back to article Plans to heat districts with datacenters may prove too hot to handle

Using waste heat from datacenters for district heating makes sense from an environmental standpoint, yet there are implementation challenges and potential pitfalls introduced with any government regulations covering it. Digital trade association TechUK has published a report on using DCs as part of district heating networks …

  1. Anonymous Coward
    Anonymous Coward

    HS3

    More useless technospout from the technology incompetent British government. Having screwed up the HS2 rail link, they are moving into HS3 - Heating Stupidity 3, all in the name of green, cheap heating. Short of placing DC's at the centre of habitation area's this idea seems a non starter.

    Maybe replacing the Westminster parliament with a large bit barn might be an improvement, although whether it would produce much more hot air than that from the current infestation is debatable.

  2. Lurko

    Locally it can work

    But it's negligible in the big picture. Using projections in an NG discussion document, we've probably got about 10 TWhe of DC energy usage. A third of that you probably can't use due to seasonality of heat demand (down to say 8 TWhe), likely net efficiency of a real-world heat recovery and distribution system is about 60%, so that's say 5 TWh heat output, compared to current heating energy demand of around 700 TWh. Even if through substitution and insulation we could halve current heating energy use, the DC contribution would be 1.4%. And then you can slash that for those DC where there's no local heat demand, or where it's simply uneconomic to capture and supply.

    1. Jellied Eel Silver badge

      Re: Locally it can work

      And then you can slash that for those DC where there's no local heat demand, or where it's simply uneconomic to capture and supply.

      You're missing the point. These 'ideas' aren't meant to make sense, they're meant to make money. Hence the special pleading for something in the budget that will shuffle money in the direction of these 'solution' providers and they'll make bank. So the suggestion that DCs might be forced to install expensive and pointless heat capture devices. It's much like the 'Renewable Heat Incentives' that lead to this-

      https://en.wikipedia.org/wiki/Renewable_Heat_Incentive_scandal

      Another whistleblower wrote a letter in January 2016 to tell Foster, who had by then become First Minister, about an "empty" farm shed that was "being heated for the subsidy". The scheme did not take into account that properties that were not previously heated could now be heated for a profit.

      1. Claptrap314 Silver badge

        Re: Locally it can work

        Just like the Cobra Bounty program in India...

    2. NeilPost

      Re: Locally it can work

      It always was nuts. Build some greenhouses next to them on colder climates and grow tomatoes etc.

      AWS Tomatoes for Amazon Fresh.

    3. Anonymous Coward
      Anonymous Coward

      Re: Locally it can work

      The solution is simple. You break up the massive monolithic datacentres and you create tons of smaller datacentres in empty retail space in town centres. Not only will this lead to more resilient and distributed architecture, it will also provide infrastructure for all the businesses around those "town centric" mini datacentres...thus reducing costs for small businesses...this will in turn make it less complicated to distribute waste heat as the waste heat won't be concentrated, it will already be distributed...it will also reduce strain on the power grid by spreading the load out rather than concentrating it.

      You also make government IT projects orders of magnitude cheaper because the base level infrastructure for a national IT system would already exist...it would just be a matter of deployment and scaling...because the majority of the infrastructure would already exist and would be maintained by engineers that are already relatively local to the "datacentres"...you also don't need massive costly "cloud" contracts if your country already is a "cloud" of sorts...you can set up twinning programmes to ensure failover in the even of local issues...so twin Staines with say Warrington...anything hosted in Staines is also mirrored to Warrington...so if the Thames floods in Staines or the power goes out in Warrington, a mirror kicks in at the opposite end of the country...no big deal, business continues...

      It boggles the mind that we can get parks, leisure centres and shopping centres galore...but god forbid if we want infrastructure to start businesses, distribute national infrastructure and localise access to tech...peasants should be frolicking around in parks and spending money at Nandos before going to the new cinema, we can't give them cheap and easy access to local infrastructure...they might get wealthy and challenge the elites!

  3. Roland6 Silver badge

    So things change

    Just today cars are advertised on emotive fluff, mobile phones on their camera (and photo manipulation capaibilities), we can see data centres will be sold not on the actual need for computing power but for their heat output. Naturally, this will be given a greenwash, so that people don’t ask the awkward question about electricity production etc

    So that Ring (whatever) home automation system, that requires cloud servers, will be using AI to ensure sufficient premises are consuming heat arising from its cloud data centres.

  4. Vikingforties
    Facepalm

    Lies, damned lies and ....

    "Servers typically have a lifecycle of three to five years, and technology improvements generally reduce the overall heat they produce."

    I suppose that's a correct statement per sum, but the number of sums being done is going up way faster than the energy required to do them is going down. Similar can be said for the power required to transport around the answers to those sums.

    1. Lurko

      Re: Lies, damned lies and ....

