A new way to keep train apart

[Adlington]

The Economist published an article (in the 1st October printed edition, the online version is behind paywall ) on a novel way of detecting where the trains are.
Here are the crucial (IMHO) points. Will somebody more technical care to comment?

Lines in a railway network are divided into blocks, and only one train at a time is allowed in a block. Rationing space this way can, though, lead to inefficiency. More precise information of each train’s whereabouts would permit trains to travel closer together without compromising safety, and therefore allow more services to be run.

Martin Lauer of the Karlsruhe Institute of Technology thinks he has an alternative. His invention, dubbed the Magnetic Railway Onboard Sensor (MAROS) , uses information encoded in the very rails a train is running on, in the form of something called their magnetic permeability. Employing this, that train’s position on the network can be determined exactly.

MAROS works by lowering a pair of detector coils over each rail and passing alternating currents through them. These currents are influenced by the magnetic properties of nearby metal objects. The production of rails for railways takes care to make them uniformly strong and smooth, but their magnetic permeability is not among the specifications. That means a train carrying a MAROS detector picks up random fluctuations from the rails beneath. Being random, the resulting pattern of permeability is probably unique to a given stretch of the network. For a 200-metre section, Dr Lauer calculates, there is about one chance in a trillion that it is duplicated anywhere on the 2.3bn kilometres of track currently laid around the world.

To get the system working on a particular route, a test train rides it a few times to map out the fluctuations in the rails. After that, trains running the route would continually compare the permeability measurements from the detectors with those on the map, and then report their position to the network traffic controller. Local changes in magnetic permeability caused by wear and tear or lightning strikes would show up during routine usage and, once confirmed, could be added to the map.

Dr Lauer’s commercial partner, a firm called ITK Engineering, has been testing the system on scheduled train services plying a 130km route in Austria and reckons it will be ready for market in 2025. Dr Lauer himself estimates that using MAROS could increase the capacity of a busy rail network by 20-30%. It would also cost less than the existing system of sensors and beacons.

 

[alistairlees]

Whenever someone says something like ‘it’s a one in a million’, ‘one in a billion’ or ‘one in a trillion’ chance, then I always take that as a signal that I should be very cautious about what they are claiming. Most people have only a very tenuous grasp of probabilities.

 

[Surreytraveller]

Whenever someone says something like ‘it’s a one in a million’, ‘one in a billion’ or ‘one in a trillion’ chance, then I always take that as a signal that I should be very cautious about what they are claiming. Most people have only a very tenuous grasp of probabilities.

And how many train crashes have been the result of the 'one in a million' chance of something happening? Or a train with a thousand people on board, that's a thousand in a million chance

 

[Shaw S Hunter]

Even if the detecting coils don't actually touch the rails they could still interfere with track circuits, couldn't they? If so then it'll be a long time before such a system could be introduced in this country.

 

[sw1ller]

What happens to the routes unique magnetic pattern when someone throws an old bike or shopping trolly over the bridge??

 

[Adlington]

What happens to the routes unique magnetic pattern when someone throws an old bike or shopping trolly over the bridge??

In this case I'd worry about the train, not about magnetic patterns...

 

[Basil Jet]

Whenever someone says something like ‘it’s a one in a million’, ‘one in a billion’ or ‘one in a trillion’ chance, then I always take that as a signal that I should be very cautious about what they are claiming. Most people have only a very tenuous grasp of probabilities.

He's saying that it is probable that if they measured the profile of every rail in the country they will probably all be different.

If they aren't, it probably wouldn't matter - the system would know from the learning stage that the St Ives Branch and the North Berwick branch had the same profile, so every time a passenger train goes from detecting the unique pattern at St Erth to detecting the pattern found at both Lelant Saltings and North Berwick, you can know with 100% certainty which of the two the train is actually at.

The only problem would be if, say the rails at Tyndrum Upper and Tyndrum Lower had the same profile, then you wouldn't know where the train was, but this problem would become known at the track learning stage, and so could not lead to a passenger train suddenly being at an unknown location on a Wednesday afternoon. I suppose in that eventuality you would have to replace one of the two tracks!

