Differences between TPWS, ATP and ETCS

Hi all

I'm trying to grasp what the differences is between TPWS, ATP and ETCS. I think I understand the benefits of ETCS from the perspective of efficiency and capacity.

But what is the difference between these systems from a safety perspective. I read that TPWS provides 70% of the protection of ATP for SPADS. What accounts for this reduction in safety? What can ATP do that TPWS cannot?

I read that TPWS doesn't stop a train before a red light but just after it but before and critical junctions etc. If that's true, why does it work that way? Is it that which accounts for the fact that it's not quite as safe at ATP?

How do they both compare to ETCS from a safety perspective?

Does each system have any other safety benefits other than preventing running red lights?

I understand that all the systems are effective, but just trying to understand the differences.

I have tried searching online but there doesn't seem to be a simple explanation.

Thanks!

trainenthus said:

Hi all

I'm trying to grasp what the differences is between TPWS, ATP and ETCS. I think I understand the benefits of ETCS from the perspective of efficiency and capacity.

But what is the difference between these systems from a safety perspective. I read that TPWS provides 70% of the protection of ATP for SPADS. What accounts for this reduction in safety? What can ATP do that TPWS cannot?

I read that TPWS doesn't stop a train before a red light but just after it but before and critical junctions etc. If that's true, why does it work that way? Is it that which accounts for the fact that it's not quite as safe at ATP?

How do they both compare to ETCS from a safety perspective?

Does each system have any other safety benefits other than preventing running red lights?

I understand that all the systems are effective, but just trying to understand the differences.

Thanks!

Click to expand...

Hello,
So I am driver who drives on ERTMS level 2.
I’ll start by clarifying that the system is called the European Rail Traffic Management System or ERTMS for short. European Train Control System, ETCS for short, is one of 3 main components of ERTMS.
In terms of the safety aspect ERTMS essentially just had ATP built into so are very similar.

We’ll start with TPWS as it’s the Moses used and simpler system. TPWS has 2 types of use, over speed grids and stop grids.
Stop grids are located at the red signal and will apply the emergency brake to any train that passes the signal when it is red.

Overspeed grids are placed a bit before signals and have 2 grids (an arm and then a trigger). The first grid (the arm) starts a timer and the second (the trigger) will activate the emergency brake if the timer has not elapsed by the time the train passes over it.
The overspeed grids will only be active if the signal is red. So let’s say the line speed is 90mph you might be expected to be doing less than 30mph over the overspeed grids towards a red signal avoid a brake application.
So this ensures the driver has sufficiently reduced the trains speed approaching the red to allow them to stop but if the driver fails to reduce speed further then they will still pass the red signal. That’s where stop grids come in to prevent the train entering a ‘conflict point’ (hopefully).

Now ATP/ERTMS, these systems continually monitor the speed of the train. That’s where my ATP knowledge ends. With ERTMS it also knows the exact location of the train at all times. Because of this it supervises that the train speed is sufficiently reduced towards any reduction in speed or End of Authority(ERTMS equivalent of a red signal).
ERTMS calculates 4 brake curve profiles, the one it expects you to make, the one where it warns the driver, the 3rd one where it applies a full service brake application and if speed is not reduced enough that it is expecting you to pass the End of Authority + the overlap then you get an emergency brake application.

Hope this helps, feel free to ask any more questions

Thank you for taking the time to reply.

Why are stop grids placed at lights, and not before them giving time for the train to stop without passing the red light?

So with TPWS, if a train was going a permitted speed of 100mph and passed a red light, it wouldn't do any intervention until that light is passed?

