Electrical Supply Problem(s) 02 May

[ChilliSauce]

It takes time to bring pumped storage power on line, although if already running their generators will have the same inertial effect as other conventional generator power stations. Preventing this kind of frequency fluctuation needs something that can respond instantly, like the flywheels mentioned by cl 317.

My understanding is that Saturday's incident was caused by the failure of an underground grid feeder. ("underground" as in below ground, as opposed to overhead pylons, not "London underground").

In all probability with a grid feeder failure isolating some of the network then the generated capacity could no longer meet the demand .. Frequency dropped so the Grid shed-load.. i.e. Disconnected The feeds to pudding Mill Lane ATFS

 

[Taunton]

Interesting - other than Langley and Iver they would be covered by ticket acceptance on London buses. No covering service west of Slough though.

But there being a bus service outside doesn't mean it goes to anywhere meaningful to the passenger. What bus goes from West Drayton to Central London? In fact, increasingly as services are rationalised, the only bus routes at rail, and other TfL, stations are those going at right angles to the rail line.

There's the same on the DLR nowadays, where rail replacement buses for the incessant weekend closures have been reduced, likely because they have run out of budget for them, with just a bland (if not patronising) statement to 'use an alternative'. Except there aren't any going that way.

 

[387131]

GWR did have special stop orders on a few of their Didcots at Ealing Broadway, Southall, Hayes & Harlington, West Drayton, Langley and Iver.

With 345's blocking the relief lines at Maidenhead and Twyford, there was no way to call Taplow & Burnham as we couldn't garentee a relief line run to call at these stops.

Also a 30 min service with 387's calling all stations to Reading would not have coped with the overcrowding.

Paddington station staff reported that 2N62 and 2N64 left completely full and standing with the extra semifast stops of Ealing, Southall, Hayes, West Drayton, Iver and Langley added.

Sadly there is only so much we can do with the limited number of 387's and crew available.

 

[Purple Train]

Apologies if I've missed this, but were 710s and 720s affected to the same degree as the 345s and 730s? And if so, how come things out East recovered much faster?

 

[Recessio]

Interestingly not seeing any news reporting about this. When I search for it, really the only thing that comes up is this thread. Was it just the railways that were affected then, no powercuts or other industries affected by the frequency dip?

 

[GraphitProjekt]

Apologies if I've missed this, but were 710s and 720s affected to the same degree as the 345s and 730s? And if so, how come things out East recovered much faster?

Apparently it was Aventras nationwide which is why the West Midlands saw some disruption too. But definitely, GA were able to recover much better than EL which makes me think there was some other problem for the Crossrail core beyond drivers having to reset every train

My understanding is that Saturday's incident was caused by the failure of an underground grid feeder. ("underground" as in below ground, as opposed to overhead pylons, not "London underground").

grid feeder from whose perspective? Power station feeding the whole transmission grid, or grid substation feeding a railway substation?

 

[Taunton]

Notably not much in the press, because we don't see any other high-tech businesses shutting down. TV stations didn't go off the air (their transmissions are hugely dependent on frequency). Internet and phone systems didn't fail. How can the railway have got to a position where they stop and are then unable to restart.

 

[mrmartin]

My understanding is that Saturday's incident was caused by the failure of an underground grid feeder. ("underground" as in below ground, as opposed to overhead pylons, not "London underground").

Not sure that could be the root cause though. Tho frequency drop was caused by not enough supply (you'd see frequency increase if a large amount of demand dropped because of a substation failing) and the grid feeder by its name feeds demand not supply. Given the frequency drop it's likely around 1GW of supply failed.

So I suspect something like a HVDC link failed or a large generation set tripped. I'm doubtful it was clouds from solar (would have to be an enormous and very sudden cloud to affect so much generation at once!). But I could be wrong.

I suspect the grid feeder you referred to tripped too "early" and shouldn't have done.

