Cooking the Books 2: Lazard's Levelized Cost of Energy Estimates for Wind
The tools you need to debunk Lazard's wind cost estimates
Jesse, it’s time to cook these books- Walter White, implausibly
This week, the global financial services firm Lazard released the 17th edition of its Levelized Cost of Energy (LCOE) analysis, which provides estimates about the cost of generating electricity from wind, solar, coal, natural gas, and nuclear power plants.
These estimates, especially the low-end estimates, are widely cited by wind and solar special interest groups to claim renewables are the cheapest forms of energy, but they shouldn’t be because they are chock-full of unreasonable assumptions that cook the books in favor of wind and solar at the expense of reliable power generators.
Today, we will discuss why the true cost of wind is nearly three times higher than Lazard’s low-end cost estimates, 44 percent higher than Lazard’s midpoint, and slightly below the firm’s so-called “high-end” estimates, which you can see in the graph below.
Next week, we’ll take a deep dive into their solar assumptions.
Hidden Costs Ignored
First order of business, it should be noted that a major shortcoming of Lazard’s cost estimates is that they calculate the cost of these technologies in isolation. Therefore, they do not account for the massive hidden costs needed to incorporate wind and solar onto the grid, including additional transmission costs, property taxes, load balancing, and overbuilding and curtailment costs.
In fairness to Lazard, they are upfront about these shortcomings, writing on page eight of their report:
Other factors would also have a potentially significant effect on the results contained herein, but have not been examined in the scope of this current analysis. These additional factors, among others, may include: implementation and interpretation of the full scope of the IRA; economic policy, transmission queue reform, network upgrades and other transmission matters, congestion, curtailment or other integration-related costs; permitting or other development costs, unless otherwise noted; and costs of complying with various environmental regulations (e.g., carbon emissions offsets or emissions control systems).
This analysis is intended to represent a snapshot in time and utilizes a wide, but not exhaustive, sample set of Industry data. As such, we recognize and acknowledge the likelihood of results outside of our ranges. Therefore, this analysis is not a forecasting tool and should not be used as such, given the complexities of our evolving Industry, grid and resource needs [emphasis added].
While Lazard is upfront about the limitations of its analysis in this respect, those who promote their LCOE numbers are not.
We detail these massive hidden costs in our piece How to Destroy the Myth of Cheap Wind and Solar, so we won’t go into more detail about them in this article.
What we will do, however, is show how the assumptions used by Lazard are overly generous for both wind and solar energy sources, even without accounting for the hidden costs that LCOE calculations typically don’t cover.
Lazard’s Latest Wind Cost Estimates
According to the report, new wind facilities in the U.S. have a mid-point cost of $50/MWh. The range of possible outcomes for wind is estimated to be $27/MWh on the low end and $73/MWh on the high end.
It is important to note that Lazard’s mid-point estimate of $50/MWh reflects the average of the high and low LCOE; it is not a capacity-weighted average of the cost of the U.S. wind fleet based on the likely installed capacity. As we demonstrate, below, this downwardly biases the midpoint cost because Lazard’s low-end cost estimates are created using entirely unrealistic assumptions.
What is the LCOE, and How is it Calculated?
Before we dig into some of Lazard’s most egregious assumptions in its cost modeling, we need to discuss what the LCOE is and how it is calculated.
As we wrote in How to Destroy the Myth of Cheap Wind and Solar, the LCOE is an estimate that reflects the cost of generating electricity from different types of power plants on a per-unit-of-energy basis—generally megawatt hours (MWh)—over an assumed lifetime and quantity of electricity generated by the plant.
In this way, LCOE estimates are like calculating the cost of your car on a per-mile driven basis after accounting for expenses like your initial down payment, loan, insurance payments, fuel costs, and maintenance.
The factors that influence the LCOE for power plants are the capital costs incurred for building the facility, financing costs, fuel costs and fuel efficiency, variable operational and maintenance (O&M) costs, such as water consumption or pollution reduction compounds, and fixed O&M costs, such as routine labor and administrative expenses, the number of years the power plant is expected to be in service, and the capacity factor of the plant, which allows us to divide the costs above by the amount of electricity the facility is expected to generate during its useful lifetime.
The easiest way to say this is:
Today, we will focus our discussion about LCOE on capital costs, expected capacity factors, and useful operating lifetimes because these are the main ways we can peek behind the curtain of Lazard’s cost estimates and see which assumptions are realistic and which ones aren’t.
Looking Behind the Curtain
Lazard's assumptions for its high-end and low-end estimates are in the appendix of its report, found on page 36. We have constructed a table below to show the cost of wind using these assumptions for the capital costs, capacity factors, and useful plant lives and compared them to more realistic inputs.
When we did this, we found that a realistic cost of wind is $72.22/MWh, which means the true midpoint cost of wind is nearly three times higher than Lazard’s low-end estimate, 44 percent higher than Lazard’s midpoint cost estimate of $50/MWh and nearly as much as Lazard’s high-end estimate.
For those who prefer graphs to tables, the one below compares the Low-End, midpoint, and High-End costs produced by Lazard to our more realistic cost projections using capital cost data from EIA, 20 years as a lifespan, and the average capacity factor for wind plants in America built in 2014 or later of 42.58 percent.
