Interminable Waits at Kennedy (Update 3)

Updated Feb 27, 6:15 pm:

At today’s TTC meeting, we learned that they expect to have repairs at Kennedy completed by Friday. I will post more information about the derailment and the nature of the damage when I receive an electronic copy of the report on this incident.

Correction Feb 26, 11:00 pm:

This evening, I received the following email from Adam Giambrone, Chair of the TTC. It was addressed jointly to me and to Ed Drass whose column I cite later in this post.


I told Ed Drass yesterday that I understood the slow order was to be off by now BUT that I was going to check with Warren Bartram of TTC during a tour of the tunnels with CTV earlier this morning (2am-5am) to confirm. I actually watched the crews doing the repairs and I called him earlier today to confirm that the slow order was still in place and will be for another week or so.

The problem is that many of the bolts (I don’t know their technical name) that bolt the rail to ground were ripped up (some 150) and there is only so much that can be repaired in the 150 minutes they have most days to do the work.

Anyway, I usually qualify a statement of fact if I am not sure and I did so in this situation.

Adam Giambrone

I expect to get more details at the TTC meeting tomorrow and will post that info here.

This post has been revised in light of Adam Giambrone’s email.

Updated Feb 26:

This morning at 8:40 am, the backlog of trains from Kennedy stretched to Victoria Park Station, and the trip from there to the terminal took 21 minutes.

According to Ed Drass’ column in today’s Metro, Adam Giambrone was advised that the slow order on Kennedy crossover was lifted last week. This proved to be incorrect based on his email quoted above.

Original post:

Ever since the Kennedy Station derailment two weeks ago, service at the eastern terminal of the Danforth Subway has been glacial, especially at the end of the peak periods. As an example, I spent more than 15 minutes this morning getting from Victoria Park to Kennedy Station, and this has happened almost every day for the past two weeks. Looking on the bright side, the TTC has figured out how to operate a reasonable headway on the SRT even when it was in “manual” mode and we no longer creep from Kennedy to STC. The combination of these two delays made the term “rapid transit” quite a joke.

The problem at Kennedy arises from the slow order which has been on the crossover. Trains move over it at low speed while TTC staff watch carefully as the trains pick their way through the special work. Riding on trains, I can’t tell whether the roughness of the crossing is due to the very slow speed or the condition of the track. With luck, we will learn more at Wednesday’s Commission meeting when there will be a presentation on the derailment.

Meanwhile, capacity on the BD subway is badly constrained. In two previous posts, I talked about the physical limitations that subway line and terminal layouts place on the frequency of service.

How Often Can Subway Trains Run?

How Frequently Can We Run Subway Trains?

The minimum headway at a terminal controls the level of service on the rest of the line unless additional trains are inserted at short-turn points. Indeed, that is how the TTC plans to fit more trains onto the Yonge line in eight years or so with turnbacks at Finch (following a northerly extension beyond Steeles) and somewhere in Downsview (following the York U extension).

Just to review, here is the sequence of events at a terminal:

  • Signal turns green
  • Train guard initiates door closing and this completes
  • Train moves off from platform and eventually clears the crossover
  • Signal system determines that the crossover is clear and realigns the switches
  • Signal system displays clear for the incoming train
  • Incoming train starts up and crosses into the station
  • Signal system determines that the crossover is clear and realigns the switches

The two longest steps in this sequence are the train movements. Today, I timed trains at Kennedy, and it takes 80 seconds for a train to move from a stationary position either on a platform (departing) or from the last approach signal (arriving) through the crossover to a point where the junction is clear for another movement.

This means that 160 seconds (2 minutes, 40 seconds) are consumed simply for train movements. Add to this about 10 seconds for signal and switch system activity, and we are up to 170 seconds before any delays introduced by the readiness of crews to depart.

However, the scheduled headway on the BD line is 144 seconds (2 minutes, 24 seconds) in the morning peak. Quite clearly, it is impossible to operate this headway given the constraints at Kennedy, and a queue of trains builds up. This affects service on the entire line unless trains are inserted along the way to bring the headway back down to the scheduled level.

The TTC should seriously consider short-turning some trains. This could be done at Warden, provided that these trains crossed over to the westbound platform so that they did not block the eastbound flow. Yes, this would require careful management at Warden, but it would reduce the backlog at Kennedy and allow a reliably frequent service to operate on the rest of the line.

I say this with some trepidation because this scheme could also foul up the line just as badly as the current arrangement if it were not managed to ensure fast in, fast out turnarounds of the short-turning trains.

A further option, applicable only to the am peak, would be to send trains that would run out of service to Greenwood into the yard eastbound.

I write this not just as someone who is personally inconvenienced, but out of concern that a long-standing operational problem affecting service capacity has not been addressed.

After Wednesday’s update at the TTC, I will add to this item as appropriate.

