33 Comments

I took a back of the envelope crack on what it would take to power a 300MW data center with solar and batteries. All the most positive assumptions. The facility would cover 25 square miles and cost $3.8 billion. Of course there would still likely be some days when you’d need gas backup.

Expand full comment

How many hours of storage did you use? And that is a lot of land for one large data center.

Expand full comment

Excellent analysis of the Talen/Amazon deal.

Expand full comment

Thank you!

Expand full comment

As someone with keen interest and no experience in the area, it is extremely disheartening to continue to see reports created that claim things like solar plus batteries will be the cheapest form of electricity production for a grid that are plainly incorrect. and yet, other than various substacks, like this and Robert Bryce or Energy Bad Boys, nobody is ever called to account about their knowing misinformation. as a financial markets guy, my analogy is that the theme will be pushed until there is nobody left to buy the story, at which point it will collapse and there will be dramatic damage. not a pretty prospect

Expand full comment

Yes, I think you’re correct; it’ll keep getting pushed as a marketing theme to keep up what is likely a long term bubble. Eventually reality will set it when the subsidies run out

Expand full comment

I did it daily. A 300 MW center would use 7200 MWH /day. 15%-30% would come directly from solar, so 5400 mwh of storage would be necessary.

Nice article, btw… forgot to say that.

Expand full comment

Thank you! Yes, you’re correct that is a generous assumption in favor of solar/ batteries. Especially since it’s cloudy some days. I don’t think most people appreciate what 5.4GWh of battery storage means. And that’s something no data center would want you to afford. Even if the battery price drops to $0, hooking up all that capacity is still expensive.

Expand full comment

I also assumed the batteries would be charged by the solar, not from the grid, which explains the huge solar field…. Even on a cloudy day the solar field would need to produce 7200 MWH. Probably an unreasonable assumption given how the grid works now, but if we are going to 100% solar and wind somebody is going to have to figure out how to charge all those batteries every day.

Expand full comment

Agreed that “ if we are going to 100% solar and wind somebody is going to have to figure out how to charge all those batteries every day.” That’s no small feat.

Expand full comment

How many square miles would be needed for 5.4 GWH of batteries, and as a very minor but important aside, how many thousands of gallons of water would be needed to extinguish a fire. I don't know zip about batteries, but is the probability of fire/catastrophic failure increase with size or connectivity? Curious!

Expand full comment

It obviously depends on the type of battery. But Moss Landing in California is a 3 GWh battery system co located at a power plant.

In the company’s investor presentation, they have pictures of the project:

https://filecache.investorroom.com/mr5ir_vistracorp_ir/315/2024_Q1_Results_Presentation_vFINAL.pdf

https://investor.vistracorp.com/2023-08-01-Vistra-Completes-Milestone-Expansion-of-Flagship-California-Energy-Storage-System

Expand full comment

As a complete aside, in looking at the photo of the container-sized unit in the Utility Dive article, it struck me how vulnerable the unit is to air strike. Back in the day, you would design containment for a nuke for an EF-5 tornado, or an airstrike by a 747. Given that batteries are "inside the fence," how would Homeland Security, FERC and (by default, EPA) regulate design for acts of terrorism or Mother Nature herself? I'm sure you've seen the folded over wind towers near Greenfield, Iowa, from last week's big wind. I doubt even a tornado could pick it up and move it any appreciable distance, but tornados can generate some piercing events such as a fence post or metal road sign.

Granted, these are all "nits", but if we are to base our livelihood on the availablity of storage on-demand, we need to set high standards for design and operation.

My two cents, adjusted for inflation. Thanks for the information, and thanks for your patience.

Expand full comment

This is one of the best parts of Substack: interacting with different people with all sorts of professional experiences in the energy business.

You do raise a good point: yes, battery storage centers are certainly vulnerable to storm damage. And tornados are a real threat to any power system in the Midwest.

If you don’t mind me asking, what kind of hydro? And I’m sure you’re familiar too with hydro pumped storage.

Expand full comment

Hi Barry. There has never been a battery installation anywhere near that big. The batteries are really a small part of the land use. They are modular, about the size of shipping containers and they stack. Not unusual to see them 4 high. With a battery of this size, fire is something you really don’t want to contemplate. It would burn for years.

Expand full comment

Thank you. So, going back to your original problem, assuming 5400 MWH are needed, you'll need 1350 MWs of storage. Right? At 1000 feet per MW (per PEs response earlier), that would add about 30 acres for storage. Also, figured 1 MW storage to last 4 hours.

The math then becomes: (Solar capacity to meet demand, 20 hours per day) + (Solar capacity to recharge batteries) +Storage = Demand. What is the efficiency rate in recharge?

Thank you for your patience.

Expand full comment

No problem!

So yeah, you might need need 300-350 MW of discharge capacity for a 300MW data center. But maybe 5.4 GWhrs of storage could give you a margin of error for daily use, but it won’t be nearly enough to cover seasonal variations in solar.

Round trip loss between 20% of electricity in the charge discharge

And as I’m sure you know, you can’t drain and fully charge the batteries quickly each day.

https://www.powermag.com/dont-neglect-round-trip-efficiency-and-cost-of-charging-when-considering-levelized-cost-of-storage/

Expand full comment

Grid batteries are about 70% - 80% efficient, mostly due to the giant cooling systems necessary to keep them from catching fire under discharge. I didn’t capture that in my back of the envelope analysis as it was ridiculous enough already, but you actually need 30% more batteries than you think you do.

Expand full comment

This is just the latest is a series of announcements of new Data Centers in Columbus Ohio. This had lead AEP to request permission to require a financial commitment to the new load before they agree to serve it.

https://www.nbc4i.com/news/local-news/licking-county/microsoft-confirms-building-data-centers-in-licking-county-topping-700-acres/

https://www.utilitydive.com/news/aep-ohio-data-center-crypto-rates-puc/716150/#:~:text=There%20is%20about%20600%20MW,by%20the%20region's%20transmission%20system.

Expand full comment

‘Under the proposal, data centers would be required to make a 10-year commitment to pay for a minimum of 90% of the energy they say they need each month – even if they use less.’

https://www.aepohio.com/company/news/view?releaseID=9539

Expand full comment

I suspect the negotiated amount will be less than 90% , but good on AEP for driving that 90% stake to start the process.

Expand full comment

Thanks for sharing the links: they’re good reads.

Yes, AEP is probably smart to have them commit before breaking ground on infrastructure. Adding nearly 4 GW of power infrastructure is a massive chunk of Ohios electricity generation.

Expand full comment