UPDATE: About 10 days after original publication, NRG got back to me with corrected and previously missing data, necessitating minor changes throughout this piece. The upshot: Gas usage was up 62 percent, not 59 percent, in Year 2, and emissions were 229 pounds CO2-equivalent per MWh in Year 2, not 225.
Natural gas consumption at Ivanpah, the controversial power plant in the California desert, increased 62 percent in its second full year of operation, and the plant emitted the equivalent of about 229 pounds of carbon dioxide for each megawatt hour of electricity it generated.
How’s that compare?
It’s about one-fifth the emissions of the average U.S. power plant and a bit over one-quarter the emissions of a recently built combined cycle natural gas-fired power plant in California.
But of course Ivanpah is considered a renewable energy source by the state, like utility-scale solar PV, which according to the National Renewable Energy Laboratory puts out 18 to 22 pounds of CO2 per megawatt hour of energy produced. In other words, CO2 emissions at Ivanpah are at least 10 times those of solar PV.*
Ivanpah used 1,251 million cubic feet (mmcf) of gas in the 12 months ended January 31, a figure based on preliminary data obtained from the U.S. Energy Information Administration.
The $2.2 billion Ivanpah Solar Electric Generating System, with tens of thousands of reflecting mirrors and giant “power towers” that heat water to generate thermal power, was fully online by February 1, 2014. It generated 430,488 MWh of electricity in its first 12 months of operation, according to the EIA, while consuming 773 mmcf of natural gas. That slow start left the plant in danger of defaulting on its PPAs with the utility PG&E (see: “Ivanpah Not Out of the Woods Yet“). In Year 2, the plant produced 655,926 MWh of electricity, a 52 percent increase.
Under its original licensing, Ivanpah’s three units were permitted to burn a total of 984 mmcf of natural gas annually. But a few months after the plant began commercial operations, Solar Partners – consisting of majority owner NRG, along with Google and technology developer BrightSource Energy – asked and then received approval from the California Energy Commission to use as much as 1,575 mmcf in a year.
So in Year 2, Ivanpah ended up using about 25 percent more gas than originally planned, but about 22 percent less than the revised limit.
The overall limit isn’t the plant’s only consideration when it comes to natural gas, however, not with PG&E and Southern California Edison counting Ivanpah electrical output toward their obligations under California’s renewable portfolio standard. To qualify as a renewable energy source, state law limits the amount of “non-renewable energy that may be included with renewable energy at no more than two percent of total fuel use.” But the law allows wiggle room up to 5 percent in special circumstances, an allowance granted to Ivanpah.
Is Ivanpah using less than 5 percent natural gas?
Depends on how you look at it.
Ivanpah relies on gas in “night preservation” and “auxiliary” boilers. As BrightSource has explained, “night preservation boilers are very small boilers used overnight to maintain seals and preserve heat.” As for the auxiliary boilers, again quoting BrightSource, they are used for:
BrightSource says “this type of natural gas does not produce electricity. In fact, it increases the amount of electricity produced through the sun.”
The gas used at Ivanpah in Year 2 would have produced 180,000 MWh at a modern natural gas-fired power plant like the Lodi Energy Center, opened in late 2012. Although that’s equivalent to 27 percent of Ivanpah’s total Year 2 electricity generation, regulators say Ivanpah has not been using more than 5 percent non-renewable fuel.
That’s because the regulators don’t count most of the fuel used for the purposes listed by BrightSource. If the gas wasn’t burned between the start of generation and the end of generation each day, it’s like it didn’t happen.
This allows NRG to say that less than 5 percent of the electricity generated at Ivanpah resulted from burning natural gas. (Under this calculation, in Year 2, 4.3 percent of electricity generated at Unit 1 came from burning gas; 4.1 percent at Unit 2; and 4 percent at Unit 3. Overall, the figure was 4.1 percent.)
But of course, the gas used for all the other purposes at Ivanpah was still burned, and presumably because it had to be in order for the plant to function. And the emissions were still emitted.
As Dan Danelski revealed last year in the Riverside Press-Enterprise, Ivanpah’s emissions are hefty enough to qualify it as a polluter under California’s cap-and-trade program.** Based on its natural gas consumption, emissions in Year 2 of the plant’s operation were about 68,097 metric tons. That’s 150,128,000 pounds – thus the 229 pounds per megawatt hour figure.
The Lodi gas-fired power plant pumps out about 837 lb/MWh.
It’ll be interesting to see what happens to Ivanpah’s gas consumption in Year 3. The most notable trend evident in Year 2 was a big increase in use at the end of the year – in the three month period from November 2015 through this January, consumption was up 142 percent.
*NREL estimates full life-cycle emissions for solar PV at 88 lb/MWh, with 21 to 26 percent attributed to operational processes, including power generation and system operations and maintenance. See this PDF.
**Danelski reported on 2014 calendar year emissions. But the three Ivanpah units began delivering electricity at various points in January that year, and NRG and PG&E are using February 1, 2014, as a starting point in measuring performance of the two units under contract to PG&E. This analysis follows suit for the entire plant.