27 thoughts on “Interminable Waits at Kennedy (Update 3)

  1. Delays at the terminus are a fact of life at Downsview, where it’s not unusual to wait 15 minutes to crawl up from Yorkdale. Strangely, the problem is worst at off-peak or near-peak times, like 10 AM or 3 PM; rush-hour traffic is nearly delay-free. I’ve often thought that the problem was an inefficient shift change, since the trains will stay in the station for several minutes before turning back southbound.

    Steve: The problem is that the trains have too much running time and get to the terminals very early. Even on the wider off-peak headways, there isn’t enough room to hold all the trains that are at the end of the line. On a surface route, this shows up as large groups of vehicles at terminals. In the subway, there’s only room for two.

    This situation will only get worse if the TTC tries to run more frequent service. The tighter the headway, the more closely aligned the schedule has to be to the actual running time conditions on the line.


  2. Thanks for the info Steve. This is definately becoming tiresome, and I too was wondering why they haven’t started short-turing trains at Warden to avoid the long queue of trains leading up to Kennedy.

    I would have no problem if every other train were a Warden-bound train, expecially if it means that once I’m on a train, it will actually move. Furthermore, eastbound ridership is low enough in the AM that the number of riders impacted would be very low. As for the PM rush, I still think it would make sense to run with two eastern terminals, because anything must be better than what we’ve been getting for the past two weeks.


  3. It’s not just Kennedy, but Finch, Downsview, and Kipling as well … but it has ALWAYS been like this, as far back as I can remember.

    The complexity of managing a turnback such as the one you suggest is almost as bad as running the Y — every train has to fit into its pre-determined pocket. Turnbacks at Warden will lead to delays similar to the ones experienced by southbound Spadina trains just north of St. Clair W. in the AM rush (as they wait for their spot with the short-turns pulling out of the St. Clair W. pocket track). A Warden turnback will be worse though because there is no pocket track, so it will affect westbound service.

    The TTC is really hung up on evenly spaced headways, so I don’t know if your proposal would work, although it was used at Eglinton Stn. for a while in the 70s. The problem has always been uneven arrivals at certain locations. That was the biggest problem at the Y as well. Even if trains were dispatched at uniform intervals from the terminals, they would not arrive at the Y with that same even spacing. The same thing would happen with a Warden turnback. It would require precise co-ordination of westbound and eastbound movements. It can be done, but the TTC has shown in the past that it simply cannot be bothered with the additional complexity of that kind of arrangement.

    Steve: Yes, the problems have always existed at terminals, and it’s one big reason why I have to suppress wild laughter whenever people talk about 90 second headways. As for a Warden turnback, the TTC would really have to get its act together and operate on the basis of first-come, first-served. We already know that they can’t push trains out of Kennedy more often than every four minutes or so, and there should be lots of gaps coming west for the short-turns to fit into. In the event that things are held up at Kennedy, then send two trains west from Warden in a row.

    Yes, it’s more difficult to manage, but it would provide better service.

    On a related note, the TTC really has got to start adjusting its schedules so that excessive terminal times are removed. We already have layover time sprinkled along the route at many hold points, and even with that, times at terminals are generous. This ain’t a bus route where four buses can lay over on a two minute headway.


  4. Steve, Out of curiosity, does Woodbine station have one of those pockets for the waiting trains talked about from St. Clair West? I have not been out on that section for a while, but it was a terminal at one time also. Just maybe the short turn can take effect here.

    Steve: Woodbine does not have a pocket track, just a crossover like Warden. The only places “pocket” turnbacks can be done in the east end are between Broadview and Chester, and at Greenwood Yard.


  5. Speaking of the Spadina line short turns, there is a recent update on the TTC Spadina subway extension site. Reading through the info, it appears as though the TTC will be using Steeles W as the alternate terminal to short turn service. However, the station will be apparently designed to properly accomodate short turns. I’m not sure what they can do other than have a cross over and pocket track north of the station, which is what they have at St Clair (and still causes some slow-downs for incoming southbound trains). So I’m not sure what exactly they will do…


  6. I wonder if the TTC could learn something from the San Francisco MUNI. The original configuration of their streetcar subway terminus used a double-crossover and was operated exactly like the TTC Subway. This arrangement was found to be completely inadequate and caused terrible congestion due to the service frequency.

    When the Embarcadaro extension was put in, the crossover was relegated to emergency use only. Instead the extended tail tracks included a huge pocket track with two routes in and out for turnbacks. LRV trains would drop off passengers on the right side of the centre-platform and immediately depart for the pocket track. The station and pocket track could therefore simultaneously hold four trains instead of only two. The additional length of the approach curve tracks between the station and the pocket track could conceivably hold at least two more trains for a crush-load total of six trains.

    I seem to remember the arrangement at Sheppard-Yonge Station planned to implement something similar to this using the northern tail track and current single-crossover. The tunnel structure clearly allows for a full double-crossover in the tail track area such that four trains could occupy the station and tail tracks at the same time. Unfortunately this line will likely never need such capacity, although a double-crossover limits the effectiveness of the arrangement.

