Reducing Pedestrian Delay at the Multi-Stage Pedestrian Crossing of Riverway at the Landmark Center Interchange
Author: Michael A. Santos
Advisor: Peter Furth
April 18, 2017
Summary
In conjunction with a project to reduce flooding risk on the Muddy River, the roads and intersections that make up the Landmark Center interchange (formally known as the Sears Rotary) were recently rebuilt with a new traffic circulation plan, opening in 2016. Where the Muddy River path, an important walking and cycling route and part of the Emerald Necklace, crosses Riverway, it was given a new three-stage crossing that offers terrible service for pedestrians and bicycles. For pedestrians, the average delay is 122 seconds, more than double what it takes to merit an “F’ level of service grade. For bicycles, there is no safe crossing provision except to dismount and walk.
This project measures pedestrian delay for this three-stage crossing using two different methods, one of them being the Northeastern University Ped & Bike Crossing Delay Calculator, a tool explicitly designed to enable engineers to determine pedestrian delay for complex crossings like this.
This study also proposes two possible changes in the traffic signal timing plan, and finds that one of these options can reduce pedestrian delay at the Muddy River Path crossing by about 75 seconds and makes possible a true bicycle crossing – all of this with less than a 1 second change in delay to motor traffic.
Introduction
The U.S. Army Corps of Engineers recently reconfigured several signalized intersections in the Fenway neighborhood of Boston, Massachusetts as part of the Muddy River Flood Damage Reduction & Environmental Restoration Project. The location of the project is shown in Figure 1. The project eliminated a left-turn jughandle from Brookline Avenue to Park Drive at the intersection of Brookline Avenue/Boylston Street/Park Drive. The project also signalized the intersection of Riverway/Riverway Connector, which was previously unsignalized and created a problematic weave as vehicles entered the intersection of Riverway/Brookline Avenue/Fenway. The project also reconfigured the intersection of Park Drive at Riverway/Landmark Center to allow northbound vehicles on Riverway to continue directly to Park Drive northbound. As part of the reconfiguration of this third intersection, new crossings were installed to allow pedestrian connectivity along the Muddy River multi-use paths. One of the new crossings involves a three-stage maneuver, resulting in significant pedestrian delays at the intersection. The reconfigured roadways and intersections are shown in Figure 2.
Figure 1: Project Location
Figure 2: Muddy River Project Roadway Geometry
Based on an analysis of the existing conditions, pedestrians currently experience an average delay of more than 120 seconds in both directions of pedestrian travel at the southern three-stage crossing. The Highway Capacity Manual (HCM)[1] considers pedestrian delays over 60 seconds to operate at level of service (LOS) F with high levels of pedestrian non-compliance with the signalized operation. When delays exceed 60 seconds, there is a high chance that pedestrians will cross against the flashing don’t walk and solid don’t walk indications, creating a safety issue.
This analysis will focus on the intersection of Riverway at Park Drive/Landmark Center and more specifically, the southern three-stage pedestrian crossing, and will present alternative timing and phasing plans that will reduce pedestrian delays by using advanced signal timing design techniques while maintaining satisfactory vehicular operations at the intersection. The goal of the alternative timing plans will also be to improve progression for the crossings to give pedestrians a better opportunity to cross without having to stop two times.
The microsimulation software VISSIM and the Northeastern University Ped & Bike Crossing Delay Calculator were used to analyze the existing conditions and the alternative timing and phasing plans.
Existing Conditions
Geometric Conditions
The intersection of Riverway at Park Drive/Landmark Center is shown in Figure 3. As shown in Figure 3, the intersection consists of four vehicular approaches, has a total of eight pedestrian crossings, and is under traffic signal control. The Park Drive eastbound approach consists of three right turn lanes – one exclusive right-turn lane for vehicles traveling southbound on the Riverway and two right-turn lanes for vehicles continuing on the Riverway Connector. The Park Drive westbound approach consists of five lanes – two through lanes for vehicles continuing on Park Drive, two left-turn lanes for vehicles traveling to the Riverway, and one left-turn lane for vehicles traveling to the Riverway Connector. The through lanes and left-turn lanes are separated by a grass island. The Riverway northbound approach consists of two left-turn lanes for vehicles traveling west on Park Drive. Due to the geometry of the islands at the intersection, the Riverway northbound approach has two stop-lines, both under traffic signal control. The Landmark Center southbound approach consists of two lanes, only allowing for right-turning vehicles traveling to Park Drive westbound, the Riverway, or the Riverway Connector.
