Comparative performance analysis of variable speed limit systems control methods using microsimulation: A case study on D100 freeway, Istanbul

Growing traffic on freeways has created demand for traffic control systems which can manage it efficiently without the need of expanding the infrastructure. Intelligent Transportation System techniques like Ramp Metering, Route Guidance and Variable Speed Limits (VSL) can help in easing out congestion on freeways. While Ramp Metering and Route Guidance system have limited scope of application, VSLs can be used for preventing traffic breakdown by regulating flow upstream of a congestion. This is achieved by imposing speed limit before critical density or flow is attained. Fixed control approaches based on occupancy, implemented on I-4 Orlando Florida, and flow, implemented on M-25 England, have been utilized in past. Many studies have reported improved performance with better control strategies however mostly in hypothetical test scenarios. Aforementioned field implementation also used simple algorithms. Thus this study presents a comparative performance evaluation of two control methods in VSLs implementation: One of the control methods is adapted and modified from Allaby et.al.(2007) which is based on flow occupancy and speed while second control method is adapted from Müller et.al (2015) based on feedback control based on occupancy to calculate VSLs. Many segments of D100 freeway of Istanbul experience recurrent congestion. A section of D100 freeway, that has heavy congestion due to merging traffic from on-ramps, is chosen accordingly as the case to analyze in this study. Microscopic traffic simulation software VISSIM is used for modeling and simulating the traffic. Calibration of VISSIM is sought using Remote Traffic Microwave Sensor data provided by Istanbul Municipality. Parameters including headway time, car following variation, oscillation acceleration are tuned until real and simulated speed profiles match. GEH test is performed to validate the calibration. VISSIM is integrated with MATLAB via Component Object Model interface. Values of flow, occupancy and volume are retrieved from VISSIM by MATLAB in real time. These values are used to determine the VSLs by running the control algorithms and are fed back to VISSIM. The present study shows that both control methods reported better results when compared no control case in terms of Total Time Spent (TTS) through the selected network even in heavy traffic flow. The feedback control strategy performed relatively better. Also, the speed profile achieved with feedback control is more homogenized when compared to resultant speed profiles from the algorithm used in Allaby et.al.(2007). It can also be concluded that application area of VSL, acceleration area, the position of signs and detectors affect the performance of VSL control systems