Truck Priority Evaluation

Transcription

1 Truck Priority Evaluation Minnesota Department of Transportation MnDOT Contract No W02 SEH No. MNTMD January 26, 2012

2 Truck Priority Evaluation Minnesota Department of Transportation MnDOT Contract No W02 SEH No. MNTMD January 26, 2012 I hereby certify that this report was prepared by me or under my direct supervision, and that I am a duly Licensed Professional Engineer under the laws of the State of Minnesota. Roger A. Plum, PE Date: January 17, 2012 Lic. No.: Reviewed by: January 17, 2012 Date Short Elliott Hendrickson Inc Vadnais Center Drive St. Paul, MN

3 Table of Contents Letter of Transmittal Certification Page Table of Contents Page 1.0 Overview Site selection System Design and Installation Controller setting modifications Truck detection validation Data collection September Results from September Data Collection Additional Data Collection November Results from Additional Data Collection (November) Potential Benefit Annual Benefits Conclusions and Recommendations List of Tables Table 1 Measures of Effectiveness and Auxiliary Related Data (September 27-29, 2011)... 9 Table 2 Measures of Effectiveness and Auxiliary Related Data (November 16-17, 2011).. 13 Table 3 Evaluation of Quantities of Trucks Receiving Benefits from Priority List of Figures Figure 1 TH 24 & County Road 8, Sherburne County... 3 Figure 2 Vehicles Detected as Trucks... 5 Figure 3 Vehicles Not Detected as Trucks... 6 SEH is a registered trademark of Short Elliott Hendrickson Inc. Page i

4 January 2012 Truck Priority Evaluation Prepared for the Minnesota Department of Transportation 1.0 Overview The purpose of this project is to evaluate the potential benefits and the actual effectiveness of implementing a detection system which would provide priority for heavy commercial vehicles at signalized intersections where these vehicles represent a significant percentage of overall traffic. The truck detection system consists of a pair of conventional 6 x6 loops located upstream of the existing normal extension detectors on each approach. The two detectors in the truck detector pair are separated by 24 feet. Using the logic processor capabilities of the traffic signal controller (Econolite ASC/3), a pseudo-detector on each approach was created which is active only when both of the truck detector loops in the pair are occupied. Because of the spacing of the pair of loops, this should only occur when a long vehicle (longer than approximately 24 feet) is passing over the detectors. Under free-flow conditions, this pair of detectors could be concurrently occupied by two vehicles only if the second vehicle is following very closely to the point of being extremely unsafe. (At 55 miles per hour, the headway between these two vehicles would need to be less than ½ second.) The pseudo-detector was then assigned to the appropriate phase to allow for extension of the green indication on the approach, providing an extra extension of the green for trucks to reach the previously-existing extension detector. The theory behind providing this extra extension is to reduce the number of trucks which have to stop. Due to their relatively low acceleration characteristics, trucks which stop (and any vehicles which may be queued behind them) experience more delay than passenger vehicles. 2.0 Site selection At the project kick-off meeting on April 25, 2011, between Short Elliott Hendrickson Inc. (SEH )and the Minnesota Department of Transportation (MnDOT), the following desired characteristics were identified for potential sites: Signalized Single-lane approaches a truck priority detection project had been conducted several years earlier by MnDOT on USH 169 in Belle Plaine. At that intersection, the benefits of truck priority were not fully realized due to the fact that both mainline approaches contained two through lanes. Consequently, passenger vehicles queued behind a stopped truck generally were able to pass the truck and minimize their delay once the queues began to dissipate. Therefore, the primary benefactors of the system were the commercial vehicles. With single-lane approaches, it was expected that commercial vehicles would benefit and passenger vehicles would benefit due to not being caught behind slowly accelerating trucks as often. Page 1

5 Truck traffic should be percent of overall traffic Intersection should not be close to another signalized intersection (to avoid the impacts of traffic platoons on the results) No advance warning flasher (AWF) system though not identified as a fatal flaw, the AWF would mask the effects of truck priority due to its own green extension characteristics Consideration of seasonal harvest traffic (not necessarily required) Prior to that meeting MnDOT had already prepared a preliminary list of six candidate intersections. Based on the above desired attributes all of the MnDOT districts were contacted, and an additional three candidate intersections were identified. Following is the list of candidate intersections. TH 5 & County Road 14, Washington County USH 12 & TH 15, Meeker County USH 12 & County Road 139, Wright County USH 14 & USH 71, Redwood County TH 55 & County Road 19, Hennepin County USH 71 & TH 7, Kandiyohi County USH 2 & CSAH 13, St. Louis County USH 14 & County Road 11, Olmsted County TH 24 & County Road 8, Sherburne County The majority of the above locations were eliminated after a more in-depth review of their traffic characteristics. The intersections in Washington County, Meeker County, Wright County, Hennepin County, and Olmstead County were found to have heavy commercial vehicles comprising less than 10 percent of the total traffic. Of the four remaining intersections, TH 24 & County Road 8 in Sherburne County was determined to be a better candidate than the rest because of its high number of commercial vehicles (over 1800 per day) relative to the other three, none of which had daily truck traffic greater than 700 per day. The Wright County intersection, with 1240 commercial vehicles per day (but only 8.8 percent of overall traffic) was the only other intersection with more than 1000 commercial vehicles per day. Figure 1 is an aerial depiction of the selected intersection, with northbound TH 24 traffic traveling from the lower left to the upper right of the figure. Page 2

