Thursday, April 30, 2009

FSLR; The Betamax of Solar?

FSLR announced their earnings today. The results were great, particularly considering the overall economy.

I've given FSLR considerable thought in the last couple years, and I remain convinced that they have unspeakable future problems. On their investor relations page, they link to the pdf associated with their Q1 conference call. In it, they mention what they consider to be risks to their business, but nowhere do they mention the risk associated with availability of their critical Tellurium supply. Ok, so maybe they have it all figured out; but nobody's asking, and nobody's telling.

Ok, I don't know, but I want to get an idea of what kind of supply issue they're up against, so I've gathered some info.

Per Greentech Media, FSLR uses 6 grams of Tellurium per square meter. (See.)

Per First Solar, the FS-277 Module is .72m2 and has a peak power of 77.5W. (See.)

77.5W / .72m2 = 107.64W/m2

So, at 6 grams / m2, the amount of Tellurium required per Watt works out to be 6g / 107.64W = .056 g/W.

Well, we know that FSLR is aiming for a bit over a GW in annual production for '09 and '10, so rounding to 1GW gives roughly 55.7 Metric Tons of Tellurium required to produce that GW of modules.

The question, then, is how much Tellurium is out there, and what does it cost?

According to the USGS, the price has ranged from $41,800/MT in 2004 to
$82,000/MT in 2007. The World Supply of Tellurium according to US Geological Survey was 132MT in 2006.

Ah, no problem. If they're using 55MT to produce 1W worth of modules, and they're paying even the high price of $82,000/MT for their supply, then they're only paying a total of $4.5 Million for their entire yearly supply of Tellurium. That's less than a penny per Watt. In fact, during the CC, Jesse Peechel stated, quite possibly accurately, that First Solar's largest cost was glass.

Wait, a problem. Solar is big. A sensible look at the required future scale of Solar Energy puts the annual Global installation rate to be around 30GWp per year by just 2012. What if FSLR wants to maintain a significant share in this market?

Well, as it is today, it appears that over a third of the World's Tellurium supply is required for the production of a single Gigawatt of First Solar modules.
If FSLR were to take 10% of that market, they'd have to produce 3GW of modules, which by today's efficiencies would require 165MT of Tellurium, or more Tellurium than the World produced in 2006! Well, maybe the price of Tellurium is a pittance when the company is demanding only a third of the World supply of material, but I can guarantee that it won't remain so when that company is demanding 33MT MORE than the World's annual supply.

A big part of this problem is that there's no such thing as a Tellurium mine. Tellurium is only produced as a byproduct of mining other commodities, such as Copper. This means that it's very difficult to increase the World Supply independently of the supply of those other materials. If you were to mine Tellurium alone, the cost would be astronomical, and yet if you were to drive up the mining activity in Tellurium's sister elements, then you'd have the affect of driving down the prices of those materials, thus making them into less desirable targets for mining.

What about efficiency gains? Sure, if FSLR is able to pull off a tripling, or even just a doubling of their efficiency, then they could make do with dramatically less material. I can imagine several possible ways that they could do this, but I suspect that it will be a tough path. As it stands, per the CC pdf, FSLR has increased the conversion efficiency of their product by .3% since Q1 of '08. That's simply not going to cut it, particularly if you look out past 2012 when the market gets even larger.

I don't know. They have some very smart people there, and they're working hard in an exciting industry. The particular technology just doesn't seem to stack up to me, though, and like I said, nobody is asking questions and nobody is volunteering answers.

Ah well, in the short term, I'm quite certain that they are going to do great. Wall Street loves them, and they have excellent margins for the time being. They very well might be able to leverage some of that temporary financial advantage in order to open up new technologies to their benefit, so we'll see.

All that said, I'm not short FSLR, and I suspect that to go short FSLR would be a very bad plan.

Also, a final note, it's pretty obvious that I think that the strongest players at this time are out of China, but it's not that I don't like some US Companies. I really like Applied Materials, and Sunpower to name a couple of domestic players.

