24 thoughts on “Analysis of 512 St. Clair, January 2010 (Part 2: Weekdays, Week 1)”

1. Now what we need is for Mitch Stambler and Rick Cornacchia to respond in the same detail as your analysis with how they are getting this fixed!

   Steve: The response would likely focus on the new schedules that went into effect in mid-February without addressing the underlying problems. Still to come is my review of the Queen car’s split operation, but St. Clair was more pressing as it is an active service for which so much was expected.

   LikeLike

2. Have there been reports of improvement since then? All I note is that complaints appear to have dropped off.

   Steve: The new schedules implemented in mid-February appear to have substantially reduced the problems. However, I won’t know until I actually get the data just what has improved.

   And, yesterday, I took the service from Lansdowne to Yonge at around 11:30. To get there, I took the subway to Lansdowne and took the Lansdowne bus up to St. Clair. I note here that the TTC was clearly struggling with the snow. There was no sign of a northbound bus, and there was a disabled southbound bus and three southbound buses coming in behind. When I noticed that one southbound bus was turning off, possibly to short turn at Wade Avenue, I started walking north. By the time I got to the next stop, I heard a bus approaching behind me, and I flagged it down. Despite the crowd that had been waiting at the northbound stop at Lansdowne station, this bus was empty. Hmm…

   Anyway, at St. Clair, I left the bus at the loop and headed to the streetcar stop. No streetcars were waiting, but one pulled in as I arrived. A supervisor was on duty to send the driver en route just about right away. We left at 11:30 a.m. sharp.

   The run was smooth and uneventful. I noticed very little bunching in the cars coming the other way, and no sign of delay-gathered crowds, although ridership was strong during this midday run. The traffic signals frustratingly gave a phase to left- turning cars ahead of the streetcars and through traffic, but this didn’t prove to be an excessive delay. In one case, we arrived at the intersection just as the left turn phase ended, and breezed across to our stop. The only delay of note was at Dufferin.

   We got from Lansdowne to Dufferin in five minutes, and from Dufferin to Bathurst in another five. We breezed through St. Clair West station and continued on our route, entering St. Clair station eighteen minutes after departing Lansdowne, finding one more car waiting in the pick-up area.

   I also noticed a very good cycling of passenger traffic en route, with people getting on and getting off at every stop, and not everybody travelling to the subway. This is a good sign that the route is being used by the locals for local trips.

   A few oddities, though:

   – One car was on stand-by in the eastbound through track at St. Clair West.
   – All westbound cars were signed 512 KEELE, except for the car I was on. When it arrived at Lansdowne loop, it was signed 512 LANSDOWNE. It took on 512 LANSDOWNE again as it arrived at St. Clair station.
   – I noticed snowplows on the right-of-way, travelling quite slowly. If anything, THEY delayed the streetcars.
   – The right-of-way was used by 90A Vaughan buses accessing St. Clair West station.

   All in all, I’d call this a very positive experience. St. Clair was operating far more reliably than 47 Lansdowne and 501 Queen, despite bothersome snowy weather. This is the level of service that I would expect from a project such as this. Did I get lucky? Or are the supervisors and the drivers getting their acts together? Either way: kudos, guys.

   LikeLike

3. You appear to be relying on GPS data for this analysis, have the TTC phased out CIS ??? I assume the GPS data is more accurate.

   Steve: Yes, I talked about this both in the introductory article and in the technical backgrounder linked from it. The GPS data is extremely accurate, but just as important, I get position information for every car every 20 seconds. This is a much more granular view of operations than the old CIS data.

   LikeLike

4. Why on earth do they send them out of the station in bunches? If there is a gap in front of the first car I can see sending out two cars, the first one being empty to spread the overflow, but this seems crazy.

   Short turning the cars in bunches seems to solve nothing except to change the direction of the bunches. The bunches have to be broken up.

   Steve: Yes, that’s rather obvious, but at least for the first week in January, that’s not what happened. Looking at the whole month, things get better gradually and week 4 isn’t anywhere near as bad as week 1. However, the weekends are a disaster throughout the month probably because of a lower level of supervision.

