Your daily dose of photovoltaic technology developments and solar news
First Solar analyst day post-mortem, Part I: 52 cents manufactured cost per watt seen by 2014
Most Popular Articles
- UK feed-in tariff is announced: payback begins April 1
- Intel outside: Chipmaker moves to install more PV power systems, equipped with First Solar modules
- Suntech predicts excess solar supply will move to U.S.
- New Product: SunPower ‘T20 Tracker Evolution’ delivers maximum energy with 128-cell, 400-watt mo
- Project Focus: GA-Solar to build 300MW solar PV power plant in New Mexico
Popular Technical Papers
Subscribe to our weekly newsletter
Now that their analysts and investors meeting has been completed, the executives of First Solar have left the casinos and watering holes of Las Vegas behind them and gotten back to running their increasingly complex company. Although no news was forthcoming on two key issues—the location and timing of the company’s expected capacity expansion moves and the hiring of a replacement for chairman/CEO Mike Ahearn (although he did say the decision won’t happen until at least the fall)—the company did outline revised conversion efficiency, balance of system, and module manufacturing cost reduction roadmaps, as well as enough financial and related info to keep the bulls snorting and the bears growling for awhile.
The market wathers have already parsed, pared, and pontificated on First’s newly modified business models, the nuts and bolts of what Ahearn calls the company’s "sustainable competitive cost advantage." But the updates on the manufacturing and technology sides of the business warrant some additional scrutiny.
In a nutshell, First Solar’s revised module manufacturing cost reduction and factory run-rate roadmaps show more of the same relentlessly continuous improvement mindset the company has become known for.
On the module side, Sohn outlined how the new five-year plan calls for 56-68% reductions in the cost per manufactured watt, driving down to a targeted range of 52 to 63 cents per watt by 2014. The largest portion of the cost cutting will be facilitated by conversion efficiency improvements (about 18-25% of the weight, with better throughputs, plant scaling, etc. also contributing). The impact of building factories in low-cost locations such as Malaysia has been much reduced in the current roadmap, dropping from accounting for 15-17% of the cost reduction improvements to 3-4% in the company’s new gameplan.
As for the factory run-rate numbers, Sohn said the current 49+MW per line per year figure will jump to more than 80MW sometime in 2014, well up from the 25MW annual numbers that the company posted when it made its initial public offering in 2006—and way up from the 10-12MW figure that was the norm in the early days.
The per-line jump from 25MW to 50MW meant that First Solar had to build 50% fewer lines to reach that 1GW capacity number. Company execs expect that trend to continue, with increases in module output coming via improved yields, throughputs, and especially conversion efficiencies in existing manufacturing facilities, pushing out the need to build factories until they are absolutely necessary.
The presence of VP of technology Dave Eaglesham at the event, who is rarely allowed out in public, added a bit of welcome techie-geekiness to the proceedings. Since this was the first time that the company shared its conversion efficiency roadmap since that legendary IPO, he had a chance to talk about, or in some cases “dance around,” the details of First’s technologies and how they align with its long-term mission.
The VP, whose banter was sprinkled with talk of a technological "horse race" and categorical "buckets" and numbers being "racked up," made it clear from the git-go that First Solar is in constant technology assessment mode, externally and internally, with many irons in the fire (my words, not his) to find ways to maintain its "sustained differentiation."
When First racks up other technologies against its own cadmium-telluride thin film PV, it does so from a really long-term (as in 50-100 years) perspective and a tech’s ability to scale in a prodigious yet affordable way. "We don’t think that anything that’s over a dollar per watt in terms of capex is going to have substantial legs in terms of its ability to go to scale," said Eaglesham.
As with anything to do with First Solar, cost is also paramount when it comes to technology evaluation. Unless a prospective improvement in the device, the process, the materials set, or whatever other bucket can match up with the company’s aggressive cost-per-watt model, it won’t make the cut.
"If adding the materials will improve the cost per watt or cost per kilowatt-hour that we produce, then it’s something we’re very interested in," Sohn told me when I interviewed him in late April. "If adding an exotic material will improve the wattage but the cost of doing so doesn’t compensate appropriately, then it’s not something we’re interested in."
"But of course we have to evaluate that at scale," he continued. "Frequently that question might be asked and the answer to that question might be very different at 20 MW than the answer at 1 GW. We constantly assess that and run those models, [and ask], How much will this cost at scale?"
When Eaglesham and First Solar frame the technology horse race, it’s always about how the other ponies do when they match up with the strengths of CdTe, now and into the future. Noting how most employees think of themselves as working in thin film PV company first, and then CdTe, he went over the shared commonalities among TFPV technologies—lower bill of materials and capex than crystalline silicon, better efficiencies in low light (but lower efficiencies in good sun as well), etc.—and then cited a couple of CdTe’s advantages, such as its sweet-spot band-gap in terms of the solar spectrum as well as its ability to be processed at a very fast film deposition rate.
