Is Transit Useful? Key Indicators

2 Is Transit Useful? Key Indicators

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 30 Ridership Arises from Usefulness To maximize liberty and oppor-

mo reland Ridership is an important goal for many reasons other than fare tunity for the greatest number of From 4th St and Oregon Ave, easton n to e l t

ÀiÛiÕi°Ìi>ÃÕÀiÃ Ü>Þ«i«iLiiwÌvÀÌ iÃiÀÛVi]>` b u

e b people, a network of routes must where could It travel to on weekdays at 12 pm?

l h i philmont how effective it is at providing alternatives to the private car and other m e pe nn byberry m sy e lv k

optimize: i a n l i n a old york w iÃÃÃ«>ViivwViÌ`iÃ° e lsh red lion

n o d chu g 1. Frequency, n i roosevelt Public transit ridership arises mostly from service that is useful to as rch t n rd u h many people as possible, for as many trips as possible. There are some haldeman woodhaven ll i easton 2. Span, m r e od p wo }iiÌÀVv>VÌÃ>LÕÌ ÜiÌÜÀÃ`Ì Ã]Ü V Ì ÃV >«ÌiÀLÀiyÞ a n p e grant re ton ee g g welsh in verr h y germantown ge s township line reviews. 3. Connections, rm a m w e an d to stenton a wn c a

d t

r l

o e f v grant

e x s

A helpful way to illustrate the usefulness of public transport is to visual- o o 4. Speed, knights rhawn o ridge r s te rd ize where a person could get to using public transit and walking from a nt cheltenham on o ho frankfo s mount airy g lme n o

e n 5. Reliability, and t ll a z given location. Figure 28 shows an isochrone map of the access avail- castor

nt mo levick able from 4th Street and Oregon Avenue at noon on a weekday. From aw godfrey h henry 6. Capacity, s n su e u tat the selected point, the map shows where someone could be, on public m ng s i a b is bustleton r r d ia a washington m and follow favorable patterns in road s transit combined with walking, in 15, 30, 45, or 60 minutes. b een ridge gr wayne n

b the built environment. r

h r A more useful service is one in which these isochrone areas are larger, b o d t r s r i a d c g rid hunting park e tacony ÃÌ >Ìi>V «iÀÃÃiÞÌw`Ì iÃiÀÛViÕÃivÕvÀÀiÌÀ«Ã°7i belmont ge frankfo montgomery roosevelt erie can quantify this by counting how many destinations—jobs, shopping, KEY INDICATORS IS TRANSIT USEFUL? hunting park The rest of this chapter and Chapter 3 will focus go sedgley in am etc.—are inside these areas, and how those numbers grow or shrink ar Ì iÃÝiiiÌÃ`iÌwi`>LÛi° >«ÌiÀ city e lehigh allegheny iv r depending on the network design. d lly e 4 will focus on the patterns of development lancaster k richmond d Even beyond usefulness, an isochrone shows the level of personal in Philadelphia. Together, network design and 33r

52nd

broad development patterns will illuminate the geo- haverfo girard freedom and opportunity afforded by the public transport network. If rd girard iÌÀVv>VÌÃÌ >Ì`iwiÌ i«ÃÃLÌiÃvÀÌ i someone at 4th Street and Oregon Avenue chooses to rely on transit, spring ga rden Philadelphia transit network. powelton vine they will be constrained by where transit can readily take them, and chestnut

th spruce ll 16 marsha market will experience the blobs in this map as walls around where they can re baltimo lombard go and what they can do. For someone to choose to rely on transit, lombard

c whitby h

34th i

s ore t

altim o

and especially for them to decide to not own a car or to share a car, b p

r c these blobs have to contain enough of the places that make people’s rgh o Areas that can be lu m s main pr macdade indbe b ing land l u lives complete like education, shopping, services, social opportunities fie yunk s accessed within: ld pass wood island oregon and jobs. 26th 15 min

essington

Expanding this freedom and opportunity is foundational for rider- broad rose 30 min pen Ã «}ÀÜÌ °7 iÌ iÀv>VÌÀÃ>Ã>vviVÌÀ`iÀÃ «]Ì Ãi>ÃÕÀiv 84th freedom and opportunity isolates the aspect of ridership that network 45 min

`iÃ}V>yÕiVi°/ ÃV>VÕ>ÌV>Li>`iÜÌ ÕÌÀiVÕÀÃi bartram to assumptions about human culture or behavior, whose effects are not 60 min predictable. Data source: SEPTA GTFS Spring 0 1 2 3 4 miles 2017

