This is the final section of a three-part post about link times on the Queen Car. Previous sections dealt with Neville to Yonge, and Yonge to Humber. A detailed description of the concepts used in these analyses is in the first post.
Long Branch has been particularly hard hit by TTC service decisions over the years. Originally a separate route, the 507 Long Branch car ran between Humber and Long Branch providing a 10-minute off-peak service and a 7-minute peak service supplemented by a few trippers than ran downtown to Queen & Church in the AM peak with outbound trippers in the PM.
This changed when the TTC amalgamated the 501 Queen service with the 507, and further with the replacement of CLRVs (50-foot cars) by ALRVs (75-foot cars). The scheduled headway on Lake Shore is now:
- 9’45” AM Peak
- 11’00” Midday and PM Peak
- 14’45” Early evening M-F
- 20’00” Late evening M-F
- Saturday service ranges from 11’30” in the afternoon up to 18’00” in the late evening
- Sunday service ranges from 14’30’ in the afternoon up to 23’00” in the late evening
The posted schedules on the TTC’s website are a complete mess with irregular headways shown throughout the day. This is clearly a problem with their schedule-production software which has been known to produce other gaffes in the generated timetables. The fact that such timetables are created for public consumption with such glaring errors tells us a lot about quality control at the TTC.
It is bad enough that riders on Lake Shore must endure much wider scheduled headways than on many other parts of the system, but as we will see, the actual service provided is much, much worse. When scheduled service is infrequent, provision of on time service, and all of the service, is essential.
Westbound:
Queensway/Humber to Lake Shore/Humber
Lake Shore/Humber to Royal York
Royal York to Kipling
Kipling to Long Branch
Eastbound:
Long Branch to Kipling
Kipling to Royal York
Royal York to Lake Shore/Humber
Lakeshore/Humber to Queensway/Humber
Round Trips Kipling to Long Branch
The trips from Humber to Long Branch are divided into four links. The first of these consists mainly of Humber Loop measured from the Queensway to the Lake Shore side. This picks up any effects from layovers in the loop itself. Outbound trips tend to be quite short, although there are a few outliers on weekends. Sometimes a Humber car has to get out of the way of a through car to Long Branch that has caught up.
Inbound times at Humber are also fairly consistent, although something obviously caused a delay between 1700 and 1800 on Thursday, December 14. Reviewing the service chart for that date, I can see that there was some major foul-up in this part of town. Several cars short turned outbound at Roncesvalles and a parade went east from Humber.
Trips from the Lake Shore side of Humber loop outbound to Royal York have a fairly consistent running time, although there is a big jump on December 14 corresponding to the inbound delay indicating that wherever event occurred, it blocked service both ways near the loop on Lake Shore. This event also shows up in the link times eastbound from Royal York.
From Royal York to Kipling, the running times are again fairly consistent. A large “wow” in the curve on Saturday, December 2 is caused by an extremely long gap in service to Long Branch and a correspondingly long running time over this segment for one car. Eastbound data over this link is also fairly consistent with only a handful of outliers, notably one early on New Year’s Eve.
Between Kipling and Long Branch, things get quite interesting. Outbound trips are clustered fairly tightly around 5 minutes, and inbound trips sit between 6 and 7. There are some problems with resolution of the departure from Long Branch in the CIS data, and some inbound trips include at least part of the layover time. This effect can be cancelled out by looking at the round trips from Kipling to Long Branch and back.
For weekdays, these are clustered in the 20-25 minutes range, although there is a very wide scatter. This gets even worse on Saturdays with values ranging from 15 minutes to above 45 minutes. We already know from the inbound and outbound link charts that the time needed to make the trip is, generously, about 12 minutes. This means that we are getting layovers from 3 minutes (almost too short for a line this long) up to half an hour! It is common to see more than one car at Long Branch Loop at the same time even when the scheduled headway is quite wide.
As we have worked our way across the city, one thing we have seen in several links is the scatter in values for some areas at some times of the day. Cumulatively over the length of the route, these can produce a wide variation in travel times, and correspondingly large swings in the layovers at terminals. However, the range of layovers we see at Long Branch far exceeds what is needed to keep service on time. Indeed, even with such generous schedules, the inbound service from Long Branch is still uneven as we have seen in the daily analyses.
For those of you have stayed with me to the bitter end, a treat, of sorts. Looking at the link data as well as the daily headway and service charts, I couldn’t help wondering what would happen if I plotted the headway information using the same template as the link times with data for every day on one set of charts. Here is the result:
Westbound at Royal York
Eastbound at Royal York
These charts should bring deep shame to anyone at the TTC who has ever claimed that service is well-managed or reliable. When I first looked at them, I couldn’t quite believe what I was seeing, and checked back in some of the source data to see that I hadn’t screwed something up.
For the four sets of weekday charts, you can see that the trend line through the headways sits roughly at 10 minutes, although it wanders a bit showing the effect of some short turning. However, the scatter of the data points is huge and is glaringly visible on page 5 where all weekdays are displayed together.
For a nominally 10 minute headway, the westbound service is a cloud of data points all over the map, gaps of over 20 minutes are common, and there are far too many above 30 minutes. By no stretch of the imagination is this the “frequent service” advertised in the timetables. Saturdays and Sundays are no better with the only difference being that the trend line sits at a higher level due to the wider scheduled headways.
