Intermittent vs. Dispatchable

“Nuclear’s obviously not the answer. It’s too expensive.”

“Renewables are awful for the grid. They’re intermittent, and they aren’t priced correctly.”

“But nuclear is dependable, dispatchable, and uses the least amount of land! According to first principles, it should be the cheapest!”

“But renewables have no fuel to ‘buy’ or ‘burn’ other than the free forces of nature. Therefore, the energy return on energy invested is super high! Just look at every report Vestas has put out in the past five years!”

“But those aren’t priced correctly. They’re fundamentally not fulfilling one of the underlying requirements of the grid which is to be dependable! Always on, all the time. How do you do that when the wind isn’t blowing?”



-Overheard in a bar somewhere in Denver earlier in 2022…

Comparison of Multiple Cash Flow Models

The discourse occurring among the energy generation technologies currently is incredible. Nearly all the players seem to be screaming past each other, with pro-energy density folks saying look no further than California and Germany for examples of how renewables destroy electricity prices and renewable advocates claiming electricity prices in Texas have been cheaper than ever thanks to abundant deployment and utilization of wind turbines. Both of them believe to be correct, but have either side actually sat down and done the math? I doubt it.

This post hopes to explore the possibilities of when and how each system could be cheaper and provide a legitimate and believable explanation for either side to stand on. The goal is to examine which system, if deployed entirely by itself, would be cheaper and thus ignore the non-confrontational baboons who are brave enough to mutter the words, “All of the above strategy.”

So who’s right? Are renewables actually cheaper than any other energy generation source? Or would nuclear power be cheaper? If not, why? If it could be, how would it be cheaper?

Before we get started, let’s step through a thought experiment. In the charts below, you’ll see one of the cases outlined as “Wind + batteries.” This is the most extreme case where wind power has to stand on its own to support the grid. Why? Because anything less results in a logical fallacy.

For example: Say you built enough wind power to support 90% of the grid with 10% dispatchable energy generation available for when the wind isn’t blowing. Unless you have enough energy storage installed to make up for the 10% you’re missing, you’ll need to build a secondary energy generation source to have on standby. This is the method currently being utilized by the grid right now.

The result can be demonstrated in the following chart:

Capital Cost and LCOE as a function of % of Wind Installed

Let’s break this down line by line (pun intended):

  • There are two “flat” lines on the chart: capital cost of gas and LCOE of wind.
    • We need a backup generation source, and that generation source needs to have as much power generation capacity as the rest of the system. Therefore, in this example, we assume it to be 300 MW of gas turbine backup at a cost of $1.8B or $45/MWh over 20 years with an 80% capacity factor (grey line)
    • The other flat line is the LCOE of wind (it’s also light blue – like the sky!). According the Lazard report (which we’ll reference several times in this article), the LCOE for wind is $20/MWh, which is super cheap. This doesn’t take into account the percent of utilization wind is on the grid or any backup generation as we’ll see later.
  • The total cost of wind (orange line) scales linearly with the percent installed. That makes sense.
  • The total capital cost (yellow line) also scales linearly with the percent of wind installed. That also makes sense.
    • Notice here: in this example, the total capital spent only gets higher with more wind installed.
  • The NET LCOE (dark blue line) scales linearly as the capital cost increases. Notice it’s not much higher than it would be without any wind installed… but it does increase.
  • Now – for my favorite part – ONE OF THESE THINGS IS NOT LIKE THE OTHER! Notice the green line which is, indeed, non-linear. It sky-rockets to an asymptote as the percentage of wind installed increases. What the hell?

This is the entire problem and why we examine each case below as a “be able to stand on your own two feet” case. It simplifies it. If an asymptote is introduced into the system near the complete utilization of renewable resources, then that has to be considered. The cost of the dispatchable power generation plant doesn’t decrease as the percentage of wind increases, and the LCOE of wind also doesn’t change. As such, the gas resource is underutilized, and the normalized cost of every MWh generated skyrockets.

So what is the answer? Can an intermittent energy resource be cheaper than a dispatchable one? If so, how? The answer is the hated, “It depends.” Given any set of inputs you can form the output to be what you want.