    The only terminal with a centre pocket track is Finch Station, and that could only handle one extra train, although Downsview has the tunnel structure available for it. Hopefully any future extensions of the line will have at least a one-train pocket track from the beginning. The York U or Steeles/407 stop probably should have one for turning back half the service there on a regular basis.

    If we’ve already crossed the line where service frequency exceeds the turn-back capacity of the terminals then any little bit will help. Pocket-tailtrack turnbacks would shave critical seconds off the operation by removing the conflicts caused by the double-crossover. Multiple sequential pocket tracks, as in the MUNI example, take this a big step further. Expensive, yes, but how costly is a system that chokes on its own success?


  7. The only way the TTC will be able to satisfactorily operate a 90-second headway is if they can learn how to dilate time. An operation like that will require split-second timing — if they try it, the operators will be running out of the washrooms at Finch with their pants half down and toilet paper stuck to their shoes (and that’s even if they use step-back crews).

    Here’s how the TTC would spin a Warden turnback in their favor …

    “Trains approaching Kennedy are almost empty in the AM rush — we would inconvenience more passengers heading west from Kennedy by cutting the service in half, and potentially delaying the remaining outbound Kennedy trains while the short-turns reverse direction in the Warden westbound track. And, aside from all this, we might crash two trains into each other”.


  8. Example story:

    I needed to get from work to the UPS service desk at Steeles/Jane before the desk closed at 7:30pm. I left work at 6 and boarded a northbound 47C near Caledonia/Eglinton at 6:05. I rode it to Yorkdale Station and transfered to the subway. The backlog of trains northbound was truly unbelievable. I managed to catch the 107 Keele North bus from Downsview in time to get to UPS at 7:25pm.

    The trip took an hour and twenty minutes, but nearly half of this time was spent on the subway going TWO STOPS! It was a miracle that my trip wasn’t wasted. Not wanting to trust reliable connections of three bus routes led to my mistake of trusting the subway in the evening rush.

    (On a side note, there was only one person boarding the VIVA Orange bus at Downsview.)


  9. There also seems to be a slow order southbound between Eglinton and Davisville. It’s not nearly as bad as what you describe (maybe it adds a minute to the trip) but that stretch probably carries more passengers.

    If I remember correctly, there was a work zone there about six months ago, but after the workers disappeared the slow order stayed for one switch. Isn’t this exactly the sort of thing that shouldn’t happen with a “state of good repair” program?

    Steve: There is a special work replacement planned at the southbound entrance to Davisville Yard in this year’s capital budget. Why a major piece of special work is allowed to remain in a condition requiring a slow order for so long is baffling.


  10. With regards to Dave’s point, I think a turnback at Woodbine instead of Warden would be good. Warden is my home station and Mimmo is right, it is pretty much empty going eastbound in the a.m. rush. Turnbacks could be done at Woodbine, allowing for a much more significant impact to be had.

    Steve: A turnback that far west would result in too many trains being on that part of the line. One could argue that Victoria Park might be ideal, given the length of the queue, but that’s a farside crossover and would be harder to manage than the nearside crossover at Warden.


  11. Last September, I posted a comment in the blog section of the SubwayNow website about how only half the trains will turn back at Finch or possibly at Steeles in order to accomodate the increased headways needed south of there that will be possible with a new signalling system combined with such turnbacks.

    Their comment to my remarks were that what I was saying was untrue. The unidentified person added, “I believe that the TTC would maintain the current schedule for the Yonge Street Subway North extension.” My comment, based on technical details was “untrue”, but their belief is the truth.

    The other possibility to allow for reduced headways would be to implement a loop beyond the terminal station and have inbound passengers exit, have the train run through the loop, and board passengers on the outbound track. This would make the terminal station like any other station as far as not having a crossover bottleneck. However, I wonder about the cost of such an installation, not to mention the added issue of complexity if the line were to be extended (as Pittsburgh is currently doing with their LRT subway).

    Looking at the specs on the TTC’s website for the Spadina extension, I see that the desirable minimum curve radius is 750 metres, though the absolute minimum radius is 300 metres. I suspect that absolute minimum is restricted to short lengths, so a turnback loop would have to have at least a 750 metre radius. Such a turnback would likely be at least 10% longer than the circumference of a circle of that size, making it about 5.2 km long!

    I suspect that the per km cost of building such a single track structure is a little more than simply half the cost of building a double track line. Even assuming just half the cost of $180 M/km, we’re looking at over $450 million for just that loop!

    Steve: The SubwayNow folks are dreaming if they think that all of the service will run northbound beyond Finch or Steeles. If there is so much demand up there, then they need their own subway, or vastly improved GO service.


