http://windfuels.com/

Solar Photovoltaic (PV)

Solar Photovoltaic (PV)
The best cost study on solar PV comes from Lawrence Berkeley National Labs, “Tracking the Sun...”, published in February 2009. Their findings are extremely close to what we first published when we opened our WindFuels website on July 11, 2008. (We do our research carefully and objectively.)

They found the mean cost of installed PV dropped 3.5% annually from 1998 to 2007, though the decline ended in 2005. Some of the additional highlights from this study for the year 2007 are:

The mean cost of installed PV before incentives was $7.60/WPE.
The mean cost of installed thin-film PV was about 7% higher than for conventional crystalline silicon, at least for installations below 10 kW.
After-tax incentives for residential PV averaged $3.1/W, and $3.9/W for commercial.
The mean installed cost after all incentives averaged $3.9/W for commercial and $5.1/W for residential.
Cumulatively, about 10.6 GWPE had been installed globally by the end of 2007. Their annual energy output was about 0.07% of the global electrical energy output.


A bleak outlook is now projected for PV.


2008 was a boom year for PV, but 2009 will be devastating. A sobering update appeared on 9/02/09 in Renewable Energy World that paints a bleak outlook of PV for at least the next three years. The mid-range projection sees PV production getting back to 2008 levels in 2011 or 2012 and then growing from there at a 15%/yr rate in dollar amount of sales. No more of the 50%/yr growth rates seen from 2004-2008, even with prices plummeting because of enormous over capacity. A recent report (9/2009) by DigiTimes projects a 26% growth from 2009 to 2010 in terms of panel MW of sales, but production capacity utilization dropping to 26%. DigiTimes also projects that half of the solar manufacturers are likely to fail by the end of 2010. If one looks further ahead, under present conditions it seems likely that another 30% may fail in 2011, and another 20% in 2012.

Essential Facts. The average capacity factor for PV in prime sites is ~14% - that is, the energy generated in a year averages about 14% of the product of the number of hours in a year and the peak power output. Capacity factors for CSP are usually between 18% and 26% (even higher if storage is included) because they normally include tracking. However, that is not justified in most PV installations. Most PV panels are fixed at an orientation that is close to the year-round optimum. Peak output power is usually ~85% of the panel rating (which is at 1000 W/m2 normal solar insolation, 25°C cell temperature) due to a combination of lower peak local insolation, non-optimal orientation, cell heating, inverter losses, and imperfect inverter impedance matching. Hence, the energy output from a PV system “rated 1 kW” averages 1040 kWhr/yr.

A rooftop PV system “rated 2 kW” will cost ~$18,000 (installed, no storage). The value of the energy it produces is $230/yr at current US average consumer prices, but this assumes the energy can be fully used. Often the energy can’t be fully utilized during the mid-day, so 3 kWhr of storage is probably justified, which adds about $3000 to the cost. Clearly, the reduction in the utility bill doesn’t begin to pay the interest on the PV system, not to mention upkeep and principle. We return with a few more comments on rooftop PV later.

There is a wide-spread belief that the price of PV will drop dramatically over the next decade because of thin-film solar (Cadmium-telluride (CdTe), copper-indium-gallium-selenide (CIGS), organic thin films, or thin-film silicon) or because of strong growth in production capacity of high-purity silicon. However, the high-purity silicon boules (ingots) from which conventional cells are made represent only about 13% of the total cost in a typical PV installation. Most of the other cost components will not be easily reduced.

First Solar is the world’s leading producer of thin-film PV, and they claim to have recently reduced the “direct manufacturing cost” to about $1/WPE for the bare panels. However, one of the lowest cost installations soon to be completed will cost about $4.3/WPE. Part of the reason is that the low efficiency of CdTe (probably under 9%) requires larger panels, which require larger support structures. Also, the “direct manufacturing cost” is only about half of the selling price of the basic modules.