Thanks

To add to what @Lurcheroo has said - TPWS can intervene and stop the train only when it passes over a track grid - actually it's a cable loop (or strictly a pair of loops) - either an overspeed loop out on the approach to a signal, or a stop loop at the signal. The rest of the time the train speed is unsupervised. With ATP (and ERTMS/ETCS includes ATP) the speed supervision applies all the time. So any time the train exceeds the speed that ATP calculates it should be doing, based on the line speed and the braking curve on approach to a signal, the system will apply the brakes. That has some important consequences. With TPWS, a train can pass a red signal if it fails to brake after the overspeed loop. With ATP it can't, because the system will intervene and bring it to a halt before the signal is reached. (Unless there is a slide). Also ATP can supervise different speeds for the main and branching routes when approaching a diverging junction, whereas TPWS can't do anything about that. And ATP knows what braking rate the train it's working on can achieve, so a freight can start braking earlier, for example.

trainenthus said:

Thank you for taking the time to reply.

So why are stop grids placed at lights, and not before them giving time for the train to stop without passing the red light?

Click to expand...

TPWS fitted signals have an overspeed pair of grids in the approach, set to the maximum that any train should be travelling at at that point on a braking curve and a stop pair at the signal.

trainenthus said:

So with TPWS, if a train was going a permitted speed to 100mph and passed a red light, it wouldn't do any intervention until that light is passed?

Click to expand...

No - it would be tripped by the overspeed loop if the signal wasn't showing clear.

DerekC said:

To add to what @Lurcheroo has said - TPWS can intervene and stop the train only when it passes over a track grid - actually it's a cable loop (or strictly a pair of loops) - either an overspeed loop out on the approach to a signal, or a stop loop at the signal. The rest of the time the train speed is unsupervised. With ATP (and ERTMS/ETCS includes ATP) the speed supervision applies all the time. So any time the train exceeds the speed that ATP calculates it should be doing, based on the line speed and the braking curve on approach to a signal, the system will apply the brakes. That has some important consequences. With TPWS, a train can pass a red signal if it fails to brake after the overspeed loop. With ATP it can't, because the system will intervene and bring it to a halt before the signal is reached. (Unless there is a slide). Also ATP can supervise different speeds for the main and branching routes when approaching a diverging junction, whereas TPWS can't do anything about that. And ATP knows what braking rate the train it's working on can achieve, so a freight can start braking earlier, for example.

Click to expand...

Thanks for the reply!

So TPWS can only interven when a red light has been passed and ATP can predict that at the speed the train is going if it will be able to stop in time or not - and if it reaches a threshold it will start to brake before the light (and stoo before the light)

Why not place track grids in TPWS further from the light so it doesn't have to wait until it reaches the light to automatically apply brakes.

trainenthus said:

Why not place track grids in TPWS further from the light so it doesn't have to wait until it reaches the light to automatically apply brakes.

Click to expand...

They do, to make TPWS more effective at higher speeds an additional loop or loops can be placed before a signal, so that if a train approaches a signal at danger in excess of the trigger speed its brakes will be activated just as if it passed a signal itself at danger.

TPWS isn't designed to prevent passing a signal at danger, but rather to reduce the consequences of such an occurrence.

trainenthus said:

Thanks for the reply!

So TPWS can only interven when a red light has been passed and ATP can predict that at the speed the train is going if it will be able to stop in time or not - and if it reaches a threshold it will start to brake before the light (and stoo before the light)

Why not place track grids in TPWS further from the light so it doesn't have to wait until it reaches the light to automatically apply brakes.

Click to expand...

“Overspeed grids” are fitted on the approach to some, but not all, signals and were also described in post #2?

hexagon789 said:

hexagon789 said:

TPWS isn't designed to prevent passing a signal at danger, but rather to reduce the consequences of such an occurrence.

Click to expand...

Click to expand...

I think this is what I'm not understanding. If TWPS has grids/loops before a red signal, how isn't that designed to prevent passing a signal at danger?

Sorry to labour the point. And feel free to explain to me like I'm an idiot

trainenthus said:

I think this is what I'm not understanding. If TWPS has grids/loops before a red signal, how isn't that designed to prevent passing a signal at danger?

Click to expand...

