Analysis of 501 Queen: January-April 2019 (Part II – Headways at Humber Loop)

In Part I of this series, I reviewed the operation of the 501 Queen car at the east end of the route, specifically to review the amount of short turning and the regularity of the service. Now in Part II, I turn to the west end of the main segment of the route at Humber Loop.

It is no surprise that the results here are quite similar to those at Neville. Specifically:

  • The average headway at Humber is somewhat wider than at Roncesvalles, just east of Sunnyside Loop where some service scheduled for Humber short turns.
  • There is a peak in the difference between these locations just after the morning peak, and again after the pm peak showing the effect of recovery actions from those periods.
  • Headways (the time between cars) are quite erratic leaving the terminal at Humber, and this pattern continues east over the route.

An important issue here is that both short turns and erratic headways can have similar effects on riders, and they might not always know the difference.

If you are riding a car that is short turned, you know about this because you get turfed off and must transfer to the following car whenever it appears. However, if you are waiting for a car and there is a long gap, this could be due to a short turn (one or more cars is missing) or simply due to bunching (all the cars are there, but running in packs). Short turns affect riders on the outer parts of routes, while bunching affects riders across the entire line.

The TTC reports riding stats from time to time, although not anywhere as often as they should, and these are calculated as hourly averages. The problem with this is that averages do not reflect the uneven wait times, nor the uneven loading that results from a gap car carrying more passengers than its follower. Indeed, a route’s average load may lie within standards, but most of the riders are actually on crowded gap cars. This is a long-standing problem on the TTC and with the reporting of demand versus service. Also, of course, riders who never board are never counted, and we do not know the latent demand if only service were provided more reliably and with capacity for all to board.

Continue reading

Analysis of 501 Queen: January-April 2019 (Part I – Headways at Neville Loop)

A recent report by the CBC on May 8 claimed that the “transition to new streetcars will help alleviate short turns on major routes”. TTC spokesman Stuart Green, citing the May 2019 CEO’s Report, said that a large increase in short turns during March 2019 was caused by, among other things, the mix of new and old cars on the route with the strong implication that things will be better once Queen goes 100% low-floor in late June.

The CEO’s Report cites various factors:

Over the five-week period, the 501 Queen route made up approximately 40% of all short turns. This increase on the 501 route was due to several factors:

First, the reduction in Run-As-Directed (RAD) streetcars beginning in Week 8 (from 6 a.m./p.m. to 3 a.m./p.m.) has hurt the operational flexibility to respond to incidents and service gaps on all routes.

Second, the Queen route is largely serviced by CLRVs. These legacy vehicles experienced a high number of mechanical delays and disablements in March.

Finally, the Queen route is in a transition period with LFLRVs making up a small portion of the vehicles on the route, mixing with the CLRVs that have historically operated on the route. The different speed and operating characteristics of the two vehicle types inherently leads to more bunching and gapping on the route. [p 37]

The situation on Queen is even more complex than this and includes factors not listed in the CEO’s report:

  • Weather varied substantially over the winter months and yet short turns were at a substantially lower level in January and February.
  • Many delays on streetcar routes were caused by the City of Toronto’s failure to clear snow resulting in many cars parking foul of the tracks.
  • There is a persistent problem with uneven headways (bunching and gapping) leaving terminals, and little evidence of attempts to evenly space service. The mix of vehicle types can compound this problem depending on the capacity of the “gap car”.

From a rider’s point of view, irregular service and short turns may be indistinguishable especially for someone beyond a common short-turn location. On the east end of Queen, this is Woodbine Loop. It does not matter whether a 20 minute gap to Neville is caused by a short-turn at Woodbine, or because the service is running in packs separated by long gaps. In practice, most of the scheduled service does reach Neville, but it neither arrives nor leaves on a reliable headway (the time between cars).

At the May 8 TTC Board Meeting, Acting CEO Kirsten Watson noted that there had been a change in operational strategy in April whereby operators who were “late” to the schedule short-turned by trading vehicles with another car near a terminal rather than physically short-turning their own car. The result is that the operator gets back on time while their original car continues to the terminal.

