Home –  Crescent Dunes
Tag Archives: Crescent Dunes

5 Ways Crescent Dunes Solar Isn’t Ivanpah (or Solana)

Crescent Dunes, near Tonopah, Nevada (photo from SolarReserve)

Crescent Dunes, near Tonopah, Nevada (photo from SolarReserve)

UPDATE: Footnotes have been added to elaborate on a few of these items, most notably the comparative price of energy between Ivanpah and Crescent Dunes.

It uses mirrors and a giant tower, like the Ivanpah Solar Electric Generating System in California, and molten salt like Solana Generating Station in Arizona. But SolarReserve’s Crescent Dunes, now in “full commercial operation” [PDF], is different from any large-scale power plant that came before it in several important ways.

1. Unlike Ivanpah, Crescent Dunes, in Nevada, was built with the ability to store energy and dispatch power when needed.

At Crescent Dunes, giant mirrors focus the sun’s energy at the top of a tower, heating a mixture of sodium and potassium nitrate. This molten salt can be used immediately to superheat water and produce electricity in the manner of any other thermal power plant. Or it can be stored in insulated tanks to drive the thermal-power process during periods of cloudy weather or at night.

At Ivanpah, the “heliostats” focus the sun’s energy atop towers to heat water, which won’t hold the heat for long. That means the vast bulk of Ivanpah’s production comes at the same time the California grid is being fed large and rapidly increasing amounts of power from solar PV plants. Ivanpah can provide a smoother flow of electricity than a PV plant, but its value is still limited by immediate reliance on the sun.

2. Solana can store energy, too, but Crescent Dunes claims an advantage over the Arizona plant.

As SolarReserve CEO Kevin Smith told me last year: “We’re using molten salt directly,” giving Crescent Dunes the ability to drive the temperature of the heat-holding salts 300 degrees higher than at Solana, where long rows of parabolic mirrors are used to first heat a transfer fluid. “They need two or three times the salt we have to get the same amount of heat storage,” Smith said. That leads to a bigger, more expensive footprint – “a whole lot more tanks, pumps and salt.” In sum, Smith said, the multistep nature of the Solana process results in less efficient operation.

3. Unlike Ivanpah, Crescent Dunes doesn’t burn fossil fuels.

In 2014, Ivanpah used 867 million cubic feet (mmcf) of natural gas. It helps jump start the system in the morning, mostly, and to get through some cloudy periods. At a typical gas-fired power plant, that would produce around 85 gigawatt-hours of electricity. Ivanpah produced 420 GWh in 2014 – so you could say natural gas use was equal to about 20 percent of the plant’s output. This is way over the 5 percent allowed by California regulations, but a California Energy Commission spokesman said much of the natural gas Ivanpah uses isn’t held against it.

“(N)atural gas used between the end of daily generation and the start of generation the next day is not considered as contributing to electricity generation and therefore, not included in calculating the percent of non-renewable fuel used at the facility,” the CEC’s Michael Ward said in an email earlier this month.

Ivanpah’s output jumped up to 652 GWh in 2015, so if natural gas use held steady, the plant’s generation-to-gas-use ratio would have improved substantially. But while we won’t know exactly how much gas Ivanpah used in 2015 for a few months, some hints are available: Between August and November in 2015, gas consumption at one of its three units was double what it was in 2014.

4. Crescent Dunes should break out of the gate faster than Ivanpah.

Ivanpah went into full commercial operation in February 2014, not even three and a half years after construction began. Despite being one-third the size of Ivanpah, Crescent Dunes took at least four years to achieve that status.NOTE A One reason Crescent Dunes was slower to start up: SolarReserve saw what happened when Ivanpah’s early production fell dramatically short of ultimate expectations.

“We certainly recognized that Ivanpah got hammered,” Smith told me. “From a management perspective, that led us to want to be more cautious.” That was a year ago, when Crescent Dunes seemed on the precipice of startup, but it wasn’t until today that SolarReserve heralded the plant’s opening.

After the fact, Ivanpah’s operators said they expected all along that it would take up to five years to hit long-term performance targets. But they didn’t make that clear at the plant’s grand opening.NOTE B And, as my reporting has revealed, performance has even fallen short of contractual obligations that anticipated a slow start.

