King Street Update: March 2019 Part II

In the first part of this review, I extended previous charts showing travel times through the Jarvis-Bathurst “pilot” area to the end of March 2019. This installment turns to the question of headway reliability – how regularly do streetcars arrive – as opposed to how long they require to cross the pilot district. A rider’s experience is affected by both of these factors, not to mention the basic question of “can I get on” when a streetcar does show up.

Although one end of a journey might be within the pilot, many trips on the King car begin or end beyond the pilot’s limits. They are affected by service quality along the whole route, not just in the core. This article begins by looking at March 2019 service at Yonge Street, and then moves further afield including the termini of the route.

The weather varied considerably over March 2019, but this did not play a big factor in the data overall, and statistics for most weeks and locations are similar. The larger issue is that service at the terminals is irregular most of the time, and this makes it impossible to have a properly “blended” service in the central part of the route.

The measurement shown in this article are taken at various screenlines along the route.

  • Southbound on Broadview crossing Danforth (leaving Broadview Station)
  • Northbound on Cherry crossing Mill (leaving Distillery Loop)
  • Westbound on King crossing Parliament (blended service from 504A/B). Note that any short turns coming south on Parliament are not included.
  • Westbound on King crossing Yonge
  • Southbound on Dundas crossing Bloor (leaving Dundas West Station)
  • Northbound on Dufferin crossing Springhurst (leaving Dufferin Loop)
  • Eastbound on King crossing Strachan (blended service)
  • Eastbound on King crossing Yonge

Headway Reliability at Yonge

The service at Yonge Street includes both the 504A and 504B branches of the King route. Their schedules normally have the same headway and so in theory this should be a blended service of cars alternating between destinations on a combined headway half that of each branch.

There are various ways of presenting the headway data each of which reveals a different aspect of the operation.

Headway Scatter Chart

Each dot on the chart represents one car with different colours for each day. The horizontal position is the time of day, and the vertical position is the headway in front of the car when it crossed Yonge Street. Note that because of the farside stop, few values are right at the zero line because cars tend to wait their turn and do not cross within the same 20-second interval used by the vehicle tracking system.

The wavy coloured lines are best fit curves threaded through the data to show how the values behave over the course of the day. Note that even though the dots for each day may be in different locations, the overall values as shown by the trend lines is quite similar for each day. In other words, one day is very much like another.

The scheduled combined headways of the King branches in March were:

  • 2’38” in the AM peak
  • 3’30” in the midday
  • 3’00” in the PM peak
  • 3’15” in the early evening
  • 4’30” in the late evening

The trend lines lie generally on the scheduled values as one would expect at the middle of the route where there should be no effects from short turns and all service is present.

The full Service Summary for 504 King is shown below.

Averages and Deviations

Another way to look at the same data is to see the hourly averages and standard deviations (a measure of how dispersed the values are around the average).

On this chart, each week’s data is displayed in a different colour (the red line corresponds to the week 1 data on the previous chart). The solid line shows the hourly average values and the dotted lines show the standard deviations. The latter values are of some concern because they lie close to the averages. This indicates that the band of data values around the averages for over half of the headways is close to the scheduled headway itself. In other words, many cars are running two headways apart, and by implication others are running very close together. This is evident in the scatter diagram, but the chart of averages and SDs shows that this is a consistent pattern for all weeks in the month.

Note that “week 1” actually is six days long because March 1 is a Friday and is included with the following week.

Quartiles

The chart below shows the week 1 data in “block and whisker” format breaking down the range for each of four quartiles.

Each column displays the data for one hour within week 1.  The green and blue boxes show the range of the second and third quartiles with the dividing line being the median value for the data. The red “whiskers” at the bottom show the range of the first quartile data, while the upper purple whiskers show the fourth quartile.

This chart shows that one quarter of the service operates well below the median headway, close to zero, and another quarter operates well above reaching values of 10 minutes and more throughout the day. The other half of the service lies within the green and blue boxes. The shorter this box is, the closer to the scheduled headway half of the service is operating, but the “whiskers” ideally should not be so long.