6 Figure 1 TH 24 & County Road 8, Sherburne County 3.0 System Design and Installation Following selection of the TH 24 & County Road 8 intersection as the test site on May 25, MnDOT developed a design and prepared construction plans and specifications based on the previously-developed MnDOT truck priority system design and guidelines developed. Following the state government shutdown, final plans and specs were released by MnDOT on August 9. Design Electric was selected and approved as the electrical contractor for the project on August 23. Notice to proceed occurred on August 30. Construction occurred and was essentially completed on September 9. A new controller (Econolite ASC/3) was installed by MnDOT, replacing the ASC/2 controller and allowing the use of logic processing to manage the truck detectors and gather other data automatically. MnDOT completed their detector connections in the cabinet and programming of the controller by September Controller setting modifications SEH made modifications to the controller settings entered by MnDOT for only two reasons: to activate or de-activate the truck priority operation, and to allow the controller to automatically collect additional data of interest for this project. No timings were altered by SEH. In the controller cabinet MnDOT wired the new loop detectors as detectors 17 and 18 for northbound traffic and detectors 19 and 20 for southbound traffic. Page 3

7 Using the logic processor capabilities of the ASC/3 controller, SEH created pseudo-detectors, the data from which can be obtained from the controller as volume and occupancy information in the same manner as conventional detector data. SEH created the following pseudo-detectors in the controller: Detector 25 was used as the northbound truck detector and was active only when detectors 17 and 18 were active simultaneously. Detector 27 was used as the southbound truck detector and was active only when detectors 19 and 20 were active simultaneously. Detector 26 was set up to be active when the northbound truck detector (#25) was active and that detector would have provided a green extension to phase 2, the phase serving the northbound through movement. In other words, this detector was active when a truck was detected and the truck would have received the benefit of the green extension. Detector 28 was set up in a similar fashion for the southbound direction, being active when a truck was detected and that truck would have received the benefit of the green extension. Detectors (northbound) and detectors (southbound) were set up to identify the frequency of conditions when the truck detector was activated but no benefit could be provided. The conditions under which no benefit would be provided to the truck are: The truck is detected when the through phase in that direction is not green, or The truck is detected during the first portion of the minimum green time, and the remaining minimum green time would have allowed the truck to reach the previously-existing extension loop, or The truck is detected after the minimum green time for the through phase has expired, but there is no call for a conflicting phase, in which case the signal indication would have remained green anyway. Detector 45 was set up to record activations of detector 10 (EB detector 475 upstream of the stop line) when the eastbound phase (4) is not green, i.e. arrivals during yellow or red. Detector 46 was set up to record activations of the detector 4 (EB stop line detection) when phase 4 is not green. The occupancy from this detector can be used as a measure of how long eastbound vehicles are waiting for service at the intersection. Detectors 47 and 48 were set up for the westbound direction to accomplish the same tasks as detectors 45 and 46 provided for the eastbound direction. Detectors 51, 52, 54, 55, 56 and 58 were set up to monitor when each of the signal phases (1, 2, 4, 5, 6, and 8, respectively) were active. 5.0 Truck detection validation Prior to starting data collection, the ability of the system to detect trucks needed to be confirmed. As expected, because the detection is based on a certain length of vehicle, not all vehicles detected as trucks were actually heavy commercial vehicles, and not all commercial vehicles were detected as trucks. Pickup (or larger) trucks towing trailers were frequently detected as trucks. This fact is not believed to be a negative consequence of the method used to detect trucks. In many of these cases, these vehicles towing trailers also exhibited low acceleration rates, perhaps not as low as tractor-trailer, but low enough to have a negative impact on following traffic to the point where extending the green for them could be beneficial to them and traffic following them. Page 4

8 Below are photographs of vehicles detected as trucks (Figure 2) and not detected as trucks (Figure 3). With the exception of the tractor-trailer in the upper left photograph, an attempt has been made to show the smallest vehicles detected as trucks and the largest vehicles not detected as trucks. It should be emphasized that during the validation process there were no cases in which a tractor-trailer failed to be detected as a truck. Figure 2 Vehicles Detected as Trucks Page 5