Tuesday, April 14, 2009

So, you want to buy a solar plant.

A Scenario.

Note: A follow-up scenario includes accounting for system degradation and inverter losses.

The cost of the install + Interest will equal some amount of money to be paid out per month. I'll call this Outgoing$Monthly.

Power generated per month will be sold on the market for some amount of money. I'll call this Incoming$Monthly.

Set Incoming$Monthly = Outgoing$Monthly.

This would be the point at which your investment broke even on a monthly basis (not including maintenance cost at the moment, this is just to include interest expense into the equation). It's not going to be quite right, because of seasonal variation, as mentioned below, but I'm not looking for anything exact, just a rough way to start gauging cost / benefits.

The end result will be a relationship between the Installation Cost per Watt, Interest Rate, and Required Sales Price of Energy produced in order to break even.

I'll skip to the chase, for those that don't want to read through the whole thing.

Cost/kWp ($/Wp) = Rate ($/kWh) * C2/C1

Note that the assumed interest rate (5%) for purposes of this post has been set and absorbed by C1, and the Insolation Ratio has been absorbed into C2.. Other assumptions are pointed out below.

To give an example of what this tries to point out, let's say you can sell the energy produced by the power plant for $.25/kWh (equal to the low range of this estimate of costs for future nuclear power plants).

Cost/kWp ($/Wp) = $.25/kWh * 146 Hours/Year / .0066 = $5,530/kWp, or $5.53/Wp.

So, if you can sell your power for $.25/kWh, then you break even (roughly) if you can complete the installation for $5.53/Wp or less. Note that the equations below DO NOT include the existing 30% Federal Tax Credit for Solar Installation. That's icing (of course, it also doesn't include lifetime performance degradation or inverter losses).

Fact: this is very much in the range of possibility in TODAY's market. Particularly in the case of mid-large scale installations.

The basis follows.

If there's one thing that I've learned being on the Internet this many years, it's that if you're wrong, somebody will point it out. Have at it with my thanks!

Here goes:

First, find the Monthly Payment required to make the loan payment for an installation of some total cost.

(1) Outgoing$Monthly = (Principle * i) / (1 - (1+ i)^-n) See

This is the Monthly Payment on the loan for the power plant with the below assumptions.

Principle = Total Original Loan amount used to finance the entire plant = the Total Peak Power of the plant * the overall Cost per Watt of the system.
i = periodic interest rate (Monthly. Assume 5% APR, so i = .05 / 12 = .0042).
n = total number of payments (Months. Assume 20 Year Loan, so n = 240).

(2) Principle = TotalPeakPower * Cost/Wp

The Principle is the amount of the loan, where the total cost of the installation is given by the Total Peak Power * Cost per Watt. Substituting for "Principle," from (2) into (1) gives:

(3) Outgoing$Monthly = (TotalPeakPower * Cost/Wp * i) / (1 - (1+ i)^-n)

For simplicity, and ease of double-checking results, I'm going to treat n and i as constants (they are part of the assumptions above), and will pull a constant out of the above equation (3):

(4) Set C1 = i / (1 - (1+ i)^-n) and substitute into (3).

(5) Outgoing$Monthly = TotalPeakPower * Cost/Wp * C1

Now, to figure out what's coming in every month on the sale of the Energy.

This doesn't include seasonal variations. On thinking about it, though, in an Energy market where consumers are paying based on momentary supply and demand, wintertime prices could actually go up based on decreased supply, and so help to balance out the annual cycle for the energy supplier. Then, in the summer where supplies were higher, the prices to the consumer would decrease to offset some winter costs.

In any case, following similar logic to my note on Insolation, the Annual Energy output of the plant can be written as below.

