Automatic Train Operation, Capacity, Night Service and Clean Stations

In a truly breathtaking display mixing unrelated subjects in one press conference, Howard Moscoe has embraced Automatic Train Operation (ATO) as the salvation of much that is wrong with the TTC.  A chicken in every pot (provided that you have a Metropass) will follow soon after.

You can read about Moscoe’s scheme on the Star website: TTC Eyes Driverless Subway.

In brief, our fearless leader claims:

  • ATO will increase subway capacity by 40% at a fraction of the cost of building a new line.
  • ATO will permit bidirectional operation on the same track which will permit maintenance to occur on one rail while the other stays in service for overnight operation
  • ATO will allow one-person train crews freeing up staff to be used as “Station Managers”

As someone involved personally and professionally with Information Technology for much of my life (my earliest computer program of any significance dates to 1963), I’ve seen a lot of huckters for technology, for “solutions” looking for a problem.  Typically, if something isn’t marketable on its own, people will lard on all sorts of additional claims and benefits.  By the time the product is actually delivered, you usually find that these claims were misleading, or that there are many add-ons required to actually achieve those ends.

I smell exactly the same pattern at work here.  After we turn off the hype, there are three basic requirements Howard Moscoe is trying to address:

  • Increased subway capacity
  • All night service
  • Much improved station management (cleanliness, working escalators, etc)

The problem arises when these are bundled together as a package.  For those of you who don’t want to plough through the technical details below, I will state my bottom line right up front:

We should not have to spend $750-million and wait at least ten years just so that tattered, out-of-date service change notices will be taken down, spilled food and drink will be cleaned up, and escalators can be restarted in less than a day.  If the TTC wants to improve the look of and service level at its stations, this needs to be done now, and it has absolutely nothing to do with automated train operation.

Casual readers can stop here, but for the detailed evisceration of this proposal …

Running Trains Closer Together

I have written on this elsewhere, but since we are discussing Automated Train Operation, I should explain what it allows compared with the present system.

Train spacing is now controlled by fixed-length “blocks” between each signal.  Generally speaking, there are always two red signals behind a train, and this effectively limits the degree to which trains can “close up” on each other.  TTC operating practice for some years has been that a single red signal means “stop and stay” where previously in meant “stop and proceed with care” (this was before the Russell Hill accident).  A double red is always a “stop and stay”.

The old rules allowed trains at congested points like Bloor Station to creep up on each other, and in fact I regularly saw the tail end of one train disappearing out of the north end of the station while the next train pulled in at the south end.  This allowed trains to take a long time loading passengers while maintaining a close scheduled headway.

This type of operation will be vital for close headways.  A computer system can monitor train locations at a much finer accuracy than the old signal blocks.  The “moving block” system takes into account both the spacing between trains and their speed, but does not depend on fixed signal locations.  This is the system used on the SRT.  As you can see, it has been around for a long time.

However, problems arise wherever train movements might conflict such as terminals or turnback locations.  There are physical contraints due to the need for slow operation through switches and the geometry of the track layouts.  In practice, these are the controlling factors on the existing lines.

Adding 40 Percent To Subway Capacity

There are three ways to add capacity to any transit route:

  • Run larger vehicles
  • Run more vehicles -or-
  • Run existing vehicles faster

The new subway cars (“Silver Snails” is my nickname pending something official) will increase capacity by about 11 percent because additional space is available in the gangways through the trains.   This leaves about 26 percent to be found through closer headways.  (Confused?  The compound effect of an 11 percent increase and a 26 percent increase is almost 40 percent.)

Current peak service on the Yonge and Bloor lines is roughly one train every 140 seconds, or about 25.7 trains/hour.  We need to increase this to 32.4 trains/hour to achieve a 26 percent increase.

However, existing terminals at Finch, Downsview, Kipling and Kennedy cannot sustain this level of service (one train every 111 seconds).  As I have discussed previously here, the current 140 second headway is very close to the practical minimum (about 135 seconds).  This is a function of track layout and train length, and, to a much lower extent, the reaction time of operators when signals change.  ATO won’t get us anywhere near 111 seconds.

If intermediate short-turn locations are provided, then the headway at any terminal is widened.  However, this assumes the existence of a turnback that can be operated on a razor-sharp basis to pull every other train out of the outbound stream and re-insert it to the inbound one. 

Vancouver’s Skytrain did some very clever stuff during Expo, but had the advantage that trains were very short (they occupied the interchanges for less time than a Toronto subway train would need), and the operation was completely automated eliminating the need for an operator to change ends on the train.  It was impressive, but doing that sort of thing on our subway would be challenging under even ideal conditions.

Next, the current order for subway cars will not even completely replace the fleet now in use on the Yonge line, let alone the Bloor line.  If we are to achieve the capacity increase across both lines, we will need many more new cars.

One optional way of reducing fleet requirements is to change the speed of operation.  For a few years in the late 1960s, the Bloor-Danforth line ran in “high rate” where the acceleration and top speed of trains was set higher than today.  This did not work out for the following reasons:

  • The then-current H-1 subway car fleet had a technical problem with the motors which failed at operating speeds around 45 mph.
  • Higher speed operation requires that the trackbed be in better condition to preserve stability, and this costs extra money.
  • Most of the system had stations close together, and the benefits of high-rate operation were limited

The H-1 cars are long retired, and we now have stations 2 km apart.  The last time I raised this option at the TTC (as part of the debate on a subway car purchase), the reaction was nearly apopletic:  running faster trains meant buying fewer new ones, and it is Toronto’s job to keep the plant in Thunder Bay running.  Although the TTC agreed to take this option into account for future fleet planning, this never happened.

