Page 8 of 11

Re: Large scale UK power cuts

Posted: August 20th, 2019, 1:38 pm
by dspp
taylor20 wrote:For info, NG's initial report:

https://www.nationalgrideso.com/documen ... 1/download


As a minimum it looks like there is a case that many loss-of-infeed scenarios are going to have to be increased to account for (typically) 0.5 GW of embedded generation connected within (a typically affected area of) the distribution network. That's at current penetration levels, and seems to have been predominantly solar PV. I wouldn't be at all surprised to see that at 1.0 GW levels within a few years. The full investigation is going to need to break down the known solar (etc) generators in the loss-of-mains area by size, as it is likely that there will need to be a requirement for increased comms & some form of improved ride-through for the larger solar farms in the future.

Anyway that would seem to imply that NG ESO needs to pretty swiftly increase the infeed loss criteria by 0.5 GW from 1.2 GW to 1.7 GW, irrespective of whether one thinks it in any case ought to be read as being 1.8 GW. Or should that now be 1.8 + 0.5 = 2.3 GW. Hmmmm........

I haven't finished reading, but at least it is now clear that the two Little Barford gas turbines are what account for the second frequency dip, whereas the coincident Hornsea loss and Little Barford steam loss were the first frequency dip. Coincident to within 20-secs, ouch. 737 MW + 244 MW.

I hope the full report also transparently analyses for public review the three similar - but avoided - recent events, if only so we can see whether there is scaremongering going on by some lobby groups.

regards, dspp

Re: Large scale UK power cuts

Posted: August 20th, 2019, 2:22 pm
by scotia
dspp wrote:
taylor20 wrote:For info, NG's initial report:

As a minimum it looks like there is a case that many loss-of-infeed scenarios are going to have to be increased to account for (typically) 0.5 GW of embedded generation connected within (a typically affected area of) the distribution network. That's at current penetration levels, and seems to have been predominantly solar PV. I wouldn't be at all surprised to see that at 1.0 GW levels within a few years.

That's the bit that astonished me. If these renewables are so sensitive, then, as you say, there will need to be a much larger reserve - and who pays for it?

Re: Large scale UK power cuts

Posted: August 20th, 2019, 3:06 pm
by dspp
scotia wrote:
dspp wrote:
taylor20 wrote:For info, NG's initial report:

As a minimum it looks like there is a case that many loss-of-infeed scenarios are going to have to be increased to account for (typically) 0.5 GW of embedded generation connected within (a typically affected area of) the distribution network. That's at current penetration levels, and seems to have been predominantly solar PV. I wouldn't be at all surprised to see that at 1.0 GW levels within a few years.

That's the bit that astonished me. If these renewables are so sensitive, then, as you say, there will need to be a much larger reserve - and who pays for it?


They are not that sensitive, and they are doing exactly what they are supposed to do.

(I'm sure you already know what I am writing below, but for the benefit of others)

What happens is that the lightning event on the transmission line caused any distribution circuit(s) connected to the transmission line to be de-energised. Any small-scale-embedded-generator (SSEG) that is connected to the distribution system 'sees' the loss of mains power and automatically disconnects itself. It does so because all SSEGs are connected in compliance with either the G83 or G59 standard, and they do this automatic disconnection for safety reasons (primarily). Most of these SSEGs will be solar PV systems and (I am guessing, based on experience) the bulk of the capacity is made up of (say) 200kW farm-scale solar arrays rather than (say) 4kW domestic house solar arrays. Anyway all these solar systems have an inverter which is either G83-compliant, or G59-compliant, and so they disconnect automatically. They then wait, watch for the grid to be restored, check that the grid parameters are all in the safe range, and then reconnect - which typically takes them a few minutes after the grid is restored. In the meantime, from the moment of the lightning strike until 2-3 minutes after grid restore there is 0.5GW of missing generation to be substituted for (or frequency falls).

