Mitsubishi Heavy Industries bets big on small turbines for datacenters Mitsubishi Heavy Industries – Earth's largest source of electricity-generating gas turbines – has tied its future growth to surging demand for datacenters spurred by adoption of AI and new semiconductor plants. The giant corp on Tuesday delivered a Medium-Term Business Plan [PDF] covering the years 2024–2026 that advised …
So now we need to site data centres somewhere there's stacks of available renewable energy next door plus an abundant supply of water to be smashed. Oooh, ooh ... and nowhere that they can be flooded or a tsunami can hit, etc, natch. This should be interesting to watch.
Interesting (my chemistry is a touch out of date since the long ago switch to IT). So there's a couple of nuances to that - a) you're using natural gas as the input (so right back at fossil fuels?, not a big winner), b) it sounds an awful lot like to make this viable you'd need a big scale cracking plant (not next to your turbine) so you're shipping lots of hydrogen around the place - not impossible but as per all the arguments about switching over central heating boilers not that long ago, it's a bugger to work with whether it's pipes or tankers. Even if you ship the natural gas to the site, crack it there it feels like you'd have to have a reliance on a grid that's subject to very similar points of failure as you're trying to escape?
I can see that Mitsubishi will sell whatever customers want to buy. But if they're proposing on site/nearsite gas turbines, driven by hydrogen or hydrogen mix, then customers have the cost and efficiency losses of a hydrogen CCGT. They also need to build gas infrastructure, since for most DCs bulk hydrogen compatible gas connections won't be readily available locally, and that has its own operating costs and new losses. Then there's the whole "where does the hydrogen come from?" which involves some form of power to gas infrastructure, meaning they need to build the generating infrastructure, maintain and operate, compress and store the gas. That's paying for three different energy infrastructures, each with its capital costs, losses, parasitic loads and standing costs.
Technically, all of this is feasible, I've seen all elements done. Whether it's ever going to be economically feasible I can't see.
Plus: Hydrogen - by definition the highest CO2 emitter fuel theoretically possible. Economically, hydrogen is produced by cracking natural gas, using energy from natural gas, not by electrolysis. Hydrogen “fuel” contains the minimum possible number of carbon atoms in a fossil fuel (zero). Most of the energy from oxidising the carbon atoms is simply flared off in the cracking plant.
'Most of the energy from oxidising the carbon atoms is simply flared off in the cracking plant.'
Not the case.
Natural gas is mixed with steam and passes through a heated vessel called a reformer. The gas and steam mix 'cracks' with the vast majority of the H2 being recovered is liberated from the steam. The remaining gas, generally referred to as offgas (heavy in CO) will combust again. It is recirculated and combusted to heat the reformer.
Yes, CO2 is produced but not in the wasteful and inefficient manner suggested.
I agree it’s not quite as bad as the naive version I described, but nor is it as good as your version either.
So, one of the great untold stories of fossil fuel is the fundamental mismatch between the random slush we get out of the ground, and the set of constituents we actually use. Cracking is what resolves that “impedance mismatch”. Forget about hydrogen for a sec, cracking globally uses about 20% of the primary energy content of the input slush we drill. Yes, it’s a complex recirculating process, but that’s the net out. The further off the required output chemical composition is from the input, the higher fraction of primary energy is used in cracking. As 0-ane, Hydrogen maximises the net energy usage in the cracking process - obviously it’s unfair to call it 20% against petrol which itself has cracking losses, and varies depending on oil well source, but 8%ish additional loss is a reasonable global average. Separately, the hydrogen also needs to be compressed/chilled for transport. The theoretical thermodynamic limit for that process loses another 17% of primary energy. So, hydrogen is 25% more CO2 for the same amount of energy, before it even leaves the production plant.
The real problem is this: people always claim “oh yeah, but Green hydrogen will be produced from renewable energy by electrolysis”. Maybe so, but then there is no argument for Hydrogen at all! Because then it just becomes a carbon-neutral cycle of central-production-local-oxidation, irrespective of how many carbons it has. So the correct question is: is Hydrogen a good working fluid hydrocarbon? And the answer is No, the optimum working fluid hydrocarbon would be one that is liquid at room temperature, which limits it into the hexane-octane range. Petrol.