The additional loops are only where the approach speed is particularly high, such that the mlst bssic arrangement of only having a loop at the signal would mean a brake intervention wouldn't stop the train before the conflict point.

That brings us to the real purpose of TPWS at signals - to prevent a train which has SPADed from reaching a conflict point, i.e. coming into collision with another train.

trainenthus said:

Thanks for the reply!

So TPWS can only interven when a red light has been passed and ATP can predict that at the speed the train is going if it will be able to stop in time or not - and if it reaches a threshold it will start to brake before the light (and stoo before the light)

Why not place track grids in TPWS further from the light so it doesn't have to wait until it reaches the light to automatically apply brakes.

Click to expand...

Worth nothing there are “TPWS+” grids on higher speed lines, placed further back, and set at a higher speed (60+ as opposed to 40+ for “normal” TPWS).

The main point to understand is that it’s a passive safety system with “speed traps” at certain locations, rather than an active system that enforces a braking curve (as is the case with both ETCS and ATP). It’s also intended to mitigate against severe consequences, rather than to prevent any possibility of a SPAD. If it’s triggered that is a safety incident in itself, taken very seriously, and could be career limiting/ending for the driver concerned, depending on history.

trainenthus said:

I think this is what I'm not understanding. If TWPS has grids/loops before a red signal, how isn't that designed to prevent passing a signal at danger?

Sorry to labour the point. And feel free to explain to me like I'm an idiot

Click to expand...

First case is that the train can only stop as fast as it can stop, any intervention before the red signal will only be as effective as the brakes on the train can realistically achieve. If for instance the TPWS on approach to a red signal activates, but the train has reduced braking performance owing to, say, a brake fault, even if the train has been commanded to stop, you are left with the laws of physics and the train will stop as quickly as it is able, but this may not be before the red signal. Trains nowadays have much better brakes though than when TPWS was designed, so modern trains will generally stop before the red.

A second case where TPWS may not stop a train before a red signal is when the train is travelling under the set speed but is not braking or braking sufficiently to comply with the red signal. If the TPWS is set at 40mph on approach to the red, and the train passes over the sensor at 38, the train could still pass the signal if sufficient braking is not applied between the TPWS overspeed sensor and the red signal, in which case the train would SPAD.

Thanks everyone. I think i understand now.

So ATP can constantly monitor the speed of a train. This means that a train can't overspeed in the approach to a signal, and if the driver doesn't apply brakes for a red then it can automatically apply an optimised brake curve to stop before the signal.

Where as with TPWS the train could be speeding before it reaches an overspeed grid, at which point because the train is travelling over the permitted speed it's braking distance is longer so my not stop before the signal? But it will stop before the conflict point?

In addition ATP accounts for the different stopping distances to different trains, so for a train with a shorter stopping distance it won't intervene until later? And TPWS is more binary - if you are going over this speed at this point then its hitting the brakes regardless of the train?

Are there any instances where conflict points are not far beyond the signal, so that whilst TPWS might have slowed the train there is still possibility for a collision? And if so, even if those scenarios are niche/unlikely, is this the advantage of ATP - because it would stop before a light?

Thanks again

trainenthus said:

Where as with TPWS the train could be speeding before it reaches an overspeed grid, at which point because the train is travelling over the permitted speed it's braking distance is longer so my not stop before the signal? But it will stop before the conflict point?

Click to expand...

Important to note they wouldn’t be “speeding” in the sense of exceeding the permanent speed restriction (ie line speed), merely in the sense of going too quickly towards that particular red, or much lower speed restriction.

There’s no TPWS at the majority of signals
on the network, only the high risk ones in terms of junction conflicts, and it also isn’t designed to guard against rear ending type accidents.

trainenthus said:

Are there any instances where conflict points are not far beyond the signal, so that whilst TPWS might have slowed the train there is still possibility for a collision? And if so, even if those scenarios are niche/unlikely, is this the advantage of ATP - because it would stop before a light?

Click to expand...