How can we help more people get to more places sooner, so that they V>`ÀiÌ }Ã¶ Figure 28: Map of travel time isochrones for walking and transit from 4th St and Oregon Avenue.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 31 Despite smartphones, realtime information and other new technolo- Ü vÀiµÕiVÞ ÀÕÌiÃ] i ,ÕÌiÃ £] Èn >` ÇÇ] Ìi` Ì >Ûi Ì i Frequency Comes First gies, frequency still matters enormously because: lowest productivity. Route 27, despite being a 30-minute frequency route, has the lowest productivity at 17.6 boardings per service hour Riders respond to many features of a service, including speed and reli- • 7>Ì} `iÃ Ì ÕÃÌ >««i >Ì Ì i ÃÌ>ÀÌ v ÞÕÀ À`i] Ì >Ã on weekdays. ability, but the dominant factor is frequency. Frequency is the elapsed >««iÃ>ÌÌ ii`°9Õ>ÞÌii`Ìi>ÛiÌ i ÕÃiÕV time between consecutive buses (or trains, or ferries) on a line, which before your departure, but if your bus is infrequent, you have to The relationship between route productivity and frequency and outlier determines someone’s maximum waiting time. choose between being very early or too late. cases will be further explored in Chapter 5.

People who are accustomed to traveling by car often underestimate • Many of the places we go do not let us hang out until our bus’s the importance of frequency, because there is not an equivalent in >ÀÀÛ>ÃiÌ°7iV>i>ÃÞ`Ì ÃÜ ii>Û} i]LÕÌ driving. A car is ready to go when you are, but public transit is not it is more awkward when leaving a restaurant or a workplace that available until it comes. is closing.

High frequency means public transit is coming soon, which means 7iV>ÃiiÌ iivviVÌvvÀiµÕiVÞÀ`- that it approximates the feeling of liberty you have with a private ership by looking at how existing services 46 vehicle—namely that you can go anytime. Frequency has three inde- perform. 11 «i`iÌLiiwÌÃvÀÌ i«>ÃÃi}iÀ\ Figure 29 shows a dot for each SEPTA bus 52 6 • Frequency reduces waiting, which is everyone’s least favorite and trolley route within Philadelphia, with 34 3 part of a trip. The basic sensation of being able to go when you the horizontal axis indicating frequency 36 75 want to go is the essence of frequency. and the vertical axis indicating productivity, 60 60 56 15 8 which is ridership divided by the quantity XH • Frequency makes connections easy, which makes it possible for 66 47 54 of service provided. High frequency means 70 KEY INDICATORS IS TRANSIT USEFUL? a cluster of transit lines to become a network. A transit line without 33 17 18 64 a short elapsed time between consecutive 16 good connections is useful for travelling only along that line. A trips, so it is to the left on these diagrams. 21 26 K H 31 network of frequent lines can make it easy to travel all over the Service quantity is measured in hours, where 23 48 29 city. This massively expands the usefulness of each line. 4 a service hour represents one bus on the 57 R road for one hour. 7 • Frequency is a backup for problems of reliability. If a vehicle 42 G L 65 breaks down or is late, frequency means another will be along 40 39 25 The highest frequency routes, like Routes 11, 45 22 43 soon. 79 30 52, and 23, tend to be the most productive. 2 53 58 20 61 Route 11 is the highest productivity route of 5 J Many people assume that today, with real-time transit arrival informa- 32 73 all with 73.6 boardings per service hour on 14 44 84 67 tion and smartphones, nobody needs to wait for a bus anymore, and 40 12 weekdays. 55 24 28 19 1 frequency therefore does not matter. If a bus only comes every 30 38 37 ÕÌiÃ]Ì >ÌÃwi]LiV>ÕÃiÞÕÀ« iÜÌiÞÕÜ iÌÃ>viÜ 15B 50 Productivity (Boardings per Revenue Hour) 47M 68 minutes away and you should start walking. 89 77 10 9 88 20 27 35

51015 20 30 45 60 Midday Frequency (minutes)

Figure 29: Frequency and productivity relate for SEPTA bus routes in Philadelphia.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 32 Figure 30 shows a the similar relationship between productivity and Productivity and Frequency Relate vÀiµÕiVÞ vÀ Õ`Ài`Ã v ÀÕÌiÃ ÓÎ ÀÌ ƂiÀV> VÌiÃ «ÕÃ Data from 23 cities Philadelphia routes (shown with green dots). The individual points for the other 23 cities are grouped into hexagonal bins, to prevent them 80 obscuring one another; darker-colored hexagons contain more routes

