With Toronto’s Congestion Management Plan in the news, we hear about various ways to improve the movement of cars and, oh, by the way, maybe speed transit vehicles too. Toronto talks a lot about “transit priority” but this more often means giving transit vehicles the leftovers, the spare time that can be shaved at some locations from competing traffic, not absolute priority for transit vehicles.

This shows up particularly where transit vehicles have a dedicated lane, but they must wait patiently while other traffic, typically left turns, gets out of the way. This has the perverse effect of making transit slower particularly when combined with the “double stop” effect of farside stops. At these locations, the nearside space a stop would normally use is occupied by a left turn lane, and traffic in that lane has first claim on a green cycle.

Two routes with this as a near-standard stop configuration are 510 Spadina and 512 St. Clair. The Spadina car only recently returned, and I will leave comparative analysis of bus and streetcar operation there until more data have accumulated. A preliminary look is not complimentary to the streetcars because, unlike them, the buses are able to combine waiting for green signals with loading passengers.

On St. Clair, streetcars returned in Fall 2024. This article reviews tracking data for March 2025. In brief, a common situation along the route is that streetcars spend more time waiting at intersections than they do serving stops once they cross to a farside platform. The traffic signals are not organized to favour streetcar movements, and the configuration with farside stops is counterproductive.

To be fair, one cannot simply throw a red signal in front of cross traffic every time a streetcar appears as this will affect both auto and pedestrian crossings that require a minimum time. When both the St. Clair and Spadina lines were designed, a concern was the frequency of streetcar service and the high probability that there would always be a streetcar demanding priority. However, the larger Flexity cars combined with TTC service cuts makes this claim less valid.

Pre-emptive streetcar signals could be challenging at some locations, but there is the separate issue of timing of the phase for left turning autos. Should it come before or after streetcars are allowed to cross the intersection? Streetcar-only phases are only provided at some locations on Spadina where cars are turning across traffic lanes, but not for through movements. There is a streetcar-only phase at Lansdowne and St. Clair westbound, but otherwise left turning auto traffic always has priority.

A related problem lies in TTC’s glacial streetcar speeds through special work which would require a lengthy transit phase to allow streetcars to clear intersections on Spadina with streetcar junctions. With streetcars crossing on the shared north-south green, this does force motorists to wait for streetcars to get out of their way.

The data on St. Clair show a common pattern: streetcars spend more time waiting to cross intersections than they do serving the farside stops. They may operate in their own lane, but the combination of farside stops and signal design works against transit operations. This has implications for the soon-to-open “LRT” lines on Eglinton and Finch.

If we are not prepared to give transit vehicles true priority over cars, how do we expect to attract more riders to the system and, in theory, reduce overall road demand?

The remainder of this article describes the methodology behind my analysis and shows the behaviour at individual intersections along St. Clair.

Methodology

All of the charts here measure the travel time on the approach to intersections, and on the farside where in most cases there is a transit stop. To do this, I created screenlines to track vehicles crossing the middle of an intersection and points 100m on either side of that. With typical Toronto streets being 20m wide (66 feet, or 1 surveyor’s chain), this puts the east and west screenlines about 90m either side of the intersection, or roughly three car lengths.

The model I have used for some time on this site for service analysis subdivides a route into 10m segments to simplify and “flatten” the geometry (a key consideration where routes do not travel in a straight line, although not a problem on St. Clair).

In an earlier version of the TTC’s tracking data, vehicle positions were reported every 20 seconds, but not all vehicles reported in every time slot. The current version gives times with any value for seconds, but records are not evenly spaced in time with occasional bursts within a few seconds, and other periods with gaps. I round the times to the nearest 20 second mark, and discard duplicates. The vehicle position is interpolated on the 10m scale between two successive reported locations base on 20 second increments for the missing slots. This produces tracking charts showing vehicle movements without gaps.

It is possible that a vehicle will be ahead of the 10m segment containing a screenline at time “t”, and after the segment at “t+20”. In this case, the assigned crossing time is “t+10”. This granularity in the crossing times means that any calculated travel time between two screenlines will be a multiple of 10 seconds. It could also be zero if both screenlines were crossed in the same 20 second timeslot. This is equivalent to a car traveling 100m in 20 seconds or 18 km/hr.

