[EnviroTO]

I'm surprised that in 2025 they haven't commoditized ATO software and the various parts that interact with it. Shouldn't it be a bunch of standard components (actuators, sensors, control circuits) you plug into a standard system and then define system speed and timing limits for? Two trains on tracks (no ability to steer, two dimensional plane, limited intersections and switches) can't be kept from colliding? I don't get it.

 

[Steve X]

I'm surprised that in 2025 they haven't commoditized ATO software and the various parts that interact with it. Shouldn't it be a bunch of standard components (actuators, sensors, control circuits) you plug into a standard system and then define system speed and timing limits for? Two trains on tracks (no ability to steer, two dimensional plane, limited intersections and switches) can't be kept from colliding? I don't get it.

From what smallspy describes the problem, it doesn't sound like a hardware issue. Through simple train detection (axels shorting the rails) you can easily detect a train is there without any fancy sensor. A train would not start moving on its own unless it has proper communication with the master control computer. Trains would not collide if it knows another train is already there AND blocks are properly mapped.

I write this as a generalization and might be be exactly how the system on the line works. ATO works with every piece of track in the system mapped into blocks. A train can occupied block(s) depending on the length of each block. A bubble (safety zone) would be defined around each train. When a train approach another slower or stopped train, they would get the command to slow down and/or stop so the "bubbles" don't overlap. A train without ATO or failed system would still be detected as something on the track and thus any approaching train would be told to stop. This is how "ghost" trains appear in the winter when debris or ice buildup shorts the rails.

Since trains have collided, that usually means the blocks are not mapped correctly. Maybe the master control thought the train was further away or on a different track cause SOMEONE programmed it incorrectly. What he was saying sound like they need to manually check that each block is actually mapped to the trackage it describes. If they need to do this for every foot of the 19 km line, it will take some time.

 

[Admiral Beez]

Since trains have collided, that usually means the blocks are not mapped correctly. Maybe the master control thought the train was further away or on a different track cause SOMEONE programmed it incorrectly. What he was saying sound like they need to manually check that each block is actually mapped to the trackage it describes. If they need to do this for every foot of the 19 km line, it will take some time.

If the collision was in the yard, wouldn't those trains be under local control, with the operator having to manually drive their vehicle into another one? Perhaps the operator hadn't noticed a track switch setting?

I wonder what happened in France earlier this year. One of the trams was apparently going backwards.

In Oz a tram operator seemingly smashed their tram into the one in front.

And then we have a Norwegian tram running a turn at high speed and driving into a shop. The report suggests a technical fault rather than operator error.

Nevertheless, all three systems above are still operating. Would the TTC shut down the Crosstown if something similar occurred?

 

[crs1026]

Ah, the joys of software....one misplaced comma in the code, and everything stops dead.

It does beg the question of whether the signalling system is vanilla CBTC, or some hybrid that ML has complicated by design spec or change orders. Or is it simply a typo in a database? It is reasonable to expect that the industry has working technology that ought to be deployable without incident right out of the box.

I don't know what the tipping point is to force a Royal Commission on this whole project. If we aren't already there, we ought to be.

- Paul

 

[T3G]

If the collision was in the yard, wouldn't those trains be under local control, with the operator having to manually drive their vehicle into another one? Perhaps the operator hadn't noticed a track switch setting?

No, there are sections of the yard where trains drive themselves.

It baffles me as to what value this adds. This driverless crap is getting real stupid. Your scientists were so preoccupied with how, they never stopped to ask themselves why.

 

[smallspy]

View attachment 690070
Just so I get this right - Line 5 Eglinton will be GoA 2 (automatic train operation, similar to Line 1) in the tunnelled section and GoA 1 on the surface. In the yard, it will be GoA 4 (unattended train operation?

Sort of. It needs an asterisk.

A line operating as GoA4 with the public needs to have a lot of additional systems and features in order to ensure that the public remains safe in all situations. And that includes emergencies that require evacuating a train. The Crosstown does not have those.

The system operates as a GoA4 in the yard, because the public will never ride trains in there - so it doesn't need those additional features and systems.

Just build a real, heavy rail SUBWAY under Eglinton Avenue in Toronto. Forget this tram train trouble it is not worth it and will never work as originally imagined.

You know that subways use signals too, right? This is not an issue specific to the Crosstown or its equipment.

Ah, the joys of software....one misplaced comma in the code, and everything stops dead.

It does beg the question of whether the signalling system is vanilla CBTC, or some hybrid that ML has complicated by design spec or change orders. Or is it simply a typo in a database? It is reasonable to expect that the industry has working technology that ought to be deployable without incident right out of the box.

I don't know what the tipping point is to force a Royal Commission on this whole project. If we aren't already there, we ought to be.

- Paul

The system is an off-the-shelf signal system, and is used in dozens of other systems worldwide.

