UniSolar and It's Power Output Warranty

 

One of the major changes is that the so-called "power output" warranty is now more complicated. Based on my reading, under the old warranty, a PVL-136, for example, was only warranted to exhibit a power output of at least 108.8W under STC (80% x 136) for 20 years from the date of sale. The new warranty apparently implies that the PVL-136 is warranted to: (i) during the first 10 years from the date of sale, produce at least 118.9W under STC (92% x 95% x 136), (ii) during the first 20 years from the date of sale, produce at least 108.5W under STC (84% x 95% x 136), and (iii) during 25 years from the date of sale, produce at least 103.4W under STC (80% x 95% x 136). See what happened? Unisolar can now finally claim that they have a 25-year warranty (which, by the way, is the industry standard term), but, in fact, they have slightly reduced the minimum warranted power output for the 20-year period (it is now 108.5W vs 108.8W before). Let me know if I have misread the old and/or the new warranty.

 

In any case, you probably shouldn't count too much on that new warranty, because the other changes to the text indicate that Unisolar could refuse warranty claims on virtually any ground. For example, if your panels have been subject to "neglect" or "excessive wear and tear," or "improper or inadequate ... maintenance or cleaning, or failure to otherwise use reasonable care in maintaining the PV Product," you are out of luck (what happened to PV being a maintenance-free source of electricity generation?). And if you thought that the Tefzel cover and the stainless steel made the laminates resistant to "vandalism" or "falling objects," you thought wrong. And make sure you know what you are gluing these things on, as "damage or corrosion caused by substrates, roofing materials, or their properties, such as corrosion resistance, thermal expansion and contraction, and water sealing properties, " would invalidate the warranty (apparently, even if the laminates themselves cause thermal stresses on certain plastic membranes, for example). And you definitely have to worry about "water pooling," if the roof surface is slightly concave in certain spots, something that does not affect optimally-tilted glass panels on non-penetrating racks. And make sure you keep and operate your panels in-house (or better, in a cleanroom) to prevent the "soot, chemical vapors, acid rain or salt water" invalidating the warranty (note that the somewhat-rough Tefzel surface may be a magnet for all those things, unlike the shiny glass surface of a typical tilted solar panel). Finally, beware of the "insects" and "animals" - it is a common knowledge, I believe, that stainless-steel-eating bugs are a serious menace, the squirrels might enjoy munching on the Tefzel, and then there are the bears.

 

So, in the unlikely event that you can somehow prove that none of the above apply to you, what can you expect from Unisolar? It used to be that if Unisolar had determined a warranty violation, they, at their option, 1) repaired or replaced, or, 2) refunded an amount equal to lost power (vs 80%-rated power) times the amount paid for the module per watt of rated power. Now they have a third option - "provide supplemental PV Product to meet the minimum power output." Of course, you will have to figure out yourself what to do with that extra odd panel or two - you can't simply add them to the rest of your system as they would unbalance the strings, and there is little extra space on the roof anyway - but, hey, maybe you can sell them on eBay?

 

If you wish to read some (even) older Unisolar warranties, take a look at page 3 of this document (modified December 2001, no major changes vs the 2005 version except the option of "additional modules to restore total wattage" and the lack of refund), and this document from December 1998 (announcing the 20-yr warranty to replace a previous 10-yr warranty).

 

And if you wish to compare the warranty to those of other PV makers, here are some samples, alphabetically:

 

Well, actually, Unisolar's current "a-Si triple junction" technology has been commercial only since 1997, see pages 40-43 here (previously, Unisolar was using "a-Si dual junction" technology). NREL got the triple-junction "stuff" only in the summer of 1998 (see this conference report by NREL describing the setup and summarizing the 6-year performance) , and since then has collected and reported back to Unisolar performance data on a quarterly basis, up to 2008 [my personal communication with NREL]. A sample of the reports Unisolar have been getting can be found in the infamous and now-removed, but still-cached-by-this-blog, Technical Report on pages 47-48. That sample also happens to be the last data point (March 2006) from NREL about this test that Unisolar has shared publicly. Unisolar appears to be actively concealing the subsequent degradation results and is not allowing NREL to publicly release the data. Wonder why? Take a closer look at those pages 47-48! Do you see that the estimated average linear PTC DC degradation rate is 1.14% /yr? As PTC is simply STC times a derating coefficient (about 0.92 for SHR-17), if NREL's PTC estimates are unbiased, it follows that the estimated STC degradation is about 1.24% (1.14% /0.92). That's not good. Remember that Unisolar's warranty used to be 20-year, 80% STC rated power, implying an average linear degradation of 1% or less (luckily, the new warranty seems to be promising only less than 1.26%/yr for the first 10 years). As the above-mentioned 6-year report showed a 0.99%/yr PTC DC degradation (data as of September 2004), the March 2006 data point looks like a disturbing acceleration of the degradation. It is no surprise then that Unisolar does not want any of the later data released to the public, especially the June 2008 result (which will be the 10th anniversary of the installation). Note, though, that so far we haven't proven a warranty violation (the March 2006 data point can still be explained away with the +/- 5% seasonal power variation and/or measurement errors).