In “forbearance agreements” recently approved by California regulators, Pacific Gas & Electric has given the owners of the Ivanpah solar power tower plant up to an extra year to reach energy production levels called for in contracts for two of the plant’s three units. Based on how those units have done so far and what will be required down the road, achieving those minimums won’t be easy.
New data from the U.S. Energy Information Administration indicate that for the first 24-month measuring period called for under the PG&E power purchase agreements, Unit 1 fell 61,692 megawatt hours shy of the 425,600 MWh production guarantee apparently set out in the PPAs. The slightly more powerful Unit 3’s production gap was even greater, 103,544 MWh, toward a goal of 469,840 MWh.
That said, in their second full year of operation, which officially ended at the close of January this year, both units performed much better than in their first year, with Unit 1 production up 29 percent and Unit 3 up 34 percent.
Here’s the issue, though: Under the PPAs, the bar rises. This snippet, from a BrightSource Energy filing several years ago, is what we know about the confidential PPAs (Solar Partners II is Unit 1 and Solar Partners VIII is Unit 3):
So instead of hitting 140 percent of “contract quantity” in a 24-month period (or 70 percent per year on average), Solar Partners now apparently needs to ratchet production up to 160 percent (or 80 percent per year on average). California Public Utility Commission documents related to the recently approved forbearance agreements note that this is a rolling 24-month measurement period, so every month from now on, for the life of the 25-year contracts, the units must hit 160 percent of the annual contract quantity for the preceding 24 months.
Those forbearance agreements push that requirement out six months, at least, and more likely for a year — so it’ll be starting next February, then going forward, when this new target will have to be met.
It will require major increases in production.
Take Unit 1. Its 24-month target — that is, the minimum it needs to produce to avoid being in default — come February 2017 will be 486,400 MWh. With 213,126 MWh in the 12 months ended January 31 this year, that means the unit will need to generate 273,274 MWh over the following 12 months – an increase of 28 percent. That’s as big an increase as it had this past year, but off a larger base.
The challenge for Unit 3 is even steeper. It will need to increase production from 220,595 MWh to 316,365 MWh to reach a target of 536,960 MWh. That’s a 43 percent increase, a significant step up from the 34 percent increase the unit managed from Year 1 to Year 2, and again it’s coming off a larger base.
Four big developments reflected in this chart: the closure of the San Onofre Nuclear Generating Station, the impact of the drought, and the rise of both solar and wind.
Output at San Onofre bounced around in the ten years before it shut down in 2012 – it hit a high of 18,399,596 megawatt hours in 2005 and then sunk to a low the next year, at 13,567,626 MWh. The average over the period was 16,332,433 MWh, about 8 percent of California’s in-state energy generation. So you can see what an emissions-free energy loss the plant’s closure was for the state. But it’s gone, gone, gone, and nothing’s gonna bring it back.
Hydroelectric is a different story, its decline a product purely of the drought. In the decade before the drought began, the state on average got nearly 17 percent of its in-state generation from hydro. In 2015, hydro’s share was 7 percent, so you could say the drought cost California 10 percent of its clean energy generation.
Solar and wind have made up for some of these losses. Wind grew substantially up to 2013 and has been stable at around 6 percent of generation since then (despite the “wind drought” of the first half of 2015), while solar, still on the move, climbed to about 10 percent of generation in 2015, from just 0.2 percent a decade before.
This year, 2016, should see a pretty hefty jump in clean-energy generation in California as wind ticks up slightly, solar continues to surge and, with a decent Sierra snowpack in place, hydropower recovers at least somewhat.
I finally caught up this morning with David Roberts’ excellent post on Vox about big solar, and noticed that in his discussion of capacity factor he mentions inverter load, saying “inverter load is important, though I won’t bore you by explaining it.”
But I will!
A good place to explore inverter load is the Solar Star Projects, a sprawl of ground-mounted solar panels straddling the LA-Kern County line in the Antelope Valley, on the edge of the Mojave, and which Dave mentions.
Technically speaking, it’s called the “Solar Star Projects” because there’s Solar Star 1 and Solar Star 2. But it’s correctly treated as one plant because 1 and 2 are conjoined, had the same builder, use the same SunPower systems, have the same power buyer (Southern California Edison) and charge the same price for their energy.
Dave labeled it the world’s largest solar plant at 579 megawatts. There is a plant in India that is said to come in around 600 MW, but I’ve never been able to get firm data on it and I’m not sure how much is actually operational, so I’m comfortable giving Solar Star the crown.
Turns out, too, that Solar Star is a little bigger than even the company that owns it says. According to the Western Renewable Energy Generation Information System database and filings by the builder, the plant is actually 586 MW. This is confirmed in a recent Fitch’s Ratings release on the plant, which noted: “The completed project’s capacity totals 586 MW of capacity, providing an additional 7 MW of capacity compared to design specifications.”
But enough of that – you came here to read about inverter load.
So as you probably know, inverters are the electronics that convert direct current (DC) output into alternating current (AC) before it goes onto the grid. This results in some loss in power, but systems are built with more DC (photovoltaics) power capacity than AC (inverter) power capacity. The ratio of PV to inverter is called the inverter load ratio, or ILR.
