Slashdot is powered by your submissions, so send in your scoop


Forgot your password?
Power Earth IT

Intel Testing Solar Power For Data Centers 66

miller60 writes "Intel has installed solar panels at a New Mexico facility to test the potential for using photovoltaic solar power in data centers. Solar has proven impractical in data centers thus far for reasons of cost (too high) and capacity (too low). Intel will test the 10-KW solar array with data center containers and as supplemental power for summer capacity challenges, and says the project is a first step toward solar data centers. The project is housed at the New Mexico site of Intel's recent research in air side economizers in data center cooling."
This discussion has been archived. No new comments can be posted.

Intel Testing Solar Power For Data Centers

Comments Filter:
  • by Rei ( 128717 ) on Tuesday January 20, 2009 @05:23PM (#26536573) Homepage

    It's a much simpler task []. You use solar heat to drive an absorption or adsorption cycle rather than a compressor. Datacenters need lots of cooling, after all.

    • by Rei ( 128717 ) on Tuesday January 20, 2009 @05:27PM (#26536655) Homepage

      Just thought I'd add: here's an example [] solar air conditioner. Some are even reversible and can become heat pumps in the wintertime.

    • by MozeeToby ( 1163751 ) on Tuesday January 20, 2009 @05:45PM (#26537011)

      Except that the summary specifically says that the new research is being done at the same place they recently tested thier non-AC cooling solutions, so this particular datacenter needs almost no cooling.

      • just because this is where they chose to test their air side economizers does not mean that the data center requires no cooling. air cooling is still cooling, and the use of a test site where no cooling is needed to test alternative cooling solutions would be pointless. and the testing was done in a temperate desert climate in New Mexico (with the average temperature in July being 92 degrees Fahrenheit). it wasn't done in Alaska or the Arctic.

        in fact, if you actually read the article on Intel's air economiz

    • I remember a story on slashdot where they seal up the entire datacenter so it's almost completely airtight then put a big turbine in the roof and the hot air rises up to it and turns it which powers the AC. Now that's an idea! But there was that other story that was a lot better which takes about some kind of phase something or other cooling where basically the processor's heat drove the cooling system on the heat sync. That'd save some money of electricity both for internal fans and for building cooling
  • by Ungrounded Lightning ( 62228 ) on Tuesday January 20, 2009 @05:26PM (#26536637) Journal

    ... but solar thermal -> heat engine has, if the size is reasonably large.

    Wind turbines ditto if large enough and the site has good winds. (For house-sized they're only past cost breakeven if you can save a few grand by not running grid power to a new rural site.)

    Progress in photovoltaic design and energy storage systems may bring both solar and small wind past the competitive-with-grid crossover, perhaps in the next few years. Rises in grid power costs could do it, too.

    • by Ungrounded Lightning ( 62228 ) on Tuesday January 20, 2009 @05:37PM (#26536851) Journal

      By the way: "Breakeven" in this context is "costs no more than installing and buying grid power" - appropriately amortized over the life of the installation.

      • Re: (Score:3, Insightful)

        by Vellmont ( 569020 )

        By the way: "Breakeven" in this context is "costs no more than installing and buying grid power"

        Depends on where you are. Many areas of the country have much higher electric costs, and a lot of sun. With tax credits available, it often times is a good investment.

      • You're wrong anyway. PV panels would pay themselves back in less than fifteen years [] back in the seventies. Today it's even better. And using thin-film solar it ought to be possible to get it in a pretty short time WITHOUT any subsidies. A typical PV install is considered to last at least 20 years (with some maintenance, but that's part of the cost) while it will pay itself back in 15 or less, and energy costs are generally rising.

  • by ducomputergeek ( 595742 ) on Tuesday January 20, 2009 @05:29PM (#26536709)

    We're not a datacenter, just a development company. We had money left over last year that was either going to be taxed and some tax credits expired in december. So we needed to be reinvested into the business some how. We put in solar panels on the office roof that meets about 60 - 70% of our power needs. This has lowered our power bills by over half. That's freed up enough cash flow to pay for another developer. We viewed the investment as a sunk cost that freed up enough to hire an additional Jr. programmer that we were really needing.

    • by StreetStealth ( 980200 ) on Tuesday January 20, 2009 @05:47PM (#26537045) Journal

      People throw the word "green economy" around without really thinking about it, but you just gave the most succinctly excellent example of it I've ever read. What a great move.