AIUI it’s always meant to stop trains before the conflict point, where fitted.

This RAIB report about a near miss at Didcot North will help with your TPWS understanding.

The arm and trigger grids of a TPWS Overspeed installation are positioned a calculated distance from each other and from the signal so that a train which passes over them in too short a period of time is brought to a stand. As long as the train does not pass over the trigger loop before the timer elapses the brakes will not be applied. The train could pass over the overspeed grids below the set speed and then fail to slow any further, or even speed up, and nothing would happen until the trainstop grids were reached.

The arm and trigger grids of a TPWS Trainstop installation are positioned butting up against each other at the signal to apply the brakes of any train which does attempt to pass a signal at red. As the grids are immediately next to each other there is no possibility of the on-board timer relapsing between passing the arm and the trigger grid, no matter how slow it is going. Obviously a train is likely to pass the signal at least a small amount if brought to a stand by the train stop grids.

ATP passes information about the line speed, any speed restrictions in place, the gradient, the distance to the next signal and the current aspect of that signal to the train. From this the train can calculate the appropriate braking curve. If the train exceeds this by 3mph a warning is given, and at 6mph the brakes are applied automatically. What ATP cannot do, however, is transmit any updated information to the train other than when the train is passing over the ATP loops/beacons. So if a signal steps up to green between passing between one beacon and the next the computer will still restrict the speed of the train until it passes the next beacon and receives updates information, even if the driver can see the signal is no longer at danger.

This YouTube video may help with the workings of ATP,

This staff information film on TPWS may help.

With ATP it also knows the performance of the train, if anything is isolated (if the data has actually been entered) and the different brake curve it with have etc but nowadays TPWS will most likely intervene before ATP even warbles (3mph over brake curve) as TPWS is set for all trains and doesn’t take into account the brake performance of individual trains.

A IET can brake faster than the ATP speedo can actually show but that’s mainly due to how old the technology is, while it is old and there are faults with the system which are mainly track based that do seem to take a while to get fixed, it’s not a terrible system, it can be a bit sensitive at times (3mph coming upto buffers at Paddington, and on approach to PSR’s where it wants you down at the target speed a little earlier than you really need to be, I think it’s good but ERTMS/ETCS will be an improvement and will prevent a lot of safety of the line incidents while also allowing trains to approach a restrictive signal in a more efficient way

800301 said:

it’s not a terrible system, it can be a bit sensitive at times (3mph coming upto buffers at Paddington, and on approach to PSR’s where it wants you down at the target speed a little earlier than you really need to be, I think it’s good but ERTMS/ETCS will be an improvement and will prevent a lot of safety of the line incidents while also allowing trains to approach a restrictive signal in a more efficient way

Click to expand...

With ERTMS, it still wants you down a bit earlier than you need to be. I’m not sure about ATP but with ERTMS we get a ‘release speed’ towards an EoA or buffer stops. This is calculated by how much space there is to the conflict point so it could be 15mph or 30mph (although we drive and work in KPH).
So let’s say approaching a bay Platform our release speed is 15mph then as long as we are below 15mph it will not force us to slow further and it is the drivers responsibility to bring the train to a stand. If you are doing 20 mph when the release speed is 15mph then it will apply a full service brake application (not emergency).

There is also a similar system to TPWS where we have 2 balises on the approach to a bay Platform. As the Cambrian was a test bed for ERTMS they chucked every different type of system it had so that others could see it and decide how they wanted to implement it.

alxndr said:

The arm and trigger grids of a TPWS Trainstop installation are positioned butting up against each other at the signal to apply the brakes of any train which does attempt to pass a signal at red. As the grids are immediately next to each other there is no possibility of the on-board timer relapsing between passing the arm and the trigger grid, no matter how slow it is going.

Click to expand...

