Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Chapter 3: Test Applications In Chapter 2, we presented the three primary areas of new research and development that were conducted during this project: ⢠Development of quantitative measures for the severity of oversaturated conditions (queue length, TOSI, SOSI) ⢠Development of a process for generating and analyzing timing plans for mitigating oversaturated conditions ⢠Development of a tool for online application of mitigation strategies using TOSI and SOSI measurements to select alternative timing plans In this chapter we describe several test cases conducted as part of the project to validate the research theories listed above. Two of the test networks (Reston Parkway in Herndon, VA and the Post Oak area in Houston, TX) were used in the development and testing of the multi-objective strategy development and evaluation methodology. Two other networks (TH55 in Minneapolis, MN and the Pasadena, CA downtown network) were used in development and testing of strategies directly related to TOSI and SOSI, including the development of an analytical procedure for directly adjusting green times. Two other test cases (an arterial in Surprise, AZ and a small network in Windsor, ON) were used to test the direct application of the guidance methodology developed in Task 6. All of the test applications were simulated using the Vissim simulation system with either the RBC controller or the Virtual D4 controller. While route proportions and demand flows were changed over time, no dynamic traffic assignment was used, i.e. vehicles in the simulation did not react to the congestion conditions by changing their route, destination, or forgo travel. The characteristics of these test cases are summarized in Table 11. Operation of traffic signal systems in oversaturated conditions Page 117
Table 11. Summary of test case attributes Test Case Config Number of Ints Spacing (ft) speed (mph) Typical phasing Test duration Causation Types of symptoms Recurrence Critical locations Types of mitigation Components tested Reston Parkway; Northern VA Arterial with freeway interchange 14 500 to 3300 40 4, 6, 8 3 hours Demand All Recurrent 2 Cycle, splits, offsets, phase reservice, gating Timing plan development framework Post Oak area of Houston, TX Network 16 400 to 1800 30-40 2, 4, 6, 8 3 hours Demand All Recurrent 8 Cycle, splits, offsets, phase reservice, gating Timing plan development framework TH55; Minneapolis , MN Arterial 5 500 to 2600 55 4 1 hour Preemption Spillback, overflow queuing Non- recurrent 2 Green extension, green truncation Calculation of TOSI, SOSI, and queue length Downtown grid; Pasadena, CA Grid 22 400 to 1000 25-35 2, 4, 6, 8 2 hours Light-rail; demand Spillback, overflow queuing Recurrent 5 Splits, offsets Calculation of TOSI, SOSI, and queue length Border Tunnel Entrance; Windsor, ON Small network 9 400 to 800 25-35 2, 4, 6 45 min, 1 hour, 2 hours Incident All Non- recurrent 1 Many Online feedback tool; TOSI/SOSI Surprise, AZ Arterial 6 2600 45 8, 6 1.5 hours, 3 hours Planned event All Both 1 Many Application of guidance Operation of traffic signal systems in oversaturated conditions Page 118
Arterial Test Case: Application of the Multi-Objective Timing Plan Development and Evaluation Framework The Reston Parkway arterial network is located in Reston, VA between Herndon and Vienna near the entrance to the Dulles International Airport. The network is significantly oversaturated during peak periods. The network consists of 14 intersections with a total length of 16,572 ft (3.1 miles). The spacing between intersections ranges from 524ft to 3,309 ft. The speed limit for the main arterial is 45 mph. Side street speed limits range from 15 mph to 45 mph. The arterial operates in a coordinated fashion during the day using traditional forward progression offsets. System detector count data was provided from The Northern Region Operation (NRO) of Virginia Department of Transportation (VDOT) which operates the traffic signal system in Reston Parkway. The current system employs 170 controllers and the Management Information System for Transportation (MIST) central software. Actual detector data from this network was taken from August 11th, 2009 to September 10th, 2009. These detectors cover almost the whole network. MIST compiles system detector data in 15-minute intervals. During the evening peak period (2:30 P.M. to 8:00 P.M.) the arterial network operates with three different timing plans. The system operates a 130s cycle until 3:00 P.M., shifts to a longer cycle of 180s, and then transitions to a 140s cycle at 6:00 P.M. A sub-network of five intersections was selected for application of the timing plan development and evaluation framework presented in Chapter 2. This portion of the arterial has significant recurrent oversaturation due to the interchange with the Dulles Toll Road ramps as shown in Figure 58. Operation of traffic signal systems in oversaturated conditions Page 119
Figure 58. Reston Parkway network Significant changes to the traffic patterns for both demand and directional distribution occur in the selected sub-network during the peak period. Turning volumes become very heavy and the dominating direction changes from south to north. The relatively short link length at the interchange with the toll road contributes to the oversaturated situation. This network was coded in the Vissim microscopic traffic simulation model. The network was calibrated to the P.M. peak period traffic using field data collected from Monday June 1st to Wednesday June 3rd, 2009. The data collected included link travel times, travel speeds, queue lengths, and queue discharge headway. Traffic Patterns on Reston Parkway During the P.M. peak period, traffic patterns change rapidly in the Reston Parkway arterial network. Figure 59 through Figure 61 illustrate the changes in the arterial northbound and southbound traffic counts at various intersections during different points of the peak period. These changes in volume patterns emphasize the need to identify the critical routes in the system. Operation of traffic signal systems in oversaturated conditions Page 120
Figure 59. Changes in traffic patterns in Reston Parkway at 3:30 P.M. Figure 60. Changes in traffic patterns in Reston Parkway at 5:00 P.M. Operation of traffic signal systems in oversaturated conditions Page 121
Figure 61. Changes in traffic patterns in Reston Parkway at 7:30 P.M. Critical Route Scenarios Six critical route scenarios were established based on the volume analysis of the actual observed counts. Throughout the day, each of the six scenarios may occur, thus requiring different signal timing plans. These scenarios are illustrated in Figure 62. The analysis of the volumes included (but is not presented in this report): correlation analysis, cluster analysis, and pattern recognition techniques. For each critical route scenario, background traffic was determined and the volume on each critical route during the peak period was estimated. This flow estimate for each critical route accounted for the maximum observed volume from the month of system detector data that was analyzed. A profile of demand volumes on each critical route was developed. The arrival demand on each critical route as well as the background traffic was modified in 15-minute increments. Operation of traffic signal systems in oversaturated conditions Page 122
Scenario 2 Scenario 3 Scenario 4 Scenario 5 Scenario 6 Scenario 1 Figure 62. Critical route scenarios on the Reston Parkway network In addition, the volume profiles were adjusted to account for the delayed demand that might not have been able to enter the network due to oversaturated conditions, and therefore would not have been reflected in the system detector data. Figure 63 illustrates the adjustment of the volume profile for an example movement. The adjustment of the profile increased the peak volume significantly, which is critical to be taken into account in the design of the mitigation timing plans. These adjustments were applied to each of the critical routes for each scenario shown in Figure 62. Operation of traffic signal systems in oversaturated conditions Page 123
Figure 63. Adjusted demand profile for Route AH to account for demand unrepresented in the system detector counts Illustration Using of Critical Routes to Determine Mitigation Strategies In this section we describe the use of critical routes to develop mitigation strategies. This discussion will focus on Scenario 5 since it represents one of the most challenging situations with severe oversaturation and symptoms. Figure 64 identifies the attributes of this scenario. Critical routes, oversaturated symptoms, and proposed control strategies are illustrated in Figure 64. Operation of traffic signal systems in oversaturated conditions Page 124
Figure 64. Problematic Scenario 5: critical movements and diagnosis In this scenario the oversaturated approach northbound on the Parkway at Sunset Hill Road for left-turn vehicles causes secondary congestion at the two upstream intersections at the eastbound and westbound toll road ramps. In addition, the primary peak direction of through traffic is southbound on the parkway. This combination of flows provides a rather challenging situation for the design of mitigation timing plans. Operation of traffic signal systems in oversaturated conditions Page 125
Extent Duration Causation Recurrence Symptoms Movement Situational Signal Timing Recurrent Starvation Approach Intermittent Geometrics Non-Recurrent Spillback Intersection Persistent Other modes Storage Blocking Route Prolonged Demand Cross Blocking One-way arterial Planned Events Two-way arterial Unplanned Events Interchange Grid Network Figure 65. Key attributes of Scenario 5 Based on observation of the critical routes in this scenario, we first determine that one of two different canonical operational strategies will be applied in this test case, either a metering strategy for the northbound critical routes or a phase reservice strategy for the northbound left turn at Sunset Hill Road. Both of these methods have strengths and weaknesses. A comparison of the two methods is illustrated in Figure 66 (a) and (b) below. (a) Northbound left-turn phase reservice at Sunset Hill Road intersection Figure 66. Comparison of phase reservice and metering strategies Operation of traffic signal systems in oversaturated conditions Page 126
(b) Metering of northbound traffic at the Dulles eastbound ramp intersection Figure 66. (continued) The main strength of the phase reservice strategy is that the additional green time for the left turn alleviates the spillback on the northbound approach at Sunset Hill Road which is the primary source of oversaturation on the arterial. The main drawback is that the green time for the southbound route will be significantly restricted and will create southbound queues north of the Sunset Hill Road intersection. The main strength of the gating strategy is that by storing vehicles on the approach link, the queues on the left turn at Sunset Hill do not grow as quickly. The main drawback of this approach is that the queue on this link can block the right turn to enter the eastbound toll road. Figure 67 illustrates the storage on the link that was selected for the metering strategy. Operation of traffic signal systems in oversaturated conditions Page 127
Figure 67. Storage capacity of the link used for metering in the Reston Parkway network The timing plan generation framework presented in Chapter 2 was used to generate timing plan parameters for the proposed control strategies. This process is described further in the next section. Cycle Length Calculations The first step of the optimization and evaluation process is to calculate feasible cycle lengths for each intersection in the network. The cycle length calculation procedure presented in Chapter 2 was used to compute the maximum cycle length that does not result in spillback at each intersection. The resulting maximum cycle times are shown in Table 12. Note that the maximum cycle times for each intersection vary widely so a judgment procedure is necessary to determine cycle times that can be implemented on all of the intersections in the system. Note also that the equation from Roess (1994) almost always estimates a longer maximum cycle than the Lieberman equation. In order to test the performance of several different cycle times, we choose several canonical cycles in the range that meet most of the requirements. If the minimum cycle time of these maximums was used, the evaluated cycles would be less than 85s. This was considered to be unrealistic to implement because of the number of phases at the larger intersections. Operation of traffic signal systems in oversaturated conditions Page 128
Table 12. Maximum cycle length before spillback occurs on critical network links Network Links Length Storage Max C Max C From To Ft Veh/lane (Lieberman) (Roess) 1 Cameron Spectrum 830 35 195 587 2 Spectrum Bowman 710 30 177 174 3 Bowman Dominion 670 28 103 150 4 Dominion Bluemont 800 33 215 352 5 Bluemont Sunset Hill 640 27 87 100 6 Sunset Hill Dulles WB 710 30 115 188 7 Dulles WB Dulles EB 600 25 85 129 8 Dulles EB Sunrise Valley 1800 75 232 236 Table 13 illustrates some of the critical geometric data for this test scenario. These distances are critical in the computation of the offsets on these links. Table 13. Critical network links and left-turn bay storage lengths Arterial Network Links Length (ft.) SB Left-turn bay (ft.) NB Left-turn bay (ft.) South Lake Rd.â Sunrise valley Rd. 435 Sunrise valley Rd. â Dulles EB ramps 1800 720 - Dulles EB ramps - Dulles WB ramps 710 - 645 Dulles WB ramps â Sunset Hill Rd. 710 - 480 Sunset Hill Rd. â Bluemont Way 640 300 380 Bluemont way â New Dominion 800 390 Table 14 lists the parameters of the shockwave model that were used to calculate the minimum and maximum offset values. From this feasible range, three sets of offset values were chosen for evaluation (i.e., progression offsets, starvation-avoidance offsets, and spillback-avoidance offsets) in each of the six different critical route scenarios. Operation of traffic signal systems in oversaturated conditions Page 129
Table 14. Shockwave Modeling Parameters Parameters Value Discharge rate (veh/hr/ln) 1,500 Desired speed (ft./sec) 36 Queue discharge wave speed (ft. /sec) 18 Acceleration rate (ft./sec2) 4 Deceleration rate (f/sec2) 4 Start-up loss time (sec) 3 Average vehicle length (ft) 25 Peak Hour Factor 0.9 Lane Utilization Factor (Fu): 1 These combinations of offsets are referred to with the following acronyms: o Sim: Simultaneous (zero) offsets o Max: Offsets to prevent starvation at downstream intersection o Min: Offsets to prevent spillback at upstream intersection o Med: Offsets that are the medium value between Max and Min Design of Splits and Offsets Figure 68 presents the split-offset calculation procedure developed in Chapter 2 for the mitigating oversaturation. The initial splits at each intersection are calculated based on the v/c ratio for each phase. These splits are then modified in an iterative procedure by estimating the expected queue lengths for each approach for v/c either greater than or less than 1.0. To mitigate the oversaturation on the critical route(s), the degree of saturation values for each approach is constrained by a pre-specified threshold. The resulting queue-to-link ratios are then used to determine the bounds on the values of the offsets that will avoid spillback and starvation. Figure 69 through Figure 71 illustrate the offset values obtained using this procedure for the southbound route and the two routes together. Operation of traffic signal systems in oversaturated conditions Page 130
Figure 68. Split-offset calculation procedure Figure 69. Oversaturation offsets for the southbound critical route Operation of traffic signal systems in oversaturated conditions Page 131
Figure 70. Oversaturation offsets for both southbound and northbound critical routes Figure 71. Scenario 5 offset design values (min and max), for northbound and southbound progression By obtaining the maximum and minimum values of the offsets using this procedure, a feasible region is identified that satisfies both objectives of spillback and starvation avoidance. The offsets obtained from this procedure can vary from negative to positive values, depending on the overflow queue length constraints, link length, and green split time. Note that in this scenario, it is not possible to find offsets that ensure that spillback and starvation are avoided in both directions. In this scenario, the most critical route, the northbound route, is selected. Operation of traffic signal systems in oversaturated conditions Page 132
Simulation Experiment Timings plans were designed for six volume and critical routing scenarios (denoted scenarios 1, 2, 3, and so on). Three cycle length values (100s, 140s, 180s) were selected for evaluation and four types of offsets (simultaneous, minimum, medium, and maximum) were evaluated, where minimum and maximum refer to the spillback-avoidance and starvation avoidance respectively. Each combination of cycle time and offsets was then given a unique number (i.e. Strategy 1, 2, and so on) are shown in Table 15. These 12 combinations of timing plans were also compared to the current baseline timings used by VDOT in the real world. Furthermore, the combination of cycle time, offsets, and splits was then combined with either metering in the northbound direction or phase reservice at the critical intersection. Table 16 illustrates the components of the control strategies for Scenario 5. For the metering strategy we chose to reduce the green time for the metered approach by 20% from its previously calculated value. Five runs, with different seed numbers, were conducted in Vissim to compute the average and variance of the performance measures for each strategy. The scenario lasts for five hours. Thirty minutes of simulation prior to the peak period, and approximately one hour of simulation after the peak period to allow enough time for the queues to dissipate and return the system to steady-state operation. Pareto fronts are then calculated to determine which timing plans are non-dominated on the three optimization objectives. Table 15. Combination of cycle time and offset values for each strategy Operation of traffic signal systems in oversaturated conditions Page 133
Table 16. Control strategy combinations with metering or phase reservice Simulation Results and Evaluation The simulation output results were collected every 15 minutes of the simulation time including total system and link-by-link delay, total number of stops, and total system throughput measured by the total number of vehicles leaving the network during the time period. Figure 72 illustrates the four-dimensional Pareto front obtained for Scenario 5. These diagrams directly demonstrate the strategy that is optimal for a specific 15 minute time period during the simulation. The Pareto front diagrams present only the optimal non-dominated solutions among all of the tested control strategies for a particular volume scenario. In both diagrams, the x-axis (horizontal) represents the total system delay time, the y-axis (extending into the depth of field) represents the total system stops, and the z-axis (vertical) represents the total system throughput. In the top diagram, the color of each dot is the simulation time. Dark colors represent early times Scenario Strategy Control Strategy Applied Location Action Cycle Length (sec) Offset Design Split 5 1 Ph as e R e- se rv ic e N B L ef t-t ur n Su ns et H ill R d. In te rs ec tio n D ou bl e th e le ft- tu rn p ha se (le ad & L ag ) 100 Simultaneous ba se d on V /C ra tio w ith p rio rit y to Sc en ar io 5 c rit ic al ro ut es 2 Spillback Avert 3 Medium 4 Starvation Avert 5 140 Simultaneous 6 Spillback Avert 7 Medium 8 Starvation Avert 9 180 Simultaneous 10 Spillback Avert 11 Medium 12 Starvation Avert 13 Base timing plan (VDOT) 140, 180, 130 Progression offset v/c ratio 1 M et er in g N B T hr ou gh @ E B r am p In te rs ec tio n R ed uc e th e N B th ro ug h gr ee n tim e by 20 % 100 Simultaneous ba se d on V /C ra tio w ith p rio rit y to Sc en ar io 5 c rit ic al ro ut es 5 2 Spillback Avert 3 Medium 4 Starvation Avert 5 140 Simultaneous 6 Spillback Avert 7 Medium 8 Starvation Avert 9 180 Simultaneous 10 Spillback Avert 11 Medium 12 Starvation Avert 13 Base timing plan (VDOT) 140, 180, 130 Progression offset v/c ratio Operation of traffic signal systems in oversaturated conditions Page 134
in the simulation (the loading period) and light colors represent the end of the simulation time (recovery period). In the bottom diagram, the color of each dot represents the cycle time of the strategy that is dominant at that time in the simulation. Blue colors represent low cycle times (in this example, 100s) and red colors represent high cycle times (in this example, 180s). For this specific test case, Figure 72 (top) can be interpreted as follows: during the processing regime timing plans that are designed to maximize throughput (orange color scale refers roughly to the processing regime) are performing best. However, timing plans that are optimized for minimizing total system delay and total stops perform the best on all three performance scales during the loading and recovery regimes (white and yellow color scale represent the recovery regime, dark red and brown represent the loading regime). In Figure 72 (bottom), illustrates timing plans with longer cycles (dark red color scale) that were designed to maximize the throughput are dominant during the processing regime. Timing plans with short cycles (blue color scale) that were designed to minimize delay are dominant in both the loading and recovery regimes of the scenario. Control strategies with the medium cycle length (140s) with different offset values (light blue, green, and yellow) are not performing well in this particular scenario since they appear infrequently as a dominant timing plan during any 15-minute time period. It is clear from this analysis that applying just one timing plan for an entire scenario, when it lasts for an appreciable amount of time (e.g. over an hour), is not recommended. Different approaches to mitigation are needed during the three regimes of operation (loading, processing, and recovery). Operation of traffic signal systems in oversaturated conditions Page 135
Figure 72. Illustration example of the Pareto front for Scenario 5: metering strategy Operation of traffic signal systems in oversaturated conditions Page 136
Pareto Front Analysis In this section we present the results of the Pareto front analysis for the 25 strategies (12 combinations of cycle time and offsets with either phase reservice or metering plus the baseline timing plan) on the six critical route scenarios. In general, it seems strategies that include metering, longer cycles perform better during the recovery regime. On the other hand, when using phase reservice at the critical intersection, timing plans with shorter cycles dominate both the loading and recovery regimes. If you consider only the throughput objective, the graphics tend to illustrate that plans with shorter cycles perform poorly during the loading regime and then system throughput starts to improve as demand increases in the network. Toward the end of peak period the timing plans with shorter cycles return to their poor performance. These temporal diagrams convey invaluable information to the decision makers and analysts regarding system performance, but take some investment of the reader to adequately digest them. In the following sections we have provided some summary bullets for each test scenario to quickly summarize the information depicted in the diagrams. The percentage improvement statistics are relative to the baseline timing plan used in the real world (130s cycle with forward progression offsets; no metering or phase reservice considerations). Operation of traffic signal systems in oversaturated conditions Page 137
Scenario 5 Results: Pareto Front Ti m e (2 :0 0- 8: 30 )P .M . Co nt ro l S tr at eg ie s 180 sec/ spillback avert offset @ 8 P.M. Figure 73. Scenario 5 with phase reservice Routing Scenario 5 with phase reservice control findings: ï§ Minimum cycle lengths and mid-value offsets minimize delay ï§ Medium cycle lengths and mid-value offsets maximize throughput ï§ The phase reservice strategy has lower total throughput than the metering strategy ï§ Throughput maximization strategies reduce delay by 29% and increase throughput by 1% ï§ Delay minimization strategies reduce delay by 29% and reduce throughput by 1% Operation of traffic signal systems in oversaturated conditions Page 138
Short cycles @ (2:30- 30:00) P.M. Ti m e (2 :0 0- 8: 30 )P .M . Co nt ro l S tr at eg ie s Figure 74. Scenario 5 with metering Routing Scenario 5 with metering control findings: ï§ Short cycle with min/mid offset minimizes delay and stops ï§ Medium cycle length maximizes throughput ï§ Longer cycle length increases delay significantly ï§ Throughput maximization strategies reduce delay by 13% and increase throughput by 11% ï§ Delay minimization strategies reduce delay by 35% and increase throughput by 7% Operation of traffic signal systems in oversaturated conditions Page 139
Table 17. Scenario 5: total improvement over the baseline strategy The following figures illustrate the dominant timing plan during each 15-minute period of the scenario for each of the control objectives based on the average performance on each objective over the five simulation runs. One thing to note is how the optimal plan for some of the objectives moves around considerably. It would be unreasonable to consider switching between the timing plans as shown here because of undesirable transitioning effects. It is also important to note that each scenario using a particular strategy evolves during the peak period differently. So the performance at any 15-minute period for a particular strategy is dependent on the performance of that strategy in the previous 15 minutes. However, the presented demand profiles are the same in every case (common random number seeds), so the trend in performance of a particular timing strategy is reasonably consistent. For the timing plans that include phase reservice, optimal strategies that were designed to minimize delay appear to be stable during most the of the peak period (Strategies 3 and 4). On the other hand, strategies that are optimal for throughput maximization show instability as illustrated in Figure 75. For timing plans that included metering, delay minimization and throughput maximization optimal strategies appear to coincide with each other for most of the peak period (the green diamonds are overlaid behind the blue diamonds). Strategy Cycle (sec) Delay Stops Throughput Phase Re-service 1 10 0 0.26 0.45 -0.01 2 0.26 0.45 -0.01 3 0.29 0.42 -0.01 4* 0.29 0.6 -0.01 5 14 0 -0.17 0.07 -0.01 6 -0.13 0.22 -0.01 7 -0.14 0.15 -0.01 8 -0.09 0.18 -0.02 9 18 0 -1.17 -0.25 -0.08 10 -1.42 -0.19 -0.12 11 -0.90 -0.29 -0.07 12 -1.22 -0.29 -0.41 Metering 1 10 0 0.35 0.45 0.07 2 0.35 0.45 0.07 3 0.34 0.42 0.08 4 0.33 0.42 0.06 5 14 0 0.06 0.07 0.1 6 0.17 0.22 0.1 7* 0.13 0.15 0.11 8 0.14 0.18 0.09 9 18 0 -0.18 -0.25 0.11 10 -0.15 -0.19 0.08 11 -0.24 -0.29 0.06 12 -0.24 -0.29 0.06 Operation of traffic signal systems in oversaturated conditions Page 140
