Union Station & Rail Corridor Capacity

At the recent Metrolinx Board meeting, staff presented an overview of planning now underway for the future of Union Station.  One background report addressed the future levels of GO, VIA and other services at Union and the surrounding rail corridors.  This report makes interesting, if unsurprising, reading because it confirms what anyone with even a modest understanding of railway operations already knows:  there are severe capacity constraints at Union as it is now configured and operated.  Too much discussion focuses on a bright future of frequent service without considering how we will fit all the trains and passengers through the hub of the network.

The full report is not online at Metrolinx, but I have obtained a copy.  Due to its size, I will not link the entire document here.  If you just want the highlights, read the Executive Summary.  For more details including a description of the evolving simulations of various levels of service, read the main report.

USRC Track Study Executive Summary

USRC Track Study Main Report

The study considered various scenarios corresponding to stages in the growth of GO and other services over coming decades:

  • Base Case:  The existing service at Union, including a reservation of two tracks out of service for the reconstruction project.  This was used to calibrate the model.
  • 2015:  Construction at the train shed is completed giving two more tracks for service.  The only new peak hour service beyond the base case would be a few VIA trains and the Air Rail Link.
  • Electrification study base case:  This configuration was used as a starting point for the recent electrification study, and it assumes two-way service on all corridors.  Three variants of this were tested to refine operations and remove constraints triggered by service at a much higher level than today.
  • Maximum capacity:  This configuration attempted to maximise service on all corridors.

The study concludes that significant changes will be required both in the physical plant (track, signals) and in train operations which will have to be managed considerably more tightly than today.

All of the simulations randomised train times from their schedules over a range of 0 to 4 minutes to reflect the actual system behaviour.  A trial expanding the range up to 5 minutes produced a noticeable drop in simulated on time performance (“SOTP”).  Similarly, when each scenario was tested for the effect of additional trains, the SOTP fall below acceptable levels.

Although each increment in train service can be handled with revisions to station plant and operations, the changes never get ahead of the demand and eventually the station runs out of capacity.  More accurately, I should say that the corridor runs out of capacity because the worst constraint is actually at John Street where there is a pinch point north of the CN Tower.  13 tracks approach from the west and there are 15 platform tracks at the station, but there are only 10 tracks at John Street with no ability to add more.

Another problem arises from scheduled bunching.  Although there may only be 29 trains in the peak hour, over 1/4 of them (8) arrive within a 10-minute period.  Because of conflicts between the routes trains take through the station approaches as currently operated, the corridor and station operate at well below the capacity one would expect for a facility of this size.

Passenger movements within the station are not modelled in this study, and that exercise is left for another consultant with the expertise to perform such work.  However, the authors of the track capacity study do acknowledge that some operating schemes may run aground on the difficulty of passenger handling.  Indeed, the movement of passengers, including the dwell time for loading, is an important constraint on overall capacity.

As a trial, one train was added on each of the seven corridors in the Base Case.  This caused the SOTP to drop below 50%.  Even just three trains were enough to bring the SOTP below 90%.  GO’s oft-stated goal is 95% on time.  This means that there is no capacity for additional service into Union during the peak hour as things now stand.

The 2015 Case

The 2015 Case adds four Airport trips and two VIA trips during the peak hour, shifts VIA operations further south in the station, shortens the dwell time of VIA trains, and restores the availability of tracks now out of service.  All but the Barrie and Milton services achieve a good SOTP, but those two routes show the effect of the Airport line’s built-in conflicts with other GO and VIA operations.

The Airport is on the south/west side of the corridor in Malton, but the platform it will use at Union (on Track 1 for easy access to a new lounge in the West Wing) is on the north/east side.  Every Airport train must cross all of the tracks used by other services in the Georgetown corridor, and at a 15 minute headway, the conflict between services is inevitable.

Adding trains to the 2015 Case causes the SOTP to fall below acceptable levels indicating that the corridor is effectively “full” even after the station has been rebuilt to handle far more passengers.  This leads to a failure of the next scenario.

Electrification Case 1

This scenario presumes full all-day service on all corridors with a total of 52 train arrivals at Union in the peak hour, almost double today’s level.  The SOTP value comes in at 84.4% as an overall average with different values in each corridor.  This scenario does not work without further changes to the infrastructure.

