• Friday, November 20, 2009 Latest Update: 4:41PM
Eric Wesoff | September 23, 2009 at 3:55 PM 15 Comments

SunPower: How Important Is High Efficiency in PV? (Updated)

SunPower's Doug Rose, the senior director of technology strategy, presented at the Silicon Valley PV Society in a talk titled, "Technology and Economics of High Efficiency c-Si PV." Of course, the thrust of the talk was the strength of SunPower's high-efficiency solar cells and panels, and the impact of efficiency on the cost and payback of a solar system.

The high efficiency of SunPower's solar cell stems in most part from its back-contact technology – a technology pioneered by founder Dick Swanson in the early 1980s at Stanford with low-cost manufacturing breakthroughs in 2001. The back contact design avoids gridlines on the front of the cell so there's no metal obscuring the cell and therefore more light gets converted to power. According to Rose, other design advantages are gained from the back-contact architecture – it allows better optimization of the front surface through texturing, an optimized backside mirror, localized contacts, and obviously backside gridlines.

The all back-contact cells allow SunPower to get to median production efficiency of 22 percent at the cell level. And while they're at it – cell thicknesses in the 150 micron range at about 6 grams of silicon per watt.

Rose raised the question: "How can high efficiency cells be cost effective? You're not using the same platform as everyone else." The response was: "Sunpower spends a little more in cell processing to deliver savings across the value chain."

That's the value proposition of high efficiency cells. The cells are more expensive but cost savings are realized all down the line. 

So how much exactly is this "efficiency bonus?"

According to research performed by crack Greentech Research analyst Shyam Mehta – gains in efficiency drive cost reductions at all steps of manufacturing on a $/W basis, from feedstock cost to module conversion – a 1 percent improvement in efficiency leads to a 5 percent to 7 percent decrease in fully loaded module cost. (Shyam's most recent report is on PV Manufacturing in the US and can be found here). His efficiency thesis is charted below:



In a solar market where prices are plunging, margins are crumbling and market consolidation is on the horizon – how much of a premium can SunPower command for its high-end product? A banker friend believes the dollar per watt premium is only 10 percent to 20 percent over conventional silicon or thin film PV.  With SunPower at a less than $2 per Watt module price in the fourth quarter of 2009 and some c-Si vendors below $1.50 per Watt – can SunPower command a 35 percent premium?

SunPower believes it can. My banker friend says no.

Here are some of the benefits of higher efficiency and the SunPower cell structure:

  • Lower area-related costs
  • Reduced installation costs
  • Reduced shipping costs
  • Reduced Balance of Plant (BOP) costs
  • Allows more Watts in area-constrained sites, which reduces the $/W cost of project costs such as sales, permitting, design, etc.
  • Delivers more energy per rated watt because of a better temperature coefficient, low light performance, broad spectral response, no LID

All factors resulting in a lower LCOE.

A Very Few Words on LCOE

A simplified formula for Levelized Cost of Energy (LCOE) is:

LCOE = Panel cost + BoP cost + O&M costs / Sunlight collection * Conversion efficiency

But, unfortunately it's not really that simple.  SunPower has detailed calculations and displayed the many factors influencing LCOE in its presentation. NREL has its own byzantine formula for LCOE.

An accurate measure of LCOE will have to include:

  • Initial investment
  • Depreciation tax
  • Annual costs
  • System residual value
  • System energy production

And LCOE calculations have a very high sensitivity to certain input variables such as:

  • Annual panel degradation
  • Differences in annual discount rate / cost of capital
  • System life (inverter replacement, etc.)
  • Annual O&M

The major contributors to LCOE are:

  • Capital costs
  • Modiule $/W
  • Area related BoS
  • Electrical BoS
  • Project related costs

"If someone says the LCOE of my technology is x cents per kilowatt-hour, it still doesn't tell you a lot," said Rose.

Differentiation and Branding in a Commodifying Market

A healthy cost structure, a good balance sheet, and the right level of vertical integration are what will distinguish winners from losers in the coming solar shakeout. Differentiation is going to help as well. And SunPower has that technical differentiation by virtue of the highest efficiency commercial solar product – a 22 percent median efficiency in 2006 looking for over 23 percent in its Gen3 cells. Combined with itss one-axis trackers which increase capacity factor by about 30 percent to match energy production with summer load, an important point for utilities – SunPower has some of the crucial ingredients for survival in the demand-constrained solar landscape.