9 Figure 3 Vehicles Not Detected as Trucks 6.0 Data collection September Stop and delay data collection initially was conducted Tuesday through Thursday, September Data collected on September 27 and 28 represent the before condition, with the truck detectors not assigned to any phases and, therefore, not extending green indications. Just prior to data collection on September 29, the truck detectors were assigned to the phases serving the northbound and southbound through traffic on TH 24. Stop and travel time data on each approach were collected manually using Jamar electronic data collection boards. Data was collected for the before (truck detectors deactivated) and after (truck detectors activated) conditions during the two hours identified as the morning and afternoon peak hours (7:15-8:15 and 16:00-17:00) and during two additional non-peak hours (11:30-12:30 and 13:30-14:30) identified through volume data as being non-peak hours with the highest level of truck traffic. Data collected on the Jamar boards was done in an unconventional fashion. The conventional method of using the Jamar boards is to press one of two buttons once for each vehicle to record that vehicle as either a stopped vehicle or a through (i.e., did not stop) vehicle, and also every 15 seconds enter a number into the board representing the number of vehicles queued on the approach upstream of the signal. However, the data needs for this project were to record separate values for trucks (or vehicles detected as trucks) and for non-trucks, and it was desired to record vehicles total time approaching and departing the traffic signal. Consequently, the conventional detectors and the pseudo-detectors described in the Controller setting modifications section were used to count the numbers of total vehicles and trucks on each approach the difference of which would yield the number of non-trucks and the Jamar boards were used to record, for each travel direction, the numbers of both Page 6

10 trucks and non-trucks which stopped, as well as the number of all vehicles within the data collection zone at the end of each 15-second interval. Trucks within the data collection zone were recorded simultaneously, on a portable computer. Also recorded on the portable computer were the locations of the trucks in the queue at the start of the green interval for the through movement. Of specific interest was the number of times the front vehicle in the queue is a truck, which means all of the traffic behind it will be slowed due to the slow acceleration of the truck leading the platoon. The data collection zone consisted of a segment of the roadway approximately 1600 feet long, extending from a point on TH feet south of the center of the intersection to a point 800 feet north of the center of the intersection. As stated earlier, the goal of collecting data both upstream and downstream of the signal was to capture the total impact of stopping trucks, both on themselves as well as on other traffic. Finally, to keep data collection as consistent as possible for the before and after conditions, the two sets of data were collected within two days of each other, and the same personnel were used to collect data in the same direction, to avoid differences between individuals as to what constituted a truck and what did not. The initial data collection, specifically the detailed data collection for determining stops, delays, and queue positioning, was conducted three different days providing only one data sample in each of four one-hour time periods under two conditions (truck priority not activated, and truck priority activated). This study is not considered statistically significant; however, it provides a synopsis of the truck priority system. Also, as mentioned above, signal timing entered by MnDOT was not altered, and some of the results may be reflective of signal timing idiosyncrasies, which will be discussed in the Conclusions and Recommendations section. 6.1 Results from September Data Collection Results from the initial manual and automatic data collection are shown in Table 1. Table 1 has been separated into four sections, labeled sections A through D. Before going into detail regarding the results, it should be noted, as shown in Section C, that the total volume entering the intersection in the after condition was greater during all time periods than in the before condition. That difference ranges from 3.0 percent greater during the 16:00-17:00 time period to 9.3 percent greater during the 11:30-12:30 time period. Since, in general, higher volumes lead to higher delays, travel times (a function of delay) in the after case would be expected to be somewhat higher than the before case, if all other factors were equal. Section A contains the primary measures of effectiveness identified in the evaluation plan. Zone travel time has been used rather than delay for two reasons. First, travel time within the data collection zone is the actual measure which was recorded. Second, with a 1600-foot data collection zone, at the speed limit of 55 mph the expected travel time of a vehicle under free-flow conditions over the length of the zone would be 19.8 seconds. As can be seen in a few instances (e.g., NB trucks in the after case), the average zone travel time is less than this value, an indication that the average speed through the zone was consistently higher than the speed limit. The results in Section A show that for the northbound direction, activation of truck priority appears to decrease travel times and, therefore, delay for trucks during all four time periods, and this effect seems to be magnified during the higher-volume periods (AM and PM peaks). Page 7

11 At the same time, travel times for non-trucks increase slightly in three of the four cases, a result which is not intuitive. The percentage of trucks which stopped increased slightly during the AM peak, but decreased (slightly during the afternoon off-peak, significantly during the midday off-peak and the PM peak) at other times. Stops for non-trucks followed these same trends. The other key measure of effectiveness is how frequently the first vehicle in the queue is a truck, causing delays to following vehicles. This is of particular interest because the detection of a truck slightly upstream of other vehicles would tend to provide a better opportunity for the front vehicle in the queue to not be a truck. For this study, trucks were slightly less likely to be the front vehicle in the queue under truck priority, with the exception of the AM peak period, during which trucks were more likely to be queued at the front. The differences between the before and after conditions for all time periods were significant. Page 8