This is a simplified view of the scatter chart above. From a rider’s point of view, wait times are important and more people will accumulate at a stop during a wide headway than a narrow one. More riders “see” the gaps (and following bunches) even though the average and median headways might be close to the scheduled values. This also affects crowding on cars.

Weekends show little difference. The Saturday data are shown below, and the Sunday data are quite similar except for the early morning hours.

All of the charts above are for westbound service. The eastbound charts are quite similar. Full sets for both directions are available below.

A classic argument about transit service is that bunching is inevitable given the vagaries of traffic and the impossibility of running precisely to a schedule. By the time the service reaches Yonge in either direction, what might have been a well-behaved service has fallen apart. At least that’s the story. However, when we look further away, this is not born out by the data.

Eastbound at Strachan Avenue

Service eastbound at Strachan is a mixture of the 504A cars originating from Dundas West Station and 504B cars from Dufferin Loop, with the services merging at King & Dufferin.

The data scatter at Strachan is similar to that at Yonge showing that service is already bunched before it reaches the core, and that this is a day-to-day situation.

Consistently through all weeks and all times of the day, the standard deviation values are close to the average values. When this occurs, cars are often running in pairs.

The hourly breakdown for week 1 at Strachan shows the familiar pattern with the first quartile reaching down almost to the zero value, and the fourth quartile regularly cresting the 10 minute line. Half of the service lies in the middle quartiles over a range of headways that sometimes breaks the TTC’s six-minute target, but half lies beyond above and below.

The full set of charts for Strachan EB is linked here:

Westbound at Parliament

Service westbound at Parliament is a mixture of 504B cars originating at Broadview Station and 504A cars from Distillery Loop with the services merging a few blocks to the east at Sumach & King Streets. The situation here is quite similar to that at Strachan Avenue even though this is close to the point where “blended” service merge. Many headways are close to zero and many range up to 10 minutes.

The weekly breakdown shows that this pattern is common across all weeks and time periods.

As at Strachan, the quartile chart below shows clearly that half of the service is operating on very short or long headways, effectively as bunches, inbound before the service reaches the pilot district.

The full set of charts for Parliament WB is linked here. Note that on Sunday, March 17 all streetcars diverted due to a collision that blocked service for the entire day on King east of Parliament. Only the occasional shuttle bus shows up in the stats for that day.

Service From the Terminals

Service at the terminals does not come close to meeting TTC standards with headways swinging wildly about the scheduled values and a lot of service running close together. The situation is physically complicated at Broadview and Dundas West Stations where there are routinely queues of streetcars waiting to enter while cars take layovers because they are early.

Conversely, short turns of service destined for Dufferin Loop, and to a lesser extent Distillery Loop, imply that cars are late and will not be able to make their scheduled trips. When the May data are available, we will see whether the planned expansion of recovery time on both 504 King branches will fix this or will merely contribute to terminal queues and even longer layovers.

With the service from the four terminals consistently so irregular, it is no wonder that the “blending” of services cannot occur properly. On top of this, there is no evidence of any effort to space cars where the branches merge, challenging though this would be with erratic service arriving on each branch.

Broadview Station

During much of the day, the standard deviation values range from 4 to 6 minutes indicating that many trips are operating at a range of headways twice this wide. The pattern is consistent through all four weeks.

The range of headways covered by the second and third quartiles is routinely wider than 6 minutes, and the first and fourth quartiles extend considerably above and below.

The full set of charts is linked below. Note that there was no service to Broadview Station on the weekend of March 9/10 due to emergency track repairs at Riverdale Park.

Distillery Loop

As at Broadview Station there is a considerable scatter in the headways operated at Distillery Loop. On some days, short turning after the morning peak pulls the trend line higher. This occurred in weeks 1 and 3 as is shown in the weekly statistics chart. Short and very long headways are not unusual at Distillery Loop.

The following chart shows data for all weekdays and with the vertical scale set at 60 rather than 30 as in the other charts. The purpose is to show the cases where a headway value over half an hour was recorded. These show up in the “block and whisker” format charts as hours with fourth quartiles that run beyond the 30 minute upper range of the chart.

The week-by-week breakdown shows that the average headways stayed generally the same with a few exceptions, but that the standard deviations bounced over a range from about 3 to 8 minutes showing a wide variation in headways.

