Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Microsoft Power Technology

Microsoft Used Hydrogen Fuel Cells To Power a Data Center For Two Days Straight (engadget.com) 79

Microsoft announced Monday that hydrogen fuel cells powered a row of its datacenter servers for 48 consecutive hours, bringing the company one step closer toward its goal of becoming "carbon negative" by 2030. Engadget reports: The idea to explore hydrogen fuel cells originated in 2018, when researchers at the National Renewable Energy Laboratory in Golden, CO used a proton exchange membrane (PEM) hydrogen fuel cell to power a rack of computers. Mark Monroe, a principal infrastructure engineer on Microsoft's team for datacenter advanced development, said his team watched a demonstration and was intrigued with the technology. Monroe's team developed a 250-kilowatt fuel cell system, enough to power a full row of data center servers, and in September 2019 installed it at an Azure datacenter near Salt Lake City, Utah. In June, the system passed a 48-hour test. The team plans to test a 3-megawatt fuel system next, which matches the size of current diesel-powered backup generators.

It's possible that an Azure data center could be equipped and run entirely on fuel cells, a hydrogen storage tank and an electrolyzer that converts water molecules into hydrogen and oxygen, Monroe said. These systems could integrate with the electric power grid to provide load balancing services. Further, hydrogen-powered long-haul vehicles could come to datacenters to refuel. By continuing to develop hydrogen fuel technology, Microsoft could eventually serve as a model for use of hydrogen fuel cells elsewhere.

This discussion has been archived. No new comments can be posted.

Microsoft Used Hydrogen Fuel Cells To Power a Data Center For Two Days Straight

Comments Filter:
  • Wherefore? (Score:4, Informative)

    by dunkelfalke ( 91624 ) on Wednesday July 29, 2020 @05:20AM (#60342429)

    If they have enough electrical power for the electrolyser, they have enough electrical power to power the data center directly.

    • by Rei ( 128717 )

      The idea would be storing hydrogen for backup (assuming they're not just reforming it from a natural gas line, which would be nonsensical, since they could just use the natural gas directly for backup power).

      Of course, storing and then later using electrolytic hydrogen means:

      * Throwing away a large portion of your energy
      * Huge capital costs on your electrolyzers ( x moderate lifespans = huge depreciation)
      * Huge capital costs on your fuel cells ( x moderate lifespans = huge depreciati

      • According to Wikipedia underground hydrogen storage is the cheapest option around 2000 hours [wikipedia.org] so I'd guess that this should be long term grid balancing, for example between summer and winter. You would be justifying the capital expenditure against the cost of purchasing and maintaining a full set of generators which you normally need for a data centre anyway. This would basically be a cheap way of setting up a long term grid storage plant with the capital reduced by the value of the generators that you don
        • by Jzanu ( 668651 )
          Precisely. This approach immediately makes sense from a project cost management perspective. Unfortunately, now that we have the technology to move forward the usual objection to the use of hydrogen (fear/ignorance) has been replaced by the self-interest objective for those who have invested in the companies adapted to the current temporary stage reliant on mostly unrecyclyable batteries (due to their waste processing hazards). While alternative batteries (flow batteries) have existed at scale for decades,
        • by Rei ( 128717 )

          I'm trying to trace back the numbers in the paper Wikipedia is sourcing (they actually cite two, one that says that it's cheapest and one that says that it's not). The one that says that it's cheapest [europa.eu] cites a LCOE of €140/MWh from "Fuel Cells and Hydrogen Joint Undertaking (2015), Commercialisation of Energy Storage in Europe" [europa.eu] But because li-ion energy storage prices are crashing, their claims of ">1000 €/MWh" LCOE for batteries are highly obsolete. The industry average in March was was $150/M [energy-storage.news]

        • so I'd guess that this should be long term grid balancing, for example between summer and winter

          In theory, yes. In practice, the vast majority of these problems is cheaper solved by overgeneration and curtailment. 2000 hours is a vast overkill [sciencedirect.com] - cost-optimal scenarios might require something between 9 and 90 hours of storage.

      • This may not necessarily be all that useful for data centers, but there's plenty of remote locations that will welcome any developments in this area.
      • by lazarus ( 2879 )

        You're correct, re: this is for backup. But I'll add a point you missed.

        A big problem with data centres is "permitting" for the gensets. Diesel needs to either be polished regularly, and/or you need to run the generators often enough that you are using up the fuel and replacing it with fresh. It is getting harder and harder to get the permits required to install massive locomotive-sized generators at data centre facilities and the emissions they bring with them.

