Massive Solar PV Deal

Sorry for the lack of postings/comments in the past few days. I have been traveling, and just arrived back in the U.S. I have lots of things to catch up on (family time, among the most important) and I have to prepare my presentations for the ASPO conference (I will be presenting on “Biofuels” and on “Tracking Public Data” such as EIA and IEA numbers). I also have a staff meeting the first week of September. My posting frequency in the next 30 days or so is likely to drop down to 1 or 2 a week, and then should go back to normal following ASPO which begins September 21st.

In the meantime, there was a major energy story yesterday that is worth noting, and it generated some numbers for crunching. Pacific Gas & Electric in California has signed an 800 megawatt solar deal with OptiSolar and SunPower. CNET has some of the details:

Massive solar photovoltaic plants are California-bound

SAN FRANCISCO–Pacific Gas & Electric has inked deals with OptiSolar and SunPower to establish 800 megawatts of solar farms in California, which could become the world’s largest set of grid-tied photovoltaic installations.

The new plants would provide 1.65 billion kilowatt hours each year, enough to serve nearly 250,000 homes, according to Jack Keenan, CEO and senior vice president of PG&E.

“This commitment not only moves us forward in meeting our renewable goal, it’s also a significant step forward in the renewable energy sector,” he said. “Utility-scale deployment of PV (photovoltaic) technology may well become cost competitive with other forms of renewable energy generation, such as solar thermal and wind.”

The supply breaks down as follows:

OptiSolar’s 550 megawatts are set to come online fully in 2013, and SunPower’s 250 megawatts should be running by 2012, both in central San Luis Obispo County. Unlike these photovoltaic projects, most large-scale solar farms feature solar thermal systems.

I found a story in the Milwaukee Business Journal that mentioned some of the area requirements:

The 550-megawatt Topaz Solar Farm to be built by OptiSolar, a venture-backed firm based in Hayward, will utilize OptiSolar’s relatively low-cost thin-film photovoltaic panels manufactured in Hayward and Sacramento. The project, which will cover about 9 square miles with solar panels, will be the largest solar PV project in the world, and is reported to cost $1 billion. It is expected to begin power delivery in 2011 and be fully operational in 2013.

How does that area requirement compare with the area I came up with in A Solar Thought Experiment for providing electricity for the entire U.S.? 550 megawatts over 9 square miles is equivalent to 62 megawatts per square mile. My calculations were that it would take 2531 square miles (a square of about 50 miles by 50 miles) to provide 882,125 megawatts (349 megawatts per square mile). If we scale that down to 9 square miles, we would come up with (882,125 megawatts/2531 square miles) * 9 square miles = 3,100 megawatts (versus their reported 550 megawatts). I would then suspect that they are taking the infrastructure, etc. into account, as my calculation was purely surface area of the panels.

However, I did a second calculation in Running the U.S. on Solar Power in which I did consider the total land requirements of a solar thermal plant. Apples and oranges, I know, but I want to know how they stack up against each other. For the Optisolar plant, we find 550 megawatts on 9 square miles. A square mile contains 640 acres, thus we have 0.89 megawatts per acre. In my previous calculation on solar thermal, I found (based on an actual plant capacity and land requirements) that they expected to get 0.147 megawatts per acre.

That suggests one of three things to me. First, it could mean that the supporting infrastructure for a solar thermal plant has a much greater requirement than a solar PV plant. Second, it could mean that I am looking at just PV surface area versus total infrastructure requirements for solar thermal, but that doesn’t seem to mesh with my prior calculation (unless they are usually very low efficiency panels in the PG&E projects). Or, it could be that since I am only looking at peak output, I am not actually getting a total picture since solar thermal has the potential for energy storage, and thus generating power far past peak output. In other words, solar PV has a pretty sharp peak, and solar PV has a much broader peak for the generation of electricity. There is a 4th option – that I have an error somewhere. But I know if I throw this out there, someone will sniff it out.

Maybe some of our knowledgeable solar experts can weigh in?

Finally, the inevitable question always seems to be whether you can invest in the companies involved. According to this article, Optisolar has been raising capital through private investors. SunPower, however, trades on the NASDAQ as SPWR. The PE is currently at 142, so if you are risk averse you are probably going to pass. The other company mentioned in the article, Pacific Gas and Electric, trades on the NYSE as PCG and a more conservative PE of 14.8.

