Back on June 22, 2005, the matter of transit priority signalling was discussed at the TTC meeting. Arising from that discussion, then Vice-Chair Olivia Chow moved the following motion:
1. That staff be requested to take the necessary action to implement transit priority signalling on Spadina by September 2005 at all locations where it is not already active, with a report back in the Fall of 2006 on the impact.
2. That recommendations 2 to 6 embodied in Mr. Munro’s submission be forwarded to TTC staff and City Transportation staff, with a joint report back to the fall meetings of the TTC and Planning and Transportation Committee.
This item has sat on the list of outstanding Commission requests ever since, but on the recent agenda, it was closed with the notation:
Memorandum dated September 2, 2010 forwarded to Commissioners.
It took a motion of the Commission and a bit of harassment on my part to get this memorandum. It was not exactly worth the wait.
Transit Priority — Signal priority on St. Clair is complete. Signal priority on Spadina will be completed by the City in December, 2010. Signal priority on Harbourfront will be upgraded when the Queen’s Quay Revitalization Project is undertaken by Waterfront Toronto (date unknown). Recommended comments and action: Mark complete, and remove from list.
With respect to item 1, it has always been the City’s position that the width of Spadina coupled with the service frequency preclude more aggressive priority for transit vehicles. Where the 2010 response speaks of completing transit priority on Spadina, this refers to complete integration of automatic track switches and traffic signals.
There has never been a report as requested in item 2 of Commissioner Chow’s motion, and it is worth reviewing what was involved in points 2 to 6 of my original presentation.
2. The operation of transit priority signaling at intersections with long cross-street green times should be reviewed, and transit vehicles should be given more frequent access to green time in the cycle.
This remains a major problem at the crossings at the south end of the Spadina route where transit service waits quite a long time to get through intersections between Front and Queen’s Quay. There are also problems at Bathurst and Lake Shore, although these have been reduced somewhat over the years.
3. The TTC and Planning & Transportation should integrate transit priority signaling with any new traffic signal installations on routes which already have priority signals. Activation of the new signals should not be permitted unless the transit priority component is operational.
4. All road and track reconstruction projects should include replacement of priority signaling detectors as an integral part of the projects so that priority signals are available when service resumes.
Both of these points refer to situations where transit priority is lost through the installation of new signals or reconstruction of the roadway. Intersections such as those on Parliament at King and at Queen, and on Roncesvalles at Dundas, Howard Park and High Park have recently received completely new track. How long will it take for transit priority signalling to be reinstalled at these locations?
5. The TTC and Planning & Transportation should determine a method to allow closely spaced priority signals to be activated as a group so that signals that are not at transit stops can give transit vehicles a coordinated green wave.
6. The TTC should investigate the provision of technology allowing operators to request transit priority when they are at nearside stops to allow signals to cycle to green while the transit vehicle completes loading.
With the addition of more signals in the central area (sometimes as replacements for pedestrian crossings), and with the close signal spacing on the Harbourfront line (existing and planned), there are locations where signals lie between rather than at carstops. Much of the debate between TTC, Waterfront Toronto and the City’s technical staff turns on the issue of how closely spaced signals will interact with transit.
It is essential that all parties find a way to create “green waves” through such areas so that transit vehicles are not repeatedly delayed. Where between-stop signals exist (e.g. King at Frederick between Sherbourne and Jarvis), the in between signal should be integrated with nearby intersections so that transit vehicles are not held by it.
For nearside stops, there are cases where a car will be loading for an extended period, but the traffic signal stays green attempting to give the car priority it cannot use. Conversely, operators should be able to identify that they are ready to leave so that a signal can react appropriately. This is particularly important where a signal is located near but not at a stop (eastbound from the York Street stop on Queen’s Quay). It should not be necessary for a car to pull forward onto a detector loop to signal that it requires clearance. Streetcars already have antennae that are used to operate the track switches at intersections, and these could also be used to interact with the traffic signal system.
Having watched service diversions around numerous construction projects, I have noticed one other shortcoming in priority signals. There is no provision for transit-only turn phases at locations commonly used for diversions and short turns. Examples include:
- Queen at Parliament, Church and Shaw
- King at Parliament, Church, Shaw and Dufferin
- Dundas at Church and Parliament
Judging by the new installation at Parliament, the TTC is now electrifying all switches, although at this intersection they are not yet active. One can only speculate about the TTC paying a pointman to operate the switch westbound at Church and Queen for the 504 diversion when repairing the existing (but out of service) electric switch might have been a lot cheaper. The pointman was not responsible for any traffic management (that would have taken a paid duty policeman), and the absence of a transit priority phase slowed transit operations here.
