On January 3, 2016, the schedule for route 510 Spadina changed from one based on the use of standard-sized streetcars (the CLRVs) to the new longer low-floor cars (Flexitys or LFLRVs). This article reviews the operation of the line before and after the new schedule.

The revised service provided an increase in capacity with a replacement ratio of new cars for old on a ratio considerably lower than 2:1, and during periods of infrequent service (early weekend mornings, late evenings), the replacement was 1:1.

In brief, the service actually operated on 510 Spadina bears little resemblance to the advertised schedule. Headways (the time between cars) are erratic and often wider than the scheduled values for both the “old” and “new” service designs. The proportion of service operated to the three destinations (King, Queens Quay and Union Station) do not always match the schedule, and indeed during January (when only two destinations are supposed to be in use at any time), service to all three persists just as in December.

However, padding the schedule, a tactic used on other routes to eliminate the need for short turns, is impractical here because terminal congestion with queued vehicles would severely interfere both with passenger travel times and with terminal operations. Unlike a route such as 501 Queen where most riders depart from vehicles before the terminal, 510 Spadina has very strong demand to its destinations at subway stations.

Line management consists of dispatching cars and operators as available, and the service levels on each branch do not match the actual design, notably the proportion of service that should operate on each section of the route. This particularly affects riders south of King and on Queens Quay whose service is erratic and below the advertised level even though the area has a large and growing population.

Notes:

• Although service emerges from the station loop at Sussex Avenue, the measurements cited here are at Harbord Street. This is necessary because the GPS on some streetcars becomes “confused” about the location while underground and do no begin transmitting valid positions until they are south of Sussex.  A similar problem with the Bay Street tunnel requires headway measurements at York rather than right at the tunnel mouth.
• An explanation of the format of all charts can be found at the end of this article.

Summary of Headways

The following charts consolidate headway statistics from six points along the route for each direction on weekdays, Saturdays and Sundays. On each chart, the service to Union (measured at York Street) is shown in red (with the standard deviation in a dotted red line). Service south of King is shown both at King and at Lake Shore in green and yellow respectively. Service between Richmond and Spadina Station (measured at Harbord) is shown in blue, violet and brown.

This clearly shows the level of service on the three branches of the route. What is immediately obvious here is that there is a considerable amount of service short-turning at King even during periods when there is no service scheduled to terminate there on weekends in December. In January, three services are visible in the averages even though only two are scheduled at all times.

The line is, in effect, being run without a schedule and what the TTC advertises simply bears no resemblance to the service actually operated. This is of particular concern south of King where riders are short-changed relative to the service they are supposed to receive.

In March 2016, the Charlotte Street Loop closed for construction for several weeks, and there was no option but to send all service further south. I will report on how the line actually operated during this period, as well as the following weeks when service was still allegedly to all operate at least to Queens Quay, in a future article.

Note: January 1 and 2 are not included below as they operated with the “old” schedules.

510_201512_MonthHeadways_TPSummary

510_201601_MonthHeadways_TPSummary

Headways Southbound from Spadina Station

510_201512_SB_Harbord_MonthHeadways

510_201601_SB_Harbord_MonthHeadways

For new readers of these analyses, a description of these charts appears at the end of the article.

Comparing the averages with the scheduled values shows that actual service is commonly not as good as the scheduled service. Cars are consistently not leaving the terminal as often as the planned headway. This problem did not disappear with the implementation of wider headways for service with the new streetcars on weekdays, but did improve on Saturdays.

• December weekdays AM peak through early evening: The scheduled headway is under three minutes, and for a good part of the day it is 2’13”. However, the actual headway crossing Harbord averages about three minutes.
• January weekdays: The scheduled headway is 3’20” for much of the day, but the average actually operated was over four minutes.
• December Saturdays: The scheduled headway is 2’20”, but the actual average exceeds three minutes.
• January Saturdays: The average headways are closer to, but still above, scheduled values than in December.

This is not a case of some days with less than scheduled service offsetting other and diluting the averages. The trend lines for headways on all weekdays lie in almost the same place on the weekday charts showing that the service was, generally speaking, similar through the months of December and January.

This begs the question of whether the line is being operated on a timetable at all, or simply being dispatched as cars are available, or at the discretion of route supervisors at the station. There is also a question of whether it is physically possible to push cars through the station at very close headways given the time needed to unload and reload. Anyone who rides the line knows that long waits to enter the station are common as cars are queued up behind those serving the platform.

