The Evolution of TTC Signaling Contracts (Updated)

Updated September 29, 2022 at 1:30 pm: Links to TTC reports have been updated to point to the “new” TTC website except in cases where the report is no longer online. In those cases, a copy from my archives is linked on this site.

Updated April 4, 2015 at 6:00 am: The review of options for consolidation of the signal contracts by Parsons is now available as part of the TTC’s report online. Comments have been added at the end of this article.

Recently much attention has focused on the runaway project to extend the Spadina Subway north to Vaughan with a flurry of questions about project management, scope creep and cost controls. Another of the TTC’s megaprojects, one that is actually far more critical to the subway as a whole, is the replacement and upgrading of the signal system controlling the movement of trains. This project has dragged on for years while riders endure service problems with antique equipment and line shutdowns for installation and testing of the replacement system.

At its recent meeting, the TTC Board approved a proposal to restructure existing contracts for new signal systems and to simplify the signaling technology that will emerge as the standard on Yonge-University (Line 1) by 2020 with the remainder of the subway system to follow.

In order to make sense of the evolving design for new TTC signals, this article will begin with a short history of the system as it existed and the limitations the new system is designed to remove.

Conventional Block Signals

The Yonge line opened 61 years ago on March 30, 1954, and its signals were quite straightforward given the simplicity of the line as it then was from Eglinton to Union. For most of the route, train detection and spacing was handled by a “fixed block” system:

  • The line is divided up into “blocks” defined by sections of track that are electrically isolated from each other. Of the two running rails, one provides the ground return for the electrical power, while the other is used to detect the presence of a train. If there is a train in a block, it provides a link between the rails, and the signal rail is grounded.
  • Blocks can vary in length and, generally speaking, longer blocks are used where trains will move at higher speeds.
  • At locations where speed control is important, signals can be timed so that they do not clear if an approaching train is operating too quickly.

Train control at block signals was enforced by a mechanical “trip arm” arrangement whereby a trackside arm would be in the “up” position at a red signal. If a train passed, the arm would engage a lever (aka “trip cock”) on the side of the leading truck. This would open an air valve and apply the emergency brakes to stop the train. It was this mechanism that failed thanks to poor maintenance and, combined with operator error and bad design of the signal system, caused the subway crash at Russell Hill in 1995.

The block signal system as implemented by TTC works, but it has a number of limitations:

  • Track circuits can fail if the trackbed is wet (either in a tunnel, or on open track with poor drainage) causing a false indication of a train’s presence. This is responsible for many of the delays due to “signal problems” especially near Davisville Station. The system is designed to “fail safe” by showing a train where there is none, as opposed to failing to detect a train that is actually present. (On mainline railway operations, there can be a minimum train length requirement to ensure that short trains provide enough of a circuit between the rails to be distinguished from random currents on long sections of exposed track.)
  • When a train passes a signal, the trip arm does not immediately rise after the last car clears it, but only after the train has cleared the entire block. This is related to a design that allows trains to approach a red block signal and pull right up to it thereby entering its track circuit (which actually begins just ahead of the signal). This causes the trip arm to move to the “down” position and allows a stop-and-proceed operation known as an “automatic key by”. The result is that the red signal immediately behind a train does not actually protect it.
  • Signals are usually arranged so that two blocks behind a train (the one it occupies plus one more) show a red signal. The trip arm will be “up” at the more distant of the two signals.
  • The spacing of trains is limited by the length of any adjoining pair of signal blocks. Originally, the TTC allowed automatic key bys as a routine move, and at the congested Bloor Station a waiting train was often in the tunnel just outside the station and would pull in as quickly as possible sometimes with the departing train still partly in the station. This procedure was banned as a safety measure, and automatic key bys are now allowed only on the instructions of Transit Control. A side effect is that throughput at Bloor is now lower than was possible with the signal system as originally designed.

There are two types of timing signals on the TTC system to control speed. The most common is “grade timing” used on hills are sharp curves. As originally implemented, the timing signals depended on indications both at the timing location and at the preceding signal. If the “far” block was timed, and the area it protected was actually clear, the far signal would show red, but the near signal would show a lunar white indicating to the driver that the far signal was in timing mode. If the lunar white was off, then the driver knew that the far signal would not clear.

This arrangement had a design flaw that was exposed by the Russell Hill crash:

  • When trains are operating closely together, it is possible that the far signal will change to timing mode after a train passes the near signal which does not show a lunar white. The next train will have moved far enough away to allow an amber indication, and in rare cases a green, at the timed signal by the time the following train arrives.
  • The approaching train’s driver must judge by experience whether what appears to be a red signal is actually going to clear.
  • In practice, operators became rather good at driving their trains so that the timing signals would be changing to amber just as they passed.
  • If, in fact, the red signal does not clear, everything depends on the trip arm stopping the train. A mechanical failure prevented this from happening at Russell Hill.
  • Signals in the Russell Hill section were widely spaced because of the operating speeds. This produced blind corners where no signal was visible.
  • The train that was struck was out of sight around such a corner.

In the wake of this accident, the system was redesigned so that a red signal that is in timing mode will flash to indicate that it will clear provided the approaching train observes the speed limit. This gives the driver a positive indication of the signal’s state. Stopping the train remains the job of the mechanical trip arm/cock. (More about this later.)

The other type of timed signal is “station timing”. This is installed typically where trains will come closer than normally would be allowed in an approach to a terminal. This permits an incoming train to pull up, slowly, right to the crossover ahead of a station so that it can enter as soon as possible when the route is clear. (A similar arrangement could be used at busy stations along the line, but the TTC does not have any of these because the automatic key by procedure described earlier provided the same function, at least until it was banned.)

