This is the first in a series of posts about service on the Queen car following on from my article about evaluating the quality of transit service. Queen is a major TTC route that includes many problems including its length, traffic congestion in certain parts of the route, and a general dissatisfaction among riders.
Just how bad is the service? A common observation from riders is that they can walk to their destination without being passed by a streetcar. On the outer ends of the route, service can be unpredictable especially west of Humber Loop where only half of the service is even scheduled to travel and some of that is short-turned.
The TTC’s goal is to operate 70% of streetcar service within 3 minutes of the advertised headway. On Queen, scheduled headways at most times lie in the range from 5 to 7 minutes, and this translates to an acceptable band of service that treats gaps of up to 10 minutes as “punctual”. In practice, the route rarely attains that 70% score.
Service at Yonge Street
The first two sets of charts show the service as it actually operated at Yonge Street eastbound and westbound.
- Each set of charts contains 15 pages corresponding to various parts of weekdays, Saturdays and Sundays. The horizontal bars give the percentage of cars at Yonge Street on various headways between 0 and 30 minutes. (Each time interval spans from 30 second below to 30 seconds above the value shown. The “30” bar contains any car with a headway greater than 29’30”.)
- If service were mainly on time, then most of the trips would be cluster in a few bars around the scheduled headway (which is shown in the legend on each page). Most of the charts do not show this pattern.
- November 2011 saw two separate schedules operated on Queen. For the first three weeks, it was the “normal” service, but effective Sunday, November 20, the schedule provided for a diversion around Metrolinx construction at Queen and Dufferin with extra running time and slightly wider scheduled headways. The diversion did not actually start until Saturday, November 26.
- On the charts, two lines are shown for each headway value with data for the period of the two schedules split apart.
During the AM peak, most of the headways westbound at Yonge lies in a band between 1 and 8 minutes. This is roughly in line with the TTC’s ±3 minute target, although there is no peak in values right at the scheduled headway.
By midday weekdays, the values start to spread out with more in the range above 10 minutes, and a relatively high number of 1 and 2 minute headways. The latter are, for practical purposes, cars that are running as or very close to pairs showing that bunching is developing as a common situation. Over 30% of the headways are 2’30” or less.
By the PM peak, things are more or less back to where they were in the AM.
In the early evening, the same effect as at midday appears with headway values spreading out and bunching re-appearing. By late evening, things get really bad with headways fairly evenly distributed between 2 and 12 minutes on a nominal 10 minute headway, and about 20% of the service is bunched.
On Saturdays, early morning service is spread over a range of headways up to 13 minutes on a nomally 7’30” schedule, but bunching is already evident. The situation continues into the later part of the morning even though the scheduled headway is now 6’15”. By the afternoon, the most common headways are from 2’30” down on a schedule of 4’20”. In effect, much of the service is running as two-car trains. This continues into the evening on a wider scheduled headway.
Sunday is even worse than Saturday with headways all over the map even though this cannot be explained by “congestion” especially in the morning.
Service to and from The Beach
The Beach provides a different view of service. Westbound, especially early in the day, service should run close to schedule and with reliable headways because nothing has had a chance to interfere yet. As the day progresses things can deteriorate. We already saw eastbound service at Yonge which was quite unreliable, but at Woodbine, we see the effect of short turns at Russell Carhouse (east of Greenwood) or at Woodbine Loop (Kingston Road).
Westbound at Neville in the AM peak, the headways are clustered around the 5 minute bar as one would expect for “on time” service. Even so, this leaves some room for variation with headways above 8 and at or below 2 minutes lying outside of the target band.
The clustering of headways around the schedule continues into midday, but by now longer gaps are showing up with several cars at over 10 minute headways. The comparable chart at Greenwood looks much like the AM peak showing that service west of Woodbine Loop is being maintained by short-turns. The pattern continues into the PM peak. Through the evening, service regularity starts to unwind, especially in the late evening with a wide range of values. (The bunching westbound from Neville is not seen at Greenwood because these are cars running in to Russell carhouse and typically following close behind a through car.)
Saturday service is not as well behaved as on weekdays, but the afternoon period is particularly bad with many wide gaps from Neville. Again, the chart for Greenwood is better (although still not ideal) showing the effect of short-turns at Woodbine Loop. Saturday evenings are a complete shambles even at Greenwood, let alone at Neville, in part because there are fewer cars to short-turn into gaps.