There are two points to this:
1) These embedded generators are doing exactly what the grid authorities (the NG ESO, and the DNOs) have insisted that they do (the G83 and G59 etc standards);
2) The lightning strike in this instance caused the transmission line to be interrupted for only 20-secs; but if it had actually damaged the line then the line would have been down for a prolonged period (say a day) until it is repaired. In the second case the 0.5 GW of small scale embedded generation wouldn't have been able to generate at all (at least not down a 'normal' grid pathway), and so there would be no 'sensitivity' issue whatsoever: there simply wouldn't be a connected grid in this area at all.

Taking the first point, you will have seen sunnyjoe and I alluding to a review of the standards, to 'relax' them in some ways for the larger embedded generators; and to put some 'ride-through' obligations onto them. This is tricky as there also need to be safety precautions taken, and so I am expecting that will lead to an increased level of communications & monitoring being required. That's a good thing, and it is coming.

Taking the second point, this is somewhat more significant. However it is not just a solar/wind etc embedded generation issue. It affects any distributed generation irrespective of whether it is a battery in my house, or a diesel engine. And yes, in extremis it is going to affect provision of reserves and there will be costs associated with it, but that is not something that is uniquely caused by renewables.

regards, dspp

Re: Large scale UK power cuts

Posted: August 20th, 2019, 10:58 pm
by scotia
dspp - many thanks for the response.
I can appreciate that a temporary loss of connection due to lightning strikes will take out the embedded generation, but, as you say, in the case of solar, this should be back in at least a few minutes (and there is really no technical reason why it should not be back sooner). But its the 500MW of embedded generation outage that astonishes me. Surely there were not lots of transmission lines down at the same time from lightning strikes? That's what made me wonder if transients caused by a single lightning strike had taken out embedded generation over a substantial area - even if there had been no loss of supply. Hence my sensitivity comment.
Its about 30 years since I actively tampered with generators on the grid - testing novel governors and load controllers.
Latterly I was much more concerned with SCADA systems - and another surprise is the vagueness of the reported timelines. On the systems I worked with, alarms and switch movements were timed with 10 or 100ms accuracy - and I would have thought this was the norm throughout the supply industry. So the timing of the initial cause and subsequent effects of the power outage should be accurately identifiable.

Re: Large scale UK power cuts

Posted: August 21st, 2019, 8:55 am
by sunnyjoe
scotia wrote:dspp - many thanks for the response.
I can appreciate that a temporary loss of connection due to lightning strikes will take out the embedded generation, but, as you say, in the case of solar, this should be back in at least a few minutes (and there is really no technical reason why it should not be back sooner). But its the 500MW of embedded generation outage that astonishes me. Surely there were not lots of transmission lines down at the same time from lightning strikes? That's what made me wonder if transients caused by a single lightning strike had taken out embedded generation over a substantial area - even if there had been no loss of supply. Hence my sensitivity comment.
Its about 30 years since I actively tampered with generators on the grid - testing novel governors and load controllers.
Latterly I was much more concerned with SCADA systems - and another surprise is the vagueness of the reported timelines. On the systems I worked with, alarms and switch movements were timed with 10 or 100ms accuracy - and I would have thought this was the norm throughout the supply industry. So the timing of the initial cause and subsequent effects of the power outage should be accurately identifiable.


As an industry insider I am also surprised, not only about the 500MW of embedded generation which tripped but also about the blase manner in which National Grid reports it as expected but didn't appear to include it in their estimate of required response and reserve. There are (and have been for several years) initiatives led by National Grid to reduce the sensitivity of embedded generation to such disturbances.

Regarding the timing of recorded events, most will have been recorded to the accuracy you suggest but local clocks are not always synchronised with GPS.

Re: Large scale UK power cuts

Posted: August 21st, 2019, 10:29 am
by dspp
This event took out about 5% of UK electricity supplies.

There is about 13 GW of installed solar in the UK. Almost all of the embedded generation in this area was solar at the time (clearly it would have been different if this event had been in the night).

https://en.wikipedia.org/wiki/Solar_pow ... ed_Kingdom

Almost all of that solar is connected to the distribution network. Since the distribution network in this area was dropped (for about 20-secs) when the transmission line got hit lightning, then all those G83 / G59 solar inverters quite correctly dropped off.

About 5% of that would therefore be 0.65 GW . Since the reported/estimated infeed loss from distributed embedded generation was 0.5 GW this suggests the numbers are about correct.

Since it takes (approx) 2-3 minutes for the G83/G59 inverters to resync following reconnection and the grid coming back into the correct range, then the loss of infeed scenario would appear to need an extra 0.5 GW of response added to the scenarios, or the G83/G59 standards will need to be relaxed/adjusted to not only permit, but require, some form of ride-through for the larger 'commercial' solar farms (and a still considerable, but lesser, amount of additional infeed response for the smaller 'domestic' solar).

If you go to the MCS MID statistics https://www.microgenerationcertificatio ... tatistics/ it shows approx 967,946 small solar PV installations, i.e. the sub 50kW 'domestic' stuff.

If you go to the Ofgem FIT statistics https://www.ofgem.gov.uk/environmental- ... statistics it shows 4.9 GW of small solar PV installations.

That suggests that a typical small solar PV installation is 5kW, which seems about right to me.

This suggests that about 8GW of the UK's 13 GW solar is the larger commercial solar farms, and 5GW is the smaller MCS domestic solar (etc) stuff.

A view might be that the loss of infeed scenarios should be revised to include (say) 5% x 5GW = 0.25 GW of dropped G83/G59 domestic infeed (as that is very difficult to arrange for ride-through). Going forwards all new solar installs above 50kW (the MCS limit) should have some form of grid stability ride-through requirement imposed (which could be - indeed should be - centrally planned/contracted/paid, but locally reallocated) and the same requirement should be imposed on any new fossil or nuclear plant. That does leave an issue to do with the 8GW of existing solar commercial capacity and the extent to which it is grandfathered or not - and if not then another 0.25GW of infeed needs to be planned for, and centrally contracted/paid, but not reallocated back down to the generator.

I would also require all generators of over 50kW capacity to report/communicate in real time what they are doing. But that gets into another discussion about whether the mobile and/or fixed line phone system can be relied upon in just these sorts of emergency situations, as being realistic you cannot have a 51kW solar array paying for a full-on SCADA connection to the NG ESO !

I am expecting these solar capacities to double or triple over the next decade.

(Wind will more than double or triple, but it is almost entirely very large wind arrays, and the loss of infeed for that is being catered for in much the same way as any conventional power plant and it is seldom a hidden embedded infeed. So it is another subject entirely.)

regards, dspp

Re: Large scale UK power cuts

Posted: August 21st, 2019, 12:19 pm
by scotia
dspp wrote:Almost all of that solar is connected to the distribution network. Since the distribution network in this area was dropped (for about 20-secs) when the transmission line got hit lightning, then all those G83 / G59 solar inverters quite correctly dropped off.

Again - thanks - that's the part that I didn't understand - it was a single transmission line which dropped off an entire distribution area which contained 500MW of embedded generation. Ouch! And you suggest this could grow?
I would also require all generators of over 50kW capacity to report/communicate in real time what they are doing. But that gets into another discussion about whether the mobile and/or fixed line phone system can be relied upon in just these sorts of emergency situations, as being realistic you cannot have a 51kW solar array paying for a full-on SCADA connection to the NG ESO !
A simple scada outstation with a backup battery supply, a gps receiver and a mobile phone connection should not cost much more than the devices in my house reading the gas and electricity meters. In the event of a comms loss, it could store the data (with modern memory sizes - for years if necessary!) And most of the data it transmitted could be dumped in a central store which could be interrogated by NG if required. Intelligent on-line control would also be a possibility.

Re: Large scale UK power cuts

Posted: August 21st, 2019, 12:38 pm
by sunnyjoe
dspp wrote:
Almost all of that solar is connected to the distribution network. Since the distribution network in this area was dropped (for about 20-secs) when the transmission line got hit lightning, then all those G83 / G59 solar inverters quite correctly dropped off.


A 400kV transmission circuit was disconnected for 20 seconds due to a lightning induced fault. I have seen no reports that any distribution circuits were disconnected, I would not expect so. If embedded generators (generators connected to the distribution system) disconnected it would be due to voltage or frequency disturbances resulting from the disconnection of the transmission circuit or from the disconnection of the large generators. In an ideal world, embedded generators should not have been sensitive to such disturbances and there is ongoing work to desensitise them.

dspp wrote:I would also require all generators of over 50kW capacity to report/communicate in real time what they are doing. But that gets into another discussion about whether the mobile and/or fixed line phone system can be relied upon in just these sorts of emergency situations, as being realistic you cannot have a 51kW solar array paying for a full-on SCADA connection to the NG ESO !


The European Network Code on Requirements for Generators (recently incorporated into UK law and and into the GB Grid Code and Distribution Code) requires that all new generators > 800W should comply with various stability requirements, and should include a communication interface which can receive an instruction to shut down (but not to send any information). Only new generators >1MW are required to "be capable of exchanging information with the relevant system operator ... in real time". The "relevant system operator" could be the distribution company rather than NG ESO

Re: Large scale UK power cuts

Posted: August 21st, 2019, 4:30 pm
by dspp
From the preliminary NG ESO report : https://www.nationalgrideso.com/documen ... 1/download

* The lightning strike also initiated the operation of Loss of Mains (LoM) protection on embedded generation in the area and added to the overall power loss experienced.

* As would be expected in such circumstances there was the loss of some small embedded distributed generation (totalling ~500MW) associated with the transient voltage disturbance caused by the lightning.

* The lightning strike also initiated the operation of Loss of Mains (LoM) protection on embedded generation in the area and added to the overall power loss experienced.

16:52:34 Frequency response initiates. Systems recorded a net ~ 500MW increase of transformer loadings (expected cause is a loss of injection of MW from embedded generation). [1481MW of cumulative infeed loss] ESO


... so, yes, the lightning strike did trip off 0.5 GW of embedded generation that was connected to the distribution network, we know because it is only being sensed at the step down transformers.

***
re SCADA, the problem is the loss of real-time 2-way info + control, the memory storage issue is a distraction. This is of course one of the motivations for smart metering to get at the demand side response, and to get at the smaller embedded generation. The smart metering comes with a comms layer, but that is not as good as it needs to be, yet. One day maybe, but we are not there yet.

***

regards, dspp

Re: Large scale UK power cuts

Posted: August 21st, 2019, 5:09 pm
by scotia
dspp wrote:From the preliminary NG ESO report : https://www.nationalgrideso.com/documen ... 1/download

* The lightning strike also initiated the operation of Loss of Mains (LoM) protection on embedded generation in the area and added to the overall power loss experienced.

* As would be expected in such circumstances there was the loss of some small embedded distributed generation (totalling ~500MW) associated with the transient voltage disturbance caused by the lightning.

* The lightning strike also initiated the operation of Loss of Mains (LoM) protection on embedded generation in the area and added to the overall power loss experienced.

16:52:34 Frequency response initiates. Systems recorded a net ~ 500MW increase of transformer loadings (expected cause is a loss of injection of MW from embedded generation). [1481MW of cumulative infeed loss] ESO


... so, yes, the lightning strike did trip off 0.5 GW of embedded generation that was connected to the distribution network, we know because it is only being sensed at the step down transformers.

***
re SCADA, the problem is the loss of real-time 2-way info + control, the memory storage issue is a distraction. This is of course one of the motivations for smart metering to get at the demand side response, and to get at the smaller embedded generation. The smart metering comes with a comms layer, but that is not as good as it needs to be, yet. One day maybe, but we are not there yet.

***

regards, dspp

I remain confused. Was there a loss of supply on those parts of the distribution network attached to embedded generation - and consequently they tripped - that's what I would call loss of mains. Or was it simply transient voltage disturbances that caused the trips? Are the embedded generators too sensitive to such transients?
On SCADA - obviously 100% two-way comms is desirable - but its better to have some comms, rather than no comms on all embedded generation! And a simple local store of data which can be read after comms is restored would be an invaluable tool for determining the timeline of a widespread fault.
And I'm still incredulous at the language used in the report "As would be expected in such circumstances there was the loss of some small embedded distributed generation (totalling ~500MW)" When I first read that sentence in the report, I thought that there had been a misprint in the amount of small embedded generation loss - did they mean 500kW - surely not 500MW - but no, it was definitely 500MW. The report writer is clearly a cool customer, writing off 500MW of generation so casually :D
Anyway - I'm long past the stage of being involved. interesting times :)

Re: Large scale UK power cuts

Posted: August 21st, 2019, 5:22 pm
by sunnyjoe
scotia wrote:I remain confused. Was there a loss of supply on those parts of the distribution network attached to embedded generation - and consequently they tripped - that's what I would call loss of mains.

No

scotia wrote:Or was it simply transient voltage disturbances that caused the trips? Are the embedded generators too sensitive to such transients?

Yes, some of them

Re: Large scale UK power cuts

Posted: August 21st, 2019, 5:23 pm
by dspp
scotia wrote:I remain confused. Was there a loss of supply on those parts of the distribution network attached to embedded generation - and consequently they tripped - that's what I would call loss of mains. Or was it simply transient voltage disturbances that caused the trips? Are the embedded generators too sensitive to such transients?
= The G83/G59 protection is supposed to disconnect if the mains goes outside predetermined limits, or if lost entirely. They are being rather vague with the terminology so it is unclear whether supply was completely lost, or just went outside the various limits. It could be a mixture of both for different amounts.


On SCADA - obviously 100% two-way comms is desirable - but its better to have some comms, rather than no comms on all embedded generation! And a simple local store of data which can be read after comms is restored would be an invaluable tool for determining the timeline of a widespread fault.
= I agree in principle, it will come in time, but I expect the comms layer to become a public utility with minimum service standards first

And I'm still incredulous at the language used in the report "As would be expected in such circumstances there was the loss of some small embedded distributed generation (totalling ~500MW)" When I first read that sentence in the report, I thought that there had been a misprint in the amount of small embedded generation loss - did they mean 500kW - surely not 500MW - but no, it was definitely 500MW. The report writer is clearly a cool customer, writing off 500MW of generation so casually :D
= your amazement is shared by me

Anyway - I'm long past the stage of being involved. interesting times :)
= these days this stuff is my day job, but fortunately the UK is only a small fraction .........



comments in red, regards, dspp

Re: Large scale UK power cuts

Posted: August 21st, 2019, 8:07 pm
by PinkDalek
dspp wrote:comments in red, regards, dspp


I've noticed many posts in recent days, not merely at Beerpig's, where posters are using a variety of colours which, for me least, are hard to read, especially on a mobile.

Would not a feature such as italics be just as good or do others find italicised sentences hard to read?

Maybe bold is better or underlined.

PD - attempting to contribute a little bit of light hearted pub related banter and discussion

Re: Large scale UK power cuts

Posted: August 22nd, 2019, 11:38 am
by supremetwo
sunnyjoe wrote:The European Network Code on Requirements for Generators (recently incorporated into UK law and and into the GB Grid Code and Distribution Code) requires that all new generators > 800W should comply with various stability requirements, and should include a communication interface which can receive an instruction to shut down (but not to send any information). Only new generators >1MW are required to "be capable of exchanging information with the relevant system operator ... in real time". The "relevant system operator" could be the distribution company rather than NG ESO

Do you mean 8000W?

I have 4000W of solar panels - the general maximum for installations in the domestic market.

First I have heard of stability requirements and presumably not retrospective if it does encompass 4kWp installs.

Re: Large scale UK power cuts

Posted: August 22nd, 2019, 12:10 pm
by scotia
supremetwo wrote:
sunnyjoe wrote:The European Network Code on Requirements for Generators (recently incorporated into UK law and and into the GB Grid Code and Distribution Code) requires that all new generators > 800W should comply with various stability requirements, and should include a communication interface which can receive an instruction to shut down (but not to send any information). Only new generators >1MW are required to "be capable of exchanging information with the relevant system operator ... in real time". The "relevant system operator" could be the distribution company rather than NG ESO

Do you mean 8000W?

I have 4000W of solar panels - the general maximum for installations in the domestic market.

First I have heard of stability requirements and presumably not retrospective if it does encompass 4kWp installs.

Have a look at the EU regulation document https://www.entsoe.eu/network_codes/rfg/
The relevant section is "Determination of Significance"
The threshold for a type A generator is quoted as 0,8kW - which (assuming the continental comma is equivalent to our dot) is 800W

Re: Large scale UK power cuts

Posted: August 22nd, 2019, 12:11 pm
by sunnyjoe
supremetwo wrote:Do you mean 8000W?

I have 4000W of solar panels - the general maximum for installations in the domestic market.

First I have heard of stability requirements and presumably not retrospective if it does encompass 4kWp installs.


Nope. 800W

It's not retrospective to installations before 27/4/19 which should comply with G83 including its stability requirements

Subsequent small generators should be type test approved in accordance with G98

Re: Large scale UK power cuts

Posted: August 22nd, 2019, 12:37 pm
by supremetwo
scotia wrote:
supremetwo wrote:
sunnyjoe wrote:The European Network Code on Requirements for Generators (recently incorporated into UK law and and into the GB Grid Code and Distribution Code) requires that all new generators > 800W should comply with various stability requirements, and should include a communication interface which can receive an instruction to shut down (but not to send any information). Only new generators >1MW are required to "be capable of exchanging information with the relevant system operator ... in real time". The "relevant system operator" could be the distribution company rather than NG ESO

Do you mean 8000W?

I have 4000W of solar panels - the general maximum for installations in the domestic market.

First I have heard of stability requirements and presumably not retrospective if it does encompass 4kWp installs.

Have a look at the EU regulation document https://www.entsoe.eu/network_codes/rfg/
The relevant section is "Determination of Significance"
The threshold for a type A generator is quoted as 0,8kW - which (assuming the continental comma is equivalent to our dot) is 800W

https://eur-lex.europa.eu/legal-content ... 1e1214-1-1

The word 'solar' in that document is not mentioned.

Re: Large scale UK power cuts

Posted: August 22nd, 2019, 1:48 pm
by dspp
https://eur-lex.europa.eu/legal-content ... d1e311-1-1

6. The power-generating module shall be equipped with a logic interface (input port) in order to cease active power output within five seconds following an instruction being received at the input port. The relevant system operator shall have the right to specify requirements for equipment to make this facility operable remotely.

=============
https://docstore.entsoe.eu/Documents/Ne ... _final.pdf

Need for synthetic inertia (SI) for
frequency regulation
ENTSO-E guidance document for national
implementation for network codes on grid connection
31 January 2018

System inertia is an essential parameter for frequency stability of the electrical power
system. It determines the initial rate of change of frequency in case of a sudden imbalance
between supply and demand (e.g. trip of a large MW source or demand).............

The IGD on High Penetration of Power Electronic Interfaced Power Sources (HPoPEIPS)
contains a detailed analysis in its Appendix 2 of the foreseen development of RES
penetration (based on an analysis associated with TYNDP2016) and consequential
calculated Total System Inertia by 2030 for each Synchronous Area......

The need for SI applies particularly for smaller synchronous areas with high penetration
of non-synchronous generation which tend to have lower total system inertia and greater
frequency volatility (such as Ireland and Great Britain).......

Consideration on limiting initial df/dt
Use of RoCoF as a Loss-of-Mains (LOM) protection is the largest concern in
respect of high initial df/dt, because of potential tripping of embedded generators
through mal operation of the protection when the embedded generation is not
islanded, but simply subject to a system wide fast frequency movement. .......


==============

You are either a rule maker or a rule taker ..........

regards, dspp

Re: Large scale UK power cuts

Posted: August 22nd, 2019, 2:16 pm
by scotia
supremetwo wrote:https://eur-lex.europa.eu/legal-content ... 1e1214-1-1
The word 'solar' in that document is not mentioned.

The document refers to all electrical power generating devices. I don't think the omission of the word solar is of any relevance.
So your solar array appears to be Class A - greater than 800W and less than 1MW (or 0.1MW in Ireland and Northern Ireland)
Sunny Joe has listed the relevant standards - G83 on installations before 27/14/19 and G98 after.

Re: Large scale UK power cuts

Posted: August 22nd, 2019, 3:19 pm
by dspp
scotia wrote: 27/14/19


eh ?