OSS and TSS arming loops have different frequencies because they initiate different processes on board the train. The onboard timer that determines whether the train is overspeeding at an OSS is only started by an OSS arming loop. The conditions that cause an intervention at a TSS are: arming loop detected, followed by arming and trigger loops both detected. Some TPWS receivers have a separate TSS timer to cater for short gaps between arming and trigger loops; this timer runs for a much shorter duration than the OSS timer.

Lurcheroo said:

With ERTMS, it still wants you down a bit earlier than you need to be. I’m not sure about ATP but with ERTMS we get a ‘release speed’ towards an EoA or buffer stops. This is calculated by how much space there is to the conflict point so it could be 15mph or 30mph (although we drive and work in KPH).
So let’s say approaching a bay Platform our release speed is 15mph then as long as we are below 15mph it will not force us to slow further and it is the drivers responsibility to bring the train to a stand. If you are doing 20 mph when the release speed is 15mph then it will apply a full service brake application (not emergency).

There is also a similar system to TPWS where we have 2 balises on the approach to a bay Platform. As the Cambrian was a test bed for ERTMS they chucked every different type of system it had so that others could see it and decide how they wanted to implement it.

Click to expand...

We get a release speeds for a Red signal which will vary from anything from 5MPH to 50mph and usually there is isn’t any explanation as why some of them are so slow, for example on line 1,2 and 3 at Paddington it’s 5mph for the first gantry but a higher speed for the other lines, I can’t really see what difference there is, the release speed for buffers would be 0 so doing anything more than 3mph about 2 coaches lengths from the buffers will make ATP warble although at Paddington I don’t think it will actually intervene at 6mph, just give you the audible warning

Does HS1 still use TVM430 or is it being converted to ERTMS?

(TVM430 being the system used on the French LGV's)

800301 said:

We get a release speeds for a Red signal which will vary from anything from 5MPH to 50mph and usually there is isn’t any explanation as why some of them are so slow, for example on line 1,2 and 3 at Paddington it’s 5mph for the first gantry but a higher speed for the other lines, I can’t really see what difference there is, the release speed for buffers would be 0 so doing anything more than 3mph about 2 coaches lengths from the buffers will make ATP warble although at Paddington I don’t think it will actually intervene at 6mph, just give you the audible warning

Click to expand...

With ATP the release speed is set by distance to a theoretical conflict point beyond the signal. That distance is passed to the onboard computer at the previous beacon/loop, and the train then works out a safe speed for that conflict distance. I don’t believe it is intelligent enough where a signal has multiple alternative overlaps beyond it; it can only give one conflict distance and therefore always assumes worst case scenario.

With the pointwork almost immediately on top of both sides of Gantry 3, that results in a lower release speed approaching those signals.

The old R149/R249 approaching Reading from the West on the reliefs were 0 release speed; and had no infill beacons or loops so you always had to crawl right up to them with ATP not stepping up until you went over the beacon at the foot of the signal.

FormerBritInMe said:

Does HS1 still use TVM430 or is it being converted to ERTMS?

(TVM430 being the system used on the French LGV's)

Click to expand...

Still TVM and will be for a long time to come I imagine

To answer the OP - all 3 do basically the same; but offer improving levels of coverage.

With TPWS the supervision is non-continuous - in simplest terms the train gets checked against a set speed at set points along the track - those checks normally only taking place when the signal they apply to is at red.

With ATP you do get continuous supervision; but non-continuous updates. The train does the hard work in its onboard computer based on data packets transmitted to it from antenna in the track - either small one-shot “beacons”, or longer stretches of “loops” which continuously transmit that information to the train. This also applies to ETCS Level 1, where data is passed to the train by trackside infrastructure not radio. Chiltern ATP only uses (used?) loops.

In ETCS Level 2 (and above) you get full supervision, AND data primarily comes to the train by radio so you don’t need to wait to be passing over fixed trackside infrastructure to receive updates.

Both ETCS Level 1 and Level 2 can do away with lineside signals (as per the Cambrian), be overlaid alongside lineside signals (as per the installation on the GWML between Ealing and Heathrow or the Northern City Line from Finsbury Park to Moorgate), or a hybrid between the two (as per the Thameslink core).