Continue reading

TTC Updates Flexity/CLRV Replacement Schedule

Over past months there has been some inconsistency in TTC statements about the fate of the “legacy” CLRV and ALRV fleets with conflicting information that

  • some legacy cars would survive into early 2020,
  • all of these cars would be retired by the end of 2019,
  • all of the buses now operating on streetcar routes would be available for bus service improvements in 2020.

It is self-evident that these statements cannot all be true.

The situation is now clarified in two reports on the TTC Board’s Agenda for May 8, 2019.

The CEO’s Report includes the following:

On streetcar services, we’ll address crowding through the continued rollout of new high-capacity, low-floor streetcars. Low-floor vehicles are expected to be on all streetcar routes by early 2020.

Supplementary bus service may be used on some routes during the busiest times.

With the continued delivery of new low-floor streetcars, we are advancing their deployment on more routes.

Currently, the 504 King, 509 Harbourfront, 510 Spadina and 512 St Clair are fully served with low-floor streetcars. We began deploying these streetcars on the 501 Queen in January 2019. We expect that all service on Queen, between Humber Loop and Neville Park Loop will be operated by low-floor streetcars by early summer.

Subsequent routes for streetcar deployment will be: 511 Bathurst (summer 2019), 501 Queen (Long Branch Loop to Humber Loop, fall 2019), 506 Carlton (late 2019), and 505 Dundas (spring 2020). Low-floor streetcar service on Kingston Road will be introduced in 2020 following a review of streetcar services as part of our Five-Year Service Plan. [pp 11-12]

The CEO’s Report now shows the decommissioning plan for all legacy cars in 2019 as “Projected” [p 39].

The 2019-2023 Accessibility Plan includes:

By the end of 2019, the remainder of the order of low-floor streetcars is expected to be received and the TTC plans to retire all high floor streetcars from regular service. [p 27]

The Five-Year Service Plan mentioned above will not be out until December 2019, but with the Capital Investment Plan now showing spending on a further order of streetcars in the mid-2020s, there will be an extended period where expansion of streetcar capacity will be limited to whatever can be provided with supplementary bus service. From King Street, we know that there is a latent demand for better service on the streetcar network, but actually addressing that will be challenging in the current climate.

Crowding is a problem on all parts of the system, but the political focus is on new subway lines that will not address most of these problems, and certainly not in the short-to-medium term. The CEO’s Report now includes a table showing crowding levels, although on a system-wide basis, not for individual routes.

These numbers should be understood in the context of “periods” as defined in TTC schedules. There are five periods through the day:

  • Weekdays: AM Peak / Midday / PM Peak / Early Evening / Late Evening
  • Weekend: Early Morning / Late Morning / Afternoon / Early Evening / Late Evening

The transition points between these periods vary from route to route depending on local demand patterns.

In the chart below, the combination of routes and periods shows that in the first quarter of 2019, 41 bus routes were overcrowded during 82 periods, but this means the combination of one route and one period. With 82 representing only 4.5% of the total, this means that there are over 1,800 possibilities for the bus fleet.

The methodology of counting weekend days individually yields 15 periods overall for most routes. (Some routes do not operate in the Early AM period on the Sunday schedules.) The reason for this is that there is a common schedule for all weekdays, but separate schedules for each of the weekend days. However, this methodology consolidates the majority of the service (weekdays) into only one third of the period count undervaluing the number of riders affected by weekday problems. Moreover, crowding that varies by day-of-week could be masked by averaging over a five-day period.

There also appears to be a mathematical problem for the subway where 7 periods are claimed to be 13.5% of the total. This implies that there are over 50 subway “periods”, but with only 3 lines and 14 periods per line (no early Sunday service), this is impossible (it is unclear where the SRT fits in here). This chart needs work to improve its content.

Reliability of the new Flexity fleet bounced back from a big dip in January 2019, but the mean distance between failures of 13,223 km is still below last year’s performance and less than half of the contracted target. This does not bode well for any move to extend the existing contract with Bombardier.

