During the last week of November, the first to see streetcars return to The Beach after almost three months’ absence for construction at Kingston Road, the line had a major disruption thanks to a broken rail near Roncesvalles. This rail damaged the track brakes on 22 Flexity cars, and while the TTC searched for the problem, the line was completely switched to bus operation.
Streetcars ran on Monday and Tuesday, November 25-26, and on Wednesday November 27 until midday. Buses ran from the afternoon of November 27 to the end of service on Friday,November 29 (actually Saturday morning).
There have been calls from certain quarters on City Council for a comparison of the operation of both modes. I published an analysis of route 505 Dundas in May 2018. Broadly speaking, it showed that buses outrun streetcars only when there is no traffic in the way and operators can drive as if they are on the suburban streets they are used to.
The substitution on 501 Queen gave an opportunity to compare the two modes over the entire route, not just over a segment running with buses due to construction. This article reviews the data from November 25-29 for 501 Queen.
The TTC has two vehicle tracking system, CIS and VISION. The streetcar fleet (and a small number of buses that often run on streetcar lines) is tracked by CIS, while most of the bus fleet is tracked by VISION. The entire system will be on VISION probably in a few years, but for the moment there are two data sources.
Until quite recently, I was unable to obtain finely-grain information about vehicle locations from VISION, but this changed in October 2019. It is now possible to get comparable data tracking vehicles from both systems. This meant that comparable data were available for both the streetcar and bus operations on 501 Queen.
The process for converting data from snapshots with GPS co-ordinates to a format suitable for analysis is described at length in Methodology For Analysis of TTC’s Vehicle Tracking Data.
In this case, we are interested in three aspects of streetcar and bus behaviour:
- How long does each type of vehicle take to get from one point on the route to another?
- What are the speed profiles for each vehicle type along the route?
- What are the dwell times for each vehicle type along the route?
For the comparatively coarse measurement of travel times between points, the route is divided by screenlines. Tracking when each vehicle crosses a screenline gives both the headways at each line, and the travel times between them.
For fine measurement of vehicle speed, the tracking data are used to calculate each vehicle’s speed as it moved along a route. The route is subdivided into 10m segments, and the speeds of every vehicle passing through that segment in each hour are averaged. This reveals locations where vehicles spend a lot of time stopped or travelling slowly, and of course locations where they move much faster.
For dwell times, the points of interest are those where vehicles are stationary. The “tick” of the clock for tracking data is every 20 seconds, and so the length of a vehicle’s stay at a point can only be calculated to a multiple of that interval. This is a fairly coarse measurement relative to the length of time most vehicles take to serve stops, and the resulting data give only a broad outline of comparative dwell times. Note also that “dwell time” is not necessarily all “stop service time” because vehicles can be awaiting a green traffic signal, or be stuck in traffic at the stop.
The distance scale to which I convert GPS positions is measured in 10m increments. Given that vehicles will not necessarily stop at exactly the same place every time, the charts here give moving averages of dwell times over 30m.
All of the analyses presented here are subdivided into hourly intervals recognizing that a route’s behaviour at 6am is vary different from midday, the two peaks, and the evening. Far too much data presented by the TTC is summarized on an all-day basis, and even on an all-route basis. This masks variations in behaviour by location and time of day, and does not give a detailed picture of what is happening.
The data reveal various aspects of bus and streetcar operation on 501 Queen, and by extension, on other routes where a substitution might be contemplated. The results for 501 echo those seen in the 2018 article on the 505 Dundas route.
- Across the entire route, buses travel faster than streetcars, but their performance varies from place to place, hour to hour.
- On sections of the route where traffic is not free flowing, and where stops are busy, buses do no better than streetcars and during some periods they are worse.
- Where traffic is free flowing, some of the advantage buses have arises from driving at above the speed limit which is 40 kph within the old City of Toronto, and 50 kph on the Lake Shore section west of the Humber River.
- The effect of streetcar slow orders at numerous locations is clearly evident in the data.
- Dwell time for buses appears to be slightly longer than for streetcars. This could be due to loading delays, but in turn that could be caused by the bus service being overwhelmed by streetcar-level demand. (There were complaints about the quality and capacity of the replacement service.) Also, buses lose time getting to and from curbside stops, but this is not necessarily reflected in “dwell times” because they are merely slow, not stopped during these moves.
- I am unable to comment on service quality with buses because many vehicles were not logged on to VISION with the 501 route number. Therefore, their data do not appear in the extract I received. However, there were enough vehicles to get a sample of their behaviour and determine travel times.
During the period of streetcar operation, route 501 operated in its standard weekday configuration with 501A service between Neville and Humber until about 9pm, and a separate 501L service from Humber to Long Branch. In the late evening, all vehicles ran end-to-end over the route from Neville to Long Branch.