Electrification Cases 2 and 3

The John Street pinch point is a severe problem coupled with the fact that the tracks at the south end of the station have only a few access routes shared among them.  This can be fixed by adding crossovers near Fort York on the Oakville Subdivision so that trains can fan out onto other routes before they hit the most congested point, and by changing the layout of switches leading to the southernmost tracks.  A minor change to the access to track 1 reduces a conflict with the Airport service.

There are no problems on the east approach because the number of tracks grows from the main lines through the yard and into the station.

These changes bring the SOTP up to 97.5% except for Lakeshore West, and that problem is resolved with yet more changes to the route through the station.  At this point, the station itself is reaching its hourly capacity.

Alternative Configurations

One alternative that was examined was to shift the Airport service to Track 3.  Leaving aside the effect this would have on access from the West Wing lounge, this would partly remove a conflict between the Airport trains and access to Bathurst Yard.  However, the overall benefit is small.

Another possibility was the creation of wider platforms by removing tracks.  This gives more space for passengers and should allow for shorter dwell times, but it requires much shorter headways to achieve the same capacity.  This would only be suitable for through trains that did not have to reverse before leaving.

Double-berthing was also studied.  In this scheme, trains from the east and west would share the same track each using half of the platform.  If a track is used for through service, the minimum headway is 10 minutes (allowing for dwell time, some leeway for delays and the time for a train to clear the platform before its follower can enter).  That’s 6 trains per hour.  With double-berthing, trains are reversing at the station, and the minimum headway is 15 minutes because of the time needed to set up a train to change direction.  That would, in theory, give us 4 trains an hour each way, or 8 trains in total.  Double-berthing could add capacity, but it poses additional problems.

First off, only four tracks (6-9) are long enough to hold two full-length trainsets.  Second, for frequent service, trains could not be spotted in advance on the platforms, but would have to pull into position with another train already on the track.  This is a safety concern and it affects operating speeds.  Finally, passengers would have to walk to the extremities of the platforms (well beyond the shelter of the train shed) to reach the outer cars.

Electric operations would improve train acceleration characteristics and save a bit of time, but the amount would be small because of speed restrictions within the station and the corridor around it.  This is not to say electrification is of no value, but that value would be obtained in other locations, not at Union.

Yard operations at Don and Bathurst would have to be confined to off-peak periods because moves to and from these yards would interfere with revenue trains.  As headways become shorter and two-way operation grows, the number of peak period yard moves should fall anyhow, but a time will come when such moves are impossible.

Summary So Far …

Each of these cases begins with a constrained configuration, shuffles operations and rearranges some infrastructure to obtain a new workable scheme.  However, each of the new states is “full” the day it is implemented, and the process must be repeated for another increment of service.

Moreover, operating so close to total capacity means there is little room for off-schedule service either through random delays (e.g. weather) or from inattention to on time departures.

Maximum Capacity Operation and Alternatives

The minimum headway a track can sustain is 10 minutes for a through service or 15 minutes for a turnaround.  [Yes, it’s possible to get lower values, but this would require changes to operating rules, removal of recovery time in the schedules and, possibly, problems with conflicts between moving trains and passengers on the platform.]

Operating the system, as opposed to each service, at its minimum headway demands that each route have a direct path to its platform that does not cross over others to the greatest degree possible.  The biggest challenge here lies with the Airport service tethered to Track 1 and the Georgetown corridor operations.

Typically, there is only one path available through the corridor for a service.  This means that the approach track to the station is effectively single track, despite the maze of switches.  This path may connect to a single platform, or to a pair of platforms.

If to a single platform, then the headway is constrained to 10 minutes through or 15 minutes for turnaround, and both the platform and the approach route are single track.  The minimum headway is constrained not just by time on the platform, but by the distance from the platform to the point where an outbound train can pass the next inbound one.

If two platforms are used, then the headway can be cut in half (5 minutes for through, or 7.5 for turnaround) by alternating platforms.  There would still be a “single track” path between the train shed and each corridor, but inbound and outbound trains could alternate in using this while one was always on the platform.

Even with this strategy, the addition of more trains to the peak hour is possible only if more platforms are created.  One scheme is a tunnel under the station for exclusive use by the Lake Shore services, and the other is a new station in place of Bathurst Yard.

Lake Shore Underground Option

In this option, there would be room for 12 Lake Shore trains each way per hour via the tunnel.  Barrie and Richmond Hill would be linked as through services, as would Georgetown and Stouffville, and each route could have 12 trains/hour (48 in total).  Milton would have 7.5 trains per hour as a stub operation.  (The hookups used in this scheme are  function of the north-to-south arrangement of services in the corridor to avoid conflicts on the approaches.)