32 thoughts on “Massive Solar PV Deal”

  1. I’ve been wondering which technology (PV or thermal) was going to be the better choice for utility-scale generation. I still haven’t seen a clear answer. To add to your list of possible trade-offs, we should also consider the cost, complexity, maintenance, and life-cycle. In principle, a solar thermal plant could gather energy using much lower-tech materials, which might be an advantage on the factors I’ve mentioned.

    Here’s another question: why can’t a hybrid plant be built that is PV + thermal + NG powered? The plant would take off as much energy as possible using PV, allow the uncaptured energy to heat a solar thermal rig, and use NG as a back-up fuel at night or during bad weather. Impossible?

  2. You might reconcile your conflicting area requirements by considering that PV cells made of x-Si are more than double the efficiency of those made using the amorphous silicon technology of OptiSolar. OptiSolar hasn’t published its efficiency figures but a-Si is less efficient than using polysilicon. Half the effficiency means twice the area is needed to produce the same output. X-Si technology has roughly 90% market share, with a-Si competing with two other thin-film technologies.

    Seldom mentioned in OptiSolar’s announcements is whether it uses high-GWP gases during its a-Si PV manufacturing process. It is typical to use these gases during PV cell (a-Si) manufacture. The global warming potential of these gases are at least 10,000 times more potent than a corresponding mass of CO2 emissions. If used and if emitted, the emissions of high-GWP gases offset the CO2 emission savings of energy from solar compared to energy from fossil-fuel combustion.

  3. Terrific post.
    What I get out of this is that we are not doomed as a species, and in fact, there are times I actually feel proud of my fellow man.
    For 2000-3000 square miles of solar, we could power up the USA? And it seems a technical reality — as this point, it is just money. This is not a fantasy technology. We are doing it already.
    Of course, we don’t need that much. We have nukes, wind, geothermal (some say that could do 10 percent), and even gas and coal plants.
    Okay, and it looks like the EV (GM Volt) is a technical reality, and soon to be commecial reality also.
    No doomster has ever answered this query: What is to prevent the developed world from migrating to EV-based car fleets, and drawing power from a green grid?
    In fact, if oil does go skyhigh, is not this the obvious, price-induced course?
    For me, that seems a better world anyway. Cleaner air, quieter streets. For the AGW crowd, a winner too.
    I just don’t see the downside here. Frankly, bring it on.
    The problem? Oil getting crushed as we speak. Could go to $40 a barrel.

  4. They are getting 25 nameplate watts per meter^2 when panels are available that do over 100. That just tells me that land is cheap and the panels aren’t dense packed. Their expectation of 2kWh per year per nameplate watt is a little optimistic for California central coast but it’s reasonable for a desert installation.

  5. Land is certainly cheap compared to the cost of PV panels. $1 billion/9 square miles works out to $174,000/acre for PV hardware. I don't think they will pay that for California desert acreage.

    I wonder how these installations will be used to meet grid loads? Base load generation requires capacity factor > 0.8. The capacity factor of these combined installations will be (1.65×10^9 kW-hr/yr)/[(800,000 kW)(24 hr/day)(365 day/yr)] = 0.24. This is typical for solar power, but much too low for base load requirements. And the cost will be $1,800/kW, way too high for peak load power. So these will only be intermediate load generating plants?

  6. According to the article I saw, the PV plants are to be installed on “grazing land”. I would guess that one possible reason for the low density is that they are intentionally spacing out the panels so that the grass under the panels gets morning and evening sunlight (so that the land can continue to be used for grazing). PV modules can’t do much with low-angle-of-incident light, but plants are pretty good at taking advantage of it. This is all speculation on my part, though, so YMMV.

    In answer to anonymous’ question:
    Here’s another question: why can’t a hybrid plant be built that is PV + thermal + NG powered? The plant would take off as much energy as possible using PV, allow the uncaptured energy to heat a solar thermal rig, and use NG as a back-up fuel at night or during bad weather. Impossible?

    The problem with this scenario is that the PV panels don’t get very hot (which is good, because heat causes loss of generation efficiency) so the temperatures that could be captured from them aren’t high enough to run a rankine cycle at reasonable efficiency. All the solar thermal-electric plants I have heard of use concentrators of some kind to get working temperatures in the high hundreds of degrees range.

  7. Appears to be a lot of PV activity going on. Following are links to 3 installations here in the Mid Atlantic area. First is a tracking installation of 10.6 MW.,0,2475409.story
    Next is a plan to power a crayon factory.,0,7167520.story
    Final is a story of local utility installing rooftop PV at a pharmaceutical facility.
    The 10.6 MW tracker is .106 MW/Acre. I suspect this is peak output on a sunny day around the summer solstice.