Figure 75. Scenario 5: optimal control strategies for each 15-minute period Finally the mitigation strategies were evaluated for each 15-minute interval with respect to the timing plan implemented in the real world. The percentage improvements for each measure are presented in Figure 76. Table 18 presents the total average improvement percentages during the entire peak period. For each scenario in the following sections, we present a representative plot for a scenario/strategy combination. Full performance data from the study is available from NCHRP. Operation of traffic signal systems in oversaturated conditions Page 141
Figure 76. Scenario 5: Example performance profiles of a mitigation strategy versus the baseline timing plan Results for the five other critical route scenarios are listed below for further illustration of the performance comparisons between the mitigations. Operation of traffic signal systems in oversaturated conditions Page 142
Scenario 1 Results The mitigation strategies in this scenario were optimized for just two critical routes: northbound and southbound on Reston Parkway. Metering and phase reservice were not applied at Sunset Hill because those turning routes were not considered critical in this scenario. Therefore only 12 timing strategies are considered instead of 24 as in Scenario 5. The Pareto front results are presented in Figure 77. Figure 77. Scenario 1: non-dominated strategies Routing Scenario 1 findings: ï§ Medium cycles with min/mid offsets maximize throughput ï§ Medium cycles with min/mid offsets dominate most other strategies ï§ During all three regimes, throughput maximization is the optimal control objective ï§ Throughput maximization strategies reduce delay by 16% and increase throughput by 13% ï§ Delay minimization strategies reduce delay by 41% , but reduced throughput by 4% Operation of traffic signal systems in oversaturated conditions Page 143
Table 18. Scenario 1: total improvement over the baseline strategy Strategy cycle Delay Stops Throughput 1 10 0 0.39 0.26 -0.04 2 0.22 0.2 -0.04 3 0.41 0.2 -0.04 4 0.39 0.25 -0.04 5 14 0 0.16 0.2 0.13 6 0.16 0.19 0.12 7 0.22 0.25 0.11 8 0.21 0.23 0.11 9 18 0 0.06 0.16 0.08 10 0.06 0.16 0.08 11 0.03 0.13 0.08 12 -0.07 0 0.12 Figure 78. Scenario 1: optimal control strategies by time Figure 79. Scenario 1: example Strategy 5 improvement percentage over the baseline Operation of traffic signal systems in oversaturated conditions Page 144
Scenario 2 Results The mitigation strategies in this scenario were optimized for just one critical route: southbound on Reston Parkway. Metering and phase reservice were not applied at Sunset Hill because those turning routes were not considered critical. Therefore only 12 timing strategies are considered instead of 24 as in Scenario 5. The Pareto front results are presented in Figure 80. Figure 80. Scenario 2: Pareto fronts Routing Scenario 2 findings: ï§ Short and medium cycles with min/mid offset maximize throughput ï§ During the processing regime, throughput maximization is the optimal control objective ï§ During the loading and recovery regimes, delay minimization is the optimal control objective ï§ Throughput maximization strategies reduce delay by 13% and increase throughput by 11% ï§ Delay minimization strategies reduce delay by 35% and increase throughput by 7% Operation of traffic signal systems in oversaturated conditions Page 145
Table 19. Scenario 2: total % improvement over the baseline plan Strategy Cycle Delay Stops Throughput 1 10 0 0.35 0.45 0.07 2 0.35 0.45 0.07 3 0.34 0.42 0.08 4 0.33 0.42 0.06 5 14 0 0.06 0.07 0.1 6 0.17 0.22 0.1 7 0.13 0.15 0.11 8 0.14 0.18 0.09 9 18 0 -0.18 -0.25 0.11 10 -0.15 -0.19 0.08 11 -0.24 -0.29 0.06 12 -0.24 -0.29 0.06 Figure 81. Scenario 2: optimal control strategies by time Figure 82. Scenario 2: strategy 7 improvement % by time Operation of traffic signal systems in oversaturated conditions Page 146
Scenario 3 Results The mitigation strategies in this scenario were optimized for just one critical route: northbound on Reston Parkway. Metering and phase reservice were not applied at Sunset Hill because those turning routes were not considered critical in this scenario. Therefore only 12 timing strategies are considered instead of 24 as in Scenario 5. The Pareto front results are presented in Figure 83. Figure 83. Scenario 3: Pareto fronts Routing scenario 3 findings: ï§ Long cycles with min/ mid offset maximize throughput ï§ During the processing regime, throughput maximization is the optimal control objective ï§ During the loading and recovery regimes, delay minimization is the optimal control objective ï§ Throughput maximization strategies reduce delay by 21% and increase throughput by 13% ï§ Delay minimization strategies reduce delay by 49% , and increase throughput by 9% Operation of traffic signal systems in oversaturated conditions Page 147
Table 20. Scenario 3: total % improvement over baseline plan Figure 84. Scenario 3: optimal control strategies by time Figure 85. Scenario 3: Strategy 4 improvement % by time for performance measures (delay, stop, and throughput) Strategy Cycle Delay Stops Throughput 1 10 0 0.49 0.57 0.09 2 0.48 0.56 0.09 3 0.49 0.56 0.09 4 0.51 0.58 0.09 5 14 0 0.43 0.47 0.09 6 0.45 0.49 0.08 7 0.20 0.25 0.13 8 0.14 0.21 0.12 9 18 0 0.20 0.25 0.13 10 0.14 0.21 0.12 11 0.21 0.27 0.13 12 0.13 0.17 0.12 Operation of traffic signal systems in oversaturated conditions Page 148
Scenario 4 Results The mitigation strategies in this scenario were optimized for five critical routes: northbound and southbound on Reston Parkway, eastbound and westbound at Sunset Hill Road, and the left turn from Sunset Hill Road on to the toll road. Metering and phase reservice were not applied at Sunset Hill because those turning routes were not considered critical. Therefore only 12 timing strategies are considered instead of 24 as in Scenario 5. The Pareto front results are presented in Figure 86. Figure 86. Scenario 4: Pareto fronts Operation of traffic signal systems in oversaturated conditions Page 149
Routing Scenario 4 findings: ï§ Long cycles with min/ mid offset maximize throughput ï§ Short cycles minimize delay during both loading and recovery regimes ï§ During the processing regime, throughput maximization is the optimal control objective ï§ During the loading and recovery regimes, delay minimization is the optimal control objective ï§ Throughput maximization strategies reduce delay by12% and increase throughput by 11% ï§ Delay minimization strategies reduce delay by 45% , and increase throughput by 8% Table 21. Scenario 4: total improvement % over baseline plan Figure 87. Scenario 4: optimal control strategies by time for performance measures (delay, stop, and throughput) Strategy Cycle Delay Stops Throughput 1 10 0 0.08 0.24 0.07 2 0.04 0.20 0.07 3 0.07 0.22 0.07 4 0.05 0.20 0.07 5 14 0 0.02 0.06 0.10 6 0.45 0.49 0.08 7 0.07 0.13 0.10 8 0.12 0.19 0.11 9 18 0 -0.46 -0.53 0.09 10 -0.40 -0.47 0.09 11 -0.46 -0.53 0.09 12 -0.11 -0.12 0.09 Operation of traffic signal systems in oversaturated conditions Page 150
Figure 88. Scenario 4: Strategy 8 improvement % by time for performance measures (delay, stop, and throughput) Operation of traffic signal systems in oversaturated conditions Page 151
Scenario 6 Results The mitigation strategies in this scenario were optimized for five critical routes: northbound and southbound on Reston Parkway, left and right turns from the westbound toll road off-ramp and the left turn from the eastbound toll road off-ramp on to northbound Reston Parkway. Metering and phase reservice were not applied at Sunset Hill Road because those turning routes were not considered critical. Therefore only 12 timing strategies are considered instead of 24 as in Scenario 5. The Pareto front results are presented in Figure 89. Figure 89. Scenario 6: network performance measures of the optimal control strategies during peak period Operation of traffic signal systems in oversaturated conditions Page 152
Routing Scenario 6 findings: ï§ Short cycles dominate the optimal solutions of this routing scenario ï§ During the process phase, throughput maximization is the optimal control objective ï§ During recovery phases, optimal solutions consist of both control objectives (delay-min and throughput-max) ï§ Throughput maximization strategies reduce delay by 29% and increase throughput by 19% ï§ Delay minimization strategies reduce delay by 38% and increase throughput by 17% Table 22. Scenario 6: total improvement % over baseline plan Figure 90. Scenario 6: optimal control strategies by time Strategy Cycle Delay Stops Throughput 1 10 0 0.20 0.30 0.08 2 0.19 0.27 0.09 3 0.15 0.25 0.11 4 0.18 0.28 0.07 5 14 0 0.29 0.33 0.19 6 0.33 0.38 0.17 7 0.38 0.42 0.17 8 0.29 0.32 0.17 9 18 0 0.10 0.07 0.18 10 0.15 0.12 0.19 11 0.22 0.22 0.18 12 0.15 0.12 0.19 Operation of traffic signal systems in oversaturated conditions Page 153
Figure 91. Scenario 6: Strategy 7 improvement % by time Lessons Learned and Guidance from the Reston Parkway Case Study The Reston Parkway case study was used to illustrate the process of developing timing plans for mitigating oversaturated conditions using the methodology developed in Chapter 2 and the evaluation of the performance of those plans on multiple performance measures. The following is a list of major findings from the Reston Parkway Study: ï§ Critical route determination is essential for determination of optimal strategies. A variety of results were obtained for the six different critical route scenarios that were optimized. In some of the scenarios, all of the mitigation timing plans outperformed the baseline plan which was not designed to perform for that set of critical flows. ï§ Cycle length should be determined to accommodate maximum queues. However, the findings for several of the scenarios indicate that longer cycle times result in worse performance in total delay but improve throughput. Side streets will be degraded to improve over-all system performance. ï§ Offsets should be determined within a range to prevent spillback and starvation. The performance results indicate that the procedure offered in Chapter 2 for determining offsets are effective in improving over-all system performance measures over a procedure which assumes forward progression and obtaining maximum bandwidth is an optimal method. ï§ Queues are to be considered as constraints in the optimization problem; throughput and delays are objectives. As part of the optimization approach, the queue-link ratio must be fixed to a specific value to solve for an offset/split combination that minimizes both spillback and starvation. This procedure is approximate since an average value of the queue ratio is used to solve one specific value of the offset. If the queue dissipates or grows due to fluctuation in demand, conceptually a new offset is needed for that situation. However, in solving for a fixed set of timing parameters that can be implemented with existing traffic controllers, this process is the most reasonable that could be expected. Operation of traffic signal systems in oversaturated conditions Page 154
ï§ During the network loading and processing regimes, optimal strategies maximized throughput. During the recovery regime optimal strategies minimized total delay. For most of the scenarios of this test case, this trend held true. In other test cases during this project, minimize delay strategies also had merit during the loading regime and maximize throughput strategies dominated during the loading period. This test case also did not consider scheduling of plans during the three regimes, so the results assume the same timing plan is running during the entire peak period. A switching strategy is needed to enable the best timing plan at the best time, either via online logic or by time of day schedule. ï§ High cycle lengths are associated with higher delay without substantially increasing total throughput. This matches findings from other research and other tests during this project, that shorter cycles have better total throughput performance than were previously assumed because of the higher percentage of lost-time. The reduction in the interaction of spillback queues using shorter cycles seems to outweigh the penalty of more frequent phase changes. Summary and Conclusions In this case study we applied the signal timing plan design and analysis framework described in Chapter 2 to a real-world network. This framework utilizes knowledge of the critical routes in the network as a key first step in the design of timing plans. The framework takes into account some basic underlying constraints on cycle, offsets, and splits for minimizing detrimental effects of oversaturation (spillback and starvation). The optimization approach generates cycle, split, and offset values that minimize the degree of saturation of the approaches on the critical routes. This procedure was then applied to several scenarios on the Reston Parkway model and evaluated. Results revealed that the use of the signal timing plan design framework led to significant improvement of system performance when measured by total system throughput, total system delay, and total stops. A three-dimensional temporal Pareto front diagram was developed to illustrate the non-dominated solutions throughout the peak period. The results indicate that since the problem is so dynamic, no one solution is optimal all of the time. This innovative type of diagramming can be used to assist in the analysis of the performance of control strategies in order to distill rational guidance for practitioners. In the next section, we apply the process again to a more complicated test network in the Post Oak area of Houston, TX and document the results and findings. Operation of traffic signal systems in oversaturated conditions Page 155
Network Test Case: Application of the Multi-Objective Evaluation Process with Explicit Consideration of Operational Regimes In this section, we describe a second test case in which we applied the methodology for design of timing plans that consider oversaturated conditions. The network shown in Figure 92 is a good example of a combination of arterial and grid operations and the influence of freeway on- and off-ramp traffic on arterial operations in oversaturated conditions. In the first test case, we learned that different plans have different levels of effectiveness in each of the three regimes of operation (loading, processing, and recovery). In this test case, we expanded our approach to explicitly consider the TOD schedule of the implementation of three successive timing plans during the scenario. This scenario also significantly increases the level of complexity by considering an oversaturated network with multiple, interacting critical routes. Figure 92. Post Oak area of Houston, TX Background Houston, Texas, is the fifth largest metropolitan area in the U.S. Traffic congestion continues to grow as the urban area expands. The Post Oak / Galleria Mall area (also referred to as Uptown) experiences serious oversaturated conditions during peak periods. Adjacent to the I-610 Loop Freeway on the west side of downtown Houston, the queuing due to on- and off-ramp traffic is To Downtown Houston Operation of traffic signal systems in oversaturated conditions Page 156
significant. Westheimer Road, on the north side of the sub-network shown below, is one of the most heavily traveled arterials in the region. During freeway incidents, the area falls under additional pressure as motorists divert to arterials parallel to the I-610 to continue their commute towards I-10 and northwest Houston. The study network is adjacent to the I-610 Loop and US59 interchange as shown in Figure 93. The West Loop is Houston's second busiest freeway (after the Southwest Freeway). The I-610 is often referred to as the West Loop parking lot because of the severe congestion that occurs between I-10 and US59. Figure 93. I-610 loop and US 59 interchange, Houston, TX The Uptown/Galleria district of Houston, Texas, is a dynamic, urban community in Houstonâs West Loop, about five miles west of Downtown surrounding The Galleria, a large, upscale, indoor mall and commercial office complex. Operation of traffic signal systems in oversaturated conditions Page 157
Figure 94. Skyline view of Uptown Houston About 42,000 residents live in the Houston Galleria's apartments, modern townhouses, and single-family homes and there is five million square feet of retail space. With the growing demand for residential property in Uptown Houston, developers have increased their activity in the area. Uptown is also host to Houston's largest hotels, which host about 20 million visitors a year. The areaâs commercial activities and proximity of major highways generates different traffic patterns during A.M. and P.M. peak periods. Development of Critical Route Scenarios The first step in the process to develop effective control strategies is to identify the critical oversaturated routes through the network. Once the critical routes, movements, and bottleneck links are identified, a wide range of control strategies are evaluated to reduce the detrimental effects of these problematic symptoms. Operation of traffic signal systems in oversaturated conditions Page 158
Figure 95. Parking lot facilities in Post Oak Network During the P.M. peak period, a large number of trips are generated from inside the network, as employees leave their offices. The large number of parking lot facilities in the area (illustrated in Figure 95) and their substantial capacities are indicators to the significant contribution of these egress flows to the P.M. traffic load. In addition to these routes, traffic passes through the network east/west to and from the I-610 loop and south towards US-59. In the analysis of this network, we considered these two routing scenarios separately. First, a critical route scenario representing vehicles passing through the network was developed. Second, a critical route scenario representing the traffic flows generated from the parking garages was developed. Both scenarios were evaluated for the P.M. peak traffic conditions. Scenario 1: Routes Passing Through the Area to Other Destinations This scenario was developed considering that the routes passing through the network on the main arterials (i.e., Westheimer, West Alabama, Richmond Avenue, Post Oak, and Chimney Rock) are the critical routes. The proximity of the network to one of the major interchanges in Houston area results in significant oversaturation on these routes. These major routes are illustrated in Figure 96 through Figure 98. The following routes are the main access routes to I-610 and US-59: 1- EB on Westheimer Road to I-610 NB 2- WB on West Alabama Street to SB frontage road that leads to I-610 SB as well US-59 EB and WB 3- SB on Rice Avenue leads to US-59 EB and WB Operation of traffic signal systems in oversaturated conditions Page 159
4- SB on Sage Road leads to US-59 EB and WB 5- SB on Post Oak Boulevard leads to US-59 EB and WB 6- EB on Richmond Avenue to I-610 NB Figure 96. Network exits and routes to the highways interchange ramps Operation of traffic signal systems in oversaturated conditions Page 160
Figure 97. Post Oak Ave./W Alabama Ave.; exit to I-610 southbound and US 59 ramps Figure 98. Richmond Ave. and Post Oak Blvd.; exits to I-610 northbound and US-59 ramps Operation of traffic signal systems in oversaturated conditions Page 161
System detector volumes were obtained for a large number of the approaches in the network from Harris County, TX. This data was analyzed to identify the critical routes in the network. To identify critical routes, we performed correlation analysis on the detector data. Correlated movements were clustered together to construct critical routes. The critical routes identified by this process were then verified and complemented with knowledge provided by local practitioners. Table 23 lists the critical routes that resulted from this process. These routes are identified graphically on Figure 99. Table 23. Critical routes for Scenario 1 Route Number Route Description 2 WB on Westheimer Rd. 3 WB on Westheimer Rd., left on Chimney Rock Rd., SB on Chimney Rock Rd. 4 EB on Westheimer Rd. 5 EB West Alabama St. to I-610 frontage road SB. 9 Exit from I-610 SB, LT on Post Oak Blvd, SB on Post Oak Blvd., and RT on Richmond Ave. 10 EB on Richmond Ave. 13 SB on Post Oak Blvd. 19 WB on Richmond Ave. Operation of traffic signal systems in oversaturated conditions Page 162
Figure 99. Critical routes for Scenario 1 Development of Arrival Demand Profiles on Critical Routes In this step, critical routes for the scenario were assigned the maximum volumes they can handle taking into account the background traffic (traffic on non-critical routes) that also traverses the same links but with different destinations. An optimization procedure was applied to generate a feasible solution (i.e., volume profiles) for each critical route. The volume optimization process considers the minimum and maximum observed volume for each movement during all of the peak periods in the system detector data that was analyzed. The maximum possible volume is assigned to the critical route, but takes into account the background traffic flow on each link that is not traveling on the critical route. The volume profiles for each of the critical routes in this scenario are shown in Figure 100. Operation of traffic signal systems in oversaturated conditions Page 163
Figure 100. Volume profiles for critical routes in Scenario 1 Once the critical routes were established the corresponding critical movements throughout the network could be identified where critical routes use common links. This is illustrated for three critical routes in Figure 101. Both routes and the movements will be considered later when developing the control strategies. For each critical movement, we protect the v/c ratios to be as low as possible to ensure that the critical routes experience the minimum levels of TOSI and SOSI that are possible. Figure 101. Critical routes 2, 3, 5 and the corresponding critical movements Operation of traffic signal systems in oversaturated conditions Page 164
Scenario 2: Critical Routes Generated from Traffic Inside the Network In Scenario 2, traffic generated from inside the network (i.e., parking lots) that is leaving the network is considered to be the critical route flows. Seven routes were identified based on the volume correlation analysis and the views and insights of local practitioners. As in the analysis of Scenario 1, the volume analysis consisted of analyzing the correlation of detector data during the peak period. Compatible movements on a possible route that were correlated to be higher or lower during the same time periods were clustered together to form a critical route. After identifying the critical routes, volumes on these routes were maximized to the highest limits they can carry. Any remaining traffic volume from the actual count data on a link that is part of a critical route is considered as background traffic. The background traffic consists of both non-critical routes traffic and the stable fraction of the traffic on the critical routes during the congested period. The minimum and maximum observed volumes are used to establish network background traffic. Table 24 and Figure 102 describe and illustrate Scenario 2 critical routes (i.e., outbound routes). Table 24. Critical routes for Scenario 2 (outbound routes) Route Number Route Description 6 NB on Sage Rd., RT EB on Westheimer Rd. 7 WB on Alabama St., LT NB on Post Oak Blvd., and RT EB on Westheimer Rd. 8 WB on Hidalgo Rd. 11 SB on Post Oak Blvd., and LT EB on Richmond Rd. 12 SB on Sage Rd., LT EB on Richmond Rd., and RT SB on Post Oak Blvd. 14 SB on Sage Rd. 20 NB on Post Oak Blvd. Operation of traffic signal systems in oversaturated conditions Page 165
Figure 102. Critical routes for Scenario 2 The volume profiles of outbound routes (illustrated in Figure 103) show that the traffic volumes start to increase substantially around 4:30 P.M. This surge in volume is due to the trips originating from the commercial offices. Shortly thereafter, the critical routes quickly become oversaturated and remain congested until the end of the peak period after 6:00 P.M. Figure 103. Volume profiles for critical routes in Scenario 2 Operation of traffic signal systems in oversaturated conditions Page 166
In the next section, we discuss common issues of oversaturation in the network under both critical route assumptions. Following this discussion, we discuss the control strategies that were applied to each routing scenario, and the timing plans and parameters that were generated from applying the optimization methodology. Development of Control Strategies for Scenarios 1 and 2 During the peak hours, the Post Oak network faces serious recurrent oversaturation due to the trips generated from the commercial office parks combined with the pass-through traffic en route to and from the freeways. The symptoms of the oversaturation are characterized by starvation, storage blocking, and entry and exit driveway blocking. Initial evaluation using Vissim simulation software indicated that the dominant symptoms affecting the traffic operation in the network were widespread spillback (SOSI > 0), overflow queue formation (TOSI >0), and secondary congestion at driveways. In particular, many left-turn storage bays spill back into the lanes for the through movements. The left-turn spillbacks are caused by inadequate green times and limited storage capacities. Problematic Symptoms of Oversaturation in the Network The following problematic symptoms of the oversaturation were observed in the analysis of the operation of the baseline signal timing strategy: ⢠Spillback and de facto red o Intersection of Richmond Ave. and Sage Rd. o Intersection of Richmond Ave. and Rice Ave. o Intersection of Yorktown and West Alabama St. ⢠Queue formation o WB and SB at Intersection of Post Oak Blvd. and West Alabama St. o SB at Intersection of Richmond Ave. and Sage Rd. o SB and NB at Intersection of West Alabama St and Sage Rd. o EB on Westheimer @ Post Oak intersection, McCue intersection, and Chimney Rock. ⢠Storage blockage o W. Alabama ⢠Entry drive blockage on: o W. Alabama St. between Sage Rd. and Post Oak Blvd. o EB Westheimer Rd. between Yorktown and Sage Rd. o Post Oak between Hidalgo and Richmond These symptoms are illustrated in Figure 104 and Figure 105 below. Operation of traffic signal systems in oversaturated conditions Page 167
Figure 104. Symptoms of oversaturation in the Post Oak network Figure 105. Simulation snapshot showing queue spillback along a critical route Operation of traffic signal systems in oversaturated conditions Page 168
Since persistent, recurring spillback of left-turn bays was identified as one of the dominating factors influencing the operation during the peak period, it was essential to develop a control strategy that explicitly addresses this problem. The left-turn phase reservice strategy was chosen to provide the extra capacity to the left-turn movements. To prevent spillback on short links, we also chose to test the use of the strategy to flare the green. Flaring the green provides extra capacity to downstream intersections in order to clear existing queues by restricting the upstream green times and holding those vehicles in upstream queues. The arterials on the perimeter of this network are characterized by long spacing between intersections and thus high capacities to store vehicles upstream of the bottleneck areas. While the spacing on the perimeter of the network allows the use of large cycle lengths, the implementation of such large cycles would not be feasible for the interior intersections that are much more closely spaced. In particular, applying longer cycles in the interior locations would cause build-up of queues that would block the many entry and exit driveways from the commercial buildings. Therefore, we determined that a critical component of the mitigation strategies would be to run interior and closely-spaced intersections on half cycles. Table 25 and Figure 106 summarizes the control strategies that were tested during the queue management regime, which intersections these strategies were applied, the improvements expected from applying the strategy, and the operational objectives of each method. These mitigation strategies were applied only during the processing regime. Operation of traffic signal systems in oversaturated conditions Page 169