This has the greatest capacity, but also the greatest cost because of the tunneling.

USRC Track Study_Underground Station Drawings

USRC Track Study_Underground Option 1

USRC Track Study_Underground Option 2

The drawings show views of a new underground station that would serve only the Lake Shore route.  It must be low enough to  be under the existing footings of the building and to clear the LRT tunnel under Bay Street.

There are two options for the station placement:  either offset with the centre of the new platforms under Bay street, or aligned to the existing trainshed between York and Bay.

The alignment has little effect on the design of the approach ramps to the station, and the  greater issue is the gradient.  If the grade is 5%, steep even by subway standards where 3.5% is the preferred maximum, then the western curve down would begin 275m east of Spadina while the eastern curve would begin just west of Jarvis Street.  If the grade is constrained to 2%, the western curve would begin 175m west of Spadina, and the eastern curve at the west side of the Don River.

Although these designs have not received detailed engineering or feasibility studies, one concern already raised is the limited availability of land at the portals which would necessarily be wider than the existing tracks.

Another issue is passenger access because the station would have to accommodate a 2000+ passenger train arriving every 2.5 minutes, and would have to provide circulation capacity for these passengers to reach other parts of Union Station, the PATH network and the surface.

Combined with a $1.3-billion price estimate, this option faces severe, (dare I say) uphill battles for acceptance.

Bathurst Yard Station

An alternative scheme would see Bathurst Yard converted into a station for the Georgetown and Barrie Services.  Because this is a stub-end station, service would be limited to 7.5 trains/hour (15 minutes on each of two platforms per service).

This arrangement eliminates the conflict with the Airport trains.  All services would have a direct route to their assigned platforms without crossing the routes of any other trains.

A “Bathurst” station would lie west of Spadina, but it is a fair hike into most of the business district.  One proposed option is for a “Downtown Relief Line” to connect under the new station enroute to a terminus at Exhibition Place.  This option is beyond the scope of the Track Capacity study, but is included in a separate Corridor Options study that I will discuss in a future article.  Some information already appeared in my summary of the recent Metrolinx Board meeting.

One major advantage of this scheme is that the new station capacity is not underground and, therefore, electrification is not a pre-requisite.  Indeed, nothing other than finding a new location for train storage prevents GO from building this station today.  A connection to downtown is essential, but a new rapid transit line into downtown is long overdue and is justified in its own right.

USRC Track Study_Bathurst Station Layout

USRC Track Study_Bathurst Station Drawings

The drawings show how the station could have a building above it and a rapid transit line below.  Oddly enough, the drawing shows Mark I Skytrain cars, not Toronto subway cars or LRVs.  I could put this down to the artist taking whatever stock illustration he had, or to Metrolinx tipping their hand on future technology selections.  We will see in time.

(The pink structure shown in some of the drawings is a placeholder for the recently installed pedestrian bridge across the rail corridor at the foot of Portland Street.)

VIA Rail

The study allows for VIA and ONR operations throughout, but this is not a major issue.  Most of the VIA trains move to the southernmost tracks in the station where they can be isolated from the more-frequent GO movements on tracks to the north.

One issue is the servicing of trains, and this may have to be done at a separate location east of Union (unspecified, but I wouldn’t be surprised to see this in Don Yard).  The location is chosen to avoid conflicts between servicing movements and other trains in the corridor.

None of the VIA changes has been discussed with that company, and it is uncertain whether they would agree to all of what is proposed.

Noticeable by its absence is any provision for a separate high speed corridor.


The study shows that the planned expansion in service at Union can be handled, but the both physical and operational changes are needed.  Track layouts and platform assignments must be changed to eliminate conflicts between trains and at full capacity the station would have dedicated areas for each service.  This is completely different from the service plan GO has discussed for the revamped Union Station with ad hoc assignment of trains to various platforms.

The Don and Bathurst yards may, in time, be replaced with other facilities, and this begs the question of where GO would move these functions.  More trains will stay in service between the peaks with all-day, bidirectional operations, but there will still be a requirement to store some trains.

Train operations must hold to schedules, and platform dwell times must be minimized.  This has implications for passenger flows in the station which have not yet been examined in detail.

Finally, the service level contemplated by The Big Move cannot be achieved without adding more station capacity.  Although two options are presented, the Bathurst Yard scheme is far cheaper and the more likely to be adopted.