CLRV reliability continues to track at under 4,000 km MDBF, and the TTC no longer publishes stats for the ALRVs as they have been out of service over the winter. The May schedule plans show a return of five ALRVs to 501 Queen, but this is tentative and the affected runs might simply show up with CLRVs or Flexitys. The CEO’s report notes:

As this legacy fleet is scheduled to be decommissioned by end of this year, maintenance staff will continue to ensure the vehicles are safe to operate in service. However, technical efforts moving forward are being shifted to the new LFLRV fleet and to providing Bombardier with additional assistance. [p 40]

TTC Service Changes Effective May 12, 2019

The May 2019 service changes bring a number of adjustments across the system:

  • Routes that serve post-secondary institutions have reduced service levels reflecting the lighter demand for summer enrollment.
  • The seasonal extension of 121 Fort York – Esplanade to Ontario Place and Cherry Beach begins, and the 175 Bluffers’ Park weekend service returns.
  • Many routes have “service reliability” adjustments which, for the most part, consist of giving more running time and/or recovery time to buses with slightly increased headways.

Construction projects beginning with this schedule period:

  • Davisville Station paving work will see the 14 Glencairn and 28 Bayview South routes interlined. They will not stop in the station. Also, peak period 97B Yonge service will only serve the southbound stop at Davisville. 11 Bayview and 97 Yonge northbound services will continue to use the station but will change loading spots as the work progresses.
  • Jane Station paving work will displace the 26 Dupont and 55 Warren Park services to Old Mill Station. They will serve Jane Station at on-street stops. 35/935 Jane services will offload in the station, but will load on Jane Street northbound.
  • Constuction work at the Wheel-Trans Lakeshore Garage will close the operators’ parking lot for several months. During this time, service to the garage will remain on 83 Jones, but a new 383 Wheel-Trans Shuttle night bus will operate from Queen and Coxwell west to Leslie and south to Commissioners. The eastbound route will use Eastern Avenue from Leslie to Coxwell.
  • Construction work at Eglinton West Station by Metrolinx will close the station during overnight hours. The 363 Ossington will be cut back to Oakwood and Eglinton and will operate as 363B from 2:12 am which will be the last southbound trip from the station.

Service on the Scarborough RT will be improved by extending the peak period service to 11 am in the morning, and to 9 pm in the evening. There is no change in the peak service level of 5’00” headways due to the ongoing reconstruction of the fleet which leaves only 5 trains available for peak service plus 1 spare.

Peak period service on 72 Pape will be modified by decoupling the 72B Union Station branch from the 72C Commissioners branch. Rather than attempting to operate the same headway on each service, the two will run independently of each other with improved service on the 72C branch and reduced service on 72B to Union.

The last of the old “Rocket” services, 186 Wilson Rocket, will be rebranded as 996 Wilson Express with no change in service levels.

The proportion of 501 Queen service between Neville and Humber operated with Flexity low floor cars will continue to increase, especially on weekends. Actual numbers could be higher than those shown in the schedule. In theory, the schedule provides for five ALRVs on the 501 service, but this is subject to availability. Either CLRVs or Flexitys would be substituted.

504 King service will see changes to the schedule during all periods, although this mainly involves adding running and recovery times, as well as some stretched headways.

  • Peak headways stay the same but with longer times through the addition of 3 cars in the AM and 4 cars in the PM.
  • Early evening service sees the greatest change with a move from service every 6’30” on each branch to every 8’00”. If nothing else, this might placate business owners on King Street who complained that service during this period was excessive.

Construction at Roncesvalles Carhouse has progressed to the point where much of the 504A Dundas West to Distillery service will now operate from that location rather than from Leslie Barns.

The growth of the Flexity fleet, combined with remaining “legacy” CLRVs and ALRVs and construction at Roncesvalles is causing problems for overnight car storage. Service on 304 King will be improved from every half hour to every 15 minutes, and similar changes will occur on other overnight routes in coming months. The reconstruction of old facilities moves to Russell Division in 2020, and so this problem is not going away soon. The TTC is also working on a plan to build a yard for 24 cars at Hillcrest as a base for 512 St. Clair, but this is only in the design stage.

2019.05.12_Service_Changes (Version 2, April 20/19 at 5:40 pm)

King-Queen-Roncesvalles Project Deferred to 2020 (Updated)

Updated February 28, 2019: The TTC has confirmed that schedule and routing for 504 King and 501 Queen will not change at the end of March as originally planned. However, the 29 Dufferin schedule had also been changed to send all buses to the Princes’ Gates loop anticipating streetcar congestion at Dufferin Loop. This schedule change will remain for one schedule period and then be backed out in mid-May.

