Transit Signal Priority: Current State of the Practice (2020)

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.

17 C H A P T E R 3 This chapter presents the results of the survey of North American transit agencies about their use of TSP. A list of North American transit agencies believed to be using TSP was developed, and 79 agencies on this list were randomly selected to participate in an online survey about their use of TSP. Of the 79 transit agencies, 74 agencies received the survey; five were ineligible to par- ticipate either because they do not work in English or because they responded outside the survey that they do not have bus TSP. Of the 74 remaining transit agencies, another 19 were marked as ineligible because the contact information found by the study team did not result in any response from the agency (often the recipient had left the agency and no alternative contact information resulted in a response—not even a confirmation that the invitation was received). Fifty-five eligible and responsive agencies remained, and 46 completed the survey, a response rate of 84%. Of the 46 transit agencies that responded, 28 reported having an active TSP deployment, but only 26 provided information about their deployment groups. Appendix A includes the complete questionnaire. Table 2 provides the full list and key operational characteristics of the 26 responding transit agencies with TSP; Figure 2 is a map showing the locations of these transit agencies. Survey Design The survey was designed to allow transit agencies to report on the characteristics, business rules, parameters, technology, benefits, and costs for up to two different TSP deployment groups by repeating the section of the survey asking for TSP deployment group characteristics. To reduce the length of the survey, transit agencies were requested to provide details only for their two most recent deployment groups. The survey was divided into seven sections: • Introduction • Questions for transit agencies with no active TSP deployments (only if the agency does not have TSP) • Background information on TSP deployments • TSP definitions and specifying deployment groups • TSP deployment group characteristics (repeated twice if the agency had more than one deployment group) – Rules for requesting priority – Rules for granting priority – Additional business rules and priority restrictions – TSP corridor infrastructure – TSP system technology – TSP benefits – TSP costs Survey Results

Transit Agency Full Name Abbreviation Region Bus Ridershipa Bus Revenue Hoursa Bus Peak Vehiclesa TSP Implement Year No. of Deployment Groups Brampton Transitb Brampton Transit City of Brampton—Greater Toronto Area, ON, Canada 27,391,889 -- 438c 2010 1 Calgary Transitd Calgary Transit Calgary, AB, Canada 105,300,000e -- 1,224e 2000 1 Capital Metropolitan Transportation Authority CapMetro Austin, TX 27,824,443 1,270,440 349 2013 1 Central Ohio Transit Authority COTA Columbus, OH 18,401,546 1,072,219 297 2018 1 Champaign-Urbana Mass Transit District C-U MTD Champaign-Urbana, IL 11,939,808 269,598 96 2007 1 Charlotte Area Transit CATS Charlotte, NC-SC 2,189,612 97,665 23 Missing 1 Cities Area Transit CAT Grand Forks, ND-MN 280,289 25,296 8 2010 2 Denver Regional Transportation District Denver RTD Denver, CO 65,266,321 2,805,769 867 2016 2 Greater Cleveland Regional Transit Authority GCRTA Cleveland, OH 30,931,712 1,180,161 299 2008 1 Hillsborough Area Regional Transit Authority HART Tampa, FL 12,901,178 655,997 162 2014 1 King County Metro Transit Department King County Metro Seattle-Tacoma, WA 121,352,294 3,511,078 1,155 1999 3 or more Kitsap Transit Kitsap Transit Bremerton, WA 2,533,304 130,054 90 1992 1 Maryland Transit Administration MDOT MTA Baltimore, MD 73,804,198 1,971,785 884 2017 1 Mass Transit Department - City of El Paso Sun Metro El Paso, TX 13,047,380 585,977 140 2008 3 or more Miami-Dade Transit Miami-Dade Miami, FL 58,383,786 2,502,559 747 2015 1 New York City Department of Transportation NYCDOT New York City, NY-NJ-PA 554,655 22,175 25 Missing 1 Ottawa Transitf OC Transpo Ottawa, ON, Canada 94,400,000g -- 978g 1997 2 Pace - Suburban Bus Division PACE Cook, Lake, DuPage, Will, Kane, and McHenry Counties, IL 28,804,740 1,720,130 635 2008 2 Rhode Island Public Transit Authority RIPTA Providence, RI 16,239,062 672,788 195 2014 1 Riverside Transit Agency RTA Riverside-San Bernardino, CA 8,315,598 623,109 163 2013 1 San Diego Metropolitan Transit System MTS San Diego, CA 49,919,496 1,823,011 512 2014 1 San Francisco Municipal Transportation Authority SFMTA San Francisco, CA 161,097,082 2,782,405 674 1998 1 Southeastern Pennsylvania Transportation Authority SEPTA Philadelphia, PA-NJ-DE-MD 169,407,096 4,078,575 1,223 2016 1 Toronto Transit Commissionh TTC Toronto, ON, Canada 261,112,835 -- 1,920i 1991 1 Washington Metropolitan Area Transit Authority WMATA Washington, DC-VA-MD-WV 123,124,352 3,949,021 1,290 2016 1 York Region Transit YRT York Region—Greater Toronto Area, ON, Canada -- -- -- 2005 1 NOTES: Unless otherwise stated, ridership revenue hours, and peak vehicles data are obtained from the National Transit Database (2017). -- indicates that the data were not provided or were not readily available. a Bus ridership, revenue hours, and peak vehicles data include all fixed-route bus modes. b Data for Brampton Transit obtained from the Brampton Transit website (2019). c Brampton transit has 438 vehicles; fewer would be used in peak service. d Data for Calgary Transit obtained from the Calgary Transit website (2019). e Calgary Transit's ridership and vehicle numbers are from 2018 and include CTrain service (bus data were not available separately). f Data for Ottawa Transit obtained from the Ottawa Transit website (2019). g Ottawa Transit's ridership includes O-train and other non-fixed-route-bus services. Ottawa transit has 978 buses; fewer would be used in peak service. h Data for the Toronto Transit Commission obtained from the Toronto Transit Commission website (2019). i The Toronto Transit Commission has 1,920 buses; fewer would be used in peak service. Table 2. Transit agencies with TSP that completed the survey.