During the bus operation, buses ran over the full route from from Neville to Long Branch diverting south to Lake Shore Boulevard at Windermere Avenue because buses cannot run through Humber Loop. Operating speeds over the diversion are not directly comparable between the modes because buses did not serve Humber Loop.
Travel Times Between Screenlines
The two sets of charts linked below compare travel times for the two modes within each segment. There are also charts for the major portions of the route (e.g. Neville to Yonge). Here is a sample from them.
- Streetcar and bus data are shown in red and blue respectively.
- The solid lines show the average values, and the dotted lines show the standard deviations in the values.
- Where the difference between travel times is small and within the value of the standard deviation, the scatter in data values will produce overlap between the ranges where most of the values lie for both modes.
The descriptions for each segment should be read in conjunction with the charts linked here.
Silver Birch to Woodbine
Note that the screenline for the east end of the route is set one stop west of Neville Park at Silver Birch to avoid problems with queues of vehicles laying over near the terminal before they actually cross the screenline.
In general, the travel times in both directions for this segment are similar with a small advantage for buses.
The westbound chart illustrates a problem with this type of analysis: the presence of major events that can skew the results.
Here is the chart with all of the data left in the analysis. The large spikes at midday were produced by a major delay for a medical emergency that blocked all transit vehicles just east of Woodbine for close to one hour on November 27. This occurred during the transition from streetcar to bus operation, and so both modes were affected.
When the data points for the delayed vehicles are removed, what remains are the trips that encountered normal conditions.
Throughout these charts, I have left in all of the data including for delays. One could filter out the outliers, but then the question would be whether this artificially trims out events that would be “normal” in day to day operations. If there were more data, for example from an extended substitution, the effect of one major delay on the overall numbers would be smaller.
Woodbine to East of Coxwell
Buses are slightly faster over this segment in both directions. One issue here for streetcars is their relatively slow operation through the Kingston Road intersection.
East of Coxwell to Greenwood
Buses are faster over this segment in both directions. Two issues affecting streetcars are slow operations through the special trackwork at Coxwell and at Russell Carhouse, as well as delays for crew changes at the carhouse.
Greenwood to West of Broadview
This is a segment where buses were notably faster than streetcars although by varying amounts through the day. Driving speeds for buses tended to be high here because of light traffic on this section of Queen Street. Streetcars crossed two intersections with slow orders due to special trackwork at Leslie and at Broadview.
West of Broadview to Parliament
Buses ran faster in this segment in part due to relatively light traffic, and partly because of the long-standing slow order on the Don Bridge for streetcars due to problems with expansion joints and the facing switch westbound at King. The expansion joint issue was resolved some time ago, but many streetcars still operate over the bridge at a slow pace.
At Parliament, there is a streetcar intersection and its standing slow order. Because the screenline is in the middle of the intersection, the effect of the slow order is allocated partly to the segment east of Parliament, and partly to the west.
Parliament to Jarvis
From Parliament westward, the difference between bus and streetcar travel times narrows. Note that as this gap narrows, the difference lies within the margin of the standard deviation and so the scatter of values will produce overlap between the ranges of typical travel times for both modes.
Jarvis to Yonge
Streetcars have a slight edge in travel times in the afternoon, while buses have a slight edge at other times. Again, the difference is smaller than the standard deviation. This segment includes two streetcar intersections at Church and at Victoria.
Silver Birch to Yonge
This chart gives the travel times over the eastern portion of the route. The cumulative effect of the shorter bus times over each segment is visible in this chart, although it is lower in the afternoon than other times of the day.
The westbound chart includes a gap for the period from 2-3pm. For some reason, the VISION data for this period is not as detailed as at other times of the day. During that hour, no buses were tracked between Silver Birch and Yonge on all three days of the substitution.
Yonge to University
Streetcars generally make better times over this segment in both directions than buses illustrating the problems that buses have in congested areas with busy stops. There is a streetcar intersection at York, but its effect is minor as there are only trailing switches that do not have the same severity of standing slow order. Moreover this is a congested area where traffic speed tends to require slower operation even without the special trackwork.
University to John
Streetcar and bus travel times are comparable in this segment which includes a streetcar intersection at McCaul.
John to Spadina
Streetcars make better time eastbound over this segment than buses, while for westbound trips times are comparable for the two modes except in the late afternoon when streetcars have the advantage.
There is a streetcar intersection at Spadina, and with the screenline in the middle of the intersection, the travel time effect is allocated partly to each segment east and west of that line.
Spadina to Bathurst
Travel times in this segment are similar for both modes. At Bathurst, there is a streetcar intersection, but its effect is mainly on westbound trips because there is no eastbound facing point switch.