        Microsoft and many other hyper-scale folks

      • They should look into LiquidAir cryogenic energy storage by Highview Power, its all tried and tested tech already used elsewhere https://en.wikipedia.org/wiki/... [wikipedia.org]
        It takes in fresh(?) air from the environment, filters it, freezes it, warms it back up to spin a turbine for power and expels clean air that could go back into the buildings.
    • Well, they were making their hydrogen from natural gas, so I guess most of the energy comes from there which is hardly "green" right now. But for future they don't need to be. Although it's been cheaper to produce hydrogen from natural gas, it seems the state of the art has advanced quite a bit [wikipedia.org] and you can get 94% efficiency converting water to gas and then you get some losses in storage [wikipedia.org] and so if you built a data centre near to a salt cavern or disused oil well where you could store the hydrogen cheaply
    • by Alworx ( 885008 )

      The text is very confusing. It would appear, reading between the lines, that the plan is to generate hydrogen via wind/solar so to guarantee constant load balancing (or to provide a backup boost if extra power becomes suddenly necessary).

      Otherwise, I agree, the laws of thermodynamics still stand.

    • Comment removed based on user account deletion
    • "If they have enough electrical power for the electrolyser, they have enough electrical power to power the data center directly."

      No such thing (yet). They used hydrogen made from natural gas, carbon-free this is not.

    • by AmiMoJo ( 196126 )

      If they are moving to renewable sources, e.g. solar on the roof of the datacentre, they likely have excess at some times and a deficit at others.

      Also they are using fuel cells to replace diesel generators used when there is a power outage. The UPS system will only run for maybe 10 minutes max, long enough to spool up the generators.

  • by fubarrr ( 884157 ) on Wednesday July 29, 2020 @05:22AM (#60342431)

    How much natural gas was used to make so much hydrogen??

    For the cost of fuel cells, one can build a small turbine power station working on natural gas.

    • Just say you're generating H2 from water using the same electricity as is used to charge EVs, to neatly head off criticism about the ultimate energy source still being fossil fuels.

      (Yes I know adding more electricity consumption just means we need to burn more fossil fuels to generate that additional electricity. The amount of renewable and nuclear generation is fixed and can't be ramped up on a whim like coal and gas can. So any time you add additional electricity consumption beyond renewable+nuclear g
      • The ultimate energy source is still mostly fossil fuels, just like the power used to charge EVs. But it is a flexible fuel, in the sense that as we start replacing fossil fuel power plants with renewables (and hopefully also with some nukes), hydrogen powered stuff will become greener as well, especially if you are using peaks in energy production to generate and store that hydrogen.

        It's good to start putting a hydrogen infrastructure in place now, even if it still mostly comes from natural gas. And hope
        • âoeCheaper ways to produce hydrogen from waterâ Physics dictates that it will always be more efficient to use the electricity directly. It has to take more power to split the water than you get back by burning the hydrogen. There is no âoecheaperâ way to do it. Those are the rules.
        • > And hopefully an increasing market will spur research into cheaper ways to produce hydrogen from water.

          I remember the first time I wrote that. It was in 1984. So 36 years ago.

        • It's good to start putting a hydrogen infrastructure in place now

          No, it is not. As long as it's less efficient than batteries, it's bad to put it in place now, because it's only going to encourage people to use it. We should only be continuing to develop the technologies now, and we can put the infrastructure in place if it ever actually makes sense. Unlike now.

        • by dgood ( 139443 )

          The ultimate energy source is still mostly fossil fuels, just like the power used to charge EVs.

          Not necessarily. From TFA:

          What’s more, he added, an Azure datacenter outfitted with fuel cells, a hydrogen storage tank and an electrolyzer that converts water molecules into hydrogen and oxygen could be integrated with the electric power grid to provide load balancing services.
          For example, the electrolyzer could be turned on during periods of excess wind or solar energy production to store the renewable energy as hydrogen. Then, during periods of high demand, Microsoft could start up the hydrogen fuel cells to generate electricity for the grid.

      • The problem is that you Chuck a quarter of the input energy in the bin making H2 from water. Then you have all the losses of conversion. How can this ever compete with battery round trip efficiency?
        • The problem is that you Chuck a quarter of the input energy in the bin making H2 from water. Then you have all the losses of conversion. How can this ever compete with battery round trip efficiency?