Table 25. Control strategies applied in Scenario 1 Control Strategy Location Expected Improvement Operational Objective Left-turn phase Reservice WBL Richmond Ave. & Rice Ave. Prevent storage blocking Queue management NBL Yorktown & Westheimer Rd. âFlare the greenâ EB on W. Alabama St. Clear downstream queues Improve progression Queue management & Throughput maximization WB on Hidalgo St. WB on Richmond Ave. Half-Cycling W Alabama St. & Yorktown Improve green utilization Prevent starvation Prevent secondary congestion Queue management & Delay minimization W Alabama St. & Sage Rd. W Alabama St. & McCue St. Hidalgo St. & Sage Rd. Hidalgo St. & Rice Ave. Richmond Ave. & Rice Ave. Richmond Ave. & Sage St. Right turn overlap Post Oak & Richmond Post Oak & Westheimer Prevent storage blocking Queue management Gating Operation of traffic signal systems in oversaturated conditions Page 170
Figure 106. Control strategies applied in Scenario 1 Scenario 1: Timing Plan Development Two strategies were generated based on the assumptions of Scenario 1 that routes through the network are the critical routes. Both strategies explicitly account for the three regimes of oversaturated operations (loading, processing, and recovery). The objectives of the first timing strategy were to minimize delay during the loading regime, implement queue control strategies during the processing regime, and then maximize throughput during the recovery regime. This is denoted later as (min delay-manage queues-max throughput). A second timing strategy was developed to maximize throughput during both the loading and recovery regimes and manage queues during the processing regime. This strategy is denoted as (max throughput-manage queues-max throughput). These strategy combinations were then tested in simulation. The cycle, splits, and offsets of the timing plans during each of these regimes were generated using the timing plan development framework presented in Chapter 2. Since the volume profiles for critical movements were used to generate timing plans instead of the movements with the highest volumes, multiple timing plans were generated for each time bin in the volume profile (i.e., a total of 17 plans for each control objective resulting in a total of 51 timing plans that were evaluated in this procedure). Operation of traffic signal systems in oversaturated conditions Page 171
An optimization procedure was then conducted to identify the optimal timing plans and their optimal switching points to meet the assumed control objectives in each regime and duration constraints for each timing plan. Figures 107 and 108 illustrate the resulting schedules of timing plans and switching times between the plans in each strategy. Table 26 describes the characteristics of the two strategies. Figure 107. Min Delay-Queue Management-Max Throughput Strategy (Strategy 1) Figure 108. Max Throughput-Queue Management- Max Throughput Strategy (Strategy 2) Operation of traffic signal systems in oversaturated conditions Page 172
Table 26. Description of strategies Strategy No. Loading Regime Processing Regime Recovery Regime Operation Objective Timing Plan # Starting time Operation Objective Plan # Starting time Operation Objective Plan # Starting time 1 Delay min 11 3:30 Queue management 22 4:30 Throughput max 43 5:30 2 Throughput Max 34 3:30 Queue management 20 4:30 Throughput max 37 5:00 Timing Plans (1-17) are plans that are designed for the delay-minimization objective, timing Plans (18-34) are plans that are designed for the queue management objective, and Plans (35-51) are plans that are designed for the throughput maximization objective. Table 27 shows the cycle lengths and timing plans numbers used in each control strategy. Table 27. Cycle lengths used in each strategy Timing Plan # Cycle length (sec) 11 150 Strategy 1 22 90 43 160/80 34 120 Strategy 2 20 100/50 37 160/80 Scenario 1: Simulation Experiment The Post Oak network was coded in Vissim. The ring barrier controller (RBC) was used to operate all of the intersections in the network and implement the control strategies. Data collection points were placed at all of the entry and exit points and along the critical routes in the network. A 30-minute warm-up period was used before data collection was started. Thirty minutes of simulation time after the peak period was added as well to clear the network of queues at the end of the simulation. Each control strategy was evaluated via five simulation runs with Operation of traffic signal systems in oversaturated conditions Page 173
different seed numbers. Figure 109 shows a snapshot of the coded network. Vissim generates the following performance parameters that were used to evaluate the control strategies: ï§ System total delay ï§ Average delay per vehicle ï§ System total stopped delay ï§ Average stopped delay per vehicle ï§ Average number of stops per vehicles ï§ Average speed ï§ Number of stops ï§ Intersection throughput ï§ System traffic load Figure 109. Post Oak network modeled in Vissim Scenario 1: Results for Critical Routes Passing Through the Network The results of Strategies 1 and 2 are presented in Table 28. The average value and standard deviation (expressed as a percentage) for each high-level summary performance measures are presented in the table. Modest improvements over the baseline timing plan are obtained for most of the performance measures with only Strategy 2 (max throughput-manage queues-max throughput) resulting in a higher number of average and total stops. Operation of traffic signal systems in oversaturated conditions Page 174
Table 28. Performance evaluation of strategies on Scenario 1 Baseline Strategy 1 Strategy 2 Performance Measure Mean St. Dev. Mean St. Dev. Mean St. Dev. Total delay time [hr] 3772.57 3.27% 3327.30 2.47% 3626.78 3.07% Average delay time per vehicle [s] 227.76 2.02% 208.77 3.81% 219.06 3.31% Total stopped delay [hr] 2645.00 4.36% 2420.48 2.02% 2448.31 3.92% Average stopped delay per vehicle [s] 159.68 4.93% 144.91 3.05% 147.88 4.35% Average number of stops per vehicle 11.65 7.72% 10.52 2.58% 12.33 2.98% Average speed [mph] 9.77 4.01% 8.79 2.14% 8.52 1.34% Total number of stops 694,942 5.90% 632,772 4.93% 734,998 5.43% Figure 110 presents the improvements of the strategies and the baseline control plan for each performance measure. These results compare the strategies against a poorly-designed timing plan (75s cycle length with simultaneous offsets, with splits allocated proportionally to each phase demand) because the baseline timing plan that was provided by the City of Houston, TX is already a rather efficient operation. While oversaturation occurs in the network and is significantly debilitating for mobility during the peak period, these results indicate that conditions could be worse Figure 110. Network summary performance measures for Scenario 1 Operation of traffic signal systems in oversaturated conditions Page 175
Figure 111 and Figure 112 present the throughput performance of Strategy 1 and Strategy 2 at the intersection level. These Figures compare the throughput at each intersection for the strategy with the baseline over the peak period. Negative values on these figures indicate that the baseline plan is more efficient at processing vehicles and positive values indicate that the mitigation control strategy is more efficient at processing vehicles at that intersection during the peak hour. Figure 111. Intersection throughput improvement for Strategy 1 Operation of traffic signal systems in oversaturated conditions Page 176
Figure 112. Intersection throughput improvement for Strategy 2 Figure 113 combines the information from the previous two figures on one display to compare the two strategies against each other. Strategy 2 demonstrates better throughput performance, as it improved the throughput of more interior intersections than Strategy (1). This is a validating result since Strategy (2) was designed to maximize system throughput before and after the processing regime. Strategy (1) was designed to minimize delay during the loading regime and maximize throughput during the recovery regime. Recall also that in Strategy 2 interior network intersections were run with half cycles. This type of operation reduced the length of overflow queues during red and improved green time utilization by reducing storage blocking and the resulting starvation symptoms. However, neither strategy could improve the throughput of intersections along Westheimer Road because in both cases the links on Westheimer Road were designated as gating links. Operation of traffic signal systems in oversaturated conditions Page 177
Figure 113. Comparison of intersection throughput between the two strategies Network traffic loads (total vehicles in the system) during the peak period varied significantly for each control strategy. The total number of vehicles in the system (network load) was calculated by subtracting the total traffic exiting the system from the total traffic entering the system. Traffic loading trajectories for the three control strategies are illustrated in Figure 114 below. As shown, the baseline timing plan is effective in processing vehicles through the system during the loading. However, as the volumes continue to increase during the processing regime and queues start building up rapidly the baseline timings splits became inadequate to process the peak load and a large number of vehicles are gated outside of the network area. As the volumes begin to reduce during the recovery period, the latent demand then fills the network considerably and it takes much longer to clear those overflow queues from the system. Strategies 1 and 2 do a better job of spreading the traffic load over the peak period by more efficiently utilizing the space inside the network for storing vehicles. During the loading regime, both mitigation strategies protected the critical routes from becoming saturated early, since the v/c ratios of the phases on the critical routes were all minimized in the timing plan design. At 4:30 P.M., both strategies shifted to queue management strategies (i.e., phase reservice, flare the green, and half-cycling). At 5:00 P.M., Operation of traffic signal systems in oversaturated conditions Page 178
Strategy 2 shifted back to throughput maximization control while Strategy 1 shifted to throughput maximization control at 5:30 P.M. The different switch points for the two strategies were determined based on the optimization process described earlier. Figure 114. Number of vehicles in the system for each strategy for Scenario 1 Table 29 summarizes system-wide performance measures for the three strategies. As shown in the table, Strategy 1 is most effective in reducing total delay and total stops. Strategy 2 provides the fastest recovery time, but at the expense of higher delay and stops. Both strategies were able to reduce the recovery time by more than 10 minutes. This theme has been found to be recurrent in other test cases presented later in this report. Mitigation strategies for oversaturated conditions are particularly effective during the recovery regime. Table 29. System-wide Summary performance measures Metric Baseline Strategy 1 Strategy 2 Clearance Time (min) 27.5 17.5 14.5 Total Delay (hour) 3777.5 3327.3 3626.7 Total Stops (vehicle) 694.942 632,772 734,998 Max load in the system (vehicle) 2,596 2,202 2,196 Operation of traffic signal systems in oversaturated conditions Page 179
Scenario 2: Critical Route Flows from Origins Inside the Network Control strategies for Scenario 2 were developed using the same methodology applied to Scenario 1, but considered different assumptions on the critical routes. In Scenario 1 we assumed that the critical routes began outside of the network and progressed through the network to external destinations. In Scenario 2 we consider that the critical routes have origins inside of the network and are progressing to destinations outside the network. Similar to Scenario 1, we developed two mitigation strategies explicitly taking into account the three regimes of the scenario. The objectives of the Strategy 1 were to minimize delay during the loading regime, manage queues during the processing regime, and then maximize throughput during the recovery regime (denoted as min delay-manage queues-max throughput). The objectives of the Strategy 2 were to maximize throughput during both the loading and recovery regimes and manage queues in the processing regime (denoted as max throughput-manage queues-max throughput). Table 30 presents the control strategies selected for Scenario 2 with their corresponding operational objectives and expected outcomes. Figure 115 illustrates spatially where those strategies were planned to be tested in Scenario 2. Table 30. Attributes of control strategies selected for testing on Scenario 2 Control Strategy Location Expected Improvement Operational Objective Left-turn phase Reservice WBL Richmond Ave. & Rice Ave. Prevent spillback Queue Management SBL Sage Rd. & Richmond Ave SBL Post Oak & Richmond Ave. SBL Chimney Rock & Richmond Ave. SBL Post Oak Ave. & Richmond Ave. NBL Yorktown & Westheimer Rd. âFlare the greenâ WB on Westheimer Rd. Clear downstream queues Improve progression Enhance traffic throughput Queue Management & Throughput maximization WB on Richmond Ave. WB on Richmond Ave. Operation of traffic signal systems in oversaturated conditions Page 180
Figure 115. Spatial illustration of control strategies for Scenario 2 Scenario 2: Development of Timing Plans The timing plans and TOD schedule for implementing the plans were generated using the same methodology followed for Scenario 1. Fifty-one plans were generated using the iterative procedure described previously. Timing plans 1-17 were developed to minimize total delay, timing plans 18-34 were developed to manage queues, and timing plans 35-51 were developed to maximize throughput. An optimization process was then applied to identify the sequence of three timing plans that produces the best performance for the objectives in each regime of operation. The resulting schedule of timing plans corresponding to Scenario 2 are shown in Figure 116 and Figure 117. Start times for each plan in the schedule are shown in Table 31. Cycle lengths for each plan in the two strategies are listed in Table 32. Operation of traffic signal systems in oversaturated conditions Page 181
Figure 116. Min Delay-Queue Management-Max Throughput timing plan schedule Figure 117. Max Throughput-Queue Management-Max Throughput timing plan schedule Operation of traffic signal systems in oversaturated conditions Page 182
Table 31. Plan start times for each strategy Strategy No. Loading Regime Processing Regime Recovery Regime Operation Objective Plan # Starting time Operation Objective Plan # Starting time Operation Objective Plan # Starting time 1 Delay min 3 3:30 Queue management 29 5:00 Throughput max 42 5:30 2 Throughput max 37 3:30 Queue management 30 5:00 Throughput max 45 5:45 Table 32. Cycle lengths used in each plan for Scenario 2 Timing Plan # Optimal Cycle length (sec) 11 80 Strategy 1 22 90 43 150 34 120 Strategy 2 20 150 37 100 Performance measure comparisons are presented in Figure 118 through Figure 121.Traffic loading profiles are shown in Figure 122. Summary network performance statistics are shown in Table 34. The results show significant improvements in throughput and delay compared to the âbadâ plan as well as comparing the mitigations to the baseline plan. This can be explained by the fact that the mitigation strategies were specifically designed to address Scenario 2, which might not be the case for the baseline plan. This is a confirmation of the importance of determining the critical routes in oversaturated networks. Table 33 indicates that Strategy 1dominates Strategy 2 on all of the summary performance measures related to delay and stops. This is not surprising since Strategy 1 focuses on the minimization of delay during the loading regime and Strategy 2 focuses on maximizing Operation of traffic signal systems in oversaturated conditions Page 183
throughput. Maximizing throughput at the beginning of a scenario that evolves slowly may apply the preferential treatment to the critical routes a bit too soon, allowing more vehicles in the system, but storing more of those vehicles on side streets. Table 33. System-level comparison of performance of the two strategies for Scenario 2 Figure 118. Comparison of performance improvements of the two strategies with the baseline for Scenario 2 Figure 119 and Figure 120 below illustrate the throughput performance of the strategies compared to the baseline plan. The two figures are then combined in Figure 121. The comparative performance is rather striking. Strategy 2, where throughput is maximized in both the loading and recovery regimes, improves the throughput for 14 of the 18 critical links where Strategy 1 which focuses on minimizing delay improves the throughput for only four critical links. This again Performance Measure Strategy 1 Strategy 2 Mean St. Dev. Mean St. Dev. Total delay time [hr] 3172.96 4.97% 3327.30 1.67% Average delay time per vehicle [s] 187.02 4.82% 198.07 1.81% Total stopped delay [hr] 2156.24 5.57% 2242.04 1.72% Average stopped delay per vehicle [s] 127.09 5.43% 133.47 1.85% Average number of stops per vehicles 9.34 6.70% 11.48 2.58% Average speed [mph] 9.77 5.19% 9.39 1.14% Total number of stops 570,284 6.86% 693,982 2.43% Operation of traffic signal systems in oversaturated conditions Page 184
illustrates the importance of selecting the appropriate optimization objectives during each regime of the scenario. Figure 119. Intersection throughput improvements for Strategy 1 on Scenario 2 Operation of traffic signal systems in oversaturated conditions Page 185
Figure 120. Intersection throughput improvements for Strategy 2 on Scenario 2 Operation of traffic signal systems in oversaturated conditions Page 186
Figure 121. Comparison of throughput improvements between the two strategies Figure 122 compares the traffic loading over time during the scenario for the baseline and the two strategies. In the loading regime, Strategy 1, which minimizes delay, follows the baseline plan initially, but then begins to store more vehicles in the system than the baseline. When the queue management plan is started at 5 P.M., the number of vehicles in the system stabilizes and remains more or less constant until the end of the simulation time. In comparison, Strategy 2 initially stores more vehicles in the system as it focuses on maximizing throughput on the critical routes. The performance during the processing regime is more peaked than Strategy 1, but still does not result in the spiking that occurs in the baseline operation. These results mimic what was found in other test cases that will be presented later. Operation of traffic signal systems in oversaturated conditions Page 187
Figure 122. Number of vehicles in the system for each strategy for Scenario 2 Table 34. System-wide Performance Results for Scenario 2 Conclusions from the Post Oak Test Case The simulation results from the Post Oak network test case showed that significant improvements are achievable by applying the methodology developed in this research. This test case extended the methodology applied to the Reston Parkway test case by explicitly considering the three regimes of operation during a scenario. In the Reston Parkway case, we considered application of only a single timing plan during the entire scenario. In this test, we proposed two canonical combinations of optimization objectives (1) min delay-manage queues-max throughput, and (2) max throughput -manage queues-max throughput. One of the goals in this test case was to determine some indication of the superiority or at least characterize the differences between the two canonical strategies. Two feasible critical route scenarios were identified from system detector data logs obtained from the City of Houston and from discussions with City and County traffic engineers. Mitigation strategies were developed for each critical scenario as a combination of ad-hoc inspection of the Baseline Strategy 1 Strategy 2 Clearance Time (min) 27.5 17.5 14.5 Total Delay (hr) 3777.5 3327.3 3626.7 Total Stops 694.942 632,772 734,998 Max load in the system (vehicles) 2,596 2,202 2,196 Operation of traffic signal systems in oversaturated conditions Page 188
networkâs critical locations and the optimization procedure developed for determining feasible combinations of cycle, splits, and offsets described in Chapter 2. The optimization process was extended further to consider the selection of a sequence of timing plans and the associated switching times between the timing plans. Each timing plan in the sequence in each regime of operation was selected to achieve an optimization objective that represented the system objective for that regime (minimize total delay, maximize total throughput, or manage queues). These two canonical strategies were then tested on the critical routing scenario using simulation. Both routing scenarios showed that significant performance improvements were possible by applying mitigation strategies. The first critical routing scenario showed less impressive improvements for the two strategies over the baseline performance for measures based on delay and stops but more substantial improvements in intersection throughput. The second scenario showed more impressive improvements over the baseline for delay and stops, but less substantial improvements to throughput. In both scenarios, the strategy that minimized delay during the loading period significantly outperformed the strategy that maximized throughput during the loading period on measures related to stops and delay. The strategy that maximized throughput during both loading and recovery regimes outperformed the other strategy on throughput. Neither strategy would be considered to completely dominate the other, so it is still an open question whether or not the system objective should be to maximize throughput or minimize delay during the loading regime. We did not test a strategy which minimized delay during the recovery regime, but it was clear from both scenarios that maximizing throughput during the recovery regime is an effective transition in optimization goals. This methodology and approach focuses on a top down method to identify and mitigate oversaturated conditions with fixed-parameter timing plans. In the next section, we revisit our âbottom upâ approach using real-time estimates of TOSI and SOSI to directly calculate modifications to green time on an oversaturated route. The theory guiding this heuristic procedure is first described and then two test cases are presented. The first test case describes the application of the process on the TH55 test network in Minneapolis, MN. The second test case describes the application of the process on a grid network in downtown Pasadena, CA. Following this section, we demonstrate the online application of mitigation strategies using real-time detector and TOSI / SOSI oversaturation estimates. Operation of traffic signal systems in oversaturated conditions Page 189
Using TOSI and SOSI Measures to Directly Calculate Green Time Adjustments In the previous section, we discussed a top-down approach for generating mitigation timing plans using an optimization methodology. This methodology is experimental and relatively complicated and could not be applied in practice without additional research and development effort. This optimization methodology also does not directly integrate the measurement of TOSI and SOSI metrics developed earlier because it was developed in parallel to those measures. In this section we address the challenge of oversaturated conditions from the bottom-up by developing a heuristic process that can compute green time adjustments for oversaturated routes. This process uses the TOSI and SOSI measures to determine the amount of green time to add and subtract, respectively, from a phase. The theory and foundation of the heuristic is first presented and then followed by two test cases. The first is a simple situation on the TH55 arterial network in Minneapolis, MN. This test case validates the basic concept. The second test case is a grid network in Pasadena, CA. This test case illustrates the extension of the methodology for two oversaturated routes that cross at a critical location. The Pasadena, CA network was selected for testing as the city has current interest in deploying the experimental hardware and software for calculating queue lengths and TOSI/SOSI measures. Forward-Backward Procedure In this section, we introduce a heuristic procedure to mitigate traffic congestion along an oversaturated route by directly applying the TOSI and SOSI theory described in Chapter 2. The procedure, denoted as the forward-backward procedure (FBP), can be applied for both online and offline signal timing adjustments. In these test cases we apply the procedure in an offline manner. In the FBP, the forward process follows the direction of traffic on the oversaturated route and aims to remove the oversaturation by changing the green and red times for all phases along the route. The backward process follows the opposing direction of traffic on the oversaturated route and evaluates the feasibility of the changes to green and red times by considering the constraints such as minimum green times and queue storage space. In this procedure, our intent is not to find the optimal solution by solving a complicated optimization program; instead, we present a logical heuristic to address oversaturation by adjusting green times and offsets. The effectiveness of the proposed FBP is demonstrated in simulation tests. For the sake of simplicity we keep the cycle length of each intersection unchanged. The control variables at each intersection of interest are the green and red times for the oversaturated phase. We should note that, although the cycle length is unchanged, the changes to the green and red times do not necessarily have the same absolute value and hold the opposite sign. As we will explain further, the changes to the green and red times can subsequently be transformed into the Operation of traffic signal systems in oversaturated conditions Page 190