A different arrangement is used at any locations where trains could come into conflict – switches and crossings between lines. A much more robust control mechanism is needed to ensure that trains cannot collide. These locations are called “interlockings”, a term derived from the original mechanical arrangements for railway switching. At railway junctions, a combination of switches and signals (mechanical semaphores) controlled train movements. These were operated from a central “tower” where a signalman had a view of the controlled area and used a set of levers that were mechanically linked to the switches and semaphores. These were arranged to a control mechanism that prevented setting a combination of route and signals that would be unsafe. The terms “tower” and “interlocking” remain  in use today even though the functions are performed out of sight of the physical tracks and with electronic rather than mechanical controls.

Subway signaling at these locations is more complex and restrictive:

  • There is no provision for automatic key by, and track circuits are arranged so that trip arms rise immediately behind a train as it passes each signal.
  • Signals indicate both the route (lower aspect) and clearance (upper aspect) to a train operator, with additional aspects for key by and timed block indications.
  • Detection requires knowledge not just of track occupancy in a complex set of switches and crossovers, but of the mechanical alignment of each part to ensure it is set as intended.

When the Yonge line opened, it included two areas with interlocked signal and switch systems. From the Belt Line bridge north to Eglinton, there are tracks linking to Davisville Yard as well as the terminal crossover at Eglinton itself. At Union, there was a double crossover in the curve east of the platform. (This was changed to a single crossover as a safety measure after a train derailed on the facing switch coming into the station. By that time, Union was no longer a terminal, and a single crossover was adequate for emergency turnbacks.)

There were also crossovers at Rosehill (south of St. Clair), Bloor, College and King. These were manually operated and, in practice, very rarely used because of their difficult operation. In time, the crossover at Bloor was electrified complete with signals and changes to power feeds to allow Bloor’s use as a temporary terminal in either direction. The other crossovers were removed. They have been reinstated as part of the current upgrade, but will not become operational until the new signal system (which controls them) is turned on.

The basic operating procedure is “red means stop”. A normal block signal can be bypassed with an automatic key by, although this is now restricted to moves authorized by Transit Control. At interlockings, a double red signal can only be passed if the signal system itself gives an automatic “call on” for very slow operation (typically part of a station timing arrangement), or if the move is authorized by Transit Control. In the latter circumstance, the operator must open the cab window and press a lever (the “key” in “key by”) mounted near the signal for the trip arm to be lowered. This complex arrangement is intended to be difficult and time consuming as a reinforcement of the safety needed in such circumstances.

All of this is rather technical, but it is an important starting point to understand the evolution of the resignaling work now in progress.

The Speed Control System

After the Russell Hill crash, the jury from the Coroner’s Inquest made many recommendations related to the TTC’s signal system and train operation. The change of timing signals to show a flashing red (described above) was one outcome.

On November 28, 2001, the Commission authorized the installation of a Speed Control System (SCS) on the existing subway system (YUS and BD) and on the Sheppard Subway which was still under construction. The contract was awarded to Alcatel for $29.5-million.

The new SCS will provide continuous on board enforcement of train operating speed limits, as recommended by the Coroner’s Inquest, and will provide enforcement of all signals. The SCS will issue a violation warning to the train Operator when a speed limit is exceeded. If this warning is not complied with by slowing the train, an emergency brake will occur. Emergency brakes will also be applied if a red signal is violated, thus enhancing system safety. The SCS will provide other important safety benefits in the subway, such as enforcement of a stop at track ends and enforcement of a speed limit when a train is travelling in the reverse direction.

The SCS will replace the current Grade Timing method of speed enforcement, which uses timer relays as well as lunar white and flashing red aspects to enforce speed limits. [p. 1]

SCS uses transponders mounted between the rails to communicate with control equipment on the trains. This system enforces speed restrictions, but track occupancy and routing control remain the responsibility of the primary signal system. SCS provides a way for the signal system to communicate directly with the trains without depending on the operator, but it does not control train operation except for enforcement of speeds and emergency stops.

Although the 2001 report claimed that SCS would replace the timing signals with their lunar whites and flashing reds, these actually remain in use in 2015. Decommissioning all of the existing timing signals and their controls would be a complex job better left to a project that completely replaced the signal system.

The South Yonge Signal Replacement

On September 18, 2008, the Commission authorized the replacement of the existing signal system on the south end of the Yonge line. The contract was awarded to Union Switch & Signal for $14.0m.

Although plans for a move to Automatic Train Control (ATC) were already in the works, the original equipment for the Yonge line was now over 50 years old. Reliability and maintenance were major concerns.

The Resignalling of the YUS subway line utilizing Automatic Train Control (ATC) will improve safety and capacity throughout the length of the line. ATC enforces a minimum safe separation between trains based on the safe braking distance from the last verified location of the rear of a preceding train or any other obstruction such as disturbed switches. This technology allows trains to travel closer together than the traditional fixed block system currently employed on the YUS line. This ability to travel closer together increases the throughput of the service and therefore allows more trains to be scheduled and more passengers to be carried. [p. 1]


To allow trains and workcars not equipped with ATC equipment to operate over the YUS Line and to mitigate against delay in the event of an ATC system failure, a new conventional auxiliary wayside signalling system is required to be installed. [p. 2]

The TTC proposed to replace the existing block and interlocking signal systems on the Yonge-University line from Eglinton to Osgoode with a computer-based equivalent.

The Contract is for the detailed design, supply, installation and testing of a fixed block auxiliary wayside signal system for the South Yonge Subway Line. The work includes a microprocessor based control system, new equipment cases and modifications to existing cases, new track circuits, Local Control Panels at each interlocking and a Zone Control Panel governing the entire South Yonge Area. Signals, trainstops, cables and other wayside equipment are being procured by the Commission under other contracts and installed by Commission Forces. [p. 2]

The decision to install two parallel systems – ATC for control of regular train movements plus a regular block signal system for non-ATC trains – is the origin of the complexity in the work now underway. The fixed block system would have to be operational both for non-ATC trains and as a potential backup system even though it used a fundamentally different and more restrictive method of detecting train locations and controlling their movements.