12 Page 9 Section D Section C Section B Section A Table 1 Measures of Effectiveness and Auxiliary Related Data (September 27-29, 2011) Time Four Time Periods Combined Before After Before After Change Before After Change Before After Change Before After Change Before After Change NB zone travel time for non trucks (sec/veh) NB zone travel time for trucks (sec/veh) NB non trucks stopped (%) 21.5% 28.5% +6.9% 31.0% 23.3% 7.7% 27.4% 29.1% +1.8% 31.7% 23.9% 7.8% 28.3% 26.1% 2.2% NB trucks stopped (%) 21.1% 21.7% +0.6% 22.5% 10.8% 11.7% 16.4% 16.3% 0.1% 20.0% 10.6% 9.4% 20.1% 14.6% 5.4% NB truck is front vehicle in queue at start of green 17.5% 19.0% +1.5% 23.7% 10.7% 13.0% 18.3% 10.5% 7.8% 12.1% 6.9% 5.2% 17.8% 11.9% 5.9% SB zone travel time for non trucks (sec/veh) SB zone travel time for trucks (sec/veh) SB non trucks stopped (%) 39.8% 33.9% 5.9% 48.1% 41.8% 6.3% 59.3% 51.2% 8.1% 31.9% 22.9% 9.0% 43.8% 36.2% 7.6% SB trucks stopped (%) 33.3% 20.3% 13.0% 23.1% 18.1% 5.0% 30.2% 17.5% 12.7% 32.1% 25.6% 6.4% 29.9% 19.6% 10.3% SB truck is front vehicle in queue at start of green 18.0% 12.1% 6.0% 18.6% 16.1% 2.6% 20.0% 12.3% 7.7% 10.3% 10.3% 0.0% 16.8% 12.7% 4.1% Occurrences of phase 1 (SBLT) Occurrences of phase 2 (NBT) Occurrences of phase 4 (EB) Occurrences of phase 5 (NBLT) Occurrences of phase 6 (SBT) Occurrences of phase 8 (WB) % time phase 1 (SBLT) 1% 1% +1% 1% 1% 0% 0% 1% +1% 1% 0% 0% 0% 1% 0% % time phase 2 (NBT) 72% 71% 1% 74% 76% +1% 73% 75% +2% 70% 74% +4% 72% 74% 2% % time phase 4 (EB) 26% 27% +1% 24% 23% 1% 27% 24% 3% 28% 25% 4% 26% 25% 2% % time phase 5 (NBLT) 11% 11% 1% 11% 7% 4% 8% 10% +2% 16% 13% 3% 11% 10% 1% % time phase 6 (SBT) 61% 61% 0% 64% 69% +5% 65% 66% +1% 55% 61% +6% 61% 64% 3% % time phase 8 (WB) 26% 27% +1% 24% 23% 1% 27% 24% 3% 28% 25% 4% 26% 25% 2% NB volume (all vehicles) SB volume (all vehicles) EB volume WB volume Total volume NB Truck Percentage 14% 13% 1% 20% 20% 0% 16% 17% +1% 9% 10% +2% 13% 14% 1% SB Truck Percentage 17% 15% 1% 17% 19% +2% 18% 24% +5% 13% 10% 3% 16% 17% 0% EB stop line occupancy 4% 3% 1% 4% 1% 3% 4% 7% +3% 10% 7% 3% 6% 5% 1% WB stop line occupancy 31% 45% +14% 38% 50% +12% 24% 21% 4% 24% 20% 4% 29% 34% 5% NB trucks detected NB trucks detected with potential benefit % NB trucks with benefit from priority 20% 26% +6% 14% 12% 2% 10% 19% +8% 40% 40% 0% 27% 29% 2% NB truck detected but phase not green 34% 30% 3% 35% 30% 5% 22% 29% +7% 22% 25% +3% 27% 28% 0% NB truck detected during minimum green 15% 22% +6% 15% 23% +7% 30% 26% 4% 29% 28% 1% 24% 26% 2% NB truck detected during green, no conflicting call 14% 23% +9% 31% 38% +7% 33% 17% 15% 9% 8% 1% 17% 17% 0% SB trucks detected SB trucks detected with potential benefit % SB trucks with benefit from priority 12% 22% +10% 17% 11% 6% 8% 23% +15% 25% 26% +1% 17% 21% 4% SB truck detected but phase not green 39% 47% +7% 42% 40% 2% 52% 39% 14% 55% 49% 6% 48% 43% 4% SB truck detected during minimum green 12% 25% +13% 17% 25% +8% 19% 16% 3% 23% 26% +3% 18% 22% 4% SB truck detected during green, no conflicting call 36% 6% 30% 23% 24% +1% 21% 23% +2% 0% 0% 0% 17% 14% 4%