   LikeLike

5. Let me add, short turning the cars in bunches seems to solve nothing except to change the direction of the bunches. The bunches have to be broken up.

   LikeLike

6. I’ve read a blog entry on Spacing Toronto about Transit Signalling Priority in Toronto. It said that it has been implemented on several routes including 510 Spadina and 512 St Clair. No offense, but it doesn’t seem like that to me. Maybe the Signal Priority systems have been switched on for certain intersection (others not)? Is this true? If so, I would understand why Signal Priority would not be switched on for Spadina at King, Queen, Dundas, or College, those other streets carry streetcars as well, and having Signal Priority for Spadina would hold those other routes up.

   It seems like Signal Priority has not been turned on for St Clair at Avenue Road. Avenue Road itself doesn’t have much transit service (only a rush-hour and a limited service route). If this is the case, why would Transit Priority not be turned on there?

   Steve: That is a wonderful question. City and TTC staff were asked five years ago to report on this matter as it relates to Spadina/Harbourfront, but have not yet deigned to produce the report. Avenue Road is almost certainly a location where the City’s traffic engineers feel the north-south road traffic is more important than the streetcar and have not done much, if anything, to give transit priority. You are correct in saying that most of Spadina does not have transit priority also, and the only thing that actually works is the white bar call-on signals for turns at intersections. Many of these are linked to the electric switches, and so at locations where the switches don’t work (another story in its own right), a streetcar has to battle its way through traffic if it is on detour.

   LikeLike

7. Two questions.

   the “Headways” chart. I read this as the time between each car, so if we see a point of 2:00, then there were 2 minutes between the leading car, and it’s following car.

   In this case, am I right in seeing a 42 minute headway just before 9pm on Friday at Yonge St? 42 minutes? On St Clair! That’s nuts. And 33 minutes about 6pm?

   Steve: If you look at the service chart, you can see these gaps. Two cars arrive eastbound and they cross Yonge together at 19:43. The next car does not arrive until 20:25. What has happened is that 7 cars have accumulated at the west end of the line. There should be 8, but one appears to be missing. There is no obvious reason such as a delay to service that explains this situation.

   The gap at 18:00 is discussed in the article. Several cars were held westbound approaching Lansdowne, and the first eastbound car after the gap was at 17:41. If you look closely at the service chart, you will see a car (orange) arriving at Lansdowne about 16:55, and staying there until 18:12 when it runs east to Vaughan and presumably back to Roncesvalles. I suspect this was a bad-order car that repeatedly circled Earlscourt Loop, although not very often. Eventually it just sits, and the rest of the service piles up behind it.

   Second question: How is the GPS data given to transit control? Is it something that they can see in real time, and detect that there are bunches, or is it stored on the vehicles and downloaded at the end of the day?

   Steve: The central system polls each vehicle every 20 seconds to get updated status information. This information drives the Nextbus displays although these are only publicly available for Spadina and Harbourfront. TTC supervisory staff at CIS control should be able to see where all of the cars are, and eventually, on-street supervisors will have handheld units to do the same thing.

   LikeLike

8. Whenever there’s a gap the first car out is overcrowded. It crawls along the route with lengthy delays at each stop while people try to clamber on and off. Even if the car is packed it must stop because someone wants off and then others try to get on at the front and the back doors. Meanwhile, the progressively emptier cars move along slowly in a bunch behind.

   When there is a bunch is there any way the supervisors can know where the car would be were things going smoothly and could they not send each car out non-stop until they reach that spot. This hearkens back to opening day on the Yonge subway.

   Of course the spotting on the subway was done before anyone got into the stations so there were no crowds getting frustrated as empty trains went by.

   This would not be the case on the streetcar lines and there would have to be some way of letting the people in on what was happening as the empty cars speed by them. In other words, the TTC would have to communicate with the passengers.

   Perhaps St. Clair would be a good place to experiment with the pixelated message boards at each stop that are supposed to be installed eventually anyway.