Looked at in quartile format, week 1 has several hours where the middle quartiles cover a fairly small range of values (e.g. 11 am), but the first and fourth do not. In particular, extremely short headways were common for much of the day especially in the late afternoon and evening.

504_201903_Mill_NB_MonthHeadways

Dundas West Station

The situation at the west end of 504 King is similar to that in the east with a wide range of headways for service departing from Dundas West Station.

The weekly statistics follow a similar pattern showing that this is common through the month.

Similarly the quartiles show that half of the service (first and fourth quartiles) operate on very short or very long headways well beyond the target range for TTC service.

504_201903_BloorWest_SB_MonthHeadways

Dufferin Loop

Service from Dufferin Loop can be quite irregular as the chart below shows. Short turns in the late morning cause wider headways leaving the loop after the AM peak and this shows up in the trend lines below.

In the weekly breakdown, the post-AM peak effect is visible together with very high standard deviations showing that even with wide headways, there are bunches where two cars are carrying a gap.

As at the Distillery Loop, the first and fourth quartiles regularly reach down close to the zero line and well above 20 minutes, cresting the half hour mark on occasion.

504_201903_Springhurst_NB_MonthHeadways

9 thoughts on “King Street Update: March 2019 Part II

  1. Steve,

    Excellent analysis as usual! This begs the question, is there line management? If so where is it and how is it done? I am guessing that “remote control” is used from the TTC central control office. Do they even have on location supervisors who monitor the merge at King and Dufferin or King and Sumach? Do they rely on the drivers to “space” themselves? Can the drivers report a crush load?

    The TTC used to run the largest fleet of streetcars in North America, you would think that they would have the institutional experience to run decent service.

    Is there a solution?

    Steve: I despair of the TTC ever getting a handle on this type of problem which is pervasive across the system. Years ago, when they first introduced the CIS tracking system, all of the on-street supervision was replaced with people watching consoles, and the content of the displays was not particularly reliable. The institutional memory of hands-on management was lost roughly at that time. I know that there are times when for some routes, nobody is “minding the store” and vehicles/operators more or less look after themselves. The merges are not managed at all, even though from a central location this would be easier to do than standing around in all sorts of weather dispatching service manually.

    TTC management “fixed” problems with short turns by padding running times, some of which needed fixing, but it has become a stock response. Meanwhile, riders see bunched vehicles that give the same effect as short turns with big gaps in service. This is a classic case of “managing to the metric” by making big improvements in one measure while totally ignoring another, the basic quality of service.

    Like

  2. Thank you for the excellent analysis.

    I frequently find myself at Danforth and Broadview and I often wonder at the lines of streetcars waiting to cross the Danforth. I assume the situation is similar at Bloor and Dundas.

    Can anything be done at these intersections? Can we give meaningful favour to streetcars?

    Can we forbid left turns and can we keep cars out of the centre lanes on Broadview? Would that even help?

    Steve: The usual reason streetcars cannot cross northbound is that there is already one in the loop and one waiting outside. The left turn advanced green northbound clears out traffic in front of the streetcars assuming there is room for them to cross.

    I will be looking at terminal operations in a separate article, and, yes, the same problem exists at Dundas West Station. The cars often have too much running time during certain periods, but the TTC’s fix is to add even more.

    Liked by 1 person

  3. Settlement reached between Bombardier and the TTC. Glad that it does not involve extra defective streetcars. Financial compensation also has more liquidity and flexibility. Can be used to purchase electric buses, articulated buses, or streetcars from another more reliable vendor.

    Like

  4. One thing about this report that strikes this long-time reader of Steve’s blog is that we’ve seen Steve’s analyses demonstrate many, many times over many years that, every time, the data conclusively prove poor line management, most especially at the termini. I have NO doubt that the analyses are accurate. Also, I have good reason to believe that TTC management read this blog on a regular basis.

    Given how long these proofs have been public here for all to see, one is forced to wonder if there is any responsible, professional TTC management at just about ANY level that feels moved in any way to attempt a solution to gaps and bunching. Certainly, after almost a decade using TTC-supplied data that prove problems at the termini, it is easy to see why Steve is exasperated. It cannot take a decade to improve management. So, one is forced to deduce that the TTC does not care about this issue in the long term; if they did, the dispatches from the termini would have improved at some point in the past decade, and the TTC’s data would have shown that.