Bathurst to Ossington
Buses have the advantage over this segment until the early afternoon. There is a streetcar intersection at Shaw.
Ossington to Dufferin
Buses have an advantage westbound in the morning, but otherwise travel times are comparable for the two modes. There is a streetcar intersection at Dufferin with a westbound facing switch.
Dufferin to Lansdowne
Travel times for the two modes are similar all day long in this segment.
Lansdowne to Roncesvalles
Travel times for streetcars are longer for most of the day westbound. This is partly affected by crew changes at Roncesvalles which count against westbound trips because they occur before streetcars cross the screenline in the intersection. Eastbound trip times are comparable for the two modes.
Yonge to Roncesvalles
Westbound trips give the advantage to buses until early afternoon when traffic conditions affect their travel speed. Eastbound, streetcars have the advantage for much of the day.
Roncesvalles to Humber East
The Humber East screenline is located west of the Humber River bridge, but east of Humber Loop.
Bus travel over this segment ran in the traffic lanes, not in the streetcar right-of-way and had the extra mileage of diverting south on Windermere. However, buses were not delayed by slow operation through the special trackwork at Roncesvalles, at the carhouse and at Sunnyside Loop, all of which is in poor condition thanks to a delayed project to rebuild in this area (now planned for 2021).
Streetcars had the delay associated with track at Roncesvalles, as well as the stop at South Kingsway which was not served by buses.
Buses were affected by delays making the left turn at Windermere and this causes the sharp increase in average bus travel times during the PM peak. Note that although the bus times more than double, the standard deviation does not change much indicating that this is a chronic condition affecting most trips, not a spike that creates a wider scatter of values. The detailed data for this segment are in the chart below.
Here are the data for streetcar operation showing much more consistent travel times.
I have not included any data for operations at Humber Loop because the streetcar and bus operations were completely different in this area.
Humber West to Royal York
The Humber West screenline is located west of the exit from Humber Loop onto Lake Shore.
Buses were generally faster than streetcars in this segment as they could operate with little hindrance from traffic on the wider Lake Shore Boulevard.
Royal York to Kipling / Kipling to Long Branch
Buses were generally faster over these two segments although less so west of Kipling.
Humber West to Long Branch
Over the portion of the route west of the Humber River, buses had lower average times almost all day except for the afternoon peak.
Travel Speeds Along the Route
501 Queen is a very long route, and the charts are too crowded with data if the entire route is consolidated on one page. The following sections deal with the eastern and western segments of the main route where it operates on a classic 4-lane city street, and then the segment west from Roncesvalles to Long Branch which has a mixture of street running and reserved right-of-way in a more suburban environment.
Here is a sample page from one of the charts showing travel speeds westbound on the eastern portion of the route during the hour from 5 to 6pm. The pattern of stop locations is quite clear because these are the points where the speeds drop down. The solid lines show the average speeds for each mode, and the dotted lines are trends interpolated through these values to show the overall behaviour. East of Yonge, the bus line (dotted yellow) is generally above the streetcar line (dotted red). The degree to which they are separated varies by time of day and location.
Here is the chart for the same area for the hour from 9-10 pm. Note how often the bus speeds peak well above 40 km/h which is the speed limit for this area. Streetcar operators would be disciplined for this type of operation, but bus drivers on replacement services appear to have free rein. It is easy for buses to “outperform” streetcars when they drive as fast as possible regardless of posted limits.
I leave it to readers to peruse these charts which show a great deal of detail about the operation of the two modes on various parts of the Queen route.
Neville to Yonge
Yonge to Roncesvalles
Roncesvalles to Long Branch
The question here is whether buses tend to spend longer at stops than streetcars. The charts linked below show the westbound comparison for each segment of the route. The values shown are the degree by which average occupancy in a location exceeds 20 seconds, the minimum possible with a 20-second polling interval for vehicle location. (A vehicle might occupy a location for less than 20 seconds, but it was “seen” there once during that interval.) As mentioned earlier, the resolution of this calculation is limited by the polling cycle of the tracking systems.
I have generated only the westbound charts for the three sections of the route as these give the general idea of how the numbers behave. There are some locations where streetcars sit for longer than buses, and others where buses take longer. The big spikes correspond to locations where there are external delays such as crew changes (which vary by mode), whereas the small ones are probably representative of the time needed at stops.
Here is a sample page showing data for the hour from 5-6pm westbound from Yonge to Roncesvalles.
The blue and green lines show the moving averages for dwell times for streetcars and buses respectively, with trend lines in red and yellow.
Buses (green) usually have higher values, but this varies by location and time of day. From the speed charts, we know that buses are driven faster than streetcars when possible, and the longer dwell times are more than made up for by higher speed between stops.