 

[TSGN Spotter]

Hmm interesting, there is a Norwich feeder station directly on the Crown Point side of Trowse bridge, so I'd guess it was offline and the whole section fed from Stowmarket. Similarly I believe Kings Lynn feeder has been long term out of use (although GA don't go that way anymore)

I am just wondering doees anyone know how long the Kings Lynn feeder station has been OOU and/or why. May explain why works at Ely seem to cause the power to be completely switched off between Downham Market and Kings Lynn preventing a shuttle service.

 

[Belperpete]

Notably not much in the press, because we don't see any other high-tech businesses shutting down. TV stations didn't go off the air (their transmissions are hugely dependent on frequency). Internet and phone systems didn't fail. How can the railway have got to a position where they stop and are then unable to restart.

Most other industries only have to ensure that their equipment can withstand the fluctuation. In the case of trains, they also have to ensure that they don't generate any interference as a result of the fluctuation that could affect other equipment, such as signalling or telecommunications equipment. And by the very nature of trains, they are limited on the equipment they can carry - it would be impractical to fit a UPS large enough to maintain the traction current for example.

== Doublepost prevention - post automatically merged: 4 May 2026 ==

grid feeder from whose perspective? Power station feeding the whole transmission grid, or grid substation feeding a railway substation?

I can't find the source now, but as I recall the information was from UK Power Networks and was unspecific about what type of feeder (other than it being a fault with an underground feeder cable that had caused the frequency fluctuation). My guess would be that it would be one of the grid feeds into the London area.

 

[najaB]

It takes time to bring pumped storage power on line, although if already running their generators will have the same inertial effect as other conventional generator power stations. Preventing this kind of frequency fluctuation needs something that can respond instantly, like the flywheels mentioned by cl 317.

Which is where layers come into play. Batteries and flywheels provide instant compensation while things like pumped storage spin up to meet the demand or absorb excess.

 

[GraphitProjekt]

I am just wondering doees anyone know how long the Kings Lynn feeder station has been OOU and/or why. May explain why works at Ely seem to cause the power to be completely switched off between Downham Market and Kings Lynn preventing a shuttle service.

The only thing I know about it is from post 50 in this thread https://www.railforums.co.uk/thread...h-fed-from-one-end.236747/page-2#post-7626812
I wonder how long they can go on like that with no N-2 redundancy for Milton Fen FS.

 

[Taunton]

Which is where layers come into play. Batteries and flywheels provide instant compensation while things like pumped storage spin up to meet the demand or absorb excess.

Pumped storage comes up within seconds. There was a TV documentary some years ago showing the power supply operators watching Coronation Street, as there was a well-known demand spike at the end of the programme as many houses switched on lights, kettles, etc. They saw demand rising on their meters at the end, said "Go", and the system in Snowdonia came up to full power right then.

 

[Belperpete]

Pumped storage comes up within seconds. There was a TV documentary some years ago showing the power supply operators watching Coronation Street, as there was a well-known demand spike at the end of the programme as many houses switched on lights, kettles, etc. They saw demand rising on their meters at the end, said "Go", and the system in Snowdonia came up to full power right then.

Having visited both pumped storage schemes, I can say that is not the case. They need significant warning. There is quite a drop between the upper dam and the power station for the water to flow through. And they have to open the sluice gates slowly, as opening (or shutting) them quickly would cause a water-hammer type surge that could destroy the generators and even the pipelines too. Only once they have got the generators up to speed and synchronised with the grid under no load conditions, can they then increase the water flow and start generating. It is only this last bit that can be done quickly, getting them up to speed can take about a quarter of an hour or so if I recall correctly. So for an expected surge in demand, like the ad break in Coronation Street, they will have had the generators up and running long before the ad break is signalled on screen. They are no use for coping with a totally unexpected spike.

And my understanding is that nowadays the costs of keeping them running "just in case" mean they are no longer used for this purpose anyway. Nowadays they are used purely to balance out peak and off peak loads.

 

[QueensCurve]

It takes time to bring pumped storage power on line, although if already running their generators will have the same inertial effect as other conventional generator power stations. Preventing this kind of frequency fluctuation needs something that can respond instantly, like the flywheels mentioned by cl 317.

My understanding is that Saturday's incident was caused by the failure of an underground grid feeder. ("underground" as in below ground, as opposed to overhead pylons, not "London underground").