We show later that the available data and analyses do not support the notion that CIGS organic, or thin-film silicon will be much less expensive than PV based on conventional silicon. (And yes, we’ve studied the latest Solyndra, Konarka, Nanosolar, and Heliovolt patents and promotional materials.) The cost of the CIGS and CdTe PV semi-conductor materials themselves is quite low (because the film thickness is sub-micron), but the cells must be protected from minute moisture ingress and must be supported. Whether or not CIGS thin-film PV can achieve energy production in the real world that is competitive with CdTe or polysilicon PV cells will not be known for at least 5 years, as the payback depends on reliable performance for 20 to 40 years, and thin-film modules have been being shipped for only a few years. We address some of the thin-film products in more detail later.

Real PV Cost and Market Data. One of the largest PV contracts recently signed and publicly disclosed (by German-based Ralos Vertriebs GmbH, with Evergreen Solar, Mass.,) calls for the delivery of about 280 MWPE of 19% efficiency panels over the next five years at an average cost of about $3.60/WPE. The balance of system (land, support, tracking, washer, installation, inverters, power conditioning) will add $1.20 to $2.00/WPE. The data publicly available on First Solar suggests they are selling their CdTe thin-film panels (perhaps up to 9% efficiency) in bulk for about $2.5/WPE.

The current global annual PV cell production capacity is about 2.5 GWPE. For several years, global capacity grew at about 40% annually, though it seems unlikely to exceed 20% annually going forward. The global annual PV cell production capacity by 2014 is likely to be ~10 GWPE, or ~1.5 GWE mean. That would be about 2% of the rate at which coal power plants were being built in China alone in 2007.

World-wide total installed electrical generating capacity in 2006 was ~4000 GW (or 4.0 TW), and it produced over 18,000 TWhrs of energy (about 51% mean capacity factor). By the end of 2014, PV will still be generating only 50 TWhrs of energy, or ~0.25% of expected total electricity. Growth in PV will likely further slow after that, as capital investments have recently fallen.

The most extensive experience with commercial-scale PV is in Europe, and these PV plants are currently selling most of their energy (when converted to current US dollars) for $0.4 to $0.8/kWhr. The owners and investors do not seem to be expecting investment paybacks of less than 8 to 12 years, and they have always had state assistance with the R&D and capital expenses.

The 40 MW CdTe solar plant to be completed this year in Waldpolenz Park, Germany, is expected to cost $170M (though it appears that doesn’t include all state subsidies) and produce 40 GWhrs of energy annually. At $400/MWhr, the annual value of that energy initially is $16M, which, after interest, leaves $8M annually to be applied to O&M and principle. No data could be found on expected monthly cleaning expenses for this 250-acre field or expected annual panel degradation rate.

Most of the dozens of “larger” PV installations planned for the next several years in the U.S. by The Solar America Initiative will still be well below the 150 kWPE level, and they will be heavily subsidized at all points along the way from a combination of private, federal, and state sources – in the PV cell manufacturing, the balance of plant costs, the installation, grid connection costs, and in the operation and maintenance costs. A few installations above 1 MWPE (in Orlando, for example) may be completed within the next year; and a plant with up to 19 MWPE may be constructed over the next few years on a 300-acre site in Lennox County, Ontario, by SunEdision and SkyPower, as they have a guaranteed minimum price of $0.44/kW-hr for their solar energy. (Many other large PV plants are being talked about, but most are unlikely to materialize, as it will be hard to get guaranteed energy prices high enough.) At least a year of operation is needed from the second 50-200 MWPE commercial plant before real costs can be determined with an uncertainty below 25%. We are not likely to be to that point in the U.S. for at least 4 years.

Rooftop PV. There is a wide-spread belief that distributed, roof-top solar PV will be more cost effective because it will avoid the transmission costs. However, residential roof-tops are seldom at an optimum orientation for solar PV, and usually solar tracking and automated cleaning are not practical. Moreover, the 60-Hz power inverters add (without storage) over 15% to the initial system cost, their losses are typically 6%, and their lifetimes are usually under 10 years. These relative costs and losses are much smaller in large installations, which further benefit greatly from tracking capability and automated cleaning systems. The combination gives large installations a 60% to 150% advantage compared to roof-top solar in cost-effectiveness – with the larger factors being location, installation costs, tracking, and other economies of scale.