By now, “low rate” is the only thing people know, and it is quite possible that the newly-ordered Bombardier cars cannot operate in “high rate”, making this all a moot point.  Getting more capacity by running trains faster is probably not a viable option, and in any event, until you deal with the terminal layouts, all “high rate” operation would do is to speed up the service and reduce future fleet requirements with, possibly, some offsetting maintenance and energy cost impacts.

Finally, if we add 40 percent to the capacity of the subway, we also add 40 percent to the pedestrian traffic through the stations.  Several of our major stations would not be able to absorb such an impact because of choke points in stairways and escalators, and minor delays would produce badly congested stations much more quickly than today.

Station Managers

Moscoe raises one-person crews in connection with his plan to provide “Station Managers”.  What is he thinking?  If he wants to have Station Managers, just hire them.  We should not have to wait until 2015 or later to have someone who will wander from station to station ensuring that escalators are running, that obsolete notices are torn down, that food spills are cleaned up, that garbage is cleaned away.

Just having someone to spot a problem is not enough:  if an escalator is not working, who restarts it?  Is there a mechanic on duty to check for problems and decide whether it requires repairs?  If a platform needs to be cleaned, is there a janitor?  What tasks does a Station Manager do on their own, and for what do they need other staff on call?

At any given time there are vastly more staff acting as conductors (“guards”) on trains than would be required to provide coverage of one Manager per five or six stations.

This is short-sighted budget manipulation at its worst: making major capital spending decisions based on a minor headcount avoidance (that’s HR-speak for “not hiring more staff”).

One Person Train Operation 

If we move to one-person crews with ATO, the doors will be monitored from the front of the train, possibly with the assistance of one of the thousands of new video cameras the TTC plans to install.

One person crews have one huge drawback for fast turnarounds (see discussion about line capacity above).  At a location where there is no step-back crewing (crews dropping back to a following train at terminals), then an operator will have to walk the entire length of a train to reverse direction.  This will make turnback operations at pockets rather tedious, and dreams of well-oiled, minimum headway operations will go up in smoke.

[An aside:  The new trains will have guards at the back of the train.  I found this rather amusing considering that for years the TTC insisted they be close to the centre of trains for safety reasons (shortest possible view along the platform in either direction).  Safety seems to be an elastic consideration at the TTC, stretched to suit the current policy directions.]

Bidirectional Operation

Yes, ATO can do this at a relatively small cost because we no longer require track circuits, relays, logic systems to be designed to deal with the idea that “ahead” and “behind” can be reversed.  It’s a software change, not a hardware change.

However, this only solves one problem — how to dodge around a maintenance crew or other source of line blockage.  It does not address a very basic problem:  the power supply system is not designed to be shut off in only one direction.  If you cut power from Bloor to Union, you do so on both rails.  Since power is commonly turned off when work proceeds at track level, this will prevent bidirectional operation on the track that’s not being worked on.

Changing the power supply so that each rail can be powered independently is a very large job.  Moreover, there are still certain types of maintenance that require access to both tracks at once or where the presence of trains on the “other” track would be dangerous for maintenance workers.

All Night Service

We are told that bidirectional operation is essential to allow all night service to co-exist with maintenance.  I have already discussed some of the operational problems of this arrangement above, but that’s not all.

The frequency of service will be dictated by the largest gap between points where single-track operation is required.  For example, there is about 4km between the pocket track north of St. Clair West and the next crossover opportunity at Lawrence West.  This means that either we are going to bunch trains together to run a convoy through the single track section, or the headway must be wide enough to avoid conflicts.  This is the same problem single-track railways face all over the world.

Are we going to operate all of the YUS and BD lines, or just truncated versions?  Are we going to close selective stations?  Will all-night bus services run into the stations or load at street entrances?  How many station staff, security, janitorial, maintenance and other workers are required to keep these stations open and the trains running? 

If trains ran, say, every 10 minutes on Bloor-Danforth, this would require 10 trains, or less if some stations closed.  If we have one person crews, there will be far more people operating the system that driving the trains.  These will eat up money that could be used to simply operate a far better surface night bus network. 

Most importantly, what do we do about surface operations?  Today, the night buses run fairly often, although we could argue for better service.  They stop at many locations between subway stations.  How attractive would all-night service be if it ran every 10 to 15 minutes and all of the minor stations were closed? 


This whole scheme is a non-starter.  If the TTC wants to move to Automated Train Control, then do it for its own sake and justify the cost.  Some new technology is essential anyhow to replace the aging equipment dating back half a century to the opening of the Yonge line.  This has been planned as a long, staged implementation as various generations of original signal equipment wear out.

But don’t piggyback a bunch of phony arguments about night service and station management onto the project so that it can be accelerated at great cost to cover the entire system.

This proposal is very badly framed, and I am astounded that Howard Moscoe hasn’t got the sense to smell the steaming mound of cow dung.

31 thoughts on “Automatic Train Operation, Capacity, Night Service and Clean Stations

  1. Will the new trains coming in 2009 only have one operator in them, or still like the old setup?

    Steve:  There will still be a guard, but in the cab at the back of the train.  There will be no cabs in the middle of the trains. 


