This article continues the series on 501 Queen car operations in November 2011 by looking at the level of service operated beyond common short turn points on the route. This should be read in conjunction with the previous articles on the route’s headways and running times.
The TTC has a target for headway “punctuality” of ±3 minutes of the scheduled value, and a target that this should be met 70% of the time. In practice, the space between many vehicles exceeds this target, sometimes quite substantially, but much is hidden by averaging of values over multiple locations, times and days. Riders, however, experience service at specific times and locations, and averages are cold comfort when they find themselves in the 30% of service that does not meet the target.
In the analysis of running times, it was clear that at certain times of certain days, the amount of time allocated for trips on the Queen route are somewhat less than the time a car actually needs to get from Neville to Humber or to Long Branch. This leads to spates of short-turning in an attempt to get cars back on time. Although the TTC’s target may be headway-based, its management tactics focus on preservation of schedules.
Readers of previous analyses will be familiar with plots of headway patterns, or rather the lack of pattern in many cases, shown as scatter diagrams of time of day versus headway for various locations. This is a bit abstract for some readers, and I have attempted to produce a simpler measure giving roughly the same information.
The amount of service, defined as capacity, can be simply expressed as the number of vehicles per hour that reach a point on a route. Ten cars an hour is twice the service of five cars an hour. What this measure loses is any sense of the regularity of service (something the article on headways addresses) because there is no definition of when those ten cars actually showed up. They could all come in a pack followed by a gap over 45 minutes long, but they would still be “10 cars per hour”. Such are the dangers of averaging data even at a more finely-grained level than is done in published statistics by the TTC.
However, for the purpose of this discussion, cars/hour does well enough to illustrate what riders on the line must put up with.
Service in The Beach
The following two sets of charts show the number of cars per hour for each direction on the east end of the route. There are separate pages with the data broken out by week of the month, and by day of the week. (It is possible to have fractional values because over one week, the number of cars in a given one-hour period may not be a multiple of the number of days in the week. This would only occur if the service were identical on each day included in a group.)
Each chart is broken down by hour, and within each hour there are three bars showing the number of cars/hour at:
- Queen & Woodbine (cars originating at / bound for at Neville)
- Queen east of Coxwell (includes cars originating/turning at Woodbine Loop)
- Queen & Greenwood (includes cars originating/turning at Russell Carhouse)
In general, the number of cars/hour originating at Neville is lower than the values further west as more cars join the route. Differences arise from various factors:
- Cars that are short-turned only appear in the counts for locations west of the turnback point.
- Observations are at each location, the number of cars per hour at each location may differ slightly even without short turns. A car crossing Woodbine Avenue at 6:59 westbound will count in the 7:00 to 8:00 total at locations further west.
On weekdays westbound, the service east of Woodbine Loop is only slightly worse than the service to the west indicating that there are few short turns. However, immediately after the AM peak, in the 9:00 to 10:00 range on most days, the cars/hour value for Neville (blue bar) drops relative to Woodbine (purple bar) showing the effect of short-turns to straighten out the service following the peak. This pattern continues through the mid-day.
One notable exception is Week 4 when new schedules with added running time for a construction diversion were in place, but the diversion had not actually started. During this period, the number of short turns dropped quite noticeably compared with other weeks.
The charts by day of the week show some differences for weekdays, but the real change comes on Saturdays. From the running time analysis, it was clear that schedules on Saturday afternoons did not provide enough time for cars to make their round trips. Although more frequent service is scheduled, the actual level of service at Neville falls off beginning at 2:00 pm and never really recovers even into the evening.
Sunday afternoons show a similar pattern, but it is correct by about 8:00 pm. (Note that this data include the Santa Clause parade’s effect on Sunday, November 20, but this does not explain all of the short-turning if one examines the data day-by-day.)
The eastbound chart shows similar information, but for eastbound service which is of particular interest to PM peak and evening riders.
