Slashdot Mirror
Can the Sun Realistically Power Datacenters?
1sockchuck writes: A massive solar array in central New Jersey provides the daytime power for a server farm delivering online financial services for McGraw Hill. The 50-acre field of photovoltaic solar panels symbolizes a new phase in the use of renewable energy in data centers. Massive arrays can now provide tens of megawatts of solar power for companies (including Apple) that can afford the land and the expense. But some data center thought leaders argue that these huge fields are more about marketing than genuinely finding the best approach to a greener cloud.
Can a certain number of solar panels power a datacenter with a given load? Sure! Just not 24/7. But that doesn't mean they can not. Your datacenter takes 1 MW/h. You receive roughly 8 hours of usable sunlight, so you need 3MW/h capacity of solar panels to produce the power you need. During the day, the power company will take your excess power, and light up factories, offices, Air Conditioning, etc... During the night, you will use the power company to power your datacenter, when it has to keep its power systems up anyways, and therefore has excess capacity. Look up the terms "base load" and "peak load", understand that not everything that plugs into the power grid needs to be up 24/7 ( not even all servers, look up "DRS"). So can that much power be produced to have a "net neutral" load on the grid? Sure!
But there is always the desire to be completely self-reliant. In this area, I always liked the idea of using the excess power during the day to lift water to a lake high up, and running hydro at night to power the datacenter. This is of course expensive, especially since power companies have excess power at night anyways, since the cycle time to stop / start producing base power won't allow the company to shut down X generators at night.
It may be a drop in the bucket now (Facebook's 100kw solar array for a facility consuming 25Mw is just that), but the infrastructure is in place to put in better panels later as they're developed. Additionally, if using otherwise "wasted" space (such as a rooftop), why not put it in place? The long-term power cost savings for such a facility (that is planned for the long term, anyway) will eventually pay for for the system a few times over, even if the impact to overall energy usage is that proverbial drop in the bucket. In other words, it makes business (read: financial) sense to do it.
My sources are unreliable, but their information is fascinating. -- Ashleigh Brilliant
A: Yes. It's called "evaporation." Next question, please.
There's no -1 for "I don't get it."
Sadly, there just aren't enough places with lakes to store anything like the amount of power we'd need to store. You also have to deal with transmission loss between the solar site and the point of use. There was this proposal a while back to use massive, carved granite/stone blocks to store power but it doesn't seem to have achieved much mention beyond its initial proposal.
I have no problem with your religion until you decide it's reason to deprive others of the truth.
Average solar insolation is more like 5 sun-hours/day, not 8, in good locations. Much less in places like Germany. If you want autonomy on the shortest day of the year, you may have less than 2 full sun hours, which means 12 MW of capacity, but that doesn't account for a cloudy day, in which case you may get less than 1 full sun hour insolation.
So, bottom line is there are a lot of ways to look at the numbers, but to be truly autonomous with no grid support, you need a lot of capacity.
Efficiency can be easy - we just need to build a Dyson sphere.
We're already harnessing the power of the sun without "batteries" in the traditional sense. Most of the recent plants built (and under construction) here in Arizona are molten salt, which provides full power for three hours after the sun is "off" -- well into peak residential hours -- on residual heat.
We're still nowhere near 24/7/365 coverage, but we're making strides.
> The fact that some arrays were done in a way that's incompatible with farming doesn't mean that it can't be done.
The light either hits the corn leaves, or it hits the solar panel. The same photon won't hit both. You don't get to use that same bit of sunlight repeatedly. Each photon is either absorbed by the solar panel, or it's absorbed by the crop. You _could_ mix 25 acres of solar with 25 acres of farming, to have 50 acres of both mixed together. The productivity of mixing them together would be precisely the same as having 25 acres of farmland on one side of the street, and 25 acres of solar on the other side of the street. Mixing them, with ten feet of farm, ten feet of solar, ten feet of farm, ten feet of solar would be silly, though, because it's awfully hard to harvest the corn with solar panels in the way.
Sadly, there just aren't enough places with lakes to store anything like the amount of power we'd need to store.
This is actually a silly concern. Electricity demand is highest in the middle of the day when the sun is shining. That is also when the spot price for power is highest. It makes no sense at all to store that power to sell it in the middle of the night, when prices are far lower.
Storing solar power is an issue in niche applications, and it is an issue in a future fantasy world where 100% of our power is solar. But it is not an important issue in the real world, and is unlikely to be for a long, long time.
I recently saw that India is taking an innovative approach to solar installations. They are installing the panels over irrigation canals. This has a few benefits... less evaporation of water because of shading and the government already owns the land for the canals so no land needs to be acquired and no land is taken out of food production. They have thousands of miles of canals.
I don't read your sig. Why are you reading mine?
Read the post again - it's insOlation, which is correct, not insUlation, which was your assumption.
Solar PV capacity planning, at least in domestic situations, is based on the amount of energy captured/generated by a panel at its PEAK capacity, and is generally calculated at 5 hours/day in temperate zones, less in frigid, more in tropical, with modifiers for local conditions and climate. Panel output throughout, for example, a clear sunny day in the mid latitudes corresponds closely to a steep-ish bell curve (more like a sine wave, though). Low output at either end of the day because the incidence of the sun's rays to the panel are more oblique.
Panels are getting better at "catching" oblique insolation, but obviously they're much better between the hours of 9-10am and 2-3pm. There is a significant amount of energy captured outside these times, but it's not really useful when calculating the number of panels needed. It's better to state that you'll capture a minimum of x on sunny days, rather than a maximum.
They sentenced me to twenty years of boredom
The example of Prineville in TFA is a good one. Here in Oregon we have a lot of base load provided by hydroelectric. We have had squabbles between the Hydro guys and the Wind guys at night in winter when the Hydro guys need to keep the turbines spinning to keep the dam levels safe, and the wind guys have to stop feeding into the grid and that hurts their bottom line.
Summer during the daytime is when Oregon fires up more of the of the coal and natural gas plants, so solar fits in well to cover these peaks times and seasons when the rain isn't falling, the snow has stopped melting, and the AC is running.
Other regions are not nearly as lucky as Oregon to have good wind and hydro options, but lets not disqualify a technology just because it isn't a perfect fit everywhere.
Sigh.
The best solar PV plants in Germany have a capacity factor of about 13%.
http://en.wikipedia.org/wiki/S...
These are the plants in the best locations. That comes to an average solar insolation of 3.12 full sun hours/day. If you had 5 hours you should have closer to 20% capacity factor. I'll let you think about why they don't see 20%, and why the average capacity factor overall is 9.5%, and try to reconcile it yourself. When you get stumped, I'll gladly explain.
primordial, a.: existing at (or from) the very beginning; first in time, earliest. Ergo, hydrogen, helium, and lithium. Everything else came later.
Ezekiel 23:20
data centers generally aren't lacking for available roof space so no taking up any more land.
Above the atmosphere, at the equator, the average insolation (that is, the amount of incoming solar energy, averaged over the course of a day) is about 400 watts per square meter. At the bottom of the atmosphere in an ideal location (like the Sahara) it's closer to 300 W/sq. m. In most places where people want to have data centers, the number is closer to 200 W/sq. m...or worse. And the efficiency of commercial solar panels runs about 20%, so you're down to 40 watts per square meter.
200 watts is (optimistically) about the draw of a single server, so you're looking at powering one server for every five square meters of rooftop. If you want to run on rooftop solar, then you're going to have to design a data center with very short racks and very wide aisles.
~Idarubicin