  2. Right on about the effects of an added 40% pedestrian traffic in the stations.  The connection between the YUS and BD lines at Yonge/Bloor can be awful when it’s packed!


  3. I don’t know if I’m remembering this correctly, but it strikes me that if the subway stopped precisely at the extreme end of the platform, then a 7th car could potentially be added to each train.  Of course I would have to assume that extending the station platforms to accomodate longer trains would be astronomical in cost.

    Steve:  This topic came up during debates on the unit-train designs of the new subway cars.  The platforms are 500 feet long, and a six-car train is 450 feet (6 x 75).  A seventh car would not quite fit.  If we wanted a short car in the middle of the consist, this would be an oddball car of a special design depending on how the underfloor gear was arranged.  Probably the best approach would be to make it a trailer.  However, this also requires that trains stop quite precisely in stations.  This is theoretically possible, but it is dependent on a completely new train control system, and we’re back to where we started. 


  4. Steve wrote:

    “The new subway cars (”Silver Snails” is my nickname pending something official)…”

    The TTC’s Oct 13, 2006 Media Release notes a name for the new Bombardier Subway cars has been chosen: The Toronto Rocket!!!

    click here for press release

    I thought the Rocket was the Express bus… I’m really confused now!

    Steve:  Innovative naming has never been the TTC’s strong point.  To me they will still be the Silver Snails — too good a name to pass up.


  5. A better option would be to replace door operators with user operated doors, much like the LRT in Calgary.  Maybe they could just have the doors operated by the driver, but install pressure and laser sensors.

    Calgary uses both of these, and it has worked quite well.  The only problem was on the old trains; when they opened, people got trapped behind the door (I have had this displeasure).  No one was injured, it was just uncomfortable for the few seconds the train doors were open.  The new trains, whose doors are like the TTC Subway doors, do not have this problem.

    Steve:  The volume of passengers on the subway is such that it’s much faster to just open all of the doors rather than having the passengers activate them.

    Also the TTC should consider a honour biased system for fares.  It would be a lot cheaper than having people show their fare upfront.  In Calgary you just enter the train with a proof of fare payment (A bus pass or a ticket).  Then Transit Cops do random checks to make sure everyone has paid their fares.  If you have not paid you get a fine (which makes up for all the people who were not caught).  I will let this site explain:

    Steve:  The TTC does not seem to understand a basic principle:  Vehicles can sit at stops loading everyone through the front doors, or they can use all-door loading and move off with better-utilized interior space.  Collecting and inspecting fares, no matter how you do it, costs money and Transit Cops are part of the cost of doing business.  It’s amazing how the TTC can grouse about streetcars stuck in traffic, but won’t deal with loading delays. 


  6. I am astounded that the TTC has not made the Spadina Car POP.  The silly statistics in the Globe and Mail attack on Spadina ROW may be even remotely accurate only because of delays at the front door.  The Spadina Car is very reliable and regular.  POP would be a great improvement.


  7. Steve: The volume of passengers on the subway is such that it’s much faster to just open all of the doors rather than having the passengers activate them.

    Aman: Fair enough. Lets use my second suggestion, the driver opens up all the doors. To make sure no one gets stuck in the doors, place lasers and pressure sensors on the doors.

    If the TTC is trying to save money, this is the way to go. Instead of having one person operating the train and another operating the doors, you will just have one person driving the train. This could save the city a lot of money, and still keep people feeling safe.

    Steve: The TTC does not seem to understand a basic principle: Vehicles can sit at stops loading everyone through the front doors, or they can use all-door loading and move off with better-utilized interior space.

    Aman: Am I understanding you correctly? Does the TTC really make people get on and off by only the front doors? That is pathetic.

    Steve:  People all have to enter by the front door to pay their fare or show a pass/transfer except at major stops on the Queen route (Proof of Payment system) or at stops where they place staff on the street/platform to check fares at the rear door.

    Steve: Collecting and inspecting fares, no matter how you do it, costs money and Transit Cops are part of the cost of doing business.

    Aman: Since Toronto already has transit cops, it might be easier to just give them an extra job to do.  The real reason for supporting such a move, is to deal with backlog of people trying to get on the trains.

    Steve: It’s amazing how the TTC can grouse about streetcars stuck in traffic, but won’t deal with loading delays.

    Aman: It’s amazing how Calgary can seem to ignore people getting hit by the LRTs on 7th ave.  In the last five weeks 3 people have been hit.  Calgary has grown.  An at grade downtown LRT isn’t in the best interests of the city.  Maybe instead of building overpasses in the suburbs, Bronconnier should do something about the LRT on 7th ave.


  8. The T35A08 can increase passenger capacity in many ways besides adding gangways between vehicles.  Bombardier has recently introduced FICAS (Fully Intergrated Carbody Assembly Syetem).  This will allow for thinner walls.  By reducing the width of the walls, more interior space can be realized.  Up to 200 mm of wall space be saved.  This will probably mean 10 more passengers per vehicle.

    You can read about it here.

    I am a big fan of ATO since computers can always drive a train better than humans.  It can accelerate faster and brake at the right time.  ATO also means that metros run smoother without the brake dive that spills hot coffee.

    Steve:  Acceleration rates are limited by passenger comfort and safety concerns, but definitely when trains are running close together, an automated system can do a better job of maintaining a regular speed if it’s properly designed.

