This article continues the series on service reliability on short routes. The common theme is that the routes in question have short trips, and recovery to scheduled times should be simply achieved. Most of them also have little traffic congestion, and that cannot be cited as the root of all problems.
On some occasions, particularly later in the year when staffing shortages hit the TTC, some buses might be missing. However, this does not explain irregular headways when all of the scheduled vehicles are in service.
When one bus is missing on a short route, this can have a big effect on the service level especially when there were only two or three to begin with.
The TTC claims that some of the gaps are actually filled by “Run as Directed” (RAD) buses, but there are problems with that explanation:
- There are far too few RADs in service at any time to fill the missing service seen on many routes across the system.
- If a RAD operator takes over an open, scheduled crew, then the bus should run with the proper route identification and show up in the tracking logs. “Route 600” RAD buses only make selected trips on routes and do not appear in route-specific tracking logs.
- If all of the scheduled buses are in service, but they are running erratically, notably with two or more buses running together, this is an issue line management and service spacing.
In this mini-series, I will review the following routes:
- 64 Main
- 92 Woodbine South
- 121 Esplanade-River (formerly Front-Esplanade)
- 124 Sunnybrook
(For those who are wondering, the next group on my radar will be many of the major routes in Scarborough.)
Note that due to the cyber-attack on the TTC and the recover efforts that followed, there are no data for the following periods:
- Friday, October 29 to Saturday, November 6
- Sunday, November 7 data begin after 10am
- Friday, November 12 through Monday, November 15
- Saturday-Sunday, November 20-21.
There are also no data for:
- Friday, October 15 to Sunday, October 17 at about 2pm.
- Saturday, October 23 from about 10pm to 11pm.
Despite these gaps, plenty of data remains to show how the route behaves.
The short version:
- Scheduled running times were too tight on 64 Main until mid-November. This was “fixed” by buses dropping trips to get back on time, and less service was provided than advertised.
- Ongoing problems with missing buses and bunching compounded the schedule issue, and persisted into December.
- Weekend service was particularly bad when only one bus was operating.
- With very rare exceptions, there are no problems with traffic congestion as a stock excuse for irregularity in service.
Scheduled Service Level
The 64 Main bus operates from Main Station just north of Danforth Avenue to Queen Street East in The Beach. This is about as simple a route as one can get, and it is the shortest route in the system with a round trip distance of 4.81km.
The schedules in effect through October to November 20 had very tight running times. On November 21 these were changed during all operating periods to use the same number of buses, but on wider headways and therefore with longer trip times.
In theory, this change should have produced more reliable service for the last schedule period of 2021.
Service Southbound from Main Station
The problem with service quality related to tight schedules can be seen very clearly on 64 Main. The charts below show the average headways, week by week, over three months (solid lines) and their standard deviations (a measure of the scatter of values around the average, dotted lines).
In October, the averages consistently run well above the scheduled headways. The reason for this, as we will see later, is that buses were actually taking longer trips than scheduled, and therefore the gaps between buses were longer. “Recovery” times at Main Station tended to be long as buses waited for their next scheduled trip. The effect is that many fewer trips than scheduled were actually operated. Note that October charts have a maximum Y-value of 40 minutes simply to fit in the range of data. Other months use 30.
In November, there is a striking change mid-month with the new schedules. Even though the advertised headways rose, the actual averages fell during almost all periods in the latter part of the month. Average headways improved because buses were no longer taking extended layovers to compensate for tight schedules. However, at the same time the standard deviation values went up indicating that service was more erratic.
December continued the improvement, although with some variation through the month due to missing buses and wide gaps in service. Even with improved schedules, service was erratic simply because a bus was often missing. It took a while for the headways to settle down. By late December the combined effect of the holiday season and reduced demand thanks to the return to stage 2 covid rules dampened traffic. This is seen in the lower value of standard deviations (dotted lines) in headways for the last two weeks.
More frequent service, on average, is of little use if there can be a wide variation in vehicle spacing. Riders might “get lucky” and catch the second of two buses after a short wait, but equally they could face a very long wait if they miss the second bus in a bunch.
Here is a chart for October 12-14, 2021 (no data are available for October 15). Note that the vertical scale here tops out at 40 so that all of the data points will fit. There is a mixture here of very short headways (points close to the x-axis) and very long ones. The detailed data for October 14 (yellow) are presented later to show what was going on.
