Should Merseyrail move to Driverless Trains of some description

[zwk500]

Note: Following this thread on the Kirkby Derailment https://www.railforums.co.uk/thread...eyside-13-03-2021.215146/page-20#post-5796876 going off-topic, I thought I'd set up a separate thread to discuss the musing of driverless trains on Merseyrail.

As a quasi-metro that is largely separated from the national network, some grade of driverless operation would seem feasible. I don't know the system or area very well, but DLR-type Onboard supervisors qualified to drive the train would seem to be a sensible compromise to me. There would be a passenger-facing member of staff on the train who can move it to a station if there was an emergency or failure, and you could even avoid the issues of driverless trains interacting at fringe areas like Chester by getting the OBS to drive the less busy section, when DOO is fine. The OBS would also be able to take the train over for the movements on and off depot if they required drivers.

Thoughts?

 

[Bletchleyite]

Complex due to the number of level crossings and the likes, so probably not, no. Or rather, not yet; there will come a point when driverless trains are feasible on the whole network - my view is that we'll see a lot of "guard only operation" within 20 or so years.

 

[zwk500]

Complex due to the number of level crossings and the likes, so probably not, no.

With Obstacle Detection and the generally lower speeds Merseyrail runs at, are Level Crossings really that big of a problem? I do also feel that with a certain amount of publicity ('these trains don't have drivers, there's nobody to stop for you!') people will have a bit more respect of the level crossings.

Or rather, not yet; there will come a point when driverless trains are feasible on the whole network - my view is that we'll see a lot of "guard only operation" within 20 or so years.

I suspect you are right, but not until ETCS is in place.

 

[Bletchleyite]

I suspect you are right, but not until ETCS is in place.

ETCS and obstacle detection fitted to trains that doesn't detect a small tree growing in the six-foot but does detect a branch across the running line?

I think 20 years is a reasonable horizon to expect that tech to exist.

 

[niggill617@gma]

One of the issues with automation on the rails is the old enemy adhesion. I suspect double variable rate sanders and electro magnetic emergency track brakes will win the day.

 

[D365]

One of the issues with automation on the rails is the old enemy adhesion. I suspect double variable rate sanders and electro magnetic emergency track brakes will win the day.

sanders yes, but I doubt magnetic on a mainline.

 

[MattRat]

With how trying to remove the guards has gone, I do feel they'll target the driver next. Less reasons to be against the removal of drivers compared to guards, as all they do is drive.

 

[Bletchleyite]

With how trying to remove the guards has gone, I do feel they'll target the driver next. Less reasons to be against the removal of drivers compared to guards, as all they do is drive.

The tech simply isn't there yet - unlike the DLR, Merseyrail is not a fully segregated system.

The "target" to fund guard retention will be booking offices.

 

[MattRat]

The tech simply isn't there yet - unlike the DLR, Merseyrail is not a fully segregated system.

The "target" to fund guard retention will be booking offices.

I didn't mean to suggest it would happen right away, but I do suspect it'll happen in the future when the tech is available. But you make an excellent point is that in the short term, they will look to other alternatives.

 

[8A Rail]

A simple answer is no. Why Merseyrail as you stated you dont know the region? May be better asking as a more general question which includes various London underground lines, Tyne and Wear Metro, Glasgow Metro etc. For that matter various tram lines too? I have no doubt sometime in the distant future, the technology will be there but not in the remainder of my lifetime for sure.

 

[Bletchleyite]

A simple answer is no. Why Merseyrail as you stated you dont know the region? May be better asking as a more general question which includes various London underground lines, Tyne and Wear Metro, Glasgow Metro etc. For that matter various tram lines too? I have no doubt sometime in the distant future, the technology will be there but not in the remainder of my lifetime for sure.

Glasgow is going driverless - indeed completely unstaffed, though I bet there'll be ASB so they end up putting rentathugs on instead.

 

[zwk500]

A simple answer is no. Why Merseyrail as you stated you dont know the region? May be better asking as a more general question which includes various London underground lines, Tyne and Wear Metro, Glasgow Metro etc. For that matter various tram lines too? I have no doubt sometime in the distant future, the technology will be there but not in the remainder of my lifetime for sure.