We talk a good line in Toronto about transit priority, but do a half-assed job of implementing it where it is most critical to operations.
I have noticed that at an intersection near me, Jane and Weston, that a green signal stays on after the pedestrian signal had counted down to zero. However, there was no bus in any direction. Sometimes, I have seen a bus ready to go in one direction, but the extended green in the other direction has no bus anywhere. There are other intersections that have to same “problem” with the extended green, as well.
Steve: This is not a transit priority problem. At some intersections, for reasons best known to the signal techies, the countdown will start even though nobody has requested a cross-street cycle (pedestrian pushbutton or car sitting on a detector loop). When it gets to zero, the signal says “oops”, stays green for the main street, and reverts to a white hand for the main pedestrian direction. This is a rather stupid way to set up an intersection because someone who is not paying attention may assume a change to a cross-street green is about to happen, and behave inappropriately. Yes, they should pay attention, but not all do. Also, of course, it makes a main-street crossing illegal (against the flashing hand) needlessly while the countdown runs.
What about SCOOT (Split Cycle Offset Optimization Technique)? Rob Ford wants to install this everywhere. That should certainly help if it’s installed on Spadina.
Steve: A fundamental premise of SCOOT is that there is enough capacity to go around, and SCOOT will figure out the optimal arrangement given current demand. So-called priority provided by extended green time in transit’s direction of travel depends on there being spare capacity to reallocate for the transit movements. Much more fundamental to the discussion is a question of policy — how an intersection should operate when it is under stress, to what degree transit moves should take precedence even if this reduces the quality of service for other traffic. This discussion has different outcomes where transit is in its own right-of-way and when it is in mixed traffic.
Part of the debate on Spadina turns on locations such as Queen Street where a route running in mixed traffic crosses one running on its own right-of-way. In brief, the argument against improved Spadina priority is that this will only make conditions on Queen worse. Spadina cars would benefit at the expense of Queen cars.
The infamous newspaper article about how travel on Spadina is slower than on Bathurst started this whole discussion, although it didn’t allow for the different level of demand and stop service times on the two routes southbound in the AM peak (the time in question). The Spadina right-of-way, however, suffers because the general flow of traffic is managed on the cycle that suits cars. The streetcars stop for long enough that they fall out-of-cycle with the auto traffic are are more likely to be held by a high proportion of signals at cross streets. This wastes the benefit of farside stops by making the streetcars waste time waiting just to cross to their platform, with the added insult that they must let left turns go first.
At locations such as Lake Shore, the east-west green time is very long in an attempt to clear traffic from connecting streets and to maintain flow in that direction. However, the transit service suffers especially when the cycle time for north-south transit moves produces delays on the order of one headway. Compounding this is the close presence of signals at Bremner, Lake Shore and Queen’s Quay. The last of these is particularly tricky because most of the service turns one way or another, and this is harder to fit into a scheme where transit vehicles share green time with motorists. The intersection only works because traffic volumes are comparatively light.
One reason that TTC staff is so strong in advocating the Bremner LRT alignment for the Waterfront West line (leaving aside whether it will ever be built) is that this eliminates the need to operate frequent service across the Lake Shore and Bathurst intersection, something that would require much more aggressive transit priority and allocation of green time than is now available. As long as traffic planning treats Lake Shore as a major arterial to be kept moving as fast as possible, transit will suffer.
Harbourfront differs from Spadina in that streetcars can only proceed at signals on their own green time because left turns are not controlled separately as they are on Spadina. When Queen’s Quay is reconfigured, this problem may disappear, although it could be offset by the increased number of signals planned for this street.
That brings us to the problem that the signal scheme used in Toronto does not appear to be capable of creating “green waves” when a transit vehicle approaches a series of signals so that it can proceed to its next stop unhindered. Each signal is treated as a separate control system, and only when a streetcar is within a signal’s local area is it detected and provided for in the priority scheme. The simplest example of this can be found westbound approaching York where a car leaving Queen’s Quay station triggers a priority request for the signal at the top of the ramp, but the request for York is not initiated until after the streetcar crosses into the next “block”. A run straight through is rare, and usually happens when the transit phase has been triggered by an eastbound car that happens to be in just the right location.
Finally, for streets that do not have transit lanes, the single biggest problem is that we use far too much space for parking, not to mention incursions for construction such as we see on King near Spadina. There is only so much capacity to go around, and we waste a lot of it. This argument is independent of whether transit is present or what technology is used.
St.Clair still has many stops where signal priority lights have not been activated; since when do left turns have priority over transit? I’ve only seen 1 or 2 at the most that gave the streetcar the right of way before left turning vehicles and then general traffic.
The TTC’s idea of transit priority on St. Clair is the same as streets without right of ways. It simply holds the green from turning to red if a streetcar is approaching. If the light is already red, nothing really changes – as long as there is a car on the left turn detector loop (or it’s broken…like at Yonge & St Clair) – the left turn signals will always activate first.
The only exception to this is intersections where streetcars have a turning phase such as St. Clair & Lansdowne, which activates before the autos’ left turn signal.
Although with far side stops, holding the green may be a BIT more helpful than near side ones where, as Steve mentioned above, you can have the greens holding for a car that is still loading, delaying everybody.
Steve: One note about St. Clair & Lansdowne (and other similar intersections such as Queen & Broadview eastbound): The transit phase is activated through the electric switch circuits. If the switch is out of service (a typical case for diversions), then the advance green does not work. Sadly, this is often the time it is most needed because more cars than usual are making the turn.
The service frequency on Spadina is so high that chances are, at any given intersection, a streetcar would be approaching from the north or south, and would therefore request priority over the E-W street. As the schedule is now, the “green waves” that you’re asking for would result in little or zero green time on the cross streets, and this problem would be further aggravated by streetcar bunching. Now maybe if we ran Transit City style cars (two double-length streetcars coupled together) and increased the headway to every 7-8 minutes, it would work, but I don’t see how it possibly can right now.
Even if we put the concept of priority aside for a minute and try to synchronize the signals on Spadina in such a way that streetcars would ride a green wave more often than not, it all starts to disintegrate the minute one streetcar falls behind schedule. In a way, it’s similar to the problems at the wye in the 60s when the subway trains would fall out of synchronization with the hard-coded ATD schedules, forcing the tower controller at St. George to intervene. If everything was on time, every train would just sail through and never get stopped … until the first delay occurred.
So, unless you have cameras and several sets of human eyes watching the Spadina line and the intersections, and adjusting the signals on the fly based on traffic conditions, you have a similar situation.
Here’s an example of how the scenario W.K. Lis describes can be beneficial.
Let’s consider an intersection in the former Toronto where the side street is a lower-volume street, and the side street green light will be skipped if a vehicle isn’t present and a pedestrian hasn’t pressed the pushbutton.
A lot of signals in the central city are set up to cycle through every 60 seconds outside of the weekday rush hours. I am looking at one example where the main street gets 33 seconds, and then the side street gets 27 seconds (but this can be cut short or skipped entirely, in which case the extra time goes to the main street). If the main street starts at t=0 seconds, then the first opportunity for the side street to cross is at t=33 seconds. The average waiting period will be somewhere around 20 seconds.
However, before the main street can get a red signal, pedestrians crossing the side street need to be given enough warning (the countdown interval). In this particular case, the countdown interval starts at t=18 seconds. If I press the pedestrian pushbutton at t=19 seconds, I’m out of luck, because now there isn’t enough time to provide that countdown interval. I need to wait until t=(60+33)=93 seconds before I will get a walk signal, or 74 seconds after I arrived — more than one full signal cycle extra.
(This applies both to pedestrians and side street vehicles, but is accentuated for pedestrians because of the slow travel speed. Over those 18 seconds, a typical pedestrian will only walk 20 to 30 metres, or less than half a block — close enough that it essentially feels like you’re at the intersection, and therefore feel as though you’re at the intersection waiting to cross for at least a couple of minutes.)
More than a couple of times this has burned me at my local intersection, where it feels like I should have plenty of time to wait for a walk signal before my bus arrives (see? Back to transit) — but I have just missed the “window”, end up waiting another full cycle, over which time my bus has come and gone. Had I arrived one second earlier, I would have been on that bus.
This problem is solved at intersections that begin the countdown automatically, but reset if no vehicles or pedestrians are waiting to cross the street — although it certainly looks unusual.
For the Transit City lines, as well as maybe St Clair, would the TTC ever consider implementing something like a railway crossing, i.e a crossing with arms to stop the flow of traffic, thereby eliminating the need for signal priority, and then always allowing vehicles to get through un-inhibited?
Steve: I very much doubt this would happen. I can think of all manner of problems including arms coming down on top of cars and pedestrians, and maintenance issues — if the arm doesn’t work, can the line operate, for example.
@M. Briganti — what you say is true for the major intersections on Spadina, but I don’t see why at least partial transit priority couldn’t be implemented at Sussex, Willcocks, Nassau, St. Andrew, and Sullivan (and Clarence Square). Sure, if there’s a string of streetcars coming in both directions, some of them will have to get a red, but it would still be better than the way things currently work, where the first car to leave Spadina station after, say, a five-minute gap can still get held by a light just turning red at Sussex. (Where there’s rarely more than 3 or 4 automobiles going east-west!)
Also, I’m not on St. Clair all that much, but is there any streetcar priority at St. Clair and Oakwood? The eastbound car I was waiting for just got caught by the light turning amber, and it didn’t seem like the green was being held at all, since the amber started as soon as the pedestrian countdown hit zero. So we had to wait…and the cycle at Oakwood takes forever!
Steve: There are many intersections that could do with fine tuning of whatever “priority” is in place, but it does not seem to be a priority for anyone. The divided jurisdiction between the TTC, plus an ongoing battle between the “experts” at TTC and the City, doesn’t help one bit either.
Unless something has changed since the EA was completed, there won’t be any additional signals for Queen’s Quay between Spadina and Bay. One light will be removed between York and Bay and one added between Simcoe and York for the new streetcar stop that will replace the York and Simcoe stops.
There is talk of adding additional stops for the new streetcar route east of Bay.
Seems like they should merge the two groups of “experts” into one team. That team is responsible for making everything working smoothly rather than two teams with one focusing on cars and the other focusing on transit…and pedestrians and bikes getting no team.
Also, out of curiosity. It would be great if the stop light data was available online. I wonder what percentage of streetlights “report back” their status to a central db. I’d imagine that most are just on timers, but some must be networked or whatever. This data combined with streetcar data would be very interesting to analyze. It would also be able to link it with the feed from the maintenance department to determine how often the priority is getting damaged in maintenance/road closure activities. Another item that would be great to get on the open-data toronto site.
Well, speaking of the devil:
TTC Nextbus Live DATA – FINALLY!!!!
Traffic Signals Data
I will be incorporating this into borkbork.com/ttc as soon as I get home…
I noticed this morning on Queen’s Quay that streetcars got green lights both just after the lights had turned red for the main roadway, and also just before they turned green. That gives streetcars two transit phases in one light cycle. I think this is new?
Steve: No, the lights on Queen’s Quay were changed some time ago to work like that. It is a particularly important adjustment where there are long cycle times because otherwise streetcars can be held as long as one headway and bunching is inevitable.
I think the problem with Spadina cars stopping unnecessarily (and wasting time at red light) can be solved by moving the streetcar platforms nearside.
Under this new system, left-turning motorists would overshoot the intersection, merge into a dedicated U-turn lane, and perform the U-turn midblock (which is usually signalised). There are examples of this in Toronto already:
– St Andrews Street at Spadina Avenue
– And a couple of locations along St Clair Ave. West in Corso Italia and Earlscourt (although I’m not sure if these stops are nearside of their street).
Here’s the animation I made to illustrate this concept.
Steve, what do you think? According to traffic engineers, the only reason not to apply this system is if there are large trucks (which have large minimum turning radii, too large to make a U-turn). However, I don’t remember seeing any large trucks on Spadina.
Steve: First off, I don’t agree with your comment about large trucks. They exist, and simply dismissing them is a fast way to lose credibility. Second, this type of hook turn would require a left turn to be made into the opposing stream of traffic which may not have capacity for it, and the turning vehicle would then block the tracks. Also, I doubt that every existing left turn location has a suitable “farside” left turn bay slot. Queen southbound is a good example. Note also that provision of the U/left bay will increase the number of traffic signals on the street, and this gives more potential delay points for the streetcars.