In Philadelphia and beyond, higher frequency is generally associated with higher productivity. The larger dataset shows the pattern very strongly, including the upward curve indicating an exponential SEPTA routes payoff of very high frequencies. More frequent services tend to have are marked in green. higher productivity (ridership per service hour), even though providing high frequency requires spending more service hours. 60

7 i> } iÀvÀiµÕiVÞVÀi>ÃiÃÌ i`i>ÌÀvÌ i«À`ÕVÌÛ- ity ratio, the higher ridership more than makes up for it. Count of Routes This is how we know that high frequency contributes to high ridership, rather than simply representing a responsive transit agency that 30 raises frequency where ridership is high. If higher frequencies were not causing higher ridership, then the dots on this chart would be a 40 20 y>Ì ÀâÌ> VÕ`] ÃÌi>` v > VÕÀÛi Õ«Ü>À` Ì Ì i ivÌ° 7 i > 10 transit agency increased the frequency on a route, its ridership would increase proportionally, and its productivity would remain unchanged. KEY INDICATORS IS TRANSIT USEFUL? Instead, higher frequency is associated with higher productivity. Productivity (Boardings per Revenue Hour) This happens because frequent service is the most useful and conve- nient service for riders; thus, transit agencies typically target this most iÝ«iÃÛi ÃiÀÛVi ÌÜ>À`Ã Ì iÀ ÃÌÀ}iÃÌ >ÀiÌÃ° 7 i vÀiµÕiÌ 20 service is available to people in a suitably dense, walkable environment, high ridership is a common result. This illustrates the power of frequency to deliver more ridership than would be expected by the increase in service hours.

0 15 30 45 60 75 90 105 Midday Frequency (minutes)

Figure 30: Higher frequency is generally associated with higher productivity. Hex plot of routes in 23 North American cities and Philadelphia.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 33 Span: Is transit there when you Weekday need it? 12am 3am 6am 9am 12pm 3pm 6pm 9pm 12am 52 For transit to be useful, it must be there at the time of day that you need it. The times of day that transit operates is called the span of 13 service. In Philadelphia, many routes provided by SEPTA run late into 11 the evening and start early in the morning. But the frequency of service varies greatly by time of day. 6 The network maps on pages 20–26 show the frequency through- 34 out the midday. Frequencies are generally even better during the peak. Headway (in ÕÌÃÌÀÕÌiÃ]Ì ivÀiµÕiVÞ`À«ÃÃ}wV>ÌÞÌ iiÛi}° 36 This is a normal response to a drop in demand, but because frequency 0-4 is so important for quick and reliable connections, this can cause a 66 very abrupt drop in usefulness. The freedom and liberty that transit 5-9 provides often arises from using two frequent routes together, so the 45 value is only present when both are running frequently. 15 10-15 of the role of frequency outlined above, this can cause a very abrupt 15B drop in usefulness. 47 16-25 KEY INDICATORS IS TRANSIT USEFUL? Figure 31 shows the 20 routes with the longest spans of high frequency Route service. Only four routes have a 15-hour span (e.g., 6 am to 9 pm) of 10 frequent service. Frequent service begins in the 6 am hour for most 26-40 routes. But there is no consistency in when frequent service ends in the 58 evening. They also differ in the beginning and ending times of service. 17 41-64 The lack of consistency in when frequent service transitions to less frequent service means that riders must consult a schedule for trips in the 33 iÛi}ÀvÌ iÞ>ÀiÌÀ>Ûi}Üiii`Ã°7 iÌ iÞ>Þ >Ûi> 65+ quick trip with a minimal wait in the morning, their return trip may be 23 much longer because their transfer may involve a much longer wait. 42

14 12am 3am 6am 9am 12pm 3pm 6pm 9pm 12am

Figure 31: Frequency and span of service for the 20 SEPTA routes with the longest spans of frequent service. Only four routes have frequent service for 15 hours a day or more.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 34 Frequent service is highly inconsistent on weekends as well. Figure Service levels on Sundays are even lower than Saturdays. Only eight 32 shows the same 20 routes with their span and frequency of service routes have any period of service that is frequent, and only Route 52 on Saturday and Sunday. Of these 20, 19 routes have some period has a period of frequent service more than eight consecutive hours. of frequent service on Saturdays, but most have fewer than 12 hours of frequent service. And there is little consistency in when frequent service begins and ends for each route. Route 52 is frequent for most of the day on Saturday, from 8 am to 9 pm. But as a crosstown route it would be even more useful if the connecting radial routes, like the trolleys or Routes 21 and 42, had frequent service in the mornings and evenings.