  12. When trains were regularly short-turned (or sent out of service at the end of rush-hour) at Eglinton, the travelling time between Lawrence (sometimes York Mills) and Eglinton was interminable, because the TTC insisted on fitting each train in it “appropriate” place. GRRR I hated that! What was normally a 2 minute ride between Lawrence and Eglinton stretched to as long as 20 minutes (I kid you not), as we waited and waited and waited at each and every signal.

    Steve, you say the plan to insert more trains on Y-U-S includes turn backs at Finch once the line is extended to north of Steeles. But this is still undecided, no?

    Steve: It is physically impossible to send all of the service to the north ends of the Yonge line on a significantly more frequent headway due to the inherent turnaround times involved with a facing crossover arrangement such as we have at Finch. If terminal operations are split by turning half of the service somewhere along the way, then the new terminals only have to handle a double headway which is unlikely to be much below 3’40” (assuming a base headway of 1’50”).

    As discussed elsewhere, it will be imperative that the “short turn” and “through” routes be operated independently, and that trains leaving the short-turn do so on a “first come, first served” basis. Total chaos will ensue otherwise with trains backed up both ways from the turnback point.

    The TTC did a fine job of sabotaging the integrated service back in 1966 by insisting on strict train sequencing. This was partly due to the way that the schedule was written with trains changing between routes and strict order being essential to making the whole thing work. If each route is treated as a seperate operation, then trains can be dispatched independently of what happens on the “other” service.

    I can hear the TTC now. Not having a strict alternation of trains will confuse passengers. In the name of avoiding such confusion (remember that the new Yonge line trains will have signs in every car), they would rather totally screw up the service. Am I too harsh? No, I have heard precisely this excuse from the TTC before, and they made the same mistake, initially, with the integrated weekend service where all trains ran via Museum.


  13. Hi Steve:-

    All of the reasons stated for being a difficult and/or costly fix to decrease turnaround times at terminals are all of the reasons why Transit City should be a single ended system from the get go. At heavily used terminal points, multiple track loops should be employed. Grade separation where short turns are required as a scheduled every day occurrance is a must; but for emergency turnbacks grade separation is not so much of a necessity; for as the line being blocked for the cars’ time taken to loop should ideally only occur when it is less than capacity service being held, otherwise there shouldn’t have been a need to loop.

    As LRTs can turn on a radius that can generally be shoehorned into their environment more readily than heavy rail’s requirement for much broader curves a generous amount of real-estate is needed to allow looping for subway trains. Certainly this is true with TTC sized vehicles, the clostrophobic NYC IRT cars can come closer to streetcar radii though. With this said, if the badly advised expansion into the hinterland of Vaughan becomes a reality, then maybe a farmer’s field or two could be expropriated for the space to build a loop.

    An upside to heavy rail loops would be simpler signalling and less costly special work trackwork. True the loop rails will require replacement on a more frequent basis than straight track, but the attention neeeded to keep them working is exponentially less than maintenance intensive switchwork requires now. I know that this is dreaming in technicolour, but I’m using your site to discuss ideals. Speaking of which, the ideal subway extension to Vaughan is LRT, n’est-ce pas?

    Boston’s Blue Line, at its Wonderland end, has a terminal whose capacity is normally more generous than its headways require, but by studying their use of extra track switches and extra long tail tracks one could find ideas there which if massaged and expanded upon could prove to be of value to the TTC. The major drawback to this is that by the time the tail tracks end, one has gone almost the distance to a next station, easily close to 1500 feet. I can only imagine the public clamouring to have a new station tacked onto the end of them.

    As to the thoughts of short turning trains to help with terminal capacity; yes it will help, but if the huge cost to build a subway has been spent, then if the end-points, which cost just as much per foot as the mid-points, have their capacities minimised along that strectch, thus limiting this piece of railroad to LRT capacities; maybe it should have been installed as the less costly, equally attractive LRT option in the first place? The short turn point’s terminal should be improved to accomodate the line’s needs if this then happens.



  14. To expand on my comments above regarding tail-track turnbacks:

    A vast improvement in operating efficiency may be acheived with the use of a three track/two platform station. The first track would be for arrivals only, the second for arrivals and/or departures, and the last for departures only. In other words, the first island platform would be easily signed as arrival and the second as departure, although nothing would stop someone from entering or exiting a train in the centre track from the unintended platform. The tailtrack area would first merge into two tracks and then into one. A computerised interlocking would easily be able to sequence trains into and out of this network on a first-in-first-out basis without any interference between trains. Incoming trains could then be stacked up quickly and the track network used only to its full capacity when struck by rush-period demand. At off-peak times the centre track could be used exclusively with doors opened on both sides.

    Another improvement to this would be to have the previous station or two use three tracks and double island platforms also such that the centre-track would be operated in the direction of demand only. The further up the line the track splits into three, the greater the number of trains that can accomodated and service stops without waiting. It also means that one broken-down train can always be diverted around. San Francisco’s BART uses such a flow-reversable third track and platforms between the Bay Tunnel and Downtown Oakland, although in their case this is to handle the volume of three overlapping routes through this section.

