Of the TTC’s many routes, 501 Queen is the longest and the subject of ongoing complaints about service quality and reliability. Two standard explanations are offered to the long-suffering riders: we cannot operate a reliable service thanks to traffic congestion, and we have no equipment with which to operate more service.
I have published detailed reviews of individual months of operation in past articles, but an accumulation of data for various periods and conditions now makes a retrospective look at the route’s behaviour possible.
The data used in this analysis come from the TTC’s vehicle monitoring system which collects GPS information on the fleet every 20 seconds. The raw data are transformed by several programs I have developed over the years so that they can be presented in a consolidated format. Interested readers should see Methodology For Analysis of TTC’s Vehicle Tracking Data for details of this process. The data were provided by the TTC, but the analysis and interpretation are entirely my own.
The last significant change to 501 Queen schedules occurred in Spring 2013 when weekday services were adjusted to address overcrowding. Since then, there have been only two basic schedules used on the route:
- The standard schedule provides service between Neville Loop in the east and Long Branch Loop at the city’s western edge. Half of the cars are scheduled to short-turn at Humber Loop, and except for overnight service, from that point westward the scheduled service is half the level of that east of Humber.
- On some occasions, construction has required that the line operate in two segments. One is from Neville to Humber, and the other (using buses) is from Humber to Long Branch. Service east of Humber is similar to that on the standard schedule. To the west, scheduled bus service is more frequent to allow for the capacity of low floor buses versus the two-section streetcars (ALRVs) used on 501 Queen. (A variation on this includes a shuttle service from Humber Loop to the condos east of Park Lawn, but it does not alter the service provided on the main part of the route east or west of Humber. This shuttle is not part of the service analysis.)
This table shows the two service designs for 501 Queen including the headways (scheduled time between vehicles), numbers of vehicles, end-to-end trip times and “recovery” time.
This last item deserves comment because it is not, as the name implies, intended to give operators on this extremely long route a break after their journeys. Instead, its primary function is to make the schedule work out so that the round trip time is a multiple of the headway. Because of the difference in trip times on the two branches of the route, this can produce long recovery times at periods in the day when they are not badly needed simply to make the schedule work out properly. I will turn to trip times and reliability in the second part of this series.
This article covers the following periods of operation on 501 Queen:
- August-October 2013 (service split at Humber Loop after Thanksgiving weekend)
- January-April 2014 (service split at Humber Loop for April)
- September-October 2014 (service split at Humber Loop)
- March-May 2015
As a general observation, service on much of the Queen route is very unreliable and, in some cases, to the point where it exists more in theory than in practice. Bunching is commonplace, and there is no evidence of any attempt to keep cars spaced apart from each other even long before they enter the most congested section of the route. If there is an operating discipline, its aim is to keep operators on time, with service to riders coming as an afterthought. In principle, if all of the service is on time, then reliability will take care of itself. However, in practice, the service routinely operates well off of its scheduled headways. This cannot be put down entirely to “traffic congestion” given how pervasive a problem this is and has been on 501 Queen for years.
Service on this route, particularly on its outer portions, has been an issue for as long as I can remember, and the TTC always has an excuse. If only they would expend one tenth of the effort to manage headways on this major route as they do to tell us about their latest of clean subway stations and other “customer service initiatives”, there would be many happy riders, and an incentive to bring even more. The route is developing medium and high intensity buildings along its length, but the service levels are unchanged since 2013 (and with only minor changes before that).
The TTC plans to introduce new schedules on Queen later in 2015 (or possibly early 2016) to address some of the reliability problems. However, without the will to ensure that vehicles on this very long route maintain proper spacing, the concept of reliability, let alone the “ten minute service” network of which Queen will be a part, will be meaningless.
Reading the Charts
The charts linked below show the average headways in hourly increments throughout weekdays with each week having its own data point (column) and each hour having its own plot (coloured horizontal lines). The solid lines show the averages, and the dotted lines show the standard deviations (SD) in the average values. Typically, about 2/3 of the actual values will lie within one SD of the average.
Where the time is given as “6:00 am”, this means that the values are for cars that crossed the measurement point within the hour starting at that time.
The TTC’s own performance target is to be ±3 minutes of the scheduled headway 70% of the time on streetcar routes. This equates roughly to an SD value of 3 minutes provided that the average value actually is the scheduled headway. However, when the average rises because of short turns, even if the SD value stays small, many individual data points will lie outside of the target band. The further the gap between scheduled and average headway, and the wider the SD value, the less the service is operating as advertised.