The standard was for ETCS to use metric measurements for speeds; however certainly the GWML installation uses Imperial speeds with Metric distances - I believe that is the standard to be adopted going forward; possibly even retrofitted to the Cambrian?

FormerBritInMe said:

Does HS1 still use TVM430 or is it being converted to ERTMS?

(TVM430 being the system used on the French LGV's)

Click to expand...

High Speed 1 still uses TVM430 - I don’t know of any plans to fit ETCS.

JN114 said:

To answer the OP - all 3 do basically the same; but offer improving levels of coverage.

With TPWS the supervision is non-continuous - in simplest terms the train gets checked against a set speed at set points along the track - those checks normally only taking place when the signal they apply to is at red.

With ATP you do get continuous supervision; but non-continuous updates. The train does the hard work in its onboard computer based on data packets transmitted to it from antenna in the track - either small one-shot “beacons”, or longer stretches of “loops” which continuously transmit that information to the train. This also applies to ETCS Level 1, where data is passed to the train by trackside infrastructure not radio. Chiltern ATP only uses (used?) loops.

In ETCS Level 2 (and above) you get full supervision, AND data primarily comes to the train by radio so you don’t need to wait to be passing over fixed trackside infrastructure to receive updates.

Both ETCS Level 1 and Level 2 can do away with lineside signals (as per the Cambrian), be overlaid alongside lineside signals (as per the installation on the GWML between Ealing and Heathrow or the Northern City Line from Finsbury Park to Moorgate), or a hybrid between the two (as per the Thameslink core).

The standard was for ETCS to use metric measurements for speeds; however certainly the GWML installation uses Imperial speeds with Metric distances - I believe that is the standard to be adopted going forward; possibly even retrofitted to the Cambrian?



High Speed 1 still uses TVM430 - I don’t know of any plans to fit ETCS.

Click to expand...

Thanks for your explanation (and everyone else). This all makes much more sense now.

My next question is how do these all compare to Indias Kavach system. From my limited understanding it seems similar to ETCS level 2?

What are the differences?

My knowledge of Kavach is limited to what was published in the IRSE News last May (not publicly available yet). There is no 'signals away' or moving block upgrade path. Some cost has been avoided by designing it as purely an overlay and not having to integrate numerous sets of operating rules. UHF is used instead of GSM-R reducing the number of base stations required. It appears that movement authority is transmitted from smaller line side station interlockings rather than centralised Radio Block Centres. RFID tags have been used instead of Eurobalises to reduce cost. It has similar function to ETCS level 2 overlay though.

It's worth pointing out that some of the terminology differs between ATP and TPWS. The objectives are different. The technology is different. Hence the costs are different. And the benefits are different.

ATP was developed by BR and in use long before TPWS was developed. But only two pilot ATP schemes were commissioned.

With the GWML ATP, all main aspect signals are provided with a 'signal' beacon. This is a rectangular construction made out of large pieces of metal and mounted in the four foot on the sleepers. It's called a 'signal' beacon because it's normally mounted right next to the signal. Electrically, it's a loop aerial.

If there is too much data for the transmission from the 'signal' beacon due to multiple speed restrictions and/or junction speed restrictions ahead, an 'additional' beacon will be provided. These are mounted in line with and adjacent to the 'signal' beacon. Although these installations are rare.