          It can't. They're hoping it can compete with batteries for capital cost though. Me, I have my doubts. Hydrogen fuel cells use frightful amounts of platinum in their electrodes, which puts a not-insignificant material cost floor on their price. Batteries don't use any material that expensive. There's ongoing research to try to eliminate the platinum, but as of May this year, nobody had done it.

          Then they're going to have to deal with hydrogen embrittlement of storage tanks, which will require expensive r

      • by Dagger2 ( 1177377 ) on Wednesday July 29, 2020 @07:46AM (#60342707)

        So any time you add additional electricity consumption beyond renewable+nuclear generation capacity, it has to come 100% from fossil fuels.

        That's only true in the short term. Over a timescale of months or longer it's not true, because we can build more renewable and nuclear power generation.

        I pointed this out to you 4 months ago [slashdot.org] but I guess you ignored my post or something? Your own post was up at +3 so you don't seem to get modded down every time you point it out, despite missing or ignoring this key insight (which I suspect is the main reason you get modded down).

    • Just a big battery (Score:4, Interesting)

      by Roger W Moore ( 538166 ) on Wednesday July 29, 2020 @07:30AM (#60342657) Journal

      How much natural gas was used to make so much hydrogen??

      None, the article says that they got it by electrolyzing water which takes power to do and makes this project just a big, likely highly inefficient, battery.

      • None, the article says that they got it by electrolyzing water which takes power to do and makes this project just a big, likely highly inefficient, battery.

        No, it doesn't. Nice reading comprehension. Of course you're modded up because everyone knows if someone said it on the internet it must be true. Here, I'll highlight the key words for you:

        It’s possible that an Azure data center could be equipped and run entirely on fuel cells, a hydrogen storage tank and an electrolyzer that converts water molecules into hydrogen and oxygen, Monroe said. These systems could integrate with the electric power grid to provide load balancing services. Further, hydrog

        • No, it doesn't. Nice reading comprehension.

          I said article, not Slashdot summary. You might want to look at your own reading comprehension skills before criticizing others: if may help you avoid looking like a bit of an arse, well next time at least.

          • I read the article, most of it is contained in the /. summary. I quoted that snippet from the article. Please, quote the part of the article where it says "they got [the hydrogen] by electrolyzing water".

            • He cannot because the article didn't say where the hydrogen came from.

              • This article [microsoft.com] (the first one linked) makes it extremely clear that the whole point of this exercise is to extract the hydrogen from water and operate it like a giant battery.
                • Quite simply you fabricated the sources of the hydrogen used in the experiment by claiming the article stated that they used hydrogen sourced from water. The article infers that they could use hydrogen sourced from water. If you think those are equivalent you have a problem.

            • Try this article [microsoft.com] (the first one linked) which makes it extremely clear that the whole point of this exercise is to extract the hydrogen from water and operate it like a giant battery.
              • Your claim was that the hydrogen for this experiment was created from water. There's nothing in that article that supports your claim. In fact, there's quite a bit about Microsoft using steam reformation of natural gas to create hydrogen that wasn't included in the Engadget article. There's also a picture of the trailers they used to haul the hydrogen to the site, if they were testing making it from water "to operate it like a giant battery" why would they need to truck it in? If there is something in t

    • https://horizon-magazine.eu/ar... [horizon-magazine.eu]
      https://www.greenbiz.com/artic... [greenbiz.com]
      trendy ...
  • They made the hydrogen out of natural gas: "They have the ability to make their own hydrogen out of the natural gas feed that they get,” Monroe explained. “They take natural gas, a little bit of water, they heat it up to 600 degrees C, which is the temperature of a hot charcoal fire.” lets hope battery tech improves fast enough to prevent too many more falling for the hydrogen scam.
    • by rossdee ( 243626 )

      so where does the carbon in the natural gas go?
      if its released as Co2 into the atmosphere then theres no point in doing this

      • so where does the carbon in the natural gas go? if its released as Co2 into the atmosphere then theres no point in doing this

        You can capture it and store it or sell it as an industrial gas but basically you are right. This is just an experiment though and it only makes environmental sense in a real installation if they link it with some large scale fixed hydrogen storage and a water electrolysis plant that makes hydrogen from electricity during low price electricity times without releasing CO2.

        • Why go through the additional step of making hydrogen out of electricity instead of using the electricity directly?
          Unless this is only intended to replace their backup generators, in which case fine.
          • Why go through the additional step of making hydrogen out of electricity instead of using the electricity directly? Unless this is only intended to replace their backup generators, in which case fine.