Survey Results 19 • TSP overall experience and operations • Participant information Appendix C contains a flowchart summarizing the overall survey design. Custom survey web links were provided to each transit agency to permit the survey to be easily shared among multiple people within the transit agency or at local partner organizations in an effort to obtain all requested information. Because the survey contains conditional logic, not every question was shown to every respondent. In addition, agencies with multiple TSP deploy- ments went through the questions in the TSP Deployment Group Business Rules and Parameters section twice. Therefore, this report refers to the 26 agency-level survey responses as agencies and to the 31 deployment-group-level survey responses as deployment groups. Also, to clarify the sample size, the total number of responses is included with all summary tables in this chapter. Deployment Table 3 presents the list of 46 survey respondents indicating the status of TSP at their transit agencies; 28 surveyed transit agencies (61%) currently have TSP. (However, only 26 of the agen- cies with TSP completed the TSP deployment survey questions.) The 18 agencies (39%) that do not currently have TSP deployed were asked about their future plans, with the responses shown in Table 4. One transit agency currently without TSP is conducting TSP pilot testing, one is con- ducting a TSP feasibility study, and 15 have no current feasibility studies or pilot tests. (One agency did not respond.) Of the 15 transit agencies with no feasibility studies or pilot tests, 11 have plans to implement TSP in the future. SOURCE: Created by the authors using National Transit Database (2017) data. 8–25 26–195 196–349 350–635 636–1,015 1,016–1,920 Figure 2. Map showing transit agencies with TSP that completed the survey.

20 Transit Signal Priority: Current State of the Practice Table 3. List of survey respondents and current TSP status. Transit Agency Service Area Does Transit Agency Use TSP? Berkshire Regional Transit Authority Pittsfield, MA No Brampton Transit City of Brampton—Greater Toronto Area, ON, Canada Yes Calgary Transit Calgary, AB, Canada Yes Capital Metropolitan Transportation Authority Austin, TX Yes Central Ohio Transit Authority Columbus, OH Yes Champaign-Urbana Mass Transit District Champaign-Urbana, IL Yes Charlotte Area Transit System Charlotte, NC-SC Yes Cities Area Transit Grand Forks, ND-MN Yes Citrus Connections Lakeland, FL No CityBus Lafayette, IN No CobbLinc Atlanta, GA No Colorado Springs Transit Colorado Springs, CO No Denver Regional Transportation District Denver, CO Yes Des Moines Area Regional Transit Des Moines, IA No Foothill Transit Greater Los Angeles, CA No Golden Empire Transit Bakersfield, CA No Greater Cleveland Regional Transit Authority Cleveland, OH Yes Greater Portland Metro Portland, ME No Greeley Evans Transit Greeley, CO No Hillsborough Area Regional Transit Authority Tampa, FL Yes Jefferson Transit Jefferson County, WA No King County Metro Transit Department Seattle-Tacoma, WA Yes Kitsap Transit Bremerton, WA Yes Maryland Transit Administration Baltimore, MD Yes Mass Transit Department—City of El Paso El Paso, TX Yes Metrolinx Toronto-Hamilton, ON, Canada No Metropolitan Transit System San Diego, CA Yes Miami-Dade Transit Miami, FL Yes Muncie Indiana Transit System Muncie, IN Yes New York City Department of Transportation New York City, NY-NJ-PA Yes Ottawa Transit Ottawa, ON, Canada Yes Pace—Suburban Bus Division Cook, Lake, DuPage, Will, Kane, and McHenry Counties, IL Yes Pee Dee Regional Transportation Authority Florence, SC No Pocatello Regional Transit Pocatello, ID No Rhode Island Public Transit Authority Providence, RI Yes Riverside Transit Agency Riverside-San Bernardino, CA Yes Sacramento Regional Transit District Sacramento, CA No San Francisco Municipal Transportation Agency San Francisco, CA Yes Sonoma County Transit Santa Rosa, CA No Southeastern Pennsylvania Transportation Authority Philadelphia, PA-NJ-DE-MD Yes TheRide Ann Arbor, MI No Toronto Transit Commission Toronto, ON, Canada Yes Translink Vancouver, BC, Canada No Unitrans Davis, CA Yes Washington Metropolitan Area Transit Authority Washington, DC-VA-MD-WV Yes York Region Transit York Region—Greater Toronto Area, ON, Canada Yes