adjustments of offsets and green splits. We first introduce the control variables and basic theory for applying TOSI and SOSI measures in order to mitigate oversaturation. Control Variables In the FBP, two sets of control variables ,n irâ and ,n igâ , namely red time changes and green time changes for phase i at intersection n , are introduced for each oversaturated phase. The two control variables have direct association with specific oversaturation mitigation strategies. Whether to change red or green time is determined by the cause of the oversaturation. Changing red times (i.e. ,n irâ ) aims to eliminate spillover and changing green times (i.e. ,n igâ ) aims to clear an overflow queue. A positive red time change (red extension) means that extra red time is added. Since the cycle length is kept unchanged, the green start would be postponed with the red extension (see Figure 123a) and the total green time is reduced. A negative red time change (red reduction) means a portion of red time is cut from the end of red, therefore, green start will be advanced (see Figure 123b) and the total green time is increased. Similarly, a positive green time change (green extension) indicates that additional green time is added to the original end of the green time (see Figure 123c), and a negative green time change (green reduction) represents that some green time is cut from the end of green (see Figure 123d). Depending on the offset reference point used for the intersection (start of yellow, start of green, barrier crossing, etc.), each case of adjusting green or red may require a corresponding change to the offset value. ,n ig , 0n irâ > Before: , 0n irâ < ,n ig ,n ig , 0n igâ < ,n ig After: Before: After: , 0n igâ > , 0n igâ > , 0n igâ < (a) (b) (c) (d) Before: After: Before: After: Figure 123. Red time changes and green time changes The values ,n irâ and ,n igâ can be easily transformed into the values of new offset and green duration, which can easily be modified in the signal timing plan. If we assume that the oversaturated approach is the coordinated direction and the green start time of the coordinated phase is the offset reference point, Eq. can be used to calculate the new offset ï¥no , green Operation of traffic signal systems in oversaturated conditions Page 191
duration ï° ,n ig and red duration ,n irï¥ . Here cn is the cycle length for intersection n. Figure 124 demonstrates an example of signal timing changes at one intersection with , 0n irâ < and , 0n igâ > . ï¥ ï° ï° , , , ,, , , n n n i n i n i n in i n i n n i o o r g g r g r c g  = + â  = ââ + â  = â ï¥ Eq. 51 ,n igâ,n irâ no ï¥no cn cn ,n ig ï° ,n ig Before After ,n igâ ,n ir ,n irï¥ Figure 124. Signal timing changes ( , 0n irâ < , , 0n igâ > ) Basic Strategies Using TOSI and SOSI Measures Simply speaking, there are two ways to deal with oversaturation: one is to increase the downstream output rate and the other is to constrain the upstream input rate. These two basic actions result in three mitigation strategies for an oversaturated phase when considering the values of TOSI and SOSI. These three basic cases will be used in the FBP to compute the changes to green and red times. Green Extension for Scenario 1: TOSI > 0 & SOSI = 0 Since a positive TOSI value indicates an overflow queue at the end of a cycle and zero SOSI value indicates that there is still spare capacity to store vehicles in the downstream link, the strategy to deal with this situation is to extend the green time for the oversaturated phase. Figure 125 illustrates this case by presenting the shockwave profiles for two intersections. After extending the green in Figure (b), the overflow queue disappears and TOSI becomes zero. The green extension can be calculated as the following Eq. 52. âðð,ð = ðððð¼ð,ð â ðð,ð Eq. 52 where âðð,ð is the adjustment to the green time at intersection n for phase i; ðððð¼ð,ð is the TOSI value at intersection n for phase i; and ðð,ð is the green time at intersection n for phase i. Note that positive âðð,ð means green extension; and a negative value means green reduction. By extending or reducing green, the start time of the following red signal will be shifted. Operation of traffic signal systems in oversaturated conditions Page 192
Distance Time #n #n+1 Distance Time Shockwave Trajectory .n igâ Shockwave Trajectory 1,n ir + 1,n ig + ,n ir ,n ig,n iO 1,n iO +1,n ir + 1,n ig + ,n ir ,n ig,n iO 1,n iO + a) Before Green Extension (TOSI > 0, SOSI=0) b) After Green Extension (TOSI = 0, SOSI=0) Vehicle Trajectory fv fv Vehicle Trajectory Figure 125. Green extension for Scenario 1 We should also note that by extending the green time at the current intersection, the queue length at the downstream approach may increase and result in spillover. This problem will be addressed in the FBP. Red Extension for Scenario 2: TOSI = 0 & SOSI > 0. If SOSI is larger than zero, it indicates that the downstream queue spills back to the upstream intersection and results in unusable green time as shown in Figure 126. But since TOSI is zero, all queued vehicles can be discharged even with reduced green time. One way to remove downstream spillover is to gate the upstream flow by extending the red time. The red extension can be calculated as the following Eq. 53. âðð,ð = ðððð¼ð,ð â ðð,ð Eq. 53 where ârn,i is the adjustment to the red time at intersection n for phase i; and ðððð¼ð,ð is the measured or estimated SOSI value at intersection n for phase i. The positive ârn,i means red extension and a negative value means red reduction. Note that by adjusting the red time, the start of the following green will be shifted. Operation of traffic signal systems in oversaturated conditions Page 193
Distance Time Shockwave Trajectory #n #n+1 1,n ir + 1,n ig + ,n ir ,n ig,n iO 1,n iO + .n irâ Distance Time Shockwave Trajectory 1,n ir + 1,n ig + ,n ir ,n ig,n iO 1,n iO + Vehicle Trajectory fv Vehicle Trajectory fv a) Before Red Extension (SOSI > 0, TOSI =0) b) After Red Extension (SOSI = 0, TOSI = 0) Figure 126. Red extension for Scenario 2 Downstream Red Reduction for Scenario 3: TOSI > 0 & SOSI > 0. A more serious situation exists when both TOSI and SOSI are larger than zero, as shown in Figure 127a. In this case, at the upstream intersection a portion of the green time is unused because of the downstream spillover. At the same time, the useable green time at the upstream intersection is not sufficient to discharge queued vehicles, i.e., an overflow queue exists. One way to address this scenario is to increase downstream capacity by reducing the red time at the downstream intersection. As shown in Figure 127b, by reducing the downstream red, positive TOSI and SOSI values for the upstream intersection will be reduced. Once the downstream spillover is removed or reduced, the unusable green time at the upstream intersection may become available and can be used to discharge the overflow queue. If TOSI < SOSI, the overflow queue can be cleared by using this strategy. The reduction of downstream red can be calculated as the following Eq. 54. âðð+1,ð = ðððð¼ð,ð â ðð,ð Eq. 54 As an alternative to reducing the downstream red, we can also deal with this situation by combining the methods for Scenarios 1 and 2 together, i.e., extending both the red and green times. This would mean an increase to the total cycle time at this intersection. In this heuristic procedure, we are attempting to avoid the re-calculation of the cycle time since the calculations for each intersection may result in different cycle values; thus requiring an optimization procedure to select the best value. Operation of traffic signal systems in oversaturated conditions Page 194
Distance Time Shockwave Trajectory #n Distance Time 1,n ir + 1,n ig + ,n ir ,n ig Shockwave Trajectory ,n iO 1,n iO + #n+1 1,n i r +â 1,n ir + 1,n ig + ,n ir ,n ig,n iO 1,n iO + Vehicle Trajectory fv Vehicle Trajectory fv a) Before Downstream Red Reduction (TOSI>0, SOSI >0) b) After Downstream Red Reduction (TOSI=0, SOSI =0) Figure 127. Red reduction at downstream intersection for Scenario 3 Among the three strategies, extending green (Strategy 1) is to increase the discharge capacity for the oversaturated phase; extending red (Strategy 2) is to gate traffic arrivals at the upstream intersection; and reducing downstream red (Strategy 3) is to remove the downstream bottleneck by discharging the queue earlier at the downstream intersection. By considering maximum/minimum green and storage space limitations on side streets, these strategies may be directly applied for an isolated intersection or two intersections in tandem. For a longer oversaturated route, the FBP needs to be applied for two reasons. First, the increase of green time of an upstream approach may create oversaturation on the downstream link and secondly, capacity constraints at a downstream phase may limit the possible signal timing adjustments for the upstream phase. The Forward-Backward Procedure (FBP) The FBP systematically evaluates the three strategies by considering the impacts of the green and red time modifications on upstream and downstream intersections. As shown in Figure 128, the FBP consists of two processes: a forward pass and a backward pass. We will first describe the calculations in the forward process, followed by the calculations in backward process. Int.1 Int.2 Int.n Int.N FORWARD BACKWARD ( )1, 1,,i iTOSI SOSI ( )2, 2,,i iTOSI SOSI ( ), ,,n i n iTOSI SOSI ( ), ,,N i N iTOSI SOSI Figure 128. FBP for an oversaturated route Operation of traffic signal systems in oversaturated conditions Page 195
Forward Process The forward process aims to eliminate both spillovers and overflow queues by reducing red time or increasing green time of oversaturated phases without considering the constraints from other phases at the intersection. The process is applied along the direction of flow and calculates the red and green time changes for each oversaturated phase on the route. Only basic Strategy 1 and 3 (described in the last section) are considered in this pass. For any intersection (except the first and last ones, which need slightly different treatments) on an oversaturated route (see Figure 128), the first step is to reduce red time according to the upstream SOSI value to remove spillover (see Eq. 55). The amount of red reduction at any intersection should accommodate not only the removal of the spillover to the upstream intersection ( 1, 1,n i n iSOSI gâ âà ), but also the increase in the arrival flow due to the red reduction made at the upstream intersection ( 1, F n ir ââ ). The second step is to extend green times to discharge overflow queues. Similarly, we need to account for the green change of the upstream intersection ( 1, F n ig ââ ). Since in the forward process we assume that downstream spillover will be removed by reducing downstream red, the unusable green time caused by spillover will become available and can be used to discharge overflow queues. The backward process will consider the situation if capacity constraints are violated. Therefore, in order to remove an overflow queue, additional green time in the amount of ( ), , ,n i n i n iTOSI SOSI gâ à is needed. The additional green time might be negative if the SOSI value is greater than the TOSI value. This situation would mean that the elimination of spillover has already given enough extra green time to discharge the overflow queue and we need to reduce green time to prevent green starvation. The signal timing adjustment for any intersection except the first and last ones can be calculated by the following equation (Note that the superscript â F â denotes âForwardâ process): ( ) , 1, 1, 1, , 1, , , , F F n i n i n i n i F F n i n i n i n i n i r r SOSI g g g TOSI SOSI g â â â â â = â â Ã  â = â + â Ã Eq. 55 For the first intersection, we only need to consider the situation when TOSI is greater than zero (positive SOSI will be handled at the downstream intersection), so Eq. 55 becomes: ( ) 1, 1, 1, 1, 1, 0 ( ) F i F i i i i r g t TOSI SOSI g â =  â = â Ã Eq. 56 For the last intersection (intersection N), since SOSI is zero (otherwise, this intersection should not be the last one), Eq. 55 becomes: Operation of traffic signal systems in oversaturated conditions Page 196
, 1, 1, 1, , 1, , , F F N i N i N i N i F F N i N i N i N i r r SOSI g g g TOSI g â â â â â = â â Ã  â = â + Ã Eq. 57 Backward Process The forward process follows the traffic direction and adds extra green time ( , , F F n i n ig râ ââ ) for each phase to discharge the overflow queue and to remove spillover. However, available green time increases for some intersections may not be achievable due to the other constraints, i.e., minimum green requirement for conflicting phases and queue storage space limitations. To solve this problem, the backward pass is designed to gate traffic when the green time changes calculated in the forward pass are not achievable. The backward pass starts from the last intersection and follows the opposing direction of traffic on the oversaturated route to determine how much green time needs to be reduced for each phase. To compute the backward adjustment, we first need to calculate the available green time , a n ig for each oversaturated phase. If only the minimum green time requirement for conflicting phases is considered, , a n ig for intersection n and phase i can be calculated by Eq. 58, where cn is the cycle length for intersection n; ,n iZ is the set of conflict phases (other phases in the same ring if dual-ring control is utilized) to phase i at intersection n; and min,n jg is the minimum green time for phase j at intersection n. Note that , a n ig may also be constrained by other conditions, such as the storage space on side streets (i.e., , a n ig can be computed by comparing the maximum queue length with the link length for side streets). , min , , , n i a n i n n j n i j Z g c g g â = â ââ Eq. 58 Next we need to follow the direction of the opposing flow and calculate the residual capacity ,n iR for each phase on the route. Positive ,n iR means the available green time can accommodate the required green time increase ( ), ,F Fn i n ig râ ââ ; negative ,n iR means available green time is insufficient. ( ), , , , , ,...,1a F Fn i n i n i n iR g g r n N= â â ââ = Eq. 59 After calculating the residual capacity for each phase on the route, the backward green time adjustment term B igâ is equal to the minimum residual capacity among all phases on the route, see Eq. 60. Note that B igâ has no subscript ânâ, which means it is the same for every intersection Operation of traffic signal systems in oversaturated conditions Page 197
along the route. If ( ),{1,..., }min 0n in N Râ ⥠, the requested green time increase ( ), , F F n i n ig râ ââ from the forward process will be satisfied at all intersections and no further adjustment is needed in the backward process (i.e., 0 B igâ = ). However, if ( ),{1,..., }min 0n in N Râ < , there is at least one phase on the route where the available green time constraint is violated and the adjustment term B igâ is utilized to make sure the available green time constraints are satisfied at all phases on the route. ( ){ } ,{1,..., }min min ,0Bi n in Ng Rââ = Eq. 60 The final changes to every phase on the route are calculated by Eq. 61, where the red time change is equal to the value calculated in the forward process and the green time change is equal to the summation of the calculated value in the forward process , F n igâ and the backward adjustment term B igâ . , , , , F n i n i F B n i n i i r r g g g â = â  â = â + â Eq. 61 FBP for an Oversaturated Network When extending the FBP to an oversaturated network, the first step is to identify oversaturated routes. For routes that cross each other, an algorithm is designed to allocate the available green time to the conflicting phases. If we have the TOSI and SOSI values for each movement, it is easy to identify oversaturated routes, since the oversaturated movements on the route will have positive TOSI and/or SOSI values. This route need not be a straight line (see Figure 129), since oversaturation at some intersections may be caused by turning movements. When two oversaturated routes intersect with each other, we call the crossing intersection the âcritical intersectionâ. For each oversaturated route, the FBP will determine the changes to the green/red times along the route. There will be a conflict between the two routes at the critical intersection since both directions will be fighting for the available green time. The available green for both phase i and j ( , & a I i jg ) at intersection I can be calculated by Eq. 62, where , &I i jZ is the set of conflict phases to phase i and j at intersection I. , & min , & , , , I i j a I i j I I k I i I j k Z g c g g g â = â â ââ Eq. 62 Operation of traffic signal systems in oversaturated conditions Page 198
Oversaturated Route Critical Intersection I Oversaturated Route Phase i Phase j Figure 129: Oversaturated route and critical intersection For two oversaturated routes that intersect, the available green time at intersection I ( , & a I i jg ) needs to be divided between phase i and j. We can split the available green time proportionally according to the requested green times from the forward process, i.e., , , , & , , , , , & , , F I ia a I i I i j F F I i I j F I ja a I j I i j F F I i I j g g g g g g g g g g  = Ã +   = Ã + Eq. 63 However, such a method of splitting the green time may not be efficient because the binding constraints for available green times on one or both oversaturated routes may not come from the critical intersection. To overcome this deficiency, we first compute the residual capacity for all intersections except the critical intersection I and the backward adjustment term for both directions, ï° B igâ and ï° B jgâ , using Eq. 60. The â^â sign is used because here we did not consider intersection I. We can then calculate the requested green time increase for phase i and j of intersection I , denoted by , R I ig and , R I jg . ï° ï° , , , , , , BR F F I i I i I ii BR F F I j I j I jj g g g r g g g r  = â + â ââ   = â + â ââ Eq. 64 Operation of traffic signal systems in oversaturated conditions Page 199
If , & , , a R R I i j I i I jg g g⥠+ , then the available green time constraint at intersection I is satisfied. The backward process adjustment terms B igâ and B jgâ are equal to ï° B igâ and ï° B jgâ respectively, see Eq. 65. ï° ï° BB i i BB j j g g g g â = â  â = â Eq. 65 Otherwise, the available green time at intersection I cannot satisfy the total requested green time increase for both directions i and j. The total available green time , & a I i jg is split proportionally to two directions, , a I ig and , a I jgaccording to the requested green time increase, see Eq. 66. The total backward process adjustment terms for the two directions is then determined by Eq. 67. , , , & , , , , , & , , R I ia a I i I i j R R I i I j R I ja a I j I i j R R I i I j g g g g g g g g g g  = Ã +   = Ã + Eq. 66 ( ) ( ) , , , , , , B a F F i I i I i I i B a F F j I j I j I j g g g r g g g r  â = â â ââ  â = â â ââ Eq. 67 The final changes to the red/green time can then be calculated using Eq. 61 for both conflicting routes. A Simple Illustrative Example To illustrate the FBP, we consider the following hypothetical scenario involving just two intersections (A and B). Calculations are illustrated in Table 35. Both intersections are in two-phase single-ring operation with a 90s cycle time. Only the route from Aï B is oversaturated. The operation at A is affected by the queuing at B because of the limited link length between A and B (about 25 vehicles storage available or about 400 ft). Assume the speed limit is 35mph so the unimpeded travel time from A to B is about 8s. Operation of traffic signal systems in oversaturated conditions Page 200
Table 35. Illustration of calculation procedure Intersection A Intersection B Phase Green Time 45s 40s Lost Time 8s 8s Phase Red Time 37s 42s Minimum Green 15s 15s Minimum Green â Conflicting phase 15s 15s Offset 0s 3s Upstream storage space 100 vehicles 25 vehicles Intersection A Cycle Green Time TOSI SOSI Overflow Queue Length (per lane) 1 55s 0 0 0 2 52s 0.1 0 2 3 50s 0.2 0 3 4 45s 0.3 0 6 5 45s 0.5 0.1 15 6 45s 0.6 0.2 18 7 45s 0.7 0.2 22 8 45s 0.6 0.2 18 9 50s 0.2 0.1 7 10 55s 0.1 0 2 Intersection B Cycle Green Time TOSI SOSI Overflow Queue Length (per lane) 1 40s 0 0 0 2 40s 0 0 3 3 40s 0 0 6 4 40s 0.1 0 6 5 40s 0.2 0 8 6 40s 0.3 0 9 7 40s 0.2 0 6 8 45s 0.1 0 3 9 50s 0 0 0 10 55s 0 0 0 To compute the recommended changes to the green times at A and B, first calculate the average TOSI and SOSI values over the study duration (neglecting zeros in determining the average). This is illustrated in Table 36 below. Operation of traffic signal systems in oversaturated conditions Page 201
Table 36. Illustration of calculations for green time modifications Intersection A Intersection B Average TOSI 0.33 0.18 Average SOSI 0.15 0 Average green time 49s 43s Intersection A Intersection B Forward Pass Delta-R 0s -(0.15)*49s = -7.3s Delta-G (0.33 â 0.15)*49s = 8.8s 8.8s â 0s + 0.18*43s = 16.5s Backward Pass Delta-R 0 0 Delta-G Min(90s â15s â 8s â45s â8.8s, 90s â15s â 8s â40s â16.5sâ 7.3s, 0s) = 0s Min(90s â15s â 8s â45s â8.8s, 90s â15s â 8s â40s â16.5sâ 7.3s, 0s) = 0s Resulting adjustments Final Delta-R 0s + 0s = 0s -7.3s + 0s = -7.3s (round to -7s) Final Delta-G 8.8s + 0s = 8.8s (round to 9s) 16.5s + 0s = 16.5s (round to 17s) Phase Red 37s - 9s = 28s 42s â(17s- (-7s)) = 18s Phase Green 45s + 9s = 54s 40s + (17s- (-7s)) = 64s Cycle Time 28 + 54s + 8s = 90s 18s + 64s + 8s = 90s Offset 0s + 0s =0s 3s â 7s = -4s (or 86s) Effective changes Reduce side street phase 9s Reduce side street phase 24s Increase phase green 9s Increase phase green 24s Modify relative offset from 3s to 86s Real-World Examples In this section, two examples are provided for applying the FBP to real-world scenarios. In both cases, average values for TOSI and SOSI over the peak hour were used to re-calculate green times on the route. Simulation tests were then conducted with the same random seed to estimate the performance differences with and without applying the FBP. An Oversaturated Intersection The first test site is an oversaturated intersection located on TH55, in Minneapolis, MN (see Figure 130). Congestion usually occurs on the westbound traffic during the A.M. peak period because traffic signals are coordinated for the eastbound direction. Because of the short link Operation of traffic signal systems in oversaturated conditions Page 202
between Winnetka and Rhode Island, queues spill back to the Rhode Island intersection creating large values of SOSI and TOSI. Signal timing plans for the intersections of Winnetka and Rhode Island are shown in Figure 131. Figure 130. Test arterial on TH55, Minneapolis, MN Figure 131. Signal timing plan (A.M. peak) for Winnetka and Rhode Island To test whether the FBP can deal with this situation, we increased the traffic demands on all approaches by 50% in order to create severe oversaturation on the corridor. Figure 132 shows the SOSI and TOSI values at the Rhode Island intersection from a two-hour simulation period. Glenwood Winnetka Rhode Island Boone Douglas TH100 2634 ft 842 ft 1776 ft 2636 ft 2764 ft Winnetka ï® Rho Winnetka Rhode Island Operation of traffic signal systems in oversaturated conditions Page 203
Figure 132. SOSI and TOSI values of Rhode Island westbound Based on the average SOSI and TOSI values, the FBP suggests increasing the green time of the downstream intersection at Winnetka by 10s. Note that the FBP becomes a simple combination of the three strategies described earlier when applied to an individual phase. Figure 133 and Figure 134 show the SOSI and TOSI values before and after the change, respectively. From both figures, it is clear the TOSI and SOSI values are reduced. Figure 135 also shows the queue length at the downstream Winnetka intersection. After applying the FBP strategy, the queue length is significantly reduced. But as expected, the queue lengths on the side streets were increased due to the loss of green time (Figure 136). Figure 133. SOSI values of Rhode Island westbound before and after the FBP Operation of traffic signal systems in oversaturated conditions Page 204
Figure 134. TOSI values of Rhode Island westbound before and after the FBP Figure 135. Estimated queue lengths at Winnetka westbound before and after the FBP Figure 136. Queue lengths on the side street (southbound) at Winnetka intersection before and after the FBP Operation of traffic signal systems in oversaturated conditions Page 205
An Oversaturated Arterial The second test site, as shown in Figure 137, is an oversaturated arterial corridor of five intersections on Fair Oaks Avenue in the City of Pasadena, CA. The length of the corridor is 0.4 mile and the speed limit is 30 mph. To create oversaturation, we assume that there is a large directional flow (3000 vph) from north to south. This may cause serious oversaturated conditions (spillovers and overflow queues) if the signal timings are not properly designed. First, Synchro was used to optimize the signal timings according to the traffic volumes shown in Figure 137. The optimized cycle length is 140s and the north-south phase is the coordinated phase. The simulation was run ten times using different random seeds and each run was for two hours. The first half hour of the simulation was considered as the warm up time, and thus the average values of TOSI and SOSI in the following one and half hours were used to represent the oversaturation condition on the corridor. Since Synchro does not describe the queue interactions between intersections, it cannot deal with the spillover situation and always assumes that the discharging capacity at downstream is available. Therefore, severe oversaturation occurs under optimized timing plans provided by Synchro due to the large directional flow from north to south. Table 37 shows the southbound average SOSI and TOSI values over ten simulation runs and the green time duration of coordinated phase (i.e., the southbound direction). Note that spillover (i.e., SOSI >0) mainly happens at the southbound of Intersection 2 and on average 12.18% of the green time is wasted. This also causes a minor spillover at Intersection 1 with a 2.52% average SOSI value. At the same time, overflow queues (i.e., TOSI >0) exist at every intersection, which indicates insufficient discharging capacities. Fair Oaks Ave. Walnut St. Colorado Blvd. Int. 1 Int. 2 Int. 3 Int. 4 Int. 5 3000 VPH 454 VPH 433 VPH 141 VPH 33 VPH 600 VPH 474 VPH 573 VPH 674 VPH 651 VPH Figure 137.Vissim simulation network Operation of traffic signal systems in oversaturated conditions Page 206
Table 37. Southbound average SOSI and TOSI values under original signal timings Inter. No. Avg. SOSI (%) Avg. TOSI (%) Green Time (Sec) 1 2.52 13.59 83 2 12.18 5.45 110 3 0.99 4.46 81 4 0.00 0.85 94 5 0.00 2.93 98 Since oversaturation exists, the FBP is then applied and the calculation process is shown in Table 38. In the forward process (FP), Eq. (49), (50) and (51) are applied to determine , F n irâ and , F n igâ for each intersection. In the backward process (BP), residual capacity ,n iR is calculated by Eq. 59 and the backward adjustment term B igâ can be determined using Eq. (54). The final red time changes ,n irâ and green time changes ,n igâ are calculated by Eq. (55) and shown in Table 38. Table 39 compares the offset and green time duration (coordinated phase) of the plan provided by Synchro and the plan after the change of the FBP. One can see that the plan from the FBP modifies the offset value of each intersection and increases the green time duration of the coordinated phase for each intersection except Intersection 2. By comparing the signal timing change before and after applying the FBP, we can clearly see the working âlogicâ of the FBP. The original TOSI/SOSI values indicate that the congestion begins from the link between Intersections 3 and 4 as indicated by the positive TOSI and SOSI values at Intersection 3 and significantly grows at Intersections 2 and 1. So the first step in the FBP is to increase the downstream capacity by reducing red time and increasing green time (see the FB column in Table 38). This explains why the green times at Intersections 3, 4, and 5 were significantly increased (see Table 39) even though they had very small TOSI/SOSI values (see Table 1). However, since the available green time is not enough for green expansion, the second step in the FBP is to âgateâ the upstream intersections so the input demand can be reduced. This is why for Intersection 2 the green time was slightly reduced from 110sec to 109sec. With the green and red changes, the offset has been adjusted accordingly. Interestingly, the offset changes for Intersections 3, 4, and 5 generated by the FBP are identical (16sec, see Table 3). This indicates that the FBP results in a âgreen flareâ solution. Operation of traffic signal systems in oversaturated conditions Page 207