    Why does the TTC need to put in a human to close doors?  The ICTS systems in Vancouver, Kuala Lumpur, monorails in Las Vegas and Tokyo have cameras that monitor doors.  Someone from central command can close the doors.  Without human operators, trains can turn around much faster.  Simply unload passengers and load them again, the train is good to go.  The operator no longer has to walk from one end of the platform to another.

    Steve:  In Vancouver, the length of time the doors stay open is part of the train schedule.  It varies by station, direction of travel and time of day.  Nobody at central control is monitoring this operation.

    Steve, I also want to discuss one more point with you.  Does it matter that the T35A08 has no high speed mode?  The iron wheel on a train does not have a lot of traction versus rubber tires.  There is a limit on how fast a train can be accelerated.  I know between Warden and Victoria Park station, faster top speed will help.  However, between places like Chester and Broadview, faster acceleration is more important than top speed.  I’ve checked the Bombardier website, it appears that most metros can do at least 80 km/h and tops out at 110 km/h.  If the TTC is purchasing “off the shelf” products, why would it be detuned to a lower speed?

    Steve:  This is a good question for the TTC, and I intend to pursue it.  To give an idea of the impact of high rate operation, I have ridden trains in high rate between Eglinton and Finch, and it cuts about two minutes off of the running time each way.  The speed up the hills at York Mills is particularly impressive.  Over the length of the entire line, I suspect that we would save at least 10 minutes out of the total round trip of 118 (peak).  Alternatively, ATO could be used to selectively operate at higher speed to make up for delays.


  9. I completely agree with you that combining the reduction in train operators with the appointment of Station Managers is not necessary. (Nor would former operators necessarily have “managerial ability”.) Though having a Station Manager is certainly not the total answer to poorly maintained stations, broken equipment etc and the Manager needs to be able to get staff to actually correct any defects s/he finds it would surely help. The “look” of the TTC does not give one a very good impression and it really is a pretty good system!

    On the question of user operated doors it is, perhaps, instructive to look at the London underground. They used to have several lines where users operated the doors (except, I think, in central London) but now all seem to be operator operated. It is, I assume, as fast to close one door as 20 so what’s the time saving with only opening some of them?



  10. Suppose, for the sake of argument, it was possible to increase train frequency by installing ATO.  Would this necessarily be the best use of $750 million dollars.  What else could you get for this money?  How many miles of service upgraded to LRT standards?  How many additional surface vehicles that also would provide more service?

    The mindset still seems to be that it is “subway or nothing”.  How many routes have the kind of passenger patterns that would justify building a subway?  I would bet there aren’t too many.

    Steve:  Moreover, that $750-million would primarily benefit downtown where there is a capacity crunch while doing little for areas in the burbs that would still have inadequate bus service.


  11. The only the TTC needs is automated doors. Installing laser and pressure sensors, will eliminate the need for a second person to operate the doors. The lasers sensor will check to see if someone is standing between the doors. The pressure sensors will sense if something is blocking the door. If either one picks something up, the door will just open again. Calgary LRT uses this model and it has worked quite well. I am sure Vancouver uses this in the ICTS.

    With the money you save, you can paint the subway cars?

    Automated drivers are a bad idea, Toronto’s Subway was designed to have someone driving the trains. It could be possible to have an automated train system, but someone would still have be on standby for safety purposes (as is the case in Montreal).


  12. The saying… what goes around, comes around… applies to ATC and wishes for lower subway headways… check out this eerily similar discussion in the 1993 EA on Northern Subway Expansion… p. 4-17 click here

    Steve:  A few points worth noting in this document.

    The cost estimate for the signalling work, allowing for inflation and change of scope, is roughly the same as today.  If anything it is better because the cost of this technology has fallen due to widespread use. 
    It clearly states that there are additional costs required to improve turnaround operations at Wilson (as it then was) and Finch terminals to achieve closer headways.
    It clearly states that there are additional costs for cars to run more trains.

    One big problem with Howard Moscoe’s pitch is that he neglected to include these and other additional costs associated with his scheme.

    In another part of the report, the consideration of alternative technologies is fascinating.  TTC has often claimed that an LRT extension of the Spadina line was considered in the earlier EA and rejected.  In fact, this EA looks at LRT (and BRT) only as a mode for east-west links between the Yonge and Spadina lines that would redistribute traffic between them, in effect as an alternative to either a Sheppard West subway, or the various loop schemes.  Subway technology is always presumed to be the mode for the Spadina Extension because the whole purpose is to build a subway loop, and you can’t do that with an LRT extension.  I will turn to this in a separate thread about the tyranny of old studies and EA approvals which keep haunting us and preventing the adoption of new alternatives.

    My thanks to Bob Brent for pointing me to this document.


  13. Automation can be very successful – just look at the Docklands Light Railway, or the SkyTrain in Vancouver – and while it’s much rarer to retrofit it onto an existing line, there’s no reason why it shouldn’t be possible. The question is whether it’s the most important job needing the money.

    Passenger-controlled doors are a bit of a red herring. Except on much lower-volume routes (or at Dupont station, maybe), their benefit has nothing to do with speed: it’s that passengers sitting near the doors aren’t blasted with hot/cold air at above-ground stations unless someone actually wants to get on or off.


