In the first part of this series, I reviewed the headways operated on 7 Bathurst Bus during the months of March and April 2014, with December 2006 for historical comparison.
This article looks at running times for the route, the time needed for buses to travel from one place to another, and the differences between each of the three months’ worth of data.
The Scheduled Service
The table linked here shows the headways and running times for the scheduled service in the three months reviewed here. (The same information appeared in Part I.)
Charts of the Link Time Data
(The term “link” as used here refers to one segment of a route. It could be a few kilometres long — St. Clair to Eglinton for example — or it could be the entire route.)
At the simplest level, the following two sets of charts show the running times between Barton (just north of Bathurst Station) and at the beginning of the loop at Steeles in each direction for the three months in question. The two points are chosen to include almost all of the route for its driving and stop service time, but to omit the terminals and their varying layovers.
7_20062014_BartonSteeles_LinkStats
7_20062014_SteelesBarton_LinkStats
Each set contains three charts, one for weekdays, one for Saturdays, and one for Sundays/Holidays.
The charts show that running times in 2014 are longer than in 2006 at most times of the day. Also weekday times in April 2014 with service provided primarily by articulated buses are slightly longer than in March 2014 with standard sized buses. However, there is a similar spread on weekends when the service was operated with standard buses in both months, and it is not immediately obvious that the artics took inherently more time to make their trips.
Unlike the situation with the headways (discussed in the previous article), the standard deviation values are much lower than the travel times showing that these values are better clustered around their averages. However, the SDs do tend to float around 5 minutes and rise much higher particularly northbound in the PM peak.
Comparison of Trip Times by Vehicle Type
During both March and April 2014, 7 Bathurst was operated by a mix of regular sized and articulated buses. Most of the charts in this article include data for both vehicle types, but an obvious question is whether they actually have the same typical trip times over the route.
Two factors may affect these times:
- Relative performance of articulated buses on hills, especially northbound, the prevailing “uphill” direction on the route.
- Longer stop service times for loading more passengers per vehicle after schedules were changed to reflect the larger vehicle capacity of artics in April.
The first two pages in this set of charts show the average running times each way from Barton to south of Steeles for March 2014 broken out by vehicle type. The next two pages show the same information for April 2014.
Generally speaking, the running time for artics is longer than for standard buses, although this is more pronounced northbound and in March. By April, the better speed of the standard buses may have been offset to some degree during peak periods by longer stop service time. (Confirming this requires a detailed review of the data which I am not undertaking here.)
Of course, the fact that there even are data for regular sized buses in April indicates that the TTC is not fielding a 100% artic fleet even though the scheduled headways presume this is so. The fifth page shows the number of trips northbound from Barton by hour for each month made by articulated and regular buses. In March, the artics were already the dominant mode on the route, but in April, regular buses show up particularly later in the day.
This suggests that there are not enough artics to reliably fill the schedule.
Comparison of Actual Averages and Scheduled Trip Times
The charts above do not include time near and at terminal locations. When this is added, we can see how the total round trip times vary over the day.
These charts, with separate pages for weekdays, Saturdays and Sundays show four components of running time for April 2014 throughout the day. These include the average times for the north and southbound trips plus the terminal time at each end of the line.
The scheduled round trips on 7 Bathurst for April 2014 were 120 and 124 minutes for the AM and PM peaks respectively, including recovery time. The midday scheduled time is 105 minutes. Actual averages are higher than the scheduled values at some times.
Some trips of greater than average length will not make the scheduled running times especially when external events such as weather or unusual congestion slows everything down. This can affect the headway reliability which, as shown in the first article, is quite poor.
Looking at More Detail
In this section, for each month, there are sets of charts for each direction of travel between Barton Avenue and the south end of the on street loop at Steeles.
7_201404_NB_BartonAve_SofSteeles_MonthLinks
7_201404_SB_SofSteeles_BartonAve_MonthLinks
7_201403_NB_BartonAve_SofSteeles_MonthLinks
7_201403_SB_SofSteeles_BartonAve_MonthLinks
7_200612_NB_BartonAve_SofSteeles_MonthLinks
7_200612_SB_SofSteeles_BartonAve_MonthLinks
There are ten or eleven (depending on the month) pages in a format similar to that used in the first article on headways.
