A Thin Film Solar Installation Revisited

(or, the kWh/kW myth, part V, as well as other myths)
A year has passed since an employee first blogged about Magco Inc.’s thin-film solar panel installation. That article [cached] generated a lot of interest and questions, so she wrote a follow-up [cached]in September. It is now a good time to take a look at the data from theremote monitoring system and report on some real-world figures.
As a reminder, in May of 2008, Magco Inc., a Tecta America company,installed 27.744 kW of United Solar’s triple-junction laminates on themetal roof of the company’s warehouse/office building. The totalinstallation cost was $215,000 (including inverter and hiring a masterelectrician). Magco, a roofing company, installed these products ontheir own building to “go green” and better understand the real-worldcommercial applications of solar technology. The home page of their installation boldly proclaims: "Let Tecta show you how to save money and the environment."So, let’s see what Magco/Tecta "learned" (or should have learned) inthe real world over the past year and what they can show us.
The analysis below relies on the recorded data from Magco’s system as well as from PV systems installed in the nearby area by Chesapeake Solar. Those data (and sources) are presented in a spreadsheet here.
  • Magco anticipated producing about US$9,000 each year from the panels. Using the average commercial electricity rates of about 13c [cached]in the state, that comes to expected annual electricity generation ofabout 69,230 KWH a year, or about 2,500 kWh/kW a year from theirsystem. However, over the past year, Magco’s system generated onlyabout 32,800 kWh, or 1,180 kWh/kW, which is an underperformance of 50% vs expectations.In fact, the system annual performance will probably continue todegrade a bit over the next few days and weeks, as the initialoverperformance wears off (Unisolar’s panels experience rapidperformance degradation in the first few days and weeks after exposureto light, until the performance stabilizes around the rated Watts underSTC, and then it declines more gradually as the years pass).
  • Magcobelieved that Unisolar’s thin-film typically performs better in low orindirect light (for example, cloudy) conditions than silicon solar. Yet in the months October, 2008 – February, 2009, a period that is not known for much sunshine, Magco’s system underperformedall crystalline (both Evergreen and GE) installations, on a monthlykWh/kW basis. The fact is that at dawn and dusk, as well as undercloudy conditions, the ambient temperatures are typically lower, thus,improving crystalline’s relative performance.
  • Magco’s understanding was that Unisolar’s panels perform well in high temperatures because of the chemicals involved [italics mine]. Butin the month of August 2008, a high-temperature month (and a few weeksafter exposure to light of the laminates), Unisolar’s system delivered worse performance, per kW, than the two crystalline (Evergreen) systems.
  • Magco thought that solar panels typically come with a twenty-year warranty. That is wrong,of course. Unisolar’s warranty is one of the worst in the industry (20year, 80% rated power) – even thin-film modules typically come with a25-year warranty.
  • Magco was convinced that Unisolar’s panels operate around 12% efficiency. That, of course, is false. Unisolar’s panel-level efficiencies are among of the lowest in the industry and range between 5.9% and 6.7%, and the PVL-136 modules specifically used by Magco, have only 6.3% efficiency.
  • Magcothought that their PV system was inexpensive, because the laminateproduction method did not use expensive silicon and mounting racks werenot required. However, silicon and racks are not all that isincluded in all-in system price per Watt. Magco’s system came to $7.75per Watt ($215,000 /27,744). That’s even higher than the $7.30 per Wattoffered by EnergyPeak aka Centria. And, based on the CSI data,the median pricing for a PV system sized 20-50KW was $7.46 (and 1/4 ofthe systems actually cost below $7 per Watt) in the main US PV market(California) at the time this installation was priced and installed.Also note that the Unisolar’s system should be priced at lower priceper Watt, by up to 10%, in order to have the same payback ascrystalline systems (as the PVL-136 are warranted for 20 years only vsthe 25-year industry standard, and may experience worse degradationwith time, over 1% vs less than 1% average linear annual degradation for crystalline and other glass panels). Thus, Magco’ system is actually quite expensive (by about 10%, adjusted, vs median) per kWhs of electricty expected to be delivered over the warranted life of the system.
  • Magcobelieved that somehow regular solar panels require extra structuralsupport and penetrations (unlike Unisolar’s thin-film laminates),particularly in locations that get snow. That is wrong, of course. Glass solar PV can be installed on metal roofs for less than 5 psf extra load using clampsor other types of lightweight non-penetrating racks, and every roofbuilt in the United States over the past 20 years has been designed towithstand up to extra 5 psf dead load, and thus, no new engineeringcalculations are required – see PV Inspector Guidelines [cached], page 14.
  • Magco hoped that the panels themselves do not require any maintenance. However, a document [cached](page 13) by a Unisolar OEM partner reveals the recommendation forregular 6-mth inspections, and the requirement for washing (water-jetcleaning) at least once a year.
  • Magco thought that thesepanels protect the roof because they’re glued right on top of it,extending its lifetime and potentially reducing roof maintenance costs. That may actually be incorrect as well, as the laminates, withtheir uneven plastic cover, could be a magnet for corrosive soot ordust while retaining moisture on the roof surface, and ungluing thelaminates (without damaging them) to repair the roof and then restorethe system could be harder than it seems.
On a side note, the laminates were apparently supplied to Magco by AGT, who now list this project as two separate ones (no doubt, to "improve" their project experience). The first instance is labeled "Magco Inc,"[cached] and the second one, "Tecta Solar" [cached]. Of course, both instances share the same roof picture – they refer to the same installation.