Survey Results 21 Of responding transit agencies with active TSP deployments, the oldest TSP system was deployed in 1991 by the Toronto Transit Commission (TTC), and the most recent in 2018 by the Central Ohio Transit Authority. Table 5 summarizes the responses to questions about the scope of active TSP deployments, including how many intersections, routes, and buses are part of each deployment group. Most transit agencies (20 of 26, or 77%) have only one deployment group (i.e., the system architecture, business rules, and parameters are similar across the whole system), but the survey sample includes TSP deployments of various sizes. Of the 31 deployment groups, eight groups are shared with other transit agencies and 23 are not. Transit agencies also indicated the criteria used to identify corridors for TSP deployment (see Table 6). Route type is by far the most common (20 of 26 responses, or 77%). Strategy, Business Rules, and Parameters This section presents findings from the survey regarding TSP strategy, business rules, and parameters. These questions cover the eligibility of buses for priority, the type of priority received, and restrictions on the use of priority. The responses are tabulated by deployment group, so agencies with multiple deployment groups will have two responses counted. Table 4. Status and future plans for transit agencies without TSP. Transit Organization Status of TSP at Transit Organization Future TSP Consideration Berkshire Regional Transit Authority No current feasibility studies or pilot tests No Citrus Connections No current feasibility studies or pilot tests Yes CityBus No current feasibility studies or pilot tests Yes CobbLinc No current feasibility studies or pilot tests Yes Colorado Springs Transit — — Des Moines Area Regional Transit Authority No current feasibility studies or pilot tests Yes Foothill Transit Conducting TSP pilot testing NA Golden Empire Transit No current feasibility studies or pilot tests Yes Greater Portland Metro No current feasibility studies or pilot tests Yes Greeley Evans Transit No current feasibility studies or pilot tests Yes Jefferson Transit No current feasibility studies or pilot tests No Metrolinx No current feasibility studies or pilot tests Yes Pee Dee Regional Transportation Authority No current feasibility studies or pilot tests No Pocatello Regional Transit No current feasibility studies or pilot tests Yes Sacramento Regional Transit District No current feasibility studies or pilot tests Yes Sonoma County Transit No current feasibility studies or pilot tests No TheRide No current feasibility studies or pilot tests Yes Translink Conducting a TSP feasibility study NA NOTES:—indicates that the respondent did not answer the question; NA = not available. Table 5. Scope of active TSP deployments. No. of Deployments Agency Count Intersections Deployment Count Routes Deployment Count Buses Deployment Count 1 20 30 or fewer 9 1 7 50 or fewer 10 2 4 31–60 9 2–10 13 51–1,000 11 3+ 2 61 or more 8 11 or more 7 1,001 or more 6 Missing 0 Missing 5 Missing 4 Missing 4 Total agencies 26 Total deployments 31 Total deployments 31 Total deployments 31 NOTES: Even if a transit agency has three or more deployment groups, the agency provided data only for their two most recently implemented deployment groups. Missing = missing data or no response.

22 Transit Signal Priority: Current State of the Practice Requesting Priority There are many potential business rules to determine whether buses are eligible to request priority at a TSP-enabled intersection. One focus of the survey was to determine whether TSP deployments required buses to be in a certain operational status to request priority (i.e., behind schedule, outside of a headway window, or carrying a minimum number of passengers). Table 7 summarizes the conditions that buses have to meet before requesting priority (assuming the bus is on a TSP-eligible route at a TSP-enabled intersection at a TSP-eligible time of day on a TSP- eligible day of the week). Thirteen deployments (42%) have at least one condition on the bus’s operational status, and the remaining 18 deployments (58%) have unconditional TSP. Many respondents indicated “other” criteria for buses to be eligible to request priority (in addition to an operational status). Typed answers for “other” responses included distance to signal, operat- ing an eligible route, and having the doors closed. However, for this analysis, these answers were coded as “unconditional.” It is important to point out that there were 18 deployment groups at 15 transit agencies for which buses did not have to meet an operational status criterion to be eligible to request priority. These deployment groups may have had some other requirements for requesting and granting priority, but those business rules were not related to the operational status of a bus. TSP deploy- ments without business rules related to the operational status of a bus result in buses’ always requesting priority as long as they meet other criteria related to direction of travel, time of day, day of week, route, and intersection. These deployments are usually focused on improving bus speeds along a corridor. Corridor Selection Criteria Transit Agency Count Percent Route type 20 77 Network or corridor type 10 38 Ridership 9 35 Bus frequency/headway 8 31 Observed service problem 6 23 Other 4 15 Agencies with a Response 26 NOTE: Respondents could select more than one answer. Table 6. Corridor selection criteria for TSP deployment. Operational Status Priority Criterion or Condition Deployment Count Priority Conditions Must Be Met 13 Schedule Deviation 13 Headway Deviation 0 Passenger Load 2 No Conditiona 18 Deployment Groups with a Response 31 NOTES: Respondents could select more than one priority condition. aHaving no condition does not mean that all buses could request priority all the time. Some deployment groups had limits on the time of day, day of week, direction of travel, and so forth. However, the buses themselves did not have to be in a certain operational status (e.g., late) to receive priority. Table 7. Frequency of business rules for buses to be eligible to request priority.