13 In the southbound direction, travel times for trucks in the zone did not demonstrate a consistent trend to increase or decrease across the four time periods, actually increasing slightly with truck priority activated in three of the four time periods. For non-trucks, three of the four time periods showed the opposite trend, with a slight decrease in travel time. These comparisons of trucks and non-trucks are interesting because they are in the same traffic stream, and the changes would be expected to be consistent for the two groups. However, when the stop data is examined, the results are much more definitive, both across time periods and across vehicle types. In all cases, southbound vehicles experienced fewer stops when truck priority was activated, with trucks showing greater overall reductions than non-trucks. Finally, when examining the likelihood of a truck being at the front of the queue when the through signal changes to green, it can be seen that in three of the four cases, that likelihood is reduced when truck priority was activated. In the PM peak case, the percentage remained unchanged. Section B highlights the signal timing attributes under both the de-activated and activated conditions. The upper half of that section shows how many times each phase was served during the hour in which data was collected. The lower half shows the percentage of time each phase was active during that hour. With only a handful of exceptions, the number of occurrences of any given phase decreased slightly when truck priority was activated. This result is consistent with expectations. As trucks are detected farther upstream, each through movement green would be extended, increasing the average cycle length slightly and thereby reducing the total number of cycles within the hour. On the other hand, the percentages of times each phase was active remained relatively stable, which would also be expected and consistent with the fully-actuated control at this intersection. As other phases green times increase, on average more vehicles will arrive during any given phase s red time. As a result, that phase s green time needs to also increase slightly to accommodate the extra vehicles. As long as the conditions remain uncongested (as was observed) the percentages should remain relatively stable within the same time periods. Section C shows the volumes on each approach as well as the mainline truck percentages in each direction and the occupancies of the stop line detectors on the eastbound and westbound approaches. As mentioned above, the interesting aspect of the volumes is the consistently higher volume in the after condition than in the before condition, which would tend to skew the results in favor of the before condition in the absence of other factors. With respect to the number of trucks as a percentage of total traffic, notice that the percentages are higher during the off-peak periods than during the peak periods. This is actually a reflection of the non-truck traffic more than the truck traffic, since the number of trucks during an hour (which can be seen in Section D) actually remained fairly constant over the four time periods and all days. The significant point of interest in section C was the eastbound and westbound detector occupancies at the stop lines, which reflect the percentage of time vehicles would have been waiting on those approaches. It was expected that the slight increases in green time due to truck priority on the mainline would have increased the wait time on the cross street. This proved to be the case only for the AM peak and midday off-peak periods for the westbound approach, which was generally busier than the eastbound approach. During the other two time periods, the percentage of wait time actually decreased slightly. Finally, section D of the table is intended to demonstrate how frequently the existence of the truck detectors would actually provide priority treatment. This was determined by using the pseudo-detectors described in the Controller setting modifications section, to collect Page 10

14 information about certain conditions. A truck was defined as receiving potential benefit from the detectors under the following conditions: 1. The signal indication for that truck is already green, and 2. Extending the green for the truck will extend the green beyond the programmed minimum green time, and 3. There is a call for a conflicting phase either on the cross street or for the opposing leftturn phase. With respect to item 1 above, if the signal is not already green, the detection of a truck will have no benefit since the purpose of the truck detector is to extend an existing green for the truck. With respect to item 2 above, if the controller is still timing its minimum green time and sufficient time remains in that minimum to permit the truck to travel from the truck detector to the conventional extension detector, the truck detection will have no impact. Finally, with respect to item 3 above, until there is a call for a conflicting phase, the signal will remain green and, therefore, the truck detector provides no benefit. As shown in section D of Table 1, in all cases the trucks which benefit from the truck detection system represents less than half of all the trucks. However, as shown in the far right columns in Table 1, where the results from the four individual time periods have been weighted based on volumes, the priority system will benefit approximately 25 percent of all trucks. In the PM peak, 40 percent of the trucks would benefit. During the peak periods, the primary reasons trucks would not receive the benefit of priority is either because the signal is not green as the trucks approach or the minimum green will provide sufficient time for the truck to reach the stop line. As traffic levels shift from higher volumes during the peak hours to lower volumes in the off-peak hours, the primary reason for trucks not benefiting from the priority is that there is less conflicting traffic to terminate the green for the truck. From these results, there appears to be a definite trend showing benefits of the truck priority system. Aside from the travel time (delay) for NB non-trucks increasing by about 1 second per vehicle, the remainder of the measures show reductions in delays and stops to both trucks and non-trucks, as well as a reduction in the probability of the truck being stopped at the front of a queue. 7.0 Additional Data Collection November Upon review of the signal timing, MnDOT determined that an omission had been made in converting the ASC2 controller settings to ASC3 controller settings. Specifically, in the ASC2 controller the added initial function of the controller, which allows for a variable minimum green time based on detector actuations during red, is activated automatically by entering values in the actuations before added initial field, the seconds per actuation field, and the maximum initial field. For the ASC3 controller, activation of the added initial feature also requires that detectors be assigned to the function. Unfortunately, no detectors had been thus assigned prior to the September data collection effort. In early November, MnDOT personnel assigned detectors to the added initial function, and SEH conducted data collection for the non-priority and priority operations on Wednesday, November 16, and Thursday, November 17, respectively. Page 11