Table 38. FBP calculation process Inter. No. , ,n i n iSOSI gà , ,n i n iTOSI gà , a n ig FP BP ,n irâ ,n igâ , F n irâ , F n igâ ,n iR Bigâ 1 2.1 11.3 14 0 9.2 4.8 -5.1 0 4.1 2 13.4 6.0 11 -2.1 1.8 7.1 -2.1 -3.3 3 0.8 3.6 15 -15.5 4.6 -5.1 -15.5 -0.5 4 0.0 0.8 22 -16.3 5.4 0.3 -16.3 0.3 5 0.0 2.9 22 -16.3 8.3 -2.6 -16.3 3.2 Table 39. Offset and green time of two plans Inter. No. Synchro FBP Change Offset Green Time Offset Green time Offset Green time 1 0 83 0 87 0 4 2 138 110 136 109 -2 -1 3 25 81 9 96 16 15 4 23 94 7 111 16 17 5 28 98 12 117 16 91 The average results for the TOSI and SOSI measures for southbound traffic under the two plans are compared in Figure 138. Overall, it is clear that the FBP plan improves the performance of the oversaturated route above and beyond what was calculated by Synchro. The FBP plan reduces the spillover time of the Intersection 2 (the major bottleneck on the route) from 13.4s to 1.5s and clears the spillover at Intersection 1. Because of the increase in discharge capacity, overflow queues at each link on the route have been greatly reduced as well. The average delay per vehicle of the FBP plan is 64s, which is a 12% decrease from the 73s average delay of the Synchro plan. Operation of traffic signal systems in oversaturated conditions Page 208
Table 40 summarizes the network performance under the two plans. With the FBP, the average number of stops has been reduced from 1.66 per vehicle to 1.31 per vehicle and the FBP plan increases the average speed 8.65%. Figure 138.: Comparison of spillover time and overflow queue discharge time 0 5 10 15 1 2 3 4 5 SYNCHRO FBP Inter. # Se co nd s SOSI * Green Time 0 4 8 12 1 2 3 4 5 Synchro FBP Inter. # Se co nd s TOSI * Green Time Operation of traffic signal systems in oversaturated conditions Page 209
Table 40. Network performance comparison SYNCHRO FBP CHANGE (%) Average Delay (Seconds/per veh.) 73.14 64.07 -12.39 Average # of stops (per veh.) 1.66 1.31 -20.90 Average Speed (MPH) 11.13 12.09 +8.65 Figure 139 and Table 41 compare the throughputs of different exits of the network in the two-hour simulation period. The first group represents the throughput of the southbound exit, the second indicates the throughput of northbound exit and the remaining five groups represent the respective throughputs of the side streets at each intersection on the route. One can see that the throughput of the southbound exit, where the large directional goes to, is increased by 4.42% during the simulation time. Due to the decrease of green time on side streets, some side street throughputs are slightly decreased. Overall, the total throughput of the network is increased by 1.58%. Figure 139. Comparison of throughput by route 0 1000 2000 3000 4000 5000 Southbound Northbound Int.1 side st. Int.2 side st. Int.3 side st. Int.4 side st. Int.5 side st. SYNCHRO FBP Nu m . o f V eh icl es Throughputs Operation of traffic signal systems in oversaturated conditions Page 210
Table 41. Comparison of throughput by route SYNCHRO FBP CHANGE (%) Southbound 4315 4505.9 +4.42 Northbound 1342.2 1318.8 -1.74 Int. 1 side streets 1945 1967 +1.13 Int. 2 side streets 916.1 953.8 +4.12 Int. 3 side street 1319.6 1289.6 -2.27 Int. 4 side streets 1779.8 1771.9 -0.44 Int. 5 side street 1646.7 1666.7 +1.21 TOTAL 13264.4 13473.7 +1.58 Similarly, because of the decrease in the side street green time, the side street queue lengths under the FBP plan are longer than in the baseline plan. This is illustrated in Figure 140. Notably the increases to the side street queues are not significant since the green time adjustments were fairly modest at each intersection. Figure 140. Comparison of side streetsâ maximum queue length in each cycle Summary A forward-backward procedure (FBP) was developed to adjust traffic signal timing to mitigate oversaturation on a route. The procedure is based on measured or estimated TOSI and SOSI values. The forward process aims to increase green time by searching for available green time which can be taken from side streets or conflicting phases to improve throughput for the oversaturated route. The backward process is used to gate some intersections to prevent overflow queues and downstream queue spillback when green time increases are insufficient. In the test cases, we calculated the average TOSI and SOSI values and applied a fixed adjustment to a new timing plan for the entire test duration. These tests indicated that the procedure can improve total delay and alleviate oversaturation along the oversaturated route. As expected, side street delays were increased to gain the system-wide improvements. 0 100 200 300 400 500 14 0 56 0 98 0 14 00 18 20 22 40 26 60 30 80 35 00 39 20 43 40 47 60 51 80 56 00 60 20 64 40 68 60 72 00 SYNCHRO FBP Time Fe et Int. 3 WB Max Queue Length 0 50 100 150 200 250 300 14 0 56 0 98 0 14 00 18 20 22 40 26 60 30 80 35 00 39 20 43 40 47 60 51 80 56 00 60 20 64 40 68 60 72 00 Synchro FBP Time Fe et Int. 5 EB Max Queue Length Operation of traffic signal systems in oversaturated conditions Page 211
The proposed procedure was applied for offline signal timing adjustment, but we envision that a similar process could be applied in an online, real-time feedback manner with appropriate interface to the signal controller or a signal control system. The approach at this time would be considered experimental in nature. Further development in a number of directions would be necessary for more holistic application of the procedure in conjunction with the timing plan design principles and rules of thumb illustrated in other sections of this research. In the next section, we will discuss a test case where the online tool was applied to a non-recurrent oversaturated scenario. Based on the status of several queue detection points, a variety of mitigations were evaluated that were envisioned to improve the network performance. Operation of traffic signal systems in oversaturated conditions Page 212
Online Application Test Case: Response to Incident at a Single Common Destination In the previous section, we presented a bottom-up heuristic procedure for processing TOSI and SOSI measurements into green time adjustments for an oversaturated route. This method was shown to be effective in improving performance on the oversaturated route in two test cases by adjusting the green time of the timing plans in an offline manner. The third component of our research methodology for addressing oversaturated conditions with signal timing plans is our development of an online tool that can use detector data including TOSI, SOSI, and queue length measures to select mitigation strategies in an online, feedback control system. This approach was described in Chapter 2. In this section, we describe a test case where this tool was evaluated. We first provided the background and motivation for the test followed by an outline of the methodology used to develop the mitigation strategies. Then we present the findings of the simulation tests and describe conclusions and future directions from the results. The City of Windsor, ON is located southeast of Detroit, MI. Two border crossings in Windsor serve significant traffic flow between the U.S and Canada, the heaviest flows of any crossing at the northern border. In fact, over 65% of the truck traffic between the U.S. and Canada flows through this port of entry. The City of Windsor maintains approximately 300 signalized intersections and one limited access freeway facility. The City is a mixture of downtown grid and sub-urban arterials, with the majority of the signalized intersections in the downtown grid area. The approach to the Detroit-Windsor tunnel is in the heart of the downtown area just north of the intersection of Wyandotte and Goyeau Streets. These features are illustrated in Figure 141. Operation of traffic signal systems in oversaturated conditions Page 213
Figure 141. City of Windsor, ON arterial and freeway network Tunnel operations can be significantly affected when the Customs and Border Protection (CBP) Agency of the U.S. changes policies. This occurs somewhat regularly during terrorist threats and other border protection events, as well as during inclement weather. In these situations, the traffic congestion at the approach to the tunnel can grow significantly and cause intersection blocking, inefficient signal timing, and major queuing and delays. At times, the City has deployed traffic police to the intersection to direct drivers and manages the backups. The intersection at the entrance to the Tunnel and surrounding intersections are illustrated in Figure 142. U.S. CBP facility Tunnel entrance Province of Ontario freeway City of Windsor arterials and freeways Bridge entrance Operation of traffic signal systems in oversaturated conditions Page 214
Figure 142. Intersections near the Detroit-Windsor Tunnel border crossing Recently, the Province and the City have designed signal timing strategies to manage the operation of the intersection when the queues are persistent and oversaturation persists. These strategies are proposed to be triggered by occupancy on queue detectors shown in Figure 143. Figure 143. Detail of detector deployment and operational strategies at Tunnel entrance Tunnel entrance Critical intersection 21 ï Monitor detector status ï Enable/disable Left turn restriction and phase omits ï Windsor, Ontario ï§To USA through tunnel Operation of traffic signal systems in oversaturated conditions Page 215
Based on the status of the queue detectors, pedestrian indications, and other inputs, a number of mitigation actions are taken, as shown in the Table 42. This table is a truth table decision making device. The top part of the table lists the status of the inputs. Y indicates a condition is true, N indicates a condition is not true, and - indicates that it does not matter if the condition is true or false. In the top part of the table, each column of Y, N and - indications must be mutually exclusive to ensure that only one column of actions will be taken at a time. The bottom part of the table then shows which actions will be taken when a certain combination of conditions in the top part of the table is true. Table 42. Proposed operational strategy at the Tunnel entrance These combinations of conditions and actions are converted into ifâ¦then rules using the tool and procedure described in Chapter 2. In this specific case, the top three conditions describe three queue detection locations. Two are shown in Figure 143 (northbound Goyeau entry link and westbound Wyandotte right-turn lane) and the first location (northbound Goyeau entry link) is just to the left of the graphic in the northbound (to the right) direction. Various combinations of queues at each location in the top portion of the table prescribes certain actions to be taken to increase vehicle phase green time and omit turning movements. This strategy was developed to maximize throughput since each set of actions keeps traffic moving on movements that can still operate when queuing exists that blocks operations on other phases. Conceptually this logic seems very reasonable. This logic only provides a mitigation strategy for the critical location. As part of the evaluation process we developed six other potential mitigation strategies and compared each strategy to the baseline operation to judge the performance and gain insight into the effectiveness. CONDITIONS Queue detected on NB Goyeau Entry Link Detector Y Y Y Y N N N Queue detected on NB Goyeau Mid- block Detector - - - - Y N Y Queue detected on WB Wyandotte RT Turn Lane Mid-block Detector - - - - N Y Y SB Vehicle Phase Call - Y - - - na - N/S Ped Call on West Side N - N Y - na - N/S Ped Call on East Side N N Y Y - na - ACTIONS Omit NB Vehicle Phase â â Extend NB Phase Green â Extend WB Phase Green â Display N/S Ped Walk on West Side (â) â (â) (â) Display N/S Ped Walk on East Side â â (â) (â) Display SB LT Turn Green Arrow â Legend: âYâ Condition is present âââ Action is taken âNâ Condition is not present â(â)â Action is taken if associated condition is present â-â Condition may or may not be present ânaâ Condition is not applicable Operation of traffic signal systems in oversaturated conditions Page 216
Scenario Modeling The baseline scenario for this test case consisted of creating an incident of one hour in duration caused by increased transaction time on the US side of the tunnel during the A.M. peak hour. During the incident, the average processing time for each customs transaction was increased from an average of two minutes to five minutes. This increase in processing time quickly results in traffic spilling back through the tunnel and out into the City streets. The model was simulated in Vissim with all intersections under control of Virtual D4 controller firmware. The toll plaza and customs plaza were approximated in Vissim using the parking lot feature of the simulation system. Using this feature, vehicles that enter the toll plaza search for a âparking lotâ with the most capacity which equates to the toll booth lane with the shortest line. Similarly at the customs plaza, each vehicle chooses the shortest line to be processed through customs. As is the case in the real-world, a subset of vehicles were designated as NEXUS vehicles and allowed to use their own special toll booth and customs plaza lanes. NEXUS is a pre-clearance system for border crossing that reduces the time to cross for very frequent travelers. During the simulation, the volumes were increased according to the time of day schedule shown in Table 43. Table 43. Schedule of volume changes during the scenario The one hour incident begins at 7:30 A.M. (3600s) and continues until 8:30 A.M. (7200s). The processing time for vehicles at the customs plaza is then returned to the original distribution with an average of two minutes. The processing time in the customs plaza equates to a dwell time in the âparking lotâ that represents each inspection station along the vehicleâs route. The resulting input, output, and number of vehicles in the system are shown in Figure 144. Notice in the figure that the output curve is closely overlaid on the input curve up until the incident starts at 3600s. Then the output curve quickly begins to drop while the input rate remains constant, creating significant oversaturation in the network. After the incident is over, the input and output curves come back to closely matching their trajectories when the input volumes drop to 75% of the peak hour volumes. The overflow queues during this time do not dissipate quickly. It is not until the input rates are dropped to 25% of the peak hour flows do the queues dissipate and the system returns to steady state operation. Time of Day (Seconds into Simulation) Volumes 6:30 (0000) - 7:00 (1800) 75% of Peak 7:00 (1800) - 8:30 (7200) Peak Volumes 8:30 (7200) - 9:00 (9000) 75% of Peak 9:00 (9000) - 9:30 (10800) 25% of Peak Operation of traffic signal systems in oversaturated conditions Page 217
Figure 144. Baseline oversaturated scenario A dynamic map was created to illustrate the spatial and temporal aspects of the oversaturated scenario over time. The resulting map is consistent with field observations at this location. This map was utilized to determine critical routes approaching the oversaturated intersection. Approximately 50% of the traffic entering the tunnel approaches from the west and turns north on to Goyeau. The remaining traffic is split, 25% each, between traffic approaching from the south on Goyeau or turning right from westbound Wyandotte. A small amount of NEXUS traffic approaches the tunnel from the north and turns right into the tunnel. During the A.M. and P.M. peaks, only NEXUS traffic is allowed to travel on southbound Goyeau. The eastbound to northbound route is the most critical because it carries the most volume. During the baseline scenario the left-turn bay fills quickly and spills back to block through traffic on westbound Goyeau and intersections upstream as well. Northbound and eastbound queues at Wyandotte extend two intersections upstream as well. The operational objectives that were identified for this test case include throughput maximization and queue management. Because the cause of the oversaturated condition blocks the receiving lane of the north leg of the critical intersection, increased green time for any of the oversaturated movements will not result in increased throughput. For this reason, mitigation strategies were chosen which had the potential for increasing the throughput of phases that do not enter the plaza (eastbound and westbound through movements). For queue management, the focus was on the northbound and eastbound queues which extend to intersections adjacent to the critical intersection. Additional mitigation strategies were envisioned in addition to the logic developed by the Province. Table 44 lists the mitigation strategies that were tested in this scenario. 0 100 200 300 400 500 600 700 800 900 1000 30 0 60 0 90 0 12 00 15 00 18 00 21 00 24 00 27 00 30 00 33 00 36 00 39 00 42 00 45 00 48 00 51 00 54 00 57 00 60 00 63 00 66 00 69 00 72 00 75 00 78 00 81 00 84 00 87 00 90 00 93 00 96 00 99 00 10 20 0 10 50 0 10 80 0 N um be r o f V eh icl es Time (Seconds) Baseline - 1 hour incident Number of Vehicles in System Input Output Operation of traffic signal systems in oversaturated conditions Page 218
Table 44. Summary of mitigation strategies Mitigation Number Name Description 1 Original Mitigation Logic Omit phases and adjust green times at critical intersection 2 Expanded Mitigation Logic Omit phases and adjust green times at critical intersection 3 Dynamic Lane Assignment Allow double left-turns on critical eastbound route 4 Westbound and Northbound Metering Meter traffic upstream of critical intersection on non-critical routes 5 Eastbound Metering Meter traffic upstream of critical intersection on critical route 6 Westbound, Northbound and Eastbound Metering Meter traffic everywhere 7 Re-routing Re-route traffic around critical intersection to alternate entrance to store vehicles on longer route These mitigation strategies are summarized in the next section. Intersection Strategies Two strategies were developed that adjust the operation only at the critical intersection. The first strategy was envisioned by consultants for the City and Province several years before this project (March 1990) began. This logic was implemented and tested on the incident scenario described here. In addition, we developed a more comprehensive set of logical conditions that would take additional actions based on the inputs from additional queue detectors. These two intersection logic approaches are described in the next sections. Original Windsor Incident Mitigation Logic This mitigation strategy used the Congestion Manager in conjunction with the Virtual D4 controller to implement different strategies based on detection of queues at three locations. Other detection points as shown in Figure 145were envisioned to be used for more extensive feedback strategies. Operation of traffic signal systems in oversaturated conditions Page 219
Figure 145. Detection points in the Windsor, ON traffic network From the original responsive logic presented in Table 44, we removed the consideration of pedestrian concerns and developed the following logic table based just on the status of the queue detectors. See Table 45 below. Table 45. Windsor queue-responsive logic Condition Queue detected on northbound Plaza Entry Link (Vissim Detector 1002) Y N N Queue detected on northbound Goyeau Link (Vissim Detectors 1006, 1007) - Y N Queue detected on westbound Right Turn at Goyeau (Vissim Detector 1004) - N Y Action Increase northbound Through Phase â Increase westbound Through Phase â Omit northbound Through Phase â Action Set Plan 2 3 4 Common Destination for all routes Operation of traffic signal systems in oversaturated conditions Page 220
A detection point was created in the Vissim model for each of the locations to be monitored for conditions as shown in Figure 145. For each action included in the table above, a coordination timing plan (Action Set Plan) was created in the Virtual D4 controller to be called by the oversaturation management logic when the triggering condition is met. The logic statements used for this strategy are depicted in Figure 146 a, b, and c. For each logic statement, the condition must be detected as true for one minute before the Action Set Plan is initiated. The Action Set Plan remains in use until the return condition is met for three minutes. We set the begin threshold for each queue detector to 85%. This correlates to a TOSI value of 100% or more. We set the return condition for each queue detector to 75%. This correlates to a TOSI value of approximately 50%. Since the queues grow back quickly and saturate the entire set of eastbound links, once the incident begins, the value of TOSI is 100% for the majority of the hour. SOSI is similarly greater than zero for most of the time at almost all of the detection locations during the simulation. (a) Figure 146. Logic engines for (a) Plan 2 [omit eastbound left turn] (b) Plan 3 [increase northbound through] (c) Plan 4 [increase westbound through] Operation of traffic signal systems in oversaturated conditions Page 221
(b) (c) Figure 146. continued Operation of traffic signal systems in oversaturated conditions Page 222
Expanded Incident Mitigation Logic The original mitigation logic was expanded to include consideration of queue detection on the eastbound approach of the Goyeau/Wyandotte intersection. Additional actions were also designed to increase the throughput of phases that do not enter the border crossing tunnel. In addition, the eastbound right-turn-on-red was eliminated during conditions where the plaza entry link experienced queuing. This action was chosen based on observations of the baseline conditions. In the baseline scenario, the westbound right turning vehicles were consistently turning RTOR during small gaps. This resulted in the entry link for the plaza filling up and then creating starvation for the critical eastbound left-turn movement. Table 46 summarizes the Expanded Windsor Logic and identifies the Vissim detectors associated with each condition. The actions were included in each response plan, and the plan number assigned to the associated group of actions. Table 46. Expanded Windsor Logic Condition Queue detected on northbound Plaza Entry Link (Vissim Detector 1002) Y Y Y Y N N N N N N Queue detected on northbound Goyeau Link (Vissim Detector 1006, 1007) Y Y N N Y N N Y Y N Queue detected on eastbound Left Turn at Goyeau (Vissim Detector 1009) Y N Y N N Y N Y N Y Queue detected on westbound Right Turn at Goyeau (Vissim Detector 1004) - - - - Y Y Y N N N Action Increase eastbound Left Turn Phase â â â Increase northbound Through Phase â â Increase westbound Through Phase â â â â Eliminate right-turn-on-red for westbound Right Turn â â â â Omit northbound Through Phase â Omit eastbound Left Turn Phase â â â Action Set Plan 5 6 7 8 9 10 4 11 3 12 Action Set Plans 3 and 4, developed for the previous responsive strategy, were also used for the expanded logic as the resulting actions were the same. The logic statements used for this strategy are depicted in Figure 147a-j below. Similarly to the logic statements used in the original responsive strategy, the condition must be detected as true for one minute before the Action Set Plan is initiated. The Action Set Plan remains in use until the return condition is met for three Operation of traffic signal systems in oversaturated conditions Page 223
consecutive minutes. We set the âbeginâ threshold for each queue detector to 85%. This correlates to a TOSI value of 100% or more. We set the return condition for each queue detector to 75%. This correlates to a TOSI value of approximately 50%. Since the queues grow back quickly and saturate the entire set of eastbound links, once the incident begins, the value of TOSI is 100% for the majority of the hour. SOSI is similarly greater than zero for most of the time at almost all of the detection locations during the simulation. We could not justify using TOSI or SOSI values for triggering different actions for this particular test case. (a) Figure 147. Logic engines for expanded mitigation logic Operation of traffic signal systems in oversaturated conditions Page 224
(b) (c) Figure 147. continued Operation of traffic signal systems in oversaturated conditions Page 225
(d) (e) Figure 147. continued Operation of traffic signal systems in oversaturated conditions Page 226
(f) (g) Figure 147. continued Operation of traffic signal systems in oversaturated conditions Page 227
(h) (i) Figure 147. continued Operation of traffic signal systems in oversaturated conditions Page 228
(j) Figure 147. continued Dynamic Lane Assignment In this mitigation strategy, the eastbound left turn at Goyeau/Wyandotte was modified to a dual left turn by converting an eastbound through lane to an additional left turn lane. Similarly, the left turn off of Wyandotte into the plaza was also converted to a dual lane movement. This was envisioned to allow additional physical storage for vehicles on the critical route, which might be possible to allow other blocked movements to proceed through the network. The lane assignments were modified by time of day to correspond to the times that the congestion management logic rules would be firing for two queue-responsive strategies. Lane control signs and indications would be necessary to implement such a strategy. Route/Arterial Strategies In addition to the three strategies that were applied at just the critical intersection, four other methods were tested. Since this scenario has a common destination for all routes, it was envisioned that metering is perhaps the only way to allow other vehicles not on the critical routes to be able to move through the network. Westbound and Northbound Metering For this mitigation strategy, the offsets at the intersections east and south of the critical intersection were modified to create a metering effect at the critical intersection. In addition, the green times Operation of traffic signal systems in oversaturated conditions Page 229
