The $1.5B Mammoth Solar Farm Will Be the Largest in the US | Hacker News Reader

The $1.5B Mammoth Solar Farm Will Be the Largest in the US

interestingengineering.com

40 pointsmardiyah16 comments

How does this compare with nuclear power?

The total power is going to be 1.65 GW, and the total cost $1.5 BN. That would be about $0.9 BN per GW.

Meanwhile Vogtle 3 and 4 will make about 2.3 GW together for a total cost of $25 BN, so a bit more than $11 BN per GW, or about 12 times more.

But that's not quite apples to apples. Solar power has an average capacity factor of about 30%, while nuclear about 90%.

What is then the cost of 1000 GWh annual net output? I picked out 4 more or less random large solar farms in the US from wikipedia [1] (Solar Star, Mount Signal, Springbok and Copper Mountain) and on average a 1 GW (DC current) farm produces about 1800 GWh annual net output. Assuming this new farm is similar to those other 4 large solar farms (which are all fairly new, commissioned between 2015 and 2020), then the cost of 1000 GWh is about $0.5 BN.

Meanwhile, Vogtle 3&4 have about 19000 GWh annual net output, for a nominal power of 2.3 GW, so if we scale to 1 GW, we get about 8.2 GWh per year. We get a cost of about $1.3 BN for 1000 GWh annual net output.

So each GWh produced by a modern solar farm is about 2.6 times cheaper than a GWh produced by an equally modern nuclear power plant.

Can nuclear become cheaper? NuScale claims to be able to deliver nuclear power plants at a cost of about $2.85 BN per GW [2], which is 3.8 times cheaper than Vogtle 3&4. If they are able to deliver, then they should be quite cost competitive with modern solar.

[1] https://en.wikipedia.org/wiki/List_of_photovoltaic_power_sta...

[2] https://www.nuscalepower.com/benefits/cost-competitive

Something people always forget when factoring solar is the cost of having the standby natural gas plant capacity that has to be ready to take over that 1 GW when the sun stops shining. Having a plant sit idle isn't much cheaper than having it running.

When you say compare you seem to mean cost-wise. My answer is that its cost is known, with no large downsides or unknown costs to be dealt with later.

Vogtle 3&4 are American plants. I don't follow the news there closely. The last I read was that the long-term storage facility at Yucca Mountain was cancelled. No no replacement or alternative in the reports I read. Did I miss anything? Has a plan for how to store the waste been formulated, perhaps even a cost estimated? How are insurance costs to cover liability costs in case of accident estimated?

Waste: on site storage. After the plant decommission, the Department of Energy has the responsibility of finding a long-term solution. The cost for the long term solution was prepaid in the form of the Nuclear Waste Fund [1], which currently has about $45 BN. The Yucca Mountain was indeed de facto cancelled, but in May this year (2021) the Biden administration stated that it will announce a new disposal plan in the coming months [2].

Insurance: currently nuclear power plants are obligated by law to buy insurance. The current scheme covers liabilities up to $13 BN [3].

[1] https://en.wikipedia.org/wiki/Nuclear_Waste_Policy_Act#Nucle...

[2] https://www.wsj.com/articles/escape-from-yucca-mountain-bide...

[3] https://www.nrc.gov/reading-rm/doc-collections/fact-sheets/n...

> 13,000 acres (5,261 hectares)

this annoys me.. unless by some 1/1000 chance the thing's area is an exact multiple of 1,000 acres, the first number is rounded to the nearest 1000 acres. If the first number is rounded, the second number should be rounded to the nearest ~500 hectares.*

Otherwise you could do something especially silly like saying '13 000 acres (52 609 180 square metres)'

* I'm not exactly sure what the protocol is for rounding between different systems - since 1000 acres is ~404 hectares, I'm not sure that 500 would be the correct rounding value to use for hectares)

Get the exact figures, and then round them in the same direction to get the same number of significant digits (others being set to zero) ?

The National Academies of Sciences just wrapped up a live stream with Elon Musk. Surprising to me, it's his opinion that we don't need to develop fusion reactors because we already have a fusion reactor in the sky. He said there is a clear path to capturing that energy with solar panels and batteries to deliver it at the desired rate https://www.youtube.com/watch?v=rLydXZOo4eA

The opinion of elon musk is intimately tied with his ownership of battery production business (that is tesla - it's more a battery company now than ever before!).

i don't see the price of batteries decreasing to such an extend that it's cheaper to solar capture and store with batteries (of the lithium-ion kind at least), than to produce electricity with nuclear.

And i would say we need a mix of everything, since they all make different trade offs.

His comment was about fusion. He said he believes we could eventually make it work, but spending the money wasn't necessary. His battery business is mostly for cars, which is unrelated to the source of the electricity.

His large scale battery plants suck up a huge number of batteries internally. It’s not all about the cars, it’s about the ecosystem around the cars.

It is cheaper today already. Battery backed solar already has a LCOE than nuclear.

Can anyone tell me what sort of environmental impact studies are done before a project like this breaks ground and where (if) I can read them?

At least where I live, you can get the papers from the same local authority that authorises building a new road or building, and perhaps the authority that handles zoning regulations if the land needs to be reclassified from e.g. "forest" or "agricultural use".

These things don't involve much new traffic to/from, don't pollute, hardly obstruct anyone's field of view, but may involve felling trees etc and building power lines. Environmental study rules are of the form "if x then study type y is required", and these have few xs but not none.

TLDR: 13000 acres, 2.85 million solar panels, eventual output 1.65GW, supposedly by 2024. Will start running at lower power earlier.

I don't know why putting them all in one place like that is preferable to decentralization.

Control. It's a business model. just like a power plant, CH4 or coal fired, or nuclear. You build it, then charge for it. Economy of scale should make it cheaper per watt. Also, different customers. decentralised is still a good idea, but not enough to power the future.

Big companies like big plants for at least organisational reasons. Fewer locations to care about, fewer local departments so simpler HR and management.

Reply on news.ycombinator.com