Eastbound, things are a bit better, if we can call headways that generally stay within twice the scheduled values “better”. However, it is an indictment of the on-time performance eastbound from Long Branch where, despite generous layovers, drivers seem incapable of leaving on time and nobody seems to be doing anything about it.
When the 507 was a separate route, the service was reasonably reliable and riding was much better than it is today, despite the direct service to downtown. The route had a lot of local demand, but sane riders today would only take the car if it were there, and then would worry about getting back again.
There is no justification for such atrocious service.
Steve Munro 
Steve, thanks again for your in-depth review and analysis of the Queen streetcar, which I have been following with great interest.
From the charts in your previous posts, it was already apparent that the biggest problem was trips that started well off schedule, or that were not evenly spaced out, whether it was regular trips (e.g. leaving Neville) or short turns inserted into service (e.g. leaving Kingston Road). If you are offering irregular service right off the bat at the start of the route, how can you expect to run regular service anywhere? Early trips just keep getting earlier, and late trips just keep getting later. This pattern was already evident from the service charts, whether on busy days, average days, or light days, and even from the start of service. The scatter plots (an apt word) at Royal York appear to reinforce this.
At the upcoming Rocket Riders meeting on the Queen streetcar, I suspect we will hear a lot of suggestions ranging from, say, extending parking restrictions in certain locations, all the way up to building subways. What these posts have shown is that the first place the TTC should be looking is making sure that the streetcars start each trip with some semblance of regularity.
a) Figure out why cars are leaving the terminals at seemingly random times
b) Determine whose responsibility it is for ensuring regular or on-schedule departures
c) Implement some sort of procedure to measure it
d) Determine what standard should be considered acceptable (e.g. xx% of trips within xx minutes of the schedule, or within xx minutes of the average scheduled headway; standard deviation of headways etc.)
e) Ensure that someone is held responsible for maintaining that standard
If we make sure that the majority of trips at least start out on schedule (or evenly spaced), we stand a better chance of having more reliable service once we get to Yonge Street and hopefully fewer short turns. As a bonus, we gain increased passenger capacity without adding cars to the route (since a convoy of streetcars has unused, wasted capacity), and with any luck potentially slightly higher speed (if loading levels even out and sardine loads become less common).
I would focus on this aspect as a starting point not only because it should easily address a major observed problem that significantly impacts riders’ overall experience, but because there are other bus and streetcar routes in the system that experience similar problems (whether or not on a similar magnitude) and would definitely benefit from a similar procedure.
Steve: I would make one subtle change in your recommendations. Instead of focussing on the schedule, focus on the headway. Riders don’t care if service is five minutes late, only that it is there on a reliable headway. Better that all of the service operate slightly more widely spaced than usual, but regularly, than the chaos when runs are missing or bunched.
Such an approach is more difficult to manage than just watching a computer screen and calling up an operator when the system says they are 5 minutes late. It requires actively monitoring vehicle spacing and making adjustments on the fly to headways and departure times. Unfortunately, CIS was never designed with such tools (stock in trade for good, old-fashioned on-street route Inspectors) as a basic part of the system.
The TTC has a fetish for on-time performance, but clearly misses even that by a mile with the service it actually operates. The drive to be on time leads to needless short turns and time wasted when cars could be providing useful service for riders.
LikeLike
Your complaint about schedules reminds me of some things I noticed.
The 501 schedule (which is the one posted at stops) includes 301 runs, marked with #. It also claims in the key for # that all WB 301 runs go to Long Branch. This is not the case. So a whole range of 301 allegedly Long Branch cars seen on the westbound schedule at, say, Coe Hill (where there is no longer a stop, of course), and even at Humber Loop westbound, vanish by Park Lawn.
Someone waiting at any InfoPost-equipped stop east of Park Lawn will have the false expectation of frequent service to Long Branch in the early wee hours; this expectation is false.
Note that the 301 schedule does get things right. But it’s not the one posted at stops. Anyway, the 501 schedule covers 301 runs too, right? Too bad it doesn’t cover them correctly.
The other interesting bit in the schedule is the 23 schedule gap in westbound service from Humber, every weekday evening: 10:37, 11:00. The headways go 15, 16, 23, then every 20 minutes. I’ve fallen into that gap and spent quality time at Humber Loop as a result.
LikeLike
I don’t think I can imagine a less controversial proposal than having streetcars start their runs evenly spaced. It’s a change that could be made very quickly and at very low cost — surely some supervisors or CIS operators could be dedicated to this task, at least to see what impact they could have.
I know frustration with current service is going to lead some people to propose more radical changes. (It’s already happened over at Spacing Wire, with other people named Matt.) But there’s a big difference between improving service and building a whole new line. Any private ROW/subway/rail corridor service would be years if not decades away — and that’s without the debate about whether it’s a net positive for the neighbourhood or factoring in time for the funding planets to align. We have the streetcars and track today, and you’ve made a strong argument it’s not being operated anywhere near peak efficiency. It’d be foolish not to try to get the most out of past investments before considering new ones.
LikeLike