From time to time, someone will tweet a complaint to @TTCHelps about a very long wait for a bus and copy me into the thread. This can set off an exchange which, to be diplomatic, can involve varying claims about what is actually happening.
For as long as anyone can remember, the TTC has a standard response to such complaints: that traffic congestion or some other transient event beyond their control is responsible. More recently a few new lines have been added to their repertoire including:
- Due to inadequacies in the schedule, buses cannot stay on time, but this will all be fixed in a coming revision.
- There are “run as directed” buses which are used to fill gaps in service and respond to problems of overcrowding. These buses are far less numerous than some at the TTC have claimed, and they are completely invisible to service tracking apps.
- Riders concerned about crowding can refer to transit monitoring apps to see if an uncrowded bus is coming down the route. Of course if you’re on a streetcar, they don’t have passenger counters and there is no online crowding info for them, in spite of ads for this service up and down Spadina Avenue.
On top of this, the TTC produces monthly on time performance stats that purport to show that, overall, things are not too bad. They have “service standards” about what constitutes an appropriate quality of service, and they hit them to some degree some of the time, on average.
This is a long-standing response of “not our problem”, backed up by “we will fix the schedule eventually”, “we are meeting our standards most of the time”, and “riders can find uncrowded buses, so what’s the problem anyhow”.
This is cold comfort to riders waiting for service.
Problems of irregular service and crowding on the TTC predate the pandemic, and were starting to attract attention by the politicians who claim to set policy and could not square complaints from riders and constituents with management reports. Then the world changed.
But the world is trying to change back, and with it the desire for transit service to actually attract riders. The time is overdue for attention to quality of service as a basic marketing tool. A shop window does not attract customers with a photos of products that might arrive soon, maybe.
Bathurst Bus Scheduled Service
In January 2021, weekday service on 7 Bathurst changed from regular-sized to articulated buses (12m to 18m), and the January 2019 schedule was restored. As we will see later, there are still several 12m buses running on Bathurst, but on schedules that assume 18m capacity.
In May 2021, peak period service was trimmed in response to actual demand, and the service in effect until Friday, September 3, was to operate every 10 minutes throughout the day (see table below). Note that the schedule includes an allowance for construction of Forest Hill Station on Line 5, but actual operating data charted later in this article shows that this is no longer a source of delay.
The January schedule with slightly more frequent service will return on Tuesday, September 7 as part of the TTC’s overall restoration of service.
On Friday afternoon, September 3, 2021, a tweet popped up asking the perennial question “where’s my bus” from a rider waiting at Glencairn and Bathurst. The 7 Bathurst is a notoriously unreliable service even though, irony of ironies, it serves the TTC’s Hillcrest complex.
This piqued my curiosity, and I turned to Transsee, a site run by Darwin O’Connor, where one can view route tracking data in real time. For the TTC, one can access detailed reports for streetcar routes free of charge, but for buses one must be a “premium” user, something that can be had very inexpensively. [Full disclosure: O’Connor gives me premium access free as a courtesy because of the work I have done on route analysis that inspired several of his site’s features.]
Among the charts one can generate is a real time service graph showing movement of vehicles. This allows one to immediately see what is happening on a route including the recent history leading up to whatever the problem of the moment might be. I started watching 7 Bathurst and was appalled by what I saw.
Here is a chart of the service as it appeared mid-afternoon on September 3. In this chart, “north” is at the top (Steeles) and “south” is at the bottom. Note that just after 1 pm there is a gap of almost one hour in service southbound from Steeles. Except for one bus that comes into the early part of the gap north of Lawrence just before 2 pm, the gap persisted over the length of the route.
(For readers unfamiliar with these charts, time runs horizontally from left to right, and distance along the vertical scale. The space between lines is proportional to the length of the headway, i.e. the wait for a bus. The slope of the line is proportional to speed with slowdowns indicated where lines run more horizontally than vertically.)
(Updated September 6 at 10:40 am: Darwin O’Connor has tweaked his software to keep the orientation of these charts more consistent, and so I have replaced the original chart here which had south and north reversed from their conventional map alignment.)