      Which begs the question, if either power or CO2 are matters of importance, are all those calculations actually adding any value to the world? I'm thinking mostly not, whether we're talking digital currencies, AI, or self driving cars (or for that matter, AI-generated pr0n, of which there seems to be an abundance, allegedly).

      1. Vikingforties

        Re: Lies, damned lies and ....

        Added to that, the inefficiency of some languages and libs. It's a winder anything productive gets done.

    2. Dimmer Silver badge

      Re: Lies, damned lies and ....

      "Servers typically have a lifecycle of three to five years, and technology improvements generally reduce the overall heat they produce."

      That is certainly correct when you go from spinning rust to SSD.

  5. richardcox13

    Balanced: Unlikely...

    > Digital trade association TechUK

    A trade association represents the interests of that trade and should not be seen as anything but biased. I very much doubt those building bit barns want to do anything except maximise their return on investment. Any additional cost are something to fight.

    OTOH to make something like this work takes political leadership to balance the medium term costs with long term benefit (in this case long term is over decades). And when has that ever been seen in the UK in recent decades?

  6. Anonymous Coward
    Anonymous Coward

    So are we saying the most logical approach would be to build datacentres to replace already full datacentres which isn't possible due to downtime and different infrastructure implementations? Maybe the way forward is to build datacentres going forward that can be easily converted once full in locations cold all year round along side housing that can also be converted at a later date. Still not going to solve the reduced heating needed in the future and the already cold location further reducing the heat already. I don't think there's an easy solution to this. Maybe build blocks of flats with the data centre on the ground floor balanced alongside traditional heating.

    1. Richard 12 Silver badge

      Retrofitting is incredibly expensive.

      District and communal heating systems need hot water (or steam) pipes in the ground between the bit that makes the hot and the places that need heating.

      The places that need heating need a connection to the trunk - more underground pipework - and a heat exchanger.

      Datacentre heat is fairly low-grade, so the homes may also need heat pumps to raise the flow temperature and/or be otherwise designed to be heated by low flow temperatures.

      Hot washing water needs to be at least 55C, so they also need some way to generate that.

      It's only really feasible if all the infrastructure is installed when the places are being built.

      1. Jellied Eel Silver badge

        Hot washing water needs to be at least 55C, so they also need some way to generate that.

        I'm keeping an eye on stats for legionella. Hot water should be kept at 60C to prevent that, 'energy saving' might be encouraging conditions for that to breed instead. Plus there'll be all the black mould from people over insulating homes, and not properly ventilating.

        1. Richard 12 Silver badge

          Regulations vary, some places say 60C, others 55C.

          Either way, it's considerably hotter than any reasonable datacentre "warm" output flow temp would be by the time it reaches a heat customer.

      2. munnoch Silver badge

        In other words, all the things that we commented on when El Reg originally reported the proposal a few months ago.

  7. Phil O'Sophical Silver badge

    This means heat exports during the early years could be light, creating uncertainty for the stakeholders in a district heating network.

    I'm not sure that's a valid argument, if you consider the whole picture.

    An apartment building will need to be heated, say from a heat pump, and will consume a certain amount of electricity which can be known/planned. If the data centre is empty, then all the apartment heating comes from the local pump, but as DC capacity increases and heat generation rises the flow to the apartments will increase, and their electricity consumption will drop. At the same time, the DC consumption will increase. Rather than consider DC heat ouput as "waste" which can be used, consider the DC + apartment block as a closed system. Worst-case the apartment heat pump will be oversized at the start.

    I'm still sceptical about the overall idea. I know people who live in buildings with a district heating system which is so uncontrollable that they adjust temperature by opening windows, even in midwinter, which is hardly an environmentally-friendly solution.

    1. rg287 Silver badge

      Rather than consider DC heat ouput as "waste" which can be used, consider the DC + apartment block as a closed system. Worst-case the apartment heat pump will be oversized at the start.

      I'm not sure you'd want to have a direct, or one-to-one relationship between source and consumer. The DC would be one source running into a district heat building which can also have a ground-source heat pump or even gas boiler to "top up" in extreme cold. That heat building can then supply multiple apartment buildings/schools/swimming pools/office buildings as required. In summer, it can also pump excess heat into bore holes as a store for winter.

      By architecting on a modular basis, you can grow the district - as Southampton have done with their network, which started off just running the council building and has since been adopted by retail sites, Solent University, the Hospital and residential developments.

      I'm still sceptical about the overall idea. I know people who live in buildings with a district heating system which is so uncontrollable that they adjust temperature by opening windows, even in midwinter, which is hardly an environmentally-friendly solution.

      This is of course anecdotal. Anecdotally, I know someone who lived in an apartment on the Southampton network and it worked very well. I also know someone living in a "conventional" apartment in another part of the country who hasn't turned their heat supply unit on in two years because some combination of modern insulation (certified not grenfell-like!) and his neighbours having their heating on leave his apartment too hot as it is. Bad heat engineering will give bad results, whether you're working on district heat or per-residence boilers.