   LikeLike

9. Rode on St Clair from Vaughan to Earlscourt loop this evening and was very pleasantly surprised. Cars came every 3-4 minutes, no bunching, passengers cycling on & off regularly. A bit of delay watching the WB car wait for lights at Bathurst then at Vaughan, but after that it was very quick. My wife complained the line was so short – but it was the speed that made the line seem short. Well done!

   LikeLike

10. Do snow plows on trucks clear the snow off the St. Clair ROW? Did they get rid of the snow plow work-cars that I saw as a kid?

    Steve: The snow plow cars (and the sweepers) are long gone.

    LikeLike

11. Apologies if this has been discussed before, but is there any reason why crossover tracks weren’t incorporated into the ROW design? In a situation described by David O’Rourke above where the first, overflowing car in a pack toddles along while empty cars bunch up behind it, would it not make sense to have designated spots where the trailing cars could skip ahead and spread out the service? I can see how this could be a potential disaster in mixed traffic, but there’s no reason such a setup couldn’t work on a ROW.

    Steve: St. Clair uses single-ended cars, and the idea of crossing over to the opposing track raises the problem that you need a way to prevent collisions. There are two basic requirements to fixing bunched cars. First, send one or two cars ahead express to different points on the line to re-establish proper spacing. Second, space out cars leaving terminals so that they do not get bunched up in the first place.

    LikeLike

12. My question about St. Clair is did they realize that they were going to 30 m long cars when the re-did the loops on the line? Both Lansdowne and Oakwood have been “beautified” by landscaping and the inclusion of planters. The result is that it is very difficult in Lansdowne for a car to pass a bus unless the bus stays to the very north side of lane right off Lansdowne. When they go to the longer vars I don’t think it will be possible for a bus to pass a car in that loop, or for an auto coming up the lane way to get past an LRV.

    I don’t know if any buses use Oakwood loop anymore but the right of way has been narrowed and it does not look like two vehicles can pass each other in the loop. The TTC also just re-built St. Clair West Station loop exactly as before. Can a new LRV pass another one at the stop using the spare track? It would have seemed to be a reasonable idea to re build the loop to handle the new cars when they had it torn up rather than just re do the status quo. The other thing I find interesting is the total lack of spare tracks or tail tracks in the new loops for storing bad order cars or ones that need to wait for time. I guess that this is because TTC vehicles never break down or get off schedule.

    LikeLike

13. With regards to transit priority, I think some traffic light controllers NEMA with railroad preemption maybe, must terminate on a left turn ahead of a transit priority call. I wounder if this is the case with Toronto’s transit preemption on streetcar lines.

    2002 ITE Signal Control Paper

    TTC 2004 Signal Control Paper

    Steve: These two papers are fascinating, although I will warn some of my non-technical readers that they presume some knowledge of traffic engineering. Several points leapt out at me.

    In Jim Sinikas’ paper, we learn that the TTC and City are not exactly in agreement on the appropriate way to manage traffic flows. One notable example is the City’s desire to preserve signal timings that, in theory, aid in co-ordinated flow through successive intersections. As the TTC points out, this approach works best on one-way streets, or on streets with unbalanced peak/offpeak directional flows. Co-ordination becomes difficult and potentially counterproductive when signals attempt to optimize for what is deemed to be the “peak” direction.

    Sinikas’ comments about internal TTC problems with maintenance of transponders and with the approach to scheduling and managing service are quite telling. Something we have seen both on Queen, and more recently on St. Clair, is an approach that simply keeps extending running times so that it is almost impossible for a car to ever be late. This produces long terminal layovers, provides so much padding that cars don’t have to run anywhere near schedule, and wastes equipment.

    The article about traffic signal timing schemes and their comparative effectiveness is quite revealing. Three signal control methodologies are modelled, and of these, the second, so-called “Transit Priority” actually closely matches what is in use in many Toronto locations. The fact that it actually produces the worst travel times for the LRT vehicles does not surprise me one bit. Several locations on our busiest routes clearly show how signals are arranged to optimize traffic flows (often on the cross streets) rather than transit flows. To be fair, some locations, notably on King Street, do tend to favour transit vehicles and noticeably improve their operation through intersections.