    The TTC should also be considering using side and centre-platforms together at two-track stations to allow for all-door loading/unloading on both sides of a train at once. This could be combined with loading-only/un-loading only signage and passenger flow as has been implemented on the Scarborough RT at Kennedy where the south side is loading only and the north side is un-loading only. Making the centre-platform the loading-only one would allow emergency double-crossovers to be practical for short-turns if space were not available for a trailing pocket track. One-way passageways are the life-blood of the Tube in London, England.

    Steve: All of these are valid design issues. The problem is what to do with existing infrastructure. Any changes to increase terminal capacity need to be factored into projects that address line capacity, not as an “oops” add-on after new signals are in place and new cars are ordered.


  15. For stations with a tailtrack like Finch, would it be possible to have (at peak periods only) the train come, drop passengers off, board a new crew at the back, pull in to the tailtrack, then let the new crew drive back onto the other platform to pick up passengers witht the old crew getting off? Would’nt that add an extra train’s capacity to the station and reduce waiting at the crossover? (Although combined loading times will probably be longer).

    Steve: A single tail track cannot sustain a headway under 3 minutes, and even such a quick turnaround is only possible with splitsecond timing. It’s the sort of thing a completely automatic train control system can do, but not operators on a day to day basis.

    The step-back crewing now used at terminals (most of the time) ensures that a new crew is available to take a train out as soon as the route is clear (in theory at least).


  16. Hi Steve and Calvin:-

    If memory serves, and sometimes my dinosauric brainpan may be a little dusty, but I believe that if the TTC were to employ a heavy rail loop at a terminal you’d probably find that they could get away with somewhere between a 150 to 200 foot radius. Where my memory comes in here is the curve at the south end of Greenwood Yard. I’m thinking it has a main radius of approximately 150 feet. Of course the wider the radius the less wear and tear on wheels and rails but the tradeoff is real-estate. Streetcars on the other hand, can turn on less than 40 feet if needs be, but most loops are in the 55 foot radius range. All curves, with the exception of extremely broad, almost straight turns, have a length of spiral or easement at their entrance and exit thus increasing the amount of space required for the overall installation.

    Too, if my memory is in gear here, the curve SB south of King is the sharpest on the system on a passenger carrying piece of track and is in the 300 foot range. I’m sure that what you read Calvin in regards to the TTC’s quoted minimums are their ideals for new construction and main tracks. Loops if employed would probably be considered non-revenue, much slower and therefore subject to different design parameters. Again, as far as designing a new subway line goes, no curves is the ideal way to go, but of course practically that cannot be the case. Therefore then, the broadest possible curve one can employ is the next design consideration, main reason being that line speeds could be compromised the tighter a mainline curve is. Secondarily rail replacement frequency and the subsequent costs incurred because of it are other factors necessary for designers to keep in mind.

    Going beyond and extending a line that ends in a loop would be an interesting exercise and what jumped up and bit me was Park Street Station on the Boston Streetcar Subway. Here the inner track turns back beyond the platforms, but the outer tracks follow around it then rejoin each other when past the loop. Inner track cars/trains use their right hand side doors to offload at Park Street before looping and then pick-up passengers after looping and stopping at the outbound platform, while outer track cars/trains share the same platforms on the opposite side and use their left hand doors. Heavy rail could do it too for only a single track tunnel width would be required on either side of the loop to allow extension of the route.

    Only money can solve these problems, but as someone else said in an earlier comment, if the money is going to be spent building heavy rail anyhow, then don’t choke it at its terminals. Somehow build them such that they can allow trains to flow in and out almost as quickly as if they were on-line stations. We definitely need some out of the box(structure) thinking here!

    I was just about to close off and I recalled the Market-Frankford Subway/Elevated Line in Philly. Both ends of the line turn back in loops but on-line turn backs are cross-overs. I don’t recall the layout of the Frankford end, but the 69th Street end of the Market Street portion is a double track loop and the yard leads are grade separated, continue on past the platform and go underneath both tracks of the loop. Photo image #67993 posted on the web site shows this loop. TTC would have to have a broader radius though I think because our cars are somewhat longer than Phillys.



  17. I hope the new trains being ordered include side destination signs next to the doors (on the outside of the train). Otherwise, how will passengers know if they’re boarding a short-turn? The “Gong Show” signs don’t work anymore.

    Regarding short-turns with strict pre-determined ordering vs. first-come, first-go on a possible Warden turnback, the TTC actually did experience both approaches back in the 60s.

    In 1966, once a delay passed the 15-minute mark, trains WERE allowed to fall out of their proper batting order. This, of course, screwed up the service pattern and the rigid alternating destination scheme went out the window. The side effect of first-come first-go operation during a major delay almost killed the integrated trial after just one week.