            From the point of view of the data centre it's for backup generators. From the point of view of the electricity grid this would be for storing electricity when prices are low and putting it back when the demand and thus price are high. It's a bit like the Tesla grid scale battery systems. It would have the disadvantage of being slower to convert hydrogen energy back to electricity but the advantage of being able to keep doing it for longer (at the same price). The thing being that it seems that hydrogen

    • by Rei ( 128717 )

      Wait, they're steam-reforming natural gas to produce the hydrogen?

      Then this is crazy. Zero logic to such a backup system. Huge capital costs and wasted energy for nothing vs. just using the natural gas directly. Heck, there's even direct methane fuel cells if you want to go that route...

      • by vyvepe ( 809573 )
        Mod parent informative. Fuel cells running on methane sounds so much better than hydrogen. There are some [gatech.edu] running at about 500 C only. Unfortunately, the efficiency is not stated in the linked article.
        • by Rei ( 128717 )

          It's always been more efficient than steam reforming to hydrogen and then running through a hydrogen SOFC. And natural gas is much easier to store, and it's safer, and it's just.... the idea of reforming to hydrogen and then storing that is just ridiculous.

          Here's a good graph [rsc.org] of system efficiencies from this paper [rsc.org]. They're using (C) - 54,4% efficiency. By comparison:

          74,9% for a CH4 SOFC
          82,6% for a CH4 SOFC with chemical looping combustion
          37,7% + 33,7% = 70,4% for methane cracking (e.g. to H2 and C, rather

      • Benefit of doubt here; The methane-derived hydrogen is just a temporary measure to test the idea of running everything off of fuel cells for an extended period of time. Once the generation part is figured out, they can replace the methane-derived hydrogen with renewable-derived hydrogen as a "Phase 2."

        But yes, currently the largest, cheapest source of hydrogen gas is from natural gas... which is why the people who have been pushing the "hydrogen economy" for the past 20 years or so have been the natural ga

        • They don't get the benefit of the doubt when they have proved time and again they they do not deserve it

      • It makes sense if you start using wind/solar to crack water - assuming all this hardware and hydrogen storage is cheaper than batteries.
        • And the lost 20% of energy electrolyzing the water to make the H2
          • by Rei ( 128717 )

            And then more in the fuel cells converting back to energy. Also the compression / storage losses (and if not on-site, transport).

            • But if their power source for electrolysis is surplus wind or solar generated, their input energy cost is basically zero and it mostly boils down to the economics of battery storage vs. hydrogen creation/fuel cell maintenance.

              My guess is the economics aren't on the side of hydrogen now, but close enough that it's worth spending the money on to see if there's long-term potential there.

              If this was such an economically bad idea with no potential, it's kind of hard to see this level of experimentation/commitmen

              • by Rei ( 128717 )

                No, unfortunately. Even if your power is free, it still affects how much power you get out of it. The figure in question is LCOE (Levelized Cost Of Energy), which includes the cost of all of your feedstocks and depreciation on your capital assets relative to how much electricity you get out of it. Getting less power out for a given capital cost means a higher LCOE.

                • So why are Microsoft presumably investing a lot of money into it if its such a bad idea? Pure research?

                  • So why are Microsoft presumably investing a lot of money into it if its such a bad idea? Pure research?

                    History and inertia. When hydrogen fuel cell research projects were started, it wasn't at all clear that battery research and development, especially production development, was going to progress as well as it has. Back then, it seemed like a good idea to try something else. These days, it's a lot harder to justify, but nobody wants to cancel a big expensive energy project before they're finished just because it's obsolete.

    • I wonder if somewhere there is someone with an idea for producing "green hydrogen" who discarded it, thinking there is no market for hydrogen.
  • A large retail store (Cabelas/Bass Pro Shop) in East Hartford, Connecticut, has been on 100% fuel cell power (800 kW) since 2008. Decade+ later, Microsoft gets hep....
    • Decade+ later, Microsoft gets hep....

      Sure, Windows gets every other virus, why not hepatitis too?

  • by bobbied ( 2522392 ) on Wednesday July 29, 2020 @06:38AM (#60342543)

    Commercially generated hydrogen gas comes from reforming natural gas.

    So, Microsoft likely just generated twice the C02 they normally would have. Nice work guys.

  • It's interesting how just about everybody came to the same conclusion, That almost never happens. I think this is something we should celebrate.
  • by freeze128 ( 544774 ) on Wednesday July 29, 2020 @07:19AM (#60342617)
    1 row of servers in a data center != 1 data center.
  • by tippen ( 704534 ) on Wednesday July 29, 2020 @07:22AM (#60342625)
    Since when does one row of servers constitute a data center? Holy misrepresentation, Batman!
  • Maybe no carbon was harmed by using the H2 power cells, but how much carbon-burning was involved in charging up the power cells to be used? I have trouble believing there's enough naturally-occurring H2 on the ground to mass-produce power cells.

