Thameslink Core Major Power Failure 28/8/25

[Dr Hoo]

The south end of Blackfriars is obviously the poster child for an intensely worked flat junction that relies on parallel working but there’s quite a bit of grade separation too, especially on the Great Northern and London Bridge-Croydon corridor axes.

 

[Bald Rick]

The south end of Blackfriars is obviously the poster child for an intensely worked flat junction that relies on parallel working

Clever bit of layout design as well.

but there’s quite a bit of grade separation too, especially on the Great Northern and London Bridge-Croydon corridor axes.

Not forgetting at St Pancras!

 

[bramling]

It doesnt say that at all!

But it does! You said that the crewing arrangements on Thameslink make it harder to recover from disruption than other networks. So unless you’re saying that the operator has, for whatever reason, chosen to have sub-optimal crewing arrangements then I’m not sure what else the statement could mean.

The whole point made all those years ago was that Thameslink, in its current form (and to be honest its original form) would not be resilient to the slightest perturbation - and here we are with exactly that.

 

[Bald Rick]

But it does! You said that the crewing arrangements on Thameslink make it harder to recover from disruption than other networks. So unless you’re saying that the operator has, for whatever reason, chosen to have sub-optimal crewing arrangements then I’m not sure what else the statement could mean.

Yes I did say that. As in, compared to some other (very reliable) railways that I quoted, the crewing means it harder to recover. I did not say, or imply that:

Thameslink in its post-2018 form is inherently unreliable!

The whole point made all those years ago was that Thameslink, in its current form (and to be honest its original form) would not be resilient to the slightest perturbation

It is remarkably resilient to slight perturbations. There has been a 5 mile long TSR on the MML for a week now and hardly anyone has noticed. My train home tonight was on time despite it. Three seperate signal faults on the MML today too. Almost no delays exported south of the river. My usual train home is often 5 minutes late through East Croydon, and on time at St Pancras. Slight pertubations are taken in their stride.

Major pertubations such as a train failing in the most difficult place for 3 hours are difficult to recover from - and as I said, more difficult to recover from than simpler networks where every driver signs almost every route.

 

[johncrossley]

Power failures seem to be common these days. There's another one out of Victoria this evening. Are they inevitable or can something be done about them?

 

[Bald Rick]

Power failures seem to be common these days. There's another one out of Victoria this evening. Are they inevitable or can something be done about them?

Power failure is a catch all term.

The one in the core last week was a failure of some of the power electronics on the train.

The one tonight (at Clapham Jn) is a failure of the third rail. The equivalent of the wires coming down with OLE, effectively. (as an aside, this kind of thing happens regualrly, but because it is not reported as a third rail issue, it causes many people to think that the third rail system is much more reliable than the OLE system. It isn’t.)

Other power failures could be the power supply to signalling, a tripping of the Overhead line due to something landing on it, or a power failure of the supply from the grid, or many many other things.

 

[bramling]

more difficult to recover from than simpler networks

More difficult to recover from than simpler networks. Thank you, only waited 7 years to hear it!

It isn’t just major incidents either.

 

[johncrossley]

Power failure is a catch all term.

The one in the core last week was a failure of some of the power electronics on the train.

The one tonight (at Clapham Jn) is a failure of the third rail. The equivalent of the wires coming down with OLE, effectively. (as an aside, this kind of thing happens regualrly, but because it is not reported as a third rail issue, it causes many people to think that the third rail system is much more reliable than the OLE system. It isn’t.)

Other power failures could be the power supply to signalling, a tripping of the Overhead line due to something landing on it, or a power failure of the supply from the grid, or many many other things.

Can anything be done about any of these things? It seems to the layperson that the railway is too difficult. Too many things can go wrong, meaning that delays are always going to happen on a regular basis.

 

[Bald Rick]

More difficult to recover from than simpler networks. Thank you, only waited 7 years to hear it!

It isn’t just major incidents either.

But this is nothing new. More complex networks are more difficult to recover from incidents than simple networks. It’s been the same for 200 years on the rsilway, and the same in all other networks.