The following week was better-behaved. This is not unusual at the changeover between “board periods” for crewing. This took place on Sunday, October 10, and there is a good chance that the operators (and their availability) changed. After things settled down, by the second week of a period can have better results. They are not ideal for October 18-22, but they are preferable to the preceding week.
In November, here are the headways for the first week of the new schedules. Note that the top of the chart is at 30 minutes, not 40 as above for October. Even with the new schedules, there is quite a large amount of scatter in headways and some bunching (very short headways with points at or near the x-axis) is evident particularly in the early evening. (Data for November 22 are incomplete.)
In December, the week before Christmas saw a combination of lower traffic and the onset of the most recent restriction on public gatherings. The service here stays much closer to the scheduled values with little bunching.
Note that unlike TTC reporting metrics, the values shown in the monthly charts are broken down by time of day and by week. The only consolidation is that data for the same hour on all weekdays are merged (e.g. from 8-9am on the five days within one week). This keeps “likes” together so that the values are more representative of what a rider might see on a specific day and time than monthly averages. On weekly charts, the data are shown for each day so that anomalies are easy to spot and can be cross-referenced to detailed vehicle tracking data.
Thursday, October 14
The headways on October 14 were particularly bad as shown in the chart above. Here are the detailed tracking data for that day showing how the service behaved.
- In the AM peak, there are three buses in service although only two are scheduled. At times, two of them (yellow and turquoise) run as a pair.
- In the morning and midday, there are two buses as scheduled, but they sometimes run together as a pair.
- A third bus joins the route after 2pm, but the spacing between buses becomes very bad as the day continues to the point that these buses are running together between about 4 and 5pm. There is a brief improvement, but one bus (turquoise) leaves service just before 6pm, followed by another (blue) at about 6:30pm leaving one (pink) to provide the evening service. It is supposed to be a 14-16 minute headway, but in practice the headway is over 20 minutes and there is little terminal time.
Friday, November 26
One late November day, as an example, did not go well even with the new schedules because of missing buses and bunching. A gap opens up after noon and it persists until by 3pm, the two buses are running as a pair. By now there should be three buses in service, but that bus does no actually appear until the transition to the evening.
Tuesday, November 30
November 30 begins well enough, but after 9am things started to fall apart.
- Although the schedule summary calls for two buses in the AM peak, there were actually three, one of which disappeared after 9am.
- The two remaining buses, yellow and turquoise, ran together as a pair for much of the next three hours even though turquoise turned back at Queen to space the service just before 10am. This did not last long, and the buses were running as a pair half an hour later.
- By noon, service spacing was straightened out, probably thanks to new crews on these runs.
- Just before 3pm, a third bus, pink, joins in, but by now there is severe bunching with all three buses running in a pack.
- Through the pm peak, service remains uneven, and the yellow/turquoise pair reappears.
- Evening service, by contrast, is quite reliable.
Tuesday, December 7
In December, service remains erratic, especially in week 2. Tuesday, December 7 show the kind of things that can go wrong, especially during a period when the TTC was short-staffed.
- Service in the AM peak runs normally enough, but one bus (blue) misses two trips starting at about 8:30am.
- Two buses run in service, albeit at times as a pair, until 11am, at which point one of them disappears for good (blue).
- By the PM peak, there are three buses in service again, and they stay reasonably spaced.
- One bus leaves service around 6pm, and another after 8pm. This, more or less, is what the evening service is supposed to look like.
Thursday, December 16
Service on December 16 runs fairly reliably until the early afternoon when a third bus, yellow, joins in. Just before 4pm, all three buses run in a pack and they are spaced out on southbound trips from Main Station. Yellow gets a long layover at Kingston Road, but in spite of this manager to run as a pair with turquoise. Some bunching continues until after the PM peak when headways settle down to a vairly regular service.
The charts below show the data for Saturdays over the three months.
- Note that October is scaled with a maximum of 40 minutes while other months are scaled at 30.
- In October, daytime headways vary considerable over a range from 3 to 26 minutes with October 2 being particularly bad.
- The late evening service on October 23 shows a particularly high peak because there are no data between 10 and 11pm.
- In November, there is data for only one day because of the cyber-attack. It is for the firs Saturday with the new schedues.
- In December, entirely on the new schedules, headways are much better-behaved and lie in the range of scheduled values through the daytime period. However, scatter is worse in the early morning and evening.
- Note that December 27 (Boxing Day observed) runs on Sunday headways. I normally include this day with Saturdays because of shopping traffic on many routes. Headways on that day are very well behaved.