Saturday Sunday 12am 3am 6am 9am 12pm 3pm 6pm 9pm 12am 12am 3am 6am 9am 12pm 3pm 6pm 9pm 12am

52 52

13 13

11 11

6 6

34 34 Headway (in minutes)

36 36 KEY INDICATORS IS TRANSIT USEFUL? 0-4 66 66

45 45 5-9

15 15 10-15 15B 15B

47 47 16-25 Route Route

10 10 26-40 58 58

17 17 41-64

33 33 65+ 23 23

42 42

21 21

56 56

14 14 12am 3am 6am 9am 12pm 3pm 6pm 9pm 12am 12am 3am 6am 9am 12pm 3pm 6pm 9pm 12am

Figure 32: Frequency and span of service on Saturday and Sunday for the 20 SEPTA routes with the longest spans of frequent service. Most routes are not frequent on Sunday.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 35 Frequent Service Standards Figure 33: Frequent service standards in four North American cities An increasing number of transit agencies use standardized spans for FREQUENT SERVICE STANDARDS all frequent routes, usually in the context of branding their frequent FREQUENT WEEKDAYS SATURDAYS SUNDAYS BRANDING network to help people see where transit is most likely to be coming 5 67891011 12 1 234567891011 12 5 67891011 12 1 234567 8910 11 12 5 6 718 9 10 11 12 2345 678910 11 12 AM PM AM AM PM AM AM PM AM soon. Figure 33 provides some examples of the frequent service stan- FREQUENT `>À`ÃÕÃi`LÞÌ iÀ>ÀÌÀ>ÃÌ«ÀÛ`iÀÃ ÀÌ ƂiÀV>°*ÀÌ>` HOUSTON, TX uses a standard of 15-minute frequency or better for at least 14 hours NETWORK on weekdays. Others provide at least 15 hours. These agencies also maintain frequent service on weekends for these branded routes, at INUTE 10-M MONTREAL, QC higher frequency and longer spans than SEPTA does. MAX >Þ>}iViÃ>Ài>Ã`iÛi«}Ã«iVwV>«Ã>`LÕÃÃÌ«Ã}ÃÌ brand and market their frequent service. Figure 34 shows the frequent RAPIDRIDE SEATTLE, WA network map and an example of a bus stop served by the frequent network in Portland. FREQUENT PORTLAND, OR The limited nature of the frequent network in Philadelphia is well illus- SERVICE trated by the set of maps in Figure 35 on page 37. At noon on weekdays, the frequent network covers much of the city. By 7 pm, FREQUENCY however, the frequent network is severely limited. By 9 pm, the fre- MINUTES BETWEEN BUSES under 5 5 - 9 10 - 15 16 - 25 26 - 40 over 40

µÕiÌiÌÜÀVÃÃÌÃvÕÃÌÌ i -] >`,ÕÌixÓ°/ ivÀiµÕiÌ KEY INDICATORS IS TRANSIT USEFUL? network at noon on Saturdays is similar to the weekday network, but less comprehensive, while at noon on Sundays it is quite limited.

Evening and weekend service is relatively inexpensive to operate compared to peak period service, and it is also crucial to a large segment of transit riders. People who work in most retail and entertainment sectors have to work on weekends and often late into the evening. Having some transit then is important to making it possible for them to rely on transit at all. Figure 34: Po r t l a n d ’s Houston recently had great success with a network redesign that TriMet is one of many extended evening service and expanded Saturday and Sunday service agencies that promotes to be the same level as weekday service, but without the peak period. the Frequent Network Ì iwÀÃÌÞi>ÀvÌ iÀiÜiÌÜÀ]->ÌÕÀ`>ÞÀ`iÀÃ «VÀi>Ãi`£Î as a distinct product percent and Sunday ridership increased 34 percent. with bus stops (left) and a map (right). 7iÜÕ`iÌÀiVi`«ÀÛi`vÀiµÕiVÞÌ iiÛi}>` on weekends for Philadelphia as part of its network redesign process, but without new resources it would require cutting the weekday network too deeply. Bringing evening and weekend service up to a similar standard to the examples noted above would increase annual costs by about 15 percent . Additional evening and weekend service should be a top priority for any new resources.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 36 Weekday Noon Weekday 7 pm Weekday 9 pm

0 1 2 3 4 miles 0 1 2 3 4 miles 0 1 2 3 4 miles

Routes with 15 minute Routes with 15 minute Routes with 15 minute or better frequency or better frequency or better frequency IS TRANSIT USEFUL? KEY INDICATORS IS TRANSIT USEFUL?