  8. Nobody mentioned cost. The 550 MW solar facility was expected to cost $1 billion. For simplicity, assume that the rate of return is only 5%. So, before contribution to capital, a $1 billion project would have to generate $50 million per year in revenue. Let's be generous and say the facility manages to peak out at 6 hours per day on average. That works out to 15.15 cents/kWh, just to generate the power. That is before amortizing the capital and assuming just a 5% rate of return, and not including transportation & distribution.

    This is surely going to drive up electric rates in California, which are already some of the highest in the nation.

  9. I agree with anonymous-5:12PM that the amorphous silicon modules that OptiSolar uses are half as efficient as the crystalline modules you based your previous calculations on, which accounts for much of the area difference. You calculated Optisolar’s 550MW thin film PV plant will take up 9 square miles or 62 MW/sq.mi.

    Another article says the two plants will cover 12.5 square miles.
    So that implies the SunPower 250MW plant will cover 3.5 square miles (2240 acres), or 72 MW/sq.mi (0.111 MW/acre). Why so low considering the high efficiency of SunPower crystalline silicon modules? It’s because, as the article says, they are mounted on solar trackers to increase the MWH generated per peak MW by about 30%. This means they have to leave more space between modules so that no module will shade any other module throughout the day. Since the land is cheaper than the PV panels they’d rather use more land than more panels to get the same GWH/year.
    Solar thermal also has the tracking/shading issue, so why does the PV at 0.111MW/acres still get less than the solar thermal 0.147MW/acre? I suspect you’re right about the solar thermal having a broader peak, so the peak rated watts for solar thermal might be lower than the peak rated watts of a PV farm that generates the same number of MWH per day.

  10. King-

    What are the costs of other plants? Nukes? Aren’t you maying 15 cents in Texas right now? I read about one Texan who had a $4,000 monthly electric bill, for a ordinary house.
    I also wonder if economies of scale will start kicking in — more production, and lower costs. I prefer pebble bed nukes, but they are not gettig buit….

  11. King, time for a new calculator.

    550 MW * 6 hours per * 365 day/yr = 1.2 billion kWh

    $50m/1.2e9 kWh = 4.15 cents/kWh

  12. doggy – you are right, I messed up the calculation, I shouldn’t have been multi-tasking while posting. If they can do it for $1 billion that might not be such a bad deal. Anything less than $.10/kWh is pretty good.

    I should have used my home solar worksheet and scaled it up instead of a back of the envelope calculation.

  13. I wonder how these installations will be used to meet grid loads?

    Simple…they will be run full-out whenever the sun is shining on them. I think this attempt to categorize solar power into pigeon-holes created based on conventionally-fueled power plants is misleading.

    Like base-load plants, the PV plants will be run whenever they are available. There is no point in taking them off-line when the sun is shining, because their are no fuel costs to be saved. An intermediate-load plant would be turned off before a PV plant would be.

    On the other hand, the sun is shining during the peak demand hours in California, so the PV plant will likely mean that fewer peaking power plants are needed. Since traditional peaking-plants are the cheapest to build (because they have fewest hours per year to spread their capital costs over) they also tend to be the least efficient, the most polluting and the most expensive in terms of fuel-cost per KWH. So while PV would be run like a base-load plant, it replaces the high cents/KWH peaking-plants.

    Here’s a perspective on what base-load is, that I hadn’t considered before, though it applies to solar thermal rather than to PV.
    Some renewable electricity sources (e.g. bioenergy, solar thermal electricity and geothermal ) have identical variability to coal-fired power stations and so they are base-load. They can be integrated without any additional back-up, as can efficient energy use.

  14. Fuel cost alone is for a single-cycle peaker is around 10 cents/kWh these days, plus amortization, O&M, etc. If these solar plants can really provide daytime power at $2-3/W capital cost (10-15 cents/kWh) they are cost-competitive from day one.

    These two aren't PG&E's only large solar deals. Some others:

    177 MW – Ausra – Carrizo plain, solar thermal (linear fresnel)
    533 MW – Solel – Mojave desert, solar thermal (parabolic trough)
    900 MW – Brightsource – Mojave desert, solar thermal (central tower)
    107 MW – San Joaquin Solar – Coalinga solar/thermal hybrid

    That's over 2.5 GW and I think there's more. I believe PG&E's total peak capacity is in the 20 GW range. This kind of scale is typically driven by economics, not PR.

