143 replies to this topic

[odyd]

http://solarpvinvest...china-report-22

[odyd]

HI there article from Solarplaza on Mexico and conference there
http://solarpvinvest...ture-for-mexico
those who have a chance to go can register here
http://www.elfuturosolar.mx/registro/ using this code EFSMX13-SOLARPVINVESTOR for 10%

[odyd]

A bit of the summary on Taiwan
http://solarpvinvest...ipments-in-2012

[odyd]

http://solarpvinvest...ces-to-increase

[explo]

"At present, the processing cost of poly-c-Si wafer is about 50% of the
total wafer cost, and that of mono is 65%. The price for solar wafers is
yet to see improvement as a result of the polysilicon increase."

This implies mono wafer processing cost is 85% higher than multi. So far wafer makers have been able to push this 8 cents per watt difference to the price, since the higher conversion efficiency of mono added corresponding value, but with high performance multi wafers now enabling same module power ratings as standard mono, but at the same low processing cost as standard multi wafers, standard multi and mono do not bring value anymore.

I think this changes the game and it will be interesting to see how it plays out. Standard multi will go away I guess, unless there's someway to reduce cost for standard quality only. Mono need to increase conversion efficiency to undo the closed gap or CZ cannot compete with multi casting.

I'm still surprised to not see more discussion around this (not just here, in general).

[odyd]

This implies mono wafer processing cost is 85% higher than multi.

you mean 30%?

[explo]

No 85%. On processing cost only. Example: if multi is at 10 cents, mono is at 18.5 cents.

[odyd]

I think including poly the difference is about 50%, in costs. You are quotting Q3 SOL processing costs yes?

[explo]

I think including poly the difference is about 50%, in costs. You are quotting Q3 SOL processing costs yes?

Yes including the polysilicon it's half the percentage difference compared to wafer step only (42.5% total wafer cost diff vs 85% wafer processing cost diff) between mono and multi. No not quoting SOL, just based on the quote from Jason. It said polysilicon is 50% of the cost for multi wafers and 35% for mono wafers. If polysilicon cost of 10 cents is plugged into that relation, then multi wafers costs 20 cents and mono wafers 28.5 cents, i.e. 10 + 10 multi vs 10 + 18.5 mono for the poly + wafer pc costs. This was my example of what that quote translates too.

As prices and cost have come down the cost difference between converting polysilicon to a mono wafer instead of a multi wafer is approaching 100% on percentage basis although in absolute cents it is around same, while the lower costs of wafer to module conversion and BOS reduces the value of the conversion efficiency gap and finally while that conversion efficiency gap has almost closed (at module level). All this spells mono trouble, relative cost penalty is approaching 100% in the wafering step and the relative added value is approaching 0%. The shift from p-type to n-type mono might have to happen even faster than predicted to prevent mono market-share loss to accelerate.

[explo]

We see more and more squared shaped p-type mono wafers. This is one way to retain more of the cell efficiency in the module level (it can add close to 5w to a 60x6 module), but intuitively this should cost more. I don't know if this is the case, but an obvious effect from pure geometric reasons is that the squaring loss to waste or recycle increases a lot. At least lowered yield and throughput should result. I think this is just a somewhat desparate way to maintain the value diff perception between mono and multi, but probably does little to improve the value per cost issue for mono. Besides geometry, oxygen concentration causing light induces degradation after initial weeks of module use and higher CTM loss are other adversities for mono vs multi in value at module level. Final nail in coffin might be that the new high efficiency multi technology (like Virtus II) turn things completely around on the traditional mono strengths on the low light conditions conversion efficiency loss and high temperature loss. Virtus II properties beats even Panda properties with wide margin here, and Panda has a specific flyer touting these proporties for effect on the end of the day produced kWh. If Virtus II beats Panda by large margin, which beats p-type mono which beats standard multi, then there's little hope for p-type mono. All these properties affecting end of the day kWh produced can be hard to evaluate, but for those who state the NOCT power in their datasheets this is where you get the indication, but NOCT is still ideal normal uses conditions (noon on a clear day) and the diff here is ampliefied on a cloudy day or late afternoon on a rooftop in Germany or on a optimal tracked plant in a hot desert.