Figure 35: Availability of frequent service at different times of day on weekdays and weekends. The frequent network Saturday Noon Sunday Noon Ã ÀÃÃ}wV>ÌÞÜii`>ÞÃ>vÌiÀÇ«°

The scope of the frequent network is relatively broad during weekdays and Saturdays at noon. But it is very limited in the evenings and on Sundays. 0 1 2 3 4 miles 0 1 2 3 4 miles

Routes with 15 minute Routes with 15 minute or better frequency or better frequency

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 37 levels and the number of stops. An unfortunate reality is that increased feet. This is consistent with the current SEPTA stop spacing standards Connections ridership will tend to decrease average speeds, as longer dwell times (detailed further on page 93) but it is not in line with current best / i- */ƂLÕÃÃÞÃÌi* >ì« > >Ã>ivwViÌ}ÀìÌÜÀ are required to board and alight more passengers. But ridership has practice for bus stop spacing, which is to space stops as follows: across most of the city that provides easy connections between most declined on SEPTA since 2014, so one would expect, absent other • Space stops more widely in dense areas, where there are likely to routes and between buses and other transit modes. This structure and changes, to see an increase in average speed. Thus other factors, like be people waiting at every single stop, and where it is relatively will be discussed more thoroughly in Chapter 5 on page 69. }iiÀ>ÌÀ>vwVV}iÃÌ]ÕÃÌLiÃÜ}`ÜLÕÃiÃ° safe and easy to walk a few blocks to a bus stop. Another factor that slows buses is the density of signals and stop signs. Buses are Getting Slower Figure 37 shows the density of signals and stop signs along major • Space stops more closely in low-density, outlying areas, where ÀÌ qÃÕÌ ÀÕÌiÃÌ ÀÕ} ÀÌ ]-ÕÌ >` iÌiÀ ÌÞ* >ì« >° ridership is lower and so it is unlikely that there will be people SEPTA bus speeds in the City of Philadelphia are typical for a dense Route 45 trips coming into Center City have 29 signals and 17 stop waiting at every stop. The closely-spaced stops therefore do not and congested urban area, but speeds have been declining. As shown signs over about four miles, averaging eleven signals or stops per mile. ÃÜ`ÜLÕÃÃiÀÛVi°7>>LÌÞÃiÞÌLiÜÀÃiÕÌÞ} in Figure 36, below, scheduled speeds have declined since 2014 and areas, so more closely spaced stops also reduce the impact on average less than 12 miles per hour during most of the day. Many In addition, bus stop spacing within the city is typically every block at passengers who might have walk to on streets without sidewalks, v>VÌÀÃ>vviVÌLÕÃÃ«ii`Ã]VÕ`}ÛiÀ>ÌÀ>vwVV}iÃÌ]ÀìÀÃ « the near-side of intersections. This results in a stop about every 500 crosswalks and lighting.

Figure 36: Scheduled bus speeds by time of day, 2014–2017. Bus speeds have declined since Figure 37: The high density of signals and stop signs along Routes 17, 23, 33, 45 and 47 are 2014 and average less than 12 mph during most of the day. major factors in the slow speed of these routes.

17 23 33 45 47

12.5 IS TRANSIT USEFUL? KEY INDICATORS IS TRANSIT USEFUL? Stop Type All Way Signalized

12.0 Time of Day AM Peak 6:30-10am Midday 10am-2pm Midafternoon 2-4pm 5 11.5 PM Peak City average speed (MPH) City average 4-6:30pm

Evening stops per mile Signals or allway 6:30-9pm

11.0 0

In Out In Out In Out In Out In Out 2015 2016 2017 Year Direction

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 38 On-Time Performance Is Poor Figure 38: A majority of SEPTA routes are not achieving the 80% on-time performance standard (conventional measure). Reliability of bus service can be measured in a few For frequent services, this measure of on-time performance is not what matters. different ways but is commonly measured by either on-time performance or headway reliability.