      • by ducomputergeek ( 595742 ) on Tuesday January 20, 2009 @09:29PM (#26540135)

        It's not green economy, it's just economics coupled with things they don't teach you at business school. (I hold degrees in German and International Business). What holds most people up is that the rate of return and ROI calculations aren't quite that good from an accounting stand point. It still takes about 7-12 years to pay for itself ceteris paribus. That is if you just calculate the cost savings off the utilities. Then you can add some future present value if you were to invest the money in the analysis if you want to get complicated. But in this market, where do you want to invest? Now it maybe true that for, say an average home owner, solar has not come down to the point where it can be declared feasible. But there are some other intangibles that are hard to put a number to down on paper. i.e., if we use the extra cash flow and spend it on marketing, or another developer, etc., what are the potential payoffs?

        What we do know is that utility prices are unlikely to go down and the new developer is already paying dividends. We have a new product release ready to ship at the end of the month. Without him, it would have been March or April. If all goes well, then the cost of the solar panels could be recovered in as little as 18 months and we're getting started on the next phase of the project 2 months ahead of schedule and we can do so with little or no change in exiting net cash flow.

        I wish I could say they taught me this at college, which taking 4 semesters of Econ did help, but I learned about sunk costs and how to factor risk into ROI considerations from our family farms when I took a year off from work before I started graduate school to learn how to run them with my Dad. (We rent the farms, but take an active role in managing the business side of things like selling on the futures market, etc..) An example was putting up grain storage bins. It was taking over 3 weeks to harvest the rice when we had to take it to the dryer, they could manage to unload 2, maybe 3 truck loads a day. That's 3 weeks where $500k worth of crops were sitting out in the field. All it takes is a hail storm or high winds, and you can cut that number by 30% or more. (It happened this year, Hurricane Ike knocked the yields down by about 30 - 40 bushels per acre, and at $5.50 a bushel it adds up). For $50k we could put up storage on the farm. It now takes 3 days (or 2 long days) to harvest all the farms. So a ready crop is now sitting out in the field a very short period of time. How long is it going to take to recoup the ROI there? Well if you look at the raw numbers, about 15 years. But if you get your crop in once before a hurricane Ike comes around, suddenly they've just paid for themselves. Now granted, my Dad is a retired executive that spent his entire career on the finance side of things at a large company. So I've grown up around looking at things...differently than my peers.

        On the farms, we hedge. In a sense, that is what we are doing here. We are hedging that utility prices in our area will continue to rise over the next 15 - 20 years. But more over, we think we an invest the savings into other projects that will in turn pay off more than the cost of the solar panels. So it made sense in our case.

    • Thank you for multiple things. You've provided an example to show that taxes do not kill business, but motivate owners to reinvest in the business. You've provided an example of alternative energy being cost-effective and good for the bottom line, even to the extent of increasing employment. Thank you for being at least a temporary antidote to the shrieking we normally get about environmentalism being the death of capitalism.

  • The cost is still too high and the output is still too low but at some point the technology is going to catch up. I just hope groups with deep pockets continue to invest cash in alternative power projects like this.

    • by icebike ( 68054 )

      > The cost is still too high and the output is still too low but at some point the technology is going to catch up.

      And that would be the time to buy it.

      Buying something now in the hopes that what ever you eventually have to replace it with might be better is sort of like planting an Egg and waiting for chickens to sprout.

  • Solar energy would be good but that's not the only clean energy solution.

    You ever stood behind a rack of servers? Those things put out a lot of heat. If we can tap geothermal energy from deep underground, we should be able to grab it from a server room and convert it to energy.

    Using that heat energy for electricity will also reduce the amount of cooling power needed.

    I plan to finalize that solution after I put the finishing touches on my perpetual motion machine.

    • Re: (Score:3, Informative)

      by DerekLyons ( 302214 )

      You ever stood behind a rack of servers? Those things put out a lot of heat. If we can tap geothermal energy from deep underground, we should be able to grab it from a server room and convert it to energy.

      Why 'should' we be able to? You think the laws of physics work like they do in Star Trek where all you have to is wish hard enough?

      Not to mention that what we are grabbing from deep underground is steam to drive turbines, not heat directly.

      • by EatHam ( 597465 )

        Why 'should' we be able to? You think the laws of physics work like they do in Star Trek where all you have to is wish hard enough? Look, I'm just a caveman blah blah blah, but in my world heat = energy which can be recovered, and geothermal taps a constant temperature, not steam.