This has real consequences. The entire human ingenuity, resources and cost that have been spent on electric vehicles and infrastructure has been entirely wasted. It’s a null cycle, achieving nothing but wasting 30% of primary energy.
We should have spent the same R&D and scale up resources on research into large-scale carbon capture CO2-to-hexane, fuelled by renewables. That *would* have saved the planet. Except, that’s a technology we sort of already know. It’s called biofuel. If you want to make it efficiently (ie without yet more uncosted fossil fuel usage via fertiliser) you need to improve on the photosynthesis cycle, by genetic engineering. We could have taken that direction, there’s really nothing that beyond science available fifteen years ago, all the component technologies are moderately understood. It’s certainly *much* easier, more predictable, lower risk, less expensive development, safer, lower environmental impact than Battery technologies.
'I agree it’s not quite as bad as the naive version I described, but nor is it as good as your version either.'
It wasn't a naive version you described, to be blunt, it was completely misleading. I wasn't aware I painted a particularly rosy picture either. I just stated the facts on how H2 in the real world is currently generated on an industrial scale.
The rest of your post? Frankly I don't have the time nor inclination to shoot it down other than to say it may superficially look impressive however having read and re read it, I realise it's just word soup with a side of numerics.
> The entire human ingenuity, resources and cost that have been spent on electric vehicles and infrastructure has been entirely wasted.
But but but..... Tom Swift and His Electric Runabout (1910):
“Hum,” remarked Mr. Swift musingly. “I don’t take much stock in electric autos, Tom. Gasolene seems to be the best, or perhaps steam, generated by gasolene. I’m afraid you’ll be disappointed. All the electric runabouts I ever saw, while they were very nice cars, didn’t seem able to go so very fast, or very far.” “That’s true, but it’s because they didn’t have the right kind of a battery. You know an electric locomotive can make pretty good speed, Dad. Over a hundred miles an hour....” “Yes, but they don’t run by storage batteries. They have a third rail, and powerful motors,”.. “But it does seem to me that if you put the right kind of a battery into an automobile, it could scoot along pretty lively. Look what speed a trolley car can make.” “Yes, Tom, but there again they get their power from an overhead wire.” “Some of them don’t. There’s a new storage battery been invented by a New Jersey man, which does as well as the third rail or the overhead wire.”
“I think I’m going to change it, and add some lithium hydrate to it. I think that will make it stronger.” “Bless my watch chain!” exclaimed Mr. Damon. “I’m going to have it fixed so I can take current from any trolley line.”
The entire human ingenuity, resources and cost that have been spent on electric vehicles and infrastructure has been entirely wasted
Shsssh you! No one is allowed to suggest practical solutions that don't cost the public, directly and indirectly (as a tax payer), vast sums of money for questionable net gain, and/or drastically reduce their level of comfort, convenience, and ability to travel great distances at will.
and they are.
Pull up a utility routing map that shows high-voltage main lines, high-capacity natgas pipelines, and fiber cable routes.
Now draw a 10-25 km circle at likely intersections.
Those are where mega data centers have been built the past 30 years and will continue to be built.
Any waving of hands with "what about the supporting infrastructure" is spurious at best.................
A physicist friend who is well into plasma got head hunted by that Aldermaston research shed, oh well over ten years ago now. He outlined the gas engine to me some time before that and I am wholly surprised NOT that we still do not have these things generating electricity in the home.
There are numerous myths still shrouding the subject. For a start, academics still assume that you have to compress the stuff before it's useful. The greedy always want to see this implemented "at scale", and including "mega power" suppliers and the inevitable power "grid".
Whist we are stuck with development being driven by greed and profit, we will never see the advantages or benefits at consumer level, unfortunately. However, I do believe my mate Neil about the efficacy of small, local, electricity production, and maybe that's why he's disappeared into a UK research facility many ears ago...........
ALF
Where does this gas come from to power the domestic gas engines?