After taking extra time to dial in performance, SolarReserve was confident enough to say today that “Consistent with the rollout plan, the facility will ramp up to its full annual output over the coming year.” The target number: 500 GWh/year beginning in Year 2. Yes, we will be watching.

5. Crescent Dunes electricity is less expensive than Ivanpah electricity.

Crescent Dunes is selling its output to NV Energy for 13.5 cents per kilowatt-hour, rising 1 percent a year during the life of the 25-year power purchase agreement. Ivanpah’s contracts with PG&E and Southern California Edison are confidential, but filings with the Federal Energy Regulatory Commission show that during the high-demand July-September period last year, the utilities paid between 20 and 22 cents per kWh for Ivanpah electricity.NOTE C During the same period, Solana sold electricity to Arizona Public Service at 12.8 cents/kWh.


A. I used the “Break Ground” and “Start Production” information from NREL’s website to determine how long it took to build each project. Another way to measure is from close of financing to the beginning of delivery of electricity under contract. That stretches the difference between Ivanpah and Crescent Dunes. Ivanpah took from April 2011 to February 2014 – about two years and 10 months. Crescent Dunes took from September 2011 to sometime in November 2015 – a period of four years and two months.

B. In all their communications around the plant’s opening as it was happening, Ivanpah’s developers didn’t talk about a long ramp-up. However, as first reported last June, in SEC filings in 2011 BrightSource Energy noted that “initial performance will be less than full design,” then would rise due to “realization of the operator’s learning curve, procedural optimization, and fine-turning of equipment and systems for increased plant performance.” The company went on to say “this ramp-up process may last up to four years.”

C. Did I overstate the Ivanpah-Crescent Dunes energy price gap? It’s worth a closer look, starting at Crescent Dunes. The price NV Energy pays, to be precise, is $134.95 per megawatt-hour in the first year of the contract and, as mentioned, it rises 1 percent each year of the 25 year contract. My math says that takes the price on an escalator to around $173/MWh by 2040.
Because they’re still confidential, the Ivanpah contracts (“contracts” because electricity from two units goes to Pacific Gas & Electric, and from another unit to Southern California Edison) are more difficult to pin down. But we do know from SEC filings that “the Ivanpah PPAs include time-of-day (TOD) pricing.” And a closer look at FERC reports reveals how dramatically that can impact what the utilities actually pay.
In the post, I gave a price range for the third quarter. Here are the average weighted prices to the penny, broken down by unit:
Unit 1: $197.33/MWh
Unit 2: $220.51/MWh
Unit 3: $201.99/MWh
But check out the figures for the fourth quarter, the most recent reporting period.
Unit 1: $134.84/MWh
Unit 2: $117.44/MWh
Unit 3: $137.91/MWh
Huge difference. Clearly when Ivanpah delivers energy has a huge impact on the price of that energy. To get a better sense of how this might work, I dug up a power purchase agreement that PG&E did to buy 48 MW of PV power from the Copper Mountain Solar Facility, a sprawling 458-MW plant in southern Nevada. Because the plant has now been delivering power for more than three years, the contract details are no longer confidential.
That contract, signed way back in 2008 when solar PV was a lot more expensive, sets a delivery price of $139/MWh – but then multiplies that price by a factor depending on the month, day and hour of delivery. For instance, if the energy is delivered from June through September on a weekday between 1 p.m. and 8 p.m., the $139/MWh figure is multiplied by 2.01, jacking the actual price up to $279.39/MWh. But if the energy is delivered on a weekday from March through May between 7 a.m. and noon, or pretty much any time on a weekend or holiday, the price is multiplied by 0.86, yielding an actual price of $119.54/MWh.
We don’t know the base price at Ivanpah, and we don’t know to what degree the Ivanpah deals mirror this one. But clearly the third-quarter prices I used were at the very high end of what Ivanpah electricity costs the utilities that purchase it. When you look at the FERC reports and add up all the numbers and average out all the mysterious up and downs in the price for Ivanpah’s electricity throughout the year, you end up with:
$162.49/MWh, or 16.2 cents/kWh.
Last thought to this ridiculously long footnote: This price comparison is a pretty crude measure of the “value” of these groundbreaking technologies. It doesn’t factor in the presumably higher cost of permitting and building in California, nor does it look at the solar resource each site offers, which can have a significant impact.