on side street movements at these intersections were set to max recall so that they would be serviced regardless of vehicle demand, to provide consistent green time on the main line as well as to attempt to manage queues by minimizing the amount of SOSI > 0 that occurs when the downstream queue has not begun to move, but the upstream light has turned green. Eastbound Metering For this mitigation strategy, offsets at the intersections west of the critical intersection were modified to create a metering effect at the critical intersection. The side street movements at these intersections were set to max recall to provide consistent green time on the main arterial as well as to attempt to manage queues by minimizing the amount of SOSI > 0 that occurs when the downstream queue has not begun to move, but the upstream light turned green. Eastbound, Westbound and Northbound Metering This mitigation strategy was a combination of the previous two metering strategies. Re-Routing To alleviate congestion at the critical intersection and to use storage across a wider area, the original modeled network was expanded to allow re-routing of vehicles. Vehicles that normally enter the border crossing from the west were redirected to enter the plaza from the north. We envisioned that with lane control signals and blank out signs, we could assume that the lane assignments could be changed in a responsive manner after the incident occurs at the tunnel plaza and the queues are detected. By re-routing those vehicles to the southbound approach to the plaza (normally reserved only for vehicles with NEXUS pre-approved v clearance) it was envisioned that this better use of the available network space would âbuy timeâ while the situation on the other side of the river was being resolved (i.e. assigning additional customs agents, capturing the terrorist, etc.). Performance Analysis Five runs for each scenario were conducted with varying common random number seeds. A wide array of performance metrics were retrieved from Vissim and post-processed. The results for the analysis of average delay are presented in Table 47 through Table 52 and Figure 148 through Figure 153. Each movement affected by the incident conditions is listed in the table. Each table reports a comparison of the performance of the mitigation strategy for 15 minute periods of the simulation. Finally, Table 53 and Figure 154 illustrate the results for the total average delay for each link for the entire three hour simulation. Because a significant difference in operation was not observed during the âloadingâ portion of this test case, only the 15 minute periods between 8:00 A.M. and 9:30 A.M. are presented. Each cell of the table is color coded to illustrate the degree of difference between the performance of the baseline scenario and the mitigation strategy. In the Figures and Tables in the remainder of this section, green indicates that the average delay experienced during the mitigation is significantly better (t-test value > 4.4) as compared to the no mitigation scenario. Yellow indicates slightly better (t-test value > 2), white indicated little or no Operation of traffic signal systems in oversaturated conditions Page 230
change, orange indicated slightly worse (t-test value < -2), and red indicates significantly worse (t-test value < -4.4) average delay results. We also present a summary graph indicating the number of links that are represented in each performance category for each mitigation strategy. For each of the figures below, the following number identifiers were used: 1 â Original Mitigation Logic 2 â Expanded Mitigation Logic 3 â Dynamic Lane Assignment 4 â Westbound and Northbound Metering 5 â Eastbound Metering 6 â Westbound, Northbound and Eastbound Metering 7 â Re-routing Operation of traffic signal systems in oversaturated conditions Page 231
Table 47. Average delay per link 8:00 â 8:15 Segment 1 2 3 4 5 6 7 EB TH at PEL-WYAN 86.3 557.6 1714.3 94.2 119.8 635.4 62.3 WB TH at PEL-WYAN 41.2 42.8 36.6 40.2 33.7 38.0 12.8 NB TH at PEL-WYAN 114.4 115.8 384.1 92.4 116.0 199.8 80.3 EB TH at OUEL-WYAN 195.6 567.1 783.6 237.8 259.8 511.6 27.6 SB TH at OUEL-WYAN 100.2 46.2 26.7 48.2 40.1 27.6 16.7 WB TH at OUEL-WYAN 11.0 7.7 14.0 11.4 14.1 28.6 2.5 NB TH at OUEL-WYAN 121.1 129.1 634.6 273.6 157.6 142.5 32.2 NB TH at OUEL-WYAN upstrem of LT bay 54.6 11.4 682.3 257.5 85.2 126.8 35.6 NB LT at OUEL-WYAN 60.5 65.8 50.6 57.1 53.1 41.8 20.7 WB TH at GOY-WYAN 18.4 18.1 93.5 12.7 16.5 19.2 3.6 WB TH at GOY-WYAN upstream of LT bay 596.9 1079.6 2038.1 1008.5 859.8 1478.3 64.0 WB LT at GOY-WYAN 470.3 741.6 1181.4 835.9 740.7 1088.5 69.8 SB TH at GOY-WYAN 78.5 3.3 8.6 3.3 3.5 5.4 4.8 SB TH at GOY-WYAN upstream of LT bay 102.5 286.0 750.0 208.4 119.4 789.2 248.2 SB LT at GOY-WYAN 0.0 0.0 0.0 0.0 0.0 0.3 0.0 Tunnel entrance RT lane 294.1 318.1 371.5 234.7 240.7 240.3 13.4 Tunnle entrance upstream from RT bay 21.2 46.5 152.5 8.5 6.0 32.8 1.1 Just past tunnel entrance on GOY 0.5 0.6 259.1 0.6 0.6 0.5 0.1 WB RT at GOY-WYAN 183.8 200.5 292.5 286.6 232.7 301.9 14.3 WB TH at GOY-WYAN 45.3 68.1 100.0 24.5 31.7 92.0 11.8 WB TH at GOY-WYAN upstream of LT bay 124.1 180.9 303.5 262.4 161.8 275.8 21.8 WB LT at GOY-WYAN 37.3 67.9 27.2 28.6 32.6 46.6 10.1 NB TH at GOY-WYAN 3710.5 1107.5 2023.5 1260.1 1125.3 1393.7 142.0 EB TH at WIND-WYAN 2.0 1.4 2.0 2.5 0.9 6.9 0.5 SB TH at WIND-WYAN 82.7 104.1 105.7 117.2 103.8 127.1 46.5 WB RT at WIND-WYAN 0.8 1.7 3.5 2.8 0.9 2.6 0.4 WB TH at WIND-WYAN 105.7 87.7 117.6 155.2 103.1 130.6 7.5 WB TH at WIND-WYAN upstream of RT bay 151.5 213.4 359.5 266.2 194.5 292.9 22.9 NB TH at WIND-WYAN 85.2 99.4 95.7 143.8 162.1 154.6 48.8 EB TH at MCD-WYAN 9.5 8.7 9.2 7.0 9.5 6.3 2.8 SB TH at MCD-WYAN 48.5 51.2 64.2 48.9 50.7 60.0 19.2 SB TH at MCD-WYAN upstream of RT bay 0.1 0.1 0.1 0.1 0.1 0.1 0.0 WB TH at MCD-WYAN 227.8 662.6 2383.7 913.7 910.8 1198.6 153.3 NB TH at MCD-WYAN 30.7 29.5 33.2 29.8 28.9 30.4 9.8 NB TH at MCD-WYAN upstream of LT bay 69.6 59.7 124.9 44.4 47.6 48.0 40.5 NB LT at MCD-WYAN 139.8 104.7 182.9 109.2 84.3 143.8 41.5 EB TH at GOY-TUC 733.6 110.5 385.3 106.1 90.7 81.8 61.3 SB TH at GOY-TUC 12.1 13.0 11.2 13.1 16.2 13.8 4.0 WB TH at GOY-TUC 1767.6 234.4 393.7 248.7 135.6 143.7 113.9 NB TH at GOY-TUC 1693.7 400.9 1149.5 447.5 391.6 630.4 91.8 EB TH at GOY-PARK 60.9 61.1 65.8 61.1 61.1 60.9 19.1 EB TH at GOY-PARK upstream of LT bay 3.4 3.4 3.4 3.5 3.5 3.4 1.3 EB RT at GOY-PARK 82.0 82.1 80.7 80.7 80.7 80.7 28.3 SB TH at GOY-PARK 13.9 14.3 15.7 14.8 14.7 13.9 21.3 WB TH at GOY-PARK 30.4 30.4 30.4 30.3 30.4 30.4 12.1 NB TH at GOY-PARK 7.7 9.2 7.2 10.0 7.4 7.6 2.7 NB TH at OUEL-PARK 114.2 212.4 129.8 134.9 125.3 161.7 71.1 NB RT at OUEL-PARK 125.8 118.5 129.8 119.6 81.2 81.1 29.2 EB RT at OUEL-PARK 85.6 85.8 85.1 85.6 85.9 86.7 30.9 SB TH at OUEL-PARK 125.7 121.8 126.0 128.4 128.1 126.1 42.5 WB RT at OUEL-PARK 32.6 34.1 31.6 32.4 32.5 32.6 12.2 WB TH at OUEL-PARK 44.9 45.6 43.6 45.2 45.0 45.5 11.1 WB LT at OUEL-PARK 33.4 33.5 32.6 33.4 33.4 33.4 11.1 NB TH at OUEL-UNIV 29.0 27.3 33.6 32.5 30.9 23.2 9.5 NB TH at OUEL-UNIV upstream of LT bay 14.8 13.7 13.0 17.0 16.8 14.6 6.0 NB LT at OUEL-UNIV 36.4 32.3 35.1 33.6 36.3 33.4 11.8 EB TH at OUEL-UNIV 25.6 25.8 22.5 23.5 23.3 26.9 8.7 SB TH at OUEL-UNIV 49.6 50.4 58.5 53.2 51.2 45.5 17.6 SB TH at OUEL-UNIV upstream of LT bay 7.9 7.5 58.5 9.2 8.0 6.7 2.9 SB LT at OUEL-UNIV 53.6 58.8 59.4 54.6 64.2 57.8 20.9 EB TH at GOY-UNIV 11.0 12.2 11.9 12.8 14.3 12.5 11.9 SB TH at GOY-UNIV 68.0 66.9 70.7 68.2 67.5 68.8 35.9 WB TH at GOY-UNIV 11.0 11.9 19.6 11.1 11.0 10.8 6.2 NB TH at GOY-UNIV 14.4 13.8 17.0 17.5 16.6 15.5 5.7 To tunnel From WEST 3425.1 5381.2 7341.6 5009.4 4887.4 5982.1 531.2 Through tunnel 6206.3 6171.4 6127.8 6125.3 6132.5 6170.2 854.5 To tunnel from East 2669.5 3797.0 5536.4 3970.8 3703.5 4188.5 348.8 To tunnel from South 5425.6 3697.6 5044.2 3877.1 3869.5 4129.7 421.3 Operation of traffic signal systems in oversaturated conditions Page 232
Figure 148. Performance summary 8:00 â 8:15 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 1 2 3 4 5 6 7 N um be r o f S eg m en ts Performance Summary of Different Mitigations 8:00 - 8:15 Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 233
Table 48. Average delay per link 8:15 â 8:30 Segment 1 2 3 4 5 6 7 EB TH at PEL-WYAN 1192.3 5199.3 2528.8 5513.5 4123.4 4924.9 1417.5 WB TH at PEL-WYAN 36.1 33.6 27.3 29.1 20.1 18.1 32.3 NB TH at PEL-WYAN 207.9 1411.3 1435.1 515.8 1147.3 1602.3 862.1 EB TH at OUEL-WYAN 692.9 2213.3 1536.1 1988.7 1993.7 2052.1 665.5 SB TH at OUEL-WYAN 41.6 38.0 72.6 141.9 25.8 27.2 29.5 WB TH at OUEL-WYAN 18.1 14.9 7.7 25.0 17.2 15.1 9.9 NB TH at OUEL-WYAN 281.7 865.4 833.1 522.9 733.4 909.3 170.5 NB TH at OUEL-WYAN upstrem of LT bay 100.4 914.5 1661.2 802.2 1172.6 997.4 691.2 NB LT at OUEL-WYAN 38.8 34.7 17.6 25.6 33.7 33.2 36.2 WB TH at GOY-WYAN 2.1 1.0 143.9 2.8 15.9 0.9 11.3 WB TH at GOY-WYAN upstream of LT bay 990.9 2756.7 1925.2 2553.4 2156.4 3215.2 630.0 WB LT at GOY-WYAN 413.0 1002.3 1449.0 1155.6 1084.4 1383.4 677.7 SB TH at GOY-WYAN 95.2 8.7 9.1 10.3 7.6 9.9 8.8 SB TH at GOY-WYAN upstream of LT bay 1100.1 1081.4 568.1 984.2 1207.0 898.8 1087.0 SB LT at GOY-WYAN 318.8 5.6 0.0 0.0 0.7 0.0 0.0 Tunnel entrance RT lane 243.0 315.3 558.8 411.1 289.0 332.3 302.2 Tunnle entrance upstream from RT bay 26.7 8.4 124.0 7.3 45.4 36.8 5.8 Just past tunnel entrance on GOY 0.7 0.5 201.5 0.6 0.9 0.5 0.5 WB RT at GOY-WYAN 155.8 447.3 270.2 499.1 586.1 762.9 301.9 WB TH at GOY-WYAN 16.0 48.9 30.1 96.6 38.4 42.1 94.6 WB TH at GOY-WYAN upstream of LT bay 160.6 710.1 453.5 540.9 810.0 653.3 413.2 WB LT at GOY-WYAN 13.4 30.3 30.1 45.1 22.5 11.9 76.3 NB TH at GOY-WYAN 808.8 1637.6 1785.9 2035.7 2101.0 1723.9 1664.9 EB TH at WIND-WYAN 3.4 5.5 0.4 3.4 2.1 2.2 2.2 SB TH at WIND-WYAN 104.4 609.5 463.9 500.7 382.9 652.3 458.1 WB RT at WIND-WYAN 3.0 2.0 0.3 2.7 1.8 2.0 0.4 WB TH at WIND-WYAN 74.8 408.2 248.9 339.9 429.0 425.6 172.4 WB TH at WIND-WYAN upstream of RT bay 252.7 1256.1 652.8 1121.0 1197.1 816.2 490.1 NB TH at WIND-WYAN 102.4 929.2 425.6 771.3 1333.1 520.5 560.7 EB TH at MCD-WYAN 9.0 8.4 8.8 0.6 8.5 0.4 7.9 SB TH at MCD-WYAN 51.1 119.8 96.9 81.4 71.8 89.9 119.6 SB TH at MCD-WYAN upstream of RT bay 0.1 0.2 0.2 0.2 0.2 0.2 0.4 WB TH at MCD-WYAN 922.1 4042.9 3479.5 3533.9 3804.2 4136.2 3087.7 NB TH at MCD-WYAN 30.3 23.1 22.6 21.0 29.3 19.5 26.8 NB TH at MCD-WYAN upstream of LT bay 56.2 639.7 1099.5 748.6 858.5 619.5 899.1 NB LT at MCD-WYAN 124.4 568.0 656.3 609.3 501.9 479.1 587.6 EB TH at GOY-TUC 2577.3 475.3 2215.5 454.0 456.2 522.0 602.2 SB TH at GOY-TUC 19.7 12.2 18.2 18.2 15.1 23.5 14.2 WB TH at GOY-TUC 3095.3 1510.2 1586.0 1186.4 1029.3 948.1 1261.7 NB TH at GOY-TUC 1567.0 1682.9 2262.9 2340.9 2627.1 1881.6 1803.1 EB TH at GOY-PARK 49.0 52.1 62.3 50.6 51.6 53.0 54.4 EB TH at GOY-PARK upstream of LT bay 82.9 3.9 3.0 72.8 15.3 3.0 3.2 EB RT at GOY-PARK 124.2 96.7 108.1 165.3 107.7 81.9 91.8 SB TH at GOY-PARK 84.1 162.0 183.1 131.5 119.7 336.4 240.2 WB TH at GOY-PARK 61.6 32.5 38.8 32.7 32.6 37.6 38.1 NB TH at GOY-PARK 19.4 9.2 5.4 5.4 10.6 6.5 6.7 NB TH at OUEL-PARK 336.3 134.7 143.6 118.7 127.5 157.2 256.2 NB RT at OUEL-PARK 86.3 66.2 66.6 60.9 92.2 90.0 88.5 EB RT at OUEL-PARK 94.4 93.7 90.1 92.6 93.5 92.1 93.2 SB TH at OUEL-PARK 119.0 122.7 120.0 120.9 119.8 127.7 122.3 WB RT at OUEL-PARK 39.9 38.1 39.9 39.4 39.6 38.9 39.5 WB TH at OUEL-PARK 33.1 32.0 30.8 32.0 31.9 33.2 31.7 WB LT at OUEL-PARK 31.7 30.9 31.9 31.4 31.8 30.9 31.7 NB TH at OUEL-UNIV 19.7 32.3 47.2 33.6 32.0 47.4 34.7 NB TH at OUEL-UNIV upstream of LT bay 11.5 11.0 15.6 13.1 17.2 14.6 14.1 NB LT at OUEL-UNIV 22.4 51.4 51.3 50.5 41.2 46.9 37.6 EB TH at OUEL-UNIV 31.6 25.2 24.3 25.1 24.6 26.8 27.0 SB TH at OUEL-UNIV 49.6 59.0 61.3 57.1 57.3 59.5 59.8 SB TH at OUEL-UNIV upstream of LT bay 8.6 12.1 61.3 12.1 11.5 11.2 11.9 SB LT at OUEL-UNIV 47.3 57.2 52.1 54.7 53.9 67.4 54.1 EB TH at GOY-UNIV 44.3 24.9 89.6 40.2 10.2 123.2 31.8 SB TH at GOY-UNIV 141.2 147.5 314.7 104.2 138.8 280.5 147.3 WB TH at GOY-UNIV 77.5 87.1 116.0 65.1 33.9 127.2 43.6 NB TH at GOY-UNIV 17.0 16.9 20.3 15.7 16.1 11.5 13.6 To tunnel From WEST 3871.0 5692.8 4588.2 6193.8 7176.3 5751.1 3898.8 Through tunnel 4570.9 4522.4 4448.5 4581.6 4690.5 4546.4 4543.6 To tunnel from East 2951.4 5065.2 3478.4 4294.2 4079.5 3245.9 4229.3 To tunnel from South 1949.7 3888.0 4243.8 4823.9 4660.9 4274.8 4434.4 Operation of traffic signal systems in oversaturated conditions Page 234
Figure 149. Performance summary 8:15 â 8:30 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 1 2 3 4 5 6 7 N um be r o f S eg m en ts Performance Summary of Different Mitigations 8:15 - 8:30 Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 235
Table 49. Average delay per link 8:30 â 8:45 Segment 1 2 3 4 5 6 7 EB TH at PEL-WYAN 802.7 4041.5 2418.1 3739.9 3621.0 2969.6 488.6 WB TH at PEL-WYAN 34.8 25.2 22.3 23.3 13.9 17.7 32.4 NB TH at PEL-WYAN 171.1 1346.5 2538.7 1008.8 2067.9 1720.2 585.6 EB TH at OUEL-WYAN 352.3 2241.2 772.0 1815.5 1810.7 1607.7 432.1 SB TH at OUEL-WYAN 31.3 17.1 187.2 102.9 20.7 18.0 50.9 WB TH at OUEL-WYAN 16.4 14.3 16.4 20.5 26.2 20.7 12.1 NB TH at OUEL-WYAN 187.7 761.9 657.1 1235.3 1037.0 1093.9 250.8 NB TH at OUEL-WYAN upstrem of LT bay 72.1 2254.1 1773.6 2281.9 1904.5 2198.4 565.5 NB LT at OUEL-WYAN 38.1 27.7 17.8 26.7 24.9 16.9 52.5 WB TH at GOY-WYAN 0.6 7.3 46.3 1.5 3.5 7.0 16.7 WB TH at GOY-WYAN upstream of LT bay 570.5 2594.0 963.0 1947.9 1776.6 2093.6 473.3 WB LT at GOY-WYAN 238.7 967.0 692.3 810.2 874.6 867.4 549.0 SB TH at GOY-WYAN 93.3 9.2 4.3 9.7 10.1 11.6 15.8 SB TH at GOY-WYAN upstream of LT bay 809.0 658.0 789.8 666.6 558.5 773.4 1184.9 SB LT at GOY-WYAN 311.1 4.0 6.5 6.5 6.8 8.8 0.0 Tunnel entrance RT lane 150.1 164.5 269.8 169.0 232.5 129.9 240.2 Tunnle entrance upstream from RT bay 3.8 10.2 45.7 8.5 25.5 16.9 7.0 Just past tunnel entrance on GOY 0.5 0.7 59.6 0.6 0.6 0.8 0.5 WB RT at GOY-WYAN 55.8 241.5 161.0 228.0 278.5 285.3 274.9 WB TH at GOY-WYAN 25.7 51.8 92.1 48.0 43.6 33.8 60.3 WB TH at GOY-WYAN upstream of LT bay 77.8 367.7 303.2 334.2 484.5 417.5 544.2 WB LT at GOY-WYAN 15.9 38.8 36.1 50.8 22.8 38.9 50.6 NB TH at GOY-WYAN 197.2 1297.2 1145.3 1146.9 1296.6 1112.4 1711.1 EB TH at WIND-WYAN 1.8 5.6 2.0 1.4 2.4 3.0 0.6 SB TH at WIND-WYAN 104.4 246.5 141.0 312.1 468.8 284.7 142.9 WB RT at WIND-WYAN 0.6 0.5 1.0 2.0 3.6 2.4 1.5 WB TH at WIND-WYAN 30.4 188.9 117.8 123.6 250.3 267.5 185.6 WB TH at WIND-WYAN upstream of RT bay 85.1 496.2 378.5 452.6 566.1 564.0 588.2 NB TH at WIND-WYAN 61.5 651.7 382.6 414.8 752.2 417.1 212.3 EB TH at MCD-WYAN 3.2 5.5 8.1 0.1 5.0 0.2 7.8 SB TH at MCD-WYAN 48.5 127.0 90.6 49.3 94.4 45.3 58.0 SB TH at MCD-WYAN upstream of RT bay 0.0 0.2 0.1 0.08 0.1 0.1 0.1 WB TH at MCD-WYAN 368.5 2390.3 1839.5 2204.7 2554.6 2573.1 3075.2 NB TH at MCD-WYAN 35.7 30.9 16.8 17.4 27.4 21.5 29.7 NB TH at MCD-WYAN upstream of LT bay 33.1 892.5 874.1 494.1 431.5 394.3 395.5 NB LT at MCD-WYAN 87.0 417.8 302.5 286.5 448.1 520.0 265.8 EB TH at GOY-TUC 417.5 263.0 1290.0 335.9 336.4 202.0 840.4 SB TH at GOY-TUC 32.3 19.3 14.9 21.0 17.8 29.0 13.5 WB TH at GOY-TUC 470.5 617.5 1066.0 1138.3 919.2 850.9 1549.8 NB TH at GOY-TUC 454.1 1237.6 1120.6 1125.8 1269.2 1160.6 1799.8 EB TH at GOY-PARK 63.9 56.6 56.7 55.2 51.3 52.5 54.3 EB TH at GOY-PARK upstream of LT bay 23.5 3.1 6.2 3.1 9.8 2.4 2.4 EB RT at GOY-PARK 73.4 82.4 86.7 80.5 110.0 105.6 100.1 SB TH at GOY-PARK 50.7 67.6 124.9 74.5 165.3 258.0 295.3 WB TH at GOY-PARK 5.58 5.6 5.6 5.6 5.6 5.3 2.5 NB TH at GOY-PARK 8.1 4.8 6.3 10.6 7.9 8.4 11.3 NB TH at OUEL-PARK 264.5 150.2 173.2 129.4 122.7 160.2 450.3 NB RT at OUEL-PARK 86.2 89.8 78.1 51.3 87.8 57.5 84.4 EB RT at OUEL-PARK 98.2 99.4 98.8 100.1 98.2 96.9 99.4 SB TH at OUEL-PARK 121.8 125.2 126.9 125.4 126.1 128.8 121.4 WB RT at OUEL-PARK 22.0 20.7 22.8 20.8 20.5 20.5 21.7 WB TH at OUEL-PARK 39.1 36.8 35.0 34.6 33.9 34.5 38.7 WB LT at OUEL-PARK 26.4 24.3 25.1 24.3 23.4 23.5 22.3 NB TH at OUEL-UNIV 24.9 42.9 51.9 46.3 42.2 47.2 21.2 NB TH at OUEL-UNIV upstream of LT bay 20.2 12.7 13.0 15.6 11.8 22.8 14.9 NB LT at OUEL-UNIV 17.0 47.6 43.9 51.7 49.6 49.3 32.5 EB TH at OUEL-UNIV 22.4 15.8 15.8 14.8 15.6 26.0 25.1 SB TH at OUEL-UNIV 47.8 55.9 61.5 58.8 53.6 55.2 49.7 SB TH at OUEL-UNIV upstream of LT bay 6.0 5.7 61.5 6.7 7.0 7.3 6.1 SB LT at OUEL-UNIV 55.9 69.8 78.7 77.2 60.7 81.2 68.5 EB TH at GOY-UNIV 41.3 16.3 84.3 43.2 10.6 52.1 117.9 SB TH at GOY-UNIV 130.5 100.7 102.0 185.7 218.9 288.8 312.0 WB TH at GOY-UNIV 47.2 46.5 76.0 37.8 21.4 65.8 35.8 NB TH at GOY-UNIV 17.0 16.8 15.3 13.6 17.7 58.6 11.2 To tunnel From WEST 2763.3 2012.9 4066.2 3673.6 4076.2 3143.5 2189.4 Through tunnel 3187.6 3221.9 3080.7 3136.4 3112.8 3119.6 3085.8 To tunnel from East 1750.8 1860.9 1000.5 1929.4 2707.0 1417.0 3072.1 To tunnel from South 577.4 2623.1 2685.4 2791.7 2755.7 2655.9 3614.7 Operation of traffic signal systems in oversaturated conditions Page 236
Figure 150. Performance summary 8:30 â 8:45 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 1 2 3 4 5 6 7 N um be r o f S eg m en ts Performance Summary of Different Mitigations 8:30 - 8:45 Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 237
Table 50. Average delay per link 8:45 â 9:00 Segment 1 2 3 4 5 6 7 EB TH at PEL-WYAN 457.5 2375.9 2035.2 2647.2 2398.7 2408.4 390.1 WB TH at PEL-WYAN 42.8 26.0 34.7 20.4 22.9 25.4 37.9 NB TH at PEL-WYAN 147.4 1209.0 1474.9 983.1 1486.4 1585.8 496.3 EB TH at OUEL-WYAN 373.9 1209.7 917.5 1475.0 1219.7 1126.9 328.4 SB TH at OUEL-WYAN 37.0 20.4 18.8 21.7 49.6 23.7 45.0 WB TH at OUEL-WYAN 25.3 26.6 19.1 18.4 29.1 18.4 16.8 NB TH at OUEL-WYAN 184.4 703.4 409.7 932.6 642.8 841.2 166.4 NB TH at OUEL-WYAN upstrem of LT bay 18.6 1909.9 1117.3 2127.7 1853.7 2301.2 267.6 NB LT at OUEL-WYAN 48.2 16.9 46.5 29.4 30.3 13.9 49.0 WB TH at GOY-WYAN 26.6 16.3 166.6 21.4 1.7 4.0 4.1 WB TH at GOY-WYAN upstream of LT bay 562.9 1213.5 1099.6 1573.6 1296.9 1360.4 990.2 WB LT at GOY-WYAN 325.2 598.5 863.9 693.3 566.5 559.0 868.0 SB TH at GOY-WYAN 5.3 7.2 8.4 5.2 2.0 10.7 9.7 SB TH at GOY-WYAN upstream of LT bay 131.8 145.9 253.0 245.7 193.2 276.5 583.9 SB LT at GOY-WYAN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Tunnel entrance RT lane 77.0 82.4 217.3 97.9 75.4 92.9 121.4 Tunnle entrance upstream from RT bay 4.3 5.4 54.7 8.0 8.8 9.8 9.6 Just past tunnel entrance on GOY 0.5 0.6 64.1 0.6 0.6 0.6 0.6 WB RT at GOY-WYAN 87.1 137.2 99.5 195.6 148.9 122.4 142.4 WB TH at GOY-WYAN 64.0 116.4 49.8 33.8 48.6 53.7 69.1 WB TH at GOY-WYAN upstream of LT bay 50.8 218.9 154.6 224.3 191.7 147.9 213.4 WB LT at GOY-WYAN 45.4 56.6 39.4 28.2 49.9 34.7 47.7 NB TH at GOY-WYAN 751.2 685.3 842.5 670.6 671.4 692.3 930.9 EB TH at WIND-WYAN 3.9 2.7 1.5 2.8 2.0 3.6 1.4 SB TH at WIND-WYAN 109.5 96.7 95.2 67.2 96.8 64.9 95.4 WB RT at WIND-WYAN 1.0 2.0 4.3 1.1 2.3 2.3 1.1 WB TH at WIND-WYAN 10.8 117.5 56.5 104.2 70.0 60.5 78.6 WB TH at WIND-WYAN upstream of RT bay 34.5 288.2 193.3 304.7 249.6 189.7 266.5 NB TH at WIND-WYAN 87.6 118.5 83.8 177.6 98.6 58.3 95.5 EB TH at MCD-WYAN 4.0 8.5 10.2 0.3 7.5 0.3 7.6 SB TH at MCD-WYAN 49.9 48.0 47.4 32.4 42.5 34.9 47.8 SB TH at MCD-WYAN upstream of RT bay 0.1 0.1 0.1 0.1 0.1 0.1 0.1 WB TH at MCD-WYAN 47.7 1419.7 1048.8 1561.9 1500.7 1288.8 1596.9 NB TH at MCD-WYAN 35.9 27.4 30.5 28.1 30.4 30.2 31.6 NB TH at MCD-WYAN upstream of LT bay 45.8 194.0 145.5 103.8 141.8 70.4 84.1 NB LT at MCD-WYAN 92.8 165.3 79.9 143.9 107.2 61.3 112.7 EB TH at GOY-TUC 736.4 81.1 855.2 63.7 76.8 61.7 169.5 SB TH at GOY-TUC 26.2 24.0 17.9 18.7 16.9 27.2 17.5 WB TH at GOY-TUC 808.0 271.5 743.2 390.5 263.3 359.7 560.9 NB TH at GOY-TUC 865.9 630.2 791.4 569.3 529.7 571.0 846.2 EB TH at GOY-PARK 54.8 54.4 61.0 54.3 54.3 54.3 50.4 EB TH at GOY-PARK upstream of LT bay 2.9 2.9 2.8 2.9 2.9 2.9 7.8 EB RT at GOY-PARK 67.8 69.5 69.6 71.7 69.5 69.5 78.8 SB TH at GOY-PARK 15.1 14.2 14.0 16.4 18.3 27.2 97.5 WB TH at GOY-PARK 11.6 11.6 11.6 11.6 11.6 11.6 11.6 NB TH at GOY-PARK 8.3 8.4 10.4 9.1 9.2 10.4 10.8 NB TH at OUEL-PARK 162.3 167.7 285.7 159.4 157.8 217.4 361.8 NB RT at OUEL-PARK 91.1 82.0 122.4 106.7 83.3 98.2 99.8 EB RT at OUEL-PARK 91.5 91.2 89.7 92.0 95.9 89.6 92.8 SB TH at OUEL-PARK 120.9 122.7 120.4 120.1 118.2 124.3 121.2 WB RT at OUEL-PARK 31.0 29.5 31.2 29.4 29.7 31.4 31.7 WB TH at OUEL-PARK 40.5 38.9 41.2 39.7 39.4 39.6 40.3 WB LT at OUEL-PARK 34.4 33.1 35.3 33.6 34.1 34.3 34.7 NB TH at OUEL-UNIV 32.9 39.9 23.0 39.5 35.6 33.6 31.1 NB TH at OUEL-UNIV upstream of LT bay 12.8 12.4 12.5 13.1 15.1 13.9 43.9 NB LT at OUEL-UNIV 33.5 37.6 23.7 36.2 32.8 39.5 35.7 EB TH at OUEL-UNIV 18.8 18.9 21.9 17.8 18.7 19.7 28.8 SB TH at OUEL-UNIV 58.3 59.5 54.3 56.4 57.5 55.5 55.2 SB TH at OUEL-UNIV upstream of LT bay 5.0 9.1 54.3 6.6 5.7 5.5 6.4 SB LT at OUEL-UNIV 68.4 80.3 76.6 57.4 64.5 80.8 60.7 EB TH at GOY-UNIV 6.2 6.9 6.3 6.5 6.9 7.7 83.2 SB TH at GOY-UNIV 70.2 69.8 73.8 70.8 72.9 84.7 220.8 WB TH at GOY-UNIV 9.3 9.5 15.8 9.7 10.0 11.8 21.7 NB TH at GOY-UNIV 22.9 26.2 20.4 22.7 24.6 22.8 18.0 To tunnel From WEST 2106.1 2495.1 2554.4 2570.4 3476.6 3593.5 2533.4 Through tunnel 2762.0 2744.5 2746.1 2787.8 2706.4 2829.7 2777.7 To tunnel from East 926.4 2280.9 1839.2 1882.3 2472.8 2174.5 2165.1 To tunnel from South 1740.1 1741.0 2121.9 1709.2 1845.2 1877.7 2325.9 Operation of traffic signal systems in oversaturated conditions Page 238
Figure 151. Performance summary 8:45 â 9:00 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 1 2 3 4 5 6 7 N um be r o f S eg m en ts Performance Summary of Different Mitigations 8:45 - 9:00 Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 239
Table 51. Average delay per link 9:00 â 9:15 Segment 1 2 3 4 5 6 7 EB TH at PEL-WYAN 91.0 551.9 784.8 639.1 704.9 723.5 402.5 WB TH at PEL-WYAN 14.5 32.9 33.1 27.5 27.0 24.4 15.6 NB TH at PEL-WYAN 87.7 375.8 481.0 297.4 695.5 590.5 114.3 EB TH at OUEL-WYAN 78.4 339.2 512.6 377.1 460.8 413.4 297.7 SB TH at OUEL-WYAN 30.9 24.9 18.1 18.7 13.8 13.8 25.6 WB TH at OUEL-WYAN 13.8 22.1 16.6 9.0 19.4 13.7 11.7 NB TH at OUEL-WYAN 63.0 236.4 419.5 319.6 248.2 249.7 184.0 NB TH at OUEL-WYAN upstrem of LT bay 0.4 337.4 670.7 576.0 452.4 559.5 138.7 NB LT at OUEL-WYAN 51.2 63.9 50.9 33.7 30.5 38.0 35.4 WB TH at GOY-WYAN 15.8 6.9 46.9 4.4 4.0 1.4 3.2 WB TH at GOY-WYAN upstream of LT bay 145.2 327.6 651.6 407.8 522.2 463.5 573.8 WB LT at GOY-WYAN 153.8 190.2 372.0 186.3 261.5 241.9 372.4 SB TH at GOY-WYAN 0.0 0.2 0.5 0.3 1.3 1.8 1.4 SB TH at GOY-WYAN upstream of LT bay 0.0 0.0 0.0 0.0 2.1 3.5 0.0 SB LT at GOY-WYAN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Tunnel entrance RT lane 31.2 37.1 70.9 34.2 40.3 37.8 58.0 Tunnle entrance upstream from RT bay 3.8 2.8 11.1 3.8 5.8 3.2 4.0 Just past tunnel entrance on GOY 0.5 0.5 13.4 0.4 0.5 0.5 0.5 WB RT at GOY-WYAN 21.0 54.5 32.7 45.6 58.9 50.2 77.5 WB TH at GOY-WYAN 32.4 42.5 27.7 18.7 33.5 19.7 42.8 WB TH at GOY-WYAN upstream of LT bay 5.0 65.8 47.4 40.2 42.1 40.7 112.9 WB LT at GOY-WYAN 26.2 44.0 40.9 17.2 27.1 28.9 30.4 NB TH at GOY-WYAN 159.8 209.5 324.3 208.7 301.5 238.3 356.9 EB TH at WIND-WYAN 1.1 0.9 1.3 1.7 0.8 2.3 0.8 SB TH at WIND-WYAN 89.8 96.8 89.6 62.2 96.9 62.1 89.6 WB RT at WIND-WYAN 0.5 0.6 0.7 5.1 3.1 5.5 0.4 WB TH at WIND-WYAN 3.0 20.4 14.3 36.1 19.9 29.6 52.1 WB TH at WIND-WYAN upstream of RT bay 0.6 58.6 35.4 83.5 37.8 57.1 126.1 NB TH at WIND-WYAN 91.3 77.9 99.4 70.8 77.8 58.9 100.9 EB TH at MCD-WYAN 1.0 1.8 1.6 0.2 2.0 0.2 1.0 SB TH at MCD-WYAN 54.6 52.4 56.6 34.7 58.5 34.9 51.5 SB TH at MCD-WYAN upstream of RT bay 0.0 0.1 0.0 0.0 0.0 0.0 0.0 WB TH at MCD-WYAN 7.8 275.8 165.8 362.6 197.5 357.1 420.2 NB TH at MCD-WYAN 54.7 44.7 58.2 34.0 42.3 34.7 44.0 NB TH at MCD-WYAN upstream of LT bay 9.7 14.3 11.6 7.3 14.9 5.2 12.3 NB LT at MCD-WYAN 63.1 44.7 64.2 51.9 40.7 64.9 53.1 EB TH at GOY-TUC 93.5 73.1 120.6 51.6 72.7 52.0 71.7 SB TH at GOY-TUC 20.6 12.7 11.9 17.3 12.7 19.1 14.1 WB TH at GOY-TUC 91.2 82.5 110.7 55.4 73.9 41.1 152.2 NB TH at GOY-TUC 66.1 45.7 106.6 63.5 58.0 65.1 179.9 EB TH at GOY-PARK 49.5 51.4 74.4 51.4 51.4 51.4 51.4 EB TH at GOY-PARK upstream of LT bay 2.2 2.1 2.1 2.1 2.1 2.1 2.1 EB RT at GOY-PARK 63.4 62.9 62.9 62.9 62.9 62.9 62.9 SB TH at GOY-PARK 8.9 7.4 8.9 7.6 7.7 6.9 8.4 WB TH at GOY-PARK 3.8 3.8 3.8 3.8 3.8 3.8 3.8 NB TH at GOY-PARK 4.2 4.3 4.7 4.9 5.4 6.3 4.6 NB TH at OUEL-PARK 100.2 330.4 225.0 212.2 216.0 258.5 187.3 NB RT at OUEL-PARK 51.8 82.9 72.3 84.2 76.3 82.8 68.7 EB RT at OUEL-PARK 114.0 112.1 113.7 110.0 112.3 118.6 113.7 SB TH at OUEL-PARK 120.6 112.6 113.5 116.4 115.7 120.2 119.2 WB RT at OUEL-PARK 20.6 24.5 24.8 25.4 24.8 27.4 23.0 WB TH at OUEL-PARK 4.2 5.4 6.7 6.1 6.8 5.4 6.4 WB LT at OUEL-PARK 24.6 31.2 28.9 29.7 30.5 29.1 28.2 NB TH at OUEL-UNIV 36.7 11.6 18.8 21.1 15.3 18.3 23.2 NB TH at OUEL-UNIV upstream of LT bay 6.2 10.1 9.6 9.1 10.5 10.7 8.3 NB LT at OUEL-UNIV 24.0 20.2 14.0 15.0 14.6 16.1 15.9 EB TH at OUEL-UNIV 13.2 16.9 15.4 16.8 16.1 19.5 14.4 SB TH at OUEL-UNIV 71.1 52.3 72.1 62.7 54.6 54.4 66.4 SB TH at OUEL-UNIV upstream of LT bay 2.4 2.1 72.1 2.4 2.0 2.1 2.4 SB LT at OUEL-UNIV 40.2 36.2 42.3 41.7 26.9 36.8 41.0 EB TH at GOY-UNIV 3.4 2.2 1.2 2.0 2.5 3.4 3.1 SB TH at GOY-UNIV 67.6 67.5 67.1 67.6 67.6 67.8 66.3 WB TH at GOY-UNIV 6.9 7.0 13.2 7.0 7.2 7.1 7.1 NB TH at GOY-UNIV 24.3 32.3 32.4 32.3 32.3 32.4 32.3 To tunnel From WEST 578.5 862.7 999.1 998.7 1474.0 1254.6 1253.4 Through tunnel 2321.5 2428.4 2201.5 2367.1 2488.4 2406.5 2395.1 To tunnel from East 292.9 611.1 549.7 322.8 439.8 676.0 801.6 To tunnel from South 463.2 593.2 802.0 556.6 682.8 604.2 806.8 Operation of traffic signal systems in oversaturated conditions Page 240
Figure 152. Performance summary 9:00 â 9:15 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 1 2 3 4 5 6 7 N um be r o f S eg m en ts Performance Summary of Different Mitigations 9:00 - 9:15 Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 241
Table 52. Average delay per link 9:15 â 9:30 Segment 1 2 3 4 5 6 7 EB TH at PEL-WYAN 14.6 13.6 10.8 13.8 37.0 28.8 13.6 WB TH at PEL-WYAN 9.4 20.1 13.3 13.9 14.6 7.8 17.3 NB TH at PEL-WYAN 57.2 63.2 157.9 65.7 71.4 55.6 60.1 EB TH at OUEL-WYAN 10.5 9.8 13.8 12.3 49.9 18.7 12.3 SB TH at OUEL-WYAN 29.6 31.5 22.0 30.1 10.8 28.8 27.9 WB TH at OUEL-WYAN 5.1 1.5 6.3 3.3 22.1 0.8 5.9 NB TH at OUEL-WYAN 62.9 55.0 47.9 67.2 31.6 68.3 69.6 NB TH at OUEL-WYAN upstrem of LT bay 0.2 0.2 0.1 0.2 0.1 0.1 0.2 NB LT at OUEL-WYAN 51.3 61.3 37.0 35.4 25.1 42.7 44.8 WB TH at GOY-WYAN 5.4 2.2 58.6 5.1 5.5 5.8 3.4 WB TH at GOY-WYAN upstream of LT bay 0.4 0.3 13.3 12.0 23.8 13.6 10.7 WB LT at GOY-WYAN 90.1 96.7 66.4 90.5 55.2 70.2 94.0 SB TH at GOY-WYAN 0.0 0.7 0.0 0.0 0.0 0.0 0.0 SB TH at GOY-WYAN upstream of LT bay 0.0 1.1 0.0 0.0 0.0 0.0 0.0 SB LT at GOY-WYAN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Tunnel entrance RT lane 3.3 0.6 8.3 1.6 4.4 1.2 1.2 Tunnle entrance upstream from RT bay 1.2 0.2 1.1 0.3 0.3 0.4 0.5 Just past tunnel entrance on GOY 0.1 0.0 0.8 0.1 0.1 0.1 0.1 WB RT at GOY-WYAN 2.6 2.4 2.2 2.4 9.7 4.0 2.8 WB TH at GOY-WYAN 7.1 8.2 10.4 1.8 5.8 2.6 4.5 WB TH at GOY-WYAN upstream of LT bay 0.5 0.4 0.7 0.5 0.5 0.7 0.7 WB LT at GOY-WYAN 7.6 8.4 5.2 5.7 6.6 9.2 3.8 NB TH at GOY-WYAN 95.3 104.1 90.2 86.5 91.8 88.0 99.4 EB TH at WIND-WYAN 1.4 1.1 1.4 3.6 2.5 9.7 1.2 SB TH at WIND-WYAN 78.6 78.6 78.6 56.6 78.7 56.6 78.6 WB RT at WIND-WYAN 1.4 0.7 2.6 0.6 0.7 2.3 1.3 WB TH at WIND-WYAN 3.5 3.3 2.9 28.0 4.4 27.3 3.5 WB TH at WIND-WYAN upstream of RT bay 0.5 0.5 0.5 3.8 0.7 4.0 0.6 NB TH at WIND-WYAN 68.6 68.6 68.6 45.5 68.6 45.5 68.6 EB TH at MCD-WYAN 0.2 0.2 0.2 0.2 0.2 0.2 0.3 SB TH at MCD-WYAN 48.4 48.7 44.0 30.9 48.3 28.8 46.5 SB TH at MCD-WYAN upstream of RT bay 0.0 0.0 0.0 0.0 0.0 0.0 0.0 WB TH at MCD-WYAN 8.1 8.2 7.8 16.4 8.1 16.5 8.3 NB TH at MCD-WYAN 49.9 47.2 56.7 21.5 52.7 23.1 46.3 NB TH at MCD-WYAN upstream of LT bay 13.6 10.8 11.1 6.9 11.8 6.9 10.1 NB LT at MCD-WYAN 45.6 47.2 65.5 18.5 55.3 12.4 51.8 EB TH at GOY-TUC 62.6 85.2 77.1 30.9 84.8 30.2 87.1 SB TH at GOY-TUC 6.0 5.7 3.2 8.9 7.7 7.4 4.7 WB TH at GOY-TUC 45.7 55.6 57.0 46.1 55.6 45.0 47.8 NB TH at GOY-TUC 3.9 4.0 3.5 19.3 3.9 19.3 3.8 EB TH at GOY-PARK 69.3 57.7 60.5 57.7 57.7 57.7 57.7 EB TH at GOY-PARK upstream of LT bay 1.7 1.6 1.6 1.6 1.6 1.6 1.6 EB RT at GOY-PARK 73.7 84.2 84.3 84.2 84.2 84.2 84.2 SB TH at GOY-PARK 8.3 9.4 9.0 8.6 8.1 7.8 8.4 WB TH at GOY-PARK 3.6 3.6 3.6 3.6 3.6 3.6 3.6 NB TH at GOY-PARK 2.6 2.7 2.8 6.3 3.9 4.9 2.7 NB TH at OUEL-PARK 99.6 101.7 88.6 88.3 67.2 90.9 88.4 NB RT at OUEL-PARK 64.4 46.5 54.5 32.6 54.3 60.9 73.8 EB RT at OUEL-PARK 109.0 112.4 115.2 117.3 111.8 115.2 115.4 SB TH at OUEL-PARK 132.6 132.5 135.5 135.2 134.9 131.9 136.5 WB RT at OUEL-PARK 11.7 13.0 10.6 13.1 14.1 12.7 13.5 WB TH at OUEL-PARK 11.7 11.9 11.2 11.1 11.6 10.7 11.5 WB LT at OUEL-PARK 15.3 12.6 12.9 12.1 16.0 12.7 12.9 NB TH at OUEL-UNIV 33.9 42.3 32.3 29.6 27.2 29.7 33.8 NB TH at OUEL-UNIV upstream of LT bay 2.5 3.1 4.6 3.7 3.3 4.5 3.4 NB LT at OUEL-UNIV 18.4 30.5 25.8 18.1 19.9 40.3 15.2 EB TH at OUEL-UNIV 10.8 11.0 11.3 9.9 12.4 11.5 9.7 SB TH at OUEL-UNIV 64.8 64.2 64.4 69.8 70.1 67.6 73.4 SB TH at OUEL-UNIV upstream of LT bay 2.4 2.5 64.4 2.5 2.5 2.5 2.6 SB LT at OUEL-UNIV 21.3 20.9 22.3 21.3 21.4 21.4 21.4 EB TH at GOY-UNIV 6.5 3.8 4.6 4.9 3.6 3.1 5.5 SB TH at GOY-UNIV 69.9 69.7 71.8 70.7 69.7 69.9 69.1 WB TH at GOY-UNIV 3.8 3.9 12.3 3.8 3.8 3.8 3.8 NB TH at GOY-UNIV 11.2 11.5 6.1 11.6 11.6 11.5 11.6 To tunnel From WEST 216.5 178.8 257.9 195.3 215.1 194.4 149.5 Through tunnel 347.8 109.8 84.9 348.4 220.6 125.9 84.2 To tunnel from East 76.3 88.9 66.7 118.8 87.9 104.3 60.2 To tunnel from South 173.3 198.8 193.8 196.3 194.2 194.0 183.0 Operation of traffic signal systems in oversaturated conditions Page 242