        • Well, your world bears little resemblance to the real one. In the real world, heat always equals energy that can be recovered only in theoretical realm - because low mass flow or low temperature differentials (or in this case, both) means that energy can be recovered only theoretically, practically and economically it can't. In the real world when I peek into a pipe leading from a geothermal well I get scalded by steam, not theoretical heat sources.

          • Re: (Score:3, Funny)

            by CheeseTroll ( 696413 )

            when I peek into a pipe leading from a geothermal well I get scalded by steam

            Caution: Do not look into geothermal pipe with remaining face.

    • Heck, the rotation of the world has got a lot of kinetic energy. If we could just fashion a generator off either pole and fix it in place in space somehow...
      • Re: (Score:3, Funny)

        I proposed that; but you wouldn't believe the whining I had to put up with. An entire bloody hemisphere: "But Mr. fuzzyfuzzyfungus, we don't want to be plunged into eternal darkness and subfreezing temperatures..."

        Spoiled, the whole lot of them. Why, in my day, we had to chip away the ice with our teeth, provide light for our crops by burning our children, and by god we liked it!
  • reasons of cost (too high)

    As opposed to it being too cheap to implement?

  • What is there to test here? They know the power requirements of a data center. They know the power output and cost of solar panels. Either it's worth it or it's not.

    Don't get me wrong, I love to see the pursuit of alternative power. But it looks as if they just tacked on a solar panel to their data center for some cheap green goodwill and publicity.

    • You may be surprised to learn that sunshine is not constant at every point on the Earth. Thus the power-output of the panels it not constant.
  • test the potential
    Solar Power

    I see what you did there...

  • by zappepcs ( 820751 ) on Tuesday January 20, 2009 @06:00PM (#26537291) Journal

    I would never think that Intel would leave data center power research to Sun? You'd think they would do their own research?

  • then we do the stupid thing and "burn" the skies to cut off power. We all know where that all leads to.
  • by MrSteve007 ( 1000823 ) on Tuesday January 20, 2009 @06:21PM (#26537617) []

    I designed, installed and maintain a 10kw solar array last year to power our businesses servers and offer a large (2900 amp hour) uninterpretable power supply during prolonged grid outages.

    We recapture the waste heat during the winter to heat our facility at night. During the summer we vent that heat directly to the outside, and only use the AC as auxiliary cooling. It works excellently. []
    • by Torfbolt ( 791759 ) on Tuesday January 20, 2009 @06:43PM (#26537939)

      ... uninterpretable power supply ...

      Yeah, those things sure are hard to understand sometimes...

      • Doh! That's what I get for using Firefox's built in spell checker to cover for my misspellings. Good catch.
    • WTG! I'm glad you went heavy on the conservation angle first. Dollar for dollar, that is the most cost effective energy solution. And good luck with the microhydro project.

    • I designed, installed and maintain a 10kw solar array last year to power our businesses servers and offer a large (2900 amp hour) uninterpretable power supply during prolonged grid outages.

      I would be very interested to know how solar is an Uninterruptable Power Supply at night. :)

  • Do the math, folks (Score:4, Interesting)

    by Ancient_Hacker ( 751168 ) on Tuesday January 20, 2009 @06:37PM (#26537847)

    The numbers on this are super dismal. To power a 300W server, you need about 5 square meters of solar collector. About $12,000 of panels to offset 2 cents an hour of power. Plus you need tons of storage batteries or substitute power for night and cloudy days. Yuck. A sensible company would only do this for PR or due to some government mandate or tax credit. Certainly not to save money or save energy.

    • Your math is off a bit. You don't need 5 square meters to generate 300 watts; maybe if you average it out to include night, it'd come to that, but in a place like Mexico I doubt it'd be that low.

      At last Mayâ(TM)s CeBIT trade show, Sun Microsystems (JAVA) demonstrated a solar Blackbox shipping container hooked to a 700 square foot array of solar panels, which produced about 10 kilowatts of power. Thatâ(TM)s barely enough power to support a single rack in a typical high-density configuration in a Blackbox (since renamed the Sun MD).

      700 square feet converts to 65 square metres? ~154 watts per square metre? That's 750w for 5 square meters - 2.5x your estimate, according to the article.

