37 replies to this topic

[joshchang]

I think using production cost to calculate existing capacity is fair since all past capital investment are sunk cost. Whether to start producing poly strictly depends on the spot price and their cash cost to produce.

For new players who want to build new poly plant, they need to look at depreciation/production cost to determine if it's a good investment.

IMHO we won't see $50 poly since we have enough capacity to produce at low cost and new entrant has even lower production cost.

[eysteinh]

Well I am not considering 50$/kg in any scenario currently so count me in on that. I am more a fan of 23$/kg, that is a very small increase in module costs.

What I am saying is costs like operational costs (adminstration, general and sales) should be included. This is stuff like shipping, packaging, insurance, sales etc.

One could also consider adding interest costs and maintainance costs to any calculations.

While for example REC has a great cash cost and production cost once you really add absolutly all costs you are looking at more than 20$/kg. Same can be said for the simens producers.

[littleguyintucson]

We should see price increase year end if demand hits 39+GW this year. If we are lower it could be well into 2014

Pops , Odyd, Explo

Can any of you clarify for me when they speak of GW are they talking global installed capacity? When rating and reporting power plants and projects this is almost always AC rated. The AC rated plants are around 80% of the actual DC rated wattage. Canadian Solar for example used some 13MW of DC rated modules for their 10MW project.

If they are stating global demand of 36-40GW for 2013 based on projects then DC shipments would be closer to 45GW. At a 45GW AC target is actually 56.25GW DC rated modules estimated for 2014.

I believe there is a disconnect in reporting as when installed capacity is reported it is generally as the power output AC rated. When companies report shipments it is always DC rated.

If correct, then the supply costs for Poly will increase, likely around 40GW AC capacity as that is 50GW of DC rating.

My expecation is a climb to over $22/KG.

[littleguyintucson]

The reason I bring this up is that DC rated demand in 2013 is likely closer to 46GW and a target for 2014 of 45GW AC rated placed DC demand at 56GW or a 10GW increase.

Off the cuff using (tsl, yge,sol,jks,jaso,ldkLostCapacity, stp, hsol, csun, csiq) I get a capacity of around 17GW. This is 25% of the total Global shipments. If the market segment is to grow 10GW and the pattern of picking up 50% of new capacity continues, the major US listed companies pick up 5GW of shipments in 2014 or a growth rate of 30%.

If one looks at capacity utilization rates most all the US listed solars need to start to expand in the later half of 2013. Currently these companies are bleeding red ink. They need to likely expand to lower the ratio of debt and opex ratio per Watt shipped

At 56GW, Poly will be above $20/kg and headed towards $25. This is going to add up to probably $0.02 in costs. That is a reduction of nearly 3% on gross margins with an ASP at $0.67.

[eysteinh]

Good point about the dc ac debate. Ironically I have not even tought about this and I feel ashamed for not seeing the obvious. Yeah the numbers are AC. This strenghtens the case even more for a meeting of supply and demand in 2013 and prices increasing to around 22-23$/kg (23$/kg is the us cost and china have around 1$/kg less mgs costs so I think 22$-23$/kg is a good estimate.) I base my numbers on both photon internatonal, installation numbers from offical FITS i read and also discussions here on the forums and other analytical estimates. I know photon international is AC because they explicitly write if the numbers are AC or DC.

I am not sure if the conversion rate is 20% less DC - anyone who knows the math on this would be very welcome! Granted many are not 1:1 And you are probably correct it is less DC to AC.

For example REC are at about 1:1 in STC dc (sometimes even more output than nameplate capacity) but I am unsure if the output number is in AC watt. I assume so since its connected to inverters and inverter output AC current. So since they measure the output current this should be an AC number.
http://www.recgroup....?epslanguage=en

Also what some people seem to forget is that the manufactures can go down to 160 um in wafer thickness. (ofcourse also lower but 160 um is probably the next thing when it comes to the sweetspot between polysilicon usage and breakage ratio of the wafers.) They are mostly using 180 um because breakage ratio is bigger at 160 um and polysilicon is simply dirt cheap now. I know REC use 180 UM currently even if they have no problem using 160 UM. So if prices increases there is a bit elasticity here also - yes demand will be down a bit 11% but so will cost. If the demand outstrip this reduction in poly usage then the cost will not have been transfered..both parties win..the poly is sold more expensive, but the module manufactures output more so the cost stays the same for them.