  8. abend0c4 Silver badge

    Unfortunately, the biggest obstacle is regulation

    Communal heating systems are not presently subject to the same rules as individual utility provision. People on communal systems don't have the right to change providers and don't benefit from the energy price cap. They also often have to rent their heat-exchanger from the energy provider. This is kind of inevitable given the nature of the beast, but it basically means people doing the "right thing" environmentally are often the effective prisoners of a monopoly supplier operating at a relatively small scale and charging higher prices than the market average.

    If you want to encourage a particular outcome, then you really need to get rid of the perverse incentives to do the precise opposite.

  9. DS999 Silver badge

    People may not want to live next to a datacenter

    And heat will be lost transporting it a distance from there.

    Seems easier to site something that requires a lot of heat next to it. Put a cement plant next to it, sized to use the excess heat from the datacenter. By using heat that would have been released into the atmosphere in a standard cooling solution that heat is not only cost free but zero carbon, so the cement costs less to produce (there will be some cost associated with concentrating the heat to the required temperature, of course) and you can probably charge a premium for it.

    1. gratou

      Re: People may not want to live next to a datacenter

      Or greenhouses

      1. DS999 Silver badge

        Re: People may not want to live next to a datacenter

        The need for heat for greenhouses, similar to the need for heat for houses, is highly seasonal. For most of the year you have a bunch of heat with nothing to do with it. That's why you want something that can use that heat 24x7x365.

        1. Ball boy Silver badge

          Re: People may not want to live next to a datacenter

          You need to find a business that has a need for a constant supply of mildly warm air - a launderette sounds like the ideal candidate. Then *they* need a client that demands cool, moist air. At this rate, all DC's will have to built adjacent to a room full of tumble driers and a mushroom farm...

          I'm pretty sure I'm taking the piss - but with Westminster being what it is, don't blame me if this gets written into an Act ;)

          1. DS999 Silver badge

            Re: People may not want to live next to a datacenter

            A datacenter would produce WAY too much heat for a laundromat, other than maybe some sort of huge laundromat that's doing all the laundry for a major hospital or something.

            No datacenters are doing pure air to air cooling, so their "mildly warm" output will be in the form of water or refrigerant like R134a. It is fairly easy to concentrate that heat to far higher temperatures if you need something in multiple hundreds of C above boiling for a particular process.

  10. Helcat Silver badge

    I'm darn sure politicians produce more hot air than data centres, so why don't we capture all the political hot air and use that instead?

    Cheaper, cleaner and...

    oh, wait, it's not cheaper or cleaner, is it. Darn. Back to the drawing board...

  11. rg287 Silver badge

    Matching supply to demand is a tricky issue, but not intractable. It's certainly easiest to do it at build time, but retrofits are possible. Southampton built their geothermal-based heat network on the basis of the council being the first (internal) customer. West Quay, Ikea, the Dell re-development and Solent University came along later.

    As far as seasons go, the trick is likely to have things like a swimming pool on the network which needs year-round heat, and can act somewhat as a buffer by saying "sure, we'll go from 19 to 21 degrees", which is quite a lot of heat in those volumes of water.

    Seasonal heat storage is also a possibility. The Drake Landing Solar Community in Alberta manages to power themselves ~97% from solar, by pumping hot water into boreholes over summer and using the heat in winter. Very occasionally they have to top up the heat with gas. Perhaps most impressively, this is all off domestic roof solar as well. Sink some deeper bore holes and you could soak up a lot of spare heat in the summer.

    The main thing here is that you're not architecting this so that data centres directly pump heat to homes. There needs to be an intermediary that accepts hot water, can pump it into storage, into a supply ring, or even vent to atmosphere if necessary. The actual source of the heat is transparent to both user and supplier - the intermediary buffers any mismatch.

    To use the Slough example, construct a heat building, get boreholes sunk for ground storage and trunk pipes laid under the roads. As datacentres replace or upgrade cooling facilities, they can tee into the trunk and dispose of their heat (since they presumably have redundancy, and chillers are mechanical hardware with a lifespan shorter than the building - just like the servers inside). This gradual growth of capacity will match the gradual growth of connected sites - schools, hotels, leisure centres, new-build residential developments, etc. It's not strictly necessary to do a big-bang switch over. A DC could be sending some of their heat to district and some to atmosphere if they've only changed one chiller.

    Given that chunks of Slough are already supplied with steam and hot water from the Slough Heat and Power CHP site (now being redeveloped by SSE), it doesn't seem like a stupid idea for Slough Multifuel to be not merely a supplier of heat, but an exchange/buffer facility, architected to receive hot water from datacentres for storage and redistribution.

  12. bernmeister
    Holmes

    Pipe Dream

    Its just a pipe dream. A lot ot pipes and a lot of dreaming.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like