Thin-film PV. The basis for most of the promises of low-cost PV in the future is that copper-indium-gallium-selenide (CIGS) PV cells can be made in very thin films and the production panels may achieve higher efficiency than CdTe. A transparent oxide conductor is used for the electrode on the surface exposed to the sun, and a number of different methods have been developed for depositing the additional layers needed of the co-deposited materials (usually by sputtering in high vacuum) and electrodes.

A major challenge is that the CIGS materials degrade rapidly in the presence of minute concentrations of moisture at the temperatures that are seen during operation. Other challenges include the large degradation in efficiency as the temperature increases and the large change in optimum load impedance as the temperature changes. As a result, average output power for CIGS has been much lower than for conventional cells, especially after a few years.

One of the few companies willing to publish much detailed information on a CIGS thin-film module as recently as mid-2008 was Global Solar Energy (GSE). Their 60 WPE panel (possibly based on the roll-to-roll metal foil processes described in their NREL reports) achieved a peak efficiency of ~8% at 25C with optimum load (when new), and it retailed for ~$300. However, it appears their focus now is on flexible products, and these generally retail for about $13/WPE and achieve about 6% peak efficiency. The warranty is usually 1 year.

The figure below shows the impressive growth in research-cell efficiencies over the past 3 decades, but it’s important to keep in mind that efficiencies of production modules are generally between 40% and 80% of best laboratory results. Efficiencies of all except the very expensive multi-junction devices have plateaued.

The expensive multi-junction devices are expected to be used only in Concentrated Solar PV (CPV), which allows better use of a smaller expensive PV cell but requires complex optics and precision tracking. Concentrix has recently announced http://www.concentrix-solar.de/news/?L=1 they are achieving 23% efficiency from their 100 kW modules, which use an array of flat Fresnel lens focusing the light onto very small cells. No cost information is available, but it seems highly unlikely that this approach could ever be competitive with conventional flat-plat PV.



Active thin-film companies include:
http://www.firstsolar.com/
http://www.uni-solar.com/interior.asp?id=102
http://www.globalsolar.com/
http://www.solyndra.com/
http://www.konarka.com/
http://www.heliovolt.net/
http://www.nanosolar.com/
http://www.signetsolar.com/
http://www.csgsolar.com/pages/technology.php?lang=en
http://www.sulfurcell.de/index.php?id=1&L=1
http://www.miasole.com/www/
http://solopower.com/index.html

First Solar produces thin-film glass-encapsulated CdTe panels up to 75 W that are probably about 0.9 m2 each and 6.8 mm thick. They don’t release information that would allow efficiency calculations. They sell only to large installations in bulk, and it appears their average selling price is about $2.6/WPE. They claim manufacturing costs of under $1/W. They also claim to have come close to 11% efficiency in some tests, but typical efficiencies are probably closer to 9%.

Data from their glass-encapsulated CdTe products from 2003 showed high degradation rates (over 5% annually) from the combination of thermal stress, sun exposure, and bias stress. The quality has improved over the past several years, but the sum of mean degradation and failure rates is still probably several percent per year. We could not find any recent test data. At just 2%/yr with no replacements, they would be at half power in 35 years.

Oerlikon Solar, a Swiss company, expects to be producing equipment in 2010 that will allow production of amorphous/polycrystalline panels that will beat First Solar in both efficiency and cost. They have recently announced plans to provide equipment to Rusnano for construction of what may be the world’s largest thin-film silicon solar panel plant, to be built in Novocheboksarsk, beginning in 2010.

Uni-Solar is producing flexible amorphous-silicon PV modules, 4 mm thick, with about 7% efficiency when new and about 6.3% after 10 weeks of exposure. Efficiency is claimed to be over 5% after 20 years.

Konarka has begun roll-to-roll production of polymeric thin--film PV cells based on organic semi-conductors. Their largest module, the KT-3000 (2.38 m x 0.65 m), is rated 26 WPE when new. That amounts to about 1.7% efficiency.