== Doublepost prevention - post automatically merged: 3 Sep 2025 ==

Can anything be done about any of these things? It seems to the layperson that the railway is too difficult. Too many things can go wrong, meaning that delays are always going to happen on a regular basis.

That’s quite a question. I’m not sure how to answer it other than “yes, but…”


Specifically for Thameslink, an awful lot of money was spent in the Thameslink programme to reduce the chance of incidents happening and to mitigate them when they do.

For example:
1) bi-di signalling in the core so that a very limited service can be provided when one line is shut for maintenance and in the event of certain faults.

2) Specially configured power supplies to the signalling that automatically re-configure around faults (I don’t think the signalling power supply has ever failed catastrophically in the core in the 15 years since it was commissioned)

3) Proceed on Sight Aspect signals (PoSA) that enable trains to be driven past failed signals / track circuits without delay

4) High performance rail that needs less attention (grinding etc) and less likely to have defects. (but harder to repair)

5) Conductor bar for the Overhead line - near zero maintenance (and so far has never failed)


Much of this is specific circumstance for the core, and all of it was rather expensive.

 

[Bikeman78]

But it does! You said that the crewing arrangements on Thameslink make it harder to recover from disruption than other networks. So unless you’re saying that the operator has, for whatever reason, chosen to have sub-optimal crewing arrangements then I’m not sure what else the statement could mean.

The whole point made all those years ago was that Thameslink, in its current form (and to be honest its original form) would not be resilient to the slightest perturbation - and here we are with exactly that.

My 80 year old father would rather walk 25 minutes to or from Horsham and use Southern than take his chances with the Thameslink Littlehaven lottery.

 

[bramling]

But this is nothing new. More complex networks are more difficult to recover from incidents than simple networks. It’s been the same for 200 years on the rsilway, and the same in all other networks.

What is new (albeit 7 years now) is that we had a less complex network before. Whether this concept is novel or not doesn’t really help people who want to get from A to B.

== Doublepost prevention - post automatically merged: 3 Sep 2025 ==



That’s quite a question. I’m not sure how to answer it other than “yes, but…”


Specifically for Thameslink, an awful lot of money was spent in the Thameslink programme to reduce the chance of incidents happening and to mitigate them when they do.

For example:
1) bi-di signalling in the core so that a very limited service can be provided when one line is shut for maintenance and in the event of certain faults.

2) Specially configured power supplies to the signalling that automatically re-configure around faults (I don’t think the signalling power supply has ever failed catastrophically in the core in the 15 years since it was commissioned)

3) Proceed on Sight Aspect signals (PoSA) that enable trains to be driven past failed signals / track circuits without delay

4) High performance rail that needs less attention (grinding etc) and less likely to have defects. (but harder to repair)

5) Conductor bar for the Overhead line - near zero maintenance (and so far has never failed)


Much of this is specific circumstance for the core, and all of it was rather expensive.

The problem is, all this is lovely, but it was said here at the time that the core probably wouldn’t be the biggest problem, albeit that when it does have its moments it will be massive - as indeed is the case.

However all the core resilience counts for nothing when there’s a fatality at Stevenage, a vehicle hits a bridge at Coulsdon, a signal failure at East Croydon or a failed train at Crawley. These thing happen, and they happen often. Frankly, it’s just not good enough that when they do happen it’s often wrecking the service for the remainder of the day - and let’s be clear, saying the service is wrecked is no exaggeration when destinations go for hours with no service even once the initial incident is resolved.

The practical effects of this are things like people stuck on platforms with gaps in their service measured in hours (especially to areas like Hitchin-Peterborough), people who have to get slow services making them late for their business, people turfed off trains which either run fast or are turned short, or people who turn up at their station and have to return home and take the car instead. This is now commonplace, whereas it was rare pre-2018. I cannot now go to my station without checking the service beforehand, which just wasn’t necessary before.