Figure 3: Intersection of Riverway / Park Drive / Landmark Connector – Existing (2017) Geometry
Two three-stage crossings are provided for pedestrians in the east-west direction. The southern crossing is the more direct crossing for pedestrians and cyclists following the Muddy River. As shown in Figure 3, the southern crossing consists of three individual crosswalks of 24-feet, 24-feet, and 28-feet in length (from west to east).
While the Muddy River Path is an important bicycling route, there is no genuine crossing provision for bicycling. Neither the northern or southern crossing has islands long enough for bicycles to safely stop or for longer bicycles to queue. Of course, bicyclists can dismount and become pedestrians, but that is highly undesirable for a major bicycle route. In addition, the northern crossing’s sharp turns make it unsuitable for two-way bicycle traffic. Without changing the intersection geometry, the only way to have a genuine bicycle crossing is to have a time in the cycle in which bicycles can make the full southern crossing in one stage, without intermediate queuing.
Traffic Signal Timing/Phasing
Figure 4 shows the traffic signal timing and phasing as presented in the as-built plans for the project. The Boston Transportation Department (BTD) has slightly modified the timings by adjusting some phases by a few seconds. Table 1 presents the timings in the as-built plans and the BTD modified timings. Phase 1 serves the Park Drive westbound movements, Phase 2 serves the Riverway northbound movements, Phase 3 serves as a clearance phase for the Riverway northbound movements, and Phase 4 serves the Park Drive eastbound and westbound through movements.
Figure 4: Phasing and Timing – Existing Conditions
Table 1: Existing Traffic Signal Timings – AM Peak Hour
| Phase | As-Built Plans | BTD Updated Timings |
| 1 | 41 s | 38 s |
| 2 | 27 s | 28 s |
| 3 | 10 s | 11 s |
| 4 | 22 s | 23 s |
| Cycle | 100 s | 100 s |
As shown in Figure 4, the traffic signal cycle is 100 seconds during both the weekday AM and PM peak hours. The western crossing (AB) and eastern crossing (CD) occur only during Phase 2 of the signal cycle, while the central crossing (BC) occurs during Phase 4 and Phase 1. Crossings AB and CD are currently allocated 7 seconds of walk (W) time and 15 seconds of flashing don’t walk (FDW) time. Crossing BC is allocated 39 seconds of W time and 10 seconds of FDW time.
Traffic Volumes
There have been no traffic counts since the new intersection was completed, and given the time the public will need to become accustomed to the new traffic circulation plan, current counts would probably be a misleading indicator of demand. Therefore this analysis uses the volumes that were projected when the intersection was planned. They are based on traffic volume measurements conducted by VHB on June 20, 2007 prior to the reconstruction of the surrounding roadway network, with traffic volumes redistributed through the network based on an iterative origin-destination matrix to account for the new traffic patterns created by the direct connection between the Riverway and Park Drive westbound. The weekday morning peak hour traffic volumes at the Park Drive at Riverway/Landmark Center intersection are presented in Table 2.
Table 2: Weekday Morning Peak Hour Traffic Volumes
| Movement | Volumes (vehicles per hour) |
| Riverway NBL | 126 |
| Park Drive EBR (to Riverway Connector) | 572 |
| Park Drive EBR (to Riverway) | 231 |
| Park Drive WBT | 488 |
| Park Drive WBL (to Riverway) | 451 |
| Park Drive WBL (to Riverway Connector) | 106 |
| Landmark Center | negligible |
| TOTAL VOLUME | 1,974 |
Alternative Signal Plans
A review of the existing traffic signal timing plans indicates that there are opportunities to increase the amount of time given to pedestrians, specifically at the southern crossing. Two alternative traffic signal timing and phasing plans were considered:
Option 1: Additional pedestrian overlaps
Option 2: Short exclusive pedestrian phase with additional pedestrian overlaps
Both of the alternatives maintained the same offsets and the 100 second cycle length as the existing conditions in order to continue to function within the existing coordinated signal system throughout the area.