The Resignalling of the YUS is necessary to meet predicted ridership figures. Utilising ATC will provide this increase in capacity. An auxiliary wayside signalling system is necessary to allow mixed mode operation and mitigate against delay in the event of an ATC system failure. Resignalling of the South Yonge area with conventional wayside equipment is the first major stage of the ATC project. This project is imperative as the state of the signalling equipment in the South Yonge area is at its life expiry date and in deteriorating condition. [p. 3]

Missing here was any discussion of the extension of the project to provide a parallel block signalling system to the rest of the subway, nor any hint of the combined cost of many works still to follow.

TTC management were quite worried that the existing South Yonge signals would reach a point of no return and the line could become inoperable. What actually happened here was that critically needed maintenance was presented as an essential part of a capacity improvement scheme so that funding would be prioritized.

The 2008 financial crisis and the cutbacks to transit spending it triggered were already underway.

How Does ATC Increase Line Capacity?

The essential difference between the fixed block system originally used on the subway (a world-wide standard in its day), the proposed ATC system is based on the concept of “moving blocks”. This has been in place in Toronto on the Scarborough RT for 30 years, and it is not exactly a new concept.

There are three major components:

  • For detection purposes, a route is subdivided into very short sections and train positions are established through communication between onboard equipment and antennae along the tunnel or right-of-way. (Conventional track circuits can also be used for critical locations such as interlockings.)
  • A train is considered to operate in a moving and changing “bubble” of clear space ahead. This is determined by the train’s speed and safe braking distance. The lower the speed, the smaller the bubble, unlike a fixed block system that enforces the same spacing on trains regardless of their speed.
  • Train locations and speed are updated through radio links to a central computer system very frequently so that the safe speed and position of all trains reflects real-time conditions.

Because trains can run closer together safely, and in particular because operations through congested areas can be done with the minimum safe train spacing, throughput is improved and with that line capacity. The basic math is simple: number of trains per hour times capacity per train. It does not matter how fast the trains move, only that the number per hour past a point is as high as possible.

If there are locations where physical constraints impose minima in space and time, then the ability of ATC to increase line capacity is limited. For the TTC, these locations are at terminals where crossover geometry and crewing practices set a lower bound on the time needed for certain moves. For example, an incoming train must be assumed to be stopped before it reaches the crossover as a clear route into a terminal is unlikely. The time needed to start up, travel through and clear the crossover is limited by the G forces passengers can tolerate moving through the switches. The TTC has partly addressed this in more recent installations with longer crossovers and gentler turnouts, but this adds to the time needed for a 6-car train to clear the interlocking area.

Operating practices must also change to minimize delays at terminals. Trains must be ready to depart with closed doors the instant a route is clear, and step-back crewing is needed to separate train turnaround from operator turnaround (a particular requirement for one person crews who must walk a train length at each terminal). Alternate terminal geometries including far-end turnaround tracks or split terminal operations might also be required. These considerations are important and essential to getting the full capability out of an ATC system.

ATC for the Yonge-University Subway

On April 27, 2009, the Commission authorized a radio based automatic train control system for the YUS. The contract was awarded to Alstom Transportation Information and Security Inc. for $50.8m.

The Contract is for the design, supply, and technical support for the installation and testing of an Automatic Train Control (ATC) system for the Yonge-University-Spadina Subway Line. The work includes the supply of processor-based Zone Controllers that control sections of the line and communicate via a radio-based Data Communications System (DCS) to the trainborne controllers on the new Toronto Rocket (TR) trains. The Zone Controllers will also communicate with the Central Signalling System at Transit Control, and the new Computer-Based Interlocking (CBI), which is being supplied under a separate contract. [p. 2]

The CBI referred to here is the work already under way by Union Switch & Signal described in the previous section.

At this point, the TR train contract was for only 39 sets, and so the ATC contract contained only enough onboard equipment for the TR order as it stood.

This contract included provision for a significant amount of add-on work, but did not include an actual funding request:

The proposal included a submission of unit prices for the supply of maintenance spare parts. Specified Option Prices were required to be submitted for the provision of: an Automatic Train Supervision System capable of supporting Communications-Based Train Control (CBTC); additional ATC-CBTC equipment for the new YUS line extensions TYSSE and North Yonge; and additional ATC-CBTC sets of trainborne equipment required to equip the extra TR trains that will be required to provide the planned increase in train throughput that will be achievable with ATC-CBTC. Consideration for the optional prices, which are valid up to 36 months from notification of award, is subject to future funding. [p. 2]

In other words, there would be extra costs, some related to the Spadina extension project (TYSSE), but at this point the TTC was only dealing with the existing YUS. TYSSE funding was already in place as of September 2008, but the signals contracts for that line had not yet been let.

On April 6, 2011, the TTC approved a change order to Alstom for 21 sets of onboard ATC gear in the amount of $4.9m. This covered the additional TR trains the TTC decided to acquire replacing the H-6 equipment.

Extending the Scope of the Fixed Block Signal Replacement Project

On March 30, 2012, the TTC approved a contract to design and install a replacement fixed block control system, including new computer-based interlockings, for the remainder of the YUS including the TYSSE. The contract was awarded to Ansaldo STS Canada Inc. (formerly Union Switch & Signal) for $30.3m to resignal the remaining portions of YUS (Finch to Lawrence, St. Patrick to Downsview) and for $26.7m to install a fixed block system on the TYSSE.

Alstom was a bidder on this project, but they were disqualified on technical criteria. The value of their bid is unknown because they were disqualified before the evaluation process came to price comparisons.