When the SD value and the headway are close to each other, a large proportion of the service is actually operating as pairs of cars. This is an important issue at the terminus where service originates and should be expected to be well-spaced, at intermediate points where one might expect supervision to intervene and hold cars as needed for spading, and at points where branches or common short-turns merge and, in theory, should produce a properly blended service.
Service at Neville Loop
For the purpose of this analysis, headways are measured a few blocks west of Neville Loop at Silver Birch Avenue. This corrects for a problem where multiple Queen cars arrive at Neville, but won’t all fit in the loop and to take layovers on street. The “arrival” time at Neville eastbound is therefore not indicative of the service experienced further west.
[A note about file naming: The prefixes “1xx” and “2xx” refer to westbound (100s) and eastbound (200s) respectively. This eliminates problems on routes which do not head in the same “direction” over their entire journey such as 504 King. The number “xx” signifies that position of each point on the line where I have measured headways. As we move across the route, this number rises. The purpose of this convention is to allow filenames to sort automatically by direction and location independently of the street names.]
Service on 501 Queen is supposed to operate roughly every 5 minutes during peak periods, and every 6 minutes at midday. However, the behaviour of service at the east end of the route is substantially different from the schedule.
The first page of charts shows the period from 6:00 to 9:59 am. What is particularly striking here is that the averages for 8:00 and especially for 9:00 are considerably higher than the scheduled value. Moreover, the SD values rarely dip as low as 3 minutes and are much higher than the scheduled headway at times. This shows an extremely irregular service from 8:00 to 10:00 am. There is little difference in the eastbound data (arrivals) and westbound (departures) indicating that little or no effort is made to even out service leaving from the terminal.
Through midday, the average values show less variation hour-by-hour, but they remain above the scheduled levels, and the SD values lie mainly in the 5-6 minute range well above the TTC’s target. Put another way, with the average running at 8-9 minutes, and the SD running at 5-6, much of the service is operating within a band of 3-15 minutes. Midday service was noticeably worse during fall 2014 when the streetcar service all terminated at Humber.
Through the PM peak from 15:00 to 18:59, the situation continues with average headways well above scheduled, and SD values generally above 4 minutes.
The evening period is only marginally better, and of particular note is the higher average headways late in the evening during the months when streetcars were all turning at Humber.
Throughout the day, the averages and SD values bounce around a lot from week to week showing that service to Neville is hard to predict. Note that if the data are examined at a more finely-grained level, this behaviour is even more marked (for example by half-hour subdivisions or by individual days), and the hourly/weekly averaging masks the worst of the situation.
The averages show a common problem for this and other routes. The AM peak starts off well enough, but in the later peak, and certainly in the transition to midday service, the headways at the outer end of the line are much worse than advertised. This occurs because of the amount of short-turning used to recover from peak period conditions. A similar situation occurs after the PM peak.
Service at Greenwood Avenue
By the time service reaches Greenwood westbound, additional cars have joined in from short turns at Woodbine Loop and from vehicles entering service at Russell Carhouse. If any attempt is made to regulate headways passing the carhouse with onstreet supervision, this should show up with good headway reliability a few blocks away at Greenwood.
Throughout the day, the average values at Greenwood are noticeably closer to scheduled headways, and the week-by-week numbers do not bounce around as much.
For the morning period, the SD values stay around 3 minutes except for the 9:00 data. Even though the average headway stays roughly where it should be (unlike the situation at Neville), the SD values are higher indicating a less regular service.
Moving on to midday, the averages stay roughly at scheduled levels, but the SD values do not settle down until after 13:00. During the PM peak, headways and SDs are better behaved. Later in the evening, the service is a bit less reliable and the SD values have risen again into the 4 minute range or beyond.
Eastbound service is quite another matter. SD values throughout the day remain well above 3 minutes showing the cumulative effect of the route’s operation across the city. This contrasts with the westbound service, at least for part of the day, showing that by the time cars reach Greenwood westbound, the service is in better shape than it was eastbound. [Note that eastbound averages in the 6:00 period are higher than westbound because the service builds up eastbound from Russell carhouse to Neville and the cars crossing Greenwood left the west end of the line almost an hour earlier on a wider early morning headway.]
Service at Yonge Street
Average headways both ways at Yonge stay mainly at the scheduled level, but the SD values are generally above 3 minutes and often overlap with the averages showing a good deal of bunching. With scheduled service running at 5 minutes during the peak period, bunching of cars can lead to gaps of 10 minutes or more rendering the “10 minute service” level that will be advertised on this route later in 2015 an empty promise.