Single axis tracking is a tremendous lever to reduce the LCOE of power plant, and to deliver significantly more power when the utility companies most want it (late afternoon in summer).

Of further interest in the differentiation department is SunPower's recent plunge into consumer branding of its panels. Ride a bus in San Francsisco and you'll see a SunPower-branding consumer ad campaign. 

Three questions for our readers:

  • Do consumers care which brand of solar panel they're buying?
  • What is the real value, the real premium for high efficiency?
  • And contrarily – what is the penalty for low efficiency?  Where does 6 percent to 8 percent efficient a-Si or OSC fit into the solar landscape?  Or does it?

We welcome your thoughts.

Comments [15]

  • Sam Jaffe 09/14/09 11:56 AM

    You say “Can Sunpower command a 35% premium? Sunpower believes it can. My banker friend says no.”

    I disagree with your interpretation of Sunpower’s beliefs. It’s obvious that Sunpower can’t command a 35% long term premium. That’s why the company is targeting a sub $1/W future cost. It will achieve that mainly through efficiency gains. That’s where Sunpower’s long term health should be judged. When it gets to 25% efficiency and under $1/W, then it has to be judged against First Solar’s 50 cents/W panels (where FS claims it will be in four years). At that point, SP has to have a premium that is worth the extra 50 cents per W. Keep in mind that by the time FS gets to 50 cents per W, it will be somewhere around 16% efficiency. That means that single-axis tracking will be economically viable. Thus the majority of costs in any system will be BOP—inverters, tracking and actuation mechanism, racking, cabling, Power control modules, etc. Having a higher efficiency panel will lead to a greater “thin film tax” than in today’s market. It’s my guess that five years from now, Sunpower and FS will be in a dead heat for overall system costs. By then, however, a disruptive CIGS technology might be even cheaper and the argument might be moot.

    Reply
  • StevePluvia 09/14/09 12:35 PM

    Typical BS sales pitch from Sunpower.  LCOE is the wrong metric for differentiating which panel to select for a project. Here’s why:  LCOE is always better if installed cost is lower.  Installed cost is the key metric to follow, with influence from the following criteria:

    1.  Site specific conditions that take advantage of lo light performance of thin film (*generally* thin film performs better than c-Si in lo light, thus paying more for hi efficiency that actually performs the same as lower efficiency thin film would be a mistake);
    2.  Foorprint:  If real estate is dear, hi efficiency is key, but only at the right cost.  For example if you can buy 18% cells at 1.80/watt vs 22% modules at $2.30/watt the cheaper modules might pencil out better.
    3.  Inverter technology; new inverter technology may improve system performance, but calculate the cost first.  For example installing expensive microinverters on every module results in a lower inverter budget when using hi efficiency modules.  If microinverters improve performance enough to justify the added hi efficiency module costs…

    Reply
  • ilyaf 09/14/09 1:55 PM

    Steve,

    I’m not sure I agree on the “LCOE is always better if installed cost is lower.”. That’s true for the same technology, which generates the same kWhr/kW installed. However, if I have a tracking panel Si at the same install cost as a non-tracking solution with the same efficiency (same peak rating), I think the LCOE will be lower, because you can (at least in a model) sell more energy per installed cost.

    On another note: FSLR and SPWR have been fairly open about their production costs and margins. Anyone know what those values are for other manufacturers?

    Reply
      • ilyaf 09/14/09 3:39 PM

        nice stuff ECD Fan, thanks.

      • StevePluvia 09/15/09 10:29 AM

        ilyaf—you are correct re tracking.  We consider tracking a form of reducing your install footprint, or said in another way, tracking = installing higher efficiency panels (as yield per installed sq ft is higher).

  • semiconductor_rep 09/14/09 2:23 PM

    ilyaf: hear you on the LCOE comment about Steve’s statement. But you are mistaken about SPWR’s openness: their actual production cost is known to all but a few. They have never disclosed it.

    Reply
      • Kevin Christy 09/14/09 2:34 PM

        LCOE is King. It’s the only way to compare differing module types, efficiencies, mounting structures, cost of maintenance, temperature coefficients, etc.. That said, you don’t want your module manufacturer quoting LCOE to you, you want them quoting price.