At this point, the TTC’s clear intent remained to allow mixed mode (ATC and non-ATC trains) on the YUS, and to treat the fixed block system as a backup to the ATC one.

Refreshing the Speed Control System

On September 27, 2012, the TTC approved a contract to update the Speed Control System on the existing YUS and to provide SCS for the TYSSE.  The contract was awarded to Thales Canada, Transportation Solutions (formerly Alcatel) for $7.9m.

The SCS was designed based on the existing signal system on the YUS, and this design and corresponding equipment installation has been complete for many years. Speed control was not implemented in the South Yonge area as this area has been scheduled for re-signalling for a number of years. Unfortunately, the speed control project suffered numerous delays primarily due to software development and system reliability problems. These problems have been successfully resolved.

As part of the Commission’s State-of-Good Repair / Safety Signalling projects, Project 2.4 – YUS ATC Resignalling replaces the existing YUS signal controls with new, state-of-the- art equipment. Coupled with the equipment replacement, the signal block design will change in order to optimize train separation and improve headway. Except for the South Yonge area, this re-signalling project was initiated after the speed control design and installation was substantially completed. The new block designs will change the number of signals and their locations, which in turn, will require an update to the existing design of the SCS, as it relies on the signal location to determine its safe operating speed limits. Given that the problems with the SCS have been satisfactorily addressed, work to upgrade the system to match the new YUS block designs and to equip the TYSSE is now recommended. [p. 2]

This description gives some hint of what is planned for the new block signal system, specifically changes “to optimize train separation and improve headway”. In other words, it was recognized that the existing block signals could interfere with the provision of more frequent service. This also suggests that despite the pending arrival of ATC, the co-existence with an existing block signal system could be problematic.

One important addition to the SCS was a provision for temporary “tags” used by maintenance workers in conjunction with the “blue lights” they deploy to mark work zones. These temporary SCS tags cause the onboard controls to implement a speed restriction overriding whatever might normally be in place on the line.

At this point, there are three active signaling contracts:

  • Ansaldo (US&S) for the replacement fixed block system and computer based interlockings
  • Alstom for the Automatic Train Control system
  • Thales (Alcatel) for the updated Speed Control system

ATC for the Spadina Extension

On February 25, 2013, the TTC approved a change order to Alstom to extend the scope of the ATC project over the TYSSE at a cost of $18.4m. There was, however, a small problem with funding:

It is recognized that the TYSSE project does not have sufficient funds within the project budget to pay for the implementation of ATC on the TYSSE. Funding is being sought for this purpose from the funding partners, specifically the Province of Ontario, the City of Toronto and the Regional Municipality of York.

The contract change will not be exercised until such time as the funding has been confirmed. [p. 1]

As noted above, there was an option in the ATC contract for sundry add-ons, including the TYSSE. However:

Due to a lack of funding, the specified option could not be exercised before its expiry in May 2011. Various funding options were explored including attempts to secure the necessary funds from the province of Ontario (in April 2010, April 2011 and April 2012). At its October 2012 meeting, the Executive Task Force instructed TYSSE staff to engage Alstom and negotiate a price and implementation strategy that would minimize the cost and schedule impact to the TYSSE. [p. 2]

Two options were explored. In one (“Green Field”), the ATC equipment would be installed and tested on the TYSSE before revenue operation began. In the other (“Green/Brown Field”), ATC installation would precede revenue operations, but testing and commissioning would occur after. The Executive Task Force (the controlling body for the TYSSE project) opted for the “Green/Brown” option thinking it would be cheaper because the TYSSE opening would not be delayed. That evaluation is ironic given what we now know of the state of the project.

It appears that external funding never arrived because this project, including the TYSSE work, is shown in the 2015 capital budget as 100% funded by Toronto. This additional cost is not included in the TYSSE project cost review because it has not been booked under that project’s budget.

Resignalling Wilson Yard

On February 24, 2014, the TTC approved contracts for the resignalling of Wilson Yard and associated works. These were awarded to Thales in the amounts of $11.8m for TYSSE effects on the yard and $24.4m for the Wilson Yard signal system.

TYSSE Wilson Yard Connection and Wilson Yard Signal System Modifications

  • $11.8m total
  • 66.7% charged to the TYSSE
  • 33.3% charged to the TR/T1 Rail Yard Accommodation Project

TR/T1 Wilson Yard Resignalling

  • $24.4m total

[pp 1-2. The breakdown above has been simplified slightly from the report for clarity.]

To put this into context: the original signal system supplied for the Spadina Subway has always been a problem for the TTC. Because of the significant changes at Wilson Yard and the fact that the signals on the remainder of the line are being replaced, this was an ideal time to extend the work into the yard itself.

Oddly, the TTC chose a vendor who was not also working on the resignalling of the line itself (Ansaldo bid high, and Thales did not bid).

Major Revision to ATC Contract Scope

On April 30, 2014, the TTC approved a large contract change to Alstom at a cost of $37.8m with the following effects:

  • Cancellation of the previously approved change order for ATC on the TYSSE (see above)
  • Inclusion of 10 additional sets of ATC train control equipment for the TR sets that were on order for future growth in demand
  • A new change order for ATC on the TYSSE at a net cost of $32.6m including a five-year schedule extension

As noted earlier, this project is funded not from the TYSSE budget, but from a separate project line that is 100% Toronto money (net of any flow through subsidies allocated to this project).

The project had encountered severe problems as outlined in the report:

The contract for ATC on the entire Line 1 was awarded prior to a confirmed schedule for replacement of the existing signal system on the entire Line 1. The contract completion date of August 2015 for the ATC contract was based on best estimates at the time, for completion of the existing signal system replacement on the entire Line 1. The delays associated with Phase 1 replacement of the existing signal system have subsequently delayed the completion of the ATC for the entire Line 1 and TYSSE.