Survey Results 23 Of the 13 deployments with at least one condition, every deployment used schedule deviation, and two deployments also used passenger loads. Schedule Deviation Parameters For each deployment group, transit agencies provided the minimum amount of time that a bus must be running late before it may begin to request priority. The time ranged from 1 second to 360 seconds. Only one deployment group (King County Metro’s MetroRapid C–D Line TSP Optimization) allowed buses to request priority when only 1 second late. The remaining deploy- ment groups required 60, 120, 180, 240, or even 360 seconds of lateness before buses would be eligible to request priority (see Figure 3). In many cases, deployment groups required different minimum seconds of schedule deviation across intersections; however, the figure shows the smallest allowable deviation. Transit agencies also provided the number of seconds late at which buses STOP requesting priority. This parameter may be different from the previous schedule deviation parameter if, for example, buses must be 60 seconds late to request priority but are allowed to continue request- ing priority until they have returned to schedule (0 seconds late). Responses indicated that most deployment groups had STOP priority request parameters within 1 second of the START prior- ity request parameter; however, three deployment groups had STOP priority parameters that were substantially different. Calgary Transit’s buses that request priority after becoming 120 sec- onds late are able to request priority until they have returned to their schedule (i.e., 0 seconds late). Capital Metro’s buses can request priority once they are 60 seconds late and continue to request priority until they have returned to their schedule. Last, the Hillsborough Area Regional Transit Authority’s buses can request priority after becoming 180 seconds late and continue requesting priority until they are only 60 seconds late. Other Business Rule Parameters for Requesting Priority Two deployment groups had a business rule based on passenger load, and one agency, the Washington Metropolitan Area Transit Authority, provided the parameter: buses must have at least 10 passengers on board. No transit agencies used headway deviation as a business rule. 0 2 4 6 8 10 12 14 16 18 20 None 1 60 120 180 240 300 360 N um be r o f D ep lo ym en t G ro up s Schedule Deviation Parameter (in seconds) NOTE: The 18 deployment groups with no schedule deviation parameter had no business rules based on bus operational status. Remaining responses include the 13 deployment groups that had a schedule deviation business rule. Figure 3. Minimum schedule deviation (lateness) parameter for buses to request priority.

24 Transit Signal Priority: Current State of the Practice Intersection Check-In and Check-Out For each deployment group, transit agencies identified how many seconds before arriving at an intersection a priority-eligible bus sends its priority request (i.e., checks in to an intersection), assuming locations with no constraints for nearby bus stops or other signals. Most deployment groups had a range of values indicating a typical minimum and typical maximum time for buses to check in to intersections. The minimum times ranged from 1 second to 92 seconds, with a median of 10 seconds. The maximum times ranged from 10 seconds to 92 seconds, with a median of 20 seconds. Many responses indicated that check-ins are triggered by geofencing, and check-in time and location may vary according to complex algorithms that account for the type of priority available at a given intersection, typical traffic speeds, average network latency, and so on. Another question asked whether check-out detection is used. This typically involves a sensor placed downstream of the intersection and can be used to terminate a green extension or special bus phase once the bus has cleared the intersection. Of 31 deployment groups, 20 have check-out detection, 10 do not, and one transit agency did not answer the question. Granting Priority Once a priority request is received from a bus or a priority request generator (PRG), logic built into the TSP system must determine whether to grant priority and, if so, what type of priority to provide. Priority Request Conflicts When two or more TSP-eligible buses approach a TSP-eligible intersection, their priority requests may conflict with one another. Table 8 summarizes the methods that are used to resolve conflicting priority requests. Although such methods have been an area of research in the litera- ture, the vast majority of TSP deployments use a first-come, first-served rule that is simple to implement. One agency, TTC, answered “other” and indicated that conflicts are handled on the fly, according to decision points built into the signal cycle in which the controller checks whether TSP-eligible buses are in detection zones. (TTC’s business rules are explained further in its case example, in Chapter 4.) Priority Types Table 9 summarizes what types of priority buses can receive, according to the 31 deployment groups in the survey. The vast majority of deployments use extended green. Table 10 shows how agencies combine these priority types. The largest group, covering 12 deployments (39%), uses only extended green. Extended and early green are used together in 15 deployments (48%), some of which also use special bus-only signals or phase insertion, rotation, or skipping. Conflict Resolution Method Deployment Count First come, first served 27 Prefer greatest schedule/headway deviation 1 Route hierarchy or corridor type 3 Other 1 Deployment Groups with a Responsea 30 NOTES: Respondents could select more than one answer. aOne deployment group was missing data for this question. Table 8. Priority request conflict resolution method.