[littleguyintucson]

Also what some people seem to forget is that the manufactures can go down to 160 um in wafer thickness. (ofcourse also lower but 160 um is probably the next thing when it comes to the sweetspot between polysilicon usage and breakage ratio of the wafers.

80% is a good general reference as STC is not optimal. I also factor in degredation of 0.5% over the lifespan and inverter loss of 5%, cable drop and thermal issues. If you read the CSIQ Q1 quarterly report, there was a line of questioning at $60M for a 10MW plant came out to $6/watt. The company stated the plant was 12-13MW DC and it comes out to about $5/watt DC rated. So they are using roughly 70-80% in efficiency conversions in sizing the system

As for Si Reduction,

Here is a nice thread from Yahoo. I posted a calculation to estimate grams per watt based on ingots and sawing and wafer efficiency.

http://finance.yahoo...tc=mb-tab-topic

I put in a spreadsheet the grams per watt formula based on ingots and wafering parameters. Using this knowing the dimensions of the bricks and wafers and wire saws you can estimate grams per watt.

Row / Col A / Col B
1 / KG / 450
2 / BRICKS / 25
3 / Brick kg / =sum(b1/b2)
4 / Length / 250
5 / Wire microns / 100
6 / Abrasive overhead microns / 20
7 / Wafer thickness / 180
8 / Total Thickness / =sum(b5:b7)
9 / Wafers per brick / =sum(b4*1000)/b8
10 / Grams per wafer / =trunk((b3*1000)/b9)
11 / 3.9 / =sum(b10/a11)
12 / 4.1 / =sum(b10/a12)
13 / 4.4 / =sum(b10/a13)




kg 520
bricks 25
weight kg 20.8
length mm 265
wire microns 100
abrassive 20
wafer Thickness 180
Total Thickness 300
wafers/brick 883
grams per wafer 23.55605889
3.9 6.0400151
4.1 5.745380217
4.4 5.353649748

change the wire size to 80 microns and you get a different ratio.

kg 450 520 800
bricks 25 25 36
weight kg 18 20.8 22.22222222
length mm 220 265 295
wire microns 80 80 80
abrassive 20 20 20
wafer Thickness 180 180 180
Total Thickness 280 280 280
wafers/brick 785 946 1053
grams per wafer 22.92993631 21.98731501 21.10372481
3.9 5.879470848 5.63777308 5.411211489
4.1 5.592667392 5.362759759 5.147249953
4.4 5.21134916 4.997117048 4.796301093 Le

[eysteinh]

Thank you littleguyintucson this is the sort of stuff I learn from

I see you suggest reducing wire thickness. This causes less tencil strenght and makes throughput lower because tension has to be less and this causes more brittle wafers and will cause more breakage ratio if throughput is at same speeds as with higher wafer tencil strenght. (27N for 140 um wire and 22N for 120 um wire. I dont know for a 100 UM and 80 um as I am basing my numbers on a scientific article from 2011, The solar wafering process, Mark Schumann et al Fraunhofer Insitute for Solar Energy Systems (ISE) ) This is also another way of lowering polysilicon consumption and can be done if polysilicon prices increases. Just as the way I suggested lowering wafer thickness to 160 um.

I would add that anyone reviewing the numbers should also keep in mind the breakage ratio (0.1-0.3% perhaps in modern production) when calculating total cost per watt in gram polysilicon.

I think the major point readers should keep in mind is that you can adjust the production process and use less polysilicon if prices increases. This elasticity means that if demand outlast supply then the supply can be stretched out over more wafers. Basically this means both that producers can produce more and that the polysilicon can be sold at higher prices while not adding as much cost to the producer as would be the case at 5.8 gram per watt.

For example if let say on average after breakage ratio is factored in use around 5.8 gram per watt for 20$/kg that is 11,6 cent per watt. But then prices increase to 23 $/kg and the producers implement measurements and lower to 5.6 grams per watt, that is now an increase to 12,9 cents per watt, 1.3 cents more per watt than with 20$/kg prices. Now breakage ratio might be higher because of more brittle wafers, but overall they profit from this than instead paying 5.8 gram 13,3 cents per watt (1.7 cents more than at 20$/kg.)