In its most recent report on utility solar trends [PDF], the Lawrence Berkeley Laboratory last fall noted that ILRs were climbing. That is, developers were installing more PV capacity per inverter capacity than they used to. Why? Well, one big reason is that, as Dave noted, PV has fallen in price so dramatically. A few more panels come cheap. Why not?
This doesn’t help the plant produce more peak power – the plant can’t produce more than its inverter capacity – but it does help it produce full power for more of the day. Instead of a production curve that peaks at somewhere between 11 a.m. and 1 p.m., you can get steady peak or near-peak production for several hours. This boosts capacity factor and can pay off for the electricity seller, as LBNL explains:
… with some utilities (particularly in California) offering time-varying PPA prices that favor generation during certain daylight hours, including late afternoon, many developers have found it economically advantageous to oversize the DC array relative to the AC capacity rating of the inverters. As this happens, the inverters operate closer to (or at) full capacity for a greater percentage of the day, which – like tracking – boosts the capacity factor, at least in AC terms…. (T)he resulting boost in generation (and revenue) during the shoulder periods of each day outweighs the occasional loss of revenue from peak-period clipping (which may be largely limited to just the high-insolation summer months).
LBNL says that back in 2010, ILRs were around 1.2, on average, and were often much lower than that. Now they’re just shy of 1.3 on average, and never under 1.2. Here’s a chart from the lab:
And, indeed, we can see this at Solar Star, where the PV panels add up to 747.29 MW of installed capacity. Divide this by the 586 MW of AC capacity, and you get a ratio of 1.28.
With this arrangement, Solar Star was able to operate at a remarkably high capacity factor in 2015: 35.2 percent.
The plant generated 1,663,593 megawatt-hours of electricity, as reported by the Energy Information Administration with – this is an important point here – a plant capacity of 540 MW. Why 540 MW and not 586 MW (or even 579)? Because the entire plant wasn’t online for the whole year – construction began in 2013 and big blocks gradually came online until the whole thing was done in late June 2015. Taking into account exactly how much capacity was online ever day of the year in 2015, the average daily capacity was 540 MW.
You might be curious how much Southern California Edison paid for this energy. Well, the PPAs aren’t public, but FERC records show the total for the year to be $157,180,118, or $94.48/MWh or, to put it in the units you see on your electricity bill, about nine and a half cents per kilowatt-hour.
This, however, masks the extreme variability in pricing. As LBNL noted, when the electricity flows to the utility can make a big difference in how much it pays. At Solar Star, Southern California Edison paid as little as $49.99/MWh during offpeak hours in the first quarter of the year. But in the third quarter – July, August and September, when California energy use often peaks – it paid $251.40 for “onpeak” electricity.
I don’t know exactly how these periods are defined in the Solar Star contracts (there’s one for each “unit” of the plant), but if you read the footnotes to my Crescent Dunes post a few weeks ago (and who didn’t), you know that one California solar PPA I did see defined the peakiest peak as weekdays from 1 p.m. to 8 p.m. during July, August and September.
So you can see why solar power developers will take readily add more cheap PV – leading to a higher ILR – in order to boost generation as the afternoon wears on.
One last thought. The overall average price of nearly 10 cents/kWh paid in 2015 emphasizes again how quickly solar prices have fallen. The Solar Star contracts are about five years old. That means that the price of solar has fallen by half or more in five years, with no change in the regulatory structure. Pretty impressive.
Two Oregon teams are among the finalists selected today in the Wave Energy Prize, a U.S. Department of Energy competition to double the energy captured from ocean waves. One is the familiar Salem-based M3 and the other is a new name in the wave game, Portland-based AquaHarmonics.
The teams, both with Oregon State ties, were selected along with seven other teams to construct 1/20th-scale models of their wave energy converters. As finalists, they’ll receive up to $125,000 to ready their devices for testing this summer at the Navy’s giant Carderock MASK Basin wave tank in Maryland. Assuming they meet a cost-efficiency threshold set for the competition, the first-place winner could receive $1.5 million, second place $500,000 and third place $250,000.
The competition began with 92 entries, then was trimmed to 66 teams that submitted technical documents. Twenty teams were next selected as semifinalists and required to perform wave-tank tests of 1/50th-scale models.
Although Oregon State has a wave tank, in order not to confer a “home-field” advantage, AquaHarmonics and M3 were sent off to the University of Michigan to test their devices.
AquaHarmonics consists of Alex Hagmuller and Max Ginsburg, both engineering graduates from Oregon State, according to their Facebook page. Their device is described as a “point absorber with latching/de-clutching control.” Point absorbers bob at the surface to absorb the vertical motion of passing waves.
M3’s design is an adaptation of the APEX device that was tested off Camp Rilea in September 2014 – submerged but not on the ocean floor like APEX. Called NEXUS, it uses the change in pressure caused by passing waves to send air back and forth through a column, spinning a turbine.
The Pacific Northwest, home to much of the wave energy development in the United States, also has a third team among the finalists, Oscilla Power of Seattle. Their floating device is completely different from anything else, taking advantage of an effect called magnetostriction, caused by the constantly changing tension in the device’s tethers, to produce an electrical current.