      Why store it? Pump it back into the grid and avoid all the costs of batteries. It's not a replacement for a generator. It's just a way to offset power demands and

      • by Belial6 ( 794905 )
        Don't forget to factor in that ridiculously low price calculated for electricity. I know that we pay a lot for electricity here in CA, but is there anywhere in the country that they are paying $0.06 /kwh?
      • by dwywit ( 1109409 )
        There's a few extra things to consider when trying to design and implement a solar PV-powered system.

        Yes, PV output of 154 watts per square metre sounds about right - during the middle of the day at low latitudes, when it's not cloudy. You have to over-supply the amount of PV to allow for such cloudy days (where PV output can be as little as 10% of peak), early mornings and late afternoons, etc.

        Putting it back into the grid would seem to be the most cost-effective solution - it offsets your energy bills,

      • Usually you want to run servers 24 hrs a day. Sunlight is only there for 12 hrs minus cloudy times. That's why the 2.5x factor. I was even generous enough to assume you could track the sun with $12K.

        And pumping it back to the grid is silly. There is no way to even pay the maintenance much less the interest on a $12K investment at 2 cents an hour.

        • Usually you want to run servers 24 hrs a day.

          No. Usually you want to run servers 5.3 hours per day, but those hours are unevenly and sometimes even quasirandomly throughout the business day (with potential for off use) which is why "cloud computing" is so ultimately desirable if it can ever be worked out to not be a gigantic PITA.

          The point here is that the servers use only idle power most of that time. And lately low power consumption has been a huge selling point for servers. Since pretty much all of the latest generation of CPUs and GPUs (less appli

    • by dbIII ( 701233 )
      It's a bit better than that if you are not in the arctic with 1960s panels. The other thing is to really think about what it is for - in at least one datacentre they are not the primary source of power but effectively a UPS that works during core business hours that can cover very long outages (much shorter ones at night depending on battery capacity). Another way to think about it is to handle the peak consumption when people are in the office even if the server power usage is fairly constant. That coul
    • by mcrbids ( 148650 )

      Running 300W 24x7 assuming a 30 day month at $0.25 per KWH (which is about what I pay here in sunny NoCal) comes to a monthly cost of $54/month. At standard loan rates, that implies a purchase of $5,400 at fairly standard loan rates. (EG: 10% -ish over 20-ish years)

      $5,400 isn't that far from $12,500. It's only a off by about ~2 1/3 times. Factor in the advance of technology or other special circumstances, and you'll find that this may well make perfect sense. And while the numbers haven't "uncrossed", it ce

      • Holy crap... I thought my 12.22 cents was high.

      • Nice try, but it's all too easy to fool oneself even with what you think are hard and fast numbers.

        First of all the solar panels provide episodic power at unpredictable times at unpredictable amounts. That kind of power is worth much, much less than steady reliable power. In most areas it's priced at about 3 to 6 cents per kilowatt-hour.

        So that's closer to $10/month, $120 per year. Not enough to even make a dent in the interest.

    • and get some accurate information to boot.

      Idiots like you spouting 1990 numbers aren't helping anybody.

    • With a 10K Solar Array and appropriate battery storage, you could run 50 eeeBoxen for a year.

      A 10K array can pull about 9000 KWh per year. An eeeBox has a maximum power draw of 20W (480Wh per day, or 175.2KWh per year).

      And that's in Seattle. Here in Phoenix, you'll be generating more like 14000 KWh per year. Which is almost 80 eeeBoxen.

      Hmmm... Asus needs to make eeeServers.

  • Solar energy can be converted to electricity. Film at 11.

    because Intel is doing it at a "data center"?

    Why do we care?

    • Solar power should be installed everywhere useful. That means roofs on correctly-oriented houses. A lot of houses are incorrectly oriented. I know you might say "it's my fucking house" but the simple truth is that if you don't have a passive solar design with a wall with proper overhangs and windows facing the equator then you are just wasting energy. (Exceptions for subterranean dwellings apply, although unless you're trying to hide the structure I suggest a solar wall.) Remember, point-of-use generation e

  • Best future ironic tech-marketing orgasm ever.
  • We need to de-distribute the power grid and use Nuclear batteries as a backup for green power sources.

  • They need to build a 10 acreindustrial solar thermal plant. This will generate about a gigawatt, 24/7. If they have the land near the fab, great. If not just sell then power into the grid and still make money instead of breaking even.

Evolution is a million line computer program falling into place by accident.