How, exactly, did this happen, and how did service evolve for the rest of the day? In retrospect we can see the whole day, although I have broken it into two overlapping charts for clarity.
8:00 am to 7:00 pm
- Bus 8392 (blue) disappears from service at Steeles at about 8:30 am.
- This creates a small gap ahead of bus 9002 (lime green).
- Bus 9012 (dark blue) disappears from service near Wilson at about 8:40 am northbound.
This is likely the end-of-peak period transition from a 14-bus to a midday 13-bus service. (7 Bathurst operates from Wilson Division.)- This run is supposed to stay out all day, and reappears briefly as 8141 at about 10:00 am, later as 9002 and finally 8130. Thanks to Darwin O’Connor for sorting this out.
- The obvious question is why did it take four separate buses to operate what is supposed to be one continuous run on the schedule.
- Bus 9002 appears to go out of service southbound at Sheppard at about 9:12. It reappears just before 2 pm just north of Lawrence. This leaves a gap which travels south to Bathurst Station.
- Bus 8141 (red) appears southbound at Davenport partly filling the gap left by 9002. It operates not quite a full round trip and disappears southbound at Glencairn just after noon leaving a gap.
- This gap is further widened by the disappearance of bus 8396 (dark green) northbound at Lawrence at about 12:45.
- At this point, the gap is about 45 minutes wide, and by the time it comes back south from Steeles, it is almost one hour wide.
- Bus 9002 rejoins the service southbound just before 2 pm, but even with this, the gap northbound from Bathurst Station behind it is 40 minutes wide.
- At roughly 2 pm, bus 9002 leaves service northbound at Lawrence and is replaced by 9019 northbound at Wilson.
- Bus 8130 (orange) enters service northbound at St. Clair just before 3 pm reducing the gap slightly, but by now there is a pack of four buses travelling together northbound. One of these will short turn at Finch, and the rest will return southbound in a pack.
- A 40 minute gap travels south from Steeles just after 3:30 pm.
- Bus 8160 (dark orange) enters service northbound at St. Clair just after 4 pm.
- There is a 30 minute gap northbound from Bathurst Station starting at about 4:45.
This is an example of how service gaps can ricochet back and forth on a route. When the scheduled headway is every 10 minutes, one missing bus, or a pair running together can make for a long wait for riders and corresponding onboard crowding.
Note also that several of the buses used on the route are not articulated vehicles (9000 series) and they will be more crowded simply by virtue of their size. The service is scheduled for 18m vehicles, but several vehicles are only 12m. Assignment of the wrong vehicle type is a chronic problem on the TTC, and it bedeviled service on 501 Queen when the ALRV streetcars were introduced.
3:00 pm to Midnight
The chart below overlaps the first four hours of the one above for continuity. From 3 pm onward, buses are travelling in packs of 2 or 3. Even when they are spaced out leaving Steeles southbound, the groups do not take long to re-form and this continues until about 9 pm. At that point a cluster of buses that has travelled together for 6 hours finally breaks apart. Service thereafter is fairly regular.
Looking at Other Days
I began by looking at two previous Fridays to see how common this problem might be.
Friday, August 20, 2021
In August 20, although there were no immense gaps, buses do commonly run in groups of two or more for an extended period. The largest gap northbound from Bathurst Station is just under half an hour long between 2 and 3 pm.
Friday, August 27, 2021
On August 27, there is a nearly half-hour gap northbound at about 2 pm.
Wednesday, September 1, 2021
Only two days before the tweet that started this analysis, the TTC achieved a huge gap northbound from Bathurst Station at about 5:30 pm. There was no bus for over an hour thanks to a gap that travelled the length of the route from Steeles to Bloor.
Thursday, September 2, 2021 is only slightly better with a 45 minute gap northbound from Bathurst Station starting at 6:24 pm.
The No-Short-Turn Rule, Gaps and Crowding
A keen observer of these charts will note that it is rare that a bus short turns before reaching the terminus. This is due to an operating practice introduced in recent years that there shall be no short turns. In theory, schedules are adequate for almost all trips to run end-to-end in the allotted time including a break for the driver. In the real world some schedules are too tight and others are very lax to the point that the TTC has started to trim some of the fat.