Service at Lansdowne Avenue
Service eastbound at Lansdowne lies roughly at the level of scheduled headways but with higher SD values than should be seen particularly fairly close to a point, Roncesvalles, where headway management would be easy to implement. Westbound service is also more or less at scheduled levels, but with a higher SD value showing the effect of the trip from the east end through the city (compare to the chart eastbound at Greenwood) and the resulting bunching.
Service East of Humber Loop
Inbound service from Humber Loop shows a similar pattern at the end of the AM peak to the one we saw in the east end at Neville with headways after 9:00 being well above scheduled values thanks to westbound short turns. This is particularly bad during the Fall of 2014 when there is no Long Branch service to fill in gaps at Humber, and the short turns affect a higher proportion of the service.
Unlike the east end, the service really does come unglued by midday with average headways well above the scheduled values, and SD values at 5 minutes or higher. This shows that even inbound from a point where the through Long Branch service should mesh with the Humber trips, headways are not managed to produce a smooth interval between eastbound cars. The situation is somewhat better in the PM peak, with average headways closer to schedule, but still with a high SD value. This is particularly troublesome for periods when all streetcars turned back from Humber because it implies that they were tending to leave, if not in pairs, then with a long gap followed by two closely-spaced vehicles.
Evening service did not improve until 2015, and even then, the SD values are well above the level one would expect from a location where properly-spaced service could easily be dispatched.
Westbound service data generally mirror the eastbound values although the SD values are higher reflecting increased bunching outbound. For much of the evening, cars are routinely arriving in pairs as seen by the overlapping average and SD values. This is a condition that existed from the downtown part of the route, not one that developed as cars fought their way from Yonge to Roncesvalles (although that could add to the problem on busy nights). The fundamental issue here is that service is irregular coming into downtown from the comparatively uncongested east end, and there appears to be no attempt to smooth things out. Headways are much wider than advertised, and the Long Branch and Humber services tend to operate close to each other.
Service at Royal York and Lake Shore
Service eastbound at Royal York should be roughly every 10 minutes in the AM peak dropping to 12 for the midday. The averages lie mainly in this range, but the SD values are also high, especially after 9:00 when the combined effect of daytime headways and variability could routinely produce gaps close to 20 minutes.
The periods of bus operation are quite clear from the lower, and better behaved headways when the line operated as a Long Branch to Humber shuttle. However, even with this simplified arrangement, the SD values lie above 4 minutes almost all day suggesting that close attention to headways was not much of a priority.
The westbound data are noteworthy because the SD values are higher than their eastbound counterparts for period of streetcar operation showing the cumulative effect of the long trip from Neville. However, the eastbound values are good only by comparison, and SD values above 6 minutes are quite common.
Service at Long Branch Loop
Service at Long Branch runs less frequently than at Royal York thanks to short turns at Kipling. This is easily seen by comparing eastbound headways from the terminus to those at Royal York (east of the short turn point). The late AM peak behaviour of wider headways due to short turns that are “fixing” the service further east also shows up as it did for westbound service at Neville. A comparable PM peak behaviour shows up at Long Branch as well with wider than scheduled average headways inbound.
As at Royal York, the SD values inbound are lower than they were for outbound service arriving at the loop showing some leavening effect of the recovery time, but the values are still well above the target range of 3 leading to wide gaps in service.
For weekend service, the values are consolidated by month rather than by week because there are many fewer data points, but the chart layouts are otherwise the same. Saturday and Sunday data are shown on separate charts broken into the same four hourly segments as on the weekday charts.
Service at Neville Loop
At Neville Loop, the average headways are well behaved early in the day for both Saturdays and Sundays, although even then, some SD values do not stay below the 3 minute line. As the days progress, the averages and the SD values both rise showing the degree to which short turns affect actual headways operated to Neville, and the unreliability of the service. On Sunday evenings, SD values of 6-8 minutes are common on a 9 minute scheduled headway with averages routinely above this value.
Service at Greenwood
The situation at Greenwood westbound is somewhat better with average headways similar to the scheduled service. However, the SD values are routinely above 4 minutes except for early morning service showing that headway reliability does not meet the TTC’s target.
Eastbound at Greenwood, the averages and SD values are indistinguishable except for early morning service. This shows that much of the service is running in pairs eastbound, and the effective headway experienced by riders is double the advertised value.
Service at Yonge
The situation at Yonge Street shows bunched service with SD values close to or equal to averages during most periods all weekend. In other words, a common situation will be for a Humber and Long Branch car to cross the city close together giving much worse actual service (most riders will see the gap, not the considerably better “average” headway) than is advertised on the schedule.