Option 1 – Additional Pedestrian Overlaps
Currently, phase 3 of the traffic signal is an 11 second long clearance phase that serves only the northbound left-turn movements (Riverway northbound left to Park Drive westbound). No other vehicular or pedestrian movements are served by this phase. The one vehicular movement that is allowed does not conflict with any of the southern pedestrian crossings, and so it makes sense to extend the time given to those crossings to include phase 3. Crossings AB and CD are currently served only by phase 2 of the cycle; it could readily be extended to include phase 3. Crossing BC currently starts in phase 4 and ends in phase 1 of the cycle; it could readily start sooner, in phase 3.
Further, it is possible to allocate more time within phase 1 to crossing BC, extending the Walk interval so that the flashing don’t walk (FDW) portion of the phase ends concurrently with the vehicular green indication.
By overlapping the pedestrian crossings with phase 3 and making full use of phase 1, crossings AB and CD will get 19 seconds of W and 15 seconds of FDW time, and crossing BC will get 57 seconds of W time and 10 seconds of FDW time. The phasing and timing diagram for Option 1 is shown in Figure 5.
Figure 5: Phasing and Timing – Option 1
Option 2 – Short Exclusive Pedestrian Phase With Additional Pedestrian Overlaps
The second option adds to the overlaps shown in Option 1 by including a short exclusive pedestrian phase between the existing phases 3 and 4. The additional pedestrian phase (phase 5) will not, in itself, be long enough for pedestrians to make the full crossing, but will, in combination with the other pedestrian phases, offer pedestrians a “green wave” to pass through the different crosswalks without long waits. It was determined that a 12-second all-pedestrian phase would be sufficient for that purpose.
Inserting this new phase while holding the signal cycle fixed at 100 s requires a retiming of the existing four phases. The 12-second exclusive pedestrian phase was added between phase 3 and 4. The phasing and timing diagram for Option 2 is shown in Figure 6. In effect, the added phase lengthens the “clearance time” after phase 2 from 11 s (phase 3) to 23 s (phases 3 and 5 together).
Figure 6: Phasing and Timing – Option 2
As shown in Figure 6, crossings AB and CD will start in phase 2 and end in phase 5 and will receive 28 seconds of W time and 15 seconds of FDW time. Crossing BC will start in phase 3 and end in phase 2 and will receive 65 seconds of W time and 10 seconds of FDW time.
Table 3 shows the W and FDW times for the three crossings under each of the scenarios.
Table 3: Pedestrian Phase Timings (in seconds)
| Crossing | Pedestrian Times (W/FDW) | ||
| Existing | Option 1 | Option 2 | |
| AB | 7/15 | 19/15 | 28/15 |
| BC | 39/10 | 57/10 | 65/10 |
| CD | 7/15 | 19/15 | 28/15 |
As shown in Table 3, the pedestrian times allocated to each of the crossings increase substantially compared to the existing conditions. However, an increase in pedestrian time does not necessarily mean that there will be significantly improved progression. The operations analysis will demonstrate the effects of the different timing and phasing plans on the southern pedestrian crossing.
Bicycle Crossing Potential with Options 1 and 2
A safe bicycle crossing requires at least 16 s; that’s 6 s for a “start” window plus 10 seconds of clearance (using a design speed of 10 ft/s for a 100 ft crossing).
Option 1 has an 11-second phase in which the full southern crossing is open, while option 2 has a 23-second interval (phases 3 and 5). Thus, Option 1 would still leave the intersection without a safe bicycle crossing, while Option 2 would make a bicycle crossing feasible.
Operations Analysis
Intersection performance for pedestrians and vehicles were analyzed using both VISSIM software and the Northeastern University Ped & Bike Crossing Delay Calculator. For purposes of this Project, the weekday morning peak hour was analyzed.
