Effects of Variable Speed Limit System on Travel Time Reliability
M A Mohammadi, G H Bham, Missouri University of Science and Technology, US
This paper evaluates the effects of the VSL system on travel time and reliability. Results indicated that travel times reduced; however, it should be carefully evaluated as the improvements were not found to affect drivers' perception.
Variable Speed Limit (VSL) systems can be an effective traffic control strategy that smooths transition from free-flow to stop-and-go conditions as well as reduce overall travel time, increase average speed and total throughput. Studies on the effects of VSL at work zones have shown that they are efficient in reducing the speed variance, increasing traffic speed and the roadway throughput. Other studies have shown that the VSL system can improve the efficiency of traffic flow on recurring congestion as well as traffic safety. The effects of the VSL system on travel time reliability; however, have not been investigated. This paper evaluates the impact of a VSL system implemented on I-270 in St. Louis, Missouri on travel time and its effectiveness in improving the travel time reliability. The system was initiated in May 2008 to reduce recurring and non-recurring congestion, and improve traffic safety. Three segments (A, B & C) on I-270 instrumented with remote traffic microwave sensors (RTMS) were evaluated using peak period data for three busiest and all days of the week. These segments were the most congested and data for nine months were used in the analysis. This paper presents an "uncontrolled" evaluation of the VSL system as the effects of incidents, inclement weather, etc. were included in the data.
The methodology in this study incorporates various measures of travel time reliability to thoroughly assess the changes in mean and variation of travel time for pre-VSL and VSL conditions. The paper will present the reliability metrics recommended by SHRP2 (skew statistic, width statistic, misery index, and on-time/failure rate)and three nationally recognized indices: travel time index (TTI), buffer time index (BTI), and planning time index (PTI) to evaluate the effects of the VSL system. This paper will also discuss challenging interpretations of the BTI through different scenarios when the change in mean and 95th percentile travel time changes are different. This aspect of the BTI, which is one of the most used reliability metrics, makes it a sensitive tool for the evaluators that require careful inspection.
The extrapolation method was used to estimate travel time. The key assumption in this method is that the speed measured at the point detector is approximately equal to the average speed for the entire zone of influence i.e. half the distance upstream and downstream to the detector. Figure 2 illustrates the concept of zone of influence for each detector. The accuracy of the estimated travel time from the speeds measured by loop detectors was verified by comparing it to the travel time measured by GPS probe vehicles.
The preliminary results of TTI, BTI and PTI indicate a decrease in all indices for VSL conditions except in segment A, however, the increase in TTI is not considerable. BTI, for segments B and C improved around 11-27%, whereas for segment A worsened by 2-20%. PTI, for segments B and C improved around 6-11%. Overall results of BTI indicate that the VSL system reduced additional time travelers need to be on-time 95% of the times. ). The VSL system affected the skewness of travel time distribution differently than it affected other statistics. results of Width statistics indicates less variation in travel time for segments B and C for VSL condition. Misery index, an indication of miserable trips (i.e. mean travel time value of the 5% of trips with the longest travel times) relative to the trips with free flow speed. Misery index showed that the travel times for the worst 5% of trips improved by 6-12% for segments B and C for VSL conditions; however, for segment A the change was not considerable. Overall, changes in reliability indices were observed for all three segments. All segments generally improved reliability metrics; however, segment A showed a decline in metrics such as the BTI, width statistic, and misery index. Mean travel time decreased for all segments in almost all cases. Variation in travel time decreased, an indication of smoother traffic with lesser variation in speed. Eightieth (80th) percentile travel time did not show a clear trend for segment A, but decreased for segments B and C by 3-8 seconds/mile. The analysis indicates that average speeds were harmonized and travel times were reduced; however, in spite of the improvements, result should be carefully evaluated as the improvements were not considerable to affect drivers? perception of the system and might not be significant for a viable system.
Association for European Transport