Another point to note in January is that the weekday standard deviation is higher, and headways well above 10 minutes are common. For a route entirely on its own right-of-way and so close to its own terminal, this is a troubling situation. Even under best case conditions, the TTC maintains neither its advertised level of service, nor a regular vehicle spacing.

Headways Northbound at Richmond Street

The situation northbound at Richmond is similar to departures from Spadina Station. The average headway is wider than that advertised on the schedule, and the range of headways is wider in January than in December. Weekends are particularly bad. Headways above 10 minutes are not uncommon.

King Street is a logical point from which to space and dispatch service, but the actual headways at Richmond suggest that the merging service from the short turns at Charlotte Loop and the through service to Union was not working well. The standard deviation in values rose in January for weekday service indicating that service was less reliably spaced with fewer cars on the line than it had been in December.

Another troubling point is that SD values grow during periods when scheduled headways are wider (weekend evenings). This begs the question of whether laissez-faire management when scheduled headways are short continues into periods when wider headways demand better on time performance (or at least regular vehicle spacing).

510_201512_NB_Richmond_MonthHeadways

510_201601_NB_Richmond_MonthHeadways

Headways South of King Street

In December, weekday service is split from the midday through to the early evening between Union Station, Queens Quay Loop, and Charlotte Loop (King Street). In theory, two out of three cars should continue south of King, and one in three should run through to Union. In January, the weekday service was changed so that half of the cars turned back at King and the other half operated to Union with no scheduled short turns at Queens Quay.

510_201512_SB_SouthofKing_MonthHeadways

510_201601_SB_SouthofKing_MonthHeadways

Service south of King is very erratic with headways over 10 minutes quite common during all operating periods. The standard deviation values are routinely at four minutes or more indicating that many cars are running close to each other separated by wide gaps. The situation is similar for both months.

510_201512_NB_SouthofKing_MonthHeadways

510_201601_NB_SouthofKing_MonthHeadways

Northbound service is even worse than southbound with the SD values closely tracking the averages. This means that pairs of cars are quite common. Headways from 10 to 20 minutes are common in a location where service should be much more frequent. During periods when half of the service is scheduled to Queens Quay Loop and the other half to Union, the pairing of cars shows not only that nobody is managing headways, but that the through and turnback cars are leaving close together northbound. This is the worst type of “service” where branches of a route supposedly blend together.

Headways on Queens Quay

Service to Union Station is even worse than on Spadina. In January, many weekday headways lie in the 10-20 minute range with a few beyond that level. Standard deviation values lie in the 5-6 minute range at which it is impossible for the TTC to achieve any of its stated targets for service reliability.

510_201512_EB_York_MonthHeadways

510_201601_EB_York_MonthHeadways

Service westbound from Union shows the same pattern as eastbound. With two routes, Spadina and Harbourfront, sharing the loop, it is impractical for cars to space themselves out properly, and so the outbound headway pattern mimics the inbound values.

510_201512_WB_York_MonthHeadways

510_201601_WB_York_MonthHeadways

Terminal Time at Union Station

Round trip times from York Street to Union Station and return are consistent through December and January with values around 8 minutes and a standard deviation usually under 2 minutes. Because 510 Spadina shares the loop with 509 Harbourfront, extended layovers at Union are not practical, and this leads to fairly uniform times over this segment.

510_201512_UnionStation_Terminal_MonthLinks

510_201601_UnionStation_Terminal_MonthLinks

Terminal Time at Spadina Station

Times shown in these charts are measured from Sussex Avenue through the station even though some trips are missing due to GPS problems described above. This minimizes any effect of the travel between Sussex and Harbord including stop dwell times.

Unlike Union Station, Spadina Station Loop has a spare track where cars can lay over. The result can be round trip times to and from the loop that are quite long depending on how much time a car spends in the loop. There is a wide scatter of round trip values for this segment, and these throw off both the averages and standard deviations.  Many cars make the trip in 5 to 8 minutes, but times up to half an hour are not unusual.

There is some difference between December and January data in that the practice of holding cars at the station seems to apply to all-day operations in December, but more so to afternoons and evenings in January.

This is another example of how the actual operation of the line is completely different from the official schedule.

510_201512_SpadinaStation_Terminal_MonthLinks

510_201601_SpadinaStation_Terminal_MonthLinks

Round Trip Time between Harbord and Union Station

The reason for operating the route so much differently from the posted schedule becomes obvious when we look at trip times over the line. The charts below give the round trip times from Harbord to Union Station and return. From the terminal times above, we know that Union Station Loop does not contribute much variability through layovers, a situation completely different from that at Spadina Station Loop.