    I was particularly interested in the European system which has far more flexibility in the definition of signal phases and, unsurprisingly, produces better results overall.

    An important part of this debate is that many streets are “full”, and at best we will squeeze a bit better performance for some types of users (transit), but may do so at the expense of others (cars). Some tweaking, especially automatic adaptation to the presence or absence of transit vehicles and recovery strategies to optimize for their passage, will help, but there is a limit to how much we can get out of the road system.

    Notably missing in both of these discussions is an ability for an operator to invoke or release a priority request. This is important for nearside stops where a car may needless hold an extended green cycle it cannot use (typically because of loading delays), but conversely would benefit from an advance on the next cycle to its benefit.

    LikeLike

14. Hi Steve,

    I’ve seen several times your suggestion for TTC to give operators the ability to request a Green light whenever they are ready to proceed. What other cities is this system used in? Does TTC make note of your suggestions?

    How does transit signal priority systems around the world compare/contrasts with Toronto? Apparently, Vancouver would use transit priority for its B-line express buses. Is that the same system used in Toronto too?

    Steve: I’m not sure which other cities have operator-activated signal priority, although the technology does exist for use by emergency vehicles. A problem in North America at least is that there appear to be a limited set of models and traffic control hardware/software with which to implement a priority scheme. Vancouver uses several different priority schemes as part of their B-line service including reserved lanes and queue jump lanes, not just traffic signals.

    An important distinction must be made between Toronto’s surface routes, especially the streetcar network, and express bus or LRT implementations elsewhere. First off, Toronto’s signals are closely spaced and operate on a comparatively short cycle time. Most but not all have associated transit stops (this has gradually changed as pedestrain crosswalks become fully signalled intersections without transit stops). Headways are frequent at most times of the day, and the probability that a transit vehicle in one direction or the other will be near an intersection is high.

    There is a big difference between clearing a signal to green for a bus or train that will speed through without stopping, and handling a situation with nearside stops in both directions. In the latter case, there are times when the transit priority needlessly holds the green for a transit vehicle, but times out before it has finished loading. It is unclear whether the fact that a vehicle is still waiting for an early green on the next cycle is actually registered by the signals, or if the waiting bus or streetcar is “forgotten”.

    Another problem in Toronto, and possibly with the implementation of the technology generally, is that each intersection is considered in isolation. Detection for oncoming vehicles can be placed no further from a signal than the next upstream controlled location, and this makes it nearly impossible to trigger “waves” for the passage of transit vehicles without stopping. For example, between Jarvis and Sherbourne on King, there is a signal at Frederick. Streetcars can be held here because they are detected only after leaving, say, Sherbourne westbound.

    A variation of this problem is on Harbourfront where a car leaving a farside stop will not be able to request priority at a nearby signal soon enough to be effective. A good example is eastbound between the stop east of Bathurst and the signal at Dan Leckie Way which often holds eastbound cars. This sort of problem is a major issue in the redesign of intersections for the Queen’s Quay West and East reconfigurations.

    Finally, at far too many locations, streetcars must wait for a dedicated transit green rather than sharing the through-traffic green phase. This arrangement gives more time for left/U turns, but penalizes transit service.

    Much greater sophistication in transit priority control is needed. I find it hard to believe that the traffic engineering and signalling fraternity worldwide has not figured out how to handle this sort of problem.

    LikeLike

15. “St. Clair uses single-ended cars, and the idea of crossing over to the opposing track raises the problem that you need a way to prevent collisions.”

    Rural highways have more traffic, higher speeds and fewer professional drivers then a streetcar ROW, but they allow crossing over to the opposing lane for passing.

    Steve: Yes, but people can pass more or less anywhere on a highway. Streetcars need crossovers to do this, and they would not be installed every few hundred metres along the route.