  14. This whole discussion (from Howard Moscoe) is maddening.

    Guys, STOP dreaming techno-toys!  IMPROVE CORE SERVICE.  More buses, more streetcars, you know — basic, boring stuff.

    The subway will operate more efficiently (not forgetting the necessary new technological improvements needed to replace worn out infrastructure) if there are more surface vehicles getting passengers to subway stations, and therefore, creating hopefully a more even flow of passengers, AND if there are alternative ways of actually AVOIDING the subway if needs be.

    As in New York, passengers should learn to ride on surface vehicles along Yonge and Bloor-Danforth, as well as on the subway.  I know the TTC tried instituting day-time service on Yonge south of St. Clair, but they did not seem sincere about it.  It’s as if, since the streetcars were replaced by the subways along Yonge and Bloor-Danforth, the idea of ANY surface operation on these streets is an excommunicable transit sin!

    Let’s look at BASIC, simple, boring ways of solving these problems.  Leave Star Wars technology to Hollywood and Washington!

    Steve:  I’m not sure a Yonge surface bus is going to make much of a dent in riding.  The real question is:  where are the additional riders coming from and where are they going to?  If they are bound for the core, is there another way to get them there?  GO Transit?  LRT lines in the Don Valley and Weston corridors?


  15. I have read the Star article and the follow up along with the comments here and I think we need to look at Howard Moscoe’s plan as a vision rather than a finished plan.  The vision includes a series of concepts that can be separated and ‘fleshed out’.

    Steve’s comments start out by doing exactly that, but in the final conclusion, I see the baby headed down the drain with the bath water. A better conclusion might be to package each of the issues then evaluate and cost them along with possible alternatives, then see where we get to.

    For example, it appears to me that ATO is a good objective simply as a safety net without any of the other baggage.  I cannot see a Russell Hill type collision occurring if ATO is in operation.  Also there may be some intermediate steps to the final objective where we can get partial results for a portion of the cost.

    Slightly off the topic, GO should probably be looking at something like ATO.  They currently do not have any signal enforcement and they have already had one case where an engineer miscalculated, fortunately at very low speed.

    Steve:  My point exactly is that the issues of night service, one person crews and station management have been bundled together with the ATO proposal.  There is no question that we need a new signal system given the age of the existing one, but this can be a gradual program spread out over many years.  Tying it up in a package where all of the capital must be spent at once places an artificial priority on yet another big ticket project that would muscle other needed improvements out of the way.

    As for GO, the signal system on its tracks is run by CN and CP, and all of their operations would need to interface with it as well.  If this were done, it could likely be a form of cab signalling with the ability to override the engineer’s controls, rather than full-blown automatic train operation. 


  16. The Yonge Line signal system is past due for replacement after 50+ years of service. The University Line signal system is 45 years of age with replacement at the budget planning stage.

    It makes sense to use the opportunity to depart from the conventional dated fixed block relay technology and to upgrade to the moving block system that will allow superior train control on the entire line.  (The Bloor Line signal system will need replacement shortly thereafter.)

    Moving block technology allows trains to operate at greater efficiencies as well as improved levels of safety.  The elimination of the signal system track circuits (they detect the train) as well as the elimination of train stop devices (remember train stops from the Russell Hill accident) will result in a more reliable signal system.  These are the 2 pieces of signal ‘ground’ equipment that cause the most delay related equipment failures.

    Lower maintenance life-cycle costs can be achieved from the elimination of track circuits, trainstops and wayside signals. The SRT moving block signal system has proven to be extremely reliable.

    Operational limitations due to terminal turnback are real but can be partially addressed through step back operation which are simplified with a single operator (or no operator at all.)

    Discussion of station managers, Platform Screen Doors and driverless trains are important side issues.  The Union members at the TTC do not want to see job loss due to automation, so the introduction of station managers concept is ‘floated.’

    Washington introduced station managers when they automated their fare collection system.  They restart escalators and provide customer assistance.

    The signal system as well as it’s associated cabling in the tunnels requires replacement as time goes on.  Replacing the entire YUS signal system on it’s own merit makes sense.  Piecemeal replacement of the Yonge Line signal system using the older technology eliminates the opportunity to upgrade to the superior train control.  It’s the perfect time to discuss ATO.

    P.S. the power rail can be de-energized in one direction at a time with existing switching.  In an emergency, the power will always be switched off and stay off in both directions.

    Steve:  I fully agree with the replacement of the existing signalling system.  My only concern is with the artificial acceleration of the timing of the project and coupling this to an unrelated issue like Station Managers.  We should not have to wait for a decade or more of resignalling work to get escalators restarted promptly.  Thanks for pulling in the technical details of what we avoid having to maintain (or build on new lines) when we get rid of the existing block signalling system.

    By accelerating the implementation schedule, we also bring forward the need to replace some of the older cars.  It is likely that the TTC would not spend the money on putting ATO into vehicles with less than 10 years’ lifespan.  This would make more work for Thunder Bay in the short term, but then there would be a very long dry spell with a fleet that needed no replacements for decades.

    The SRT signalling system itself is very reliable, but the equipment it has been running on is antique and fails regularly in cold weather.  Finally, the TTC is getting around to upgrading, something Vancouver did a long time ago.

    I was unaware that the power rails could be de-energized in one direction separately and will have to verify this.  In any event, there are sections where for safety the TTC would want to cut both directions such as anywhere that the two rails are close together such as the open cuts.  They would not want to chance having workers stray into the “live” rail.


