This post examines the details of operations for the 512 St. Clair route from Monday, January 4, to Friday, January 8, 2010. The next post in this series will review the last week of January for comparison to discover what improvements took place over the intervening weeks.
Future articles in this series will review weekend operations, as well as a month-long overview of the line’s behaviour.
When February 2010 data are available, I will examine the effect of new schedules introduced on Sunday, February 14.
Unlike the New Year’s Day operation discussed previously, the weekdays in week 1 were a mess, and service poor a great deal of the time.
January 4: Cold with some snow showers.
January 5: Cold with snow before the AM peak, cloudy in the AM, snow showers in the afternoon, snow in the evening.
January 6: Cold and mostly cloudy.
January 7: Cold and mostly cloudy.
January 8: Cold with snow in the AM peak and late morning, snow showers in the PM peak and early evening, mostly cloudy otherwise.
Monday, January 4
The Service begins fairly well, but a gap begins to develop just after 7:30. A car (dark blue) enters service and is closely followed eastbound by a second car (turquoise). These cars leave St. Clair Station together and remain close to each other until 9:15 when turquoise is short turned. Meanwhile, the gap in front of dark blue has been growing. It is short turned on its next eastbound trip and goes west at 9:38 in a pack of no less than five cars. At this point, most of the service is running in two packs, and the situation is not sorted out until late morning.
By mid-afternoon, bunching and short-turning reappear, and by 15:30, over half of the cars are running in a pack.
Between roughly 19:48 and 20:00, there are no cars east of St. Clair West Station.
At 20:55, two cars (light and dark green) form a pair eastbound from Lansdowne, and they remain together for two round trips at which point dark green runs out of service.
Tuesday, January 5
Ths service chart for this date includes some of the buses running between Keele and Oakwood. Although 8 are scheduled for the AM peak, only two appear on the chart. I suspect that most of the buses running the 512 shuttle don’t have GPS units yet.
This date had the most snow within the week and, as noted later, the weather produced wider variations in running times than on other days.
By about 8:00, the service has clustered into two groups of cars. Six cars leave Lansdowne eastbound between 8:06 and 8:12, and six more leave between roughly 8:24 and 8:31. The situation becomes even worse by 9:30 when eight cars, two of which will leave service eastbound at Vaughan, are running as a pack.
A great deal of short-turning follows through the day, but by 15:00, another group of five cars consolidates eastbound to Yonge. On their westbound trip, they are joined by one car entering service and three cars that short turned. All eight of them come east from Lansdowne between 15:25 and 15:40. Half of these are short-turned at St. Clair West. The problem with parades of cars followed by long gaps persists through the PM peak.
During the evening, when one might expect something like reliable service, cars continue to run in pairs.
Just after midnight, two streetcars are held near Yonge. The first of these appears to lay over at St. Clair West until just after 1:00 am when it goes out of service.
Meanwhile, two cars are short turned at St. Clair West, and a bus makes a trip east to Yonge Street. Note that the absence of data from all of the buses means that cases where the shuttles were pressed into service east of Oakwood is often missing from the data.
Wednesday, January 6
Only one of the shuttle buses on this date is reporting its location.
Some clustering of service during the AM peak is visible, but not as bad a situation as on the preceding two days. Through the midday and early afternoon, some gaps to Yonge are created by cars short turning at St. Clair West.
Starting at about 20:10, a gradual buildup of cars shows quite clearly that nobody is minding the store, or if they are, what passes for service management leaves much to be desired. At 20:05, “green”, which has been laying over during a short turn at St. Clair West, heads out closely behind “purple”. At Lansdowne, they catch up to “turquoise” which has been taking a lay over there. The three cars head east to Yonge, and “yellow” gradually catches up to them. On the westbound trip, “dark purple” joins the parade. Between 21:10 and 21:13, five cars leave Earlscourt Loop eastbound at a time when the scheduled headway is 5’30”. Various short turns sort this mess out, but regular headways are not visible again until about 22:30.
Thursday, January 7
As on Wednesday, the AM periods are better behaved than early in the week, but again we see the accumulation of groups of cars beginning after 13:00. By the PM peak, there are large gaps followed by parades of vehicles. Quite clearly, there is no attempt to space out service, only to short turn cars, often to join a parade in the opposite direction.
At 16:15, there is a delay eastbound near Dufferin which holds five cars that were already running more or less in a pack. Although some short turning is done eastbound at St. Clair West, the pattern of bunching continues, and service after the delay is almost indistinguishable from other times during the afternoon.