The City of Toronto announced today that the work planned for this summer at the King-Queen-Roncesvalles intersection and westward on The Queensway to Parkside Drive would be deferred to 2020 because of complications with the project.

Email from Chief Engineer, Michael D’Andrea.

“As you know, in early February, the City issued the tender for the planned project at the intersection at King / The Queensway / Queen / Roncesvalles. This project included: sewer and watermain replacement, replacing the entire TTC overhead and track infrastructure within the intersection and west along The Queensway, Streetscaping, road and intersection works along the Queensway, rehabilitating the bridge over Parkside Drive, removing the right-turn channel at Queensway and King Street and overall intersection / road improvements within the area. Based on feedback received to date from contractors considering the tender, there are areas of construction and design that require additional review and clarity to ensure the construction delivery schedule and budget can be upheld and delivered according to plan.

As a result, the City of Toronto is rescheduling the delivery of this project to 2020.

Efforts are underway between several City Divisions and TTC to firm up the design, schedule, and tender and reporting to the Infrastructure & Environment Committee. We expect to provide additional information to all stakeholders involved (Parkdale BIA / Roncesvalles BIA / St. Joseph’s Health Centre) in April – with more details to follow for the residents in the area at a later date.

We understand that the wait and anticipation for this construction has been a long time coming; however, the City and TTC wanted to ensure that the planned construction will be delivered according to the plan, schedule and budget that works to mitigate traffic and TTC service impacts as much as possible. We look forward to meeting with you and stakeholders soon, to further discuss these measures.”

I await clarification from the TTC whether the route changes contemplated to go into effect with the March 30 schedules will or will not take place. They are probably far enough away from finalization of those schedules to avoid having an inappropriate service design for diversions that are not required now.

When I hear definitively, I will update this article.

Meanwhile, the expected release of streetcars from the west ends of 501 Queen and 504 King will not occur, and this will prevent the change back to streetcar operation on other routes until more of the new cars are available.

504 King vs 501 Queen Speed Comparisons (Part II)

This article continues the analysis of transit vehicle speeds on King and Queen Streets downtown over the past two years. The first installment of the article compared travel times between the vehicles on each street at specific times of the day and periods over the course of two years. Here, the same data are arranged to show the evolution of travel times on each street over time both before and after the implementation of the King Street Pilot.

King Street

Westbound

Here is a sample chart showing King Street westbound in the hour from 8 to 9 am.

Three periods in 2017 (January, September and early November) are plotted in “warm” colours (pink, red, orange), while periods in 2018 with the pilot in operation (January, July and October) are plotted in “cool” colours (green, blue and purple). This makes it easy to distinguish the groups of data that belong to the “before” and “after” periods.

As this is westbound data, the chart is read from left to right. A common pattern that shows up here is the different location of low speeds corresponding to stops once the pilot is active with “before” data dipping ahead of the intersection, and “after” data dipping following. The degree to which “after” data also includes a nearside dip indicates how traffic signals can compound the stop service time with farside stops. Note especially the green line which shows data from before the re-activation of Transit Signal Priority (TSP). At Jarvis, for example, there is a decided reduction in nearside delay comparing the blue (July 2018) and purple (October) lines with the green (January) one. There is also an improvement at York Street and at University Avenue.

Continue reading

504 King vs 501 Queen Speed Comparisons (Updated)

Updated December 2, 2018 at 8:00 am: A sixth set of charts has been added covering the last two full weeks of September 2017. The intent is to show fall conditions before the pilot, but also before construction on King (at least some of which was to set up the pilot itself) slowed King cars in the early November 2017 data already published here.

With a year’s worth of the King Street Pilot now behind us, attention turns to two basic questions: should King Street remain as it is with a degree of priority for transit, cyclists and pedestrians; and should this scheme be extended to other major streets, notably Queen Street.

In previous articles I have reviewed the behaviour of the King and Queen routes:

This article provides a detailed look at average weekday speeds along the streets during six separate periods, and with hourly breakdowns from 6 am to 1 am.