For both December and January, this round trip averages about one hour through weekday daytime periods until after the PM peak, and similar times for midday and afternoon operations on weekends.

At least 10 minutes is needed for a layover-free trip from Harbord to Spadina Station and return making the full round trip a 70 minute proposition. With a standard deviation typically running at 5 minutes, the majority of the round trip times will lie within a band from 65 to 75 minutes. This is greater than the scheduled time for round trips over the entire route (56 to 62 minutes), and it is impossible for cars to remain on time.

510_201512_RT_HarbordUnionStn_MonthLinks

510_201601_RT_HarbordUnionStn_MonthLinks

Average Speeds

The average speed of vehicles along a route seen at a fine granularity can reveal locations where cars are routinely held by congestion, traffic signals and stop service time. In the case of 510 Spadina, congestion as we usually know it does not exist because the route does not operate in mixed traffic. The two sets of charts below illustrate average vehicle speeds by hour for the weekdays in the middle of January 2016. Stepping through the pages provides a “flip chart animation” of the evolution of these values from 6:00 am until after midnight.

In both charts, Union Station is at the left and Spadina Station is at the right. The vertical scale is in km/hr. Where there is a “notch” in the chart, this will be a location where vehicles move slowly or stop, and these will usually be at stops or intersections.

510_201601_SB_SpeedStats

Reading the southbound stats, think of the vehicles as moving from right to left across the page. Among the effects that can be seen here are:

• The degree to which cars are held southbound approaching College, Dundas and Queen. All of these have farside stops, but cars slow on the approach to the intersection (often stopping as regular riders know), then the average speed goes up as they cross, and then it drops again at the stop. There is no transit priority at these locations.
• The section from Richmond to King is particularly slow because of the combined effect of the signal (and some streetcar turns) at Adelaide and the nearside stop at King.
• Operation between Spadina/Lake Shore and the entrance to Queens Quay Loop is consistently slow both because of the traffic signals and the eastbound stop.
• Average speeds on Queens Quay are slightly higher than on Spadina probably because stops are further apart, and there is some transit priority in the traffic signals.

510_201601_NB_SpeedStats

Reading the northbound stats, think of the vehicles as moving from left to right across the page.

• As with the southbound service, operation on Queens Quay is a bit faster than on Spadina.
• The Spadina/Queens Quay area is a location of slow operation northbound as it is southbound.
• Cars operate slowly on average northbound to Bremner where they are often held by the traffic signal and then cross to serve the northbound farside stop. The average speed north of Bremner is at times higher than the speed to the south showing how the traffic signal works against the transit service. A similar problem exists northbound at Front.
• King, Queen, Dundas and College all show the effect of cars held nearside for traffic signals and then again for stop service farside.

Reading the Charts

Many regular readers will be familiar with the charts used here, and so I have placed this explanation at the end. Each set contains the following pages:

• Four or five pages plotting the values for weekdays that are not statutory holidays. Each dot represents one value (a headway or a trip (or “link”) time), with a separate colour for each day. The trend line is interpolated for each day’s data and to the degree that these lines lie on top of each other, this shows the similarity (or not) in day-by-day operations.
• One page consolidating all of the weekday data points. The purpose of this is to show the overall shape as a “cloud” and in particular the degree of scatter in values.
• One page each for the Saturday and Sunday/Holiday data in the same format as the weekday charts.
• One page for each of weekday, Saturday and Sunday/Holiday data giving monthly averages and standard deviation values calculated on an hourly basis. Note that weekend values can be more “spiky” because there are fewer observations within each hour, and so one delay can have a bigger effect than with the weekday values.

To the degree that values are tightly clustered around the trend lines, the values don’t vary too much, and most observations will fall close to the average. This shows up also in low values for the standard deviations. Conversely, when the data points are spread out, this shows that headways are erratic (long and short values being as common as average ones), and the SD value can rise close to the average. In that situation, service is generally operating as pairs of vehicles (or worse) separated by gaps in the case of headways. For link times, a low SD value and tightly clustered values show that the travel time is consistent, or not.

The ideal in the case of both headways and link times is that SD values be low and observed values stay close to the average.

For a full description of the process by which I have analyzed the TTC’s vehicle tracking data, please see Methodology For Analysis of TTC’s Vehicle Tracking Data.