That's the basic GWML ATP system. When specified, in-fill devices can be added. There are two types. In-fill beacons or in-fill loops. When provided, one or the other is provided on approach to a signal. They only provide limited data, but importantly, signal aspect data. An in-fill beacon can provide a spot update to the onboard ATP computer. The advantage is that a beacon is easier to maintain and less likely to suffer damage. The disadvantage being that the train only gets an update at that specific location. An in-fill loop is a long length of two single core cables that runs down the four foot. Typically for 500 to 800 yards. Again, electrically, it's a loop aerial. The advantage is that any time the train's aerial (mounted under the cab/unit/loco) is over any part of the loop, the on board ATP computer can receive updates. Hence if the signal aspect steps up from a more restrictive aspect to a less restrictive aspect (say from yellow to green), the computer will recalculate allowing the driver to adjust the speed of the train before getting to the 'signal' beacon. The disadvantage is that these 'loops' of cable are an absolute pig to maintain and a nightmare to repair. It can take one or two shifts to get a non-functional loop back in working order.

The Chiltern ATP only uses loops. As I've not worked on this system, I can't comment in detail on it.

The objectives of TPWS are:
* the primary purpose of which is to minimise the consequences of a train passing a TPWS fitted signal at danger.
* minimise the consequences of a train overspeeding at certain locations,
* cost is kept low by only fitting signals where there is a point of conflict (junction/points) ahead of the signal.

The objectives of ATP are:
* to prevent trains passing any signal at red (apart from at a slow release speed) on ATP fitted lines,
* to prevent trains from exceeding any and all speed limits (rolling stock/loco speed, line speed, permanent speed limits, temporary speed limits, emergency speed limits and junction speed limits),
* to be both a safety critical system and a linked system. If a 'signal' beacon is missing, damaged or malfunctioning, or for any other reason the on-board computer is unable to receive the data message, it will display a fault code to the driver.

You may find these web sites interesting: http://www.traintesting.com/ATP.htm and http://www.old-dalby.com/ATP.htm

Annetts key said:

The Chiltern ATP only uses loops. As I've not worked on this system, I can't comment in detail on it.

Click to expand...

*used. ATP is no longer in use on the Chilterns.

I was trying to think back to my days of signalling materials contracts. I seem to recall ATP on Chiltern and GWR, then the introduction of TPWS. Our company was often asked for kits of TPWS material, and had to ask "but what is in the kit". Over time, we were able to predict most of the items and pre plan material ordering, so as to hold stock, particularly of the longer lead time items, such as relays. I guess where ATP suffered was, TPWS became the standard and materials support for the more restricted ATP was withdrawn by manufacturers.. Often happens with electronic type systems. The World moves on faster than the railways. . Then I wondered if TCAIDs stopped being used. (In those slippery , leaf covered track areas

The GWML ATP equipment on the infrastructure is mainly very reliable.

A lot of the problems in the early days were due to problems with wheel slip or wheel slide (on board computer distance travelled data became inaccurate, so the next beacon message was outside the limits of when the computer was looking for it) and problems with the aerial on the train.

There have always been problems with the in-fill loop cables being damaged.

The beacons occasionally fail. Normally it's the feedback circuit that tells the system if the beacon is working, but trains still continue to receive the data packet okay. But at the same time, there are various beacons that have been clobbered and are now misshapen, but which still work okay.

A few years ago, replacement beacons were still available. They were often ordered in small batches rather than individually.

Over the last twenty odd years, some of the interference suppression filters (actually an off the shelf part) have suffered from what I believe are paper capacitor failures.

And occasionally an encoder (the dual processor 'box of electronics') will fail. Often it's the power supply section that fails.

The biggest problem with encoders, is that there are a number of different configurations, which are not interchangeable. So a depot may have say six or seven in stock, but not have the correct type when one fails...

TPWS is a much, much newer system. But the failure rate is poor. The loops (fibreglass 'grids' with a multicore cable looped around inside the outer sections of the fibreglass) are far more easily damaged compared to ATP beacons.

But the main problem is the baseboard/plugboard contacts on the equipment in the location cases/cupboards. If there is moisture in the air, a reaction occurs between the different metal contacts in the baseboard and the plugboard of the TPWS module. So it's fairly regular for a S&T team to replace an existing 'failed' module with a new one on their area every week or two. Not the for the same signal/PSR but it's not unknown for a signal/PSR to become a repeat failure.