(6) Annual Energy (kWh) = TotalPeakPower (kW) * 20% * 8760 Hours/Year * 1 Year

Start by writing down an equation to relate the Installation's Total Peak Power, to it's Annual Energy Output. I'm plugging in an assumption of a 20% Insolation Ratio, which would include a broad swath of non-sunbelt States. The Insolation Ratio Assumption for this post applies to such shady states as Tennessee, Missouri, and even North Dakota.

(7) Incoming$Yearly = Annual Energy (kWh) * Rate ($/kWh)

Multiplying the Annual Energy Output by the Rate at which it sells for, gives the Total Income for the year. Divide by 12 (below) and you have the Average Monthly Income.

(8) Incoming$Monthly = Incoming$Yearly / 12 Months

(9) Set C2 = .2 * 365 * 24 / 12

Once again, I'm going to pull all of the Constants out of the equation (6) to come up with C2.

(10)Incoming$Monthly = TotalPeakPower (kW) * Rate ($/kWh) * C2

Ok, so now we have the Monthly Outlay required for loan payments, and we have the Monthly Income from energy sales.

To break even - let's set them equal to each other.

(11) Set Outgoing$Monthly = Incoming$Monthly

(12) TotalPeakPower (kW) * Cost/kWp * C1 = TotalPeakPower (kW) * Rate ($/kWh) * C2 (Hours/Year)

(13) Cost/kWp ($/kWp) * C1 = Rate ($/kWh) * C2

Canceling out TotalPeakPower (kW) from both sides of the equation, gives a very simple equation relating the Rate at which the energy is sold, to the Cost/kWp of the initial plant installation.


Ok, so to an example and a factcheck.

First, Calculate out C1 and C2.

(14) C1 = i / (1 - (1+ i)^-n) = .0066 (i = .0042, n = 240)

(15) C2 = .20 * 8760 Hours/Year / 12 Months/Year = 146 Hours/Month

Then, pick a target Sale Price for the power that is produced by the Installation, and solve (13) for Cost/kWp. I'm using $.25 in this case, so:

(16) Cost/kWp = Rate * C2/C1 = $.25/kWh * 146 Hours/Month / .0066 = $5,530/kW

Now to check it, or at least check the Interest Calculations:

Since TotalPeakPower was canceled out of the above equation, I'll pick a value to use for the factcheck, say, 1000kW.

So, using (3), Outgoing$Monthly = (TotalPeakPower * Cost/Wp * i) / (1 - (1+ i)^-n) = 1000kW * $5,530/kW * .0042 / (1 - (1+ .0042)^-240) = $36,617/Month.

Then, using (6), Annual Energy (kWh) = TotalPeakPower (kW) * 20% * 8760 Hours/Year * 1 Year = 1,752,000kWh/Year and Dividing by 12 to get a monthly Energy Output, gives 146,000kWh/Month.

Multiplying this by $.25/kWh gives $36,500/Month

Pretty Close. Exponentials are subject to rounding errors. Another way to check would be to put the total cost, or Principle (in this case, $5,530,000) into any number of online mortgage calculators.


Monday, April 6, 2009

Big news out of Applied Materials!

Applied Materials says value of solar order cut.

'Applied Materials Inc (AMAT.O) said on Monday the value of a sales agreement to supply solar production equipment to a private buyer has been slashed to $250 million from $1.9 billion due a worsening global economy.'

This is almost certainly in relation to AMAT's deal with Best Solar out of China. If so, this dramatically reduces the risk to Mr. Peng of default on this deal, for which he put up a large number of his personal shares of LDK as collateral.

Kudos to ibcnu2nite of Yahoo for the find!

Saturday, April 4, 2009

The Market - A Bucketshop.

So, a fellow on Bloomberg was talking about Bucketshops this morning.

We modernised ourselves into this ice age.

Wikipedia on the Bucketshop.