  15. Their expectation of 2kWh per year per nameplate watt is a little optimistic for California central coast

    Robert, I used to live in San Luis Obsipo county. These plants are in the Carrizo plain which is not considered central coast. Carrizo is sheltered from ocean fog and clouds by coastal mountain ranges. It's not quite desert but it's pretty close. Furthermore, PG&E has a couple of 550 kV lines running right through this area (they connect the Diablo Canyon nuclear plant on the coast to the state's main north-south intertie near Bakersfield).

  16. Well doggydog I used to live in Lompoc. I figured they checked the solar insolation before they spent their billion dollars. Green Engineer says that it's "grazing land". There's tons of cows on the central coast but they don't do so well in the desert.

    Raw land in silicon valley was over 2 million bucks an acre when I left.

    They are generating a billion kWhs a year for their billion dollar investment so if their only cost is a 5% return on capital, that would imply a price of 5 cents a kilowatt/hour. PG&E is charging 18 cents so there is room there for their profit margin.

  17. Solar may be more expensive than gas, but we have to consider 1) external costs and 2) possible future problems with supply.
    I love the price mechanism, but it does not take into account the cost of pollution. Gas plants emit pollution, and solar plants do not (although in the making of panels, some may be emitted).
    And we do not know what gas will cost 10 years from now. Coal burns dirty too, and they flatten mountains to get it.
    I am not a doomer, but it might be time to start backing away from fossil fuels, when possible. Go with solar, wind, nukes (I don’t mind glowing in the dark).

  18. I’m all for this. It won’t maintain industrial civilization at its present level, but putting lots of renewables in place now will help us achieve a soft landing.

    In the case of solar power there is one very mundane problem that needs to be considered: Dust. It’s one thing to measure the output of a solar panel in a laboratory and calculate that one acre of such panels will yield this much power, but you can easily lose 10% output (which I actually read somewhere) from dust accumulation. Especially in arid regions where huge arrays are planned and built, there is little rainfall to help wash off the collectors. I’m surprised by how much dust collects on my solar hot water heater panels, and we get a lot of rain here in Japan.

    So I wonder if this is included in their output calculations, or if they plan to have people going through the installation with big brushes or mops to keep the equipment clean. Unless this problem of dust is taken into consideration, it could lead to serious disappointment later. Just consider how big a problem dust has been to the exploratory robots on Mars.

  19. Doggy might have to check my math, but at $1 billion and 550 MW that puts the cost at less than $2/W. The best I can do for a home system is around $6/W. The panels alone run around $5.

    So why so much cheaper? Economy of scale? Big PV panel order? Maybe I should call up the developer and see if he would throw an extra 30 panels or so at $2/W for me. Sort of a big group buy. Maybe we could get the R-Squared discount.

    This implies to me that somewhere there is a big solar factory that is underutilized. Or just maybe the costs are really a lot higher.

  20. anyone having interests in financials and or investment–

    1]-PG&E are purchasing power in this proposition only. from some vendors which may be OPTISOL, SUNPWR, or some other utility supplier.

    2]-no financial detail has been released, nor has date for release been announced.

    3]-someone must provide capital, gain regulatory approval, handle EPC, manage facility throughout life.

    if/when items 1, 2, 3 are satisfied, EP &G can buy power to sell to CA residents. they appear to be free of all primary responsibilty for all front end make happen effort.

    record shows SUNPWR has done this in Germany and Nevada[Nellis AFB]. it looks like they may do sme for FPL in FLA[25 MEGW PV]. the other outfit has no advertised track record.

    this whole item looks good, but too many details in financials missing to assess business proposition. transmissin detail and tax credit status also unclear.

    if successful, PG & E gets jump on state requirements; two vendors get big start-up boost.

    if unsuccessful, who gets to walk away free and clear; who loses what and how much.

    too many open business questions for investors.


  21. It will be interesting to see how the Sierra Club reacts. PhotoVoltaics has a huge footprint compared to conventional power sources. As utilities found out with nuclear power plants, a few aggressive lawyers can delay a project for decades and destroy the economics.

    As for the technology, even if successful this venture still leaves PhotoVoltaics on the margin — unable to scale up to replace more reliable power sources. Enthusiasts need to focus their attention on energy storage — the missing link in so-called renewable energy’s chain.

  22. King, this article says $1b+ so I figure $2-3/W. Optisolar makes their own thin-film silicon panels. They focus on large farms, I don't know if they even sell on the open market.

    I listed some other PG&E deals. No one ever quotes firm prices but $2-3/W seems to be the ballpark. They now say Optisolar won't install panels in this location until 2010, so they're forward pricing.

