Hacker News Reader

Comment by philipkglass | Hacker News Reader

According to this article, the project cost $2.13 billion for 3.5 gigawatts-peak of installed capacity:

https://www.pv-magazine.com/2024/06/06/worlds-largest-solar-...

It's expected to generate 6.09 TWh per year, for a capacity factor of 19.8%. That's equivalent to $3.07 per real annualized watt. For rough comparison purposes, last year's contract for China to build a 1.2 GW nuclear reactor in Pakistan was for $4.8 billion [1] (4 dollars per watt, $4.44 per real annualized watt assuming a 90% capacity factor [2]).

[1] https://www.voanews.com/a/pakistan-signs-4-8-billion-nuclear...

[2] https://pris.iaea.org/PRIS/CountryStatistics/ReactorDetails....

View on HN

That looks pretty good for nuclear! I assume the cost of batteries to shift a annualized watt from day to night is more that $1.50

Don't know how it is for Chinese projects, but in the west you should not compare actual cost of A with projected cost of B, because with such projects we repeatedly see that the actual cost can be a multiple of the projected costs, especially for nuclear power.

For Chinese projects we can assume that the price tag also doesn't match the actual costs (similar to French nuclear projects) due to heavy government subsidies.

> For Chinese projects we can assume that the price tag also doesn't match the actual costs (similar to French nuclear projects) due to heavy government subsidies.

Why would China subsidize a $5 billion reactor in Pakistan? That's not their MO with the Belt and Road initiative - China prefers to inflate sticker prices and loan more money to the buyer, building more economic influence.

The project certainly benefits from the economies of scale thanks to all the reactors China builds, but it hardly helps to conflate that with actual subsidies.

Once the central government has decided what they want, China absolutely has a track record of extremely heavy subsidies, both for domestic market and for export. Like you say, it is to get to that sweet 'economies of scale' (or to say it more honestly, pricing others out of the market).

There is one other interesting tidbit about the pricing of energy projects. Renewables have been scrutinized about their price to an unprecendented extent. Before renewables, costs were usually a minor concern in the public debate when it came to build-out of energy infrastructure.

In the nuclear heavy country like France (which did benefit from the so called 'economy of scale' most, in their case it was more the will for nuclear armament), the government decides the prices for households. They had to increase it steeply in the last decade, and a government commission suggested to further increase it by 74% this year. The government decided to only increase it by 10%. I guess they are a bit afraid of the protest-loving population. Nobody cares for costs, if they don't see it.

In Germany meanwhile the subsidies (for renewables mind you, not for all the other forms of energy) actually work in a way that they are added to the price of all electricity. So while German and French prices get closer and closer now, the German household electricity prices reflect the total cost much better than the French prices.

Now we have enough data to not just project costs, but also we know what we actually pay. And contrary to what everyone believed, renewables are actually the cheapest form of energy we can get. And when you look at the fundamental physical process, it even makes sense. No need to dig anything out of the ground, because the source that put all this into the ground actually is accessible on the surface. No need to create a process that biological beings never used for energy. Wind and solar are just technological variations on the theme that evolution put into live and proven for billions of years. It just makes sense to leverage for the bulk of our energy needs until such time when our energy needs overtake what is accessible so easily. That may be a century from now or even farther in the future.

The battery packs in my local hardware store are all close to €1/Wh ($1.09/Wh but I'm rounding more than that anyway), so they'd have to last a mere 265 cycles to be as expensive as $1.50/Wh for every day of a year.

https://www.obi.de/technik/power-stations/c/4850

I don't think that's correct because of battery only makes one round trip a day while the other numbers are for continuous production. To compare units Apples to Apples I believe you'd have to multiply the battery cost by a factor of 24.

You are comparing $4/(watt-day) to $1/(watt-hr).

Hm.

It's certainly possible I'm mixing up things. I will go back to the linked figures and start from scratch.

(Aside: assuming your point, it would be 12, not 24, as the batteries only shift from day to night and not vice versa).