Survey Results 25 Green Extension Parameters Transit agencies identified by how many seconds the green light is typically extended, assum- ing locations with no constraints for nearby bus stops or other signals. Transit agencies pro- vided ranges (for example, 5 to 10 seconds) for 20 TSP deployment groups (eight deployment groups did not have data for the green extension length parameter). The minimum extension per deployment group ranged from 1 second to 30 seconds, with a median of 7.5 seconds. The maximum extension per deployment group ranged from 3 seconds to 30 seconds, with a median of 10 seconds. Figure 4 displays the possible green extension lengths for the 20 deployment Table 9. Priority types used in TSP deployment groups. Type of Priority Deployment Count Extended green 28 Early green 16 Phase insertion, rotation, or skipping 4 Special bus-only signal 7 Other 4 Deployment Groups with a Response 31 NOTE: Respondents could select more than one answer. Table 10. Priority combinations used in TSP deployment groups. Type of Priority DeploymentCount Extended only 12 Extended + early 8 Extended + early + special 5 Extended + early + phase 2 Special only 1 Extended + phase 1 Early + phase + special 1 Other only 1 Deployment Groups with a Response 31 NOTE: Unless the deployment group had only an “other” type of priority, “other” is not considered a unique type of priority for this table. Figure 4. Possible green extension lengths. 0 2 4 6 8 10 12 14 16 1–5 6–10 11–15 16–20 21–25 26–30 N um be r o f D ep lo ym en t G ro up s Possible Green Extension Lengths (in seconds) NOTES: Based on 20 responses. Columns in the chart count up any deployment group in which the green extension range falls within the end points of the column range. For example, a green extension of 5 to 12 seconds is counted in three columns: 1–5, 6–10, and 11–15.

26 Transit Signal Priority: Current State of the Practice groups that had parameter data. Deployment groups with ranges larger than a single column range are counted in multiple columns in the chart. The most common available range for a green extension length was 6 to 10 seconds, with 14 deployment groups (70% of 20 groups that had data) allowing green extensions at least within this range. Early Green Parameters For deployments that provide early greens, transit agencies identified how many seconds earlier the green light is typically activated (parameter ranges were provided for 13 of the 16 TSP deployment groups that grant early greens). The minimum early green activation per deploy- ment group ranged from 1 second to 10 seconds, with a median of 7.5 seconds. The maxi- mum early green activation per deployment group ranged from 3 seconds to 60 seconds, with a median of 10 seconds. Figure 5 displays the possible early green lengths for the 13 deployment groups with parameter data. Deployment groups with ranges larger than a single column range are counted in multiple columns in the chart. The most common available duration for an early green was between 6 and 10 seconds, with 11 deployment groups (85% of 13 groups with data) allowing early greens at least within this range. Limitations Placed on Priority Transit agencies reported what additional business rules are in place to restrict the request- ing or granting of priority (see Table 11). Lockout restrictions were the most common (16 of 31 responses, or 52%). Lockout rules deny priority requests for a certain amount of time or a certain number of signal cycles after a successful priority request. Their purpose is to allow sufficient time for the signal to return to its normal timing and to dissipate queues on other approaches that accumulated because of their phases being skipped or shortened. Of the 16 deployments that have lockout rules, seven are based on the time since the last granted priority request, three are based on the cycles since the last priority request, three use both time and cycles, and three responses were missing. Figure 6 shows the distribution of lock- out durations for agencies that reported a time value for deployments using time-based param- eters. Figure 7 shows the distribution of cycles among granted priority requests for deployments that use cycle-based parameters. 0 2 4 6 8 10 12 1–5 6–10 11–15 16–20 21–25 26–30 31–60 N um be r o f D ep lo ym en t G ro up s Possible Early Green Length (in seconds) NOTES: Based on 13 responses. Columns in the chart count up any deployment group in which the early green duration range falls within the end points of the column range. For example, an early green duration of 5 to 12 seconds is counted in three columns: 1–5, 6–10, and 11–15. Figure 5. Possible early green lengths.

Survey Results 27 Restriction Type Deployment Count Lockout 16 Location (some intersections/corridors are not eligible) 10 Direction (some route directions are not eligible) 11 Day of week (some days are not eligible) 6 Time of day (some times of day are not eligible) 11 Service type (certain types of transit services are not eligible) 9 Some other rulea 1 Deployment Groups with a Response 31 NOTES: Respondents could select more than one answer. aThe one “other” response suggested that emergency vehicle requests may limit transit priority requests. Table 11. Business rules that place restrictions on priority. 0 1 2 3 4 5 6 120–179 180–239 240–299 300–359 360–419 420–479 480–539 540–599 600 Possible Lockout Durations (in seconds) N um be r o f D ep lo ym en t G ro up s NOTES: Based on 10 responses from deployment groups that include a time-based parameter for intersection lockout. Columns in the chart count up any deployment group in which the lockout duration range falls within the end points of the column range. For example, a lockout duration of 120 to 180 seconds is counted in two columns: 120–179 and 180–239. Figure 6. Lockout duration parameters for time-based lockout rules. 0 1 2 3 4 5 1 2 3 N um be r o f D ep lo ym en t G ro up s Lockout Duration (in cycles) NOTE: Based on six responses from deployment groups that include a cycle-based business rule for intersection lockout. Figure 7. Lockout duration parameters for cycle-based lockout rules.