  • January 16-27, 2017: Winter operations downtown deal with less traffic, especially on Queen Street. This is a “before” snapshot of the two streets.
  • September 18-29, 2017: Fall conditions after TIFF and before the beginning of construction at Queen & McCaul.
  • November 1-10, 2017: Just before the implementation of the pilot.
  • January 15-26, 2018: Winter operations with the pilot in place.
  • July 16-27, 2018: Summer conditions for the pilot. (2017 was a construction year on Queen and a direct comparison to 2018 is not available.)
  • October 15-26, 2018: Current fall conditions.

This gives a view of “normal” conditions, but does not capture all of the seasonal and special event variations, many of which are evident in the data in previous articles. The periods have been chosen to avoid skewing the numbers with special events such as TIFF, vacation periods and construction.

A Caveat About “Average Speed”

The values show in these charts are derived from TTC vehicle tracking data. This originates in GPS format, and after cleaning up the “rogue” points that are off route due to GPS errors, what is left is mapped onto a version of the route broken into 10m segments. The speed of a vehicle at each segment is determined by the “before” and “after” locations of adjacent GPS observations and the time difference (usually 20 seconds) between them.

These values give the average speed within each segment of all vehicles whose GPS data placed them there. This is not the same as the average speed over the entire pilot area on any given trip. What the numbers show is the locations where vehicles tend to be making good time between stops, the places where they are bogged down, and the places where they stop. Even at stops, the average is never zero because it includes observations where vehicles are just stopping or just starting up within a segment, not to mention variations in stopping location and cases where vehicles do not stop at all.

The Effect on Riders

Vehicle speeds on Queen are often comparable to those on King, but Queen is notorious for less frequent, unreliable and crowded service. Riders are very sensitive to waiting times at stops, and this is compounded if they cannot board the first vehicle to appear, let alone if they walk some distance without a streetcar ever passing them.

With the King Street Pilot there has been a reduction in the variability in travel times which has led to less irregular service. Scheduled service comes more often and the line’s capacity has been improved with new streetcars. Speed alone is not the only measure of a service improvement.

It is vital that debates about the effectiveness of changes to King are debated not just on speed but also on reliability and capacity for demand. The early focus on speed changes which were, in some cases, unimpressive, opened the field to critics who argued that there was a great upheaval for little benefit. It is the combination of more reliable travel time, increased speed, reliability and capacity that has affected the riding experience and drawn more people to use this route. Any move to expand either the scope on King or to another street must take all of these factors into account.

Reading the Charts

Each chart contains data for one hour’s operation in one direction. Here is a sample from the PM peak westbound in October 2018:

The blue line shows data for King while the green line shows Queen data. Trend lines are interpolated through the data for King (red) and Queen (yellow).

The limits of the chart are east to Parliament and west to Shaw, both beyond the Jarvis-Bathurst pilot, for a few reasons:

  • The wider scope shows conditions on the approaches to the pilot area.
  • The nature of the trend line calculation in Excel can cause misleading behaviour in the lines as Excel tries to project beyond the charts using values near the edges. Having a “buffer” area on either side of the pilot ensures that the trend lines do not include this problem through the Jarvis-Bathurst section.

Queen and King run parallel to each other with most north-south streets at the same location on both. This allows data from the two routes to be lined up on a single chart. A few notes about special cases:

  • Sherbourne Street is used as the reference point for aligning the data. The distance from Sherbourne to Bathurst is the same on each street.
    • Ontario Street is slightly offset on Queen relative to King.
    • Parliament Street is slightly further east of Sherbourne on King (measured as travel distance) than on Queen due to the shift in King’s direction just west of Parliament at Berkeley.
  • Victoria Street has a small jog at Adelaide. The line on the charts reflects its location on Queen which has a transit stop and traffic signal.
  • The stop and traffic signal at Queen and Augusta have no equivalent on King.
  • Niagara Street curves eastward between Queen and King, and so the crossing points are different on the two streets.

Westbound charts as above are read from left to right. Eastbound charts are in the same format and should be read from right to left.

Comments on the details of these charts are included later in the article.

Continue reading

501 Queen and the Effect of the King Street Pilot (Updated)

Updated November 27, 2018: A major section has been added comparing travel times between Jarvis and Bathurst on Queen and King Streets.

With the King Street Pilot a year old, one area that also deserves a review is the effect it has had on transit service on Queen Street. The short answer is “very little” with the usual caveats about seasonal variations and unusual events that complicate direct before and after comparisons.

