Performance Evaluation of Adaptive Ramp Metering Algorithms in PARAMICS Simulation Lianyu Chu, Henry X. Liu, Will Recker California PATH, UC Irvine H. Michael Zhang Department of Civil and Environmental Engineering, UC Davis
Presentation Outline Introduction Methodologies Evaluation study – Calibration & Validation – Ramp metering algorithms – Evaluation results Conclusions
Background California PATH program Project Objective – Evaluating ramp-metering algorithms in a micro- simulation environment
Introduction Categories of ramp-metering control Fixed-time Local traffic responsive – ALINEA Coordinated traffic responsive – BOTTLENECK – ZONE
Methodologies Choosing an ITS-capable model (PARAMICS, VISSIM, AIMSUM2,…) Developing ATMIS modules Good calibration of studied network Development, design, calibration and optimization of ramp-metering algorithms Performance evaluation under different scenarios
Methodologies Micro-simulator PARAMICS Scalable, high-performance microscopic traffic simulation package ITS-capable API programming => Capability enhancement through API development
Methodologies API development: A Hierarchical Approach Provided API Library ATMIS Modules Developed API Library Advanced Algorithms Data Handling Routing Ramp Signal CORBA Databases Adaptive Signal Control Adaptive Ramp Metering Dynamic Network Loading Demand
Methodologies Evaluation framework
Evaluation study study site
Evaluation study Network coding in PARAMICS
Evaluation study Model calibration Accurate Network Geometry Vehicle characteristics & Performance The proportion of vehicle types Driving restrictions The signposting setting for links Driver behavior factors in car-following and lane-changing models
Evaluation study Model Validation (volume-occupancy) Real worldSimulation Loop 3.04
Evaluation study Model Validation (volume comparison)
Evaluation study ALINEA maintaining a optimal occupancy on the downstream mainline freeway Calibration: – K R = 70 – O desired = 20% – Location: 60 m
Evaluation study BOTTLENECK System level metering rate – Occupancy at Downstream > Desired occupancy – Vehicle storage in the section Local level metering rate:Occupancy control Calibration: - Area of influence of each section - Weighting factor of each on-ramp
Evaluation study ZONE System level metering rate: volume control Local level metering rate:Occupancy control Calibration – Identify bottleneck, divide the network into zones – 6-level metering plan for each entrance ramp
Evaluation study Assumptions & experimental designs Override strategy. Metering rate restriction No diversion Same occupancy control calibration used in BOTTLENECK and ZONE. 15 simulation runs for each scenario Compared with fixed-time control
Evaluation Study Performance measures Total vehicle travel time (TVTT) Average mainline travel time (AMTT) Total mainline delay (TMD) Total on-ramp delay (waiting time) (TOD)
Evaluation study Scenarios Morning peak hour (6:30-10:00) – highly congestion – lower congestion Incidents : block the rightmost lane for 10 minute – at the beginning of congestion – at the end of congestion
Evaluation study algorithms to be evaluated ALINEA Traditional BOTTLENECK Improved BOTTLENECK: replacing the local control strategy, i.e. occupancy control, with ALINEA control ZONE Improved ZONE
Evaluation study Total vehicle travel time
Evaluation study Average mainline travel time
Evaluation study Total mainline delay
Evaluation study Total on-ramp delay
Evaluation results All algorithms can be used for improving freeway congestion. ALINEA shows very good performance under all scenarios. The two coordinated ramp-metering algorithms, i.e., BOTTLENECK and ZONE, are a little more efficient than ALINEA under normal conditions. Compared with ZONE, BOTTLENECK can identify a bottleneck dynamically. Coordinated algorithms can be improved by integrating a better local algorithm, such as the ALINEA algorithm.
Conclusions A capability-enhanced micro-simulation laboratory has been developed for evaluating ramp-metering algorithms, potentially, some ATMIS applications. Adaptive ramp-metering algorithms can ameliorate freeway traffic congestion effectively. Compared with local algorithm, coordinated algorithms are more efficient, but the improvement is limited.
More Information PATH reports: – Liu, X., Chu, L., and Recker, W. PARAMICS API Design Document for Actuated Signal, Signal Coordination and Ramp Control, California PATH Working Paper, UCB-ITS-PWP – Zhang, H. M., Kim, T., Nie, X., Jin, W., Chu, L. and Recker, W. Evaluation of On-ramp Control Algorithm, California PATH Research Report, UCB-ITS-PRR