And if they're to realise the benefits of "small, local, electricity production" they'll need something to share capacity between properties (assuming there's no madcap ideas to size domestic units to meet peak island demand with no grid connection)....oh, hold on, so they do need a grid. But it will be carrying local balancing loads until it can't, and the whole grid cost then gets spread over fewer units and....nahh, this idea of generating your own power, it's pants.
As for "academics still assume that you have to compress the stuff before it's useful", what a load of bollocks - gas is compressed for storage and transport, because the asset cost is otherwise unfeasible.
"Whist we are stuck with development being driven by greed and profit, we will never see the advantages or benefits at consumer level,"
What utter, utter garbage.
Where does this gas come from to power the domestic gas engines?
Hydrogen! Just ignore the costs and efficiencies in creating 'green' hydrogen and then distributing and storing it. But I suspect the hydrogen part of the Mitsubishi release is greenwashing given turbines can be multi-fuel.
As for domestic. Problem is cost, space, and attempting to retrofit any new tech into existing properties. I'm currently having a house designed though that is built around a Stirliing engine/generator with multiple heat sources and stores. So solar PV, large hot water tanks, incinerator/wood burner and an indoor swimming pool. But this isn't cheap, and the house has to be designed around it. Plus it's been fun dealing with some pseudo-myths. Solar PV fed into hot tanks is cheap and cheap to maintain. Resistive heaters don't care if they're being fed AC or DC. But this is penalised in the UK and our 'energy efficiency' ratings because of heat loss. I don't really care about that because tanks are pretty well insulated, and if the heat's lost within the structure, it's not really lost. I'm going to utilise that. Downside is most modern homes in the UK weren't designed with space for hot water tanks, so can't use this as an energy store or 'cheap' hot water.
"Resistive heaters don't care if they're being fed AC or DC."
The simplicity of using a water tank to act as a heatsink is a big advantage, especially in winter when there's a need for lots of heat. In the short term, I'm planning to install a tankless water heater running on propane as I got one very cheap and my current hot water heater is nearing the end of it's useful life. The next iteration will be a thermal battery added to the system to store heat so I don't need any or as much propane for heating water. A thermal battery can be better than just heating and storing hot water since a medium can be chosen with better characteristics. The cost to rebuild the house around an optimized energy fabric isn't feasible. There would be no ROI on that and it might only make sense for a completely new build. It's like the "smart homes" with so many automated gadgets that the build cost is astronomical and the companies that built the gadgets don't often stay in business or continue to support the products long enough to make them worthwhile.
It's like the "smart homes" with so many automated gadgets that the build cost is astronomical and the companies that built the gadgets don't often stay in business or continue to support the products long enough to make them worthwhile.
Yep, this is part of the challenge. Or FUN! The design I want is to keep things as simple as possible, with as few moving parts and gadgets as possible for that reason. It'll be in Alaska, which has it's own set of challenges and advantages. Like having a walk-in freezer just by building an old-skool ice house in the permafrost. Then kitchen freezer won't need to be as large and won't need as much energy. There'll be propane for feeding the Sterling engine if needed, and cooking, assuming gas stoves are still available. Which being Alaska, the probably will be. But also fending off people trying to flog me heat pumps because I'm not sure I'll need one. There's plenty of wood on the land for fuel, plus there'll be plenty of offcuts and sawdust from my attempts to make furniture.
But I've always been a fan of heat stores, which was another advatange of older homes with hot water tanks. Use those as airing cupboards or for raising bread, fermenting booze etc. Heat is not wasted if we find a use for it. I've never owned a clothes drier for pretty much that reason. Why would I want space or waste energy doing something that can be done naturally, or as a byproduct? But that's also current challeng. Ventilation. There's been news articles in the UK about people having problems with black mould after insulating their homes. Which means those homes aren't being adequately ventilated if condensation is building up. Not sure how much of that is user error, ie not ventilating or opening a window when cooking. Idea for that is to have fresh air intakes going via the heat store so I can use heat exchangers to warm it. Which has also been FUN!, like being sold on fancy heat exchangers, and asking why I couldn't just use truck radiators instead.
"I've never owned a clothes drier for pretty much that reason."