On GN we also have the added own-goal of the 379 fiasco, which isn’t down to Thameslink, but is ultimately something which traces back to the DFT. Again, the practical consequence of the 379s at the moment is trains not turning up when they’re supposed to, for whatever reason.

Can you see why a lot of people are of the view that all this just isn’t good enough, and why Thameslink isn’t popular?

 

[Bikeman78]

However all the core resilience counts for nothing when there’s a fatality at Stevenage, a vehicle hits a bridge at Coulsdon, a signal failure at East Croydon or a failed train at Crawley. These thing happen, and they happen often. Frankly, it’s just not good enough that when they do happen it’s often wrecking the service for the remainder of the day - and let’s be clear, saying the service is wrecked is no exaggeration when destinations go for hours with no service even once the initial incident is resolved.

Stations such as Littlehaven, Ifield, Salfords and Earlswood effectively had no service after 1500, other than a handful of peak Southern trains in one direction. Looking specifically at Littlehaven, the last down train to run as booked departed at 1436. There are 18 scheduled Thameslink departures in the down direction after that time. Only four ran.

There are three Southern peak trains that stop at Littlehaven but not Ifield. The 1505 and 1535 from Victoria lost a few minutes between Crawley and Horsham so it's fair to assume that they called additionally. Nothing else did. I can't find any evidence of extra stops in the up direction.

I don't know what it was like on the GN side but I doubt that Ashwell & Morden did much better.

 

[johncrossley]

Stations such as Littlehaven, Ifield, Salfords and Earlswood effectively had no service after 1500, other than a handful of peak Southern trains in one direction. Looking specifically at Littlehaven, the last down train to run as booked departed at 1436. There are 18 scheduled Thameslink departures in the down direction after that time. Only four ran.

There are three Southern peak trains that stop at Littlehaven but not Ifield. The 1505 and 1535 from Victoria lost a few minutes between Crawley and Horsham so it's fair to assume that they called additionally. Nothing else did. I can't find any evidence of extra stops in the up direction.

I don't know what it was like on the GN side but I doubt that Ashwell & Morden did much better.

Surely every cancellation should mean extra stops? It is unacceptable to wait one hour between trains.

 

[Bikeman78]

Surely every cancellation should mean extra stops? It is unacceptable to wait one hour between trains.

It does in Wales, but apparently not on GTR.

 

[Earle Grey]

I don't know what it was like on the GN side but I doubt that Ashwell & Morden did much better.

In the event of such an incident on Thameslink, GTR have "enough" trains Kings Cross - Cambridge so they apply SSOs in those for Ashwell & Morden and the like. IIRC its the 2Cxxs which pick up the pieces. A shuttle service Peterborough - Stevenage is also in the agreed Contingency plan, resources permitting, to cover off the stops on this route too.

 

[Hadders]

In the event of such an incident on Thameslink, GTR have "enough" trains Kings Cross - Cambridge so they apply SSOs in those for Ashwell & Morden and the like. IIRC its the 2Cxxs which pick up the pieces. A shuttle service Peterborough - Stevenage is also in the agreed Contingency plan, resources permitting, to cover off the stops on this route too.

The problem with this plan is that the four 12-car Thameslink trains an hour that serve Stevenage and Hitchin are cancelled and replaced by SSOs on two 8-car trains an hour.

Yes, the stoppers also operate (although some were cancelled in last weeks disruption) but taking last weeks disruption as an example, there simply seemed to be insufficient capacity provided between the problem occurring at 14:30 And end of service, even though NRE reported that the power issue was fixed at 18:00.

 

[galwhv69]

Surely every cancellation should mean extra stops? It is unacceptable to wait one hour between trains.

They are looked into and added where possible, however there are lots of limitations (e.g. driver going on tight PNB after, or would miss it's slot at a key junction etc.). Also worth noting in general that extra stops will not show on RTT

 

[saismee]

It is unacceptable to wait one hour between trains.

Welcome to most other parts of the UK!