When a gap forms, for whatever reason, short turns are (or should be) used to fill the hole. However, taken to excess short turns can create service reliability problems on the outer ends of lines while fixing service on the inner parts. They can also be an incentive for drivers to deliberately run late in hope of a short turn and a nice siesta.
What we see on Bathurst is an extreme case where simply leaving buses to make full trips allows gaps to widen to the detriment of service. There has to be a trade-off somewhere including managing service and headways. This is not the same as simply saying “run on time”, or worse as the TTC does, “run more or less on time”. Much of the service shown in charts here meets TTC standards because it is very, very bad only occasionally. But for those with the bad luck to land in one of the exceptions, things can be very bad indeed.
Gaps are a particular problem because more people are affected by them than “the average rider”. Many people will stand at a stop for a very long time, or give up and leave, in a one hour gap. But after it passes, service is back to normal. Only one bus out of many is in a big gap, and for the stats, this is perfectly ok. Riders might not agree.
Unevenly spaced buses have uneven crowding. We don’t actually know how bad this is because the TTC still does not make historical crowding data available for analysis (come back in 2022, maybe). However, a “gap” bus will be more heavily loaded, and so more passengers will experience crowding than vehicle-based stats would indicate. Only one bus in three might be in a gap, but more riders are on that bus than on the other two. The average experience is crowded even though statistically the overall service is not.
This is a fundamental flaw in the TTC’s analysis of crowding – it is based on vehicles, not on riders.
The Problem of Missing Runs
Another problem besetting TTC service is that buses and streetcars will simply disappear from time to time because there is no relief operator to take over at a shift change. This happens due to a shortage of spare staff and an attempt to minimize overtime.
When a vehicle is missing on a route with only modest service (every 10 minutes or worse), the effect for riders can be catastrophic with very wide gaps and no option to “wait for the next bus” because the next one might be over 20 minutes away.
The TTC claims that it uses some of the “run as directed” buses to fill these gaps. Complaints from riders (more tweets!) suggests that the RAD buses are as invisible to riders as they are to the vehicle tracking apps. In any event, as I wrote recently, there is only a limited number of these buses, and on weekends they are likely to be occupied with subway shuttle duty, not with filling service gaps.
I will turn to the problem of missing runs in a future article.
A Note For Twitterers
If you are going to post a comment/complaint about service, be sure to include the route (name or number), location, direction and time. The vehicle number is a nice-to-have, but not vital for tracking down gaps that should be obvious in the tracking data. However, for “I was on bus xxxx and here’s what happened” reports, the vehicle number is very important.
How Long Has This Been Going On?
Some decades ago, before the Sheppard Subway existed, I was at a TTC Board meeting and the then-Chair, Gordon Hurlburt, was late arriving. When he did come into the room, he complained that he had been waiting for half an hour on Sheppard at Leslie for a bus that he could actually board.
In due course, management reported that they had reviewed the route and found no reports of overcrowding, in effect telling the Chair that what he had experienced did not occur.
Hurlburt took up smoking again not long after that, and did not seek reappointment for another term.
Steve Munro 
There’s still a “7” Bathurst bus? Which rapid transit is next to be given the “7” to its line number? Guess I’m getting ahead of myself.
Seriously, the 7 Bathurst is a l-o-n-g route. Any incident along the route would cause bunching and/or short turns. Maybe they should consider breaking the bus routes into several shorter routes, using the Line 5 Forest Hill Station as one “terminal”, to start. Unfortunately, there is no off-street loop at Eglinton, unless they can figure out how configure an on-street loop in the neighbourhood (and upset them at the same time).
Steve: Many routes operate from the BD subway to Steeles, and there are east-west routes of comparable length also. We can’t chop them all in half or thirds to make them easier to manage. Moreover, the TTC has proved that they can make a mess of short routes too (my next article will look at that).
LikeLike
If you click on a dot on one of these lines in the charts (which can be difficult even with a mouse) you can see the details of the trip path. From there you can click the link to see the schedule for this trip. By looking at the Trips in this block you can see the bus is scheduled to run all the way to 4:40AM the next morning, a total of 23 and a half hours.