Again, I must caution that the numbers here are averages over one-hour periods with data from 3-to-5 days included on each chart. This can mask a worse distribution of actual data for individual days and times.
Service to Humber and Long Branch
The following charts show comparable data for the west end of the Queen route.
Each chart is broken down by hour, and within each hour there are four bars showing the number of cars/hour at:
- Lake Shore east of Long Branch Loop (cars originating at / bound for Long Branch)
- Lake Shore & Royal York (includes cars originating at / bound for Kipling Loop)
- The Queensway east of Humber Loop (includes cars originating at / bound for Humber)
- Queen at Triller Avenue (just east of Roncesvalles; includes cars originating at / bound for Roncesvalles Carhouse or Sunnyside Loop)
Unlike the east end of the line, the west end has a scheduled service that ends at Humber Loop. It is supposed to be 50% of the service with alternate cars running through to Long Branch. The infrequent 508 Lake Shore cars which operate to/from downtown via King in the peak periods are not included here.
In Weeks 1-3, the amount of short turning east of Humber is evident from the differences in cars/hour values for Triller (east of Roncesvalles) and Humber Loop. Counts for Long Branch are generally close to those for Royal York (with the wide and irregular headways on this part of the route, a one-hour window at one location will not “count” the same cars as at another some distance away).
In Week 4, as in the east end, the severity of short-turning drops somewhat. In Week 5, there is no data shown for Triller because all of the service was diverting via King.
On weekend afternoons, short-turning at Kipling reduces the level of service operating through to Long Branch.
On Tuesdays, the charts shows an anomaly with more service at Long Branch than at Royal York. This was caused by data from Tuesday, November 1 when a bus shuttle operated on the outer end of the route providing a regular, fairly frequent service that persisted for several hours after streetcar service was restored. This caused the vehicles/hour count at Long Branch to be higher than would normally be seen, especially when both services were operating. (I left this in as an example of the danger of looking at “standard” analyses without trying to understand factors that could produce distortion in the “average” results.)
As with east end data, there is a caveat that these are averages that mask the effects of individual days and of uneven headways within each hour.
The Regularity of Blended Services
Some readers have commented about the unevenness of headways that supposedly blend at Humber Loop eastbound. More generally, it is fairly common to see cars from short turns “blend” into the service unevenly spaced within the gap they supposedly are filling.
To get a sense of this, I took the data from Saturday, November 7 westbound at Greenwood and eastbound at Roncesvalles and separated it by the origin point of each trip. If, for example, cars from Humber Loop consistently showed a short headway, this would indicate that they were always pulling out right behind a through car from Long Branch. A similar consideration would apply to short turn cars leaving Woodbine Loop.
For the purpose of measuring the headways, I used points where, if someone were actually managing the service, there was an opportunity to space the cars directly in front of the carhouses they pass. (At Russell Carhouse (Connaught westbound) traffic is light, and a car holding for time could easily be passed by traffic. At Roncesvalles Carhouse, an eastbound streetcar does not block through traffic on The Queensway.) No such effort is visible in the data.
In these plots, the vertical position of each dot represents the headway of one car, and the horizontal position is the time. The colour/shape of the dot indicates the origin of the trip.
There is no real pattern to any of these data except for the onset of short-turning in the afternoon when running times are shorter than required. The headways of cars, regardless of their origin, are scattered over a wide range. Sometimes people will see a Humber car carrying a short headway, other times a Long Branch car. Fairly large gaps compared to scheduled service are not unusual as are pairs of cars entering the most congested part of the route together.
Westbound from Neville and Woodbine, there is no pattern to the spacing between vehicles.
This shows quite clearly that the primary effect of short-turning is to get cars back on time, not to fill gaps in service and provide regularly spaced vehicles for riders. The width of the “cloud” of data points also shows quite clearly how badly the TTC misses keeping all of its service within a three-minute band of a target headway.
I will return to these metrics in future articles looking at more recent time periods.