- The first four or five pages are weekly plots of actual running times with trend lines interpolated for each weekday. Unusual days show up here because the data do not follow the same pattern as other days in the week. For example, there was a storm on March 12, 2014 that affected trip times from roughly noon to 8:00pm in both directions.
- The next page combines the data for all weekdays. This “cloud” of data points gives a sense of the dispersal of values (this is also reflected in statistics with lower standard deviations for more tightly clustered values).
- The next two pages show the data for Saturdays and Sundays/Holidays in the same format as the weekly pages.
- The last three pages provide averages and standard deviations on an hourly basis for the weekday, Saturday and Sunday data. This is the same information that is consolidated in the summary charts above.
It is worth noting that, unlike the headway data in Part I, the standard deviations of the link times are quite low compared to the values. Where the SD does rise, this indicates more scatter in the data values. Generally speaking, the SD is five minutes or less indicating the spread of typical values around the averages.
The day-by-day values are fairly consistent as shown by their trend lines. In other words, although individual vehicles might be affected by events such as breakdowns, the time needed for trips is predictable within an expected range of values.
However, there is a noticeable pattern that running times for days early in the week, especially Tuesdays, are shorter than for days later in the week. I will explore this in more detail in the next article.
From a planning point of view, the challenge is a tradeoff between giving enough running time for “typical” trips as opposed to trips under “unusual” circumstances such as bad weather, not to mention the variation in what is “typical” simply for each day of the week.
Behaviour Within Individual Links of the Route
The MonthLinkStats files contain twelve pages each which looks at segments of the route northbound and southbound. Averages and standard deviations are given for weekdays, Saturdays and Sundays.
7_201404_MonthLinkStats
7_201403_MonthLinkStats
7_200612_MonthLinkStats
Each segment of the route has its own colour to allow easy comparison from page to page. In almost all cases, there is not a large change in values for the times required for vehicles to cover each segment, and familiar patterns such as the weekday peaks and the weekend afternoon rise in running times show up across the route.
The one exception is with links at Wilson Avenue, mainly the Lawrence to Wilson link. The break between this and the next link to the north, Wilson to Sheppard, is defined just north of Wilson Avenue and specifically beyond the farside northbound stop. This means that all of the layover time taken for crew changes in either direction is included in the Lawrence to Wilson link times.
Looking at the daily details of the route’s operation, something I will address in the third article, the higher averages for this link arise from a combination of congestion around Highway 401 (northbound, PM peak only) and crew change layovers.
Terminals
Vehicles in and near terminals do not behave in a uniform manner notably that the exact location where they lay over is not always the same, especially for on-street loops. To avoid problems with identification of where a northbound trip ends and a southbound trip begins, the terminal areas are, for the purpose of these analyses, treated separately as a round trip from a nearby point on the route.
For northbound trips, this point is at Bathurst and Carpenter, the bottom end of the on street loop at Steeles. For southbound trips, this point is at Barton Avenue, the traffic signal just north of Bathurst Station.
7_201404_BloorTerminal_MonthLinks
7_201404_SteelesTerminal_MonthLinks
These charts show the detailed breakdown, trends and statistics for operation at the two terminal locations. What is particularly striking here is the very high values for time spent near and at the terminal during some periods of the day. These values tend to be higher for the same weekdays when running times over the route were shorter (e.g. Tuesdays). The higher standard deviation values relative to the averages also reflect the wide range in values.
A common sight for passengers is to see a bus arrive and the operator disappear for an extended period. This does not happen on all trips, but it is frequent enough that riders, although annoyed, will not be surprised. Are these layovers excessive, are the buses still managing to run on time with the operators taking advantage of excess time in the schedule on certain days? The erratic headways leaving terminals that were shown in the first article of the series suggest that, at a minimum, a common behaviour is to take whatever break is, shall we say, comfortable and then make up time along the route if possible.
Coming Next
In the third article, I will review day-of-week effects on the lengths of trips times and layovers, and will look at detailed operations on some days as examples of headway management problems.