The charts presented here cover the period from March 2016 to October 2018, the same period as King Street charts in recent articles. Selected images are included in the post, and more extensive sets are linked in PDFs.

(Note that the level of detail in the three-year charts is such that they suffer when reduced in size to the scale of a blog post. Click on any of them to open a full size version in a new browser tab.)

Major Events on Queen Street

More so that on King Street, Queen has been affected by many construction activities over the three years, and of course it also has more special events disrupting service in the area parallel to the pilot on King. The major events affecting travel times between Jarvis and Bathurst Streets are highlighted in these charts.

  • April 2016: Reconstruction of King and Charlotte diverted service onto Queen adding to streetcar running times there primarily through delays where King cars turned off of Queen without assistance from traffic signals.
  • May to December 2016: Water main reconstruction on Queen west of Spadina triggered a long-running diversion of service via Spadina to King. The times shown are measured between King and Bathurst and Queen and Jarvis for this period where vehicles were diverted.
  • September 2016: Although TIFF closed part of King Street, the service was split rather than diverted. There is no effect on Queen running times beyond that already seen for the diversion of the 501 route.
  • May to August 2017: Construction continued for another summer on Queen including completion of work started in 2016 and replacement of the pedestrian bridge at Queen and Yonge. Buses operated on 501 Queen.
  • September 2017: Streetcars returned to Queen. Reconstruction of Dundas from Church to Yonge caused some traffic to shift south to Queen. King services diverted onto Queen during the opening weekend of TIFF.
  • October 2017: Reconstruction of Queen & McCaul caused Queen service to divert via King between Church and Spadina.
  • November 2017: King Street Pilot begins.
  • June 2018: Construction delays.
  • September 2018: TIFF diversions delay Queen service.

Chart Format

As with the King Street analysis, the charts here show the 85th percentile values for travel times in orange. This includes most trips across the area parallel to the pilot on King. The median value (50th percentile) is shown in blue. Half of the trips took longer and half shorter than this value. When these values track close together, most trips will lie in a fairly narrow band of values (see details for October 2018 later in this article). Where the lines pull apart, travel times seen by riders will have more variation in length.

Where the route was on diversion in 2016, the time shown is measured between Bathurst & King to Jarvis & Queen via Spadina.

For the diversion in 2017, the time is measured between Bathurst and Jarvis on Queen, but includes the diversion both ways via Spadina, King and Church.

Where a value drops to zero, this indicates that no car was observed between the two points usually due to a short-term diversion around a major blockage.

AM Peak 8 to 9 am

Westbound

The beginning of the pilot in November 2017 had no effect on AM peak travel times. The spike in late June 2018 was caused by construction delays according to TTC eAlerts.

Eastbound

Continue reading

King Street Update: August 2018

This article continues the series reviewing the operation of transit service on the King Street transit priority pilot. August brings two major events that affect service on King Street, although the traffic problems are concentrated at the western part of the line: the Caribbean Carnival and parade on Saturday, August 4th, and the CNE from mid-August onward, especially with the air show in the final days of the month. Both of these bring congestion through Parkdale notably at the approaches to The Queensway and to Jameson Avenue westbound.

Peak Travel Times

As usual, we begin with the PM peak travel time chart westbound from Jarvis to Bathurst. The 85th percentile line has higher spikes in August, and the three largest relate to specific events:

  • Wednesday, August 8: A delay near Church Street held a few cars causing a jump in the 85th percentile value although the change to the 50th (median) percentile was much lower. The cause of the delay is unknown because the TTC did not issue a service alert.
  • Tuesday, August 21: Severe congestion westbound to Spadina from about 5:40 pm onward drove up both the 85th and 50th percentile values. Again, there was no TTC alert indicating a problem.
  • Friday, August 31: “Police activity”, as the alert put it, required diversion of streetcars in both directions due to an incident west of Yonge Street. The spike in the 85th percentile was caused by one car that crossed Jarvis at about 5:30 pm but was not diverted. As a result its trip, including the delay, was included in the 5-6pm data for the pilot district. As with the August 8 data, note that the change in the 50th percentile is small and on a par with typical day-to-day variations.

For comparison, here is the eastbound chart.

Here are the full sets of charts:

Continue reading