Let’s examine four distinct cases and compare them against each other:

  1. Wind + Storage
  2. US Nuclear
  3. Chinese Nuclear
  4. Microreactor Deployment

Our underlying assumptions:

  • Project lifetime is 100 years starting 5/1/22. Why? Because many large scale nuclear power plants stand a chance to survive for at least that long, and many dispatchable power sources will need to be replaced in that same time span. For this analysis, wind turbines are replaced every 20 years and batteries are replaced every 10 years.
  • Required power demand is 1,000 MW.
  • 14 days of back-up storage capacity is required. This is a reasonable mid-range estimate for
  • Cash flows are cumulative. All charts demonstrated are un-discounted cash flows, but discounted cash flows are also calculated and demonstrated in the attached spreadsheet.
  • Sale price of electricity is constant for cases 1 through 3, but can be higher for the Microreactor case (case 4). Why? Because the market is fundamentally different. They will be utilized to replace diesel generators throughout the world and can demand a different (read: higher) price than utility scale generation.
  • Because it’s referenced frequently by renewable energy advocates, the Lazard report is utilized as the source for the cost inputs both for wind and for nuclear.
  • We assume we need 1/(Capacity Factor) * Power Demand for all power generation sources to meet the base load demand at all times. For example, wind’s capacity factor is assumed to be 33%, so we’ll need 1/33% * 1,000 MW = 3,000 MW of wind installed to meet the 1,000 MW power requirement.

If you disagree with any of the assumptions and would like to modify them to see how the outcomes change (or you’re just a nerd and love crunching numbers), then give it a whirl. This is the spreadsheet with all the work in it:

For our base case scenario of $20/MWh for utility scale pricing and $100/MWh for microreactors (MR):

Base Case: $20/MWh for Utility; $100/MWh for MR

Let’s break this down:

  • The orange line is Case 1: Wind + Storage. You’ll notice it never breaks even and loses money at an astonishing rate over the course of the project. The large steps every 20 years are the replacement of wind turbines. Batteries are assumed to be the mode of storage in this model, so they’ll need to be replaced every ten years. The battery cost is assumed to improve over time and can be adjusted.
    • Total invested over 100 years: $207B
    • Total lost: $193B
  • The grey line is the current US nuclear. It also never makes money, and in fact loses money over time because the sale price of electricity for $20/MWh is below the assumed operating and maintenance cost. This is (obviously) problematic and unsustainable.
    • Total invested: $13B
    • Total lost: $18B
  • The yellow line is the cost of Chinese built nuclear. Notice it both eventually goes cash flow positive and breaks even after about 35 years. While the true cost of China’s operating cost is difficult to discern, their reported capital cost is much, much cheaper than the US’s.
    • Total invested: $3.7B
    • Total earned: $5.9B
  • The blue line is an example of a Microreactor design. The design selected is the Holos modular unit in development by HolosGen. The capital cost is assume based on a report by the US Army, and the operating expense is assumed to be minimal. Note again a different electricity price was utilized for this scenario because it’s a fundamentally different application than the other cases, but it’s interesting to visualize side by side.
    • Total invested: $58B
    • Total earned: $29B

Now consider if you think adding in the cost of storage is unfair for wind. No problem. We can reduce the storage requirement to zero days which results in this:

$20/MWh for Utility; $100/MWh for MR; 0 Days Storage Required

You’ll notice wind originally does better than US nuclear, but all gains are wiped out by replacement turbine costs.

What if the price of electricity were higher? Does that change the outcome? What would the price need to be to breakeven with US Nuclear after 100 years? About $42/MWh:

$42/MWh for Utility; $100/MWh for MR

What does the price of the Microreactor need to be to breakeven at 100 years? About $67/MWh:

In summary:

  • Each case demonstrates a theoretical cash flow model.
  • The inputs can be modified for each case to more closely match a user’s assumptions or real world knowledge.
  • Wind + Storage is outperformed by all of the nuclear scenarios.

I encourage you to inspect the spreadsheet. Challenge the assumptions. Provide feedback on why they’re correct or how they can be improved.

I love to be proven wrong.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top