47 thoughts on “Union Station & Rail Corridor Capacity

  1. @ Ray L
    If we are to believe the Metrolinx reports than GO Transit is moving to a S-Bahn style system. So we might as well do it right and just built a central city GO Transit tunnel under the CBD.

    The only reason Milan’s tunnel took so long, is because it is in Italy, and sadly corruption rules. But if we wanted, we could build a tunnel in no time.

    At the end of the day, the fact remains that huge double decker commuter trains are not going to accomplish the goals set out in the regional transport plan. Moving to an S-Bahn frequent service regional rail network will.

    Not to mention, that using smaller trains at higher frequencies might actually address some of the crowding issues at Union, because we won’t have 2,000 people getting off all at once in bunches, or even have everyone getting off at Union, if we had proper secondary stations located around downtown.

    If anything, the TTC should have secondary stations at this very moment at Spadina and Jarvis Street. Just simple platforms, nothing fancy. It would probably be a hit with commuters.

    I still maintain we have capacity to grow the system through Union however. If Philly can push through something like 12 regional rail lines through a double track tunnel and then into 30th Street station, including all the track switch changes that Toronto trains have to go through. Then why not Toronto? The problem with Toronto is we don’t go and look how other much busier rail systems handle vastly more train traffic on less tracks than we have.


  2. I have been digging through some old GO documents to find clearances for 25 kV electrification. The preferred clearance above rail head to the underside of overhead structures is 7395 mm, 24′ 3″ for mixed traffic. They want 270 mm, 10.5″ above the top of the overhead. (L1903) The preferred height of the OCS (overhead catenary system) is 7000 mm above the rail head and the minimum for passenger only service is 5135 mm. The vertical clearance must be greater than this to provide electrical clearance above the OCS and is 5500 mm for passenger only service.

    The tunnel clearances that I can find, not GO’s, that seem to meet TC/FRA requirements want a minimum distance of 8′ 0″, 2439 mm, from the track centre line starting at a height of 4′ 0″ above rail head to a height of 16′ 0″ above rail head for tunnels. The minimum clearance from track centre line at rail head is 5′ 0″, 1524 mm. These give widths of 16′ 0″, 4877 mm, and 10′ 0″, 3048 mm, respectively.

    This means that the tunnel must be at least 16′ wide from 4′ above rail head to 16′ above rail head or for 12 feet. Using a little trigonometry and trail and error this would give a tunnel diameter of about 20′ compared to the Spadina extension’s diameter of 17′ 9″.

    Since the cost of tunnelling varies as the square of the diameter, a GO tunnel would cost 30 % more to build. I have been going cross eyed from reading different sources, some in metric and some in Imperial and converting units then trying different scenarios. I send it along for your perusal. It might explain why the GO tunnel will never get built under TC/FRA rules.

    I have also been going over the Canadian Railway Operating Rules (CROR) to see what is required as a brake test when changing ends. A brake continuity check has to be performed. The engineer has to apply and release the brakes while another crew member checks the brakes on the last car to see that they work properly. When CN operated GO with 3 man crews they had an engineer at either end and a conductor on the handicap car. The one engineer would shut down his end and the other would start his. The lead person would apply the brake test and the rear one would watch the brakes. This would take as long as a TTC end change. The Bombardier crews have an engineer and a conductor in the lead unit and a Customer Service Agent in the handicap car to work the doors. When they change ends the engineer walks to the other end of the train, about 1100’ for a 12 car train, performs the brake test, and then the conductor walks to the front end before they leave. Since each person walks about ¼ mile, the change of ends takes up to 10 minutes.

    The operating speed from Bathurst St. to Cherry St. is restricted to about 10 mph. It takes a train about 90 seconds to pass any point. Since the throat tracks can be up to a ¼ mile long it would take the train 180 seconds or 3 minutes to clear. Add to this the time to clear all interlocking blocks and reset the switches and a train would occupy the throat for 5 minutes easily. This is why it is paramount for the platforms to be assigned so that none of the trains need to use the throat to change tracks.

    383onthetree says:

    “Also Montreal still uses the very light single decker GO trains (so does the Northlander) so how do these old light trains pass FRA standards ? “

    Easy! When the cars were built the collision test only required the train to withstand a buff load of 1 000 000 pounds. This is accomplished by having a very strong centre beam running from coupler to coupler. It says very little about the shell of the car staying intact and protecting the occupants.