    The article also has a site photo. Not quite desert but I don't see any cows munching away, either. It does green up when the spring rains come. The area averages 8 inches of rain per year. There is some ranching up there but it's not what I'd call big business.

  23. Kinu, storage won't be an issue for a while. PG&E will use this for peaking power. When they need more peak power than these projects produce they'll use the single-cycle NG peakers, just as they do today.

    When storage becomes an issue note that more than half of PG&E's deals are with solar thermal plants. Thermal tanks are a pretty cheap way to store solar energy and all the solar thermal vendors plan to add storage as an option as they scale up.

  24. Doggy – with that big an order I guess you could build the factory and base load the facility with this order.

    If we can get panels down to $2.50/W then home solar starts to make sense. I keep up with kit prices. At 5% rate of return and $6/W home systems are still down around $0.20/kWh, still higher than a long term contract with the local utility.

  25. "PG&E will use this for peaking power."

    Not for winter night peaking, obviously. May help on summer afternoons, but output will start falling off shortly after midday as the sun heads towards the horizon. So PG&E will have to line up other peaking power sources as well.

    This is what makes the true economics of intermittent power sources so difficult to estimate — especially for large scale applications.

    In effect, PG&E is trading off lower fuel costs (when they can use solar) versus higher capital costs (duplicating the peaking power source & reserving fuel supplies for it, even if only used occasionally).

    Any way you cut it, there are additional costs that have to be added to intermittent power sources before we can make an apples-to-apples comparison.

    Which bring us back to the necessity for very large-scale low-cost safe energy storage for intermittent sources. We should all have learned from the ethanol debacle that we can't just avert our eyes from technical issues and hope it all turns out OK.

  26. PG&E has winter night peaking? That's news to me. Even in the winter of 2000-2001 during the California electricity crisis, we had rolling blackouts during the day, between noon and maybe 4pm, but not at night.

  27. I have wanted to critique solar for some time, but it is a daunting task. I would like to add just a couple of observations though:

    Solar cannot take the place of other systems as a peak load system. California’s power grid operator posts a daily system status graph. Loads start to rise in the morning, but they do not peak until after 3 pm. They remain above noon levels well into the evening.

    The sun reaches the zenith around noon (it depends upon where the plant is in relation to 120 degrees west longitude and on the season of the year). After that it heads towards the horizon. This is a significant mismatch with the diurnal power cycle.

    If you are building a solar thermal system, it is fairly low tech to heat the working fluid, store it for a few hours in an insulated container, and then use it to create electricity. Solar PV will need some other storage system — type and cost yet to be determined.

    Until that happens, PV is not ready for prime time — literally.

    Second. If you check NREL’s solar power maps you will find that the California central Valley area has an insolation of about 6 kWh/m^2/day. If I did all the conversions correctly, 9 square miles in that area should have an annual insolation of about 50*10^9 kWh.

    The gross efficiency of these projects therefor seems to be about 2%. I am sure that much of the 9 sq. mi. will be taken up with access roads, auxiliary machinery and the like. I suspect that no matter what the efficiency of the individual solar cells is, this factor will not allow a tremendous reduction in land costs for this use.

    Another way to look at it is that the US uses about 4000 billion kWh/yr. If you get power at the rate of 1 billion kWh/10 sq. mi. you would need about 4000 sq. mi. to cover 10% of that. And that is before the storage problem is solved. It sounds very expensive.

  28. Fat man, CA solar peaks at 1pm in the summer due to daylight savings time. That’s a 3 hour mismatch from the 4pm peak. As you note thermal storage is cheap. In fact, Ausra says storage reduces total system cost because they can downsize the turbines. PV could work in conjunction with solar thermal or CA could shift the peak during the decade+ it’ll take to scale PV up. The diurnal mismatch is an issue but not a showstopper.

    Don’t put too much stock in these land area numbers. These companies often lock up more land than they initially need to allow for future expansion. Note that Ausra’s deal in this same area is 177 MW for 1 square mile.

  29. “The diurnal mismatch is an issue but not a showstopper.”

    There are no show stoppers, but every step you take costs money. The thing that bothers me most about solar is the failure of any of its proponents to define the real costs of the system.

  30. What’s the “real cost” of anything? Fossil powerplants pumping previously-sequestered carbon into the atmosphere? Nuclear proliferation or long-term storage? Or meltdown? Just because we can push these costs off onto future generations doesn’t mean the cost is zero, as so many like to pretend.

Comments are closed.