How much continual energy/power is required to keep a site like this up and running?

It's not a dumb question, if the complexity of the answer is anything to go by.

With the battery fully charged, the maintaining power for the electronics and any cooling might be a few kW, then there is the self-discharge of the batteries, maybe 10kW? These are just my estimates, the real figures could be higher.

The point is that the system expects to be cycled, and that's where the inefficiencies creep in. There are losses in the power converters, from AC to DC and back again. Power converters at this scale are normally quite "agricultural" - they might use 12-phase thyristor invertors, which involve a fair bit of smashing voltages together. This is what is used for the UK-France DC link, with efficiency of maybe 95%. Then there is the battery coulombic efficiency, which is very high for Li-Ion batteries, at low current - but drops significantly if you run them at C/2 or C/4 rates (i.e. so they charge or discharge in 2 hours or 4 hours respectively).

It is this 2-hour rate and 4-hour rate that gives you their published figures, the overall round-trip efficiency. This includes all the losses described above.

Quoted figures for the Tesla megapack are

2-hour duration:

Power & Energy: 1,927 kW / 3,854 kWh per Megapack

Round Trip Efficiency: 92.0%

4-hour duration:

Power & Energy: 970 kW / 3,878 kWh per Megapack

Round Trip Efficiency: 93.5%

source: https://electrek.co/2022/09/14/tesla-megapack-update-specs-price/

"diesel generators are quite compact, very cheap, well understood, generally reliable"

Yeah, about that ... the few times when the poo has really hit the fan in my career have been when those generally reliable gennys decide to not be reliable, and not so well understood. And the operators haven't understood the cutover/cutback procedures sufficiently well. Wrecked weekends, whole nights of lost sleep

Agreed. When I read that, I thought Mr Buss had never had to use one in anger.

His linkedin profile confirms that. https://www.linkedin.com/in/andrewbuss/?originalSubdomain=uk

Perhaps get out of the consultancies and into the sharp end before pronouncing efficacy of technology?

Yup... Last time we ran a test, a room full of smug gits suddenly became a herd of "whoah shit"s as they charged down the very very long corridor to where the back up generator was. Meanwhile a clock was ticking.. 5,4,3...

When I worked at a large facilities management company, we did a disaster recovery test. The diesel generator fired up lovely and then a few minutes later conked out due to lack of fuel. The chap who drove the van to deliver the tape offsite, used to fill his van from the generator tank, rather than going to the garage. Now you might have thought the finance people would have wondered why he wasn't putting receipts for diesel, or why the monthly account at the garage was so small, but apparently not.

Nope, I've never had finance asking why money wasn't being spent on expense items like this.

The question that should be being asked though is - Was road fuel duty paid on the Genny fuel?

Something that seems to be missing:

If the data centre batteries are providing energy to the grid, if there is then a powercut, they'll be out of energy to provide backup.

Choosing one or the other task would seem advisable.

Let's get something out of the way: I'm in favor of this, and I'm generally in favor of any attempt to contribute to solving the energy problem. I do not like comments on the line of "they should be doing X instead", because I think that we ought to deploy all credible solutions, aggressively and at the same time.

That said, what about decommissioning? This is a site full of hazardous chemicals, that don't have an infinite useful lifespan, and the recycling of which is not yet working perfectly. Has decommissioning been factored in? We make quite a big deal of it when talking about nuclear; it seems only fair that we should talk about it in the context of other energy-related technologies too.

Not to mention that to make wind energy actually reliable in the UK, you'd need to build somewhere around 9,000 of these - that would allow you to power 30 million homes for about a week.

Which, in turn, would require near-doubling the amount of energy production in order to both charge the storage and also meet regular demand. This is just one of many reasons why grid-scale battery storage is a stupid, stupid idea. It's only touted because of the need to compensate for the unreliable nature of wind and the fact that solar (already a poor proposition in the UK) supplies at about a third of its nominal capacity when demand is highest.

It's only touted because of the need to compensate for the unreliable nature of wind and the fact that solar (already a poor proposition in the UK) supplies at about a third of its nominal capacity when demand is highest.

But this is the point. It's about compensation and collecting subsidies, not doing anything to overcome the fundamental problems with 'renewables'. As rg287 points out..

196MWh is a nice amount for frequency management and smoothing renewable inputs.

Especially when the grid/system operator will pay you a LOT of money for that service. So it massively inflates the cost of 'renewable' energy, along with providing some arbitrage ability, ie charge when prices are low, discharge when they're extortionate because there's no wind and impending blackouts. We may get a taste of this next week, depending on how our weather develops.

I agree that decommissioning needs to be included in any fair assessment, but it's more complicated than that.

Firstly, I don't believe that Li-Ion batteries are particularly toxic, certainly nothing at all like nuclear, and their reprocessing is already implemented on a large scale.

Secondly, there is a good case for taking "exhausted" car batteries (Li-Ion) that have fallen to 70% capacity, and rather than recycle them, just continue to use them in fixed installations till they drop to 50%. This would easily double their working life.

Thirdly, flow batteries are possibly a better contender for long-term storage, like keeping a weeks worth of power stored.

I worked on Dinorwig pumped storage power station (near Snowdon in North Wales) - that can store about 9.1GWh - over 46 times as much energy.

A side note - the original computer system monitoring the Dinorwig power station was a PDP 11/34 that handled approximately 5000 plant inputs and drove 3 displays and 6 line printers. Handling all that on a small 1/3 MIP computer with 160kbytes of RAM and 4.8Mbytes of disk storage was an interesting challenge!!!

That works both ways. Based on the capacity they seem to be using about 60 megapacks. Assuming a generous spacing that's 60 X 100 square metres or just over half a hectare. I'm guessing Dinorwig is slightly larger than that. At a rough estimate I'd say just the reservoir is about 30 hectares. So this compares favourably in size and has the bonus that you don't have to find a mountain.