The best efficiency reported on a laboratory sample (see US patent 7,414,188, or Advanced Functional Materials, 16, Wiley, 2006) was ~4.5%, but these tests on glass slides have limited relevance to panels from a roll-to-roll process.

A series of many complex processes is required, and very expensive substrates (such as molybdenum-coated thermoplastic polyimides) are needed (because high processing temperatures are required) along with expensive nano-particle-based dopants. From a press release in June 2008, it appears that 80% of some samples “on flexible substrates” encapsulated “with low cost packaging materials” degraded by less than 20% in what was similar to about one year of normal sun exposure in a humid climate. (Of course, that’s not exactly the way they presented the test results.) The packaging material had a water permeation rate of 0.1 g/m2/day. (That’s an order of magnitude lower permeation than the best available 2-mil polymeric conformal coating at 37C, Parylene-C, which is not cheap.)

No aging or warranty information is available on the Konarka products. Since the Uni-solar flexible products (which are 8 times thicker and use a much a much more robust semi-conductor), are expected to degrade by about 30% in 20 years, a reasonable estimate is that the Konarka products might degrade by 30% in 2 years of steady usage.

Buying a flexible PV cell that costs $12/WPE, lasts 3 years, and gets used continuously under excellent conditions is like paying $600/MWhr for electricity, or $22/gal for gasoline. But of course, if you’re comparing the energy cost to that of disposable batteries, it’s a completely different story. You only need to use the PV cells for 10-20 hours a year for them to pay for themselves in a few years – and they might last 15 years at that usage level.

The Konarka plant is expected to soon be capable of producing 10M m2 of PV film annually. Perhaps ultimately it will have 3 times that capacity, which might be sufficient to produce 1 GWPE of annual capacity. However, it’s hard to imagine much of a market for such low-efficiency cells that are likely to drop below half of initial performance within 4 years in the field at steady usage.

Solyndra has generated a lot of attention because of the very strong funding they have received to produce and market thin-film PV CIGS cylindrical cells. They expect to achieve long lifetime by having the thin film inside of sealed, glass tubes.

For a product that apparently may have already booked about $2B in advance sales, it’s surprisingly hard to get real numbers on this very secretive company. Their claim of “more solar power per rooftop” is blatant disinformation, and it raises suspicion about all their other claims.

They use 40 of their 1”-diameter tubes, spaced 1“ apart, on a 1 x 2 m2 frame. The thin-film CIGS PV material may be deposited on a glass rod in a sequence of layers and then enclosed in a glass tube, or the cells may be made on a flexible film substrate that is then slipped inside the glass tube. Many options are discussed in their pending and issued patents. Although their patents are better written than most in this field, the efficacies of the processes are still difficult to assess, and it’s not possible to determine what approach they are emphasizing.

Only half of the roof will be directly covered by the tubes, and almost all of the radiation (all the time) will be striking the tubes at angles that result in a lot begin reflected back into space. Moreover, energy production is likely to degrade within a year, as the tubes will be difficult to keep clean on their lower sides, where light reflected from the roof is received. (The Nevada Power One station finds it necessary to wash their mirrors weekly. Granted, surface dust is much less critical here.)

Their data sheet specifies initial peak output power of the 1.08x1.82 m “panels” (arrays of tubes) ranging from to 150-191 WPE, which corresponds to 7.5-9.7% efficiency by the normal (gross) method. These specs (like most PV specs) assume the panels are aimed directly at the sun and the cell temperature is 25C (which requires an air temperature of 0C and a good breeze). They have a 5-year limited product warranty. They also warranty that the products will continue to produce power for 25 years if not broken, but they don’t say how much power.

They claim 13% solar efficiency, but they also claim their 1 x 2 m “panels” (arrays of tubes) will generate 180 WPE, and this would correspond to 9% efficiency by the normal (gross) method. However, this is still higher than seems likely (at least after the first year), and Solyndra won’t release anything about the conditions of their tests.