>Aside: assuming your point, it would be 12, not 24, as the batteries only shift from day to night and not vice versa)

I think you are still missing the point. you will use the battery exactly once per day and it provides 1 watt-hour each time. It isnt providing 1 watt-hour every hour of the day, nor is it providing 1 watt-hour every 12 hours. you discharge it once every 24 hours, or 365 times in a year.

Compare this to nuclear, which is providing 1 watt-hour every single hour in a day. (1 watt-day = 24 watt-hours.

My use of "aside" was supposed to indicate that it was not critical to your core point that you re-emphasise here.

Going through the numbers one step at a time and keeping track of the units, I believe the core problem is that the original was a meaningless unit in the first place, taking the total system construction cost and dividing by the power output (but expressing things in the middle in terms of energy produced over the whole year).

(Not that it matters except to help me avoid such mistakes later, but because the numbers which are important are in units of $/Wh not $/W, I just assumed and that assumption led to me doing the thing I did).

So, we have: PV for $3.06/W, nuclear for $4.34/W

But that's not what energy is normally priced in, energy is normally priced in [money]/[energy] rather than [money]/[power], to convert between them I have to account for system lifetimes.

The IAEA says nuclear plants have a lifetime of "between 20 and 40 years"[0], so I'll pick the half way value of 30. PV is generally guaranteed to last at least 20, but expected to last 30; as very few have been around that long and the tech is rapidly changing, I'll be pessimistic for that and pick 20.

With those assumptions, this gives nuclear a cost per unit of energy of $16.5 / MWh (which is extremely cheap, so much so it's outright suspicious!); and PV a cost per unit of energy of $17.49 / MWh (which would also be extremely cheap, bordering suspicious, though by a smaller margin as the normal LCOE is lower anyway).

For comparison, this graph suggests PV is usually about 3 times that and nuclear usually around 10-11 times that: https://en.wikipedia.org/wiki/Levelized_cost_of_electricity#...

Now, with these numbers, you don't really need to ask about the cost of storage, the nuclear option is just directly cheaper than the PV. I assume the surprise here is that maintenance costs are not included in construction costs? But then I'd be going down a completely different rabbit hole trying to dig into how Lazard reached those numbers.

Would be good for everyone if China's actually done what people keep asking for with cheap nuclear power.

[0] https://www.iaea.org/sites/default/files/29402043133.pdf

>My use of "aside" was supposed to indicate that it was not critical to your core point that you re-emphasise here.

Your aside, while not critical, was also no true. The time between sequential discharging of a battery is 24 hours, not 12. The time between charging and discharging is irrelevant, just how many times you do it a year.

I think you are also misinterpreting the parent post. I think they are intentionally providing [money]/[power]. This is a common format when talking about the cost of installed grid capacity.

I should have converted your watt-hour numbers to [money]/[power] and used to for better comparison.

>For comparison, this graph suggests PV is usually about 3 times that and nuclear usually around 10-11 times that:

LCOE is a radically different form of calculation than what you used, and not comparable. It takes into account many other factors, most notably the time value of money.

But let’s not ignore the average down time of 35-40 days per year for refueling nuclear PP.

https://www.statista.com/statistics/185276/us-nuclear-refuel...

What am I missing? 36 down days a year is 90% duty, same as parent post

Is that something that needs to be considered separately, since the operation factor is given in % of hours per year?

Which is why most installations have more than one reactor on site.

That’s terrible for nuclear, and I’m pro-nuclear. That’s the “Chinese reactor made in Pakistan price”

GA famously took 7 years and spent $17B.

That is equally terrible comparison. It is the first new reactor in US in 30 plus years , the experience and expertise is simply not there to do it cheaply and on time .

France , India and some other countries regularly builds reactors and they don’t cost anywhere close to the cost GA had

https://en.m.wikipedia.org/wiki/EPR_(nuclear_reactor)#Flaman...

France, 13.2€ billion, 17 years, still not producing. Started loading fuel this month.

Initial costs were estimated at 3.3bn euros, blew out to 13.2bn.

That is the first next generation EPR reactor, the second one of its kind in Europe and first in France. This is just reinforcing my point the first one is hard to predict the cost and timeline for.

France built 20 P4/P'4 reactors, most were built on time and at reasonable costs.