In order to continue with the ATC contract in a phased integrated sequence with the signal system replacement work, the ATC contract schedule must be completed under a new schedule that revises the completion date from August 2015 to July 2020. This schedule was developed based on working with both the signal system replacement Contractor (Ansaldo) and the ATC contractor (Alstom).

The overall ATC schedule requires extension from the original six years to 11 years. Completion of ATC work on the Line 1, including ATC on the TYSSE line, is based on a revised six phase implementation approach. This revised schedule addresses a number of complex scope and scheduling issues between the signal system replacement and ATC contracts of which some of the more significant issues are as follows:

The difficulties of introducing ATC to an operating subway service without causing any extended delays to normal revenue service, while continuing the maintenance of an old unreliable system, were originally underestimated in terms of technical and implementation complexity;

The unplanned impact of other state of good repair programs to existing subway facilities required for the ATC and signal system replacement equipment have arisen, such as north Yonge asbestos abatement and deferral of Davisville Area Rehabilitation Program (track reconstruction between St Clair and Eglinton stations);

A new test strategy to significantly increase testing of signal system replacement subsystems independently and then in parallel with ATC prior to a combined commissioning. This mitigates unacceptable risks associated with commissioning two new systems simultaneously;

The introduction of an extra phase of commissioning. Splitting the original Phase 4 commissioning was required to reduce the risks associated with the commissioning process during service closures. This introduced more software releases;

Interdependencies of ATC on Line 1 and TYSSE require the introduction of ATC onto TYSSE to be completed sequentially and not in parallel, thus adding additional time to the overall schedule requirement. This change was not anticipated or known prior to the award of the ATC contract to Alstom;

Increasing the fleet of TR’s from 70 to 80 (approved by the Board at the March 26, 2014 meeting); and

The increased number of computer software baselines due to additional phases and revised implementation strategy that requires extensive software programming. [pp 4-5]

All of this speaks not only to the complexity of project management within the multi-vendor environment of the signal project itself, but with other work planned and in progress on the subway line.

Another troubling component to this report is that “you must approve the contract now to lock in the price”. The TTC Board might have misgivings, but it has no scope to question what is proposed. This situation was repeated a year later.

Consolidating the Contracts

On March 26, 2015, the TTC Board approved a major reorganization of the signaling contracts such that all of the work would now be under one vendor, Alstom. The cost of this change is priced at $74.6m, although this is expected to be offset by savings from cancelled contracts and from simplification of project management and integration.

This will have the following effects:

  • The original scheme whereby a conventional fixed block system would be used to handle non-ATC trains and act as a backup in case of ATC failure has been abandoned. ATC is now treated as sufficiently robust in its own right that a backup system is not required.
  • Provision of computer based interlockings will now be done by Alstom instead of Ansaldo.
  • Provision of speed control will now be integrated with the ATC system under Alstom instead of Thales.
  • Work cars that will operate on the system will be ATC equipped so that they can be monitored by the central signal system in the absence of conventional block signals.

It appears that the work related to Wilson Yard is still in Thales’ hands.

The project’s overall complexity outlined above has proven even worse than thought:

ATC System – Moving Forward Plan

In almost all subways around the world, signal system replacement and ATC contracts are between the Subway owner and a single supplier. However, given the age (60 years) and the condition of the signal system on Line 1, a different approach was undertaken by TTC in 2008 to separately tender the CBI and ATC contracts. The CBI contracts are also separated into three contracts – Phase 1, Phase 2-4 and TYSSE. As this implementation evolved, delays occurred to the Phase 1 CBI schedule and incompatibilities arose between the two main suppliers, Ansaldo and Alstom. These incompatibilities would result in TTC not getting the maximum benefits from ATC and potentially an unacceptable signaling solution in that the capacity issue, i.e. the number of trains per hour, currently being offered doesn’t meet the capacity requirements of the City of Toronto. This would result in an inadequate customer experience should the current arrangement continue. There are also significant risks to cost and schedule if the TTC continues with the current arrangement as further technical compatibility issues continue to be discovered between the two suppliers. [p 6]

To review this situation, the TTC retained Dr. Alan F. Rumsey, P. Eng, Vice President, Rail and Transit Systems of Parsons. (See the report for his resumé.) His report was distributed at the Board meeting but is not yet online. A short summary is included in the public report.

As a result of this review, the TTC will move to a single supplier and single technology – Alstom’s ATC – for the subway, and will not install a parallel block signal system. Interlockings will continue to have computer-based management, but with Alstom as supplier and integrator to the ATC system.

This work will take the TTC up to 2020, and barring a complete screw-up of their contract, Alstom would obviously be the chosen vendor for the Bloor-Danforth line’s ATC conversion and upgrade in the early 2020s.

What remains unclear is how a 2020 completion date for the revised Yonge line’s ATC project meshes with a 2017 opening date for the TYSSE or if the migration to ATC operation will be phased in with some sections remaining under the old block signal system for a period of time. I will pursue these issues with the TTC.

As on the TYSSE project itself, what we do not have is a consolidated tracking of the various contracts, change orders and funding sources. These would be essential to any project management, not to mention oversight by the TTC Board.

Updated April 4, 2015 at 6:00 am:

The review by Parsons of the alternative options for consolidation of the signalling contracts has been added to the TTC report online [begins at p. 18 of the linked pdf]. This report does not address the “how did we get here” problem which is beyond its scope, but does review the goals and potential risks of various ways to reorganize the project.

Given the evolution of various contracts over time, it is important to remember just what this project is supposed to achieve.