Figure 153. Performance summary 9:15 â 9:30 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 1 2 3 4 5 6 7 N um be r o f S eg m en ts Performance Summary of Different Mitigations 9:15 - 9:30 Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 243
Table 53. Average delay per link total (3 hours) Segment 1 2 3 4 5 6 7 EB TH at PEL-WYAN 74.8 169.2 177.2 163.1 161.5 174.0 62.3 WB TH at PEL-WYAN 13.0 12.8 12.4 12.2 11.2 11.1 12.8 NB TH at PEL-WYAN 37.3 136.7 181.7 91.6 142.9 159.4 80.3 EB TH at OUEL-WYAN 37.2 61.4 72.5 60.7 63.0 68.4 27.6 SB TH at OUEL-WYAN 18.4 15.3 19.3 21.7 15.4 8.0 16.7 WB TH at OUEL-WYAN 3.5 3.4 3.4 3.1 3.7 11.9 2.5 NB TH at OUEL-WYAN 33.7 58.5 64.2 63.9 59.3 57.6 32.2 NB TH at OUEL-WYAN upstrem of LT bay 7.5 123.3 149.1 149.7 124.8 146.5 35.6 NB LT at OUEL-WYAN 20.3 20.8 19.6 19.7 19.3 12.5 20.7 WB TH at GOY-WYAN 2.9 3.0 11.5 2.6 2.7 2.2 3.6 WB TH at GOY-WYAN upstream of LT bay 66.7 90.9 131.3 87.2 86.5 95.0 64.0 WB LT at GOY-WYAN 53.0 67.2 84.0 67.3 67.9 68.0 69.8 SB TH at GOY-WYAN 49.1 4.7 4.4 4.6 4.2 4.6 4.8 SB TH at GOY-WYAN upstream of LT bay 158.9 206.9 219.5 179.5 174.4 217.2 248.2 SB LT at GOY-WYAN 276.6 4.6 3.2 6.5 7.5 9.0 0.0 Tunnel entrance RT lane 11.4 11.8 17.3 11.9 12.3 12.1 13.4 Tunnle entrance upstream from RT bay 1.2 1.0 5.8 0.9 1.4 1.3 1.1 Just past tunnel entrance on GOY 0.1 0.1 9.5 0.1 0.1 0.1 0.1 WB RT at GOY-WYAN 9.3 14.3 14.0 14.3 15.2 16.8 14.3 WB TH at GOY-WYAN 8.5 11.0 11.9 8.8 8.9 13.7 11.8 WB TH at GOY-WYAN upstream of LT bay 10.4 21.4 20.7 18.1 20.3 21.1 21.8 WB LT at GOY-WYAN 7.9 11.2 9.5 8.6 9.0 12.5 10.1 NB TH at GOY-WYAN 155.4 129.9 152.1 132.2 136.1 129.8 142.0 EB TH at WIND-WYAN 0.6 0.5 0.4 0.6 0.5 1.3 0.5 SB TH at WIND-WYAN 32.4 56.5 50.3 52.5 54.6 57.1 46.5 WB RT at WIND-WYAN 0.5 0.4 0.6 0.6 0.6 0.8 0.4 WB TH at WIND-WYAN 3.8 7.2 7.6 7.7 7.3 7.9 7.5 WB TH at WIND-WYAN upstream of RT bay 9.1 22.2 23.2 22.0 22.3 22.9 22.9 NB TH at WIND-WYAN 27.2 74.6 47.3 65.3 89.8 52.0 48.8 EB TH at MCD-WYAN 2.6 2.9 3.1 1.8 2.8 1.8 2.8 SB TH at MCD-WYAN 16.5 21.0 19.8 16.8 17.9 17.5 19.2 SB TH at MCD-WYAN upstream of RT bay 0.0 0.0 0.0 0.0 0.0 0.0 0.0 WB TH at MCD-WYAN 47.8 151.6 159.2 156.1 151.8 158.6 153.3 NB TH at MCD-WYAN 10.3 9.7 11.0 9.2 9.9 9.5 9.8 NB TH at MCD-WYAN upstream of LT bay 15.8 52.0 61.2 40.2 45.3 33.6 40.5 NB LT at MCD-WYAN 26.8 44.4 46.1 40.9 40.0 39.3 41.5 EB TH at GOY-TUC 169.0 47.3 153.1 46.1 48.6 45.0 61.3 SB TH at GOY-TUC 5.3 4.1 4.1 4.6 4.1 5.2 4.0 WB TH at GOY-TUC 192.9 91.7 126.2 94.9 83.7 83.9 113.9 NB TH at GOY-TUC 109.1 82.5 96.2 83.4 82.8 82.5 91.8 EB TH at GOY-PARK 19.6 19.3 21.2 19.1 19.0 19.2 19.1 EB TH at GOY-PARK upstream of LT bay 3.5 1.2 1.2 2.0 2.1 1.1 1.3 EB RT at GOY-PARK 28.9 27.6 27.9 28.5 28.7 27.6 28.3 SB TH at GOY-PARK 8.0 9.0 11.5 8.8 9.5 15.2 21.3 WB TH at GOY-PARK 13.7 11.7 12.1 11.7 11.7 12.1 12.1 NB TH at GOY-PARK 2.8 2.6 2.8 2.8 2.8 3.1 2.7 NB TH at OUEL-PARK 55.5 53.4 59.7 45.9 44.5 71.2 71.1 NB RT at OUEL-PARK 31.3 30.6 33.7 31.5 29.9 32.8 29.2 EB RT at OUEL-PARK 31.0 30.9 30.5 31.0 31.0 30.9 30.9 SB TH at OUEL-PARK 41.9 42.0 42.1 42.1 42.0 42.2 42.5 WB RT at OUEL-PARK 12.1 12.1 12.3 12.0 12.0 12.2 12.2 WB TH at OUEL-PARK 11.1 11.0 11.0 11.1 11.0 11.2 11.1 WB LT at OUEL-PARK 11.1 11.1 11.2 11.0 11.0 11.1 11.1 NB TH at OUEL-UNIV 9.6 9.9 9.6 10.1 9.8 8.8 9.5 NB TH at OUEL-UNIV upstream of LT bay 5.1 4.8 4.8 4.9 5.0 4.9 6.0 NB LT at OUEL-UNIV 11.2 11.6 11.6 11.7 11.8 10.6 11.8 EB TH at OUEL-UNIV 8.4 8.2 8.1 8.1 8.0 9.1 8.7 SB TH at OUEL-UNIV 17.4 17.8 19.0 18.0 17.8 17.2 17.6 SB TH at OUEL-UNIV upstream of LT bay 2.9 3.1 19.0 3.1 3.0 2.9 2.9 SB LT at OUEL-UNIV 20.0 21.2 20.8 20.5 21.0 21.2 20.9 EB TH at GOY-UNIV 5.7 4.1 7.3 5.2 3.4 6.9 11.9 SB TH at GOY-UNIV 26.2 25.4 28.8 26.3 26.9 31.9 35.9 WB TH at GOY-UNIV 7.1 7.4 11.2 6.4 5.2 9.5 6.2 NB TH at GOY-UNIV 5.7 5.8 5.4 5.7 5.8 7.1 5.7 To tunnel From WEST 480.5 566.8 689.6 584.2 608.3 601.4 531.2 Through tunnel 850.3 850.4 883.5 852.5 848.5 865.0 854.5 To tunnel from East 260.4 318.2 412.6 343.8 344.5 319.1 348.8 To tunnel from South 412.8 390.9 445.7 398.7 411.0 403.0 421.3 Operation of traffic signal systems in oversaturated conditions Page 244
Figure 154. Performance summary total (3 hours) Once the situation is oversaturated, phase failures are unavoidable and TOSI quickly grows to 100% and higher on every link in the system. A significant number of links experience SOSI > 0, sometimes up to 50% or more of upstream green time is wasted when the downstream queue cannot move. It appears that more extensive mitigations such as the combined metering approaches make the situation worse than doing nothing at all, but most mitigation strategies had a combination of movements that were better and worse and mostly cancelled each other out. Doubling the capacity of the left-turn movement (Strategy 3) is a particularly poor concept that tends to make situations much worse at more than 25 movements, without any corresponding improvements on other movements. Only the re-routing strategy produced significant improvement during the first 45 minutes following the triggering of the incident. After the 8:00-8:15 A.M. period, the initial queue caused by the prolonged processing time clears and improvement is not apparent after this time because all tunnel traffic is being re-routed to enter through the north. Interestingly, the simplest approach envisioned by the Province and the City initially (without any simulation modeling or extensive analysis), Strategy 1, appears to improve conditions on the largest number of movements. This is an important lesson learned for the research. Throughput Analysis The number of vehicles in the system was calculated by comparing the vehicle input and output data recorded by the simulation. The average input rates for each mitigation strategy as well as the Vissim demand input are shown in Figure 155. This figure illustrates that overflow queues on the arterials are inhibiting vehicles to even enter the system at the rate that the model is demanding. 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 1 2 3 4 5 6 7 N um be r o f S eg m en ts Performance Summary of Different Mitigations Total Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 245
As shown in the figure, each mitigation strategy resulted in very similar input rates. Most, but not all, mitigations are able to slightly outperform doing no mitigation at all. As reflected by the average delay performance results presented above, the Windsor Logic (Strategy 1) had the highest input rate of the mitigations during the time of the incident. Recall that the incident begins at time 3600s and extends until 7200s. Due to the length of the tunnel (approximately one mile), it is not for approximately 45 minutes after the incident starts that vehicles are restricted from entering the system. Figure 155. Average input rates under different mitigations The network output graph shown in Figure 156 illustrates that each mitigation strategy impacts the output differently as compared to the baseline no mitigation condition. The Windsor Logic strategy and the Re-Routing strategy have the greatest impact on output. This matches the performance results presented for the average delay by link. 50 100 150 200 250 300 350 400 450 30 0 60 0 90 0 12 00 15 00 18 00 21 00 24 00 27 00 30 00 33 00 36 00 39 00 42 00 45 00 48 00 51 00 54 00 57 00 60 00 63 00 66 00 69 00 72 00 75 00 78 00 81 00 84 00 87 00 90 00 93 00 96 00 99 00 10 20 0 10 50 0 10 80 0 N um be r o f V eh icl es Time (Seconds) Network Input Under Different Strategies Baseline Windsor Logic Expanded Windsor Logic Dynamic Lane Assignment WB and NB Metering EB Metering WB, NB and EB Metering Re-Routing Vissim Demand Input Operation of traffic signal systems in oversaturated conditions Page 246
Figure 156. Average output rates under different mitigations The average number of vehicles in the system was calculated using the input and output data shown above. Figure 157 shows the resulting number of vehicles in the system for each strategy. Figure 157. Average vehicles in the system Determining the best strategy from this presentation of data depends on the desired outcome of the strategy. For example, if the objective is to reduce the number of vehicles stuck in the system, one would choose a strategy which results in a data line which falls below the baseline. However, if the objective is to utilize the storage space within the system, choosing a mitigation which falls above the baseline date would be appropriate. The Dynamic Lane Assignment mitigation strategy 100 150 200 250 300 350 400 450 30 0 60 0 90 0 12 00 15 00 18 00 21 00 24 00 27 00 30 00 33 00 36 00 39 00 42 00 45 00 48 00 51 00 54 00 57 00 60 00 63 00 66 00 69 00 72 00 75 00 78 00 81 00 84 00 87 00 90 00 93 00 96 00 99 00 10 20 0 10 50 0 10 80 0 N um be r o f V eh icl es Time (Seconds) Network Output Under Different Strategies Baseline Windsor Logic Expanded Windsor Logic Dynamic Lane Assignment WB and NB Metering EB Metering WB, NB and EB Metering Re-Routing 0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 30 0 60 0 90 0 12 00 15 00 18 00 21 00 24 00 27 00 30 00 33 00 36 00 39 00 42 00 45 00 48 00 51 00 54 00 57 00 60 00 63 00 66 00 69 00 72 00 75 00 78 00 81 00 84 00 87 00 90 00 93 00 96 00 99 00 10 20 0 10 50 0 10 80 0 N um be r o f V eh icl es Time (Seconds) Number of Vehicles in System under Different Strategies Baseline Windsor Logic Expanded Windsor Logic Dynamic Lane Assignment WB and NB Metering EB Metering WB, NB and EB Metering Re-Routing Operation of traffic signal systems in oversaturated conditions Page 247
appears to result in the highest number of vehicles in the system while still recovering in a time consistent with the no mitigation scenario. Performance improvements for total average delay and throughput do not support that the dynamic left-turn strategy is particularly effective. The Windsor Logic mitigation strategy keeps the number of vehicles in the system lower, and, as a result, appears to recover sooner than any other strategy. This appears to indicate a strong lesson learned from this test case that it is typically better to use a simple solution in an extremely challenging situation. Summary This test case demonstrated the application of the online congestion management tool to oversaturated conditions. A number of logic statements were constructed to select different types of mitigation plans based on the TOSI and SOSI values measured from detectors near the incident location. The first strategy used logic conditions envisioned by the City and the Province that enact actions just at the critical intersection. An extension of this logic was theorized that brought in data from additional detectors and enacted additional actions. A number of other offline mitigation strategies were applied that considered the start time of the incident would be easily able to be identified shortly after the queuing in the tunnel began using existing CCTV resources. Because of the single point of failure condition at the tunnel entrance, the mitigations that were tested had very similar performance and did not show significant improvements to the âdo nothingâ strategy. Some were able to improve total system input and output and reduce the total number of vehicles in the system. Notably, as stated earlier, the original Windsor logic which takes the fewest number of actions possible was the most effective. This validates the principle that in a challenging situation with a single point of failure, taking a minimum number of mitigation actions may be optimal. Since more than 50% of the vehicles were destined for the same location, it was not possible to clear enough storage blocking queues to get other (<50%) vehicles destined for other locations moving enough to offset the delays experienced by the vehicles destined for the tunnel. We had originally envisioned that we might be able to construct a strategy where the vehicles not destined for the tunnel could have been provided greater mobility, but the situation was too challenging. In such a situation, it appears that choosing the smallest number of actions was the most effective. However, it is important to note that we did find that the Windsor mitigation logic was statistically more effective than the baseline âdo nothingâ operation. This shows again, as has been shown in all of the other test cases, that mitigation strategies can be effective in improving system-wide conditions during oversaturation. There is no reason to resort to the aphorism that âthere is nothing that can be done. There is simply too much trafficâ. In the next section, we will present another test case that illustrates that significant improvements to performance are possible with mitigation strategies. In this test, we followed a process based on engineering judgment, with the intended outcome to gather additional evidence and experience with a wide range of potential mitigations. Operation of traffic signal systems in oversaturated conditions Page 248
Test Case: Arterial with Special Event Traffic This real-world test case concerns a heavily traveled arterial that becomes significantly oversaturated on several critical routes. This oversaturation happens intermittently during A.M. and P.M. peak periods due to surges in traffic due to day-to-day variability but also when there are crashes or incidents. In addition, the arterial experiences oversaturation during P.M. peak periods when event traffic is overlaid on the already heavy through flows. In this test case, we studied the application of various mitigation strategies at the five intersections near the event location. Figure 158. Location of test case in the Phoenix, AZ metropolitan area Bell Road is located in the northwest Phoenix, AZ metropolitan area as illustrated in Figure 158. Portions of this roadway carry 70,000 vehicles per day. Because of its location relative to major freeway connections to central Phoenix, there is a pattern of high commuter traffic eastbound in the A.M. peak and westbound in the P.M. peak. The relative P.M. peak travel directional flows are illustrated in Figure 159. Figure 159. Illustration of relative flows along the arterial during P.M. peak Operation of traffic signal systems in oversaturated conditions Page 249
The current signal coordination timing during P.M. peak period consists of a 130s cycle length and offsets which favor forward progression in the westbound direction. The splits, offsets, and corresponding progression pattern are illustrated in Figure 160. The event traffic is destined for Bell Road and Bullard Avenue which is the fourth intersection down from the top of the figure. Figure 160. Progression patterns during the P.M. peak During late February, March, and early April, the commuter traffic along Bell Road is elevated due to additional vehicles traveling to the baseball Spring Training facility for the Texas Rangers and Kansas City Royals located just south of Bell Road. These additional critical routes are illustrated in Figure 161. The result is a significant increase in westbound left turning vehicles at Bullard which causes blocking and starvation for the through phases at the intersections upstream of Bullard to the east since the primary access to this part of the Valley is via the northbound or westbound 101 Loop freeway. During night games, the game traffic is coupled with the normal commuter traffic and the arterial operation breaks down quickly. Some oversaturation is also caused by the increase in traffic heading eastbound towards the facility, but these backups are not nearly as significant as the westbound problem. The forward progression offsets fail to operate efficiently just a few minutes after the game traffic begins to arrive for the start of the game. Queue lengths quickly approach the entire length of the links between Bullard and Sun Village (2600ft), and between Sun Village and Litchfield Road (2600ft). With the baseline 130s plan, forward progression offsets, and split times, it typically takes two to three cycles for vehicles to traverse each of the three links between Litchfield and Operation of traffic signal systems in oversaturated conditions Page 250
Bullard. What is typically a four and a half minute trip along the three mile segment becomes more than 15 minutes; with more than 12 of those minutes spent in queues between Litchfield and Bullard. Figure 161. Critical routes during game overlaid with P.M. peak flows This case study is allocated to the oversaturated scenario taxonomy as shown in Table 54. Table 54. Allocation of Bell Road game traffic case study on the oversaturated scenario taxonomy Stadium Extent Duration Causation Recurrence Symptoms Movement Situational Signal Timing Recurrent Starvation Approach Intermittent Geometrics Non-recurrent Spillback Intersection Persistent Other modes Storage Blocking Route Prolonged Demand Cross Blocking One-way arterial Unplanned Events Two-way arterial Planned Events Interchange Grid Network Operation of traffic signal systems in oversaturated conditions Page 251
This scenario is definitely a two-way arterial problem with several significant critical routes. The queuing lasts for at least an hour and a half as game attendees typically arrive approximately an hour ahead of the first pitch and continue arriving approximately 15 minutes after that. Another 15-20 minutes is required before the queues dissipate and normal traffic operations resume. A dynamic map of the typical queue construction and dissipation process is illustrated in Figure 162 and Figure 163. Figure 162. Queue growth at the beginning of the arriving event traffic START 6:00pm 6:15pm 6:30pm Oversaturation extends further upstream, blocking right turn at upstream intersection Oversaturation on the left turn movement begins Operation of traffic signal systems in oversaturated conditions Page 252
Figure 163. Queue dissipation as the event traffic flows subside This condition is caused by three major factors (a) the event traffic, (b) the existence of a single lane left-turn bay with only 225 ft of storage, and (c) the lack of adequate split time for the left turn. In addition, the common cycle time along the arterial cannot (or was not being) be adjusted because further upstream to the east of the Litchfield intersection is an intersection that is managed by another agency. The situation is recurrent and predictable, since the start time of every Spring Training game is published and known in advance. Attendance at each game is relatively stable as the attendance nears or meets the capacity of the stadium for most night games. Simulation Test Configuration To simplify the test conditions for simulation analysis without onerous data collection, some basic assumptions were made about the game traffic and their origins based on agency and engineer experience with the location and the travel conditions. The baseline condition consisted of increased P.M. peak hour volumes due to event traffic with no change to the P.M. peak coordination timing plan parameters. An additional 7,000 vehicles (approximating that two persons per vehicle attend the event; the stadium holds 14,000) were overlaid on the background vehicles in the P.M. peak period. Of that traffic, 85% was assumed to approach from the west and 15% from the east. The over-all traffic profile was adjusted according to the time of day profile shown in Figure 164. 6:45pm 7:00pm 7:15pm Oversaturation almost fully cleared Oversaturation begins to dissipate upstream Oversaturation extends further upstream, affecting Westbound and Northbound approaches Operation of traffic signal systems in oversaturated conditions Page 253
Figure 164. Traffic arrival volumes and turning percentage profile during game traffic As shown, in addition to the ramp-up and ramp-down of the arrival volumes, the route proportions at Bell and Bullard in the westbound direction were adjusted in the following manner to represent the change in the mix of game and through traffic destinations. The eastbound approach volumes were adjusted to represent the additional game traffic, but the turning percentages were not modified. Five simulation runs were executed for the baseline case and for all mitigation strategies the performance data was averaged over the iterations in the presentation figures. Common random number seeds were used for all five runs for all strategies to reduce the variance effects. The resulting input, output, and number of vehicles in the system are shown in Figure 165. The x-axis units are the number of seconds since the beginning of the simulation time. Figure 165. Number of vehicles in system and I/O rates of baseline scenario 4:00 PM 4:30 PM 5:00 PM 6:00 PM Game Begins 6:30 PM 7:00 PM 50 % 75 % 100 % 75 % 50 % Time Input % Normal day turning % (90/10) * Turning percentage at Bell Road/Bullard Avenue between vehicles turning on to Bullard Avenue and through traffic on Bell Road. Normal day turning % (90/10) RT/LT: 65% TH: 35% * RT/LT: 65% TH: 35% * Game day peak hour turning % (50/50) 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 30 0 60 0 90 0 12 00 15 00 18 00 21 00 24 00 27 00 30 00 33 00 36 00 39 00 42 00 45 00 48 00 51 00 54 00 57 00 60 00 63 00 66 00 69 00 72 00 75 00 78 00 81 00 84 00 87 00 90 00 93 00 96 00 99 00 10 20 0 10 50 0 10 80 0 11 10 0 11 40 0 11 70 0 Nu m be r o f V eh icl es Time (Seconds) Game Day No Mitigation Scenario Number of Vehicles in System Input Output Operation of traffic signal systems in oversaturated conditions Page 254
Notice that until the game traffic begins to arrive (approximately one hour or 3600s into the simulation) the traffic condition is relatively stable. At 75% of the peak hour total volume, the number of the vehicles in the system remains relatively constant, but the queues begin to form at the westbound left-turn movement and quickly spill back into the next two intersections causing SOSI > 0 and subsequent queuing on northbound left turn at Sun Village and right turn blockages at Sun Village and Litchfield as illustrated in the dynamic maps in Figure 162 and Figure 163. The left-turn queue at Bullard also blocks the through movement at Bullard westbound causing starvation. Mitigation Strategy Development The process of choosing mitigation strategies began with addressing the most obvious problem(s) first and working outward to identify other symptoms and potential mitigations. In this test case, we considered only one change to timing plan parameters to implement the mitigation (i.e. we did not expressly consider timing plans for loading, processing, and recovery). Each of the mitigation strategies was implemented by time of day schedule one hour and 30 minutes before the scheduled game time. This start time was selected based on the previous experience which indicated that the most severe increase in volumes begins one hour before the first pitch. An additional 30 minutes was provided to allow the controllers along the arterial to transition and settle into the selected mitigation operation during the time that the volumes were beginning to ramp up at 75% of the peak hour flow rate. In mitigation strategies which include changes to offsets, the offset at Bullard remained constant to avoid transition time at the critical intersection. Five iterations of each mitigation strategy were run and the results were compared to the baseline to determine better and worse conditions in terms of average delay, throughput, and travel time. All simulations used Vissim and the Virtual D4 traffic controller at all intersections in the model. This oversaturated scenario is generated from the inadequacy of the left-turn phase split westbound off of Bell Road at Bullard Avenue toward the Spring Training facility, so the major objective of the mitigations was to solve this problem first. All of the mitigations included some level of green time re-allocation to increase the left-turn split. Additional objectives were to attempt to alleviate the westbound through movement blocking at Bullard, managing westbound queues interactions at Sun Village and Litchfield and managing the eastbound queues developed because of the adjustments to the other parameters to help the westbound movements. Other mitigation strategies were theorized and developed as the results of the previous mitigations were analyzed. The mitigation approaches were developed in the order shown in Table 55, but the purpose of some of the strategy development was done simply to evaluate the differences between the various mitigation strategies. In some cases, one strategy would create queuing or delay in another segment of the arterial so additional strategies were tested which attempted to mitigate the ânewâ congestion as well as the original critical movement. Operation of traffic signal systems in oversaturated conditions Page 255
Table 55. Mitigation strategies evaluated in this test case Mitigation Strategy Description Extreme Left-Turn Split at Bullard WB Left-Turn was significantly increased to address the queue causing blocking. Negative Offsets In addition to the increased left-turn split at Bullard, negative offsets were implemented in the westbound direction to potentially clear downstream residual queue. Simultaneous Offsets In addition to the increased left-turn split at Bullard, simultaneous offsets were implemented in the westbound direction. Double Cycle at Bullard Cycle length at Bullard was reduced to 65 seconds to serve the critical movement more frequently. Resonant Cycle 135 second cycle length along the entire corridor Moderate Left-Turn Split at Bullard Game traffic is also generated from the west. The greatly increased left-turn split westbound at Bullard caused significant backup for the eastbound direction at Bullard. To address this, a more moderately increased left-turn split was tested. Dynamic Lane Assignment In this mitigation strategy, the WB left-turn movement changed from a single turn lane to duals (converting one westbound through lane to an additional left-turn lane) by a time of day schedule. Reduced Cycle Length A cycle length of 90 seconds along the entire corridor was implemented to test if the residual queue lengths could be reduced by reducing the queue at downstream intersections by reducing the available green at upstream intersections. Design of Negative Offsets ⢠130s cycle is used to coordinate with the Arizona Department of Transportation (ADOT) signal to East of evaluated system ⢠Natural spacing is ½ and ¼ mile grid network (90% of all of Phoenix) ⢠Adjust offsets appropriately; provide all additional split (5s) to coordinated movements Dynamic lane assignment ⢠Convert left-most through lane at Bullard to left-turn only lane during game time ⢠Five-section head, blank out sign, upstream DMS sign warning traffic to merge RIGHT if not going to game ⢠Easy in Vissim by modifying signal head model and vehicle route logic by TOD (no mods necessary to Virtual D4 controller) ⢠Illustrated in Figure 166, below Operation of traffic signal systems in oversaturated conditions Page 256