Clicking on the next scheduled trip and then the Trip path link shows the run reentered service on bus 8141 (red) that was only in service for a couple hours.
Later on the run was served by 8130 and then 9002.
Steve: Thanks for the tip. I noticed after I posted the article that there was no corresponding vanishing run in August data, and therefore this is an unfilled crew, not a peak period only bus. The fact that the same run is served by different buses later in the day shows how it was partly filled as the day went on, but not with normal crew changes. I will change the text now.
For readers: Note that this description only applies to actions directly on the transsee site. The charts in the article are screen grabs and they do not contain the hotlinks. Also, because this is a bus route, you will not be able to display the linked charts Darwin describes unless you are a premium user on transsee.
LikeLike
Wilson division doesn’t have the right amount of hoists to repair enough of the articulated buses. If they end up losing the route to Mt Dennis (Artic division), service might be worse with the distance buses would have to travel to route.
LikeLike
The irony. Crossing Eglinton is still a huge source of disruption for the 7, and no amount of data will change the daily experiences of every rider who’s been dealing with these problems for years now.
Steve: I was surprised to see so little disruption at Eglinton in the data knowing the situation there. There was actually more of a problem further north where there is road construction. In both cases, the delays that did occur are relatively short and do not occur in both directions all of the time. Even so, buses become bunched and remain so over the course of round trips. This cannot be explained by whatever construction-related congestion does exist.
LikeLike
In theory, shorter routes are easier to manage, but that would only work if the TTC actually manages their service and does it well. Experience shows that service on some short TTC routes can be just as erratic and terrible – Coxwell, Woodbine South, Ossington, Wellesley (east of Yonge), the Broadview shuttle, Cosburn, just to name a few. Even the Sherborune bus, with only two buses running on Sundays, could come in a bunch. Poor service is the number one reason why I don’t use transit as often as I could right now.
LikeLike
Theoretically the Bathurst North route would cover parts of the service gaps north of Wilson. Though that’s cold comfort for passengers looking for through service south of Wilson. Could this rationalize the behaviour of line management for leaving such wide gaps in service?
I’ve noticed that some RAD crews will inherit the scheduled run so as to remain visible. It does beg the question as to why it took four buses (9012, 8141, 9002, 8130) to make service on that run (mechanical issue? RAD crew? change-off? operational/staffing issue?)
Steve: On September 3, 160 Bathurst North had its own problems with a one hour gap southbound at Steeles between and 2 pm thanks to a missing bus, and other general irregularity in headways. With a scheduled afternoon headway of 24 minutes, it is hardly in a position to provide relief for gaps on 7 Bathurst.
As for four buses on one run, yes something strange was going on there.
LikeLike
I was thinking about the recent events of the near-miss subway incident and how the TTC Board noted there was no formal reporting that it happened. The TTC agreed there would be in the future, with the most serious incidents requiring immediate notice to the board.
I think it would light a fire under the TTC route management’s comfortable chairs if there was a required metric in the CEO report that reported any extraordinary delays on surface routes and an explanation of what caused them too. Immediate notice is not required, but this is information would be useful to the board, and further to all the councillors that are impacted. I don’t know an exact metric to use, but as an idea, if any surface route has a gap in service that is more than 30 minutes beyond the schedule, there should be a detailed explanation available and included in the reports given to the board. They don’t have to present it, but then at least the councillors can choose to ask about it if they want, and even challenge the explanation. I’m thinking these would be the worst of the worst level, so the “30 minutes” mark can be adjusted up or down to capture those.
Do you think that would be of value?
Steve: What must happen first is that the TTC must get beyond the covid-era “we’re just keeping the lights on” mentality to one of operating, and provably operating, the best service possible.
Service reliability has been a problem for as long as I can remember, but it has become worse in recent years. Management claims that it’s not as bad as riders often report are an insensitive way to address the issue, and put the TTC in a position at odds with riders rather than supporting them. This is a problem of corporate culture.