Saturday, October 2
Service in the early morning and through the evening is provided by one bus, with a second scheduled through the afternoon.
- The second bus (blue) arrives just after 9am, but does not run consistently spaced with the first one (yellow) until after 10am. Yellow disappears from service at about 12:30pm, and is later replaced by pink after 2:30pm.
- Pink mises a trip just after 5pm.
- In the evening, pink is the only bus in service, but it operates less often than scheduled with layovers at Main Station.
Saturday, October 23
The pattern on October 23 mirrors that of October 2, except that there is an hour’s missing data in the late evening.
Saturday, December 11
December 11 shows the new Saturday schedules in operation. The service is supposed to run every 20 minutes in the early morning and evenings, every 12’30” in the late morning, and every 10′ in the afternoon. The transition from the morning to the afternoon headway is achieved simply by cutting five minutes of recovery time out of the schedule.
By the evening, the pattern of quick turnarounds at Queen and layovers at Main Station is quite clear, but at least the advertised headway is provided most of the time.
The situation on Sundays is similar, although the scheduled service is less frequent.
- Note that October is scaled with a maximum of 40 minutes while other months are scaled at 30.
- In October, headways are scattered and their values lie generally above the scheduled frequency of 14-16 minutes.
- In November, for the two days with available data, one (November 7) is on the old schedules, while the other (November 28) is on the new. Neither is particularly good.
- In December, service on Christmas weekend is generally well-behaved, but earlier in the month shows signs of missed trips pulling the trend lines upward during the afternoons and evenings.
Sunday, October 24
On the October Sunday schedule, there is supposed to be one bus all day on a 14-16 minute headway (roughly four trip/hour, or 12 for each three-hour interval). The number of trips actually operated varies through the day. A second bus joins the route for about one hour after 1pm, but vanishes an hour later.
Sunday, November 7
Data on November 7 begin at about 10:45am. This day shows the same pattern as on October 24 above with a single bus whose headway varies through the day and is worst in the mid-to-late afternoon. A second unscheduled bus joins in after noon for a few hours, but spends much of its second hour running nose to tail with the other bus thereby providing little real added service.
Sunday, November 28
In the afternoon and evening, the actual headway is wider thanks to extra layover time taken at Main Station. From 1pm onward there are consistently fewer than the 9 trips one would expect in each three-hour interval from a 20 minute headway.
Sunday, December 5
As on November 28, the headways widen in the afternoon reaching a point where a scheduled 20 minute service extends to half-hourly.
Full chart sets for 64 Main southbound from Danforth:
Service Northbound from Queen
Service northbound from Queen shows similar patterns to the southbound operations.
Full chart sets for 64 Main northbound from Queen:
Travel Times and Terminal Layovers
Unlike the headways shown in charts above, travel times are quite consistent. They are shown in two formats below. First is the travel time over the route, second is the time spent at the terminals.
There was a change in mid-October where the south end of the line was no longer used as a layover point. This affects layover times at Main Station as well as the travel times between Danforth and Queen which no longer include a south end layover.
Also visible in the stats is that the standard deviations, a measure of the scatter in values, are quite consistent and small for travel times, less so for terminal layovers at Main Station.
Southbound from Danforth to Queen
The spike in the average and standard deviation in the last week of November (purple) is caused by unusually long travel times on November 30 (see detailed service chart earlier in the article). Because Week 5 contains only two days (November 29-30), a major disruption on one day can affect the “weekly” values more than it would in a regular five-day week.
In December week 3, there is a spike in the standard deviation caused by a bus holding for its time enroute (as shown in the service chart for December 16 earlier in the article).
Full chart sets:
Northbound from Queen to Danforth
Northbound travel times mirror the southbound values. The spike in the standard deviation for week 2 of October is caused by the service disruption shown in the charts earlier in the article.
Full chart sets:
Main Station Terminal Time
The bulk of layover time on 64 Main is taken at Main Station, and the average values are in some cases quite long as buses drop trips to get back on time. There is a marked change in values in mid-November with the new schedules. Average terminal times went down because buses were operating all of their scheduled trips rather than sitting until their next due departures.
Full chart sets:
Queen Street Terminal Time
At Queen, the terminal is a round-the-block loop. Because the location of a layover could vary, the times are measured from the northeast corner of the loop around the circuit.
For the first two weeks of October, buses took some of their layover time at Queen, but shifted to using Main Station more from mid-October onward. The standard deviations of times at the south end loop are small and fairly consistent showing that buses typically made the circuit and left again.
Full chart sets: