Load Shedding (ALCS? Kaluza?)

  • 20 August 2019
  • 14 replies

Userlevel 4
So the initial report on the Friday power cut that left the railways in chaos is out now.

The 1GW of backup was not enough to save the grid from having to remove 5% of customer demand.

In his book Without the Hot Air, Prof McKay indicates that UK fridges represent an average load of 500MW. It looks like had we been able to load shed the fridges using ALCS the 1GW of FFR would probably have saved the grid from the outage.

But we seem to be nowhere with ALCS :(

So to the point which is a series of questions

1) Did OVO receive any signals from the grid to shed load or even to add generation via V2G?

2) Given sufficient EV charge points could Kaluza respond quickly enough to protect the grid from what we hope will be a rare event?

3) How can I connect my fridge to Kaluza?

4) Is ALCS a dead parrot?

Best answer by Darran_OVO 11 September 2019, 13:26

Hi all, been busy checking into this with our teams to provide some answers @NoPoke

In short, here are the responses to your questions:

1) Did OVO receive any signals from the grid to shed load or even to add generation via V2G?The V2G units measured the grid frequency during the event and detected it dropping as Hornsea and Little Barford tripped. The V2G units are not yet providing response to National Grid based on these measurements but in the future during such an event they would discharge energy from connected EVs to counteract the system imbalance as effectively as possible.

2) Given sufficient EV charge points could Kaluza respond quickly enough to protect the grid from what we hope will be a rare event?
Absolutely, we expected Kaluza connected EVs to be a key response resource for such events in the future.
View original

14 replies

Userlevel 7
Badge +3
Well I'm glad someone has raised this Topic. I actually wrote a report on the Grid Shut-down which I sent to a senior Government Minister on Monday 12th. So I managed to beat the official report by a week!

A. You are correct to ask if ALCS connected devices could've saved the situation. The answer is most certainly yes... if we actually used that system in the first place.

As far as I know, there is no Auxiliary Load product available which connects to ALCS. That includes devices from Kaluza, who have better reason than most to have embraced that technology.

ALCS isn't so much a dead parrot, as a parrot that is yet to to be born.

B. The issue which resulted in National Grid throwing the breakers to remove load across large areas of the UK was the drop in frequency. The 50Hz we use is the only constant that makes our National Grid operate. Voltage and Current can fluctuate without loss of functionality.

It is not possible to connect a source of generation to the electricity grid unless it is first synchronised to the 50Hz frequency. If you attempt to do so, the surge current will throw the breakers and remove that source of generation.

If faults elsewhere mean that the frequency is falling and/or erratic, then other stand-by generation cannot be brought in to boost supplies.

Keeping the 50Hz stable must be the highest priority for National Grid. On Friday 9th, the frequency fell to 48.8Hz.

C. The National Grid does have some large scale storage available, such as the hydro-electric pumped-storage generators at Ffestiniog and Dinorwig in North Wales.

Even so, these could take up to a minute to come on-stream, and may already have been operating at capacity during the Friday evening peak. (National Grid must pay to retain spare capacity on standby). But whoever saved the Grid on Friday 9th had only seconds to act.

D. You mention about connecting your fridge to the Kaluza platform. It's the right question to ask.

Yes, the Kaluza platform could've acted to remove loads, but not in the way it has currently been implemented.

As for putting fridges & freezers onto a system using Software Management for Distributed Load Control, there are better devices to implement. You need things that are power-hungry but don't suffer from deficiencies if they are not supplied with power within a certain time-frame.

Storage Radiators and Immersion Heaters are better examples of load-control devices which might've saved the blackouts on 9th.

There is no device yet commercially available which would permit such pre-existing load devices to be connected to Distributed Load Control systems. However I had a meeting with a business colleague yesterday where we discussed how one might be created, and how its implementation could be undertaken at very low cost to the end-user.
Userlevel 4
Dynamic Demand produced a device ages ago. AFAIK no one took it up. The frequency meter on their website is mesmerising. but you do have to enable flash, which gives you some idea of how old their technology is.

Shedding storage heaters would not have helped much on Friday as it was a peak time event. You can guarantee that there are a lot of fridges running at any hour of the day.

I'm hoping someone from kaluza will join in.
Userlevel 7
Badge +3
That device from Dynamic Demand could've been useful. But there would need to have been some incentive for customers to have wanted to install it.

The beauty of using Smart Meters to be the messengers for Load Control is that they are already installed widely and numbers are increasing.

Yes, you're right that Storage Radiators wouldn't have been any good if we tried to switch them off at 5pm on a Friday. I was looking at a diagram I have here, but failing to notice that it assumed we were already using 30-minute Time Of Use tariffs. We would've had Storage Radiators running in my area if that had been the case, because we had oodles of renewable power available.

I'll nudge someone in Kaluza to have a look at this Topic, but there's no guarantee they'd want to join in of course.

I found another frequency analyser here which doesn't require Flash.

Further down the page I found the text "The greatest fluctuations occurred regularly on the hour change..." which is quite understandable.

One of the features built into SMETS2 is that ALCS switch-on commands are staggered by combining a fixed delay and a random one. This avoids surges when 1000's of households all try to set the same time for devices to run.
Userlevel 4
That on the hour change is a big deal on the continent.

Also would the SW have been able to export its excess generation?
Userlevel 7
Badge +3
And the on-the-hour-change will be even more significant as we are encouraged to buy EVs.

It seems very odd to me that DCC and Ofgem expend effort to design an ALCS system that avoids the problem, but then don't back it up with a regulatory structure to prohibit the connection of EV chargers which allow customer choices to generate surge currents.

No, the SW Region would be unable to export the excess renewable generation because there's no spare capacity on the 400kV Transmission Grid.

However, if customers were choosing TOU Tariffs, then we wouldn't have nearly so significant a peak in early evening. Thus the power required from National Grid would be substantially lower if there was a supply deficit.

And if West Country consumers also had managed battery storage within their immediate locality, then this could be configured to deliver power onto the 11kV Distribution Grid if there was an alert signal issued from elsewhere in GB.

And that brings me to two issues where I simply don't understand OVO's strategy:

a. Why isn't OVO taking the lead in offering TOU tariffs?

b. Why do Kaluza's storage products have such a low max-output current?

The original spec for the Indra-designed Storage Battery was that it could deliver 10kW peak output. That is sufficiently in excess of what would be consumed in the host-house that it would be useful in keeping other houses supplied which were on the same sub-station feed and phase.

Now we're being invited to consider the Home Storage Battery products available from a partnership with Sonnen GmbH. But those are limited to 3kW output.

This what I'll call "selfish technology". Instead of the customer becoming part of a community-wide response, the marketing strategy is to offer us products whose output power equates to merely removing our own house from the evening peak.

This differs significantly from the ethical technology strategies championed by Stephen Fitzpatrick, which attracted me to become an OVO customer in the first place.

It feels like OVO is losing its soul.
Userlevel 6
Badge +1
I doubt if Fridges or Freezers would have saved the day; I recently bought an A+ freezer which lives in the garage and it seems to use about 30W every two hours.

Similarly, my "alcohol" fridge which also lives in the garage uses about 35W every two hours, this is soon to be replaced by an A+ variety as it is over 30 yrs old, my guess is that this will use less that 30W.

Incidentally, both appliances were "Freecycled" and have gone to new homes.
Userlevel 7
Badge +3
Unfortunately the energy-rating on fridges and freezers is unrelated to their efficiency.

Cooling cabinets are scored on the effectiveness of the insulation. The test involves a measured quantity of hot water being placed in the cabinet and measurements taken of how quickly it reaches the external/ambient temperature.

You can buy a fridge/freezer with an A-rating and still have it gallop through the watts due to a poorly design condenser. There's no incentive for the manufacturer to improve the temp-decrease per watt figure.... in fact - it's not usually even specified.
Userlevel 7
Badge +2
Hi all, been busy checking into this with our teams to provide some answers @NoPoke

In short, here are the responses to your questions:

1) Did OVO receive any signals from the grid to shed load or even to add generation via V2G?The V2G units measured the grid frequency during the event and detected it dropping as Hornsea and Little Barford tripped. The V2G units are not yet providing response to National Grid based on these measurements but in the future during such an event they would discharge energy from connected EVs to counteract the system imbalance as effectively as possible.

2) Given sufficient EV charge points could Kaluza respond quickly enough to protect the grid from what we hope will be a rare event?
Absolutely, we expected Kaluza connected EVs to be a key response resource for such events in the future.
Userlevel 7
Badge +3
I've resisted commenting overnight, just to see if @NoPoke was going to reappear first 😉

I'm really glad to see the aspirations suggested in these two responses from the Kaluza Teams. However, I think there are technical reasons why the V2G charger may fail to meet those aspirations.

I'm deliberately keeping these technical explanations "light" so that others can join in. Please don't have a pop at me for seeming to not understand the underlying physics, else I'll ask @Amy_OVO to sort you out!

A. Imbalance in our electricity supply occurs when we connect heavy loads to a single-phase, such as 99.9 of our houses have.

The imbalances are caused either by high-current loads taking energy, such as a standard EV charger, or by supplying energy, as with a PV Solar-Panel or V2G charger.

Because an imbalance is due to the single-phase connection, it is most evident at the local substation.

As you progress further up the chain of transformers, the effect of imbalances are lessened because the random distribution of single-phase high-current loads is dissipated across the wider area served by the higher order 3-phase transformers.

B. It is possible to use a high-current load to counter an imbalance at a local substation. However, to do so it is essential to know what phase you're on!

There is no GB map depicting which house connects to which phase. Nor does the Kaluza Platform have an input to its software to derive that information.

C. The max output from OVO's V2G grid-connection is around 5kW. On a Friday evening at around 5pm (when the outage occurred on 9th Aug), it is the start of the evening peak. A typical house will be drawing around 2.5kW.

If the Kaluza Platform had the capability to respond to an emergency demand-call from National Grid, it would be able to supply the host-house and one other. That's not nearly enough power being pumped back into the system.

I could offer a greater effect to National Grid by having ALCS-connections to other in-home loads such as an electric shower (10kW), storage radiators (3kW each) and an immersion heater (3kW). Taking those off the Grid via an ALCS command to SMETS2 meters would've enabled National Grid to have countered the outage on 9th Aug.

It took only 8 minutes for National Grid to bring enough spare generation back on-line. There would be complaints from a few thousand people whose electric showers stopped working, but I doubt if anyone would even notice the loss of 8 minutes power from their storage radiators!

Let me stop there.

I would very much like to see the OVO/Kaluza V2G charger play a part in reducing the risk of a future grid-outage. But it will need to be developed a long way beyond the current product to be of any practical use in supporting the grid.
Userlevel 4
I wasn't too sure how to respond, not least because I suspect that the smart meter roll out will be an expensive white elephant. I don't anticipate mandatory adoption of TOU tariffs, which leaves ALCS as the upside for smart meters and with little (zero??) deployment of ALCS after several years that is not looking likely.

Regarding lack of knowledge of phase connection at local substations. Phase voltage varies not just with time of day but between phases as loads are not balanced across the three phases. What this means is that if you can measure voltage at customer houses you can establish which houses share the same phase using correlation. Then either by installing a reference monitor at the substation of by knowing (small database) the actual phase connection for a handful of houses per transformer it is possible to establish the phase of all smart meter connected houses for that local transformer.

I don't know if OVO subcontract smart meter installation. It would be useful to start collecting phase information as part of the smart meter install and certainly as part of the V2G trial.
Userlevel 7
Badge +3
I too am disappointed at the energy industry's response to ALCS, @NoPoke.

However, unlike you I think that there will have to be regulatory pressure to ensure that ALCS and genuine variable tariffs are implemented. As the proportion of renewable (intermittent) energy increases, and EV's become more common, it will become apparent that the existing methodology behind the electricity grid is unsustainable.

Turning to your last point, it isn't possible to ascertain the phase feeding a house when a Smart Meter is installed. There's no signal which can be detected to enable such a conclusion to be reached.

Even if we used a satellite-based clock to determine when Phase L1 reached peak +ve voltage at a particular power station, you would still need to know the configuration of each transformer on the Transmission and Distribution networks in order to correlate this with the phase feeding a house.

Transformers used on the grid are either Dyn1 or Dyn11, which means that the phases are shifted +/- 30º between the primary and the secondary. There are 230,000 ground mounted substation transformers in GB, and a similar number of primary, bulk, and grid transformers upstream of those. There is no master-map to show all the phase shifts.

I agree that it might have been possible to implement a phase-detection algorithm into SMETS2 meters, but it can't be added into the existing hardware designs. Our electricity meters provide a 20-sec cumulative output to the IHD. They produce a per-minute cumulative total which is stored internally. This is used to generate the 30-minute totals which OVO collect just after midnight.

Voltage fluctuations between phases happen faster than this, and cannot therefore be recorded by Smart Meters.

As I've been monitoring a sub-station for almost a year now under the OpenLV Project, I've had access to 146 different parameters which are collected at intervals of 1 minute by the LV-CAP monitor in the substation. Current differences (ΔI) show greater promise as a method of deriving a statistical method to determine phases by analysing Smart Meter data, but on their own, measurements of ΔI wouldn't be accurate enough.

Of course, you'd still need to have an LV-CAP monitor in each and every local substation. We've had eighty available for the last year, but this is about to be reduced to less than fifteen.

EV Chargers would be better suited than Smart Meters to having phase-detection built into them. But I don't know of any EV Charger manufacturer who's yet bothered to investigate this.
Userlevel 4
No you don't try to establish the phase directly. The daily variation in voltage allows you to work out which premises share the same phase. Then either by installing a voltage monitor at the transformer or by knowing the phase of a handful of premises it is possible to work out the phase connection of all the other smart meter connected premises on the same transformer. No need for precision clocks. Average voltage over a several minute interval collected over a day will be more than enough data to perform the cross correlation.

In this paper they went even further up the distribution network.
Userlevel 7
Badge +3
I'll agree that others have been exploring this problem, and many thanks for posting the link.

May I share here two graphs that illustrate my point. Both are from the same 11kV/440v substation.

This is average voltage on all three phases taken at 30min intervals over a week. In order to see these small fluctuations I have expanded the y-axis and put 240v at the origin.

For comparison, here are current fluctuations on all three phases on one of three feeds from the same substation, also over a period of a week:

If I'm looking for a parameter where there is a large enough swing to discriminate between the three phases, then current wins hands down.

In fact the voltage drop on the underground cables is very similar to the fluctuations we'd be trying to detect.
Userlevel 7
Badge +3
I've now read the research paper you referred to above @NoPoke.

Yes, it's an interesting approach. But the researchers have used the USA local distribution network for their study. This has SCADA monitoring equipment close to their substations, which we don't have in the UK.

Secondly, they required voltage measurements from Smart Meters in homes to an accuracy of 0.1v which our SMETS meters don't record.

Having said that, I do like the statistical analysis approach they've taken. This works well with the very large data-sets they obtained.