So in summary mono used to have two module selling points compared to multi

• Much more Wp
• More kWh per Wp

This could easily motivate the 8-10 cents higher price per watt for mono wafers to cover the 8-10 cents higher production cost, especially when there was a lot BOS per kWh to be saved with those two points when BOS costs were high. With multi tech like Virtus II both points above are muted. The hope for mono is that Virtus II like tech capacity remains much lower than total PV demand.

[explo]

Panda flyer:
http://www.yinglisol...Benefits_GB.pdf

Panda datasheet:
http://www.yinglisol...l_Series_EN.pdf
260w average STC, 189w average NOCT, 3.5% low light loss, 0.42%/℃ high temperature loss

Standard multi (YGE) datasheet:
http://www.yinglisol...l_Series_EN.pdf
240w average STC, 174w average NOCT, 5% low light loss, 0.45%/℃ high temperature loss

Virtus II datasheet:
www.renesola.com/File/download/id/409?t=en
255w average STC, 189w average NOCT, 1.25% low light loss, 0.40%/℃ high temperature loss

The kWh values are in the datasheet details. Virtus II production cost is the same as YGE standard multi, but the kWh production value is the same as Panda.

Virtus II preso:
www.renesola.com/File/download/id/213?t=en

[odyd]

Honey is right up there, they make wafer themselves, although I think that GCL wafers could be used for the same. JA and now YGE can produce similar results of HP wafers. PANDA has been notoriously achieving more than their datasheets show. Due to bifacial structure, each panel can get as much as 15W more on the top of the rating due to reflection effect of the roof etc. We beat this to death. The range concentration, average, whether is 255W, 260W, 265W etc is really an enigma today. I wish this statistic was published somewhere, but it is not. PANDA was the second best panel rated just last week or so Solarwatt had the best, the company I think is semi-bankrupt. At any rate. I think that selling more watts (due to efficiency) at less logistical costs (shipping, storing etc) should benefit. Producing more efficiency should add to the savings, if processing has the same cost. So if the cost per watt, versus ASP per watt, has improved gross margin, then this is the winner, somewhat regardless of efficiency at the end or rather inclusively.
I am looking forward to see what margins will be in Q4 and what processing costs will be then. Yingli's costs are really in silicon, Trina did pretty good job getting poly at $0.12 level.
It looks like last week prices for wafers and cells got the lift, while modules were actually selling off, for the first time in three weeks.

[explo]

Odyd, like Klothilde says I was only looking at Jason's comment on cost difference between multi and mono wafer processing. I'm not considering cell tech here and looking at Trina and Jaso thus becomes difficult, since they partially outsource wafer production and their focus is more on the cell tech.

The questions I posed were:

• Will the growing market share of multi wafers relative to mono wafers seen past two years accelerate?
• Will HP multi grow quickly within multi and will standard multi soon be gone?
• Will the emerging growth seen and strong growth projected for n-type vs p-type within mono accelerate as p-type costs too much compared to limited relative value it can add compared to HP multi?

Klothilde, SOL is focused on high R&D spending and properitary technology. GCL is the multi wafer giant and have a similar approach. LDK is another big multi wafer supplier, but depends more on partner technologies (e.g. GTAT). SOL has a patent portfolio, but generally I think it is hard to have success with a strategy depending on first discovery and then putting up fences around the find. You have to continuously be at the forefront of technology (not just idea and prototype, but implemented technology platform with good scalability properties for mass production economics) so that you keep having the best commercial value/price offer to bring to the market. Right now I would say that SOL is in this position for multi wafers and maybe for wafers generally until n-type mono has matured more. As the market sees the availability of this new value how will it adapt and what will that reshaped form look like in 2015? This could be one way to drive wafer capacity rationalisation, a lot might be forced out if it cannot keep up or adapt with the new HP multi.