 

[Bikeman78]

They are looked into and added where possible, however there are lots of limitations (e.g. driver going on tight PNB after, or would miss it's slot at a key junction etc.). Also worth noting in general that extra stops will not show on RTT

It's easy to tell if trains have not called additionally. No train is going to get from Horsham to Crawley or vice versa in the scheduled time if it makes two extra stops. The opposite is less conclusive. A train can lose three or four minutes for many reasons. It's also worth noting that the up Arun Valley trains have lots of recovery around Three Bridges and Gatwick, so they can make the two extra stops and still depart Gatwick on time.

 

[renegademaster]

What GTR normally does during a meltdown on the Brighton Mainline is introducing extra stops on Southern Victoria or London Bridge services

 

[Ken Ickie]

All, my first post. Been a lurker for a while, however, registering is free, and some things bothered me tonight.

I guess all forums have this to an extent, however there is stuff up thread that is unsubstantiated tripe or simply not thought through. Perhaps folks posting in haste, and I I'm not immune to that.

I could use actual quotes, though on my first post, this might be considered confrontational.

"People stranded for hours". Perhaps if the Thameslink core goes down and you stay still. Let's say I'm in Sutton and want to go to Bedford. Looking at my map: Sutton to Wimbledon to Richmond to Willesden Junction keeps me moving and I end up comfortably north of the river. With respect to formal ticket acceptance, surely a "Thameslink let me down" is all that is required? (Even if the issue was caused by say, Network Rail). Did Joe Strummer wait for the owner to arrive and ask for permission to squat? (There is a blue plaque outside the property now).

"Frequent fatalities at Stevenage". Not what the poster means, of course, however, word for word that is what was stated.

Lisbon was clearly a tragic accident. I'm guessing most UK rail fatalities are presently not. Can these be 'designed out'? To a tiny extent you might marginally improve security at railway entry points, e.g. manned stations. If the frequency of rail fatalities has noticeably increased, for myself that's a societal issue. I personally cannot see magic wands or silver bullets today.

(Paraphrasing): third rail is not materially more reliable than OLE. I'd like to see the data on this. I've led FMECA studies in the process industries and, to my mind, the physics and engineering point to divergence. But, I respect good data.

Apologies for a long post!

 

[Horizon22]

What is new (albeit 7 years now) is that we had a less complex network before. Whether this concept is novel or not doesn’t really help people who want to get from A to B.

It is if people going from A to B had to previously go A to C, then C to B. Or even A to C, C to D and then D to B. In this regard C/D is analogous to a London Underground service or a change where now a through journey is available. And many people are more appreciative of that, as it runs well the vast majority of the time. Just that when it is disrupted, it is potentially worse (scope and timeframe) than before.

Exactly the same concept is true of the Elizabeth line, and TfL Rail (or the predecssor GWR/FGW and Greater Anglia services) required changes which substantially increased the time for through journeys.

It also frees up platform availability for the major providers at London terminals to increase their service as platform availability is often a key capacity constraint.

It - as @Bald Rick discusses, does have risks during disruption because of crew and fleet working and Thameslink has a) longer end-to-end services and b) more branches. However the clock isn't going to be re-wound and outside of more robust and crucially deliverable contingency plans (e.g. a reduced, but confirmed number of tph from Kings Cross / Moorgate / London Bridge), what else would you suggest. Blocking the core of any route for several hours is likely to lead to widespread disruption.

== Doublepost prevention - post automatically merged: 7 Sep 2025 ==

The problem with this plan is that the four 12-car Thameslink trains an hour that serve Stevenage and Hitchin are cancelled and replaced by SSOs on two 8-car trains an hour.

Yes, the stoppers also operate (although some were cancelled in last weeks disruption) but taking last weeks disruption as an example, there simply seemed to be insufficient capacity provided between the problem occurring at 14:30 And end of service, even though NRE reported that the power issue was fixed at 18:00.

This is true of any contingency plan. Where the infrastructure is limited, naturally not as much of a service can be provided. Ticket acceptance is often implemented and disruption messaging sent to try and move demand to other places, where possible.