  17. Passenger-controlled doors are a bit of a red herring. Except on much lower-volume routes (or at Dupont station, maybe), their benefit has nothing to do with speed: it’s that passengers sitting near the doors aren’t blasted with hot/cold air at above-ground stations unless someone actually wants to get on or off.

    Exactly. Getting one of those freezing Prairie Northeastern wind drafts at 20 below is none too pleasant.

    Additionally, doesn’t some of it have to do with the fact that the Siemens designed vehicles in Calgary & Edmonton also serve as “pre-metro” car-sets in Germany and Europe (streetcar service above ground, with some underground subway-like service on the same route)?

    Based on personal experiences in Cologne and the Ruhr (IIRC there are DuWag carsets there), the main change would be an altering in the passenger entrance design – usually steps rather than a platform.


  18. My reference to buses along Yonge and Bloor-Danforth was a “for instance”.  Yes, the key is knowing who is going where and when.  Greater use of the rail corriders is of utmost importance.

    But how long will it take, and can the TTC implement more service, sooner, like the Premium Express services?  Can it be done without charging Premium Express fares?  Just like the old Coach services along Avenue Road and Mount Pleasant were meant to divert passengers from the over-crowded YONGE streetcar (and yes, alas, the Coach lines were “premium-fare”), can’t the TTC do the same to the subway lines?  I know there were attempts in the ’70’s and ’80’s that were not successful, but I always got the feeling they were designed to fail.

    Maybe, just maybe, they should think about channelling passengers away from the subway.  You can implement as many technological improvements to the [subway], there will come a point where it cannot handle any more than it can.

    Steve:  There is a huge problem with capacity.  To achieve a 10% increase (say 3,000 passengers per hour), you would need about 60 buses per hour (at a design capacity of 50 each on average).  That is a very frequent bus service, although it could be divided into multiple north-south routes, and they would all have to end up somewhere downtown.  One big problem is that the major collection points for riders on the Yonge line do not correspond with easy jumping off points for a parallel express bus service. 


  19. Setting aside the questions you raise as to whether the scheme would deliver the promised benefits, it’s disconcerting to read of the ‘apoplectic’ reaction you describe in response to the suggestion to run trains faster.

    Steve:  The apoplexy came from my suggestion that they needed fewer trains just when they were locking up a contract with Thunder Bay.

    Not only does it seem that passenger considerations are secondary to others, but the type of reaction sends a clear message that innovative ideas are likely to get a frosty reception.

    The way I see it, if the TTC could increase capacity by 26% by implementing this type of automation, any jobs lost in operating the trains would be made up in increased jobs for:

    – subway car maintenance (more cars – more staff)
    – bus drivers (more subway passengers would mean more demand on the surface routes that feed the subway.

    I can’t see much more capacity being available – especially on the B-D line – without expanding the capacity (platform + stairs/elevators) at Yonge & Bloor.

    Has building new platforms on the outside of the B-D line at Yonge & Bloor ever been looked at?

    Steve:  There was a very complex scheme years ago of which the only part actually built is the widened platforms on Bloor Station.  The design involved moving the tracks further apart and adding a centre platform (this would have closed Bloor Station for a very long time), and adding new platforms on the outside of the existing Yonge Station tracks.  The engineering involved was ferociously complex because the whole station sits in a streambed (the Hudson’s Bay building is actually sitting on an underground bridge!), and parts of the station are hemmed in by building footings.  The new eastbound platform would have had a support column for the Bay right in the middle of it!  This is an example of “we can do anything” designing that ignores the fact that at some point, there’s just no more room.


  20. At Bloor-Yonge, would it be possible to add an eastbound platform to the south of the B-D line, while converting the centre platform to a westbound platform and widening the escalators?  This is what will be done at Union Station.

    Steve:  This was proposed as part of the TTC’s mad scheme to rebuild Bloor-Yonge Station (I have the report, and I have seen the scale model — this is not a joke).

    There are many, many problems with what you propose.  First off, the new eastbound platform would conflict with the supporting pillars of the Bay (the subway is physically inside of its parking lot).  Even if we allowed for that, you now have to get vertically to the Yonge line.  This would involve expanding the two existing transfer areas to the south out under Bloor Street.  (As a point of reference, those silver fluted columns about 1/3 of the way down the Bloor Station platform are right under Bloor Street, and they used to hold up the transferway to the Bloor Streetcar.)  I’m not sure, but believe that some serious utility relocation may be involved. 

    The very east end of the eastbound platform is at roughly Park Road and Bloor (the street at the east end of The Bay).  This would possibly put the east end of the new platform somewhere under the existing buildings on the south side of Bloor.  Very messy.

    At Union, they had the advantage of “the moat” in front of Union Station rather than a whacking great building.  Even so, there is a very large sewer now being moved out of the way of the subway expansion.  That work alone is a good chunk of the total project cost.


  21. Well if money was no object, could they build another layer downward? One of the B-D levels EB and one WB? Or would tunneling below existing trackage be such a disruption that it’s not even feasible?

    Admittedly though it’d be an absolute labyrinth of a station – even moreso than now.

    If there was one thing I would like to see changed in the Hudson’s Bay Centre would be the entrance (on both the LCBO and Shoppers Drugmart sides) – it seems so counter-productive to such a key station.