[odyd]

Hi Both,
My original reply was to Klothilde, I quoted those companies i believe they posses the technology internally or are able to purchase the wafer externally. The article I had written on HP was based on conversations with insider who works for one of the companies in waferring business. SOL`s technology is not unique I was told, and because of it I do not believe this is just SOL`s niche. Clearly concept of HP, or moving to HP is being a focus of many. However, Chinese market in particular but also global commercial installations are continue to use lower or average efficiency modules. I doubt that mono will disappear. In fact mono inventory has been reduced dramatically, probably because of Japan demand. I am trying hard to talk to ECN about their visit in China , and inquires made by LDK about producing n-type wafers. LDK has also its HP wafers. So is GCL or they claim to have it.
I think that there is also one more factor. Mono efficiency has been long-time proven, HP wafers well have not. Adoption could be a barrier and buyers may simply prefer mono because of it. SOL is the best to monitor this.

[explo]

Mono efficiency has been long-time proven, HP wafers well have not. Adoption could be a barrier and buyers may simply prefer mono because of it. SOL is the best to monitor this.

Can you clarify this? Why is multi not proven? It has higher market-share than mono. Increases in multi efficiency should be viewed in the same way as increases in mono efficiencies, right? Wouldn't a shift from p-type to n-type be a bigger risk, since p-type has dominated last 30 years?

[odyd]

Explo, HP concept is a new concept? Can you please understand my statements for what they are instead of assuming what they may refer to?
I said HP wafers do not have a history of efficiency nor the modules made of HP do. This is not the question about multicrystalline family of products but another sub-category. Do you think that purchasers are that much interested in taking on the risk, rather than buy something what is familiar and known to have high efficiency? There is always a group which will buy a new product offering advantage, but there is also a group, which will pay more to get the sure thing. If quasi-mono showed as anything, it was basically a "no go". Companies first could not stop talking about it, now they had dismissed its value. Again from perspective of a buyer, nobody wanted this to a point it matters to a supply chain. N-type mono cells have been used in space for the longest time. I am not here to say that n-type is going to be a next big thing, nor I say that HP will not become a mainstream pushing away legacy type multi-wafer.
I am only saying that adoption curve is something one needs to consider, despite high specs. As time progresses I have no doubt, when it comes to efficiency upgrades, what companies will do next will tell us what they see value in.

[explo]

I still don't understand this point fully. Quasi-mono was a different case so let's focus on the HP multi. HP is just short for high performance, right? So it just means it is multi wafers with higher conversion efficiency than current industry standard.

Now that YGE announced "significant progress in ingot casting" and multi was close to 90% of their shipment in 2012, will this mean that they can't use this to offer 250w multi panels instead of 240w due to a high adoption risk?

[odyd]

Quasi-mono was a way to get the mono qualities using the multicrystalline process. HP involves sorting existing wafers and upgrades involve ingot production, by way of furnace upgrades including graphite cores etc, to produce anything beyond 17.3%. Statement one: No matter what specs you will print on the multi-module some people will choose mono module, anyway. If you take the Wafer II module and you put PANDA beside each other, some will still pick PANDA. They make a choice because there is a huge literature support mono product, and beside the blog and datasheet from ReneSola very little on HP is being said. The spec is also new, Explo, and such it is viewed as novelty.
So you ask theoretical question and I am giving answers which are very theoretical as well. If I assume that SOL represents HP revolution, I can tell that their increase of selling Virtus II modules will be the way to benchmark the adoption. If they happen to expand their current and add more capacity of virtus modules, this is going to show that HP wafer technology will be pushing out legacy process. They do sell other modules now, so there must be a reason for it. Many believe that efficiency has very little to do with solar markets at large today. So if even the process is cheaper or equally cheap to legacy, spending on upgrading of existing processes may not be of interest, if you are selling other modules. This goes for mono and so on. Imho, I believe that there is no other way but efficiency to separate winners from losers. As an investor I see pursuit of cost reduction as something which offers advantage to a company, but I understand from the perspective of achieving success in this pursuit, the new technology, may not have an immediate advantage to the adoption curve. This is what brings me to your comment on Yingli. So they have the way of getting a 17.75% multi ingot. Based on the demand for their products they may not have a need to produce the ingot this way, and save $30M in upgrades. On the other hand this is the only things which is selling, I am sure they will do it. I am not sure how to describe this, demand is separated from cost savings. As a consumer you do not care what manufacturer paid to make it. For manufacurer if you happen to produce products which are in both areas, they are cheap to produce and they get bought, then you have a success.