    Sydney Australia is looking at putting a new tunnel under the bay for city rail and rail link trains, 1500 VDC electrification IIRC. If they do this they would convert everything to single level cars to cut down on tunnelling costs and loading times. Their commuter rail uses 75′ long GOish style double deckers with the doors in the end sections for high platform loading.


  3. The recent US-Canada border deal included harmonization of rail safety standards. Anyone know if there will be an impact on GO?


  4. At times like this I wonder how the UK’s Southeastern Railway company managed to electrify hundreds of km of outdoor railway with 3rd rail power – running all-day service using 4 or 8 carriage trains approximately 20 minutes apart.

    And yes, they have problems with switches and rails not working on the occasions that it snows, but it is still such an accomplishment.

    Cheers, Moaz


  5. How feasible would it be to route the DRL under Eastern Avenue and Richmond or Adelaide Streets instead of Queen Street?

    Steve: Eastern was the proposed route when it was going to be an ICTS line, but that was so that it could get to a potential carhouse site on Eastern. As for Richmond/Adelaide, either of them may work except for the problem of getting to the Bathurst North GO station if that’s going to be the west end of the line. Where there were once freight yards, there are now many condos to tunnel under. Wellington is probably the best bet as it avoids conflict with Union subway station, but gives a good walking distance connection to the south ends of King and St. Andrew stations. Richmond would work with connections to Queen and Osgoode stations, but would be harder to connect with GO at Bathurst yard.


  6. Jay says:

    December 9, 2011 at 3:38 pm

    “The recent US-Canada border deal included harmonization of rail safety standards. Anyone know if there will be an impact on GO?”

    The FRA and TC standards are pretty much the same because so much interlining of equipment occurs. There will probably be little impact to existing standards with the possible exception of Positive Train Control. This is a result of a commuter train collision in California that resulted when an engineer was texting his girl friend and missed a red signal. By 2015 all lines that have passenger trains on them or exceed 50 mph must be PTC (Positive Train Control) equipped. I down loaded a copy of the proposals for PTC from the FRA and read them. I found it interesting that no part of the PTC act can over ride any parts of the “Steam Boiler Safety Act” of 1949. I am glad that steam powered commuter trains are included in this act.


  7. Barrie-Richmond Hill, Georgetown-Stoufville, Lakeshore… if they’re each run through (A they should be) they each need only two tracks apiece. Call it two more for Milton and two more for ARL. This is generous. That’s eight. There are 14.

    VIA’s Corridor service could easily fit entirely into two tracks if VIA had modern equipment and operated competently. This might require high platforms, and would certainly require modern operating practices, but I can’t see any possible problems with it. Of course, all the VIA Corridor services would also have to be run through, and to balance demand an eastern yard might be needed for VIA, but this is managable. If you want to keep it simple, give VIA four tracks, two for trains headed on Lakeshore West, two for trains headed on Georgetown.

    You still have two tracks.

    Ontario Northland service sadly appears to be ending.

    That leaves the Canadian and the Maple Leaf, which admittedly could easily occupy two tracks each due to timekeeping problems — but on the other hand, with such delays, each one can wait until the previous one clears the station, so really these long-distance, long dwell trains should only occupy a single track.

    So it seems like there are plenty of tracks in Union Station. Unless contractual obligations with respect to freight are occupying an undue number of peak timeslots?

    What needs to be done are three things:
    (1) through-running of as much as possible
    (2) reduce platform dwell times for every service (for the Canadian and Maple Leaf this may be impossible, for every other service it can be done)
    (3) Pick the most appropriate tracks for each service to eliminate conflicting movements.
    (3) Rearrange the approaches so as to eliminate remaining conflicting movements. This might require a flyover or two, but these could be placed further away from Union Station in strategic, less crowded locations.
    For example, if the ARL and Georgetown lines used directional operation, with an ARL northbound flyover where it branches off, they could easily run on the same tracks at the John St. chokepoint; they won’t both be running at 5 minute headways.
    For another example, central flyovers to the VIA and GO yards would eliminate a lot of conflicting deadhead moves on the Lakeshore West line.

    Philadelphia shoves its entire SEPTA regional rail system, with far more branches and far more frequency than Toronto, through a four-track station (two through, two terminating). Short dwell times are key. Most of Amtrak’s Northeast Corridor stations are four-tracked. Short dwell times and through running are key.


Comments are closed.