  • Headway/Capacity

The ATC project is to deliver a capability for a “step change” increase in passenger carrying capacity on the line through Automatic Train Operation (ATO) and the use of a modern, service-proven, moving block, train control solution supporting safe, short-headway operations.

A capability for a sustained operating headway of the order of 105 seconds (approximately 34 trains-per-hour) under driver-supervised ATO, is desired.

The train headway will always be constrained by the safe train separation requirements of the Automatic Train Protection (ATP) functions of the ATC system, with the recognition that the achievable headway involves certain factors that are outside the control of the ATC system, such as track alignment, gradients, civil speed limits, train acceleration and braking rates, station dwell times, terminal track configurations, etc. [p. 7 of report, p. 27 of pdf]

This goal includes two key points. First is that the new signal system should permit a sustained headway of 105 seconds yielding an hourly capacity of 34 trains. This is very different from permitting shorter headways for shorter periods where the dynamics and track geometry permit. At times during discussion of what signalling might achieve, there have been citations of 90 second headways although these generally cannot be achieved on a sustained basis on most rapid transit lines. The 105 second goal translates to an hourly design capacity of 37,400 assuming that, for planning purposes, the sustained capacity of trains over the hour is 1,100.

The TTC has talked of moving to 7-car trains on YUS and this would add a theoretical 10% to capacity. However, such a change may have effects elsewhere in the system because the entire network’s design (including carhouses, pocket tracks, etc.) was based on 450-foot long trains and 500-foot long stations. The TTC has not recently touted this option when discussing future line capacity on the YUS.

  • System availability: The new system should be extremely reliable with a very low failure rate, the ability for remote diagnostics of problems, and a short mean time to repair.
  • Maintainability: Maintenance requirements should be minimized by reducing the amount of equipment installed along the line (as opposed to in centralized equipment rooms).
  • Automatic Train Protection (ATP): Fail-safe protection against hazards
  • Automatic Train Operation (ATO): The ability to perform all of the functions now handled manually by the operator
  • Automatic Train Supervision (ATS): Schedule and headway management
  • Increased operational flexibility such as support for bi-directional train operation
  • Online by 2020
  • ATC should not affect the opening of the Spadina extension (TYSSE)
  • Provided within the currently approved budget and with minimized life-cycle costs

It is self-evident that a 2020 target date for ATC  and the current plans for opening the TYSSE at the end of 2017 do not line up. Although this is not discussed in the report, a temporary operational plan could involve a simplified “protected mode” operation of the extension pending completion of the ATC project. There are no details of what this will entail.

As discussed in the main article, there are two implementation scenarios:

  • Scenario “A” provides for mixed operation of trains with and without ATC gear on board and the existence of two parallel signal systems to manage each type of movement while protecting both types of train from each other.
  • Scenario “B” provides an ATC-only environment in which all control functions rest with one system and any trains moving over the line during revenue service hours would be equipped with ATC gear.

Various configurations of contractor and project management are combined with the two scenarios, and it is no surprise that the preferred combination includes the simplest options: an ATC-only signal system and a single provider of technology and project management.

  • Continued provision of separate systems for automatic operation, conventional block signalling and speed control require interfaces between each of the systems, extensive design and testing before implementation, and additional maintenance requirements in operation.
  • The use of multiple vendors places the TTC in a project management and system integration role. Although the report does not say this explicitly, this requires a degree of expertise in signalling technology and understanding of multiple proprietary systems that the TTC may not possess.
  • The addition of any new vendor poses problems of familiarity with the existing and planned technologies, not to mention yet another potential risk of a multi-party environment for technology compatibility and project management.

The risk assessment process is fairly straightforward: assign a severity to each of the goals (is it critical to the project, or merely important), and evaluate the risk of various possible problems/failures within the project. This produces compound scores for each permutation of operating and provisioning giving a more formal view of the benefit of a single technology, single provider project. [See Sections 5.10 and 5.11 beginning on page 53 of the report, page 73 of the pdf.]

Not included in this report, although it is a stated goal of the project, is that the work be done within the existing project budget. The covering staff report claims this will occur, but does not provide details. These are likely subject to negotiation of the termination of existing contracts and the takeup of the full project by Alstom.

19 thoughts on “The Evolution of TTC Signaling Contracts (Updated)

  1. So, the geek in me wants to know how, in the rare occasions the St. Clair, Dundas/College, and King crossovers were used, the trains were safely allowed in and out of these zones. Why, in fact, were they installed in the first place? I had been told by an old timer or two that there was still a bit of streetcar thinking going on with regards to potential short turns.

    Steve: That sounds plausible, but it could also be that someone decided to save money. In any event, the power gaps were not set up properly because the crossovers were the break between sections. If the power was off, say, south of College, the crossover there could not be used. When Bloor was electrified, new gaps were created south of Bloor platform and half way around the curves to Rosedale. This allowed an isolated power feed to both sides of the crossover.


  2. To allow trains and workcars not equipped with ATC equipment to operate over the YUS Line and to mitigate against delay in the event of an ATC system failure, a new conventional auxiliary wayside signalling system is required to be installed.

    There are situations when it’s better to put all your eggs in one basket, and this is surely one of them. I can’t understand why the TTC would suffer from such a severe lack of confidence in the ATC system — technology that has been proven for several decades — that they would view it worth the cost and complexity to maintain two different signalling systems. It’s unfortunate that it took so many years to reconsider this decision.


  3. James said:

    “There are situations when it’s better to put all your eggs in one basket, and this is surely one of them. I can’t understand why the TTC would suffer from such a severe lack of confidence in the ATC system — technology that has been proven for several decades — that they would view it worth the cost and complexity to maintain two different signalling systems. It’s unfortunate that it took so many years to reconsider this decision.”