      • ilyaf 09/14/09 2:56 PM

        semiconductor_rep: Referring to a recent talk by SPWR (Doug Rose), who said their production cost today is at $2/W with a target of $1/W by 2014. Not sure about the accuracy of this number, but at least it’s been stated. A drop of $1/W is quite aggressive over a four year span. FSLR did a drop from $2.40 to $1 in five years, but that’s definitely a different technology and mostly given by economy of scale.

      • StevePluvia 09/15/09 12:28 PM

        LCOE includes several calculations that do not improve module selection decisions and result in inaccurate data, including specifically on the cost side: cost of money, maintenance of inverters (which technically should be modeled separately based on the inverter technology).  On the power side, power production will be dependent on performance at site specific conditions, (lo light, hi heat) and inverter technology selection.  These power calcs really can’t be predicted without historical data at this point because the numbers differ *substantially* depending on the data source (try asking 5 different inverter tech manufacturers about power production, or module mfgrs about power production in different light & heat). Therefore, LCOE ends up being both inaccurate and overly complex.

        A simpler and more accurate metric for considering module selection is installed system costs per watt using fixed budget for inverters.  For tracked systems, you simply increase the installed wattage. e.g. if your tracking increases power production by 30%, your installed watts increase by 30%, add the tracker expense and modify the footprint.  Once you narrow your module selection, site specific criteria (heat, light etc) and very specific inverter modeling should enlighten you as to the best module for the site and provide fairly accurate ROI info, assuming your source data is credible.

        Most people who rely on LCOE either never calculated LCOE or forgot the devil was in the details.

  • ECD Fan 09/14/09 2:57 PM

    To StevePluvia:  I can’t disagree with your points, however, thin film degrades worse than crystalline over the long term (particularly when comparing poor quality thin film aka Unisolar vs high quality crystalline aka SunPower) - we are talking of something like 0.5% differential in annual linear degradation, but over 20 years that gets to 10% difference in power output.  As far as First Solar is concerned, have you looked at the actual data from the Tucson study that First Solar touted at the Investor Day?  I suggest you do.  You will then see what degradation does to the output even after a few short years and then you will start having second thoughts about the management’s “honesty!” 

    To Eric:  Of course, there is a premium, as efficiency directly affects BOS costs.  The question is how to model it.  I don’t think it is % of module price, it is more like cents per Watt.  But, hey figuring out the precise premium to pay makes the difference between profits and losses for people like Phoenix when they chose between First Solar and the Chinese, so don’t expect the correct answers from your readers (if they truly know it and their business depends on it, they won’t tell you)

    Reply
      • StevePluvia 09/15/09 12:38 PM

        ECD, I read the FSLR presentation on investor day, but that was then.  What stuck out as a problem re degradation?  All PV is know to degrade; the FSLR spec sheets (along with all other PV mfgrs) disclose this information and warranty against degradation beyond the norm.  As I recall FSLR panels were performing more than 100% of namaplate on several sites, well above the Light Induced Degradation (LID) threshold that would triggered warranty concerns?  What did I miss?

  • Real Science 09/30/09 10:49 AM

    Today solar is subsidized, and whether LCOE is king or cost per Watt is king depends on the nature of the subsidy.
    For feed-in-tariffs, LCOE is king because you get paid for energy output, not Watts installed, and ignoring the complexity introduces significant inaccuracies.
    For per-Watt rebates, cost per watt is king but LCOE still matters.
    For per-Watt rebates and fixed rooftop systems I’d agree that per-Watt is so dominant that other LCOE differences can be ignored for all but large commercial installations.

    Reply
      • StevePluvia 09/30/09 11:59 AM

        Real Science—Excellent points, agree 100%;

        “whether LCOE is king or cost per Watt is king depends on the nature of the subsidy”.

  • Solvida 09/30/09 1:26 PM

    I agree as well but would add that considering project development/install decisions re: components/ types of materials- for ex sch 40 vs rigid pipe vs EMT, thicker wire for less V drop, multiple small inverters vs central vs micro, and install methods- conduit over edge and down wall or thru roof jack, etc… all effect initial $/w and $/kwh over time. Less maintenance, less oversight,  more power may mean quicker return on investment for any PV system. LCOE as an acronym is getting bashed with some valid and some not so valid rebuttals (IMHO)  but thinking about any capital investment over its cycle time seems to make sense.

    Reply

Green Light

Greentech Media's Green Light blog covers the full-scope of the greentech world, while expanding the range of our daily news reporting with brief and insightful blog posts from our Greentech Media editors, GTM Research analysts and numerous guest bloggers.

.