The next step is to decide just what constitutes “good” service with metrics that reflect what riders experience and at a granularity riders can relate to. Saying that “the Dufferin bus is really crowded in these hours, and at these locations” means a lot more than “on average, over 24 hours of operation, on all routes, buses are crowded less than 10 per cent of the time”. The same could be said of the 401, and we all know just how credible such a claim would be.
The current service standard for routes with headways between 5 and 10 minutes (6 to 12 buses/hour) is that:
For services less frequent than 10 minutes (fewer than 6 buses/hour):
These are demonstrably ludicrous standards first because only 60% of the service has to meet them overall, and because wide gaps are still possible within the standard as I have discussed in previous articles. Briefly, for a 12 minute headway, one bus could be one minute early and the next five minutes late producing an 18 minute gap. There is no standard regarding maximum allowable gaps or bunching, nor is there any reporting or standard regarding lost trips due to missing vehicles. Bunching is almost the same as a lost trip because second and third buses in a parade may provide little service to riders even though they would count against a vehicle/hour metric.
The TTC does not even achieve these rather generous standards.
There is no standard for vehicle availability beyond meeting the day-to-day requirements of scheduled service. However that service, especially for streetcars, is routinely constrained by the number of vehicles available. Moreover, a shortage of operators causes cancellation of entire crews causing service gaps that are particularly bad on routes with wider scheduled headways.
A big impediment to better standards is that by just about any metric I can think of, the first report in the new format would look really, really bad and management might stop getting their gold stars. Tough. We are hoping to see the city rebound, and it will need good transit service to do this. The problems of erratic service in areas dominated by bus services will accentuate the two-tier service quality so often complained of by riders outside of the core. Even “downtown” there are serious problems.
Management gaslights complaining riders with stats that prove nothing, and with claims of run-as-directed buses that nobody can see or prove how they actually improved service. Crowding stats are available through a real time feed to some providers, but not for the general public on an archival basis to correlate operations with onboard conditions. Maybe next year. And, of course, it is now a rider’s job to possess a smart phone and use it to see when an uncrowded vehicle might show up. This is an abdication of responsibility to run reliable service adequate to demand.
The TTC is good on PR, but not so good on actual delivery, and few of the TTC board members look beyond the rosy reports.
I will be the first to agree that presenting operating stats on a meaningful granular basis is difficult because there are multiple dimensions: location, time of day, type of vehicle, external conditions. Overall stats show the general state of affairs, but riders experience service one bus at a time, and predictability is key. Moreover, we are long past the era when detailed analysis required an army of clerks poring over hand-written reports that took days to arrive from the field. We have computers now, and this sort of thing can be automated, if only there is the will to understand what is really happening on the street.
LikeLike
A simple way to see if management is doing their jobs running the system is to measure median distance between buses and to count the number of times that the distance between 2 buses is 1 standard deviation above the median. Every modern transit system has GPS in their buses and it would be easy to measure how far away the buses are from each other. This is not impossible to measure with modern GPS and GIS systems and would be a start to see who is playing games and who is asleep at the wheel at TTC control.
Steve: There is a flaw in this plan in that the standard deviation of disorganized service is larger than for service that is spot on time where the SD would be zero. Also, from a political/publicity point of view, SDs are not clearly understood by politicians or most riders.
LikeLike
Much of the problem with the Bathurst Bus can be attributed to the fact that there is now major construction on Bathurst between Lawrence and Wilson. Limited short-turning of the odd northbound Bathurst Bus at Lawrence could help to regularize the service south of Lawrence. As you said, Eglinton does not seem to be as much of a problem as I thought it would be because of the fairly generous green light timing given to north-south Bathurst traffic through the intersection.
Steve: What is troubling, however, is that problems with the reliability of the service start to develop before the major effect of slow operation north of Lawrence kicks in, compounded by missing buses and the wide (10′) scheduled headways with artics. There appears to be very little effort to deal with these issues by regulating service.
LikeLike
Another report available in TransSee is the Route Performance report. Using data for route 7 Bathurst from last week you can see no point where schedule adherence becomes suddenly worse. The route, like many in Toronto and across North America, is general runs ahead of schedule.