    Steve:  No.  Aside from the extreme difficulty of working that far underground in a watery environment, you would need a junction with existing tracks somewhere well away from Yonge Station to provide for the graded approach.  There is no place to put it, especially to the west where the subway is underneath buildings.


  22. I can see the need for close spaced trains on the Yonge line, maybe even BD, but isn’t the Spadina line running under capacity already?  If trains are running more frequently down Yonge could they be short-turned at Union to go northbound?  Or maybe turn around at St. George?  They could even implement somewhat of the integrated subway system by having half of the trains coming up University turn east bound onto Bloor and then north onto Yonge.  I just don’t see the need for increased capacity on Spadina, and don’t the two lines need to be looked at as one.

    Steve:  I won’t go into analytical details (there is a bit more in a reply to another comment), but I need to point out a few flaws in your assumptions.

    If you short turn trains, you great a gap of 2x the basic headway.  That would not be enough to provide service northbound on University (which is quite busy, thank you, during the height of the rush hour) nor would there be enough service going north from Union to accommodate any sort of integrated operation at the Y.  Turning at St. George would be very demanding as there is no pocket track to use as a staging area for a turnback.  As for going north on University, east on Bloor and north on Yonge, that is physically impossible.  There are no track connections to the Yonge line at Bloor Station, and don’t even think about trying to build them with the subway structure sitting physically under and inside buildings.


  23. With regard to passenger platform/stairway choke points at Bloor-Yonge and St. George with ATO and increased volumes, couldn’t that be alleviated through the Y?  With ATO in place on both lines, I imagine the Y could be reopened and made workable.

    Steve:  Back when the Y was operating, most of the traffic was bound for the core.  Now there is a very large volume transferring to and from the north.  The proportion of the BD traffic that would benefit from the Y is lower than it was in 1966.

    Also, even at a headway of say 110 seconds, you still need to fit the BD-downtown trains around some sort of service coming down Spadina, a line we plan to dump more riders on.  I’m not going into a complex operational analysis here (yes, I have thought about it and rejected it as unworkable even in an ideal situation), but the short answer is that it won’t work.  Once the Spadina line was hooked onto University, any thought of running an attractive service in from Bloor-Danforth was lost because there isn’t enough track time left over for the Bloor trains on University.


  24. In that case, do you think a third north-south line into the core is justified?

    Steve:  Yes, but as LRT, not as full-blown subways.  Two candidates are a Don Mills line and a Weston line.  Also, of course, more capacity on GO so that we are not trying to handle trips from the 905 on the subway lines.

    Back in the grand days of planning an extensive rapid transit network, it was politically incorrect to propose more capacity into the core because this would divert resources from suburban subway expansion.  Meanwhile, some politicians (including one who has gone on to Federal leadership) has the naive idea that not building more transit would actually constrain growth in the core.


  25. As a subway operator I find this thread very interesting.

    Automation and laser doors will never solve the problems.

    A lot of people were discussing short turns and I thought I would add to those thoughts.  When a train is short turned we go “out of service” which requires both the guard and operator to leave their respective positions and make sure that the train is clear of passengers before we go into the pocket or centre track.  Inevitably there will be one or two passengers which did not hear the announcement or noticed the people leaving the train.  We must then escort them through one of the staff doors which can add a lot of time to the turnaround.  Cameras simply will not do this and a single operator will take twice as long.  For instance a simple short turn a St Clair West on a weekday would involve an operator walking the train twice before leaving the platform and then when in the pocket or center track would have to walk back to the other end before they could pull out again.

    Steve:  This is an important point, and it is the cause of much delay at short turns.  The TTC needs to decide whether clearing the train before entering a pocket is really worthwhile, or if the “lost” passengers will have to figure out what is happening on their own when the train re-enters service.

    With the doors being user operated, by camera, or by laser is not very practical.  We have enough people holding the doors open for their friends and for strangers delaying the entire line because the 2 minute wait for the next train could be too long.  It would be scary to see such a system at work downtown at rush hour.

    I am not trying to protect my job with these arguments.  I would just like to shed a little light on the facts.  The people who promote automation are probably the same ones who said we would be living in a paperless world by now back in the 1980’s or thought the dot coms would not become the dot gones.  I have a few suggestions to improve service:

    Having the operators at either end will improve end terminal turnarounds and can be done now.
    POP service will be an improvement and should be done only with a large increase in enforcement (similar to GO).
    Having maintenance workers work in the daytime to keep things clean.
    More special constables on the subway system doing regular patrols will avoid a lot of long delays for security problems. (this works well with the POP idea)
    Adding additional “call on” lights to bring trains closer at busy stations.
    24 hour service can be done on the Sheppard Line right now it has reverse signals in place already. (we only close it for two and a half hours anyway)
    We can do 24 hour service in segments on the mainlines with shuttle service planned in advance through sections closed for maintenance.  Which would allow for improved maintenance; because they could work longer on a section (maybe an entire weekend).
    We could offer express service for the first trains that depart in the morning.  For example the first train leaving Finch southbound in the morning goes all the way to Union empty.  Why not operate an early bird express?
    At night we take empty trains from Finch back to Wilson.  We could remain in service until 3:00 am right now.

    I do not like seeing paying passengers being left on the platform late at night while I take an empty train in the direction they want to go.