For those who wonder whether First Solar might be getting into copper indium gallium (di)selenide (CIGS) or other thin-film or nano-thin film PVs, Eaglesham provided a bit of insight into how the company scrutinizes them. "When we look at how we rack up against other thin-film technologies and when we assess other thin-film technologies that we’re interested in getting into ourselves, one of the primary issues that you have to face is that it is very difficult to find other semiconductors that you can deposit anywhere near as fast [as CdTe]."
Apparently First Solar believes CIGS has some fast-deposition and scaleability potential. When I asked Sohn about this during our interview, he would not confirm or deny whether the company is involved in R&D on that or any other non-CdTe thin film. But several anonymous sources told me during the recent IEEE PV Specialists Conference that it's an "open secret" that First has a small team, led by former Solyndra chief scientist Marcus Beck, working on CIGS at an undisclosed California location.
But when it comes to another area of possible R&D focus at First Solar--multijunction devices--Eaglesham noted during his presentation that the company is "not that wild" about any such technology for the following reason: "The second junction has to deliver at a cost per watt that makes sense for the incremental power [gain]. Let’s say my first junction is delivering 11%, my second junction gives me another 3%. Well I gotta figure out how to pay for that second junction at 3% incremental efficiency. I gotta figure out how to manufacture that device at a cost that makes sense."
"At First Solar, every time we’ve gone through this, if you had a device that made sense as a second device, in a multijunction, it would always economically make more sense to only make that device," he concluded, since it’s such a challenge to make multijunction approaches work out at the cost per watt required at the module level under the company's strict metrics.
So two or more junctions don't necessarily stack up better than one, in First Solar's cost-conscious estimation.
In the final part of this two-part blog-column, I will take a look at First Solar's conversion efficiency history and current efforts, including some of the technological pathways the company is exploring to push efficiencies ever closer to the theoretical limits. Click here to go to part two.
Cart
There appears to be nothing in your cart!
CA: You paint a very bleak picture, my friend, but unfortunately it is quite true and is supported by similar historical developments. Nevertheless there is one major difference this time; the US “price floors” are supported by unprecedented rivers of money, compliments of Obama’s financial and energy packages. So I predict that the US energy frenzy will continue much longer this time, but will die off very quickly when the rivers dry off, taking a number of solar companies with it. EU and China, however, will mostly continue the solar and wind power build-up mostly of desperation, since they have no other energy resources. But on a personal level, while I like talking to you and don’t mind you sharing your knowledge with the masses, I need to talk to you privately (business, you know), so I’ll be patiently awaiting your email or phone call. Thank you in advance.
ablazev - thanks again for your comments. since the article is about economics (cost per watt) I will offer yet another comment in that regard. i agree that demand "will" outpace supply then reverse with an over supply outpacing demand. fundamental economics. what is important here is to understand there is a price floor which will exisist despite the "demands" or "supply" in the industry. investors and institutional financing (banks) have minimum thersholds for returns. more importantly you hit the proverbial nail on the head - there is a lack of funding in the renewable marketplace. the proejcts I permit (and am aware of in the pipeline) average $1 billion in development costs. it follows that "demand" for product will never materialize unless there is project finance to start with (beyond permitting costs). what you will see is 90% of the proposed, large utility scale proejcts being stranded with no funds, transmission, or permitting (due to financing exhaust). the implications for many manufacturers is BK. they are all expanding production capacity with the hope of someone showing up with capacity to buy large quantities 100 mw - 1 GW projects. when buyers do not show up it will be a disaster for manufacturers. they operate with significant leverage that cannot be serviced. my earlier comment about CEQA and NEPA are meant to address the real timing issues that I personally deal with in terms of environmental permitting process. it takes years. although i'm happy to trade knowledge with you i much prefer to give it away to the general masses. hopefully others will do the same. there are enough hurdles to this business already. lastly, to address the origninal comment about tracking systems: they are the most economical project choice. this is demonstrated in real life by the SEGS facilities in so cal. which i have personally toured, several times. although they are solar thermal the economics of land use, additional cost of tracking system, comapred with added output have already been examined. i hope my comments have been useful to the other readers. talk with you later.
CA: Education is not free, my friend, so it will cost you. Or wait, how about trading services? We here can use some development and finance (starting with finance) education and hands-on help. When do we start? Just one last comment on the "bottom line". While you are right in saying that it is the starting point--which is the main reason why the US solar industry was dormant for 30+ years after a false start in then 70s--lately things have been changing quickly. Demand for solar in the US, EU and China has increased dramatically, thus creating a large and growing gap in the supply chain. So pretty soon the supply and demand equation will shift heavily in favor of the suppliers, which will alter the "bottom line" logic in a way, and to a degree that we have never seen before. The rest of the story after we agree on the services trading deal.