Basically, they were businesses on the sidelines that would play bets with customers on the stock market, but were not actually connected to the stock market. It's as if I were to bet someone $50 on LDK to go up, and vice versa, but neither of us would actually ever trade a share of LDK, and certainly we wouldn't be regulated as if we were actually trading in the market. It's very close to what has happened with Derivatives in the last 10 years. A great many of them, Trillions of Dollars had no fundamental basis in any physical ownership of ANYTHING whatsoever. They're side bets, pure and simple, and many of those making the wagers had no ability to pay up in the case of losses. The idea of running bucketshops didn't stop when they were outlawed... it was expressed later by those that led the US Government to deregulate via the Gramm-Leach-Bliley Act, and it was implemented by the "Derivatives Desk."

Of course, the Bucketshop is illegal, but the insideous concept finds its way even into the regulated markets, by way of the DTCC. Is the DTCC just throwing your trades in a bucket in the back room? In some cases, at least, it certainly is; only, we the customers don't ever get to look behind the curtain to see for ourselves. Does the share that my brokerage claims on my account really represent a legitimate link to a physical asset? All I know is what my broker tells me. If my broker were a bucketshop, would it be obvious to me, the customer? Would they admit it?

The DTCC needs to get cracked open. Let's find out what's going on in there. The Investing Public has the RIGHT to know how the DTCC handles their PROPERTY.

Friday, April 3, 2009

Part II : Percentage Land Area required for 100% Replacement of 2006 Energy Demand.

Yesterday I posted a chart showing a rough estimate of how much land area would be required by each State in order for that State to replace 100% of its Energy Demand (per DOE numbers).

I posted it at DailyKos, and on the LDK board for comments.

Apsmith of DailyKos makes a good point that there are generator losses, etc., which should be used to reduce the overall total energy required to be replaced, and China_s2 of Yahoo agrees, and points out a different set of data, which is based on retail electricity use, so should closely represent actual electricity delivered, as opposed to total Energy Input.

So, I copied over the old data to a new sheet, plugged in the new data, and came up with a rough estimate of the total land are required to replace 100% of US 2007 Electricity demand.

Thursday, April 2, 2009

Percentage Land Area required for 100% Replacement of 2006 Energy Demand.

The following chart represents the percentage of land for each State, and the USA as a whole (without Alaska), that would be required to replace 100% of that State's Annual Energy Demand.

Make no mistake, the numbers are huge. Then again, nobody is actually talking about 100% replacement by Solar, Ever. This is just to give an idea that it is physically possible, at all.

Assumptions and references follow.

here's the spreadsheet.


State Energy Data.
State Land Area Data.
State Insolation Estimates.
Sunpower Power/Area Claim.

Assumptions / Notes:

The percentages reflected in the Graph are based on a Stationary system, though the value for Power/Area is based on a Sunpower claim related to their tracking system. This should be irrelevant, as Power is independent of whether the system tracks or not. Since these are Sunpower numbers, the Panel's Conversion Efficiency should be around 22%.

The Demand cited is irrespective of source, and so includes existing production of renewables such as Hydropower. Here's a very interesting page from the DOE giving detailed map-based information on US Energy sources. There's a "Select a State" dropdown that will take you to a close-up of the individual State including facts and demographics.

In order to work out an the Area, I used the equation:

Annual Energy Output = 1 Year * Power/UnitArea * Insolation Ratio * TotalSolarArea * 8760.

For more info, see A Note on Units of Energy and Insolation. Solve for TotalSolarArea, and divide by the State's Total Land Area, and you will get the percentage. Most of the trouble here is just in the conversion of units. On a political note, can we just all go metric please?

The Insolation values were eyeballed from the map. If anybody's got some better data on State Average Insolations, I'd love to see!

The base data does not seem to include Transportation Energy, though it didn't specify.

Of course, this assumes nice flat areas of land, on which to set up installations, and it also assumes that each state takes care of its own needs irrespective of local conditions or capacity. It's a brief look from 1000 miles up above. It's not exhaustive, but it's fun, and maybe interesting.

By all means, if my basic math is way off, let me know.

This post is followed by Part II, which calculates the same area percentage, but only for the replacement of Electricity End Use.