    I wonder whether it is a lack of confidence in the actual signalling system, or a lack of confidence in actually having a large enough and reliable enough fleet of car sets, and maintenance equipment that will be ATC capable? I wonder whether the TTC lacks confidence, in both the security of their budget to fill the details like maintenance cars and the like, and their own ability to ensure they are ready from other managerial perspectives.

    I look at the fact that basic maintenance has been so delayed, and certainly the budget issue, would be a valid concern (and this would be reinforced by the relative shortage of new buses or orders for them). The issues with Streetcar delivery also points to the concern, that something would fail and thereby mean they were not ready to do ATC, when they really needed to be.


  4. James said:

    “I can’t understand why the TTC would suffer from such a severe lack of confidence in the ATC system — technology that has been proven for several decades — that they would view it worth the cost and complexity to maintain two different signalling systems.”

    The last time the TTC had absolute faith in a subway signalling system that had been proven for several decades, it resulted in the Russell Hill crash. Even though it’s been almost 20 years, the TTC is still paranoid over another accident like that happening.


  5. Steve wrote:

    When a train passes a signal, the trip arm does not immediately rise after the last car clears it, but only after the train has cleared the entire block. This is related to a design that allows trains to approach a red block signal and pull right up to it thereby entering its track circuit (which actually begins just ahead of the signal).

    I don’t mean to argue, as my original understanding may have some inaccuracies, but I was under the impression that the feature of automatic key by was a side effect of a necessary requirement of how trip arms work, combined with positioning the block boundary slightly in advance of the signal and trip arm.

    When a train passes into the next block, the signal changes to red, but the trip arm must remain in the lowered position because prior to TR trainsets, there would be a trip cock at the front of each car with a cab facing forward (i.e.: the third and the fifth car). Only when the block in advance of the signal is clear will the trip arm rise.

    This also forces the need to have two red signals behind an occupied block. If a train were to enter the block immediately behind a train, then both the block in advance and the block in rear of the signal would be occupied and would cause the trip arm to drop. Performing an automatic key by is only possible because the insulated rail joint that forms the block boundary is a couple of metres in advance of the signal. Moving slowly over this boundary causes both blocks to become occupied and the trip arm drops.

    Steve: We are both describing the same thing, just in slightly different ways.


  6. I watched a video from japan where a few thousand labourers and engineers converted a surface train station into a lowered subway station in a matter of HOURS. They started at 8PM and the train was running on the new track at 8AM the next morning.

    Why is the signal replacement on TTC not taking a similar approach? Instead of doing a little bit of work every month over a 10 year span, just get a huge number of engineers & labourers to install new signalling system over a weekend or two?

    Steve: Because replacing a signal system is a lot more delicate a process than a station rebuild. The million monkeys approach does not work here. Indeed, if it were practical on a general basis, we could build entire subways in weeks.


  7. Kentlee said:

    I watched a video from japan where a few thousand labourers and engineers converted a surface train station into a lowered subway station in a matter of HOURS.

    Did your video also show the months of enabling works? Construction at the station has been ongoing since 2008 (video of March 2013) and half the station has been closed for the last two years. It’s a good YouTube story, but it is pretty standard engineering practice to minimize impact on mainline operations.


  8. Looking at it objectively (i.e., without considering the history of the Russell Hill crash), couldn’t the TTC reinstate automatic key-by at the last signal before station platforms? I can’t see what the risk would be for that specific application, as by that point it would be visually apparent whether the next block (the platform) is clear or not, and the train is not proceeding beyond the platform into the subsequent block. (In addition to helping with peak capacity at pinch points like Yonge and Bloor, I have also thought it would be nice on the occasion that I have been stuck at the last signal eastbound before getting off at Vic Park… where you can see the bus bays from the train and wonder if you are going to miss your connecting bus!)

    I know it’s not going to happen (and will be a moot point anyway if the re-signalling ever gets completed), but is at least an interesting thought exercise.

    Steve: Actually at Bloor this change is not as easy as it seems. This area is now an interlocking, and automatic key bys are not allowed at double red signals. An automatic call on can be built into the signal system (the flashing amber one sees at terminal approaches), but this would require changes to hardware that is about to be replaced.


  9. Kentlee wrote:

    Instead of doing a little bit of work every month over a 10 year span, just get a huge number of engineers & labourers to install new signalling system over a weekend or two?

    How long something will take is not linearly related to the number of people available to do the task. There is a law of diminishing returns and the more complex the task, the greater this effect.

    I know of one daily task in my home where two people can get it done in about 25-35% of the time that one person can do it, but any more will not shorten the time needed. With complex tasks, such as changing a signalling system, there is a point where additional bodies get in the way and actually make things take longer.

    Also, there can be outside events, such as testing and break-in periods, where other tasks must wait. More people simply means paying more people to stand and wait.


  10. Sorry if the question was answered before. I know Alstom has plenty of experience to upgrade signal system to CBTC WITHOUT closing the system. I have heard their CBTC is compatible with the old system. Why Toronto need to close the system to upgrade their signal system? It is primary because of cost?

    Steve: I don’t know how any new CBTC could be “compatible” with the old system that dates from the 1950s. With a relatively new block signal system, maybe, but the whole idea of a digital system simply did not exist when the Yonge line was built.

    Liked by 1 person

  11. Alstom CBTC is being used in Toronto. The “no closure” claim comes from 2006 to 2008 work on Line 2 in Beijing. They used 3-hour night shifts during their normal 11PM to 5AM closure period. The TTC is 90 minutes shorter from 1:30AM to 6AM. Also, Line 2 was finished in 1987.


  12. I did not find anything useful on Alstom but I research on Siemens’ “Trainguard MT CBTC” system. NYC L line (operate 24/7) successfully implemented it few years ago. This really surprises me…

    In addition to that, I research on the compatibility of the CBTC system. I find out Siemens, Bombardier, Alstom and Thales all support overlay re-signaling. Alstom even commits to “providing disruption-free installation and commissioning.” Again, why TTC can’t do it?