While I am not thrilled with the way the current design works, the real issue is that streetcars move at a different average speed from the rest of the traffic, but the signal timings suit the cars, not the transit. Streetcars should not be caught by the end of a green cycle that has no extension provision, and the likelihood that they hit a red on approaching an intersection should be minimized.
I’m not trying to outright dismiss the existence of large trucks on Spadina. I assumed that since Spadina already has U-turn spots at two locations (Sullivan and St Andrews) that Spadina Avenue is designed to accommodate vehicles that could perform U-turns.
The EA for the Eglinton Crosstown LRT proposes a similar configuration as well for selected intersections. What makes it ok for them to assume that large trucks would not have to contend with downstream U-turns?
Midblock U-turn signal would be in phase with the streetcar signal back at the streetcar stop (i.e. when it’s green for the streetcar stop, it’d be green for the streetcar at the U-turn spot as well). Also, U-turn locations would be fully signalised; oncoming traffic that could potentially block U-turners would face a red-light, while U-turners proceed.
I don’t expect this to work along the entire stretch of Spadina. Just north of Queen.
Steve: I don’t like the design for the Eglinton LRT U turns either. I think that this was one major reason for the lack of credibility in the plan for surface transit. That it was proposed for Eglinton does not justify using it on Spadina.
I found a whole discussion on the re-routed left turns on another blog post about the Eglinton LRT design. It seems that opponents to this design believe that the rerouted left-turns inconveniences motorists more than conventional left-turn systems. That’s understandable (more turns, added distances, more signals, etc.).
There was much skepticism throughout the U.S. when Michigan implemented the rerouted left-turn system too. However, an actual traffic study found that:
-left-turn-related collisions decreased by 20-50%
-through-capacity increased, also by 20-50%
-merging-related collisions decreased slightly (can’t remember statistic for that)
These benefits are especially realised during rush hour and times of heavy traffic. Yes, the Michigan-left becomes more of an inconvenience when traffic is light.
Plano, Texas offers an anecdotal evidence of the system’s benefits:
“Backup at the intersection had been reduced by 60 percent, which means that vehicles cross through the intersection about 35 seconds faster than before. ”
However, I whole-heartedly agree with you that it is silly to adopt a foreign system for an Eglinton transit line, and simply because it works in an unrelated case in a different location, without running an actual traffic model for Eglinton. But I would disagree with ruling it out immediately.
Steve: The difference with the Toronto proposal is that it has the extra traffic of the LRT line right through the middle of the left turn lane, and this invalidates any modelling or “real world” experience. Apples and oranges.
Just wanted to add a correction:
The reduced delay was probably experienced by through-traffic, not by left-turners themselves. Left-turners would probably be more inconvenienced.
Jacob Louy said:
Your animation only shows a streetcar going one way. Given that the schedule shows headways that alternate between 2 and 3 minutes, and that there will be cars going both directions, you will need to synchronize a north of Sullivan turn with not only the lights at Sullivan, but also with the southbound streetcars. Do you honestly believe that this can be done in a way that will not impact the TTC service?
Also, as Steve noted in a reply, you need to stop traffic in the southbound lane (in this case) to insure that turning cars have a way to enter the traffic stream. If the traffic is heavy and the crossover is close to Sullivan, this light would have to be slightly out of sequence with Sullivan to ensure that the southbound traffic has a chance to move enough to leave space for a large vehicle to get fully onto the southbound lanes. Also, if lights are too close, not only is it very hard to synchronize them, but also it becomes much easier for a driver to focus on the more distant light and miss the local one (and this becomes even more likely if there is not a street at the local light). The last thing we need is a turning vehicle being hit by someone who accidentally runs the light while the rear of the turning vehicle is still on the tracks.
If the turn approaches the midpoint between the two stops, then it will seem to make no economic sense to have separate crossovers for the two directions, and the designers will make a single turn for both directions, and this will have to synchronize with both Sullivan and Dundas. I rather doubt that this can be done easily, and expect it would become another point of delay for at least one direction, which would negate the whole reason for the moved turn.