The question stemmed from a discussion of the Kirkby derailment. London Underground, absolutely yes should be driverless. Glasgow Metro is mentioned about as going to ATO. Tyne and Wear would be nice but I think the interface with NR would be operational problematic to resolve without a driver in the front. However the T&W could fit quite well with my suggestion for Merseyrail of having an OBS who is qualified to drive, and takes the controls for sections identified as high-risk or operationally required (e.g. the NR shared sections, or sections with lots of Level Crossings). This would permit One-Person operation of the train, but give the benefits of ATO in the busiest sections whilst preserving the ability to move the unit should there be problems. There wouldn't be questions of leaving the unit unstaffed in tunnels or anything, the staff member would stay with the train, just sometimes drive and other times be a Train Captain/OBS/PSA whatever you want to call it.

Tram systems are driven on line of sight and share roadway with general traffic, until hazard identification technology has moved to a different realm, they need a pair of eyes in the cab to keep them safe.

 

[deltic]

Tram systems are driven on line of sight and share roadway with general traffic, until hazard identification technology has moved to a different realm, they need a pair of eyes in the cab to keep them safe.

Autonomous vehicles have clocked up millions of miles with few incidents - I would have thought automated trams would be far easier than cars

 

[zwk500]

Autonomous vehicles have clocked up millions of miles with few incidents - I would have thought automated trams would be far easier than cars

A car isn't fixed in it's path. Trams can't swerve out of the way. A tram could easily be driven on ATO with driver in cab to hit the big red button though.

 

[tomuk]

Autonomous vehicles have clocked up millions of miles with few incidents - I would have thought automated trams would be far easier than cars

Yes it is surprising that all the research into autonomous vehicles hasn't been applied to heavy rail. On the surface it would appear to be an awful lot simpler as you don't need to steer the thing. A mmW radar and lidar fitted to the front of the train processed by some on board ai.

 

[deltic]

A car isn't fixed in it's path. Trams can't swerve out of the way. A tram could easily be driven on ATO with driver in cab to hit the big red button though.

Automatic detection is far more likely to react quicker than a human eyeball that might have been distracted etc.

 

[DerekC]

Yes it is surprising that all the research into autonomous vehicles hasn't been applied to heavy rail. On the surface it would appear to be an awful lot simpler as you don't need to steer the thing. A mmW radar and lidar fitted to the front of the train processed by some on board ai.

The problem with trains is that they can't stop within line of sight, so until somebody invents radar that can see round corners we will still need a system that tracks all the trains and calculates and issues movement authorities. Going back to the subject of the OP, one of the key requirements for automation is high reliability. Generally speaking main line railways have far too many failures to be fully automated. An onboard steward like DLR doesn't solve the problem. Think of a full and standing 8 or 12 car train, failed between stations with the steward stuck somewhere in the middle. For the foreseeable future we are going to need a capable person in the cab, even if the train drives itself most of the time.

 

[bramling]

Note: Following this thread on the Kirkby Derailment https://www.railforums.co.uk/thread...eyside-13-03-2021.215146/page-20#post-5796876 going off-topic, I thought I'd set up a separate thread to discuss the musing of driverless trains on Merseyrail.

As a quasi-metro that is largely separated from the national network, some grade of driverless operation would seem feasible. I don't know the system or area very well, but DLR-type Onboard supervisors qualified to drive the train would seem to be a sensible compromise to me. There would be a passenger-facing member of staff on the train who can move it to a station if there was an emergency or failure, and you could even avoid the issues of driverless trains interacting at fringe areas like Chester by getting the OBS to drive the less busy section, when DOO is fine. The OBS would also be able to take the train over for the movements on and off depot if they required drivers.

Thoughts?

I think this is a solution looking for a problem.

The main benefit of automation is to screw more capacity out of fixed infrastructure. Merseyrail doesn’t run a sufficiently intense service for this to be beneficial, and in all fairness probably isn’t likely to. The main pinchpoint is Liverpool Central low level, and that is more to do with space on the platforms for passengers than throughput of trains.

If anyone was looking to increase capacity on Merseyrail, the first step would be running maximum length formations, which at the moment only applies to a tiny handful of services.

Aside from that, what benefit is automation going to bring? Much as I wouldn’t particularly want to advocate DOO, it would be very viable to have DOO on Merseyrail, with the main staff presence being on the stations, as on London Underground.

 

[niggill617@gma]

sanders yes, but I doubt magnetic on a mainline.

The problem with trains is that they can't stop within line of sight, so until somebody invents radar that can see round corners we will still need a system that tracks all the trains and calculates and issues movement authorities. Going back to the subject of the OP, one of the key requirements for automation is high reliability. Generally speaking main line railways have far too many failures to be fully automated. An onboard steward like DLR doesn't solve the problem. Think of a full and standing 8 or 12 car train, failed between stations with the steward stuck somewhere in the middle. For the foreseeable future we are going to need a capable person in the cab, even if the train drives itself most of the time.