On-time performance measures the percent of Conventional On-time Performance trips that arrive within a given window around the 90% ÃV i`Õi` >ÀÀÛ> Ìi° - */Ƃ ÃÌ>`>À`Ã `iwi Ì i “on-time” window as 0 minutes to 5 minutes and 59 seconds late. By this standard, a bus that arrives SEPTA OTP Standard: 80% at a timepoint 6 minutes behind schedule would be 80% counted as late. A bus that arrives 5 minutes behind schedule would not be considered late.

/ Ã`iwÌÃÀi>ÌÛiÞÃÌÀVÌLÞÌ iÃÌ>`>À`Ãv 70% US transit agencies. Any bus that arrives even one minute early is considered not on-time and would count against this metric. More typical standards allow a window of up to one minute early before counting 60% an arrival as not on-time.

Figure 38 shows the on-time performance of most KEY INDICATORS IS TRANSIT USEFUL? SEPTA routes in Philadelphia. SEPTA performance 50% standards call for buses and trolleys to achieve 80% Ìi «iÀvÀ>Vi° ƂÃ Ì i w}ÕÀi Vi>ÀÞ Ã ÜÃ] the majority of routes are not achieving this standard 40% today. 54 of the 83 routes analyzed do not meet this standard.

But this raises a critical question about the value of this metric for most SEPTA routes in Philadelphia. 30% Route 47, for example, achieves only 73% on-time «iÀvÀ>Vi°9iÌÌ iÀÕÌiÃÃV i`Õi`Ì«iÀ>Ìi>Ì least every 10 minutes or better from about 6 am to 6 Percent of trips on-time (less than 6 minutes late) 20% «°7 ii>ÃÕÀ} } vÀiµÕiVÞÃiÀÛViÃ]v£ä or 15 minutes or better, earliness and lateness relative to the schedule is usually a poor measure of reliability. Earliness and lateness matter if somebody is really 10% going to expect the bus at 10:10 am. But on high- frequency services almost no one does that. They just go out to wait for the next bus and trust that it will be

along soon. So the thing most customers care about 0% 6 46 59 73 79 53 19 MFO 29 30 BSO 5 28 35 39 45 60 88 L 13 15B 8 20 24 70 36 54 67 11 18 H 3 34 50 68 XH 14 33 37 56 64 66 J 48 R 10 40 55 58 75 84 G K 7 12 2 78 1 22 43 47 52 26 31 57 89 38 4 21 23 25 65 15 17 77 32 42 47M 61 16 27 62 44 9 80 is the maximum wait time, not earliness or lateness compared to the schedule.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 39 Judging service on earliness and lateness is an easy measure to apply Better Measures of Reliability to vehicles. This means that operations management has an easier Suppose that at a particular stop on a particular day, buses are sched- time monitoring performance on a route and dealing with disruptions. uled to come every 10 minutes as shown in Figure 39. There are three Changing to a headway reliability standard would shift the onus to ways to describe this situation: active management because success is the result of teamwork rather than individual vehicle performance. Changing work and management • Conventional on-time performance, also called schedule adher- habits is a major cultural shift and bus operations managers may not ence, would be 14%, because only one of the seven buses is on have the training, or the tools, to do it that way. However, many transit time. The others are all six minutes late. agencies are seeking to make this transition, because the conventional on-time performance measure just does not describe what matters in • Headway regularity would be 86%. Only one of the buses is more frequent bus operations. than 10 minutes behind the one in front of it.

• Probability of waiting no longer than the scheduled 10 minutes would be 90%. If people appear evenly throughout the 60 minutes of the hour, only 10% of them will arrive at a time when the next bus is more than 10 minutes away.