There is a pumped hydro storage project in the works in Scotland at Coire Glas that would have 1.5GW power and up to 30GWh capacity (for reference, the peak power consumption in the UK is about 30GW).

https://www.coireglas.com

Tiny

I worked on Dinorwig pumped storage power station (near Snowdon in North Wales) - that can store about 9.1GWh - over 46 times as much energy.

And even Dinorwig (incredible engineering feat that it is) isn't that big in terms of national demand - 9 hours worth of one nuclear reactor output, capable of running itself dry in about 4hours if you really open the taps. If the wind stops blowing for a prolonged period (e.g. one of those pesky high-pressure winter systems that leave us with dark, flat weather for a week), storing energy for a week's worth of demand is simply unfeasible. The storage required in itself is monumental, before we consider the surplus generating capacity required to fill it.

196MWh is a nice amount for frequency management and smoothing renewable inputs. It could even take the edge off the evening demand peak. It might also have prevented [1]the 2020 blackout , which was predominantly down to a cascade of perfectly good generators isolating themselves due to frequency excursion (given this is in Hull, where the Hornsea field supplies power into, I suspect the 2020 incident is actually a key reason its being built!).

But it's not storing meaningful amounts of energy for actually filling demand.

And somewhere along the line it needs charging up first. Storage is no replacement for generating capacity. Which means if we "go green" and spurn nuclear, we'll need to keep a fleet of gas turbines around the country to fill in for mid-winter.

[1] https://www.theregister.com/2020/01/03/thameslink_trains_software_design_error_2019_lightning_strike/

One of the things that the big rotating machinery is good at is frequency maintenance. My understanding is that the inverters on wind farms are network-synced, and so can't drive the frequency.

So, can the monstrously-huge inverter on this lot do that?

One of the things that the big rotating machinery is good at is frequency maintenance. My understanding is that the inverters on wind farms are network-synced, and so can't drive the frequency.

So, can the monstrously-huge inverter on this lot do that?

Yes, it's [1]one of the roles they tout on the website . Renewable smoothing; Voltage & Frequency Regulation & Demand Support. The batteries can cut in on extremely short notice (milliseconds) to avoid frequency excursion.

Stability was one of the key reasons that Tesla built the Hornsdale Power Reserve in Australia back in 2017/18. Wind and solar were causing instability and brownouts as the grid didn't have enough inertia. If I recall, the Hornsdale battery ended up paying for itself in something like 18months through stability controls - more so than actual storage of surplus energy.

The new 450MWh Victoria Big Battery (also Tesla Megapacks) is likewise being built as much for stability and inertia as it is for storage.

[1] https://www.tesla.com/en_gb/megapack

If you install a backup system for a datacentre, I would assume you specify and install a system that suitably sized for the datacentre in question, the aim being to keep it running for as long as you need to until the grid capacity is restored. Unless the power requirement of a datacentre has reduced, I am missing the "excess capacity" bit.

If you can use "excess" battery capacity I assume the facility is too big in the first place. If you use any capacity that is part of the spec then your own resilience drops as you've offloaded some capacity to the grid.

I can only see this as deliberate over specifying a system - buy electricity at £x and suddenly "find" capacity which you "generously" donate to the grid in return for a generator tariff at £X+n when required.

In fact, if your batteries are already charged, and you simply do a AC-DC-AC conversion, rerouting the equivalent of your battery capacity back into the grid, does that qualify for generator tariff? Do you even need to demonstrate the DC conversion? I'm getting too cynical in my old age but we're talking about Bezos business models to make as much wonga as possible ...

"buy electricity at £x and suddenly "find" capacity which you "generously" donate to the grid in return for a generator tariff at £X+n when required."

formula is wrong at the end they'll be wanting £X*n

where n is always 2+p

where p stands for how much piss can we take.

When homes are preparing the Sunday roast on a bitter winter day with the heating full on, they are using roughly 10KW. So 196 MWh of energy for 300,000 homes will last roughly 40 minutes. I am unimpressed.

So you're assuming that none of them have a fireplace where they can burn wood to heat the house instead of electricity ?

Out of 300,000 homes I'd suppose that there is a portion of them that have one. That would lower their electricity consumption for heating drastically.

> When homes are preparing the Sunday roast on a bitter winter day with the heating full on

Gas heating and cooking here. I'm pretty sure I don't use 10kWe while cooking dinner.

I always hate it when we're told some fantasy number of homes that power schemes will power. In this case they're saying that 196MWh will last 300,000 homes for two hours but that allows each home less than 1/3kW. Is 1/3kW a standard per-home figure in such calculations or do they just make numbers up to suit?

That's enough to keep my CPAP running as I freeze to death.

well considering most electric cars state X miles, but in reality do about 60% to 80% of that unless you drive miss daisy, possibly worse in winter.

Seems to be within piss take standards.

They make the numbers up, of course! Otherwise the real answer (it can replace a single power station for about 6 minutes) sounds less impressive.

One trick is to use the daily consumption and divide by 24. Like we cook and wash through the night too.

Homes use between 8-10kWh/day (Ovo, repeating a figure from BEIS suggests 10.2kWh). So at 10kWh, that's going to be 10/12 kW (about 830W) mean consumption during the daytime. So the actual answer is ~236000 homes for 1 hour. Their figure assumes average consumption is 327W, a daily waking hours consumption of about ~4kWh. Fantasy.

Using rechargeable batteries to supply the National Grid is really not that much less crazy than using disposable batteries to supply the National Grid.

You might have heard of a place in Wales that can store nearly 50 times as much energy, without any of the noxious chemicals .....