The last gen N4 had only 4 units, even then the second batch at CIVAUX costed only $4B.1[2] i.e. $1,349/ kWe overnight costs

Vogtle(GA) 3/4 is expected to cost ~ $8,000 /kWe [3] [4] and the EPR is at Flamanville 3(France) is expected to come in at $6,500 kWe [5].

France projects to achieve 40% cost reduction and faster commisioning for EPR designs over next few being built to achieve $ 3800 /kWe overnight costs [5]. The report details on how they this is achieved and gains already seen at Taishan(China) ( partly owned by EDF)

OECD Nuclear Energy Agency reports in the modern era the overnight costs to be between $2,157 kWe [South Korea] to $6,900 kWe [Slovakia] [6]

Reduction in time and cost is normal to expect in any repeated project development. The first one is path finder, next few should benefit from them, however if we do just built only a few then costs or timeline shouldn't be an objective to compare.

---

[2] https://www.power-technology.com/projects/civaux/?cf-view

[3] https://world-nuclear-news.org/Articles/AP1000-remains-attra....

[4] https://ifp.org/nuclear-power-plant-construction-costs/

[5] https://www.sfen.org/wp-content/uploads/2020/04/EN-The-cost-...

[6] https://world-nuclear.org/information-library/economic-aspec...

Moving goal posts? Your assertion was "France and India regularly build cheap reactors quickly".

With nuclear power is always "is expected...", "projections are...", "it's hard...", "the next one would be...". You know what is delivering in time and costs right now? Solar and wind.

I made two consecutive comments in the thread . My point was split between the two comments and i replied was in combination of both. Comments and timestamp are available for you to see

I realize that from this thread point of view it seems I am changing the argument, that’s on me . It is not always the commenter is trying to the move the goal post sometimes things to do happen .

—-

To the actual point, It is not about whether nuclear is best option, I never compared it with any other option. It was not defense of nuclear , it was of project and costs management .

Nuclear’s costs cannot be baselined by costs at GA , every developed country is doing it half the costs or less.

With respect to “projected” gains are already realized in Taishan, yes it is china but that does not make EDF components being built in Europe faster or, state subsidies can make projects cheaper not faster .

France and other countries have proven track record of building plants . Citing one outlier with brand new tech plant whose costs and time have ballooned is not valid argument .

Reject or dislike nuclear for any number of reasons ideological, economic or environmental, but building a single plant after 40 years and saying it is too costly and slow is a dishonest argument .

90% capacity factor is possible but quite generous, France is below 70%. As nuclear penetration goes up it becomes impossible to sustain high capacity because demand itself is variable, so you either need to overbuild nuclear and run nuclear at low capacity factors, or you need to build storage.

I'm very shocked to see that China's utility scale solar project is $600 per kwp. I almost want to say something must be wrong with that number. It's higher than retail rooftop PV costs which is $450 per kwp.

Presumably for rooftop costing you regard the roof itself and the land it sits on as being "free", whereas for a power plant you have to account for everything? Still seems a bit of a stretch for it to make that much difference but... possible?

It's higher than retail rooftop PV costs which is $450 per kwp.

Which country is that in? I don't know of any country where you can get e.g. a 5 kilowatt rooftop system installed for $2250 (except perhaps with special government tax credits/incentives). The costs for this Chinese project include not just the PV modules but the land, racking, inverters, transformers, grid connection, and installation labor.

Any developing country that has cheap labor costs and no tariffs. The retail cost for panels is $205 per kW, inverter $115 per kW, then add another 30-40% for labor, cables and mounts.

Meanwhile Georgia spent $17B and took 7 years for ~2.2GW.

What a joke. Nuclear is great and all, but not when it’s run like this.

That cost and time is less about nuclear and more about this being bespoke one-off for US with little recent experience in building new reactors.

It is like saying designing and building a single new car is costing few millions and taking years when a Corolla is 20,000 and you drive off the plot today .

As you comment the same, I respond the same: the nuclear behemoth that is France is also burning (public) money to build nuclear reactors with worse delays than GA.

So like 10 attack jets... Damn, what a steal.

Reply on news.ycombinator.com