15 Data was collected at the same times of day as the September effort: 7:15-8:15, 11:30-12:30, 13:30-14:30 and 16:00-17:00. Eastbound and westbound vehicle stop and stopped delay time data on County Road 8, the minor street, was also collected during this effort. 7.1 Results from Additional Data Collection (November) The measures of effectiveness (MOEs) from this data collection effort are shown in Table 2. The new minor street stop and delay data appears in section C of the MOE table. For the majority of MOEs, the results obtained in November provide further support for the conclusions derived from the September data. Average travel times and, therefore, delays to both trucks and non-trucks on TH 24 tend to be slightly reduced, and the reduction to trucks appears to be slightly better than for non-trucks. Similarly, the percentages of trucks and non-trucks which stop for the signal appears to decrease due to the truck priority, and the reduction for trucks is markedly higher than for non-trucks. As shown in section B, in general the percentages of phase times for each phase did not change significantly between the non-priority and priority cases. For the minor street in the westbound direction, the average delay per vehicle did increase by a few seconds in each case, a result which would be expected with the lengthening of the mainline phases due to trucks being detected. Somewhat contrary to what was expected, the delays to minor street vehicles eastbound (approaching from the west) decreased slightly in the truck priority scenario. While not initially obvious, field observation of the intersection reveals that the majority of eastbound traffic is turning right onto TH 24. The lengthening of the mainline green for trucks (because it is lengthened for either northbound or southbound trucks) actually provided a greater opportunity for those eastbound right-turning vehicles to turn right on red. As was also observed in the September data, the data collected in November revealed a significantly higher overall traffic level on Thursday, over 12 percent higher than the traffic on Wednesday during the same four time periods. Page 12

16 Page 13 Section D Section C Section B Section A Table 2 Measures of Effectiveness and Auxiliary Related Data (November 16-17, 2011) Time AM peak Four Time Periods Combined Before After Before After Change Before After Change Before After Change Before After Change Before After Change NB zone travel time for non trucks (sec/veh) NB zone travel time for trucks (sec/veh) NB non trucks stopped (%) 36.2% 22.2% 14.0% 18.2% 13.4% 4.8% 14.8% 17.3% +2.6% 20.3% 21.9% +1.6% 22.2% 19.6% 2.5% NB trucks stopped (%) 25.9% 7.3% 18.6% 25.0% 1.9% 23.1% 7.4% 14.5% +7.1% 15.2% 13.0% 2.2% 18.1% 9.5% 8.6% NB truck is front vehicle in queue at start of green 20.0% 5.8% 14.2% 25.0% 3.9% 21.1% 6.5% 14.9% +8.4% 7.7% 5.5% 2.2% 14.9% 7.3% 7.6% SB zone travel time for non trucks (sec/veh) SB zone travel time for trucks (sec/veh) SB non trucks stopped (%) 30.7% 21.9% 9% 14.7% 21.1% +6% 22.8% 26.7% +4% 25.4% 31.0% +6% 24.2% 25.4% +1.3% SB trucks stopped (%) 29.0% 25.5% 4% 26.7% 12.3% 14% 23.9% 9.4% 15% 18.2% 37.0% +19% 25.2% 20.2% 5.1% SB truck is front vehicle in queue at start of green 18.3% 12.3% 6% 19.2% 17.2% 2% 19.2% 8.9% 10% 8.5% 17.5% +9% 16.1% 14.0% 2.1% Occurrences of phase 1 (SBLT) Occurrences of phase 2 (NBT) Occurrences of phase 4 (EB) Occurrences of phase 5 (NBLT) Occurrences of phase 6 (SBT) Occurrences of phase 8 (WB) % time phase 1 (SBLT) 0% 1% +1% 1% 1% 0% 0% 0% 0% 1% 0% 0% 0% 0% 0% % time phase 2 (NBT) 73% 73% 0% 78% 76% 2% 78% 78% 0% 75% 72% 3% 76% 75% 1% % time phase 4 (EB) 26% 26% 1% 21% 23% +2% 21% 21% 0% 23% 27% +3% 23% 24% +1% % time phase 5 (NBLT) 12% 12% 0% 6% 8% +1% 10% 9% 1% 12% 15% +2% 10% 11% +1% % time phase 6 (SBT) 61% 62% +1% 72% 69% 3% 68% 69% +1% 63% 58% 6% 66% 64% 2% % time phase 8 (WB) 26% 26% 1% 21% 23% +2% 21% 21% 0% 23% 27% +3% 23% 24% +1% NB volume (all vehicles) SB volume (all vehicles) EB volume WB volume Total volume NB Truck Percentage 17% 12% 5% 24% 17% 7% 19% 17% 2% 10% 8% 2% 16% 12% 3% SB Truck Percentage 19% 14% 5% 23% 22% 1% 22% 19% 3% 9% 14% +5% 17% 17% 1% EB stop line occupancy 3% 1% 1% 2% 2% +1% 2% 2% 0% 9% 7% 2% 4% 3% 1% WB stop line occupancy 34% 41% +7% 17% 20% +3% 22% 27% +5% 33% 46% +14% 27% 34% +7% NB trucks detected NB trucks detected with potential benefit % NB trucks with benefit from priority 17% 27% +10% 9% 17% +8% 6% 14% +9% 32% 26% 6% 22% 22% 0% NB truck detected but phase not green 38% 33% 5% 35% 25% 11% 26% 35% +8% 32% 33% +2% 33% 32% 2% NB truck detected during minimum green 28% 24% 4% 29% 32% +3% 24% 13% 10% 24% 30% +5% 26% 25% 1% NB truck detected during green, no conflicting call 19% 16% 3% 28% 26% 2% 37% 35% 2% 11% 11% +1% 18% 21% +3% SB trucks detected SB trucks detected with potential benefit % SB trucks with benefit from priority 2% 11% +9% 5% 6% +1% 7% 11% +3% 15% 15% 0% 9% 11% +2% SB truck detected but phase not green 52% 64% +12% 42% 40% 2% 42% 42% 0% 39% 57% +18% 42% 52% +10% SB truck detected during minimum green 44% 27% 16% 18% 28% +9% 30% 31% +1% 30% 28% 3% 29% 29% 0% SB truck detected during green, no conflicting call 3% 0% 3% 35% 26% 9% 21% 16% 5% 15% 0% 15% 21% 9% 12%