VISSIM Analysis
The VISSIM analysis was conducted by using the traffic signal timing and phasing plans provided by the U.S. Army Corps of Engineers and BTD. The traffic volumes and origin-destination data were input to accurately reflect the projected traffic patterns throughout the area. Pedestrian volumes were not provided in the data. Therefore, in order to analyze pedestrian behavior at the southern crossing, a total of 200 pedestrians per hour were used as inputs for both directions. Note that, since pedestrian phases are automatically recalled, the volume of pedestrians does not affect vehicular operations.
The results of the VISSIM analysis are presented in Table 4. Delays for the southern pedestrian crossing and all vehicular approaches to the intersection are shown for each of the three scenarios (Existing conditions, Option 1, and Option 2). The delays for the pedestrian crossings represent the total delay for all three stages.
Table 4: Operations Analysis – Weekday AM Peak Hour
| Movement | Average Delay (seconds) | ||
| Existing | Option 1 | Option 2 | |
| Southern Pedestrian Crossing (east to west) | 123 | 112 | 42 |
| Southern Pedestrian Crossing (west to east) | 121 | 110 | 53 |
| Riverway NBL | 38 | 40 | 45 |
| Park Drive EBR (to Riverway) | 45 | 45 | 45 |
| Park Drive EBT (to Riverway Connector) | 41 | 41 | 41 |
| Park Drive WBT | 33 | 33 | 33 |
| Park Drive WBL (to Riverway) | 24 | 24 | 24 |
| Park Drive WBL (to Riverway Connector) | 17 | 17 | 17 |
| All pedestrians | 122 | 111 | 46 |
| All vehicles | 34.8 | 34.9 | 35.3 |
Table 4 shows that the vehicular movements are almost completely unaffected by either Option 1 or Option 2, with average delay increasing by less than a second for both cases.
However, Table 4 shows some dramatic changes to pedestrian delay. In existing conditions, average delay is 122 seconds (approximately 2 minutes). Pedestrian delays are slightly reduced to around 110 seconds under Option 1, with the addition of the pedestrian overlaps with Phase 3. Despite an increase in the pedestrian walk times under Option 1 when compared to the existing conditions, the average pedestrian delays remain significant and at an overall LOS F, because pedestrians still have to wait for a long time at the islands after their first and second crossings. Under Option 2, however, pedestrian delays are reduced by 76 seconds on average, because adding the short exclusive pedestrian phase gives pedestrians something close to a “green wave,” allowing them to make one crossing after another with little or no wait in between.
Videos from the VISSIM simulations for each option are shown in the following links below:
Video Simulating EXISTING CONDITIONS
Northeastern University Ped & Bike Crossing Delay Calculator
The Northeastern University Ped & Bike Crossing Delay Calculator was used to estimate delays experienced by pedestrians at the southern crossing for all three scenarios. The inputs needed for this calculation include crossing lengths, cycle length, walk start time, and walk duration. It also assumes that the pedestrian walk speed is 4.5 feet per second – a typical average walking speed – and an additional 4 seconds of effective “walk” time, recognizing that most pedestrians will still enter the crosswalk for the first few seconds of the FDW period. Pedestrian volumes are not used as an input for the calculation because it is assumed that pedestrians do not experience any delays from queuing behind each other (this effect is significant only at very crowded crosswalks).
The results of the delay calculations using the NU Ped & Bike Crossing Delay Calculator are shown in Table 5, with the VISSIM results shown for comparison purposes.
Table 5: Pedestrian Delays – NU Ped & Bike Crossing Delay Calculator (AM Peak Hour)
| Pedestrian Crossing | Existing Conditions | Option 1 | Option 2 | |||
| NU Ped & Bike Delay Calculator | VISSIM Results | NU Ped & Bike Delay Calculator | VISSIM Results | NU Ped & Bike Delay Calculator | VISSIM Results | |
| West to East | 124 s | 120 s | 112 s | 110 s | 42 s | 48 s |
| East to West | 123 s | 122 s | 111 s | 111 s | 41 s | 41 s |
As shown in Table 5, the NU Ped & Bike Crossing Delay Calculator verifies the results obtained from the VISSIM models. As stated earlier, pedestrian delays are slightly reduced in Option 1 and are substantially reduced in Option 2.
