Green Apples May 21, 2012Posted by Maury Markowitz in solar.
Tags: cost of electricity, kilowatt hour, LCoE, solar panels
Last week Greenpeace pulled another one of their publicity stunts (do they do anything else these days?) by “landing” on the Apple campus and complaining that Apple’s data centres don’t use renewable energy. Only days later, Apple makes a big announcement about how they’re making all their data centres run 100% on renewables. One may be forgiven for concluding these events are connected.
Out in the blogosphere there’s been a lot of debate about the “why”, why is Apple spending all this money to “go green”? Is it pressure from the likes or Greenpeace? Or maybe Jobs’ buddist leanings from beyond the grave? Pure publicity on everyone’s part?
Maybe. Or maybe it’s just about the money… you know, green.
Green, as in leaves
I’ve posted on the Levelized Cost of Energy (LCoE) before, so I’m not going to do that again. Instead, I’m going to walk you through a real-world example of how to use LCoE on Apple’s data centres. The LCoE article was full of math and looked complicated. Now you’ll see just how easy it really is.
LCoE is simply the amount of money you pay for the system divided by how much energy that system produces over its lifetime, after adjusting everything for inflation and interest. For no good reason, I’m going to start with the energy production side.
Apple has two data centres they’re converting, Maiden, North Carolina and Prineville, Oregon. Ok, so now we open two new browser windows, one onto Google Maps, and another onto PVWatts 2.0 - go ahead and click both of those links. Ok, go to Maps and search for “Maiden, North Carolina”, then zoom out two or three levels. Notice that it’s just northwest of Charlotte. Ok, switch over to the PVWatts page and find the star on that map that’s closest to Maiden – right on Charlotte in this case.
Click that star and a small black-background pop-up appears in the upper right. Click on “Zoom to PVWatts”. This takes you to the actual heart of the PVWatts system. Selecting the weather station (the star) feeds the last 20 years of sun angles, clouds, snow cover and everything else you can think of into this page, where you can run calculations.
For reasons I’ll never understand, PVWatts defaults to a 4.0 kW system, which is dumb. Change that “DC Rating box” to 1. The next box is another oddity, PVWatts says you’ll lose about 23% of all the energy before it gets out of the system. My tiny system on my garage is losing only about 18%, and large systems are even lower, around 15%. This bit of frugality on PVWatt’s part is well known and often commented on. So change it to 0.85.
If you read the Apple article carefully, you’ll notice “These sites use high-efficiency solar cells and an advanced solar tracking system”. So in PVWatts, change the “Array Type” to “2-Axis Tracking”.
Hit the “Calculate” button. Up comes a new page. Over on the right is the output section. At the bottom you’ll see some totals. The only useful one is the “AC Energy” column, which reads 1,907 kWh. That means that PVWatts predicts that a 2-axis array with 1 kWp of panels will produce 1,907 kWh of energy a year. Apple is building a 20 MW array, which is 20,000 kW, so 20,000 times 1,907 is 38,140 kWh a year, or 38 MWh. Apple’s press release says 42 MWh, so they’re saying that PVWatts is still on the low side even after we changed it to 0.85. If you play around with the calculator a bit, you’ll find Apple is assuming a DC Derate of 0.94, which is damn impressive.
Ok, so now we need to know how long this system will run. That’s easy, it’s 25 years, minimum. So we expect to get 42 MWh a year for 25 years, or 1,050 MWh over the lifetime of the system. Well that’s basically that. We have almost the math done. See, that wasn’t so hard, was it?
Ok, now the cost. Right now over at the day job we’re selling complete kits for $1.79 a Watt. These are for small systems on the order of 10 kW or less. For large systems like Apple’s, one can shave 25 cents off the panels and another 25 cents off the inverters. However, the trackers will add another 50 cents, so I’m going to roll all of this together, up it, and call it $1.75 a Watt. Install costs on a system of this size are about 35 to 50 cents, so I’m going to round up and call the all-in cost of the complete system $2.50 a Watt.
Ok, ($2.50/W x 20,000,000 W) / 1,050,000 kWh = 4.7 cents/kWh
Now comes a bit of complexity. This calculation assumes that the system does not need maintenance, there’s no inflation and they don’t pay any interest on the capital they spent to build it. The first is not that far from the truth, as PV requires very little maintenance, especially compared to other energy sources that have moving parts. But the second and third, no inflation and no interest, is obviously not true. That gets a little more difficult to calculate, but it’s really easy if you use a spreadsheet. So I’ve run the calculation for you (and attached the spreadsheet, below).
And -drum roll please- the LCoE for Apple’s system is 9.45 cents/kWh using my PVWatts estimate, and 8.41 cents/kWh using Apple’s.
Ok, what about the other data center? Ok, I Googled up Prineville, Oregon and found that it’s just northeast of Bend. And now over to PVWatts we go… oh perfect, there’s a star right between Prineville and Bend. Ok, same assumptions, 1 kW system, 0.85 derate, 2-axis tracking and… 2119 kWh/kW of panels. And if we use Apple’s derate of 0.94? 2341 kWh/kW of panels. So I’ll plug all of that into my LCoE calculator and out pops 8.5 cents at my PVWatts estimate, and 7.63 using Apple’s.
So using these quickie calculation, Apple’s going to get power for somewhere between 7.5 and 9.5 cents.
Green, as in cashola
So, ask yourself this question: do you think Apple pays more than 7.63 cents/kWh and/or 8.41 cents/kWh for electricity during peak hours?
Yes, you do?
Then maybe the real reason Apple’s building a big solar array isn’t to feel good about themselves, but feeling good about their bottom line.
Here you go, use my Apple LCoE play sheet and see what you can see. Inputs are in green, calculated outputs in yellow, and the “bottom line” are the two yellow boxes on the right.
Change the prices and interest rates in the green boxes to see what sort of effects they have.The duty cycle input at the bottom should be between 22 and 25% to match mine or Apple’s outputs.
If you want to know more about this calculator, it’s more fully explained in the LCoE article.