    The incident, which was highly publicized at the time, involved about 100 Canada Post workers who were physically stranded on the platform at Union waiting for northbound Keele-bound trains when the subway closed. Six Woodbine trains passed them in a row starting at 1:30am. This was the result of an earlier delay which led to a huge gap in Woodbine->Eglinton service. Upon arriving at Eglinton, those trains were supposed to switch to the Eglinton->Keele route, but they never got there.

    So, if passengers expect alternating short-turns, or whatever the case may be, they could get really pissed off if, let’s say, 4 Warden short-turns showed up in a row.

    Steve: The situation at Kennedy is an emergency, and under these circumstances the TTC wouldn’t and shouldn’t be trying to schedule anything. The choice is simple:

    If there is a gap available westbound from Kennedy, turn a train into it and let the next one go through. If the eastbound track is blocked by a train sitting on the platform at Warden in a queue, then turn whatever shows up eastbound at least until that train moves off.

    Under ideal circumstances, the headway westbound from Kennedy would be double the normal value, and it would be fairly easy to insert short turns. This only applies to peak period operations, and the stranded posties scenario does not apply.


  18. Lol this is the same problem I hear every where I go. Nice to see Toronto is also dealing with it.

    One thing I have thought about potentially is [to] create an extended track behind the last station in the line. This might be easier on an LRT System to accomplish but it might work on a subway as well. Anyways that last point would be where the ‘rush’ hour cars are held.

    What would happen is that in the morning the cars are taken out of the rail yard, where ever that is located. Then they are put into service as usual. As the morning rush hour starts to die down and these cars need to be taken out of service, they will be stored at the end of the line, behind the last station. Then when the evening rush hour hits they bring them back into service, and do the same thing as the evening rush hour dies down. Then once the system is closed down for the night, the last step would be bringing the ‘rush’ hour cars into the rail yards.

    When Whitehorn Station was the Northern Terminus to the NE line it did this and Dalhousie Station currently does it as well. McKnight-Westwinds does not do this and you are starting to now see the system crawl in the early morning hours as trains are being shuffled between McKnight-Westwinds and Anderson Storage Garage. Also many commuters in the South have been complaining that they get to many out of service trains before they can actually get an operation train in the morning.

    Steve: There are roughly twice as many trains on the BD line in the peak than off peak (42 vs 20). Even if half of the peak trains were stored at each end, this would require room for 11 trains at each end of the line. Operationally, the issue is that schedules need to address terminal congestion at the end of the peak when trains are at times arriving faster than the scheduled departures. This is caused by the transition in the number of trains in service, the move from peak to offpeak scheduled running times and the limitations of the signal system on dispatching trains on very close headways.


  19. Terminal modifications, signal replacement, and platform lengthening and widening will significantly increase capacity on the Yonge and Bloor subway lines, and there is no doubt that they are necessary. Signal replacement and the new subway trains alone offer a substantial increase in capacity (56% for reducing peak headways from 141 to 90 seconds; another 8% for new articulated trains that fill the entire platform). However, these modifications are expensive, and in 10-20 years ridership growth will again lead to overcrowding. It would probably make more sense to spend the money on new capacity, particularly a downtown relief line.

    Steve: I don’t believe that a 90 second headway is actually possible on the Yonge line given its station spacing and configuration. It might be sustained for a brief period over a section of the line, but the smallest disruption would bring everything to a halt. I agree that if we really do need this much capacity, what we need is more lines. The debate, then, is how to provide that capacity.

    We seem to put far too much faith in a technology “solution” that nobody has explained, credibly, will work. If we get to the day when that 90 second headway is desperately needed and it doesn’t work, we sill suffer for years waiting for an alternative.


  20. In light of Adam Giambrone’s email regarding the situation at Kennedy, it does not make sense why they would not shut down the portion of the line for a much longer portion of the day so more work can be done to rehabilitate the track. i.e. have skeletal service to Kennedy only during rush hours and ensure the track is worked on at all other times. It’s a lot better than days and days of delays while they can only work on the track for 3 hours at a time. That and the special provisions for turning back trains would have helped to put an earlier stop to this situation rather than dragging it on.

    Okay I admit I live close to and use mostly the Yonge line, so some people may be a little pissed towards my opinion.

    Steve: That mess at Kennedy makes shuttle buses look attractive by comparison, and if we could get normal operation back sooner, it would be worth the annoyance. Also, the problems with maintaining headway affect the entire BD line. This is not just an issue for people riding east of Warden.


  21. Dennis Rankin wrote, “Loops if employed would probably be considered non-revenue, much slower and therefore subject to different design parameters.”

    Not if the loop were part of a continuous overall revenue operation, which is what we are talking about here. The need to allow a train to pass once every 90 to 110 seconds through the loop to match that sort of headway on the rest of the line means that the radius of the curve has to be subject to more or less the same specs on the rest of the line.