Headway regularity is about how often a headway (the elapsed time between consecutive buses) is what it should be. Probability of waiting Minutes Actual Minutes Between Departures Behind translates this into the customer experience: How likely am I to be Departures Schedule >vviVÌi`LÞ>`i>Þ¶ The customer experience of reliability is better 10:00 +0 KEY INDICATORS IS TRANSIT USEFUL? 7Ì ÀiViÌ Õ«}À>`iÃ Ì Ƃ6 >` VÕV>ÌÃ ÃÞÃÌiÃ] - */Ƃ than conventional on-time performance stan- 16 now has the technical capabilities to track and manage the reliability dard makes it look. The key is to look at the of frequent service in this way. Changing to a new standard, however, 10:16 -6 probability of waiting more than a certain time. ÜÕ`ÀiµÕÀiÃ}wV>Ì>`ÕÃÌiÌÃ«VÞ>`Ì iVÕÌÕÀiv Ü 10 Conventional on-time performance: 14% operators, supervisors, dispatchers and others cooperate in the daily 10:26 -6 Only 1 of 7 buses are on time. management of service. 10 Headway regularity: 86% A key factor in changing this policy is that the individual driver cannot 10:36 -6 Only one bus is more than 10 minutes from be held responsible for it. A driver is only responsible for when her 10 the one ahead. bus gets to the stop, but she obviously does not control the buses in 10:46 -6 front of her and behind her. But an aggressive operations manage- 10 Probability of waiting no more than the ment, using current tools such as GPS to tell them where every bus scheduled time: 90% -6 is, can monitor these things, giving the necessary real-time direction 10:56 The next bus comes in 10 or less minutes for to drivers about how to adjust their operations to keep buses more 10 all but 6 minutes of the hour. evenly spaced. 11:06 -6 An example of more aggressive operations management would be Figure 39: Different ways of describing reliability. to more regularly instruct operators to switch to drop-off only when they start falling behind the regular headway of the bus ahead of them and the bus behind them is approaching quickly. This will upset some riders, who will get skipped by that bus, but the overall system will function better as the aggressive management of headways will reduce bus bunching and even out passenger loads.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 40 Probability of Waiting at SEPTA Figure 40 describes the reliability of frequent routes in the SEPTA system Wait in multiples of the headway 6% LÞ>º«ÀL>LÌÞvÜ>Ì}»ÃÌ>`>À`°/ iw}ÕÀiÃ ÜÃÃÞÃÌiÜ`i 1.5–1.9 performance for frequent routes (those that average every 15 minutes 2–2.9 or better) during different time periods of the day, because those are the routes where this standard is relevant. The analysis is limited by the 3–3.9 small number of days included in the analysis and missing vehicle loca- >= 4 tions, but it helps illuminate where reliability problems are greatest.

Unlike the abstract example above, we use a standard of 1.5 times the scheduled headway. Thus, if the scheduled headway is 10 minutes, a wait of up to 15 minutes would be considered acceptable. This helps 4% ÀiyiVÌÌ iÀ>Û>À>LÌÞ`ÕiÌÌÀ>vwVÃ}>Ã]ÃÜL>À`}Ã]iÌV°] that is impossible to prevent even in the most optimized network.

The probability of waiting at least 1.5 times the headway is more than 6% from 4:00–6:30 pm and more than 5% from 2:00–4:00 pm1. For trips in the PM peak, a rider should expect to wait at least 1.5 times the headway one or two workdays a month. A delay of twice the headway or worse could be expected every other month.

The results of this “probability of waiting” analysis suggest that per- KEY INDICATORS IS TRANSIT USEFUL? 2% formance is much better than one might think when reviewing the ÀiÃÕÌÃvÌ ivwV>- */Ƃ"/i*iÀvÀ>Vi>>ÞÃÃ}ÕÀi 38. SEPTA is achieving better than 95% headway reliability on high

frequency routes in the AM, Midday, and Evening. Probability of waiting X times longer than headway

iÛiÀÌ iiÃÃ] «iÀvÀ>Vi VÕ` Li «ÀÛi`° / i * *i> >` Mid PM (or afternoon) reliability levels are lower than many customers might like. Imagine being a shift worker trying to get home to pick up ÞÕÀV `vÀ`>ÞV>Ài°>Ìi«VÕ«ÃÀiÌ >ÌÜVi>Ì ÜÕ` likely get your child dropped from typical day care centers. This forces 0% riders to leave earlier than they want so they can avoid a missed pick up—or a missed shift. It adds wasted time to their day, takes away from 11:30pm– 6:30– 10:00am– 2:00– 4:00– 6:30– time with family, and takes away some of their freedom to do what they 6:30am 10:00am 2:00pm 4:00pm 6:30pm 9:00pm wish.

Chapter 3 describes key steps that could be taken in this direction.