    Clearly, TTC management have no shame when it comes to seeing long-term repetition of their dirty laundry being displayed in public.

    Like

  5. Trevor wrote:

    “Given how long these proofs have been public here for all to see, one is forced to wonder if there is any responsible, professional TTC management at just about ANY level that feels moved in any way to attempt a solution to gaps and bunching.”

    Yes, this is something that I have been wondering about for a long time. No company in a competitive industry would ever tolerate this kind of lousy service. Their competitors would quickly put them out of business.

    But the TTC has a monopoly… sigh…

    However, even government-run monopolies with unionized workforces can deliver efficient customer service. The LCBO is famous for being well-run, efficient and responsive to customer demand. So what’s up with the TTC?

    Is there a TTC management belief that being effectively managed for better customer service will create some kind of employee issue with their unionized workforce? I genuinely do not know.

    Steve, can you shed some light on this issue?

    Steve: Part of this is a long-standing accepted behaviour from drivers, but a big part is that I don’t think TTC really wants to undertake the effort required to restore regular service. The whole idea was lost decades ago. Andy Byford hired Rick Leary to “fix” surface ops, but all we got was “standards” that match what management thought they could hit easily, and padded running times to reduce short turns without dealing with regularity of headways.

    Like

  6. This analysis you’ve done Steve is brilliant!! A consultant would have charged the TTC thousands of dollars for this work. The mayor should be privy to this blog and the information being discussed here to so they can see the work you’ve done over the years to help improve the TTC. Unfortunately, it’s all politics down there man. You should be a member of the TTC board, period.

    Like

  7. Regarding the TTC being a monopoly: if Metrolinx/GO wished to, they could compete with TTC with services from Kipling, Long Branch, Main Street, Kennedy, Weston, Bloor/Dundas West, Agincourt, Sunnyside (if station rebuilt), Humber Bay / Park Lawn (if planned station was actually built), etc. Run shorter trains more frequently and charge the same or less than the TTC – that’d be some competition. They choose not to, though.

    Steve: And for the simple reason that this would be a waste of track and platform time on their network. A massive change to GO infrastructure and operating methods would be needed to run frequent “in town” service. This would still not deal with the basic structural problem that many stations are not located where there is much opportunity for walk-in traffic, and riders would still depend on the TTC just to access GO services.

    Like

  8. Kipling, Main Street, Park Lawn, Bloor all have ready walk-in customer bases from nearby towers, and Long Branch, Weston, Sunnyside areas are no less dense than some of TTC’s subway stations. You could add Oriole to that list with the towers south of Sheppard, but I accept its corridor has a right-of-way that makes it hard to compete. There would also be the possibility of competing on reverse commutes from downtown core to at least some of the employment lands.

    “A massive change to infrastructure” illustrates _why_ management and planning of rail-running transit is effectively a monopoly, though. Complain about poor management and lack of political will (*cough* transit priority), not the TTC being a monopoly.

    Like

  9. I was a conductor on crew operated London Transport bus routes in the early 1980’s. The culture then was very much of honouring the schedule in the breach. Early running in particular was a curse and it used to infuriate me such that I had frequent arguments with my drivers over it. It’s not that they were deliberately destructive, they just didn’t care and supervision was poor. They saw their job to just get the allocated vehicles out on the road. That was good enough. That they would adhere to time was too much to ask. Like I say maddening. Much less of a problem now.

    I left LT some years ago but it’s clear that all buses are now fitted with sophisticated kit which warns the driver if he / she is running early and they clearly do react to it, as in they obey the warnings. On road supervision is invisible so the technology is effective. I do know the London bus control room is sophisticated and well run. I visited the TTC tram control room in 2008 and it seemed pretty grim although everyone there meant well. I don’t know if it’s still the same. Your analysis highlights and proves the issue which potentially undermines good work the TTC does on overall service, promoting tram priority initiatives and so on so this ought to be a top priority for TTC management.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.