Steve Munro 
Rear Doors
There is a safety feature built in to the buses rear exit doors. The brakes lock up until the doors are closed, and stay locked for another instant. You can hear the air brakes release. It may not seem to be related to the post, but the modern buses have much slower rear doors, and this
coulddoes add trip time.Some drivers feel the pressure of the clock, and try to close the doors sooner. This results in a worse delay: passengers force the rear door, which locks it and causes an alarm. It also makes the door unreliable, and it won’t fully close. The driver walks to the door, and pulls it shut by hand.
Lastly, the rear door brakes won’t release occasionally. The driver will do some things, but in the end, the bus will be evacuated and taken out of service. It is forbidden to carry passengers if the the rear door safety feature is defeated.
Rear doors are used more during heavy periods, so the stats don’t change the graphs, they just offer a suggestion of an additional excuse for slower service.
Steve: Yes, in general some things in “modern” transit vehicles work against better stop service times. It will be interesting to see the effect of all-door loading, when it finally arrives, on the operation of bus routes.
LikeLike
I find this more than moderately disturbing. Is this service being managed, or is it being resourced? This plus the other article make it seem as though someone is calculating the number of buses and drivers required and allocating them, and that is all.
LikeLike
Do the trip times include the amount of time spent waiting for the bus? I’m not so interested in how long the bus takes to get from Point A to Point B. I want to know how long it takes ME to get from Point A to Point B, including the amount of time I have to wait at a stop for a bus (that has space for me to board).
Steve: The trip times shown here are the time for the bus to get from A to B. For the wait times, you have to look at the first article that dealt with headways which are only distantly related to information in the printed schedules.
LikeLike
About the TTC exit doors on buses. Look at these videos on the exit door of a London bus. Toronto’s buses, in comparison, take forever.
Steve: Important differences are immediately obvious. The doors slide sideways rather than pivoting, and they are moved by door motors using arms that pull the doors back and forth into position. They are actively restored to the closed position rather than depending on the timing of sprung doors to close, but not too quickly, on passengers.
LikeLike
A rode the Peter Witt car at the Beaches Easter parade and got off at the rear door at Neville Loop and was amazed at how quickly the door closed. Effortless. The interior is spacious especially compared to the new cars. I think the Witt could handle the same number of passengers!
BTW 4403 was out on St.Clair this afternoon (saw it at Earlscourt) NOT IN SERVICE said the sign! The operator on my car made an announcement to riders of its approach.
LikeLike
The “dwell time” can be defined as the time a transit vehicle (bus, streetcar, light rail vehicle, heavy rail vehicle, commuter train) spends at a stop or station awaiting continued movement.
With the wider doors, starting on the T-1 subway cars, and the set of four doors on the new low-floor streetcars, they hope to reduce the dwell times.
However, if the articulated buses continue to use the pivoting doors on the exit doors, I don’t see much reduction of dwell time on them.
Steve: A related problem for longer vehicles will be aligning the bus with the stop so that there isn’t too much of a gap between the door and the curb. On streetcar routes, the question will be whether the TTC persists on demanding that ops pull right up to the stop (something the better ones ignore already as this usually wastes a traffic signal cycle).
LikeLike
Steve, so in essense it appears that you are saying that larger vehicles are not solving, by their mere presence, the headway management issues. While they may have made running time slightly longer, the variance (as a portion of trip time) does not appear to have changed much.
I was hoping to see some data that backed the TTC’s previous apparent position that the headway issue would be addressed by larger vehicles. It would appear that larger vehicles do not address dispatch control issues.
It will be interesting to see if over time the absolute variance in headway gets larger, as they convert to 100% larger buses and drivers become completely comfortable with the new operations. If drivers are deliberately driving together, or merely driving into a gap I would expect this to get worse over time, with larger vehicles, as a reduction in the number of buses on the line should exaggerate the issues.
Again the TTC needs to work on line management, especially with an eye to the fact that if you reduce the number of vehicles servicing a line having 2 together means an even larger increase in the size of headway. Management should be viewing the larger vehicles as potentially (likely) compounding problems in service levels, and working to preempt this. Wasting streetcars meant to carry 150, because they are hard behind another, on a busy line will be a disaster.
Steve: Against all evidence, I hope to see the TTC improve their line management, but this will not be possible if they cling to their mythology regarding the behaviour of transit routes and the limitations on their own efforts. Their new route map includes designation of the “frequent service network” with thicker lines on the map, but this will be an empty promise if line management persists in using short turns as a standard mechanism to keep cars and buses on time.