    The only leeway for this is that since passengers will not likely be on board, the level of jostling around as the train enters and leaves the curve is only subject to what the crew can take, which would likely be higher than what some passenger busy reading or taking to someone else would be able to take.

    On the subject of extending a line ending a loop, I mentioned Pittsburgh. When the “subway” portion of the LRT opened, there was some PCCs in service, so the Gateway Center station was built with a loop (and a low platform at one end). The downtown portion of the line is more or less on an east-west orientation. The inbound trains are on the north track that parallels the south track until about half way down the platform at Gateway Centre which is on the south track side. Here the north track curves to the north to enter the loop, then curves around to the west, south, then to the east where it enters the single-platform Gateway Centre station. (There are a few photos of this on The new extension will be running north from the current end of the line, so I suspect that the connections will be on the loop for each direction where the track is aligned north-south. I have not heard how or if a second platform will be built for Gateway Center.


  22. in order to add more capacity to the Subway system on all lines going in and out of downtown, we need to return to interlining. Its as simple as that. Back in the 60’s and late 50’s people started to realize that demand was far outpacing capacity on the bloor danforth and yonge – university lines, so they broke up into 3 lines, the yonge – university – danforth , bloor- danforth, and yonge univeristy bloor lines. It took the stress of yonge and still allowed people to get to their destination without having to transfer. My point here is that the only way to get rid of the capacity issues, we can put those so called gaps to good use and send every other train downtown, it allows for short turns on the bloor line, while still keeping a direct route downtown available during short turns. Anyways, sorry for going off topic but I wanted to voice my opinion seeing as you guys are talking about headways and building a ” whole other line ” to deal with the capacity issues in relation to headways.

    Steve: I don’t know what city you were living in in the 50’s, but the Bloor-Danforth line didn’t even open until February 1966. It wasn’t a question of breaking the system into three pieces. When BD opened, there was a trial integration of three services: Eglinton to Woodbine, Eglinton to Keele and Woodbine to Keele. For reasons that have been extensively discussed here already, this arrangement, as operated by the TTC, did not work, and they changed to the separate YU and BD lines we have today in September 1966.

    We now have the Spadina leg coming in at St. George and, trust me, it is not physically possible given the constraints of the signal system and of TTC’s operating style to concoct an integrated service that includes the Spadina line and provides headways worth waiting for on an integrated service. [Please don’t try to argue this issue as we have done it to death already in previous threads.]

    We keep hearing about closer headways on YUS, but nobody talks about related issues:

    The number of new subway cars now on order is not sufficient to run the entire Yonge line, especially if it is extended to York U. More cars are needed, but there is no provision for them in the capital budget. Capacity improvements depending on the new fleet will be constrained until the fleet is expanded.
    There is no provision for additional fleet (or carhouse space) to get the headway down below two minutes.
    The BD line is incapable of absorbing additional outbound transfer load in the PM peak that will be generated by improved service on YUS.

    I believe that people who think that we will magically solve the capacity problem simply by cramming more trains onto the Yonge line are (a) missing a collection of significant, unbudgeted expenses, and (b) ignoring the very real limitations on the BD line which will see no significant improvement for a decade or more.

    Additional capacity into the core is needed. The question is how to provide it, and on a related note, where are the passengers originating.


  23. I had the experience of taking a train from Kipling to St. George this morning.

    There was an H4 train sitting by itself at Kipling, and I had a fair amount of time to get on.

    By High Park, it was full rush hour load. By Ossington and Christie, the majority of people waiting on the platform couldn’t get on.

    Is this a result of all the trains being stacked up at Kennedy? I’m not a regular rider of the subway at this hour, but I don’t recall too-full-to-board trains on Bloor-Danforth in the past.

    Then the southbound train at St. George tried to close its doors while large numbers of waiting people were still trying to board, but that’s a separate topic.

    Steve: If you had a single train languishing on the platform at Kipling, this indicates that the headway arriving outbound is worse than it should be, probably because of the Kennedy situation, but also possibly due to other problems on the line.

    I am waiting until next week when all is supposed to be back to normal at Kennedy to see how the line behaves and the level of crowding with no extraneous issues to affect operations.


  24. Hi Steve and Calvin:-

    I’m sorry if I misunderstood that you meant in-service track Calvin. I was only addressing a terminal operation whereby the trains would scoot around the loop only after off-loading. If too the loop were double tracked then capacity of the loop increases by about an extra 80%, allowing for signal proving times to throw and clear the switches, thus a not 100% increase. But this still means that the terminal can handle more trains than the line could feed it.

    As to the time taken to go around a sharp loop versus a broad one, it would be interesting to sample a variety of radii via computer or even with prehistoric pen and paper mathematical modelling and see the length of time taken to leave the in-bound platform empty and return to the out-bound platform ready to load. I’m going to speculate that all will take approximately the same amount of time since a sharper loop with slower speeds is also a shorter run as compared to a higher speed allowed over a much broader curve radius but requiring a greater distance to travel; therefore any loop size would be a positive impact on terminal times especially if two tracks were available.