 

[whoosh]

It - as @Bald Rick discusses, does have risks during disruption because of crew and fleet working and Thameslink has a) longer end-to-end services and b) more branches.

c) Maybe differs in the number of 'Standby/Cover' duties?

 

[Surreytraveller]

Maybe the Elizabeth Line only has driver changes at locations where trains can be put away, that the driver on the train, will always sign. Unlike Thameslink.

 

[Nicholas Lewis]

It - as @Bald Rick discusses, does have risks during disruption because of crew and fleet working and Thameslink has a) longer end-to-end services and b) more branches. However the clock isn't going to be re-wound and outside of more robust and crucially deliverable contingency plans (e.g. a reduced, but confirmed number of tph from Kings Cross / Moorgate / London Bridge), what else would you suggest. Blocking the core of any route for several hours is likely to lead to widespread disruption.

The randomness of contingency planning is the central issue here and needs reviewing particularly on the service groups involving ECML.

This is true of any contingency plan. Where the infrastructure is limited, naturally not as much of a service can be provided. Ticket acceptance is often implemented and disruption messaging sent to try and move demand to other places, where possible.

In the case of this incident 4 tosh had been removed from Welwyn to Hitchin section so spare capacity theoretically existed. Given the rave reviews given to Luminate traffic management system on GW perhaps it should applied on this end of ECML?

 

[Bald Rick]

Given the rave reviews given to Luminate traffic management system on GW

Depends who you speak to about how rave the reviews are.

 

[iphone76]

Maybe the Elizabeth Line only has driver changes at locations where trains can be put away, that the driver on the train, will always sign. Unlike Thameslink.

Very few driver changes en route. Also have standby drivers at key locations such a Ealing Broadway, Stratford and Shenfield. A few at Gidea Park, where, as you say, trains can be put into the yard.

 

[Horizon22]

The randomness of contingency planning is the central issue here and needs reviewing particularly on the service groups involving ECML.

In the case of this incident 4 tosh had been removed from Welwyn to Hitchin section so spare capacity theoretically existed. Given the rave reviews given to Luminate traffic management system on GW perhaps it should applied on this end of ECML?

I don’t think it’s “randomness” but it’s the gap between specifying a contingency plan service (relatively easy) and delivering that service within the crew and fleet constraints you have and fitting in an amended timetable on the fly (much harder). There are too many variables to discuss as to why one day it might work better than another. There are some methods to help “normalise” this and whether GTR are actively using them is another matter.

Luminate is fine but can suffer with the “garbage in, garbage out” approach and like ARS, over-reliance on features and a lack of flexibility when needed. For example clashes are often a big no, even if it causes a recoverable 1 minute delay. As with anything it depends how it is used (and who by).

 

[Nicholas Lewis]

I don’t think it’s “randomness” but it’s the gap between specifying a contingency plan service (relatively easy) and delivering that service within the crew and fleet constraints you have and fitting in an amended timetable on the fly (much harder). There are too many variables to discuss as to why one day it might work better than another. There are some methods to help “normalise” this and whether GTR are actively using them is another matter.

Luminate is fine but can suffer with the “garbage in, garbage out” approach and like ARS, over-reliance on features and a lack of flexibility when needed. For example clashes are often a big no, even if it causes a recoverable 1 minute delay. As with anything it depends how it is used (and who by).

My service is provided by the 9Rxx's and is generally the first service to be culled from the timetable when there is any problem along its line of route and even when its not occasionally. I understand why this is the most sensible service to take out first as its completely duplicated North of St Pancras by the 9Txx (Brighton-Bedford) it causes very little angst on that side. However, on the South side the 9R doesn't have an entirely overlapping calling pattern but there is a ready made solution with putting stop orders on the 9Jxx (Horsham-Peterborough) services. This where the randomness comes in as to whether they do this or not they just aren't consistent.

 

[Bald Rick]

I understand why this is the most sensible service to take out first as its completely duplicated North of St Pancras by the 9Txx (Brighton-Bedford) it causes very little angst on that side.

Believe me it does!