17 Also consistent with what was found in the earlier data collection, the implementation of priority for trucks also slightly increases the percentage of trucks which may benefit from the priority. While the reason is not obvious, it is believed that increasing the green slightly for the phase serving the truck movements lengthens the green time available for additional trucks to be detected. The percentage of trucks which could benefit from truck priority is significantly different for the northbound and southbound directions. On average, 22 percent of northbound trucks would be expected to benefit during the four time periods, while only about 10 percent of southbound trucks would be expected to benefit. The reason for this can be found in the signal timing data (in section B) and the reasons priority is not provided (in section D). Northbound left-turning traffic is significant, as evidenced by the approximately 10 percent phase time for phase 5. Southbound left-turning traffic is very light, with average phase time for phase 1 being less than one percent. (On both Wednesday and Thursday, there was demand for phase 1 only six times in the four hours data was collected.) Because the northbound left turn opposes the southbound through traffic, the northbound left turns increase the red time experienced by southbound through traffic (relative to northbound through traffic). Consequently, southbound trucks are much more likely to approach the intersection when their phase (6) is not green than are northbound trucks; and priority can only be provided when the phase is green. There was one data sampling period (the 16:00-17:00 period after truck priority was implemented) which skewed the results to some extent in favor of the non-priority operation. In particular, the southbound MOEs showed a worse operation for the priority cases than for the non-priority case. However, it appears the major reason for these results is a 63 percent increase (from 33 to 54) in southbound truck traffic during that one-hour period on Thursday relative to the same period on Wednesday. Compounding that problem was a decrease of five percent in the amount of phase time provided for the southbound through movement (phase 6) from 63 percent to 58 percent due to increased times for the northbound left turn (phase 5) and the minor street (phase 4 and 8) traffic. Based on field observations at the time, it appeared that phase 6 was extending to maximum green time on several occasions, building up the northbound left turn and minor street queue lengths, which in turn increased the phase times for those movements. Because phase 6 was already reaching its maximum, those other movements which phases were not reaching maximum received a higher percentage of the overall time. 8.0 Potential Benefit In addition to attempting to identify how much benefit may be achieved by implementing a truck priority system, the other question is How many trucks may benefit from such a system? In an attempt to answer that questions, over 700 hours of detector data were collected and uploaded from the ASC3 controller at the intersections. Table 3 shows the day-by-day results of that data collection, with aggregated results shown by day of week and weekly. Page 14