For route level analysis, these simplifications reduce the complexity of the data, but for very fine-grained views of a route, such as here, the granularity shows up in the calculated travel times between nearby points.

Reading the Charts

For each intersection and direction, there are four charts. On the left are the raw data points for all weekdays in March arrange by time of day (x-axis) and duration (y-axis). On the right are the average and standard deviation values (a measure of scatter in the values).

The top pair of charts show the times between the screenline on the approach to an intersection and the midpoint. The bottom pair show the times from the midpoint to the screenline marking the departure.

For intersections with farside stops (“FS”), the top pair show the time spent on the approach and waiting for a clear signal, and the bottom pair show the stop service time plus a small interval beyond the stop itself.

For intersections with nearside stops (“NS”), the top pair show the approach, stop service and waiting time, while the bottom pair show the time from the centre line of the intersection to the departure screenline.

The screenlines are named “East”, “Centre” and “West” corresponding to their relative locations.

As a general note, times on the approach side of an intersection (east side westbound, west side eastbound) are longer than on the departure side. Streetcars are caught by a red signal and have to wait to the next cycle to cross to their stop. This can easily happen thanks to time spent serving a farside stop during which the streetcar falls out of the traffic wave and arrives at the next signal too late to pick up any green extension that might be active there.

The differences between average times on the approach and departure side of intersections might appear small, but they are cumulative across the route.

Westbound

Avenue Rd (FS)

Travel times on the approach (east) are clearly higher than on the departure side of Avenue Road even though the latter includes the actual stop.

Although the values of the averages are not that far apart, the standard deviation of approach values is much closer to the average thanks to the wider range of values on that side of the intersection.

The spikes in the standard deviation values were cause by a delay where one car held at the stop (west side) and its follower held on the approach (east side). The delay was 11 minutes long and so the data point does not appear on the charts.

Spadina (FS)

As at Avenue Road, the travel times on the east side are scattered over a wider range than on the west side.

Bathurst (FS)

Bathurst sees extended travel times both on the approach and departure side of the intersection.

Christie (FS)

Although Christie is a minor intersection, times on the east side are longer than on the west.

Winona (FS)

Travel times on the east and west side of Winona are not as markedly different as at some other locations, but there are still more scattered values (higher standard deviation values) on the approach (east).

The standard deviation spike at 11am on the west side of the intersection was caused by a delay near Oakwood that held service both ways on March 28. There is a corresponding spike eastbound at Dufferin.

Oakwood (NS)

The stop at Oakwood westbound is nearside, and so travel times here include stop service times. There were also a few delays that skewed the standard deviation values on the east side.

By contrast, times on the west side are low. Although cars will cross the intersection at a walking pace due to the standing slow order on all special trackwork, the data points are clustered around the 20 second line.

Dufferin (FS)

Dufferin westbound is unusual in having longer times on the west side including the farside stop than on the east side. This could be a case where detection for transit priority (green time extension) is working better than at other locations.

Lansdowne (FS)

The scatter in east side times at Lansdowne is higher than on the west side although the average times ar lower. This can occur due to streetcars waiting to make left turns south on Lansdowne to short turn at Earlscourt Loop.

Old Weston Road (FS)

Old Weston shows the familiar pattern of longer and more scattered times on the east approach than on the west which includes the car stop.

Keele (FS)

Keele repeats the pattern seen elsewhere with more scatter in the eastern approach times and a slightly higher average.

Eastbound

Eastbound data show quite similar patterns to westbound.

Keele (NS)

The stop at Keele eastbound is nearside, and so the stop service time is part of the western approach to the intersection. On the east side, the times are very low.

Old Weston Road (FS)

Lansdowne (FS)

Approach times from the west at Lansdowne can be high because this is a crew change point.

Dufferin (FS)

Oakwood (FS)

Times at Oakwood are generally high because these segments include the slow order on special track work at Oakwood itself and at Robina, just east of the stop.

Winona (FS)

Winona is unusual in having low approach times. This could also be due to this being a minor intersection where proportionately more green time is available for all east-west traffic.

Christie (FS)

Bathurst (FS)

As with the westbound data, values for Bathurst eastbound are scattered over a wide range on the approach (west) side of the intersection.

Spadina (FS)

Avenue Road (FS)

Eastbound approach times at Avenue Road are scattered over a wide range, even worse than westbound data shown earlier.