Figures 7, 8, and 9 show time-space diagrams that illustrate pedestrian progression under each scenario, along with other output from the NU Ped & Bike Crossing Delay Calculator.
Figure 7: Existing Conditions – NU Ped & Bike Crossing Delay Calculator (AM Peak Hour)
As shown in Figure 7, progression for pedestrians is very poor. In both directions (blue for crossing from A to D, black for crossing from D to A), pedestrians finishing any stage of crossing have to wait, sometimes a long time, before they can start the next stage. Because the the first and last crossings (AB and CD) operate during the same single phase, it takes pedestrians an entire cycle to complete their crossing (that’s after waiting for a WALK to begin their crossing).
Figure 8: Option 1 – NU Ped & Bike Crossing Delay Calculator (AM Peak Hour)
As shown in Figure 8 (Option 1), the added overlaps lengthen the green “windows” within which pedestrians can enter crosswalks AB and CD. However, nearly all pedestrians are still required to stop and wait at an island after each stage of the crossing. This Option demonstrates that additional pedestrian time does not necessarily lead to better progression or significantly reduced delays if the pedestrian phases are not sequenced efficiently.
Figure 9: Option 2 – NU Ped & Bike Crossing Delay Calculator (AM Peak Hour)
Figure 9 (Option 2), in which a short pedestrian-only phase has been added, shows vastly improved progression in both directions. In both directions, there are pedestrians who, depending on their initial arrival time, can pass through all three crossings without delay, and the majority of pedestrians experience only a short delay at the first island they meet and then no delay at the second island. The only pedestrians that finish their crossing a full cycle after beginning are the roughly 20% whose initial arrival time is during the latter part of “green” (that is “WALK”) for their first crossing.
Discussion
This study has focused on pedestrian delay, assuming that pedestrians comply with the WALK signal (although allowing that they will treat the first 4 seconds of Flashing DON’T WALK as if it were WALK). In reality, it is recognized that when delays become inordinately long, non-compliance becomes more likely. That makes long delays a safety issue. At this particular intersection, non-compliance means crossing multilane roadways that sometimes have fast traffic and in which the direction from which traffic might approach is not always clear. Therefore, signal timing options that reduce pedestrian delay can also be expected to contribute to improved pedestrian safety.
This crossing is also important for bicyclists using the Muddy River Path. The current provision is safe only if bicycles dismount and become pedestrians; however, the majority of cyclists will not do that, and are therefore making unsafe crossings. Option 2, with a 23-second interval that could be used for a bicycle crossing, makes a safe bicycle crossing possible. To make that crossing work as it should, bicycle signals should be installed with the bicycle phase running only during that 23-second interval. Bicycle signals are needed because the time at which it’s safe for bicycles to begin crossing is different from what will be indicated by the pedestrian signals.
Conclusion
Multi-stage pedestrian crossings present challenges to designers and engineers to provide acceptable levels of service for pedestrians. The intersection of Park Drive at Riverway/Landmark center was recently redesigned with vehicular optimization as a priority, with little thought toward pedestrian progression through the multi-stage crossings. By implementing an alternative signal timing time expressly designed to provide good progression for pedestrians, overall delay to pedestrians can be substantially reduced with little negative impact to vehicular delay. By reducing the delay to pedestrians at the multi-stage crossing, pedestrians are less likely to cross against the flashing don’t walk and the don’t walk phases, improving the overall safety at the crossing and at the intersection.
In this particular case, good pedestrian progression was created by adding a short exclusive pedestrian phase – not long enough to make the full crossing, but long enough that, in combination with (existing) concurrent WALK intervals, it creates a “green wave” for pedestrians through the three stages of the crossing. Compared to existing conditions, pedestrian delay is reduced by 76 s while average vehicular delay changes by less than a second. At the same time, this case showed that simply lengthening the WALK intervals without paying attention to progression offered very little reduction in crossing delay.
This study also shows the utility of the Northeastern University Ped and Bike Crossing Delay Calculator. It enables one to calculate pedestrian delay at multistage crossings without having to resort to (more complex) simulation. In addition, the progression diagrams it creates are a valuable tool for visualizing the quality of progression for multistage crossings, which can help a designer see what changes to signal timing might be needed to improve progression.