Note to Steve: I, too, didn’t like the Michigan turns that Transit City posited on the western portion of the LRT. Partly this was because I do not trust the statement that they only need 200 m from the signal point (station) in order to cycle the lights (though this might be correct in this case as there would be no pedestrians to be accommodated) and because I doubt the ability to synchronize for LRTs in both directions. A second problem I foresee is the potential for accidents involving turning trucks that would leave the truck still partially blocking the tracks.
I strongly believe that if they are going to use this sort of turn, the U-turn portion needs to be on the ‘side’ street. I know that they chose the main street because the median with the tracks on it gives a much larger turn radius, but this can be overcome by creating a short wider stretch on the side street at the point where the turn is to be made (similar to a mid-block bus bay). The cost of expropriation (if it is actually required) of such a small area should be minimal.
What if new LRT lines were to be implemented in the reverse direction (i.e. instead of having LRV’s travel on the right side of the ROW, they would travel on the left side). That way, more stations can be nearside while motorists still have their conventional left-turn system.
The only disadvantage I can think of is that it might be confusing to passengers, but this would not be the only example of a “reverse-direction” line (Hong Kong has a line that is reversed). With doors on both sides of LRV’s, I don’t imagine too any technical difficulties.
What are your thoughts?
Steve: My main concern relates to the confusion that “wrong way” traffic will have. Motorists, pedestrians, cyclists, everyone is used to the idea that vehicles approach from the left. This would certainly be a messy situation anywhere that a line had to slip into mixed running and cars needed to switch to the “correct” rails. A mixup at a left turn where both the auto and the LRV started at the same time would result in a head on collision, not a sideswipe or lateral blow at lower relative speed.
It’s conceivable that if the system had always worked like this for many decades, it could be workable, but not as a net new addition. We have enough troubles getting people to accept LRT without the complexity of left hand running in the middle of streets.
Thanks for your feedback on the concept of wrong-way LRT’s. I certainly don’t expect the TTC to actively pursue this alternative until a credible study has been published on the effects of wrong-way LRT’s.
This definitely would not work if the LRT line was not entirely on a private ROW.
As for collision severity at intersections, an LRV starting from a nearsided station would travel through an intersection at a slow speed as it accelerates from rest, compared to an LRV deccelerating from a high speed toward a far-sided platform.
An opening in the LRT ROW (at minor intersections without stations) for left-turning motorists would be more critical locations for severe head-on collisions, if motorists do not obey their signals.
On the matter of transit “priority” is it just my imagination or have all of the signals previously fitted with this advanced technology been unfitted during the past few years? I remember streetcars that used to hold lights green for a good time but that doesn’t seem to happen anymore.
Sometimes however it’s the signal timing itself that needs to be checked rather than some kind of prioritization scheme. For example, traveling late at night and going east on King, the light goes green at Jarvis only to turn red before any vehicles get to George Street. This is just stupid.
I’ve noticed the same thing early in the mornings traveling on King and Queen where streetcars will pick up a passenger and proceed through a fresh green, stop at the next block and pick up one more and have the light turn red, move to the next block, pick one person up and light turns red again. This was not downtown! This was at minor cross streets leaving downtown. Someone at city traffic needs to have their head examined.
Steve: Yes, I have also noticed that gradually the transit priority signalling is failing at various locations, and nobody at the TTC seems to make a point of checking this. Also, it is likely that there is no budget to fix problems like this (e.g. the loop detector failing in the street) in a classic case of having money for capital (installation) but not repairs.
As for George and King, this has never seemed to have any kind of priority, nor is there much if any attempt to sync the lights here with the operation of Sherbourne or Jarvis. Part of the problem with transit priority where signals are close together is that the detection loop for an oncoming transit vehicle is too close to the intersection, and the streetcar has only a brief interval to request an extended green. The whole issue of “green waves” for transit seems to be completely beyond the abilities of Toronto traffic engineers. It’s a big problem on Queen’s Quay (existing and proposed sections) as well.
Especially puzzling is the implementation on Carlton Street, as signals hold green for Carlton cars as passengers board/alight. The signals ‘give up’ and turn red as the car is almost ready to proceed.
Operators seem either oblivious or indifferent to signal priority. My impression is if they would promptly shut the door and hit the ‘gas’ they could take advantage of the held green. While I acknowledge Toronto has poorly implemented signal priority, perhaps operators should be better trained regarding the system as installed? Bizarrely, some operators don’t know the system exists.