Service at Lansdowne Westbound
With the service already bunched at Yonge, it is no surprise that the same situation exists westbound at Lansdowne. This situation continues further west.
Service East of Humber Loop Eastbound
In theory, the service from Humber Loop should consist of regularly spaced cars from Humber and from Long Branch (during periods when the entire line operates with streetcars). In practice, except for morning periods, SD values creep up into the range of headways, and service east from Humber is not well spaced despite this being an ideal location to regulate headways.
During Fall 2014, when all streetcar trips terminated at Humber, there is little difference in headway reliability compared with other periods. Erratic headways came east from Humber even without the need to mesh two services.
Service at Royal York
Weekend service outbound at Royal York is very unreliable. Not only are average headways at times above the scheduled values, the SD values can be 10 minutes or worse. Eastbound service is somewhat better with SD values that stay below the average values, but are still well above 3 minutes.
In this context, a “target” of ±3 minutes is laughable and from a rider’s point of view, there is no sense of when a car might show up. This is particularly bad when headways are already wide on the schedule, let alone the service as experienced on the street.
Service Arriving at Long Branch Loop
Service arriving at Long Branch is even worse than service westbound at Royal York, notably in early 2014 when Toronto was in the grip of a bitter winter (for example, compare Saturday afternoons). This shows the effect of short turns at Kipling.
Detailed Data for May 2015
Of the data for the 2013-15 period, the “best behaved” is from Spring 2015. However, “best” is a relative term, and severe problems with wide variations in headways remain. The charts below show the scatter of headway values for the month of May 2015 to give a visual sense of what is happening at various points on the route.
Each set of charts has 10 pages with the data subdivided into four sections:
- Four pages show the headway data for weekdays, one page per week. Each day has its own colour, and a trend line is interpolated between each day’s data points.
- One page shows all of the headway data for weekdays on a single page to give a sense of the dispersal of values overall. To meet the TTC’s ±3 minute target, the values should lie mainly within a band 6 minutes wide.
- Two pages show the Saturday and Sunday data in the same format as the weekdays. Victoria Day is included with the Sunday data.
- Three pages show the averages and standard deviations by hour for the data. Note that the weekday values are for the entire month, while on the Headway Histories earlier in this article, they are calculated for each week separately.
Service at Neville Loop
For westbound service at Silver Birch, just west of Neville, note how consistently the weekday trend lines lie at values of 9-10 minutes even though the scheduled headway is 5-6 for much of the day. This is an indication of how pervasive short turning at Woodbine Loop (and further west) is on this route. One might argue that all of the service is not required so far east, but this is a chicken and egg situation. Service is so unreliable at times that only die-hards (or those who luck onto a passing streetcar) will use it. Poor reliability drives away demand and, perversely, creates political support for parallel, if infrequent, premium express bus service.
The “cloud” of weekday headways (page 5) shows many values well over 10 minutes and several over 20, not to mention many that are well below 5. This is service just as it has left the terminal, and it is irregular from the outset.
Weekend service is even worse with headways in the 20-40 minute range not uncommon.
Eastbound values are more spread out with many, many cars arriving on headways of under 5 minutes (effectively right behind the car ahead). Weekend headways are also spread over a wide range.
Quite clearly, there is no attempt to provide reliable service on the east end of 501 Queen during the weekends.
Service at Yonge Street
At Yonge street in both directions on weekdays, headway values are smeared out over a range from 0-10 minutes, and with several above that range. This is reflected by an “average” that matches the scheduled service, but an SD value almost as high because of the range of values. On weekends, many cars arrive on very short headways, and gaps of 20-30 minutes or more are common. Even here, in the middle of the route, the TTC provides very unreliable service.
Service East of Humber Loop
At Humber Loop, the effects of short turns further east start to appear because the trend line values are above the scheduled headways. Individual values continue to lie over a considerable range, and very short headways inbound again show the complete lack of management of the through Long Branch and turning Humber cars.
Service at Long Branch Loop
At Long Branch, the service is truly appalling. The weekday headways are spread over a range of 0-30 minutes with some values beyond even the 40 minute line. By the weekends, headways are even more dispersed. Inbound values are somewhat more compact than outbound, but the SD values are still very high with Sunday values of 10 minutes routine through the day. As with the service at Neville, but on an even wider scheduled base, it is quite clear that no attempt is made to provide a reliable service on this part of 501 Queen.