Sunday early morning service headways cluster around the scheduled value, but there instances of cars running well off-headway to be troubling. By late morning, the service westbound at Neville and at Greenwood has become spread out into a wide range of headways. By the afternoon, things are even worse at Neville with gaps over 20 minutes common, and although Greenwood benefits from short-turn activity, it has a lot of very short headways suggesting cars running in pairs. Sunday evening isn’t much better, although by the late evening, some degree of clustering around the scheduled headway reappears in the data.
Service eastbound at Woodbine is considerably worse than what we saw eastbound at Yonge in previous charts revealing the effect of short-turns. Headways are spread all over the map with multiple-headway gaps common. Service gets to Neville almost by accident.
This confirms what anyone living in The Beach has known for years — the Queen car does not provide reliable service because so many of them short-turn and the headways are unpredictable ranging over a wide span. This service is miles outside of the target any transit system, even the TTC, would consider acceptable.
While it could be claimed that less service is required to the eastern end of the line, the problem is that this service is unpredictable and very wide gaps are common. The Beach is an affluent neighbourhood where driving is a common option for many residents, and the TTC service does not provide an attractive alternative.
Service in Parkdale
Most of the service westbound from Yonge reaches at least Roncesvalles although headways may be ragged by the time it arrives. Inbound service is a mixture of through cars from Long Branch and Humber mixed in with short-turns from Sunnyside Loop.
As in The Beach, service east and west of Roncesvalles differs due to the effect of short-turns at Sunnyside. Triller is just each of Roncesvalles, and Parkside is the eastern edge of High Park where the open track on The Queensway begins.
Weekday service does not show the peaking in data values we saw at Neville and Greenwood, and headways are scattered over a considerable range, especially late in the evening. At that time, a 10’00” headway is scheduled, but many cars are well below and above the ±3 minute band. Service at Triller is somewhat better than at Parkside, but the difference is not as marked we see in the east end where Woodbine Loop short turns are quite common.
Saturday service is very unreliable with headways scattered over a wide range especially in the afternoon and evening. Service at Triller is slightly better than at Parkside. (Note that there was no Saturday service at Triller for week 4 because of the diversion via King.)
Sunday service is more or less the same as Saturday’s. Particularly galling is the almost uniform distribution of headway values during the late evening period.
This data suggests that little or no effort is made to manage the service to anything close to the scheduled headway, and cars will show up more or less randomly.
And Finally in Long Branch
Service west of Humber Loop is scheduled at half the level of the main Queen line to Neville. It is subject to the compound effects of any problems east of Roncesvalles and, as in The Beach, gets service at times almost by accident.
Service westbound from Humber to Long Branch shows the effect of cars making a one-hour trip across the city. Even in the AM peak, a nominally 10 minute headway is smeared out over a wide range of values from 0 to 20 minutes. This pattern continues throughout weekdays and is especially bad late evenings when the 21 minute headway can be anything 5 minutes to almost half an hour with equal probability.
Weekend service is totally unpredictable even first thing in the morning.
Eastbound service from Long Branch Loop shows the effect of built-in recovery time built with headways clustered around the scheduled values. The eastbound service is better behaved than the westbound service at Humber. However, by late evening, inbound service is a shambles with headways ranging fairly uniformly from 0 to 30 minutes.
Saturdays start out with service reasonably closes to target, but they quickly fall apart and don’t recover for the rest of the day. Sunday is more or less like Saturday except that the late evening service recovers some reliability.
What is particularly frustrating for would-be riders on Lake Shore between Humber and Long Branch is that outside the peak period over half of the trips are local, they are not going east of Humber. However, the service, if we can call it that, is a shambles thanks to a combination of the hook-up with Queen, the effects of short-turns on what actually gets west of Humber, and absolutely no attempt to regulate headways.
I will continue this series with data for other months and with different presentations of the information to dig into what is actually happening.
Postscript — Data Preparation
This is for the benefit of comparatively recent readers of the site who have not seen my service analyses before. It gives a once-over-lightly explanation of how these charts were produced.
- Raw vehicle tracking data are supplied by the TTC giving the GPS co-ordinates of all cars on the line at (usually) 20 second intervals.
- The GPS locations are “cleaned” by discarding values that are off-route (caused by diversions and by GPS resolution errors).
- The locations are mapped into a linear set of values representing the distance from Neville to Long Branch. This allows the creation of graphic timetables showing the “as operated” service (this method of presentation has been around since 1885).
- An “as operated” timetable can be constructed by finding the times when each car crosses a point of interest (say Queen & Yonge).
- From the “timetable”, headways between each car can be calculated. The timetables for two points on the line can be used to calculate travel times from one location to another.
- The initial analysis is done one day at a time, and the results are consolidated into monthly overviews.
Yes, it’s a lot of work, but I’ve been programming since the early 1960s and much of the work is now automated.