    Steve: Don’t forget that the TTC was doing two upgrades at the same time with different vendors providing the technology. The original block signal system was being replaced by Ansaldo while Alstom was doing the ATC work. On top of that, the Thales Speed Control System was also to be upgraded. The decision to have three parallel projects from different vendors created conflicts that were difficult to resolve. This has now led to consolidation of the contract under Alstom.

    The project should have been scoped and managed from the outset as one job with one vendor, but in the early days the TTC was getting a lot of pushback on big capital projects that did not include system expansion. They “sold” the first contract, resignalling the original Yonge line, as a pre-requisite for capacity increase in anticipation of the Richmond Hill subway. While I must question the wisdom of how the work was split up among multiple vendors and with overlapping designs, a related problem is the changing political environment in which management tried to “sell” bits of projects as being part of whatever the pet scheme of the day might be.

    Liked by 1 person

  13. This might be little bit off topic. I think in Toronto maybe it is time to consider outsourcing the subway operation to a experienced company. In my option, TTC don’t have resources and experience to improve the subway system and service. This signal upgrade project is not methodical; the public always complains about the subway. Then why not just outsource the subway operation to someone that commits to improve the service quality? Why not give this job to someone that has more experience in terms of upgrading systems, and has more experience to run the system (Like Hong Kong MTR and Veolia Transport)?

    Stockholm metro is a good example: the city outsourced the operation to MTR Corporation while keeping the same fare system and brand. Hong Kong MTR’s on time performance is 99.9%. The company improved the subway service in Melbourne significantly, and they are going to run the London Crossrail next month. This works all around the world, and why not Toronto?


  14. asiaworldcity said:

    Then why not just outsource the subway operation to someone that commits to improve the service quality? Why not give this job to someone that has more experience in terms of upgrading systems, and has more experience to run the system (Like Hong Kong MTR and Veolia Transport)?

    Stockholm metro is a good example: the city outsourced the operation to MTR Corporation while keeping the same fare system and brand. Hong Kong MTR’s on time performance is 99.9%. The company improved the subway service in Melbourne significantly, and they are going to run the London Crossrail next month. This works all around the world, and why not Toronto?

    I’m not sure how MTR improved subway service in Melbourne, when they don’t have a subway. Stockholm runs on 10 minute headways. Even the London Crossrail won’t be in full operation until 2018. MTR (1975) has much less experience than the TTC (1954).

    The main issue with the Toronto subway is that it’s been starved for cash for decades. If we gave the TTC the same funding as an outside contractor would want, we’d have higher levels of supervision and less service irregularities.

    Liked by 2 people

  15. Matthew Phillips said:

    “The main issue with the Toronto subway is that it’s been starved for cash for decades. If we gave the TTC the same funding as an outside contractor would want, we’d have higher levels of supervision and less service irregularities.”

    I would also suggest that the type of funding they receive makes it harder to approach other issues beyond the subway as well. Where perhaps there should be a long hard look at how to help drivers know whether they are on headway, the money is not there, therefore the workforce will believe it is not important, and not make a real effort either. The money and measures of performance need to be present, in order to make a system work well, regardless who is doing it. Toronto, should be look at how to make the system more efficient, not just looking for “efficiencies” . If you want to get more out of your buses, without adding staff, – you make them run faster (ie BRTs and Signal priority), you get the spacing, so the drivers need to know they are closing up or gaping out, so they can react themselves without having to have a person standing beside the route to let them know.

    Steve: Actually, with the new vehicle location system the TTC is installing over the next few years, it should be possible to give drivers info about headway maintenance. The question, however, is whether this will be included in the design or omitted because of the TTC’s fetish for “on time performance”.

    Liked by 1 person

  16. Steve said:

    “Actually, with the new vehicle location system the TTC is installing over the next few years, it should be possible to give drivers info about headway maintenance. The question, however, is whether this will be included in the design or omitted because of the TTC’s fetish for “on time performance”.”

    Yes, it will be critical whether they include it, and whether they pay attention to as a real metric. On time performance in my mind is without meaning on a headway governed system, and I really do think if you have enough service for 6 buses per hour they should be (where ever possible) set up to be every 10 minutes, and run on a headway basis. Headway, then should be the measure – and 2 buses 2 minutes apart is bad cause the next bus does come for 18.

    Steve: In an even more bizarre turn, it has been argued that ops should not be told if they are on time or not because they dry to drive more quickly (sometimes speeding – gasp!) to stay on schedule. Therefore, some in the TTC actually think that it’s a bad idea to try running on time because it is unsafe. Of course this presumes that the schedules are reasonable in the first place.

    Anyone who rides the system knows that operators drive at considerable speed for reasons other than being “on time”, notably to get longer breaks at terminals.


  17. Yes, there is a lack of apprication for money that has no visible effect, but can greatly improve the quality of service.


  18. Matthew Phillips said:

    “Yes, there is a lack of appreciation for money that has no visible effect, but can greatly improve the quality of service.”

    The problem this would also be true of a really involved signal priority system. I suspect [it] is also one of the reasons no space was saved in the former railway lands for a route for an LRT. Doesn’t look like much so well…


  19. You never hire multiple signalling contractors for the same line (or even lines which interact). I can’t imagine what the TTC was thinking. Signalling is fundamentally a coordination problem; as such, having multiple contractors just makes the coordination problem worse. You want one central “signalling czar”.

    It’s actually best to do signalling mostly in-house, but few agencies have the expertise these days (well, actually that’s been true for roughly 100 years).

    Anyway, this made for an easy recommendation by Parsons: tell the TTC to adopt standard practice!


Comments are closed.