Figure 40: SEPTA’s frequent routes are performing relatively well by the headway reliability standard. But reliability is low in the afternoon. 1 These headway reliability measures probably underestimate the real impact on customers. Many scheduled time-points were missing from the data. For missing timepoint crossings it is impossible to know if they were missing because the bus did not make that trip or because the Ƃ6ÃÞÃÌiv>i`Ì`iÌiVÌÌ iÌiÌ >Ì>LÕÃVÀÃÃi`>Ìi«Ì°-Vii>V ÃÃ}LÃiÀÛ>- >VÕ>Ìi`>`ÃÕ>Àâi`vÀÜii`>ÞƂ6Ìi«ÌVÀÃÃ}ÃvÀ >ÞnÌÕi£È]Óä£ÇvÀ i>`Ü>ÞÃÃV i`Õi`>>ÛiÀ>}iv£xÕÌiÃÀLiÌÌiÀÜÌ >>ÝÕ i>`Ü>ÞvÓäÕÌiÃ° tion causes uncertainty in the actual headway experienced at a stop, we dropped the affected headways from the analysis. Detection failures may be especially common when on-time performance is poor, introducing bias into reliability measures.

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 41 All door boarding

There are a range of things that can be done to improve bus speed and reliability. Most of these require cooperation with the City of Philadelphia, but there is one important area where SEPTA can lead.

Many transit agencies are looking at ways to reduce the amount of time that buses dwell at stops. A potential solution is to allow customers to board at any door, rather than requiring all customers to board through the front and interact with the driver. San Francisco does this on all bus services citywide, and many other big-city transit agencies are using it on major lines or Bus Rapid Transit services. It is nearly universal on light rail; in fact, SEPTA’s trolleys are one of the few rail services that does not use the method.

The key challenge of all-door boarding is fare enforcement. Staff must be assigned to randomly ride buses and check fares, issuing citations to those who are not holding a valid proof of payment. Fare inspection Figure 41: San Francisco’s all-door boarding reduced boarding time by is a challenging job that is part law enforcement and part customer ensuring that all doors are in use. Without all-door boarding, passengers service. Careful management is required to ensure it is done equitably, must use the front door while the rear doors are idle. Photo: SFMUNI LÕÌ>ÞVÌiÃ>Àiw`}Ì >ÌÌ iLiiwÌÃv>`ÀL>À`}>Ài worth the effort. KEY INDICATORS IS TRANSIT USEFUL?

1 The Trolley Modernization project also envisions new longer trolleys and Vancouver. San Francisco’s dwell time per boarding passenger that will also use all-door boarding, so fare enforcement will be needed dropped by 38% when all-door boarding was introduced. This trans- there as well. Since SEPTA will be expanding its fare enforcement lated into a 2% average increase in speeds, which would amount to capacities anyway, it should consider extending all-door boarding ÀiÛiÀÃ}Ì i>ÃÌÌ ÀiiÞi>ÀÃv- */Ƃ½ÃÃ«ii`ÃÃiÃ° iiwÌÃÜÕ` across much of the bus system. be highest on the busiest corridors and times of day. All-door boarding iÜ9À½Ã-iiVÌ ÕÃ-iÀÛViÀÕÌiÃV>ÕÃi``ÜiÌi«iÀL>À`- On some systems, particularly those with few stops and relatively high ing passenger to drop by 42-51%. levels of infrastructure, ticket machines on platforms can further reduce the need to pay cash, and some systems, including most light rail All-door boarding also reduces the variability in boarding time. This lines, require tickets to be purchased in advance. This approach may translates into more evenly spaced buses, which reduces bunching be appropriate for trolleys, Direct Bus services, and stops with high and thus more evenly distributes customers across buses. numbers of cash boardings throughout the system. SEPTA is already studying all-door boarding for Boulevard Direct But ticket vending machines are not essential. In the simplest version of service, but the agency should also consider the evolving experience all-door boarding, customers board at the front door only if they need on local buses, especially San Francisco’s and consider extending the to pay cash. SEPTA Key validators would be installed on rear doors so idea to major routes across the city. that passengers using the card, or carrying some other prepaid fare, could use any door.

/ i >Ì> ƂÃÃV>Ì v ÌÞ /À>Ã«ÀÌ>Ì "vwV>Ã Ƃ /"® recently released a white paper on the subject, looking at the experi- £ Ƃ /"]º iÌÌiÀ >À`}] iÌÌiÀ ÕÃiÃ\-ÌÀi>} >À`}>`>ÀiÃ°» ÌÌ«Ã\ÉÉ>VÌ°À}É iViv->À>VÃV] iÜ9À]-i>ÌÌi] V>}]ƂÕÃÌ] i>«Ã] Ü«VÌiÌÉÕ«>`ÃÉÓä£ÇÉäÓÉ Ƃ /"Ú iÌÌiÀ ÕÃiÃÚ >À`}°«`v

JARRETT WALKER + ASSOCIATES SEPTA Philadelphia Bus Network Choices Report | 42