Nor can the mark 1 eyeball see around corners. How long to walk 12 coach lengths?

 

[tomuk]

The problem with trains is that they can't stop within line of sight, so until somebody invents radar that can see round corners we will still need a system that tracks all the trains and calculates and issues movement authorities.

Like ETCS which is being rolled out on the Southern end of the ECML and has run daily on the Cambrian for years.
Arguably the Autonomous bit fitted to the train doesn't need to be as highly SIL rated as the signalling system as it only needs to add the protection to reduce the harm of running someone over.

 

[niggill617@gma]

sanders yes, but I doubt magnetic on a mainline.

Magnetic emergency brake systems are on the RSSB agenda.

 

[43066]

I think this is a solution looking for a problem.

The main benefit of automation is to screw more capacity out of fixed infrastructure. Merseyrail doesn’t run a sufficiently intense service for this to be beneficial, and in all fairness probably isn’t likely to. The main pinchpoint is Liverpool Central low level, and that is more to do with space on the platforms for passengers than throughput of trains.

Indeed. If LU isn’t going driverless any time soon (and it isn’t, despite what a certain ignorant Secretary of State might say to stir the industrial relations pot and score points with the Daily Fail readership), it’s difficult to see how this kind of tech could be feasibly brought to Merseyrail in the foreseeable future.

The same goes for the wider rail network. There will no doubt come a time when truly driverless trains roll out across the network. I’m not yet 40 and I’d bet good money I won’t see it before I retire… In fact I’d also bet good money there will still be parts of the national network signalled by absolute block semaphore signals long after I’ve been pensioned off!

 

[tomuk]

Indeed. If LU isn’t going driverless any time soon (and it isn’t, despite what a certain ignorant Secretary of State might say to stir the industrial relations pot and score points with the Daily Fail readership), it’s difficult to see how this kind of tech could be feasibly brought to Merseyrail in the foreseeable future.

The same goes for the wider rail network. There will no doubt come a time when truly driverless trains roll out across the network. I’m not yet 40 and I’d bet good money I won’t see it before I retire… In fact I’d also bet good money there will still be parts of the national network signalled by absolute block semaphore signals long after I’ve been pensioned off!

No that I'm a Tesla fan but you can go out now and see on the UK highways plenty of Tesla cruising about 'autonomously' why is it so much more difficult in a much more constrained environment?

 

[43066]

No that I'm a Tesla fan but you can go out now and see on the UK highways plenty of Tesla cruising about 'autonomously' why is it so much more difficult in a much more constrained environment?

Teslas might be driving around semi-autonomously, but certainly not driverlessly.

It’s not that it’s more difficult to automate trains than cars (in fact it’s easier), it’s the cost of implementation and proving the safety case v the potential (small) savings. Also consider about the respective size of the markets in question: how many private cars are there in the world, versus how many train drivers to replace? That’s why companies like Tesla will invest billions in the required R&D for driverless road vehicles, yet HS2 is being built to still require train drivers.

ETCS will take 30 + years to roll out, as above HS2 isn’t even built yet but will have drivers etc. As a young(ish) train driver in 2022 my biggest fear isn’t being replaced by automation, it’s the interim period where ETCS/ATO comes in and makes the job soul crushingly boring, as per the way most of London Underground has gone. When I can’t avoid ATO any longer that’s when I’ll retire, or become a manager .

 

[bramling]

Teslas might be driving around semi-autonomously, but certainly not driverlessly.

It’s not that it’s more difficult to automate trains than cars (in fact it’s easier), it’s the cost of implementation and proving the safety case v the potential (small) savings. Also consider about the respective size of the markets in question: how many private cars are there in the world, versus how many train drivers to replace? That’s why companies like Tesla will invest billions in the required R&D for driverless road vehicles, yet HS2 is being built to still require train drivers.

ETCS will take 30 + years to roll out, as above HS2 isn’t even built yet but will have drivers etc. As a young(ish) train driver in 2022 my biggest fear isn’t being replaced by automation, it’s the interim period where ETCS/ATO comes in and makes the job soul crushingly boring, as per the way most of London Underground has gone. When I can’t avoid ATO any longer that’s when I’ll retire, or become a manager .

ATO boredom on LU isn’t a problem for those who are able to drive equivalent to or faster than it, as no one is going to hassle them as long as they’re not clocking up incidents. However this is of course easier in a tunnel environment where one can generally rely on adhesion to be perfectly constant.