Bunching continues into the mid-evening. At around 22:45, there is major problem at St. Clair West and service does not resume until about 23:15. As noted above, it is possible that buses were redeployed to fill the large gaps in streetcar service, but the only bus transmitting data stayed resolutely on the Oakwood to Keele shuttle.
Friday, January 8
The pattern visible through this week shows up here too. Morning service is fairly reliable, but by early afternoon, parades begin to form. By 15:00, there is a cluster of nine cars westbound from St. Clair West in the space of less than 20 minutes. Four of these short turn eastbound, but they join another group of westbound cars. The remaining five continue as a pack to St. Clair Station.
The familiar pattern of two bunches of cars continues through the PM peak. Just after 17:20, several cars are held westbound approaching Lansdowne. This causes a large gap eastbound from Lansdowne affecting service at Yonge between 17:50 and 18:20. This is not sorted out until about 20:30.
Except for the delay at Lansdowne, the service problems are caused by the cumulative effect of bunching, short turning, and a total absence of headway regulation.
Although the service charts discussed above show the bunching, the headway charts give the riders’ perspective. In all of my service analyses, I have not seen such appallingly bad service and a complete abdication by the TTC of their duty to actually manage the vehicles on the street. The TTC mantra, “traffic congestion”, is not applicable on St. Clair.
In these charts, the bunching of cars shows up as clusters of points near the base line with spikes in between showing wide gaps. Particularly telling are the headways westbound at Bathurst immediately after a point where service regulation, including short turns of eastbound cars, might be expected to privide some regularity.
Rather than publishing each set of link charts for each day, I provide here the “week 1” page from the monthly consolidations. These show the comparison from day to day as well as the scatter in link times for each segment.
The Yonge to Lansdowne chart measures the time from crossing Yonge Street to arriving just east of Lansdowne. This eliminates peculiarities of moves at the terminals. Visible here is the longer running time required during the afternoon and early evening. This corresponds to the period of the most chaotic service when short-turning is constantly in use to get operators “on time”. Note how, despite all the problems with headway reliability, the actual travel times over the route stay in a fairly tight and predictable band, although the PM peak is not as good as the rest of the day.
In the individual link charts, we can see which parts of the route contribute the greatest variation to the overall running time. Note that the scale on the individual link charts is different from the one for the whole route.
Yonge to Tweedsmuir (just east of St. Clair West Station) shows a small rise in the PM peak. This looks the same as on the overall route chart, but the latter uses a 60 minute scale rather than a 20 minute scale. The trend lines range between roughly 6 and 8 minutes through the day.
Tweedsmuir to Bathurst shows a consistent link time, although there is a lot of scatter in the data on Tuesday, January 5, probably caused by layovers within St. Clair West Station.
Bathurst to Oakwood is also consistent, but with a lot of scatter. Oakwood to Dufferin shows extremely well behaved times over the days, as does Dufferin to Lansdowne (with the exception of data for Friday, January 8 showing the delay westbound approaching Lansdowne in the PM peak. This translated to long link times for the affected cars.
What is quite striking in these charts is that the greatest variation for individual segments lies east of Oakwood. I will return to this in a later article.
The times from Lansdowne toYonge eastbound predictably show a slighly higher bump in the AM peak and a correspondingly lower one in the PM peak as compared to the westbound times.
Running times from Lansdowne to Dufferin and from Dufferin to Oakwood are quite consistent. The delay eastbound near Dufferin on Thursday, January 7, shows up as extended running times for the Dufferin to Oakwood segment after 16:00.
Between Oakwood and Bathurst, the trend lines are flat, but there is a wider scatter in the data than for the sections west of Oakwood. This is comparable to what appeared in the westbound data.
Bathurst to Tweedsmuir data stay fairly close to the trend lines, although some outliers indicate cars laying over at St. Clair West Station before continuing further east.
From Tweedsmuir to Yonge, we again see scattered data, and the AM peak trend lines show a slight rise that is not present in the PM peak.
Overall, the greatest scatter in running times is usually found on Tuesday, January 5 (pink on the charts) when the heaviest snow for that week fell.
In the next article, I will examine the operations for week 4. By that period, route supervisors might have more experience with line management, operators would be more familiar with the route, and (according to info I received from the TTC) some changes in traffic signal behaviour should be evident.
For a future article, thanks to the fine-grained data available with GPS, I will review small line segments around intersections where dwell times due to signals and/or loading contribute to delays.