    If the facing track switch for this ideal terminal was a generously long one, thus allowing a decent approach speed, and was positioned a minimum of one train length ahead of a double sided off-loading platform and the loop length allowed for at least two trains on each track, then it would be unlikely that few if any hold-ups attributable to terminal operations would occur; barring of course some sort of catastrophy occurring. Similarily the loading platform should be double sided and if the trailing switch beyond it was likewise a generous length, then trains leaving the station would be able to match acceleration rates almost the same as any of the on-line stations.

    If while designing and building such an ideal terminal it was thought that the line may be extended beyond that point then all work should allow for that eventuality. If the central two tracks were designated for the potentially through route, then beyond the platforms another six track switches would route trains on to and off of a pair of tail tracks initially, but could be considered the beginning of the next extension. Of course grade separation would be a must here with the loop and tails going over and/or under each other (doesn’t matter which way either goes unless there is some geographic feature beyond the station that may dictate how the extended line needs to fit into its environment). Initially, as tail tracks, a much greater flexibility of operation and temporary storage to this the ultimate heavy rail end-of-line could be realised.

    Put it all together and a terminal that can handle more trains than the line can run safely spaced would give a fantastic extra bit of leeway benefitting the whole route. Whew, I always knew I could spend someone else’s money quite quickly and easily!


    Steve: Can we please end the discussion about terminal loops here? I think that this is a complete non-starter given the geometric constraints at existing terminals, and I am far from convinced that it is a viable option even for future extensions.


  25. If the TTC were serious about increasing capacity, each terminus should have three tail tracks. Now, this makes for a messy switch (it is different from the Chester-Broadview case since the far-left side track in the 3-track segment can switch over to the right side track in the 2-track segment… Chester-Broadview requires going via the pocket track between the two), but it allows for much better capacity issues than today at terminals. Ideally, terminal station platforms absolutely should not be islands. The TTC got it right the first time with station layouts in Keele and Woodbine as termini (even though they didn’t use tail tracks for turn-backs). The side-platform layout dedicates one track for detraining and one the other for boarding, and avoids the crowds in opposite directions fighting for space on the same platform (the same problem is also very pronouced on Yonge’s platform and the St.George platforms, even though they aren’t termini). The train pulls into arrivals, unloads, and leaves into one of the three tail tracks. The crew does its turn-around-related duties while in the tail track (the tail track area should include a provision for crew changes at the back, using the front door that normally is for emergencies), and then pulls out into the departure track. This can accomodate much tighter headways with a properly designed switch (they are more space-consuming length-wise (no difference in width), but they don’t require the slowdown that standard TTC cross-overs experience), instead of the switches terminals are stuck with today. This is partly a technological solution since it involves more recent switch designs than currently employed, but otherwise this is just a common sense solution, and highlights the management shortfalls at the TTC.

    Steve: Woodbine and Keele are designed as line stations, not terminals, because they were only used that way for two years. Neither of them had provision for run-through operation to a far-end turnback, and there was never any intent for this type of operation. The only place, actually, where this is possible on the BD line is at Victoria Park where there is a farside crossover.

    As I said to an earlier post, enough of this discussion of terminal geometry. Expanding or revamping existing terminals would be very difficult and there are other ways to deal with this problem if and when lines are extended.


  26. The slow order through Kennedy appears to be back on, as the train I was on this morning crawled through the cross-over into the station. The queue wasn’t as bad as it was a few weeks ago, but it was certainly slow from Warden to Kennedy. The outbound train from Kennedy also proceeded out very slowly.

    Steve: My trips through Kennedy today both in the am and pm were at normal speed.


  27. Paris and Mexico City metros have many 3 track termini. This allows trains to alternately reverse in the platforms using the centre track, or reverse in a siding beyond the platform using the outer tracks. This allow for frequencies of less than 2 minutes to be operated.

    The Moscow Metro reverses at all but two termini in pocket tracks, with drivers at both ends for the reversing manouvre. This combined with short dwell times in the arrival platform (i.e trains are not checked for passengers still onboard) allows for 95sec frequencies on the busier lines.

    Although I don’t know of any examples, it would also be possible to build a grade seperated crossover at a two track terminus instead of the usual double crossover. This could easily allow 90sec frequencies (even if the rest of the line can’t!). It would obviously be considerably more expensive to build!

    Steve: Yes, if the termini are built for it, you can get the headway down below 2 minutes. However, one wrinkle to keep in mind is train length. Systems with shorter trains need less time for them to pass through the junctions.

    Also, signal design can foul up turnaround times if it’s not done properly. At Kennedy, if there is a train stored beyond the station, there is a speed restriction on trains entering on the same track and this slows operations, not much, but enough to make a difference on a tight headway.


Comments are closed.