28 Transit Signal Priority: Current State of the Practice Of the 11 deployments that specified a time of day rule, three restrict TSP use to a.m. and p.m. peak periods; one restricts TSP use to a.m. peak, mid-day, and p.m. peak; and six permit TSP use around the clock. System Architecture The main distinction in TSP system architecture is between decentralized systems, where buses communicate directly with traffic signals, and centralized systems, where priority requests are processed through a centralized transit management system. Decentralized systems can be referred to as vehicle-to-infrastructure systems, and centralized systems can be referred to as center-to-center systems. Table 12 presents the findings from the survey, showing that decen- tralized systems are currently the most common. Table 13 summarizes what other commonly used transit technology systems are integrated with TSP. Seventeen deployments have integrated AVL systems with TSP, which is inherent in some architectures. (For example, AVL is necessary to support a centralized TSP system so the transit management system knows where buses are when communicating priority requests to the signaling system.) Five deployments integrate passenger counts with TSP, either through automated passenger counters (APCs) or through fareboxes. However, only two deployments use passenger counts to determine priority request eligibility. Agencies were also asked questions about the hardware and vendors used for TSP systems. Table 14 displays the vendors of components on board transit vehicles, and Table 15 displays the vendors of signaling system components. Additional Infrastructure Design Factors Transit agencies also provided information about other corridor design factors that are likely to affect bus performance, including stop location, queue jumps, dedicated bus lanes, and route overlap. Transit agencies estimated the percentage of stops in the deployment group that are located near side, far side, or mid-block. The values for each stop location type ranged from 0% to Bus Communication with Signals Deployment Count Vehicle-to-infrastructure 24 Center-to-center 7 Deployment Groups with a Response 31 Table 12. TSP system architectures. Transit System Deployment Count AVL 17 APC 4 Farebox 1 None 12 Deployment Groups with a Response 30 NOTE: Respondents could select more than one answer. Table 13. Transit technology systems integrated with TSP.

Survey Results 29 100%, and the averages across all deployment groups were 45.8% near side, 42.6% far side, and 11.6% mid-block. Figure 8 shows the distribution of stop locations on TSP-enabled routes. (For example, 21 deployment groups had between 0% and 9% mid-block stops.) Transit agencies also reported how they handle priority requests at near-side stops. Twenty- four of 31 deployments (77%) use TSP at near-side stops, although 10 of the 24 have buses request TSP after leaving the stop to reduce the uncertainty related to dwell time (see Table 16). Of the 31 deployment groups, 12 had queue jumps for buses at some TSP-equipped inter- sections. Transit agencies with bus queue jumps reported the proportion of TSP-equipped intersections that also had queue jumps; responses ranged from 1% to 100%, with a median of 4%. The distribution is shown in Figure 9. (For example, seven deployment groups used queue jumps at 0% to 9% of intersections.) Of the 31 deployment groups, 15 had dedicated bus lanes along at least a segment of the bus routes included in the deployment. Transit agencies with dedicated bus lanes reported the proportion of the entire TSP-enabled corridor length that had dedicated bus lanes. The propor- tions ranged from 1% to 70%, with a median of 23%. The distribution is shown in Figure 9. Vendor Response Count Trapeze 4 INIT 6 ACS/Conduent/Orbital 3 Routematch 0 TripSpark 0 Clever Devices 3 ETA Transit Systems 0 GTT-Opticom 10 GTT-Infrared 1 EMTRAC 1 Other, specifya 4 I do not know 6 Total Responses from All Deployments 38 NOTES: Respondents could select more than one answer. aTwo of the “other” vendors reported were Multilek and Novax. Table 14. Vendors of onboard TSP system components. Vendor Response Count Econolite—ASC/2 or ASC/3 9 Econolite—Cobalt or EOS 6 Siemens—M50/M60 4 PEEK 6 D4 1 Siemens—NextPhase 1 McCain 2 Cubic/Trafficware/Naztec 2 SCATS or SCOOT 2 Other, specifya 6 I do not know 13 Total Responses from All Deployments 52 NOTES: Respondents could select more than one answer. a“Other” vendors reported included Intelight, Kimley-Horne KITS, Thompson Technologies, and GTT. Table 15. Vendors of signaling system components.