A large part of the issue, however, is that these systems need to be thought more, as an application, like Excel - the software gives the the framework with which you configure it for your use.

As well, the problems experienced by the Crosstown are certainly not unique to the system that has been installed on it. There have been lots of issues over the years, including collisions, on lots of other vendor's systems. It all comes down to set-up.

It baffles me as to what value this adds. This driverless crap is getting real stupid. Your scientists were so preoccupied with how, they never stopped to ask themselves why.

It's pretty easy if you think about it - cost avoidance. Don't have to pay for nearly as many hostlers to shuttle trains around the yard.

Dan

 

[crs1026]

A line operating as GoA4 with the public needs to have a lot of additional systems and features in order to ensure that the public remains safe in all situations. And that includes emergencies that require evacuating a train. The Crosstown does not have those.

The system operates as a GoA4 in the yard, because the public will never ride trains in there - so it doesn't need those additional features and systems.

The optimist in me thinks that maybe it's a good thing that the incident happened in the most complex level of the system. Possibly this implies that the system level that protects passengers is robust and dependable, and it's only the higher tech segment that has a flaw. How's that for looking for a silver lining in all this.

As well, the problems experienced by the Crosstown are certainly not unique to the system that has been installed on it. There have been lots of issues over the years, including collisions, on lots of other vendor's systems. It all comes down to set-up.

It's not extravagant to expect that the software that provides safety to thousands of transit riders every day be as well qualified and tested as avionics software. The risk level is comparable. I wonder what the process is to do this qualification. As we saw with Boeing, accepting incidents as part and parcel of perfecting software is not really acceptable, although it may be reality.

It's pretty easy if you think about it - cost avoidance. Don't have to pay for nearly as many hostlers to shuttle trains around the yard.

I will digress and post this nostalgiac shot of the hostler complement at Roncesvalles Division coming to a complete halt due to the non-operability of PCC car 4580 that was being moved to dockside for sale to Egypt.
I dunno, letting a bunch of humble people earn a living by hostling cars is not the worst outcome, IMHO - but I don't have the exact cost comparison.

- Paul

 

[Admiral Beez]

I will digress and post this nostalgiac shot of the hostler complement at Roncesvalles Division coming to a complete halt due to the non-operability of PCC car 4580 that was being moved to dockside for sale to Egypt.

I wonder if our PCC's are still running. The climate should be forgiving.

Trams in Alexandria - Wikipedia

en.wikipedia.org

 

[T3G]

I wonder if our PCC's are still running. The climate should be forgiving.

Trams in Alexandria - Wikipedia

en.wikipedia.org

Long gone. The last of them were withdrawn in 1984.

The climate may have been forgiving, but the maintenance conditions were anything but. And some units never even got to enter service and were destroyed in the 1967 Arab-Israeli war, too.

 

[jordanmkasla2009]

Long gone. The last of them were withdrawn in 1984.

We have a couple preserved but they're currently at the Halton Country Radial Railway while they do upgrading work at Hillcrest to reactivate it as a carhouse and storage yard.

 

[cplchanb]

cant they run on manual control in the tunnels until they can transition in the fix if it is indeed a bug? why must they rely on ATO in the tunnels? run it like they do at queens quay.

 

[T3G]

We have a couple preserved but they're currently at the Halton Country Radial Railway while they do upgrading work at Hillcrest to reactivate it as a carhouse and storage yard.

Without wanting to drag this discussion too far off topic, I was talking about the cars in Alexandria. We used PCCs in Toronto until 1995.

 

[nfitz]

cant they run on manual control in the tunnels until they can transition in the fix if it is indeed a bug?

They probably could after they spend $1 billion and 10-years to install the necessary signalling system.

2040 opening anyone?

 

[jordanmkasla2009]

It's not extravagant to expect that the software that provides safety to thousands of transit riders every day be as well qualified and tested as avionics software. The risk level is comparable. I wonder what the process is to do this qualification. As we saw with Boeing, accepting incidents as part and parcel of perfecting software is not really acceptable, although it may be reality.

- Paul

Routes within the yard track are far more complex than out on the main. With ~50 or so switch points, there are quite literally hundreds of possible routing combinations so it's not unlikely for one of those routing combinations to have an overlooked checkbox that should be checked but isn't, or be unintentionally configured to interlock the wrong switch. The fact that it's been discovered now probably has more to do with the controllers at Davisville doing regular fleet rotation movements or whatever and finding an edge case in those route configurations that wasn't discovered until they started doing RSD.

 

[A 6264]

Without wanting to drag this discussion too far off topic, I was talking about the cars in Alexandria. We used PCCs in Toronto until 1995.

Less than 10 years ago I was on a PCC car on Carlton. It was during rush hour, and I presumed they had a few old cars for when extra are needed. Or was I dreaming?