    Most of us want to do a good job and take pride in what we do.  Unfortunately the TTC is a large company with a lot of good and talented people at all levels that is controlled by local government and is pulled in a lot of different directions at the same time.  I do not know how to improve this.  I appreciate the passion that a lot of you have and wish that I could say more.

    Steve I am a big fan of yours and I enjoy your articles, keep up the good work.  I would like to see someone with your knowledge sit on the commission

    TTC Subway Driver


  26. Regarding the Bloor/Yonge transfer point, wouldn’t it be possible to reconfigure the stations so that Bloor would only permit transfering to/from Yonge North to/from B/D, Spadina would have a moving sidewalk reinstalled and be used for transfering to Spadina from B/D, and St George to be used as the sole station for going downtown from the B/D line?  This way, the traffic from B/D would be split, relieving Bloor/Yonge, and St George could be modified/expanded more easily than Bloor/Yonge I assume?

    Steve:  What you are saying, in effect, is that if someone is at a station south of Bloor and they want to transfer to the BD line, they must ride to St. George.  This takes them a long way out of their way especially if they are at College or Wellesley.  How exactly are you going to prevent them from getting off northbound trains?  Even assuming you could do this, all you are doing is switching the transfer traffic between the transfer stations and making the system infinitely more confusing to riders all in the name of better efficiency at stations.  Mind you, that’s the sort of thing some of the folks at TTC might come up with given how rider-unfriendly it is.

    And a side note, with ATC and increased frequencies, and assuming there’s a way to correct the timing, wouldn’t it be possible to divert some trains on the Yonge line to Don Mills via an improved wye at Sheppard, so there’s through service from both Don Mills and Finch to Downtown, reliving Finch and making more use of the Sheppard line?

    Steve:  Installing a west to south curve from the Sheppard Line to the Yonge line would be immensely expensive because it would have to go under the existing Yonge tracks and merge in somewhere south of Sheppard Station.  Also, the last thing the Yonge line needs is more riders fed in by the Sheppard line.  That was one of the original fallacies of the Sheppard route because if it actually achieved its purpose, it would hopelessly overload the Yonge line.


  27. TTC has 4 interchange stations between B-D and Y-U-S (only 3 are in use) and they are all lined up in a row.  What about short turning Yonge-University line trains at the wye, and having passengers get off at Lower Bay station?

    The transfer to B-D would still be possible, and being able to board Y-U-S trains at Lower Bay could relieve plaftorm pressure at St. George and Yonge.

    Steve:  This is physically impossible.  There is no connection between the Yonge line and Bay Station, and one could not be built because of conflicts with existing building foundations.

    I suppose that there is no chance of TTC separating the Yonge-University segment from the Spadina segment?

    Cheers, Moaz

    Steve:  There was a proposal years ago to do this by four-tracking Union Station, but it was never pursued.  The proposed design would have kept the construction folks busy for years, but operationally I don’t think it would have helped.  There is an advantage in having “around the corner” travel that distributes demand between the two lines.


  28. Hi Steve:

    This note is sent because of your comment (paraphrased here) that ‘power cannot be cut off in only one direction and still allow for the other track to be used’. With co-ordination with holding trains out of the area in question, power can be cut in one direction only for maintenance needs; but the method to shut it off requires both directions off until the appropriate time when the other direction can be restored for use. Probably about a three to five minute operation.

    There are many more details in this procedure than the simplistic explanation that I’ve given here and I’m sure that since I trained people in the system how to cut power for maintenance needs there have been numerous subtle changes, I would really be surprised if that major a change would have occured.

    Indeed your other comments are accurate for safety for the work crews, trains are required to pass a power off work zone need to run slowly by it for its length plus about 1,000 ft. on either end, thereby increasing travel time greatly and therefore really minimizing capacity.


  29. Hi Steve,

    I found this article and read through the entire thing, including comments, and one comment that stuck out to me was the mentioning of increasing capacity by adding a 7th car. Would it not be possible to add a 7th car, but have trains stop at each station so the first door is near the beginning of the platform? Essentially, the front of the train from the edge of the first door should be in the tunnel.

    I know they do this in some stations in New York. Perhaps if more room was needed, the last door in the train could be inoperable, allowing the train to stop further in the tunnel (at the end).

    I look forward to hearing your comments!


    Steve: Current 6-car trains are about 50 feet shorter than the stations and could almost fit a seventh car on the platform. However, this would require that trains stop quite accurately, something that is unlikely with the current manual control system. We won’t have any form of automatic operation on the subway for almost 10 years, and then only on the Yonge line.

    Meanwhile, the new trains on order will have about 10% more capacity because the car-space is one continuous unit rather than being broken up along the train. In theory, it should be possible to put one short car in the middle of the train to build it out to 500 feet. However, the TTC is not planning such a change at this time, and it would not be practical until they can guarantee that a train can stop quickly and accurately exactly where it should. If the train has to slow down to allow a precise stop, this adds to the station dwell time and slows down the service.


  30. All very interesting.

    Big debate raging in AUs re ATO and the benefits.

    My question: With a legacy system with a single flat junction where a pair of tracks merge with no grade separation how many trains an hour can be operated with a modern signalling system and ATO. Would love reference cases if possible.

    Am doing a rail capacity thesis in Curtin Uni in Western Australia.


    Steve: To be clear, I assume you refer to two pair of tracks merging — four into two — so that there is the conflicting traffic across the diamond to consider, not just switching. That will be the controlling factor.


Comments are closed.