In the north bound chart the gap data shows a gradual increase in bunching along the route.
In then south bound chart the gap data shows a gradual increase in bunching in the south of Lawrence.
Both directions show a sizable amount of missing service and bunching approaching Wilson, but looking at the operation chart it appears to be due to radio problems in the area.
Steve:
It’s not radio problems. 7 Bathurst runs out of Wilson, and when a bus goes out of service, it will often do so there.Sorry, I misunderstood your point as demonstrated in the next comment.LikeLike
If you look at the details of the operation chart you can see buses that disappear and reappear twenty minutes later exactly where you would expect it if it was running normally.
Steve: Intriguing that this only affects northbound buses. In any case, your plot does not show them disappearing from service, only not appearing at stops. This would affect a stop-based headway report, but not an overall view of service.
LikeLike
Many cuts on routes where more bunching
What’s wrong with making sure drivers have proper washroom time
And maybe two minutes to grab a quick bite and hydration
Treat drivers like human beings
Many delays to service are traffic poor city planning make city more transit friendly
Steve: The biggest problem with TTC schedules on that account is that they have never wanted to schedule a formal break at the end of a trip that operators would be entitled to take even if they were running late. Where there is “recovery” time, it’s all to make the headways come out right especially on branching routes. Whether ATU has ever fought for this, I don’t know, but it has been an issue for as long as I can remember.
LikeLike
Should Bathurst be part of the 900 Express network and will it help the service levels during peak periods? Thanks
Steve: The problem is that Bathurst is already comparatively infrequent, and if it were split into local and express neither component would be very attractive. It’s a bad choice for artics because of the wide headways.
LikeLike
I’m on the 501 eastbound, right now. Car 4539, can’t see the run number from inside. I know it’s running 10 minutes late because a) Google told me b) the car behind us is effectively right behind us c) the car in front of me – which I just missed – was WAY ahead of us.
Despite that, with literally no cars in sight ahead, we are doing a GPS verified 23km/h through leslieville.
Uuuggghhhh. It’s this kind of service – waiting nearly 20 minutes for the streetcar to show up, only to have it plod along at a snails pace – that makes me wish I’d driven.
Steve: I just had a look at the operating chart on TransSee and can see the pair of cars you describe. What is so annoying is the every car gets a long layover at Neville, and there is no excuse for service running in bunches.
LikeLike
Edit: the car behind us, 4553 / run number 02, was in fact literally following us by the time I got off in the beaches.
We would slow to a crawl at every stop, even if there was no one there. I might just be paranoid, but I swear our driver was even trying to see how many yellow lights we could hit, to maximize our waiting time…
Steve: 501 Queen has a very padded schedule probably in anticipation of construction delays that are not in fact happening. The mid-evening schedule includes 90 minutes travel time for a Neville-Spadina round trip plus an astounding 18 minutes of recovery time. The result is cars that travel slowly and have very long layovers.
When the route is one very long trip from Long Branch to Neville, there is a good rationale for terminal layovers, but Spadina to Neville is a short route with relatively little congestion. The excessive running time just wastes resources and gives riders bad service.
LikeLike
Tonight just after 10pm 4405 (Run 08, did not appear on Nextbus) and 4538 (Run 01) both left Charlotte loop to head east as a pair. For some reason the trailing vehicle was the one picked to turn back before Neville. It seems like a nightly occurrence.
Steve: Looking at the operating chart for the period in question, this looks like a change-off. 4538 disappears from tracking at around Sherbourne and is replaced by 4405 shortly later.
Also visible on the chart is a gap of over half an hour westbound from Neville caused by car 4596 taking an extended siesta. There was no eAlert about any problem on the 501.
LikeLike
I don’t know about this change-off business but 4538 was definitely still in service and following 4405 at Broadview. Signage was programmed for Woodbine loop short turn. Is it a tracking anomaly?
Steve: 4538 does not show up later in the tracking data and probably ran in to Russell. If it was a changeoff, or a crew handover, it would not be able to logon with the same run number as its replacement.
LikeLike