Keefwivanef: Vandalism is a problem with any PV installation, and this is where CPV trackers have a huge advantage as well, for: a.) there are no useful parts in these to tempt the average thief or home PV-enthusiast, b.) with the lowest point of the tracker at 10-12' above the ground, vandals have to get very creative, so most would look for an easier target, and c.) just in case, we rely on our friendly insurance agent, 'cause you just never know. So, look for the rows of CPV trackers popping up along the banks of your neighborhood canals in the near future. But seriously, this is not going to happen any time soon, mostly because of complex political and logistics problems.
ablazev - thanks for your comments. one last comment in this forum then i will contact you off line. perhaps you can educate me about pv efficiency. my background is in development and finance. i start with the bottom line profit (commanded by threshold requirement of investors min IRR) then back into the generating system that fits. agreed efficiency and life cycle are different thats why I said need to compare. higher up front cost/ higher output vs lower cost / lower output? this effects IRR. also factor in ITC accelerated under section 1101. not sure what factors you speak of other than replacing inverters at year 11 and 20. economy of scale at 20 MW (1 plant operator, no washing, manufacturer garunteed). although contribution to NPV and IRR is low for years 25-30+, it is still relevant - thus life cylce cost is where the focus is. PPAs are awared based upon NPV of payment structure thus start with life cycle cost - per MW...then back into the product selection, based upon efficiency and comparative cost per MWh output. i would be interested to see comparison between additional land cost to facilitate tracker vs additional output. land cost is relatively low part of total cost. i can't image business decision to forgo tracking system. i welcome comments. i'm here to learn and share knowledge. license solar thermal but am exposed to large scale pv projects also. working on several 250 MW projects in CA. personal view - pv is the future. key lies in CEQA, NEPA and long term cost trends.
10,000 miles of canals. CPV collectors all along them. Terrific. You don't think theft or vandalism might be a slight problem?
keefwivanef: yes, the canal row is an extreme case, but a feasible one. The point is that CPV is very flexible as far as land utilization is concerned. There are many reasons for that, but I'm afraid we are in the wrong place for such discussion.
Mr Ablazev, Can you please clarify that for me. A single string of collectors eg along a canal is a special case. If the string runs North South then I can see that a tracked panel would be about 30% more efficient than a fixed panel ( either CPV or conventional). In the case of a multi-row field (normal for solar farms) the checkerboard layout leaves a large area of wasted space between trackers. Every photograph of a tracking solar farm shows this situation quite clearly. (Not, trying to pick an argument, trying to find the truth)
CA: Efficiency and life cycle cost are different variables with their own meaning. The starting point for evaluating the performance of a PV system is its efficiency (just like GPM rating on your new car's sticker). Life cycle cost will be calculated later on--again on the basis of the system efficiency--but also taking into consideration a large number of other variables. Hiding the efficiency and mixing it up with all these other numbers is a misleading practice, which will hurt the reputation of the US solar industry. I, however, think that we should take this discussion offline, because I want to discuss your other points, but they are too wide and deep for this forum. I also need to find more about your large scale projects licensing, so please email me at ablazev@sst-usa.net. Thanks.
interesting. efficiency must be weighed against life cycle cost. PV is up against concentrated solar thermal which has a life cycle of 30-40 years. PV degrades much faster and has max life cycle of 25 years. coverage is approx. 50% of land for both. "output" is what needs to be taked into account when considering land use efficiency. trackers are preferred. run a cash flow model and it will become apparent. i license these large scale projects for a living. I've seen the project design; output and cost data which leads me to these conclusions. good stuff.
In response to Keefwivanef's challenge: yes, under normal circumstances CPV trackers use approx. 2/3 of the land used by flat panels. CPV, however, is much more flexible. I.e., CPV trackers installed on the bank of an irrigation canal (and we have approx. 10,000 miles of these in the Western States), or on the North edge of an agricultural field will need practically zero useful land. And there are many scenarios in between as well. In all cases, CPV trackers can be arranged so that to use less land than any of the existing solar technologies.
"if you had a device that made sense as a second device, in a multijunction, it would always economically make more sense to only make that device," - this logic makes perfect sense, unless they have a way of producing a effective multi-junction device with the extra layers built into the manufacture of the original device and were able to cherrypick units that "made it" but sell duds as standard product.
I think Daniel is parroting the line that CPV produces more power per acre than flat panels. This is part of the hype around CPV and it is patently false. Trackers are essential to CPV and spacing is required between arrays to prevent shadowing. The solar collection area will be only 20% or less of the total land area. Go ahead and prove me wrong...I'll be delighted.
Amazingly enough, with all the techno talk and loud promises, there is a little, if any, mention about the efficiency of First Solar's panels, which is the lowest in the industry. And which also means that the cost per unit area of land is the highest, if not prohibitively high for any significant power generation purposes. Is this just a blind ignorance, a cover up, or am I missing something here?
Manufacturers seldom talk about the cost per unit area of land over a 25 year timescale. That is understandable, the production line is their emphasis. The installed cost in KwH per acre/hectare over a 25 year timescale might be a useful cost figure. In that context, a multijunction device might reduce mounting structure costs, might last a few years longer, or respond across the spectrum with a better fill ratio.