Also just for some data: In 2012 may issue of photon international there was a market survey of current production equipment for wafers. All where producing at 120-130 um and grit size 8-9,5 (size of the grains used to actually cut the wafer, they are spread on the wafer wire with oil and erode down the wafer) in mass production. With 120 um wire size your looking at 5,9 5,6 and 5,2 (3,9 Watt per wafer, 4,1 Watt per wafer and 4,4 watt per wafer) But all the major producers (meyr burge, advanced materials, rena etc) had options for 100 um wire diameter. So I think you perhaps are correct now in 2013 we are down at those sizes. And that is like 0.37 gram per watt savings at 3.9 watt per wafer. (allmost a cent saved.)

[explo]

To me it would be a surprise if they are not rating PV panel demand projections in GWp DC. It would be good news though.

[eysteinh]

Well since the very job of an inverter in a solar system is to convert DC to AC i would assume numbers are in AC for the finished installation.

I sent an email to the german ministry who do the counting for german installations asking them exactly this. Wonder if I will get a reply

http://www.bundesnetzagentur.de/

[explo]

Well no wonder people see chronic supply/demand imbalance if the demand is counted in AC and the supply in DC (adding 25% to the AC supply).

[eysteinh]

Well we dont know yet. I know photon international report in AC because they write so. But is this the case for offical installation statistics like from germany italy etc? Thats why I sent the email and I am curious of the answer if I do get one. Perhaps someone who is german speaking could try to email them?

[eysteinh]

http://www.nrel.gov/...parameters.html
" Overall DC-to-AC derate factor 0.77"

http://solarexpert.c...rmance-factors/

Many factors. Inverter and cables loss are two major causes.

But I would not say 25. I asked earlier in this post if anyone knew any good number for this. Personally I have read 15%.

One factor against the numbers being reported in AC is that this is less comparable number since the different conditions make different AC numbers. So I really dont know yet if installations are in AC or DC. I am perhaps leaning towards DC. I really prefer how for example photon international actually state clearly if it is AC or DC watt so the reader can know what is going on. From that source I allready know that some installations are reported in AC numbers, while others are reported in DC...

[littleguyintucson]

Why is it 25% less? Inverter efficiencies are close to 99.5% now for the good ones. So where does the 25% drop off comes from?

As a tester(not in Solar) tests are designed to show functionality at optimal performance.

The standard STC testing was designed to do just that. Optimal thermal temperature, Optimal light angles and light frequency. Real world has varying temperatures, varying angles of light depending on locations and seasons and different color spectrums that impacts the overall performance negatively.

Cabling can cause power loss of up to 3% when designed within codes depending on run lengths amperage driven and conduit fill factors that create heat within the wires. That is why High Voltage is desired as low amperage is pushed.

A Flat mounted system vs a system tilted south southwest at a 30 degree pitch can create a 10-15% variance in power output on average. On peak even more.

Temperature in the mid 90's can cause a 15-20% power loss vs a 70 degree weather. Units mounted on black shingled roofs with 3 to 6 inch clearance operate in higher temperatures that ground mount systems.


These are many of the reasons why a module rated at 250W STC DC rated will only generate some 200W of actual peak power.

[eysteinh]

Here is an example of a project reported in AC watt.
http://www.adb.org/s...-01-tha-iee.pdf

"The Solarco Solar project has total production capacity of
9.5 x 6 = 57 MW-AC, which will be sold to Provincial Electricity Authority or PEA under a very small
power producer (VSPP) contract."

Here is it in the news:
http://www.adb.org/n...entral-thailand

If you look at page 10 you see the exact DC installation numbers. It adds up to 72,28 MW.

The same news in REC reported in DC watt:
http://www.recgroup....totaling-72-MW/

But yes I am also wondering how the offical records in for example FIT's are calculated. DC or AC. Thats why I sent an email asking about this.

[redsolar]

25$/kg generates a lot of graavy for all the big players...and they can easily scale the capacity up... if warranted.

Low 20's may be a sustainable number I believe.