And this is where automation runs in to problems, as 50+ years after the first production ATO system came in, the problem of varying adhesion has yet to be properly cracked for open-air running. Hence why automation remains confined to specialist metro applications where the capacity benefits are required.

 

[tomuk]

Teslas might be driving around semi-autonomously, but certainly not driverlessly.

It’s not that it’s more difficult to automate trains than cars (in fact it’s easier), it’s the cost of implementation and proving the safety case v the potential (small) savings. Also consider about the respective size of the markets in question: how many private cars are there in the world, versus how many train drivers to replace?

Well of course the train market is much saller but if you replaced drivers with Train captains you would save £2.8bn a year.

That’s why companies like Tesla will invest billions in the required R&D for driverless road vehicles, yet HS2 is being built to still require train drivers.

On HS2 and NPR that the core HS sections will be ETCS/ATO press the button and go.

ETCS will take 30 + years to roll out, as above HS2 isn’t even built yet but will have drivers etc. As a young(ish) train driver in 2022 my biggest fear isn’t being replaced by automation, it’s the interim period where ETCS/ATO comes in and makes the job soul crushingly boring, as per the way most of London Underground has gone. When I can’t avoid ATO any longer that’s when I’ll retire, or become a manager .

Yes, if I was a driver I'd rather they went from 'real' trains direct to driverless then put up with the boredom of ATO.

 

[43066]

ATO boredom on LU isn’t a problem for those who are able to drive equivalent to or faster than it, as no one is going to hassle them as long as they’re not clocking up incidents. However this is of course easier in a tunnel environment where one can generally rely on adhesion to be perfectly constant.

I’m not sure how likely ATO is on the mainline, even with ETCS (as you say it’s only really needed for high capacity locations), but I like the “old school”, organic aspect of the job: knowing the speeds, the gradients, looking out for signals, junction indicators etc., driving entirely to my own discretion with no active monitoring (other than by the OTMR and passive safety systems), which requires *really* knowing the route. A screen in the cab telling me how quickly to drive doesn’t appeal in the slightest.

Well of course the train market is much saller but if you replaced drivers with Train captains you would save £2.8bn a year.

How do you arrive at that figure? How much would it cost to achieve driverless operation over the entire U.K. network (and is it even possible?). I suspect many, many tens of billions…

Keep in mind DLR train captains (as a good comparator) don’t work for free and, in fact, are on circa. £50k these days. Rightly so because what a horrible job it must be.

Yes, if I was a driver I'd rather they went from 'real' trains direct to driverless then put up with the boredom of ATO.

Agreed.

 

[SynthD]

No that I'm a Tesla fan but you can go out now and see on the UK highways plenty of Tesla cruising about 'autonomously' why is it so much more difficult in a much more constrained environment?

Tesla drivers still need to be in the seat, eyes forward, foot near the break and hands near the wheel. That sounds like current train control to me. A Tesla driver is a private individual with insurance, a 2-3 ton vehicle and the gigantic permissiveness we give drivers. Trains are none of those. The environment isn't more constrained if it has level crossings. The ECTS on the Cambrian line doesn't 'solve' level crossings either, as there was an incident of overspeed over them. https://www.gov.uk/government/news/...al-signalling-data-on-the-cambrian-coast-line

 

[DerekC]

No that I'm a Tesla fan but you can go out now and see on the UK highways plenty of Tesla cruising about 'autonomously' why is it so much more difficult in a much more constrained environment?

Roads are very much less safe than railways. From memory (sorry haven't time to check the RSSB website this morning) the risk per person-km is about ten times less by train than by car - and automation of cars has the potential to improve safety by a lot, by eliminating idiots and fatigue. On rails, ATO/ETCS (or whatever) won't suffer from overspeed incidents or SPADs (apart from adhesion problems as highlighted by @bramling) but the difference in overall safety of the railway will be very small, because it does nothing for platform-train incidents.

And both road and rail will suffer from the problems of what to do when the system goes wrong. Until we have systems that have very high availability and can cope with everything met on any trip, you will need a very bored driver sat in the seat, trying to stay alert and ready to take over. That's where rail is much easier, because you can at least apply the emergency brake at any point without the train behind running into you or having to swerve into the path of another vehicle.

 

[skyhigh]

Keep in mind DLR train captains (as a good comparator) don’t work for free and, in fact, are on circa. £50k these days. Rightly so because what a horrible job it must be.

Drivers at my TOC are also on circa. £50K. So the suggestion is to spend billions on driverless trains, to save money by replacing drivers with train captains who get paid similar amounts...?