    "It’s possible that an Azure data center could be equipped and run entirely on fuel cells, a hydrogen storage tank and an electrolyzer that converts water molecules into hydrogen and oxygen, Monroe said.
    Right, so how much carbon was burned operating the ele

    • by PPH ( 736903 )

      Right, so how much carbon was burned operating the electrolyzer?

      It depends on where you are. Not much carbon in the Pacific Northwest, since we are primarily hydro powered. Not much carbon if solar/wind power is used to produce H2 when the resource is available. The article implies that the electrolyzer is installed on the data center site. So it has the exact same carbon footprint as the utility power running the center. They didn't make it clear if this was intended to replace diesel generators. Which would make the backup power source somewhat cleaner. But backup pow

  • by 140Mandak262Jamuna ( 970587 ) on Wednesday July 29, 2020 @07:41AM (#60342689) Journal
    It is possible to make hydrogen from solar or wind. Selling the electricity to the grid is more profitable.

    100%of hydrogen sold today is derived by cracking ammonia from fossil fuels using heat from burning fossil fuels. Hydrogen is a con game set up by Oil and Coal to maintain their grip on automobile market by keeping pollution far from the cities. Well, too bad for them, the battery and electro-chemistry advanced too fast for them.

    They managed to get nickel-metalhydride chemistry under patent law bottled up. [wikipedia.org] This is the reason we are using lithium ion today. They kept nickel-metalhydrive starved of R&D funds.

    Well, soon they these patents will expire, or they will be sold off in fire sale prices. Nickel Metalhydride might never overtake li-ion even after that. But some of the electro chemistries not used in li-ion due to fear of litigation will be opened up.

  • So you've got magic black panel that spits out free money for years
    OR
    The lowest chemical energy fuel, an atomic non-chain non-molecule that you basically have to make because it isn't just sitting around anywhere by itself in pure form. Hmmmm.
    • Unfortunately, those magic black panels don't work very well at night. So we'd need some way to store excess electricity for later. Giant banks of batteries work, but have a lot of drawbacks. The best, most stable energy storage tends to be chemical energy (everybody loves hydrocarbons). Gee, I wonder if there's a way we could somehow transform electricity into a chemical energy store of some sort? It would probably have to be something simpler than hydrocarbons, which aren't easily synthesized. Somebody sh
  • That's probably less than 300kW, you could run that off the farts in your office building

  • Let's use up all the water powering cars and servers... and become the second Mars. At least rising ocean levels won't be a problem then.
  • Check out that photo. 3 huge trailers of high pressure H2 tanks to power 1 row of a DC. You'd need huge tank farm for enough H2 to power the entire DC for 2 days. What city / fire department is going to give you a permit for that?!

    Granted, maybe they didn't use the full tanks... but that sure looks like a whole lot of hardware.

  • First off, energy blends should be regionally relevant. Where there are dams, the focus should be hydro-electric. Where there is great wind, there should be wind turbines. Where there is great sun, there should be rooftop solar and solar farms.

    As with all renewable sources, of course, there are imbalances of demand and supply. With enough solar, the west/southwest of the US can overproduce electricity in the day. Wind tends to overproduce at night. Hydroelectric tends to overproduce seasonally.

    Instead of tr

    • Okay. Show us data comparing the energy loss from inefficiency of electrolysing seawater (don't use fresh water, that's valuable!) to get hydrogen for fuel cells versus energy lost as waste heat charging banks of Li+ cells. Be sure to include both the energy/carbon costs of transporting that hydrogen to where it needs to be, and also the energy lost to heat discharging those Li+ cells when the power is needed (because there is waste heat generated there, too, from both the cells and the inverters needed to
  • This is purely a PR initiative.

  • The hydrogen that a fuel cell uses has to come from somewhere, and that 'somewhere' uses more power to produce the hydrogen than you get from the fuel cell.
    Even if you use a front end on the fuel cell so you can use a hydrocarbon source like natural gas, it still takes more energy to get that fuel to the fuel cell than the fuel cell generates, therefore it's not 'carbon neutral', not even close. Laws of Thermodynamics still apply, kids.
    As much as I've been an early enthusiast for the idea of hydrogen fuel

No spitting on the Bus! Thank you, The Mgt.

Working...