Judging from what Solyndra reports for the mass of their panels (16 kg/m2 of roof area), it appears that they may be using solid glass rods coated with the thin-film CIGS cells inside glass tubes. The cost of the glass alone will probably be about $6 per cylinder (rod and tube), and the cost of the caps and connectors may be one-third that amount. Manufacturing is quite complex and requires a number of sequential high-vacuum processes, etching (for the electrodes), and interconnections.

The cost of producing the Solyndra solar cylinders is anyone’s guess. The selling price of mass-produced manufactured items is often ~6 times the materials cost. (An item in mass production that is outwardly somewhat similar to the Solyndra receiver tubes is the fluorescent light tube, which sells for about $4 each in bulk; but there the glass is only about 0.9 mm thick. Another useful reference item is the common CRT used in oscilloscopes, which requires glass, sealing, electrodes, and simple coating deposition processes. They start at ~$100 each. Another useful reference point is to note that the cheapest ICs cost about ~$5/cm2. The Solyndra PV process, though simpler, is similar in many ways. However, their processing cost must be over three orders of magnitude cheaper than IC processing per unit area to have any chance of being competitive.)

Assuming $20 per tube, the panels would need to sell to the end user for at least $1200 each, which is $6.7/WPE if the roof happens to be at the optimum orientation. For the more typical flat roof, it’s about $10/WPE– prior to installation and balance-of-system costs. Several posts by other PV experts have indicated they also think Solyndra’s manufacturing costs are in the $8-10/Wp range.

Clearly, Solyndra must be selling their 160-W panels at under $700 each, or they could not have booked any advance sales. How much they are losing on these sales will not be known for at least another year. When their current plant is up to capacity, it is expected to produce 50,000 panels per month.

Solyndra is set to receive a $535M load guarantee from the DOE to build another plant when no data are yet available on the actual production capacity of the first plant or on the production costs of panels from it.

The engineers at Heliovolt have been talking about making CIGS thin-film PV cells faster and cheaper for more than a decade, but its hard to find much useful information on their progress. They have a lot of patents (6500733, 6559372, 7148123, etc.), but it’s not really possible to tell yet if Heliovolt has ideas that will translate into improving CIGS processes. They have demonstrated over 12% efficiency in a CIGS test sample on a glass slide by a process reported to be fast, but as yet there is no useful information available on any products that might be produced. The cost-dominating issues with CIGS have been how the thin film will be supported, protected, and packaged.

Signet Solar has just begun manufacturing thin-film-silicon modules up to 5.7 m2 that are 8.3 mm thick, and they achieve 5.5 - 6.2% peak efficiency. The polysilicon film is protected on the front and back sides with glass sheets, each 3.2 mm thick. It appears that this company does not have any patents, pending patents, or technical publications. They have a 25-year limited warranty.

CSG Solar has apparently produced thin (poly) crystalline silicon on glass (CSG) that showed exceptional longevity, but very little other useful information is available. Pricing information is not available for either of these thin-film-silicon companies. It’s hard to imagine how these low-efficiency panels can compete with conventional silicon getting three times higher efficiency.

SulfurCell is delivering 50-60 W modules based on CuInS2 that achieve 6-7% efficiency. They have a 20-year warranty, but output power decreases faster than most at high cell temperatures.

Nanosolar is another thin-film company with a lot of VC money, patents, and history, but no products or useful information currently on their website. They are focusing on processes that would be analogous to ink printing using coated nano-particles (see USP 7306823). For the last four years, Nanosolar has been saying production would begin the next year. There are indications (in notoriously unreliable media) they have shipped some products that may have achieved 12% efficiency. They keep saying they will soon be producing flexible PV film at under $1/W. It will be interesting to see what it really costs, what its efficiency is, and how long it will last.

Somewhat similar ink-printing processes are described in the NREL report 37284 by ISET (International Solar Electric Technology). They achieved efficiencies up to 13% on very small test CIGS cells (1 cm2) on rigid molybdenum foil, but efficiency dropped 15-30% within 3 months under mild aging conditions (apparently, just lying around in the lab). The performance degraded over 5% when held at 100C for just 1 hour.