FYI: The claim that fewer, larger vehicles would provide more reliable service comes from a research paper that looked at a completely different, and very specific situation, namely frequent streetcar service on King Street with closely-spaced traffic signals and busy stops. In that situation, fewer, larger streetcars meant that traffic in general was blocked for stop service less often and could move slightly more freely. This does not extrapolate to wider headways on relatively uncongested streets with fewer traffic signals, especially with buses that pull out of the traffic flow to serve stops. However, TTC mythology ignores the differences and says “bigger buses are better” as an article of faith. More like a false god, I think, and the TTC has too many of them.
LikeLike
Unfortunately this sort of thinking seems to be a problem elsewhere in statistical analysis as well. See correlation, assume causation, without having a reasonable understanding of how.
The King Street situation makes perfect sense, as the streetcars themselves would become a material part of traffic. On that point however, does the TTC not run CLRVs on King and ALRVs on less busy and frequent Queen?
Steve: You are not supposed to notice that there are real world counterexamples staring them in the face. When you get a study that “supports” your bias, you stop asking questions. Isn’t that the way it’s done?
LikeLike
So you are saying that the TTC is like a vegan reacting to the correlation between shark attacks and ice cream sales? {Since ice cream is a sin, it must be the cause of the attacks, and therefore they are safe, and it neither has anything to do with the hot day}.
Steve: There have been problems with the Queen car for decades since the introduction of ALRVs and the widening of headways. The TTC has steadfastly refused to acknowledge that their laissez-faire approach to line management compounds the headway issue, not to mention the length of the route.
LikeLike
So how does the TTC approach the issue of line/route management on the subways? Is the timing entirely based on signals geared only to safety, counting on trains running at the same rate and allowing averages to do the job, or do they work to maintain headway in off peak hours-where trains are not as close as they can get them? Are they better at maintaining headway on subways because they can do so with signals (and therefore it is easier) or because they think that actually doing so is more important?
If it is the former, it would argue strongly for introducing at least some sort of start signals for buses and streetcars at the terminus points, that could then later be tied into a tracking, fleet, headway/schedule management tool, unless they are going to have said tool very shortly. I worry that this will have to be a tool developed by and for the TTC (because they will either have or create the perception of unique needs), and thus it will be a multi-year project even after they have done a study on requirements etc.
Meantime, there are likely some more basic things that could be done at the end points of the line that would mitigate (not solve) the issues.
Steve, has the TTC gone through any sort of recent analysis that would mean they have identified the sources and type of data, nature of required communication and feedback to and from operators? I worry a little about this, purely because they do not appear to even acknowledge that they have an issue.
Running a smaller vehicle on a route that is capacity limited seems silly, especially when they already believe that this is part of a headway management issue. Wasting the larger vehicles on a route that is seen to have thin service is ridiculous. This seems very nearly purely stubborn and ornery (like short turning vehicles that are more than half full several blocks before the terminus).
For any system they bring in to really address the long term needs of the city, the basic notion of what service is intended to be, what it should look like, what the metrics that are reasonable representations of good service are all critical. They should be using data that it would appear at this time they are not collecting. Before you choose to short turn a vehicle, you need to know how full it is.
Knowing the detailed loading patterns of routes, would help tremendously in terms of being able to dispatch vehicles to match, without radically altering the number of vehicles or loading standards. Some of the major causes of large and disruptive point loads will be quite regular and operating in a manner to preposition near empty vehicles will allow them to be more reasonably absorbed. Management systems being able to identify these issues dynamically is critical. Actually being of the mind that delivering that service is even more important.
Steve: Subway trains are dispatched based on scheduled times, not on headways, from the terminals and from various time points along the routes. However, the time a signal turns green is based not on the specific train, but on whichever one happens to be there. This effectively gives dispatching based on headways. However, that only works if the service is on time or slightly early. If service is late, everything is dispatched the moment it appears.
The idea of a train just sitting because the operator feels like it is unknown. That said, it is quite common for crews to be “short turned” by swapping trains enroute. This gets the crews back on time but leaves a continuous through service for riders. Actual short turns of trains are much rarer, and are generally a response to the need to fill a gap.
LikeLike