30 Transit Signal Priority: Current State of the Practice TSP Handling at Near-Side Bus Stops Deployment Count Yes, bus requests TSP before bus stop 14 Yes, bus requests TSP after leaving bus stop 10 No, TSP is disabled at near-side stops 6 Response missing 1 Total Deployment Groups 31 Table 16. TSP provision at near-side stops. 0 5 10 15 20 25 Percent of Bus Stops Near Side Mid-Block Far Side 0% –9 % 10 %– 19 % 20 %– 29 % 30 %– 39 % 40 %– 49 % 50 %– 59 % 60 %– 69 % 70 %– 79 % 80 %– 89 % 90 %– 99 % 10 0% N um be r of D ep lo ym en t G ro up s Figure 8. Distribution of stop locations. 0 1 2 3 4 5 6 7 8 Percentage of Corridor Queue Jumps Dedicated Bus Lanes N um be r of D ep lo ym en t G ro up s 0% –9 % 10 %– 19 % 20 %– 29 % 30 %– 39 % 40 %– 49 % 50 %– 59 % 60 %– 69 % 70 %– 79 % 80 %– 89 % 90 %– 99 % 10 0% Figure 9. Distribution of queue jump and dedicated bus lane provision.

Survey Results 31 (For example, four deployment groups had between 20% and 29% of TSP corridor miles with dedicated bus lanes.) TSP Performance Monitoring, Operations, and Maintenance This section presents the results from survey questions about TSP performance monitoring and operations and maintenance practices. Transit agencies were asked about their TSP performance monitoring practices; Table 17 summarizes their responses. The four most common practices for performance monitoring were related either to system diagnostics (quantity of priority requests granted and sent) or to system outcomes (travel time and schedule adherence). The agencies were asked how often they evaluate and update business rules. Most agen- cies do not have a regular update cycle and reevaluate only in response to reported problems (12 responses, or 39%) or have not yet done so (10 responses, or 32%), likely because the systems are still relatively new. Of the six agencies that do have a regular update cycle, three (10%) evaluate business rules every 1 to 3 years, and three do so every 4 to 6 years. Table 18 summarizes the metrics used to measure TSP benefits. Twenty-one of 31 deploy- ments (68%) reported that they quantitatively measure benefits. Travel time and schedule adherence were the most common metrics. Performance Monitoring Indicator AgencyCount Quantity of priority requests sent 14 Quantity of priority requests granted 12 Travel time 11 Intersection delay 7 Schedule adherence/on-time performance 11 Headway adherence 5 Travel time variability 5 Othera 1 Transit Agencies with a Response 24 NOTES: Respondents could select more than one answer. aThe “other” response suggested that the transit agency monitored communication latency and the average amount of priority granted. Table 17. Indicators used for performance monitoring. Benefit Evaluation Metric Deployment Count None 10 Travel times 13 Intersection delay 5 Schedule adherence/on-time performance 10 Headway adherence 3 Travel time variability 9 Headway variability 1 Other 6 Deployment Groups with a Response 31 NOTE: Respondents could select more than one answer unless they responded none. Table 18. TSP benefit measurement.

32 Transit Signal Priority: Current State of the Practice Lessons Learned, Benefits, and Challenges This section summarizes responses to survey questions about perceived benefits, perceived challenges, and lessons learned. Perceived TSP Benefits Agencies rated the impact of TSP on various facets of bus operations on a 5-point Likert scale, from “1—made significantly worse” to “5—made significantly better.” Between 24 and 27 responses were received for each factor; the results are summarized in Figure 10. Not surpris- ingly, agencies found that TSP had the greatest positive impact on intersection delay. In one deployment group (Sun Metro’s Brio), TSP reportedly contributed to worsening of headway adherence and headway variability, which is why the minimum value for these two benefits was 2 (i.e., “made somewhat worse”). The responses to the questions shown in Figure 10 were used to sort deployments into three benefit groups: 1—Lowest Improvement (average rating between 3.0 and 3.5); 2—Modest Improvement (average rating between 3.5 and 4.0); and 3—Most Improvement (average rating between 4.0 and 5.0). Missing responses were ignored when calculating average ratings. No deployment had an average rating below 3.0. Two deployments did not answer the benefit questions and are categorized as “NA” in Table 19. There were eight deployments in the lowest improvement group, eight in the modest improvement group, and 12 in the most improve- ment group. Table 19 cross tabulates the combinations of priority types by benefit group. Deployments in the lowest improvement group use only extended green or early green priority. Deployments with phase rotation, insertion, or skipping and special bus-only signals report higher benefits. A simplified chart that combines priority combinations into two groups (extended green or special signal only versus combination of priorities) is displayed in Figure 11. The results suggest that deployments with single priority types (e.g., extended greens only) were less likely to achieve operational improvements from TSP. Figure 12 shows the near-side TSP business rules by benefit group; however, results should be treated with caution given that near-side stops accounted for fewer than 20% of all stops for 3.4 3.5 3.6 3.8 3.9 4.1 0 1 2 3 4 5 6 Headway Variability Travel Time Variability Headway Adherence Schedule Adherence Travel Times Reduction in Intersection Delay Benefit Rating Be ne fit T yp e NOTES: The blue bars display the average rating for each benefit type; the error lines on the chart indicate the minimum and maximum ratings for each benefit type. Figure 10. Perceived TSP benefits.