If the Konarka products will lose half their performance in 4 years when well packaged, the cheap (poorly protected) Nanosolar products may only last two years. Perhaps they could be made to last 15 to 20 years with expensive packaging.

Applied Materials likes to hype the future of PV to promote its sales of equipment that is used in various semiconductor and PV processes. It does not appear that they plan to produce in PV material.

Thin-film Conclusion. It’s clear that all of the above companies (and others) have a lot of very smart people on staff, and they have all made some useful progress – but it falls far short of what they are continually saying to their investors.

Konarka clearly has a serious lifetime problem with organic semi-conductors on polymeric films – probably only 4 years. They also have serious efficiency (1.7%) and cost challenges – probably over $12/WPE. Assuming a 4-year lifetime and $12/WPE initial cost, their energy costs would be about ten times those of polysilicon.

Solyndra understands that their CIGS solar cells need to have at least a 25-year lifetime if they are to be competitive, but they have a poor appreciation for the cost of mass production of large, massive items. However, we suspect that their approach is better than those of many other CIGS competitors that don’t yet have products. Global Solar, however, is partnered with a large, mature company – Dow.

The bottom line is that we think it will be at least another 10 years before CIGS or organic PV competes with conventional polysilicon – if ever. It’s hard to fully evaluate the recent CdTe products because we can’t find any recent aging data, and the aging data from a few years ago was not good. Are we overly cynical? No – we just have 27 years of experience in development and manufacturing of complex systems and processes. We try to look at the science, engineering, technology, and economics very carefully and objectively.


Useful references:

Recent PV projections, 9/2009:

http://www.digitimes.com/print/a20090903VL200.html

http://www.renewableenergyworld.com/rea/news/article/2009/08/the-pv-industry-2009-in-search-of-stability-and-sustainability1?cmpid=WNL-Wednesday-September2-2009

Mid-west feed-in tariffs
http://www.renewableenergyworld.com/rea/news/article/2009/08/indianapolis-power-light-proposes-modest-midwestern-feed-in-tariff-program?cmpid=WNL-Wednesday-September2-2009

Feed-in tariffs in Sacramento of $0.17 won’t attract any solar:
http://www.renewableenergyworld.com/rea/news/article/2009/08/smud-announces-feed-in-tariffs-but-can-program-deliver-as-promised?cmpid=WNL-Wednesday-August12-2009

http://www.greentechmedia.com/articles/read/amorphous-silicon-solar-losing-the-shakeout/

http://www.greentechmedia.com/articles/read/solyndra-adds-238m-contract-brings-backlog-total-to-2b/

http://www.greentechmedia.com/green-light/post/lets-talk-about-cigs/

http://www.greentechmedia.com/articles/read/intersolar-will-oerlikons-silicon-rival-first-solar-and-more/

Feb 2009 PV cost study by LBNL
http://www.greentechmedia.com/assets/pdfs/berkeleylab.pdf

http://en.wikipedia.org/wiki/Photovoltaic_module
http://en.wikipedia.org/wiki/Copper_indium_gallium_selenide
http://en.wikipedia.org/wiki/Cadmium_Telluride_Photovoltaics

US patents and pending patents:
2007/0240760, Solyndra
2008/0110491, Solyndra
7407831, Konarka
7148123, Heliovolt
7306823, Nanosolar

http://www.greentechmedia.com/articles/solyndra-rolls-out-tube-shaped-thin-film-1542.html

NREL reports:
http://www.nrel.gov/pv/thin_film/pn_techbased_cadmium_telluride.html
NREL/SR-520-37284 (ISET, CIGS non-vacuum), 2005
NREL/SR-520-39119 (GSE, CIGS roll-to-roll), 2005
NREL/SR-520-38681 (GSE, CIGS manufacturing), 2005

The Energy Supermarket
http://shop.solardirect.com/index.php?cPath=69_88_199


http://ecoworldly.com/2008/03/05/worlds-7-biggest-solar-energy-plants/

http://www.greentechmedia.com/articles/duke-chops-100m-distributed-solar-project-in-half-5052.html