Capital Costs
Lazard believes the cost of building a wind turbine ranges from $1,300/kW ($1.3 million per MW) to $1,900/kW. This is at odds with the U.S. Energy Information Administration estimates the cost of wind is $2,098/kW, and our friends in the utility have told us the most recent bids they have received for wind projects are also more than $2,000/kW.
This means that Lazard’s high-end estimates are still too low to accurately encapsulate the numbers we’re seeing on the ground, and the low-end estimates are 38 percent lower than reality. In other words, Lazard uses a range of capital costs that doesn’t include the average cost of a wind turbine as reported by EIA.
Capacity Factors
This is where Lazard’s decision to take the midpoint of its two ranges of possibilities really comes into play. Lazard assumed a low-end capacity factor of 30 percent, which is actually fairly conservative, but its high-end capacity factor is entirely unreasonable.
The graphic below from Lawrence Berkeley Labs (LBL) shows a map of all the wind turbines by performance in the United States, with capacity factors ranging from 0.2 percent (yes, you read that right) to 62.1 percent.
Lazard’s low-end wind cost estimate relies on an assumed 55 percent capacity factor, which causes the costs to be divided by more megawatt hours and drives down the LCOE. However, the LBL data show that almost none of the wind turbines operating in the nation actually produced at this level in 2022.
In fact, just 474 MW of capacity operated above 55 percent, all of them located in some of the windiest areas of the country. For context, U.S. EIA data show that as of 2022, the country had 141,402 MW of installed wind capacity, meaning 0.33 percent of the nation’s wind fleet was operating at Lazard’s assumed 55 percent.
It gets worse. LCOE estimates assume that a wind facility will perform at its stated capacity factor for its entire operating lifetime. This is problematic because prior to 2022, which was an especially windy year, the LBL data did not show any wind facilities operating at 55 percent capacity factors.
For our more realistic assumptions, we used a value of 42.58 percent because it was the capacity-weighted average performance of wind turbines installed from 2014 to 2021. We chose this time frame because the LBL data show that newer wind facilities are operating at a lower capacity factor than those installed from 2013 to 2018, and the 35.9 percent average capacity factor for the whole wind fleet includes older turbines that are not reflective of newer builds.
If we used the U.S. fleet-wide capacity factor instead of our more generous assumption, the realistic LCOE of wind would jump to $85/MWh.
Useful Plant Lifetimes
Lazard’s estimated 30-year lifetime also downwardly biased the cost of wind turbines because it gives them a longer amount of time to generate electricity than we see in the real world, thus allowing them to divide the upfront costs by more MWh.
The 30-year lifetime is especially egregious because Lazard’s estimates assume new natural gas plants will only last for 20 years. If you think a wind turbine will last 50 percent longer than a gas plant, we have a beautiful bridge to sell you.
For our realistic assumptions, we used the National Renewable Energy Laboratory’s estimate of a 20-year useful lifetime for wind turbines even though we are seeing a growing trend where wind turbines are being repowered after just ten years of service because it allows the facility to requalify for generous federal tax subsidies.
As we noted in our piece The Death of a Wind Farm, it comes as no surprise, then, that data from the U.S. Department of Energy shows wind facilities undergoing repowering ranged in age from nine to 16 years old, with the median age being ten years.
Conclusion
Lazard’s $50/MWh mid-point LCOE estimate for wind turbines is predicated on using entirely unrealistic assumptions for its low-end cost estimates and averaging them with their more reasonable, but not entirely reasonable, high-end estimates, without respect for the fact wind turbines likely won’t last 30 years and only 0.33 percent of the existing wind fleet operated at its low-end capacity factor assumptions.
In other words, Lazard cooked the books to get a bogus number that the pro-wind crowd could champion to influence public policy, even though they have little basis in reality.
Lazard’s Levelized Cost of Energy analysis. Could you tell?
If you begin reading government documents, your standards of brevity must change.
The incident started when a surge arrester failed at a combined cycle power plant. This caused a circuit breaker to trip, and the 192 MW combined cycle power plant went offline. The disturbance this caused on the grid (mainly a frequency drop) cleared within a couple of electrical cycles. However, many generators that used IBRs tripped due to the disturbance. Over 1100 MW of solar and wind (mostly solar) went offline.
Rodeo Clowns by
The world harvests approximately 14 billion bushels of soybeans per year using roughly 18% of global cropland, and each bushel can deliver 11 pounds of soybean oil. Assuming it requires 8.5 pounds of soybean oil to produce a gallon of renewable diesel and that there are 42 gallons in a barrel of the stuff, we arrive at a maximum theoretical number of 1.2 million bpd from the crop, or roughly 4% of current global diesel demand. Certainly not nothing, but the hype-to-impact ratio is unmoored from reality.
Thank you for this clear analysis. People are asking me what I think of the Lazard data, and I am glad to have this to reference.
And thank you for the reference to my post about IBRs on the grid. And the quotes from it!
Your assumptions for capacity factor are unrealistic. Great Plain's cap factors on the best sites can be 50% on average for the year. But DTE's newest wind farm in MI has an annual cap factor of 25%. But MI govt has committed to 100% clean energy. This means MI must have enough wind farms to power the grid during the worst month of the year. In July, MI wind's cap factor drops to 10%. This means MI will need 10 wind farms to replace 1 coal plant with the same nameplate capacity. But even 10 wind farms will come up short during extended wind droughts. You can't say "on average, the grid will be powered". Wind and solar just don't work in many places even if you ignore the outrageous costs.