Figure 166. Illustration of dynamic lane allocation for two-lane left-turn movement Average Delay Analysis The results for the average delay analysis are presented in table format per system link for each half hour of the simulation and are color coded to illustrate the degree of the impact on average delay. In Figure 167 through Figure 171and Table 56 through Table 60, green indicates the average delay experienced during the mitigation is significantly better as compared to the no mitigation scenario. Yellow indicates slightly better, white indicates little or no change, orange indicates slightly worse, and red indicates significantly worse than average delay results. The number of links experiencing each level of performance is also included. In addition, Figure 172 and Table 61 illustrate the results for total average delay per system link for the three hour simulation. For each of the figures below, the following number identifiers were used: 2a â Extreme Left-Turn Split at Bullard 3 â Negative Offsets 4 â Simultaneous Offsets 5 â Double Cycle 6 â Resonant Cycle 7 â Moderate Left-Turn Split at Bullard 8 â Dynamic Lane Assignment 9 â Reduced Cycle Length Convert this lane to LT only Operation of traffic signal systems in oversaturated conditions Page 257
The results of the average delay comparison indicated that each mitigation strategy reduces delay on links on the east end of the system while increasing delay on links located on the west end of the system. Table 56. Average delay per link 4:30 â 5:00 Segment 2a 3 4 5 6 7 8 9 EB TH at Reems-Bell 21.8 21.6 22.1 22.2 20.4 22.0 22.1 20.0 EB TH at Reems-Bell Upstream of LT Bay 1.3 1.3 1.4 1.4 1.2 1.4 1.4 1.3 NB RT at Reems-Bell 8.9 8.2 8.3 8.2 8.6 8.5 8.4 8.6 NB TH at Reems-Bell Upstream of RT Bay 0.9 0.9 0.9 0.9 1.0 0.9 0.9 0.9 SB LT at Reems-Bell 38.0 42.1 41.7 41.7 43.2 41.7 41.7 32.7 SB TH at Reems-Bell Upstream of LT Bay 1.8 2.6 2.6 2.6 2.6 2.6 2.6 1.3 EB TH at Parkview-Bell 4.2 5.3 13.3 16.2 12.1 12.9 13.0 14.4 EB TH at Parkview-Bell Upstream of LT Bay 2.2 3.4 4.6 9.1 3.7 4.4 4.2 5.6 NB RT at Parkview-Bell 3.8 4.0 3.6 4.4 3.6 4.1 3.7 3.6 NB TH at Parkview-Bell Upstream of LT Bay 0.2 0.2 0.2 0.3 0.2 0.2 0.2 0.2 SB LT at Parkview-Bell 44.0 39.2 41.5 39.4 43.3 40.1 41.3 21.9 SB TH at Parkview-Bell Upstream of LT Bay 5.9 3.1 3.3 3.3 3.1 3.4 3.4 1.1 EB TH at Bullard-Bell 15.8 23.1 24.5 21.1 24.9 15.2 16.0 11.6 EB RT at Bullard-Bell 12.3 18.5 17.3 22.7 17.9 11.9 12.0 13.1 EB TH at Bullard-Bell Upstream of RT Bay 3.9 5.3 6.6 17.3 7.6 2.8 3.0 3.6 WB LT at Bullard-Bell 16.8 13.5 14.1 15.4 12.3 19.4 19.1 15.8 WB TH at Bullard-Bell 0.9 1.2 3.8 2.7 4.0 3.4 3.7 2.9 WB TH at Bullard-Bell Upstream of LT Bay 13.1 8.7 12.1 12.6 8.6 35.1 32.4 7.8 NB LT at Sun Village-Bell 60.2 55.4 53.6 60.9 64.8 56.6 59.3 39.5 NB TH at Sun Village- Bell Upstream of LT Bay 1.3 1.4 1.4 1.4 1.4 1.4 1.4 1.4 WB TH at Sun Village-Bell 0.8 3.0 2.1 2.0 1.9 3.7 3.6 5.9 WB TH at Sun Village- Bell Upstream of LT Bay 1.6 7.7 1.8 1.7 1.7 3.5 3.3 5.2 SB TH/RT at Sun Village- Bell 0.1 1.9 1.1 0.9 2.4 1.1 1.1 2.0 SB TH at Sun Village-Bell Upstream of LT Bay 1.8 1.8 1.8 1.8 1.8 1.8 1.8 1.8 NB LT at Litchfield-Bell 59.3 58.0 58.8 60.1 62.2 58.8 56.8 48.2 NB TH at Litchfield-Bell Upstream of LT Bay 0.7 0.7 0.7 0.7 0.8 0.7 0.7 0.6 WB TH at Litchfield-Bell 19.1 19.0 18.0 18.3 18.2 17.8 17.9 26.1 WB TH at Litchfield-Bell Upstream of LT Bay 15.1 17.8 13.7 14.4 14.4 13.1 13.6 68.4 SB TH at Litchfield-Bell 45.2 43.5 48.2 47.6 51.8 48.2 48.2 31.4 SB TH/RT at Litchfield- Bell Upstream of LT Bay 1.1 1.1 1.1 1.2 1.2 1.2 1.2 0.9 Operation of traffic signal systems in oversaturated conditions Page 258
Figure 167. Performance summary 4:30 â 5:00 0 2 4 6 8 10 12 14 16 18 2a 3 4 5 6 7 8 9 N um be r o f S eg m en ts Performance Summary of Different Mitigations 4:30 PM to 5:00 PM Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 259
Table 57. Average delay per link 5:00 â 5:30 Segment 2a 3 4 5 6 7 8 9 EB TH at Reems-Bell 23.1 22.6 23.3 27.6 23.0 22.0 22.3 23.9 EB TH at Reems-Bell Upstream of LT Bay 3.2 2.9 3.0 4.9 3.4 2.9 3.0 2.9 NB RT at Reems-Bell 15.9 16.1 16.5 41.6 22.2 13.3 13.8 15.7 NB TH at Reems-Bell Upstream of RT Bay 12.2 8.5 9.0 19.2 16.1 9.0 11.5 5.0 SB LT at Reems-Bell 54.2 54.0 54.4 54.1 59.7 54.6 54.6 34.0 SB TH at Reems-Bell Upstream of LT Bay 27.4 33.3 33.6 31.1 49.2 36.0 33.3 4.1 EB TH at Parkview-Bell 17.3 17.2 24.1 29.8 25.4 21.2 22.6 22.0 EB TH at Parkview-Bell Upstream of LT Bay 67.4 71.7 74.6 115.9 82.2 59.6 66.4 75.1 NB RT at Parkview-Bell 5.4 5.3 6.2 5.3 5.5 7.4 6.1 5.5 NB TH at Parkview-Bell Upstream of LT Bay 0.4 0.4 0.5 0.3 0.3 0.4 0.3 0.3 SB LT at Parkview-Bell 59.6 52.3 54.3 62.5 60.9 50.6 52.5 24.8 SB TH at Parkview-Bell Upstream of LT Bay 41.4 19.3 26.1 41.6 55.4 21.5 18.8 1.9 EB TH at Bullard-Bell 18.4 42.3 23.8 25.2 30.3 13.0 12.9 13.8 EB RT at Bullard-Bell 22.7 23.7 23.3 29.6 26.7 18.5 21.2 22.4 EB TH at Bullard-Bell Upstream of RT Bay 40.1 40.3 40.9 61.1 47.9 30.5 35.5 41.2 WB LT at Bullard-Bell 19.9 16.5 18.2 16.1 14.6 25.7 25.4 21.9 WB TH at Bullard-Bell 2.3 2.1 2.9 3.0 3.1 2.9 3.0 3.0 WB TH at Bullard-Bell Upstream of LT Bay 104.0 86.0 93.9 94.8 72.4 183.4 99.2 80.8 NB LT at Sun Village-Bell 164.8 70.5 145.6 130.2 153.2 134.1 118.5 49.4 NB TH at Sun Village- Bell Upstream of LT Bay 5.0 1.5 7.6 2.0 5.2 13.7 2.1 1.5 WB TH at Sun Village-Bell 13.0 13.1 13.5 13.2 11.0 25.5 16.4 14.5 WB TH at Sun Village- Bell Upstream of LT Bay 71.5 67.9 71.6 67.4 59.5 127.7 99.0 62.3 SB TH/RT at Sun Village- Bell 3.8 4.3 2.7 2.3 3.9 2.7 2.8 3.3 SB TH at Sun Village-Bell Upstream of LT Bay 1.8 1.8 1.9 1.8 1.8 1.9 1.9 1.8 NB LT at Litchfield-Bell 112.1 90.8 140.1 102.3 144.3 187.6 178.5 80.5 NB TH at Litchfield-Bell Upstream of LT Bay 18.2 7.3 41.2 10.8 39.2 111.8 94.6 6.6 WB TH at Litchfield-Bell 30.0 28.0 29.6 28.8 25.9 38.2 35.4 35.9 WB TH at Litchfield-Bell Upstream of LT Bay 181.9 171.8 177.6 169.9 165.6 212.3 229.6 229.6 SB TH at Litchfield-Bell 68.0 70.3 74.8 80.4 80.2 78.4 79.7 38.7 SB TH at Litchfield-Bell Upstream of LT Bay 13.3 16.1 22.5 18.7 18.8 27.1 22.5 1.8 Operation of traffic signal systems in oversaturated conditions Page 260
Figure 168. Performance summary 5:00 â 5:30 0 2 4 6 8 10 12 14 16 18 2a 3 4 5 6 7 8 9 N um be r o f S eg m en ts Performance Summary of Different Mitigations 5:00 PM to 5:30 PM Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 261
Table 58. Average delay per link 5:30 â 6:00 Segment 2a 3 4 5 6 7 8 9 EB TH at Reems-Bell 50.3 50.4 49.9 83.1 63.5 32.3 43.5 64.6 EB TH at Reems-Bell Upstream of LT Bay 42.7 38.2 37.1 167.3 78.8 6.4 23.4 72.6 NB RT at Reems-Bell 129.3 196.9 277.5 307.4 288.0 98.3 216.3 223.4 NB TH at Reems-Bell Upstream of RT Bay 137.3 203.4 224.1 417.9 326.8 74.0 117.7 136.6 SB LT at Reems-Bell 59.5 54.0 52.6 71.1 57.0 55.1 55.9 42.3 SB TH at Reems-Bell Upstream of LT Bay 109.6 87.1 116.6 120.1 146.5 135.2 129.0 11.2 EB TH at Parkview-Bell 41.3 40.3 40.0 63.9 49.8 29.3 35.7 45.2 EB TH at Parkview-Bell Upstream of LT Bay 366.0 359.0 352.5 586.6 445.9 229.6 301.4 412.8 NB RT at Parkview-Bell 5.6 5.1 6.1 5.8 6.1 5.4 5.7 6.0 NB TH at Parkview-Bell Upstream of LT Bay 0.5 1.6 0.6 0.8 0.5 0.5 0.8 0.4 SB LT at Parkview-Bell 67.3 61.6 69.1 87.5 83.9 51.7 64.8 29.0 SB TH at Parkview-Bell Upstream of LT Bay 41.4 30.5 44.5 118.1 108.7 13.6 34.4 2.1 EB TH at Bullard-Bell 29.4 33.6 28.2 23.6 38.8 12.6 13.6 16.5 EB RT at Bullard-Bell 23.6 23.9 24.2 29.7 26.6 19.4 23.4 23.9 EB TH at Bullard-Bell Upstream of RT Bay 118.6 116.1 116.4 186.6 148.9 73.1 103.8 122.0 WB LT at Bullard-Bell 16.7 16.3 17.2 14.7 13.6 28.1 24.6 18.9 WB TH at Bullard-Bell 3.0 3.3 2.6 3.5 2.7 2.1 2.7 3.7 WB TH at Bullard-Bell Upstream of LT Bay 276.5 259.0 260.5 239.4 208.0 442.5 164.3 279.3 NB LT at Sun Village-Bell 355.2 194.4 295.3 255.4 361.0 177.2 179.7 135.2 NB TH at Sun Village- Bell Upstream of LT Bay 215.7 25.8 114.9 66.3 199.4 69.5 13.5 8.5 WB TH at Sun Village-Bell 34.6 33.1 32.3 28.2 23.8 60.3 21.3 33.9 WB TH at Sun Village- Bell Upstream of LT Bay 230.9 213.6 208.4 190.8 166.5 425.0 167.7 210.4 SB TH/RT at Sun Village- Bell 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 SB TH at Sun Village-Bell Upstream of LT Bay 1.9 1.9 1.9 1.9 2.1 2.0 1.9 1.8 NB LT at Litchfield-Bell 298.0 205.0 263.9 221.0 291.4 403.7 255.1 189.6 NB TH at Litchfield-Bell Upstream of LT Bay 426.3 220.2 508.0 302.3 505.6 763.0 628.0 224.7 WB TH at Litchfield-Bell 53.1 51.0 52.2 48.2 40.1 76.0 48.1 57.9 WB TH at Litchfield-Bell Upstream of LT Bay 364.2 339.1 352.4 333.7 281.8 497.2 401.5 394.0 SB TH at Litchfield-Bell 82.6 93.7 103.1 92.7 96.0 115.9 94.8 37.3 SB TH at Litchfield-Bell Upstream of LT Bay 35.6 58.6 111.8 30.1 28.9 109.1 38.1 1.8 Operation of traffic signal systems in oversaturated conditions Page 262
Figure 169. Performance Summary 5:30 â 6:00 0 2 4 6 8 10 12 14 16 18 2a 3 4 5 6 7 8 9 N um be r o f S eg m en ts Performance Summary of Different Mitigations 5:30 PM to 6:00 PM Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 263
Table 59. Average delay per link 6:00 â 6:30 Segment 2a 3 4 5 6 7 8 9 EB TH at Reems-Bell 89.3 80.9 93.3 116.6 99.8 29.9 73.7 85.6 EB TH at Reems-Bell Upstream of LT Bay 297.4 239.0 247.4 739.4 517.8 2.6 113.0 408.5 NB RT at Reems-Bell 601.2 569.3 651.0 792.2 672.6 218.5 426.9 597.7 NB TH at Reems-Bell Upstream of RT Bay 999.2 1194.8 1213.2 1510.3 1346.3 390.9 953.3 1043.4 SB LT at Reems-Bell 66.0 65.9 60.4 81.5 72.3 40.5 51.9 51.3 SB TH at Reems-Bell Upstream of LT Bay 111.4 83.1 81.9 353.1 210.9 44.2 66.8 33.2 EB TH at Parkview-Bell 48.0 46.3 51.4 71.4 56.8 29.5 43.7 49.0 EB TH at Parkview-Bell Upstream of LT Bay 745.9 714.9 761.6 1038.0 839.9 424.0 652.3 738.8 NB RT at Parkview-Bell 4.2 3.8 4.3 4.2 4.5 4.9 4.4 3.8 NB TH at Parkview-Bell Upstream of LT Bay 0.2 3.4 0.3 0.2 0.2 0.2 0.8 0.3 SB LT at Parkview-Bell 72.9 76.1 80.7 99.5 76.7 51.8 72.2 29.9 SB TH at Parkview-Bell Upstream of LT Bay 38.4 43.8 112.5 228.3 140.0 7.6 76.9 1.4 EB TH at Bullard-Bell 27.5 31.1 30.6 24.5 43.5 13.6 16.0 16.6 EB RT at Bullard-Bell 24.0 23.9 23.8 29.6 27.5 19.3 24.0 24.0 EB TH at Bullard-Bell Upstream of RT Bay 133.3 139.7 151.6 199.1 165.2 87.0 126.3 133.6 WB LT at Bullard-Bell 16.3 15.3 15.9 14.6 13.1 26.2 19.5 17.8 WB TH at Bullard-Bell 2.1 2.3 2.3 3.6 2.8 2.2 2.5 2.8 WB TH at Bullard-Bell Upstream of LT Bay 166.7 168.7 174.3 168.6 150.8 312.4 119.4 189.6 NB LT at Sun Village-Bell 316.6 131.5 272.9 222.3 315.8 125.6 102.0 63.1 NB TH at Sun Village- Bell Upstream of LT Bay 219.8 2.3 105.6 53.7 350.2 22.2 1.6 1.4 WB TH at Sun Village-Bell 19.4 20.5 21.1 20.7 18.2 40.1 15.7 21.6 WB TH at Sun Village- Bell Upstream of LT Bay 164.6 169.7 165.5 163.9 140.7 372.8 133.9 175.9 SB TH/RT at Sun Village- Bell 6.2 4.7 6.7 6.1 9.6 6.6 6.6 3.4 SB TH at Sun Village-Bell Upstream of LT Bay 1.8 1.8 1.9 1.8 2.4 1.9 1.8 1.8 NB LT at Litchfield-Bell 210.1 162.1 233.5 185.3 235.6 355.4 221.9 169.4 NB TH at Litchfield-Bell Upstream of LT Bay 729.9 473.7 810.6 600.3 839.9 1058.8 788.1 520.7 WB TH at Litchfield-Bell 37.4 37.6 38.0 38.9 33.1 68.2 37.9 46.7 WB TH at Litchfield-Bell Upstream of LT Bay 271.1 273.4 273.1 281.7 220.5 569.4 306.0 340.5 SB TH at Litchfield-Bell 62.8 80.7 77.2 52.5 58.8 93.3 69.8 33.0 SB TH at Litchfield-Bell Upstream of LT Bay 31.7 65.8 165.0 1.3 2.5 182.1 42.1 0.8 Operation of traffic signal systems in oversaturated conditions Page 264
Figure 170. Performance summary 6:00 â 6:30 0 2 4 6 8 10 12 14 16 18 2a 3 4 5 6 7 8 9 N um be r o f S eg m en ts Performance Summary of Different Mitigations 6:00 PM to 6:30 PM Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 265
Table 60. Average delay per link 6:30 â 7:00 Segment 2a 3 4 5 6 7 8 9 EB TH at Reems-Bell 26.8 23.4 26.1 52.7 39.3 19.5 20.5 27.8 EB TH at Reems-Bell Upstream of LT Bay 21.8 2.6 19.5 581.7 192.0 0.5 0.9 42.3 NB RT at Reems-Bell 73.2 34.1 56.1 386.1 153.1 7.1 17.8 82.6 NB TH at Reems-Bell Upstream of RT Bay 652.1 591.4 597.7 1822.7 1122.1 2.2 279.9 733.4 SB LT at Reems-Bell 39.7 40.8 39.3 44.2 45.4 39.4 40.5 25.5 SB TH at Reems-Bell Upstream of LT Bay 3.9 1.6 1.8 43.5 2.8 0.6 0.7 0.5 EB TH at Parkview-Bell 13.6 13.1 15.5 23.3 17.2 12.4 13.2 20.7 EB TH at Parkview-Bell Upstream of LT Bay 273.7 269.4 280.9 545.0 366.2 57.1 192.9 316.0 NB RT at Parkview-Bell 3.3 4.2 3.8 4.6 3.7 4.3 3.6 3.1 NB TH at Parkview-Bell Upstream of LT Bay 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 SB LT at Parkview-Bell 39.3 35.5 35.4 35.1 40.9 32.6 34.4 23.2 SB TH at Parkview-Bell Upstream of LT Bay 3.6 0.6 0.8 1.9 0.9 0.6 0.6 0.5 EB TH at Bullard-Bell 9.9 11.9 9.9 8.6 5.4 10.8 6.3 7.7 EB RT at Bullard-Bell 12.7 11.8 12.8 17.8 12.6 9.2 11.2 15.3 EB TH at Bullard-Bell Upstream of RT Bay 26.0 29.0 27.7 52.4 31.4 8.7 17.8 28.6 WB LT at Bullard-Bell 11.3 7.5 11.1 12.5 12.4 15.4 12.4 13.3 WB TH at Bullard-Bell 1.0 1.0 2.4 1.8 2.2 2.4 2.3 4.0 WB TH at Bullard-Bell Upstream of LT Bay 13.9 12.2 14.4 8.3 3.8 79.4 8.4 23.2 NB LT at Sun Village-Bell 62.2 43.4 73.9 51.4 54.5 62.5 53.0 36.8 NB TH at Sun Village- Bell Upstream of LT Bay 1.2 1.2 1.3 1.2 8.7 1.3 1.2 1.2 WB TH at Sun Village-Bell 1.7 5.2 2.0 1.6 0.9 8.0 1.3 6.4 WB TH at Sun Village- Bell Upstream of LT Bay 8.8 11.0 6.4 2.3 1.0 73.5 1.2 11.1 SB TH/RT at Sun Village- Bell 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 SB TH at Sun Village-Bell Upstream of LT Bay 1.8 1.8 1.8 1.8 1.8 1.7 1.7 1.8 NB LT at Litchfield-Bell 63.2 62.7 64.0 66.1 66.1 105.9 68.9 44.2 NB TH at Litchfield-Bell Upstream of LT Bay 23.5 2.0 15.6 4.2 11.7 281.3 11.5 1.3 WB TH at Litchfield-Bell 18.7 18.6 17.8 18.7 17.4 22.6 18.1 20.5 WB TH at Litchfield-Bell Upstream of LT Bay 5.7 3.3 6.1 7.0 3.2 115.5 4.0 17.0 SB TH at Litchfield-Bell 42.0 43.7 61.1 40.9 42.3 66.0 44.8 24.3 SB TH at Litchfield-Bell Upstream of LT Bay 0.6 3.0 72.7 0.6 0.6 127.5 0.6 0.5 Operation of traffic signal systems in oversaturated conditions Page 266
Figure 171. Performance summary 6:30 â 7:00 0 2 4 6 8 10 12 14 16 18 2a 3 4 5 6 7 8 9 N um be r o f S eg m en ts Performance Summary of Different Mitigations 6:30 PM to 7:00 PM Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 267
Table 61. Average delay per link (3 hour total) Segment 2a 3 4 5 6 7 8 9 EB TH at Reems-Bell 37.7 36.3 38.1 48.6 42.1 24.7 33.6 39.9 EB TH at Reems-Bell Upstream of LT Bay 59.3 46.9 49.0 231.2 125.1 2.9 24.0 87.3 NB RT at Reems-Bell 64.5 65.2 66.3 97.9 79.7 40.6 55.5 67.1 NB TH at Reems-Bell Upstream of RT Bay 249.3 267.5 273.6 435.7 352.7 82.1 200.7 255.3 SB LT at Reems-Bell 50.3 49.7 48.9 54.1 52.9 46.2 48.2 37.5 SB TH at Reems-Bell Upstream of LT Bay 48.1 40.4 46.3 92.8 76.2 45.9 47.2 9.0 EB TH at Parkview-Bell 20.5 20.3 24.1 29.8 25.3 19.2 22.0 24.9 EB TH at Parkview-Bell Upstream of LT Bay 250.5 244.9 251.6 347.9 287.8 146.0 216.4 264.0 NB RT at Parkview-Bell 4.7 4.7 5.1 5.0 5.0 5.6 5.0 4.8 NB TH at Parkview-Bell Upstream of LT Bay 0.3 1.3 0.4 0.4 0.3 0.3 0.5 0.3 SB LT at Parkview-Bell 54.5 53.2 54.9 61.4 59.2 46.2 52.7 27.7 SB TH at Parkview-Bell Upstream of LT Bay 26.9 19.9 34.6 75.2 60.3 10.2 25.6 1.5 EB TH at Bullard-Bell 15.9 23.2 18.3 15.6 20.9 11.1 10.6 10.8 EB RT at Bullard-Bell 19.5 20.8 20.8 24.5 21.9 17.1 19.3 20.1 EB TH at Bullard-Bell Upstream of RT Bay 52.3 54.6 56.0 74.2 61.4 37.6 48.3 53.3 WB LT at Bullard-Bell 16.3 14.6 15.5 14.7 12.9 22.6 20.4 17.8 WB TH at Bullard-Bell 1.4 1.6 2.5 2.5 2.6 2.3 2.4 2.8 WB TH at Bullard-Bell Upstream of LT Bay 86.4 82.3 83.7 81.1 70.3 138.7 65.3 86.8 NB LT at Sun Village-Bell 186.4 107.0 166.2 146.6 181.8 116.7 111.2 73.7 NB TH at Sun Village- Bell Upstream of LT Bay 89.0 7.4 44.7 25.3 103.0 22.3 4.2 2.9 WB TH at Sun Village-Bell 10.4 11.6 10.8 10.2 9.0 17.5 9.0 12.7 WB TH at Sun Village- Bell Upstream of LT Bay 71.6 72.1 68.1 65.3 58.8 127.2 59.2 68.7 SB TH/RT at Sun Village- Bell 57.0 52.3 50.6 47.4 73.8 50.4 50.7 49.3 SB TH at Sun Village-Bell Upstream of LT Bay 1.8 1.8 1.8 1.8 2.0 1.9 1.8 1.8 NB LT at Litchfield-Bell 115.7 103.0 119.7 109.7 122.0 148.4 124.8 94.6 NB TH at Litchfield-Bell Upstream of LT Bay 174.9 115.3 201.2 143.9 199.5 298.5 232.6 123.3 WB TH at Litchfield-Bell 27.1 26.7 26.8 26.6 24.4 32.7 26.6 31.6 WB TH at Litchfield-Bell Upstream of LT Bay 124.1 121.4 122.9 122.4 109.4 177.8 134.5 157.1 SB TH at Litchfield-Bell 61.8 67.4 71.2 64.5 67.2 78.4 68.1 34.7 SB TH at Litchfield-Bell Upstream of LT Bay 16.4 27.8 63.6 11.6 11.7 73.4 20.4 1.2 Operation of traffic signal systems in oversaturated conditions Page 268
Figure 172. Performance summary (3 hour total) As shown in Figure 167 through Figure 171, each of the mitigation scenarios performed similarly in terms of average delay per link. Because the logical first mitigation strategy that would be implemented by a traffic engineer facing this problem would be to extend the westbound left-turn phase at Bullard, the results of each mitigation strategy was also compared to the results of the extended left-turn split strategy. The results of this comparison are shown in Figure 173 and Table 62. 0 2 4 6 8 10 12 14 16 18 2a 3 4 5 6 7 8 9 N um be r o f S eg m en ts Performance Summary of Different Mitigations 4:00 PM to 7:00 PM Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 269
Table 62. Average delay comparison with extended left turn at Bullard (3 hour total) Segment 3 4 5 6 7 8 9 EB TH at Reems-Bell 36.3 38.1 48.6 42.1 24.7 33.6 39.9 EB TH at Reems-Bell Upstream of LT Bay 46.9 49.0 231.2 125.1 2.9 24.0 87.3 NB RT at Reems-Bell 65.2 66.3 97.9 79.7 40.6 55.5 67.1 NB TH at Reems-Bell Upstream of RT Bay 267.5 273.6 435.7 352.7 82.1 200.7 255.3 SB LT at Reems-Bell 49.7 48.9 54.1 52.9 46.2 48.2 37.5 SB TH at Reems-Bell Upstream of LT Bay 40.4 46.3 92.8 76.2 45.9 47.2 9.0 EB TH at Parkview-Bell 20.3 24.1 29.8 25.3 19.2 22.0 24.9 EB TH at Parkview-Bell Upstream of LT Bay 244.9 251.6 347.9 287.8 146.0 216.4 264.0 NB RT at Parkview-Bell 4.7 5.1 5.0 5.0 5.6 5.0 4.8 NB TH at Parkview-Bell Upstream of LT Bay 1.3 0.4 0.4 0.3 0.3 0.5 0.3 SB LT at Parkview-Bell 53.2 54.9 61.4 59.2 46.2 52.7 27.7 SB TH at Parkview-Bell Upstream of LT Bay 19.9 34.6 75.2 60.3 10.2 25.6 1.5 EB TH at Bullard-Bell 23.2 18.3 15.6 20.9 11.1 10.6 10.8 EB RT at Bullard-Bell 20.8 20.8 24.5 21.9 17.1 19.3 20.1 EB TH at Bullard-Bell Upstream of RT Bay 54.6 56.0 74.2 61.4 37.6 48.3 53.3 WB LT at Bullard-Bell 14.6 15.5 14.7 12.9 22.6 20.4 17.8 WB TH at Bullard-Bell 1.6 2.5 2.5 2.6 2.3 2.4 2.8 WB TH at Bullard-Bell Upstream of LT Bay 82.3 83.7 81.1 70.3 138.7 65.3 86.8 NB LT at Sun Village-Bell 107.0 166.2 146.6 181.8 116.7 111.2 73.7 NB TH at Sun Village- Bell Upstream of LT Bay 7.4 44.7 25.3 103.0 22.3 4.2 2.9 WB TH at Sun Village-Bell 11.6 10.8 10.2 9.0 17.5 9.0 12.7 WB TH at Sun Village- Bell Upstream of LT Bay 72.1 68.1 65.3 58.8 127.2 59.2 68.7 SB TH/RT at Sun Village- Bell 52.3 50.6 47.4 73.8 50.4 50.7 49.3 SB TH at Sun Village-Bell Upstream of LT Bay 1.8 1.8 1.8 2.0 1.9 1.8 1.8 NB LT at Litchfield-Bell 103.0 119.7 109.7 122.0 148.4 124.8 94.6 NB TH at Litchfield-Bell Upstream of LT Bay 115.3 201.2 143.9 199.5 298.5 232.6 123.3 WB TH at Litchfield-Bell 26.7 26.8 26.6 24.4 32.7 26.6 31.6 WB TH at Litchfield-Bell Upstream of LT Bay 121.4 122.9 122.4 109.4 177.8 134.5 157.1 SB TH at Litchfield-Bell 67.4 71.2 64.5 67.2 78.4 68.1 34.7 SB TH at Litchfield-Bell Upstream of LT Bay 27.8 63.6 11.6 11.7 73.4 20.4 1.2 Operation of traffic signal systems in oversaturated conditions Page 270
Figure 173. Performance summary comparison to extended left-turn split at Bullard (3 hour total) The results of this comparison illustrate the various mitigation strategies do not produce significantly different results. The mitigations which do reduce average delay on some links do so at the expense of other links which experience increased delay. Throughput Analysis The number of vehicles in the system was calculated by comparing the vehicle input and output data recorded by the simulation. The average input rates for each mitigation strategy as well as the Vissim Demand input are shown in Figure 174. This figure illustrates that the arterial is so congested that vehicles cannot enter the system at the rate that the model is demanding. Each mitigation strategy tested results in input rates which are higher than the no mitigation scenario indicating that each mitigation strategy allows for more vehicles to enter the system. 0 2 4 6 8 10 12 14 16 18 20 22 24 3 4 5 6 7 8 9 N um be r o f S eg m en ts Performance Summary of Different Mitigations 4:30 PM to 5:00 PM Significantly Better Slightly Better No Difference Slightly Worse Significantly Worse Operation of traffic signal systems in oversaturated conditions Page 271
Figure 174. Average input rates under different mitigations 200 300 400 500 600 700 800 900 30 0 60 0 90 0 12 00 15 00 18 00 21 00 24 00 27 00 30 00 33 00 36 00 39 00 42 00 45 00 48 00 51 00 54 00 57 00 60 00 63 00 66 00 69 00 72 00 75 00 78 00 81 00 84 00 87 00 90 00 93 00 96 00 99 00 10 20 0 10 50 0 10 80 0 11 10 0 11 40 0 11 70 0 N um be r o f V eh icl es Time (Seconds) Network Input Under Different Mitigations Game Day No Mitigation Extreme LT Split Increase at Bullard Negative Offset Simultaneous Offsets Double Cycling Resonant Cycle Moderate LT Split Increase at Bullard Dynamic Lane Assignment Reduced Cycle Length Vissim Demand Input Operation of traffic signal systems in oversaturated conditions Page 272
Similarly, the network output graph shown in Figure 175 illustrates that each mitigation strategy improves the baseline condition. The greatest impact of the mitigation strategies can be seen during the ârecoverâ portion of the curve which occurs after the peak hour at approximately 7800s into the simulation. Without any mitigation, the output continues to decrease until the input volumes decrease to 75% of the peak hour volume while each mitigation strategy causes the output to increase and return to steady state sooner than the baseline condition. The Dynamic Lane Assignment strategy resulted in the highest output rate and also yields the earliest recovery. Figure 175. Average output rates under different mitigations 200 300 400 500 600 700 800 30 0 60 0 90 0 12 00 15 00 18 00 21 00 24 00 27 00 30 00 33 00 36 00 39 00 42 00 45 00 48 00 51 00 54 00 57 00 60 00 63 00 66 00 69 00 72 00 75 00 78 00 81 00 84 00 87 00 90 00 93 00 96 00 99 00 10 20 0 10 50 0 10 80 0 11 10 0 11 40 0 11 70 0 N um be r o f V eh icl es Time (Seconds) Network Output Under Different Strategies Game Day No Mitigation Extreme LT Split Increase at Bullard Negative Offset Simultaneous Offsets Double Cycling Resonant Cycle Moderate LT Split Increase at Bullard Dynamic Lane Assignment Reduced Cycle Length Operation of traffic signal systems in oversaturated conditions Page 273
The average number of vehicles in the system was calculated using the input and output data shown above. Figure 176 shows the resulting number of vehicles in the system for each strategy. Figure 176. Average vehicles in the system 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 30 0 60 0 90 0 12 00 15 00 18 00 21 00 24 00 27 00 30 00 33 00 36 00 39 00 42 00 45 00 48 00 51 00 54 00 57 00 60 00 63 00 66 00 69 00 72 00 75 00 78 00 81 00 84 00 87 00 90 00 93 00 96 00 99 00 10 20 0 10 50 0 10 80 0 11 10 0 11 40 0 11 70 0 N um be r o f V eh icl es Time (Seconds) Number of Vehicles in System under Different Strategies Game Day No Mitigation Extreme LT Split Increase at Bullard Negative Offset Simultaneous Offsets Double Cycling Resonant Cycle Moderate LT Split Increase at Bullard Dynamic Lane Assignment Reduced Cycle Length Operation of traffic signal systems in oversaturated conditions Page 274
Determining the âbestâ strategy from this presentation of data depends on the desired outcome of the strategy. For example, if the objective is to reduce the number of vehicles âstuckâ in the system, one would choose a strategy which results in a data line which falls below the baseline. However, if the objective is to utilize the storage space within the system, choosing a mitigation which falls above the baseline date would be appropriate. The Double Cycling mitigation strategy appears to result in the highest number of vehicles in the system but Figure 177 also indicates that this strategy take the longest to recovery. The Dynamic Lane mitigation strategy keeps the number of vehicles in the system lower during the loading portion of the simulation and appears to recover sooner than any other strategy. Travel Time Analysis Strategies for mitigating the Bell Road corridor under game conditions was also analyzed in terms of average travel time through the network. The results of this analysis, shown in Figure 178, are similar to the delay analysis which implied that strategies that improve the performance for the westbound direction of travel, will negatively impact other directions of travel. Figure 177. Average travel time under different mitigation strategies 0 200 400 600 800 1000 1200 No G am e Ga m e Da y No M iti ga tio n Ex tr em e LT S pl it In cr ea se at B ul la rd Ne ga tiv e O ffs et Si m ul ta ne ou s O ffs et s Do ub le C yc lin g Re so na nt C yc le M od er at e LT S pl it In cr ea se at B ul la rd Dy na m ic L an e As sig nm en t Re du ce d Cy cl e Le ng th Tr av el T im e (S ec on ds ) Travel Time Comparison for Entire Corridor EB Bell Rd WB Bell Rd Operation of traffic signal systems in oversaturated conditions Page 275
Figure 178. Travel time comparison to Bullard from eastbound and westbound directions Summary In this test case, we applied the guidance process to a real-world situation with event traffic overlaid on normal heavy P.M. peak flows. This situation, unmitigated, produces extensive queuing which increases the travel time on the arterial by 400%. A number of different mitigation strategies were applied, working up from the basic mitigation to increase the left-turn split time at the critical intersection. All of the mitigations were found to be effective in reducing the westbound travel time. Some detriment to eastbound travel time resulted from the improvements in the westbound direction. The largest effects of the mitigations were observed during the recovery period. 0 100 200 300 400 500 600 700 800 900 No G am e Ga m e Da y No M iti ga tio n Ex tr em e LT S pl it In cr ea se at B ul la rd Ne ga tiv e O ffs et Si m ul ta ne ou s O ffs et s Do ub le C yc lin g Re so na nt C yc le M od er at e LT S pl it In cr ea se at B ul la rd Dy na m ic L an e As sig nm en t Re du ce d Cy cl e Le ng th Tr av el T im e (S ec on ds ) Travel Time Comparison to Bullard from East and West W end to Bullard E end to Bullard Operation of traffic signal systems in oversaturated conditions Page 276