Survey Results 33 Table 19. Benefit groups and priority combinations. Improvement Group Extended Only Special Only Extended + Early Extended + Phase Extended + Early + Special Extended + Early + Phase Early + Phase + Special 1—Lowest Improvement 5 3 2—Modest Improvement 3 1 1 3 3—Most Improvement 3 5 2 2 NA 1 1 Total 12 1 8 1 5 2 1 NOTE: NA = not available. Figure 11. Distribution of different priority type business rules across improvement groups. 0% 10% 20% 30% 40% 50% 60% 70% 80% Lowest Improvement Modest Improvement Most Improvement Extended Green or Special Only Combination of Priority Types Pe rc en ta ge o f D ep lo ym en t G ro up s NOTES: Based on 28 deployment groups that had benefit data and a specific priority type (e.g., extended green). The Extended Green or Special Only group included 12 deployments, and the Combination of Priority Types group included 16 deployments. Percentages are calculated using the total number of deployments within an improvement group. Figure 12. Distribution of different near-side stop priority business rules across improvement groups. 0% 10% 20% 30% 40% 50% 60% 70% Lowest Improvement Modest Improvement Most Improvement TSP is disabled Request made AFTER stop Request made BEFORE stop Pe rc en ta ge o f D ep lo ym en t G ro up s NOTES: Based on 29 deployment groups that had benefit data. Percentages are calculated using the total number of deployments within an improvement group. Chart should be interpreted with caution, given that most deployment groups have a very small percentage of near-side stops.

34 Transit Signal Priority: Current State of the Practice 24 of 31 deployment groups. Disabling TSP at near-side stops appears to be associated with lower benefits, while requesting TSP after leaving near-side bus stops appears to be associated with higher benefits. Deployments that request TSP before near-side stops appear to have mixed results. Perceived TSP Challenges Agencies rated a list of challenges related to TSP deployment on a 3-point Likert scale, with values of “1—not a problem at all”; “2—slight problem”; and “3—major challenge.” Depending on the question, between 21 and 23 responses were provided. Figure 13 presents the average “challenge scores” for each statement. As can be seen in Figure 13, challenge ratings varied across transit agencies; however, the two issues rated most challenging were measuring the TSP system’s benefit, and maintaining long-term TSP operations in the face of changing priorities and financial pressures. Transit agencies also perceived coordination with traffic signal owners, determining optimal business rules and parameters, and generating buy-in during procurement and installation as challenges. Lessons Learned Transit agencies also provided, in their own words, lessons learned about maintaining and operating TSP. A few themes emerged: • Coordinating responsibility and aligning priorities among transportation agencies are critical to TSP implementation and success. 91% 57% 54% 52% 39% 35% 30% 22% 17% 5% 30% 25% 35% 57% 48% 57% 48% 50% 5% 13% 21% 13% 4% 17% 13% 30% 33% 0% 50% 100% Coordinating with other transit operators that use the corridor Adjusting bus schedules to account for TSP Maintaining TSP hardware/software onboard vehicles Implementing and operating the system according to desired business rules and parameters Generating buy-in with all stakeholders to begin procurement and implementation Determining the optimal business rules and parameters to maximize benefit Coordinating with jurisdiction(s) that manage the traffic signals Maintaining long-term TSP operation in the context of changing priorities, budget pressures, etc. Measuring the TSP system’s benefit Not a Problem at All Slight Problem Major Problem Figure 13. Perceived TSP challenges.

Survey Results 35 • Constant monitoring of software and hardware advances is important to keep TSP systems up to date. • Selecting the appropriate business rules and parameters is critical and takes time to refine. • Evaluating TSP systems early and often is important to ensure TSP systems are producing the desired benefits. (This includes having reliable pre-TSP data to serve as a baseline.) Capital and Operating Costs Transit agencies also provided information about the cost of their TSP systems, including costs for intersection equipment and software, transit vehicle equipment and software, and ongo- ing operating and maintenance. This study found that transit agencies had difficulty providing implementation (capital) cost information—either because TSP system costs were lumped into other projects, or because implementation cost data were no longer available. Also, different TSP deployment groups included different components as a cost of implementation (e.g., adding roadside PRGs versus implementing a center-to-center software change). Capital (Implementation) Costs Eight transit agencies provided cost estimates for the equipment, equipment installation, and software for transit agency vehicles and back-end transit management systems to implement TSP. The provided costs were divided by the number of buses in the deployment group. The cost per bus to implement TSP ranged from $4,484 to $32,033, with a median of $8,133. Five of the agencies’ per-bus costs were $10,000 or less. Ten transit agencies provided a cost per intersection for signal equipment and software to implement TSP. Costs varied significantly, from a low of $2,350 per intersection to a high of $100,000. (One respondent indicated a per-intersection cost of $300,955, which is likely an error and was excluded from this analysis.) According to nine seemingly valid estimates, the average per-intersection cost ranged from $2,350 to $65,000, with a median of $19,415. Seven of the agencies’ per-intersection costs were $20,000 or less. Operating Costs Three transit agencies provided a specific annual budget for TSP to support ongoing TSP system operations and maintenance; responses ranged from $500 to $1,000 per TSP-equipped intersection.