If you do the public tour at Dinorwic or Cruachen they boast about how quickly they can come online, but it is still too slow to prevent these problems.

 

[Nicholas Lewis]

This was caused by a significant drop in power across three interconnectors of nearly 2GW from Belgium, Holland and Denmark. It happened over a few minutes and with Drax on line along with 8GW of CCGTs there was plenty of spinning reserve to help arrest frequency decline. NESO haven't commented and as its above 49.5Hz minimum they dont really have to.

Certainly another event that the wider railway industry should look into as whilst its a rare event it nonetheless is a credible event that could repeat itself.

 

[Dazza12]

Hi,

Having visited both pumped storage schemes, I can say that is not the case. They need significant warning. There is quite a drop between the upper dam and the power station for the water to flow through. And they have to open the sluice gates slowly, as opening (or shutting) them quickly would cause a water-hammer type surge that could destroy the generators and even the pipelines too. Only once they have got the generators up to speed and synchronised with the grid under no load conditions, can they then increase the water flow and start generating. It is only this last bit that can be done quickly, getting them up to speed can take about a quarter of an hour or so if I recall correctly. So for an expected surge in demand, like the ad break in Coronation Street, they will have had the generators up and running long before the ad break is signalled on screen. They are no use for coping with a totally unexpected spike.

And my understanding is that nowadays the costs of keeping them running "just in case" mean they are no longer used for this purpose anyway. Nowadays they are used purely to balance out peak and off peak loads.

If the generators are synced, spinning in air then Dinorwig will from from zero to full load in less than 20 seconds - that is from zero water flow to full water flow. As you note it takes somewhat longer to get those generators synced. One of the main reasons for its existence was to allow Sizewell B to be built - the answer to the question "What happens if Sizewell B (what would have been the largest generator on the grid) trips and we loose 1200MW of nuclear?". Of course, its ability to support peaks like the infamous Coronation Street tea breaks was also handy.

As noted, it isn't used in quite that way any more with more of an energy storage role rather than fast response.

 

[edwin_m]

Apparently it was Aventras nationwide which is why the West Midlands saw some disruption too. But definitely, GA were able to recover much better than EL which makes me think there was some other problem for the Crossrail core beyond drivers having to reset every train

Speculating, it could be something to do with the multiple complex train protection systems on the 345, or that the reset process is more difficult, or that the crews were less well trained in using it.

 

[sharpener]

In the current issue of Private Eye, their energy correspondent "Old Sparkay" has a piece about grid stability and solar power, particularly the issues in dealing with sudden fluctuations from solar farms. Obviously no link available as PE is printed only (it's in issue 1674, p13).

One problem is allegedly that a lot of solar farms connect to local networks not at Grid level, so NESO has limited access to or control over their output intentions.

As well as the infamous Spanish grid collapse, it says there was a German problem earlier this year when solar generation was 30% different from the prediction.

I would guess proper solar farms will be connected at 11 or 33 kV at least, the problem is increasingly with domestic-level solar at the nominal 230V. There is already so much of it in our village that I am regularly seeing more than 253V in the middle of the day, and am awating a response from NGED as to what they are going to do about it.

From the stability point of view domestic solar is no help, as out-of-spec frequency will simply cause this generation to shut down, thus reducing supply when it is needed most. It is embodied in the specs of domestic size inverters (with significant variations in the freq and voltage limits in various European states).

When I worked in the wind turbine industry 20 yrs ago there was a requirement for wind turbines to keep generating in low voltage/low freq situations to help stabilise the grid. And I am sure this is the case today with even more bells and whistles. There is a lot of rotational inertia in a wind turbine bc the blades are so massive.

 

[Vexed]

This was caused by a significant drop in power across three interconnectors of nearly 2GW from Belgium, Holland and Denmark. It happened over a few minutes and with Drax on line along with 8GW of CCGTs there was plenty of spinning reserve to help arrest frequency decline. NESO haven't commented and as its above 49.5Hz minimum they dont really have to.

Certainly another event that the wider railway industry should look into as whilst its a rare event it nonetheless is a credible event that could repeat itself.

I had emailed NESO the day after the frequency drop. They just replied and this agrees with what you said - a 1.6GW loss of interconnectors combined with an increase in demand. No details on the cause of the interconnectors dropping.

Our statutory frequency limits are 49.5Hz and 50.5Hz and frequency remained within these limits.

On 2 May at approximately 16:00 hrs the frequency was at the lower end of this range following an indicated demand increase of approximately 1GW coincident with a net transfer drop on interconnectors of approximately 1.6GW.

We aim for operation closer to 50Hz using many tools, services and interventions and these were employed and the frequency returned to 49.8Hz by 16:05hrs and to 49.96Hz by 16:08hrs.

If anyone is particularly interested, NESO might provide more details if you submit a FOI request.

 

[a_c_skinner]

We are coming at this from the wrong angle. Grid frequency is going to become harder to maintain in the tight limits of the old coal fired days. So we need any electrical kit (even trains) to be hardened against failure because of such modest frequency dips. This must be possible, though I couldn't say how.

 

[SansHache]

Most modern electric locomotives and multiple units have a voltage monitoring transformer connected between the pantograph and the main circuit breaker (VCB). This is used to monitor that the supply voltage and frequency are within the acceptable limits. If detected out-of-range, the fault handling control should immediately open the VCB and hold it open until the supply is back in range. Unfortunately it appears that some types of rolling stock treat this a a more serious fault that locks the VCB open until a power-on reset of the train is performed (or even require technician reset with a laptop in the case of the Class 700s).
As a former designer of Traction drive protection systems my philosophy was always based on how would I want the train to perform if I was travelling home on it after a long day at work!

 

[Belperpete]

We are coming at this from the wrong angle. Grid frequency is going to become harder to maintain in the tight limits of the old coal fired days. So we need any electrical kit (even trains) to be hardened against failure because of such modest frequency dips. This must be possible, though I couldn't say how.

Trains with old-school traction control systems would produce predictable interference frequencies. However, it was soon realised that solid-state traction control systems could produce a whole range of frequencies, which could cause unacceptable interference with signalling and telecommunications equipment. So modern traction control systems are required to shut down if they start generating unacceptable frequencies. The problem is that if the frequency of the incoming traction supply varies too much, the traction control system can start producing frequencies outside of what is acceptable, and so will shut down.

To increase the bandwidth of acceptable frequencies probably requires a lot of the signalling and telecommunications equipment installed in the 1980s and later to be replaced, as much of it relies on frequency sensitive equipment. Such equipment was widely used for track circuits, point detection and remote control and indication systems, to make them immune to both AC and DC traction.

Modern signalling systems are designed such that if a wrong-side failure is detected, the system will shut down until a technician replaces or resets the affected equipment. Allowing the system to automatically restart is unacceptable, as it introduces the possibility of a second fault masking the first. I suspect that as the traction control shut down is designed to protect signalling equipment, it employs the same methodology if it assumes that unacceptable frequencies are a result of a wrong side failure of the traction control system.

If such shut downs are to become more common if frequency variations become more common, then it might require significant rethought on how such shut downs are triggered and recovered from, without affecting the integrity of what they are intended to do. Requiring a technician to come out and reset the train if the equipment has failed may be acceptable, but shouldn't be necessary if the problem isn't with the train. But equally, the train shouldn't be allowed to carry on producing frequencies that could potentially cause a wrong-side signalling failure.

 

[davews]

As somebody very aware about EMC and its affect on normal radio I am a bit puzzled by your explanation. How can a few percentage change in the 50Hz frequency affect the accuracy of things like track circuits (presumably working at dc)? Or does the signalling system itself use similar low frequency data which can clash from the close proximity of magnetic radiation from the traction motors? And at 50Hz all the coupling will be by magnetic fields rather than normal RF frequency coupling.

 

[edwin_m]

Trains with old-school traction control systems would produce predictable interference frequencies. However, it was soon realised that solid-state traction control systems could produce a whole range of frequencies, which could cause unacceptable interference with signalling and telecommunications equipment. So modern traction control systems are required to shut down if they start generating unacceptable frequencies. The problem is that if the frequency of the incoming traction supply varies too much, the traction control system can start producing frequencies outside of what is acceptable, and so will shut down.

As somebody very aware about EMC and its affect on normal radio I am a bit puzzled by your explanation. How can a few percentage change in the 50Hz frequency affect the accuracy of things like track circuits (presumably working at dc)? Or does the signalling system itself use similar low frequency data which can clash from the close proximity of magnetic radiation from the traction motors? And at 50Hz all the coupling will be by magnetic fields rather than normal RF frequency coupling.

I'm also a bit confused. Are you saying either the traction system or the signalling equipment uses the mains as a frequency reference?

For traction I think that's unlikely, not least because the same equipment can also work on a DC supply. The electronics in a modern system (as fitted to 700s and Aventras) need to generate a wide range of frequencies to feed into the motors, and this is surely best done based on an accurate internal timer.

For signalling, the 50Hz track circuit is presumably referenced from the mains, but all traction packages operate at higher frequencies. Otherwise Reed track circuits and FDM transmission systems are the most obviously susceptible to interference as enough of a single harmonic at the critical frequency could cause a wrong-side failure. But these use something like a mechanical tuning fork as a frequency reference. I don't know how newer types like the TI21 get their reference but again I'd expect it to be some sort of internal timer.

 

[GraphitProjekt]

Or does the signalling system itself use similar low frequency data

I wonder if they were talking about the equipment used to remotely control interlockings from the central signalling centre. Older installations could be using coaxial or perhaps twisted pair with FDM.

 

[a_c_skinner]

soon realised that solid-state traction control systems could produce a whole range of frequencies

Thanks for a full reply, something of which I was not aware. I'm intrigued that we do not find ourselves able to (?easily, ?economically) prevent this either at train or signalling level and obviously at the current (sorry, present) level of problems it isn't urgent but it seems likely to become more intrusive as time goes on. Not needing a technician to reset is an obvious precaution even if it would mean some sort of degraded operation as preferable to being stationary in the middle of nowhere.

 

[andyjhatton]

I thought most modern traction rectifies the 25kV50Hz to DC which then goes into the traction inverters.
Which is partly why it's fairly easy to make dual voltage 750vDC/25kV50Hz units; as either way you're feeding the inverters with DC.

In which case, why is the frequency of the AC supply so critical?

 

[Pigeon]

There will almost certainly be a switching stage on the input side as well, for power factor correction and the like, which necessarily does see the raw 50Hz, so it is reasonable for the operating frequency and duty cycle of that stage at least to be directly affected by variations in the input frequency. But that's beside the point really. The internal frequencies used by the power converters vary continuously over a massive range according to output power and input conditions; the fundamental frequencies and their harmonics go all the way from sub-audio frequencies up into megahertz at least (theoretically to infinity) and they can be anything within that band - they don't leave convenient gaps coinciding with frequencies used for AC track circuits or to which the signalling system is otherwise unusually susceptible. (And if they did leave such gaps, it would have to be done by the control software of the power converters deliberately skipping certain frequency bands, which it would do based on its own internal timing reference.)

And this is all in any case beside the larger point, which is that irrespective of the intricate details, the specification for the power supplied to the railway states that the frequency shall be 50Hz +/-0.5Hz. Therefore the stock must be able to cope with that much variation without complaint. Since it evidently doesn't, it's the fault of whoever built it for not doing a proper job.

 

[edwin_m]

The internal frequencies used by the power converters vary continuously over a massive range according to output power and input conditions; the fundamental frequencies and their harmonics go all the way from sub-audio frequencies up into megahertz at least (theoretically to infinity) and they can be anything within that band - they don't leave convenient gaps coinciding with frequencies used for AC track circuits or to which the signalling system is otherwise unusually susceptible. (And if they did leave such gaps, it would have to be done by the control software of the power converters deliberately skipping certain frequency bands, which it would do based on its own internal timing reference.)

The traction electronics and its controlling software do indeed have to avoid the signalling frequencies and their sub-harmonics.