But if there is only a one in a trillion chance of two adjacent branches having the same profile, the one in a trillion chance that you might have to replace one of the tracks is nothing compared to the insane cost of installing and maintaining conventional signalling, all of which could be ripped out, and the insane cost of signalling failures, which could never happen again.

 

[P Binnersley]

That means a train carrying a MAROS detector picks up random fluctuations from the rails beneath. Being random, the resulting pattern of permeability is probably unique to a given stretch of the network.

So every train on the network will need to be re-programmed when you re-lay a section of track?

 

[Bald Rick]

So every train on the network will need to be re-programmed when you re-lay a section of track?

Exactly!

 

[Adlington]

So every train on the network will need to be re-programmed when you re-lay a section of track?

My reading of the Economist's article is that every train reads the magnetic pattern and reports it to the control centre. So there is no need to reprogram anything on the train.

 

[Ediswan]

My reading of the Economist's article is that every train reads the magnetic pattern and reports it to the control centre. So there is no need to reprogram anything on the train.

The Economist quote includes:

After that, trains running the route would continually compare the permeability measurements from the detectors with those on the map, and then report their position to the network traffic controller.

The train does the comparison, then reports the position. To do that, the train needs a local copy of the map.

 

[The Planner]

So its a way of delivering moving block, surely you still need a method of speed controlling the trains via movement authority? How is that solved?

 

[6Gman]

I have a simple rule of thumb view on these high-tech ideas.

Given how difficult the railway seems to find it to get a working system for electronic seat reservations I find it difficult to imagine the practicality of a system that requires moving trains to "report" their locations to a control/signalling centre which will then be able to process that data quickly and accurately enough to operate a railway safely.

 

[Randomer]

So its a way of delivering moving block, surely you still need a method of speed controlling the trains via movement authority? How is that solved?

I'm guessing some kind of radio based system as a few companies have been deploying for European Train Control System level 2. It just removes the need for balises so you can get ETCS level 2 much cheaper without the expense of the reliability of the radio systems needed for ETCS level 3.

Having said that isn't there an ETCS level 3 trial somewhere in Italy that uses satellite positioning with some overlay balises in stations with movement authorities broadcasted over GSM-R? If this system was implemented it could achieve level 3 with moving blocks without the issue of tunnels and limited satellite navigation accuracy. This might be what they are aiming towards.

I don't think its going to be particularly practical for the reasons overs have outlined above. I think a national level database might be practical especially if the positioning is only required to be accurate to within a tolerance. However it would be a huge single point of failure if, for example, the train needed to positively check for any database updates every time it was prepared to go into service.

 

[fishwomp]

And how many train crashes have been the result of the 'one in a million' chance of something happening? Or a train with a thousand people on board, that's a thousand in a million chance

The number of passengers does not increase the number of crashes.

 

[davews]

Loads of wishy washy things there, clearly the chap who wrote it knows nothing about magnetic permeability. Strange that that fundamental characteristic of materials is affected by lightning... It may be a sensible idea but you would need the proper technical details not something some journalist has made up. I would have thought that the magnetic properties of the train itself, something pretty variable especially with electric ones would have a far greater influence than some bit of track.

 

[100andthirty]

My take is that just because you can do something, it doesn't mean you should. This sounds like a spectacularly complicated way of locating trains.

 

[Dstock7080]

I wonder if Dr Lauer has read the spec of the original Victoria Line system, tested on the District Line and fitted in the 1960s:

The alternating current flow in the running rail produces a fluctuating magnetic field, which can in turn generate an induced current in a coil located just above the running rail. Two safety coils are fixed to the lead truck on each driving motor car, one coil over each rail, to prove that the current flowing in each rail is the same. The weak currents flowing in the detector coils are then amplified in an amplifier tuned at 125Hz to provide a square wave output at code frequency for the use of the trainborne ATC equipment.

seem rather similar?

 

[lyndhurst25]

Why rely on the random fluctuations in magnetic permeability of the rails? You could treat the track as a giant length of magnetic tape (stereo as there are two rails) and use a train fitted with erase and record heads to record location, and all sorts of other useful information, onto the tracks. (Joking...)