    LikeLike

16. The question about 28m streetcars on St. Clair might depend on which routes the Flexities will be phased in on. After all, *LRVs will exist in the TTC network for a long time after the first Flexity arrives.

    The Vancouver Olympic Line Flexity from STIB was pleasant to ride (albeit at barely above cycling pace due to the limitations of the pilot trackage) but I’m glad the TTC ones will be wider. All door boarding (as a free service) caused them to suck up huge queues in short order.

    As for the Nextbus link, I clicked it around 2 and noticed a huge gap both ways in service between Front St and Union Loop. Never mind, Union Loop is such a pleasant place to wait, right?

    Steve: It’s useful to turn on both routes 509 and 510 on Nextbus to see the degree to which route management on Spadina presumes that the 509s will fill in on Queen’s Quay. The problem for inexperienced riders, of course, is that they will pass up southbound 510s rather than riding down to Queen’s Quay and transferring, and people may wait an eternity at Union for a westbound Spadina car. This is a classic problem in TTC customer information. The “next car” signs presume that everything is working normally, but don’t tell people of an alternate strategy when things are askew.

    LikeLike

17. I drove up Spadina from Queens Quay at 12:30 today. The 510 cars were operating in 2 or 3 car convoys but there did not seem to be any major gaps in either direction. I left Queens Quay just after an CLRV and did not pass it until Harbord St. It made all the stops and picked up a lot of passengers and still kept up with me. Car traffic seemed a little heavier today than normal but not unusually so. The 1 minute 53 second headway is too close to the normal cycle time for lights north of Front St. I believe that Spadina will work better with slightly fewer cars when they get the longer ones. All door loading will speed up the service and they will not bunch up at signals. It will be impossible for 2 of them to load up at the same stop at the same time as ALRV’s do now.

    I came home along Eglinton from Caledonia to Matheson Blvd. It took at least two lights and some times three to get through every intersection. With the PRofW the LRT should be able to save 30 to 60 seconds at each light over the buses.

    LikeLike

18. Steve, your explanation really helped me understand why sometimes a streetcar can be held by a light when I thought it shouldn’t. I must say, besides the unexpected red light, service on 512 St Clair is satisfactory, and quite enjoyable to ride.

    LikeLike

19. Steve points out the challenge is “handling a situation with nearside stops in both directions. In the latter case, there are times when the transit priority needlessly holds the green for a transit vehicle, but times out before it has finished loading.”

    Absolutely. The usual maximum green extension in Toronto is 16 seconds, which is just not enough to load passengers at peak times at many stops downtown. In fact, at peak, transit priority should probably generate a RED SIGNAL not green. Parallel car traffic is going to be stopped in one direction anyway to permit boarding, so far better to stop it all and allow the side street traffic to clear, then switch to green when loading is over. Of course this applies only to big intersections like say Dundas and Bathurst that have lots of loading and lots of side street traffic. (Is this is a completely kooky suggestion? Probably.)

    In contrast, when I rode the 512 on Saturday afternoon, TSP worked impeccably. Though most lights were red on approach (a fluke I guess), the operator only had to slow down a little to permit it to cycle to green. This worked even where there was an advance left turn signal, though admittedly traffic was light. In all, I was really impressed by the service.

    Steve: I know that there have been adjustments made to the TSP at some locations. Possibly in the February data, when I receive it, this will show up in a reduction of holds at signals along the line. I have some comments about this in the next article, re week 4 of January, now in preparation.

    LikeLike

20. There’s the additional problem on streetcar routes of left-turning cars. At priority-enabled interesections, the streetcar is ready to go, but can’t because there’s a left-turning car ahead of it.

    The car can’t turn because the held green allows a constant stream of vehicles to come from the other direction.

    Under these circumstances, the light should allow the waiting car to turn left, for example by a flashing green phase or a green arrow. Alas, trailing turn phases are rarely used, and therefore rarely understood. (Carling and Richmond, in Ottawa, had a trailing turn phase in the ’90s. Only one I can think of in Ontario.)

    LikeLike

21. Ed says:
    March 2, 2010 at 2:52 pm

    “Under these circumstances, the light should allow the waiting car to turn left, for example by a flashing green phase or a green arrow. Alas, trailing turn phases are rarely used, and therefore rarely understood. (Carling and Richmond, in Ottawa, had a trailing turn phase in the ’90s. Only one I can think of in Ontario.)”

    There is one in Mississauga SB on Central Pkwy and Bloor. Mind you there is a separate left turn lane. I remember a couple of others but they are not common. The reason is that if too many cars are waiting for the turn then the queue extends into the left through lane and holds up through traffic.

    LikeLike

22. Actually, the left-turn signal is last in the sequence – which goes roughly as follows:

    1. Green for E-W (no left turn allowed from St. Clair)
    2. Left turn signal (in some cases) for vehicles on the N-S
    3. Green for N-S
    4. Left turn/U-turn signal for vehicles on St. Clair

    Steve: Any of the four phases can be called “last” depending on where you start in the cycle. I can’t think of an existing St. Clair intersection where phase 2 in your list actually exists because street geometry in the old part of the city does not permit N-S left turn queue lanes that would be served by such a phase. In any event, one issue from a traffic management perspective is the time needed for the transitions between phases (which effectively lengthens them) and the degree to which a green that transit can use (however it is presented) appears promptly when a streetcar approaches an intersection.

    The farside stop configuration tends to put transit vehicles “out of sync” with the wave of general traffic, presuming that the signals are timed for traffic in the direction of transit travel, and the traffic is moving freely enough for it to progress through multiple intersections. This effect has different implications for signals at transit stops and for signals between stops. The latter tend to be at minor intersections where shortening (or completely pre-empting) the N-S green would have a lesser effect than at, say, Dufferin Street.

    Green time extension for transit works well at locations where a streetcar initiates the process far enough from a signal that it is still in the portion of the E-W green where the transition to red can be delayed or pre-empted, and where the traffic management scheme acknowledges that some green will be taken from the N-S flow. The longer the approach area, the greater the probability that the streetcar will be detected early enough to influence the signal. However, Toronto has still not addressed the problem of signals located closely beyond a transit stop or strings of closely-spaced signals that should clear for what is really a single, continuous transit move.

    When transit vehicles miss the green because they are not detected soon enough, then waiting for the left turns becomes an additional annoying delay.

    LikeLike

23. Yes – my point was that there is no 1st or last – it’s a continuous cycle.

    I think there is an advanced green for Southbound on Dufferin at St. Clair.

    St. Clair has such a hodge-podge of intersection spacing and geometries – which makes optimizing movement a challenge from any perspective.

    There is a delayed green at Lakeshore and Leslie for Westbound traffic to turn south. This works fine because there was plenty of room for a long queueing lane – hence the vehicles waiting to turn do not block other vehicles proceeding straight (i.e. West.) That isn’t the case on most of St. Clair W.

    LikeLike

24. “Steve: St. Clair uses single-ended cars, and the idea of crossing over to the opposing track raises the problem that you need a way to prevent collisions. “

    Well, one normally does this on railways by using bidirectional signalling.

    I’m guessing the problem is that the blocks are too short to put a pair of crossovers, with room to hold a car between them and room to hold a car before and after them, within a single road block. Is that right? Because if the blocks are too short you’d end up with streetcars crossing cross streets on the “wrong track” which would confuse everyone and be dangerous.

    Steve: And you may have noticed that streetcar lines don’t have signal systems other than their limited interactions with traffic lights and electric switches. This is a net additional cost to make your proposal workable for comparatively rare events.

    —
    The information about the “transit priority” is interesting. It looks like Toronto is Doing It Wrong. Maybe they should get some advice from Seattle, which set up a system to allow their light rail vehicles to run down the length of Martin Luther King Jr. Way at speed without hitting any red lights unless something went wrong (emergency vehicles, people late getting out of crosswalks, delays at stations). But then again it sounds like the Roads department in Toronto is simply unwilling to really give transit priority, which may be the real problem.

    LikeLike

Comments are closed.