18 Table 3 Evaluation of Quantities of Trucks Receiving Benefits from Priority Page 15

19 The data is not continuous because the controller is capable of storing only approximately five and a half days of detector data before it begins overwriting the data already stored. Nevertheless, information to answer the question as to how many vehicles may benefit can be estimated. The bottom portion of table 3 has taken all of the data collected from the 34 days shown and aggregated it on a day of week basis, with the weekly results shown on the bottom line. From that table, it can be seen that 13 percent of all traffic on TH 24 is truck traffic, not surprisingly higher on weekdays than and weekends. Of the estimated 12,150 trucks on TH 24 during the week, it is believed that 16 percent (1947) would benefit through the implementation of truck priority at this intersection. 9.0 Annual Benefits In an attempt to quantify the benefits which would be obtained by implementing truck priority at this intersection, the data collected in November and summarized in Table 2 was used and applied to truck rental rates for truck operations. According to the Minnesota Department of Transportation (April 4, 2011) the tractor-trailer rental rate, including the prevailing wage plus fringe benefits for tractor-trailer drivers, is $94.43 per hour. Based on the average travel time reduction of 3.7 seconds to northbound trucks and the average increase of 1.7 seconds to southbound trucks, the average weekday volume of trucks (1034 northbound, 938 southbound, from Table 3), and 255 weekdays per year, it is estimated that the truck priority system will result in an annual reduction of 158 truck-hours, corresponding to an annual operational cost savings of $14,924. The same calculations were performed on the September data, before MnDOT changed the signal timing parameters. From that data, the average travel time reductions were 5.3 seconds and 1.9 seconds for northbound and southbound trucks, respectively. In that situation truck priority would have provided an annual reduction of 514 hours and $48,577. With respect to non-trucks, the November data collection showed a 1.1 second/vehicle reduction for 6054 northbound vehicles each weekday and a 0.5 second/vehicle increase for 4630 southbound vehicles each weekday. Expanded to a full year, this corresponds to an overall reduction of 308 vehicle-hours for non-trucks Conclusions and Recommendations Based on the results of this project, including personal observations and data collected before and after implementation, the overall conclusion is that providing special detection to implement truck priority in the form of extending the green interval for that movement does provide benefit, both to the heavy commercial vehicles themselves and to other vehicles which may be following them. However, the benefits derived are not clear-cut. In some cases travel times (delays) actually increased slightly for both types of vehicles when truck priority was implemented. Rather than delays, it appears that greater benefits derived through implementation of truck priority are a reduction in stops (both to trucks and to non-trucks) and a reduction in the probability that at the start of the green indication, the front vehicle in the queue will be a truck. Although relatively small improvements in delays were attained through truck priority, and although the number of trucks triggering the priority request at this location (approximately 16 percent) was also relatively small, as demonstrated in the Annual Benefits section, when multiplied by the total number of trucks over the course of a year, the time and monetary savings provided by truck priority become significant. Therefore, it is recommended that Page 16

20 priority operation be expanded to other intersections with high truck activity, with total truck volumes being the primary criteria used in determining precedence of those intersections. With respect to this particular intersection, the conclusions in November above were qualitatively the same as those drawn from the original data collection effort in September. At that time, it was believed that a possible cause of these mixed results was the signal timing omission mentioned earlier with respect to the conversion from the ASC2 controller to the ASC3 controller. The controller programming was corrected prior to the November data collection, and the November data yielded trends similar to those in September. Based on the magnitude of the time and monetary savings between the September and November data, the change in the controller settings were found to have a significant impact on improving the operation, and the implementation of truck priority provides an additional benefit. At this intersection, one concern that remains in the controller programming is that the mainline phases (2 and 6) have been programmed with soft vehicle recalls rather than hard vehicle recalls. Soft vehicle recalls will place a call to that phase only if there is no activity on any conflicting phases. Consequently, phase 6 (for example) would only be called if a southbound vehicle passed over the detector 550 feet upstream of the stop line or if there were no calls on phases 4, 5 and 8. As a result, a slow-moving southbound vehicle approaching which arrives at the intersection just as the signal turns red may not receive a green indication until after the controller has cycled back and forth several times between phases 4/8 and phase 5. This was actually observed several times in the field under low-flow conditions. This potential problem was exacerbated by the added initial timing features not being implemented for the September data collection. While the corrected controller programming has reduced this problem, it was still observed a handful of times in the field during the November data collection. Even if added initial detection is implemented as suggested, it is recommended that hard vehicle recalls be used for phases 2 and 6. Considering the high percentage of trucks at this location and their low acceleration rates, there is a relatively high probability of trapping a truck between the extension loop and the stop line on several occasions during the day. Based on field observations and the latest results from the data collection, it is believed that further improvement in overall operations at this intersection can be achieved by increasing the maximum green time for phases 2 and 6 (especially phase 6). As described earlier, it appears that the maximum green time on several occasions during the afternoon peak hour is limiting the controller s ability to fully accommodate the southbound demand. Finally, it was difficult to find locations within the state which met the two primary criteria for consideration high percentages of trucks, and single-lane approaches. Finding locations with single-lane approaches was essential for this project to verify that non-trucks are not adversely affected by truck priority. That was confirmed by the overall reduction in vehiclehours to non-trucks. Going forward, however, single-lane approaches need not be a prerequisite for implementing truck priority. With multi-lane approaches, although drivers of non-trucks can minimize their delays by maneuvering around trucks slowly accelerating from a stop at the traffic signal, this project has demonstrated that the savings to trucks alone from priority treatment can justify the installation. Page 17