 

[100andthirty]

Dr Lauer implies that the coils on the train induce the current into the rails and read the magnetic signature whereas the Victoria line system and all the others that use coded track circuits have the train read the electrical signal that has been input from the trackside.

 

[Railsigns]

The article begins by comparing the new system, which is a form of moving block signalling, with fixed block signalling. The benefit of moving block signalling is already understood. It ought to be telling us why this new system is better than existing moving block systems.

 

[43096]

Why is it that when I see "Dr Lauer" I read it as "Dr Lardner"? As in the crackpot Dr Dionysius Lardner who came up with various theories that were utterly debunked by Brunel.

 

[Adlington]

the crackpot Dr Dionysius Lardner

He wasn't totally crackpot. He was professor of natural philosophy and astronomy at UCL. Also FRS & FRSE.

 

[Ediswan]

The article begins by comparing the new system, which is a form of moving block signalling, with fixed block signalling. The benefit of moving block signalling is already understood. It ought to be telling us why this new system is better than existing moving block systems.

I read the article differently. They are suggesting an alternative means of locating where trains are, to be used with existing (actual or proposed) signalling systems.

 

[HSTEd]

So every train on the network will need to be re-programmed when you re-lay a section of track?

Just as every train driver must be reprogrammed when anything on the line they drive is changed.

Luckily computers are rather easier to reprogramme than people.

 

[Railsigns]

I read the article differently. They are suggesting an alternative means of locating where trains are, to be used with existing (actual or proposed) signalling systems.

It is an alternative way of locating where trains are (or at least where the front ends of the trains are), but the article pitches it as a solution for overcoming the inefficiency of the fixed block system and improving capacity without saying that solutions already exist.

Just as every train driver must be reprogrammed when anything on the line they drive is changed.

Train drivers don't need to update their route knowledge following a like-for-like track renewal.

 

[Peter Sarf]

So every train on the network will need to be re-programmed when you re-lay a section of track?

I'm guessing some kind of radio based system as a few companies have been deploying for European Train Control System level 2. It just removes the need for balises so you can get ETCS level 2 much cheaper without the expense of the reliability of the radio systems needed for ETCS level 3.

Having said that isn't there an ETCS level 3 trial somewhere in Italy that uses satellite positioning with some overlay balises in stations with movement authorities broadcasted over GSM-R? If this system was implemented it could achieve level 3 with moving blocks without the issue of tunnels and limited satellite navigation accuracy. This might be what they are aiming towards.

I don't think its going to be particularly practical for the reasons overs have outlined above. I think a national level database might be practical especially if the positioning is only required to be accurate to within a tolerance. However it would be a huge single point of failure if, for example, the train needed to positively check for any database updates every time it was prepared to go into service.

Well. A bit simplistic but :-

If the train encounters a new bit of track it would be the same as a new or missing balise. Just report unsure of location. Last location is already known. Block does not move forward until the leading train encounters a recognised piece of track after the new/unknown track. Remember also that the signalling system probably would already know a section of track has been relaid and so has become unknown so no alarm there just cation. That caution would only apply to trains failing to report so would recede as trains were updated.

Otherwise, how quickly could a train learn a new bit of track ?. Or how quickly could it be told about a new bit of track. To be on the safe side the train could be comparing with GPS signal anyway.

 

[HSTEd]

Otherwise, how quickly could a train learn a new bit of track ?. Or how quickly could it be told about a new bit of track. To be on the safe side the train could be comparing with GPS signal anyway.

In theory a matter of seconds after validation of the new profile, using GSM-R.

But it would depend on just how much information is required to identify the rail.

 

[Irascible]

As an additional system it seems a reasonable idea, it at least depends on the most critical infrastructure, unlike using GSM or GPS for location. Using that and GSM and GPS ( and whatever else you can think of that's not completely nuts for a train - I think INS might be a bit much! ) starts layering continuous position detection which is really not a bad thing. The database would be a fairly hefty thing though.

Theoretically this might also pick up developing problems in a rail?

 

[Ediswan]

Theoretically this might also pick up developing problems in a rail?

That gets a brief mention as "In development" at 00:59 in this video: