literature either require extensive data collection, such as current status of shock waves, or time-consuming model calibration, which may not be realistic under the current environment. The previous phase of this study had developed a practical variable advisory speed limit (VASL) system, which has been implemented at the I-35W freeway corridor of the metro network in Minnesota. In this study, the effectiveness of the I-35W VASL system is assessed with the real traffic data and an enhanced version is developed to improve the responsiveness of the VASL algorithm in reflecting the current field conditions. The second part of this study has refined and implemented the adaptive ramp metering strategy, which was also developed in the previous phase of this research. While the adaptive ramp metering has long been considered as the most effective ATM strategy, the common issues found from the existing algorithms developed to date can be summarized as follows: • The time-variant bottleneck structure of a given freeway is not explicitly identified in real time. While the location of the active bottlenecks, thus their association with upstream ramps, can vary through time depending on the scope of the congestion in a given roadway, most coordinated algorithms currently in operation are based on the fixed metering zones bounded by pre-defined bottlenecks or have a pre-assigned association between ramps and potential bottleneck stations. • The pre-determined association between a potential bottleneck and upstream ramps, or the inherent fixed-zone structure with pre-defined bottlenecks may restrict the flexibility of traffic operations in dealing with unexpected incidents and detector malfunctions, which can happen frequently in real field operations. • Most operational systems try to maintain the flow rate or the density level at a bottleneck point to the pre-specified target value and the traffic conditions at the segments between bottlenecks are not explicitly considered in determining the metering rates. Depending on the type of a bottleneck, e.g., a diverge bottleneck as pointed out by Cassidy the relationship between the traffic conditions at a given bottleneck point and those on the upstream mainline segment may not be same throughout a given corridor. • Recently a new metering system called HERO, adopting a heuristic coordination approach, has been implemented in Victoria and Queensland, Australia. According to Dekker et al., HERO starts with a local feedback metering, called ALENEA, for all the ramps in a given corridor and starts a coordinated metering when a ramp becomes a ‘mater’ ramp, i.e., the traffic condition at that ramp area meets a set of the pre-specified congestion thresholds. This scheme is similar to that of the HELPER algorithm, but in HERO, each ‘master’ ramp has preassigned set of ‘slave’ ramps whose metering rates are coordinated in a heuristic manner to improve the traffic occupancy level at the mainline of the master ramp area. The new coordinated adaptive metering strategy, which is enhanced and implemented in this study, addresses the above issues by identifying the time-variant bottleneck structure for a given corridor in real time and tries to reflect the behaviour of the traffic flow reacting to the control in determining the metering rates. Further, the operational limitations of the metering control, e.g., ramp wait time and queue size restrictions, are explicitly considered in the algorithm. Chapter 2 includes the assessment and enhancement of the I-35W variable advisory speed limit system. The field traffic data from the I-35W corridor collected before and after the activation of the VASL system was analysed. Based on the analysis results, an enhanced version adopting a variable parameter is developed to reflect the current field-specific traffic conditions in determining the variable speed limits. The simulation study to adapt the enhanced version to a new corridor is also conducted in Chapter 2. Also, a strategy to locate an advance warning sign before the first VASL sign was developed in Chapter 2 to improve the responsiveness of the drivers approaching the speed control section. Chapter 3 contains the refinement and field implementation results of the new adaptive coordinated ramp metering strategy, which was developed in the previous research. Finally Chapter 4 contains the conclusions and future study needs. (Author/publisher)
Abstract