 

However, in July 2007 an NREL researcher inadvertently disclosed another data point (December 2006) in a presentation on outdoor module and array testing. Buried on Slide 9 is the innocuous-looking chart below:

 

Yup, that's our system (with some of the previous data points only slightly different from the 6-year and the March 2006 reports, likely due to data clean-up). So, as of December 2006, the estimated average linear PTC DC degradation rate had accelerated to 1.22%/yr (or implied 1.33%/yr STC). Now, it is true that there are a few more (seasonally) low-power data points in this set and that the implied 1.33% linear STC degradation is still not enough to suggest a warranty violation under the old warranty, given a 5% seasonal variation, but it appears enough to suggest an imminent violation under the new 10-year warranty threshold.

 

Actually, we can be a bit more precise: the 72 x 17W-shingle system is rated 1,224W STC DC (maybe slightly higher, as the shingles are actually rated up to 17.3W STC DC each, based on the 6-year NREL report) and about 1,130W PTC DC. Assuming the SHR-17's production tolerance is similar to that of the current laminates (5%), the new Unisolar warranty promises power of at least 1,070W under STC (1224*92%*95%) in the first 10 years after the sale (that is, until June 30, 2008). In PTC terms, that promise is about 980W (1070 *92%). Take a look at chart above again - under that assumption (and also assuming NREL's PTC estimate is unbiased), it appears the system could have been in violation of that warranty for more than one seasonal cycle (from September 2005 till the last reported point, December 2006).

 

So, when is Unisolar going to release the June 2008 result to the public? And what is the basis (and supporting evidence) for that new warranty?

When Mr. Morelli, ECD's CEO, fleetingly mentioned on the Aug 28th, 2008 earnings call the "durability problems ... faced in the past that [Unisolar] have worked through," probably very few listeners cared to notice. After all, total warranty claims for ECD have been less than $0.22mill in each of the fiscal years 2006, 2007 and 2008 (according to the 10K, page 73). However, solar product shipments in calendar 2008 were 109MWs, compared to less than 5MWs five and ten years ago (see slide 4). And, as Part II indicated, the warranty concerns may not manifest themselves until only after the 6th year following the sale, thus those claims could start rising despite the skilfully-crafted new warranty document (particularly, if the durability problems have been "worked through" the same way the efficiency "problem" appears to have been solved).

 

Here are a few examples of what Mr. Morelli may have meant by the "past:"

1. An Australian PhD student analysing PV modules' performance for her thesis was surprised to find out that one of her two Unisolar US-64 modules exhibited severe, unexpected degradation long after the initial 8 to 10 weeks (when it was supposed to). On pages 75-76, she explains: "At the end of the [two-year] study the module [manufactured in 1999] had a maximum power at STC 12% below its rated value... The prolonged degradation was unexpected... Communications with the manufacturer of the US-64 module were unable to identify any cause or reason for the lower than expected performance... The module behaviour is not unique. LEEE have experienced and reported a similar result for two US-64 (3j, a-Si) modules. The manufacturer requested the return of one of the modules to determine the cause of the low Wp value. The module was tested by the manufacturer and found to have a Wp value of 52.1 Watts, or 18.6% below its rated value [my note: almost certainly, an imminent warranty violation]. LEEE subsequently received another module from the manufacturer and found, that, when stabilized it produced 25% more energy that the earlier module [Cereghetti, et all, 2000. Behaviour of Triple Junction a-Si Modules. In Scheer, H (Ed.) Proceedings of: 16th European photovoltaic solar energy conference, Glasgow, May. 2414-2417. James & James; 2000]... This inconsistency in power output poses difficulties for users wishing to reliably predict or model the PV output." Another US-64 module in her possession, made in 1997, or two years earlier, apparently was doing ok. Looks like a manifestation of progress, the Unisolar way.
2. Our LEEE friends did it again in 2002. Look at Tables III and IV in this report. The US-32 panel was down 23.4% vs rated power after only 15 months of outdoor testing. And Unisolar had the arrogance to present this in 2008 (for the Santa Cruz part of that "piece, " see this post).
3. The Agder University College in Norway, the host of a 20KW Unisolar installation completed in 2000 (see page 2 in the "Large Scale UNI-SOLAR® PV-Installations" section), analyzed, in a separate test set-up, the seasonal variations in yield for different types of PV modules measured under real life conditions in Northern Europe in 2004 and 2005. After correcting for an earlier error, the extrapolated data showed that the US-64 panel was generating just 51W under implied STC (Norway just doesn't have the 1000 W/m2 sunshine required for STC), and that's 20% below rated power.
4. NREL wrote a report after they visited India in 2003 to validate a 21-kW installation there consisting of five different types of thin-film modules, including 56 US-64 modules. Correcting measured power for irradiance and module temperature to bring it to STC resulted in the Unisolar modules delivering 8% less than rated power, basically from the get-go.

If would be very interesting to find out what percentage of the triple-junction a-Si MWs sold ten or more years ago are still producing any electricity (especially if they are producing within warranted parameters). Of course, we should exempt this installation, as it never generated any electricity to begin with. For crystalline modules, at least, there is some decent history of durability (see here and here, for example).

 

So what was the basis (and supporting evidence) for the new warranty, again?