At my last home, I air dried my laundry, but where I live now (owned) there could be an issue with two-legged rats helping themselves to things off the line. I'm waiting for the lot next door to go up on county auction (maybe next year) and I'll put up a fence around the yard/garden that will impede anybody just walking up and grabbing a pair of (expensive) jeans or having a bit of fun by flinging things in the dirt. In the summer, many things will dry so fast that once I'm done hanging up a load, I can start taking some things down. Jeans take a while and those are often a target. In winter, it can take hours for things to dry even on a sunny day so I'm not keen on deleting the dryer all together.
Scale almost always beats smaller. At the most basic for heat engines, surface to volume ratio decreases with bigger so less heat loss. Also friction losses are relatively less for bigger moving engines - another surface to volume effect. Ancillary equipment often doesn't scale similarly so the same clever controller works just as well for a big system as for a household or community setup.
And it may have uses in some cases/some places but it's no magic bullet (because if there was a magic bullet somebody would be making billions of it right now).
I'm all for local/less centralised power generation closer to the point of demand because it ticks a lot of boxes in terms of resilience/not having to build and maintain massive amounts of grid capability/not ramming up wind turbines in rural parts hundreds of miles from where the demand is (because subsidies and free money).
But hydrogen and ammonia - tricky customers at the best of times and as already mentioned by another commenter, they need a significant amount of supporting infrastructure (per site) to be safely handled. As for electrolysis - at a time when large parts of the world are generally dealing with too much (floods), not enough (drought) or bouncing back and forth between the two somebody rocking up and saying they want to steal some of it to generate electricity is going to make for a fun conversation (plus where are they putting the damn wind turbines to feed the electrolysis process in the first place?).
"I'm all for local/less centralised power generation closer to the point of demand because it ticks a lot of boxes in terms of resilience/not having to build and maintain massive amounts of grid capability/not ramming up wind turbines in rural parts hundreds of miles from where the demand is (because subsidies and free money)."
In which case, demand has to be built where there's power to be had, rather than hoping to use often meagre local resources to fit demand that's evolved in relation to the economy and policies of the past few hundred years.
It sounds nice but I seriously doubt the funds will be given to clean up their emissions as well as a larger generation plant; small producers will probably balk at the increased cost of their microgenerators once, say, EU-required post-burn treatments are factored in. I fear these microplants will only add back to the carbon footprint which we've been working so hard on controlling.
I'd imagine the belief set here is that "surplus" renewable energy is used to convert electrical energy that we can't efficiently store into chemical energy which we can. There's a critical assumption here of assuming that there will for a long time be a lot of renewable power that's near enough free. Unfortunately, the concept of surplus renewable output being free is merely a reflection of low levels of battery storage and a 1960s design of power system. Hang enough BEVs on the grid, and invest in grid reinforcement, and suddenly the days of "surplus" renewables are suddenly gone, and renewable power always has a positive price. The same threat applies to historic oddities like electric storage heaters linked to simple time of day tariffs - they're living on borrowed time, and the whole concept of very cheap off peak power should vanish.
Germany are planning on importing ammonia from Canada. The ammonia will be made from power from offshore wind turbines off the coast of Newfoundland. Newfoundland have a big offshore oil industry, and they see ammonia from wind as a long term industry to replace oil when the oil fields dry up. The island of Newfoundland is relatively isolated so there aren't good alternative markets for the wind power electricity otherwise. The German PM was in Canada last year signing deals with his Canadian counterparts on this subject, so it has high level support in Germany.
Ammonia (NH3) can be turned into hydrogen relatively easily, or it can be burned directly in turbines. Mitsubishi have outlined both options in the above article.
Personally I'm not convinced that ammonia is a great idea as a synthetic energy material. A better solution for countries like Germany or Japan would be nuclear power, but there are political roadblocks in that respect. There's not much support for using ammonia for electric power in Canada itself, they're very big on nuclear being the future. They're happy enough to sell ammonia to others though.
Canadian wind-powered ammonia is certainly an energy source, but that adds various further complications. There's the basics of ocean gas transport - so a long and potentially vulnerable physical supply chain, the cost of which is linked to cyclical world gas transport markets. Then we've got currency risk for somebody in this mix between the Euro and the Loonie. There's also a complex question about who's investing in the wind farms, who owns the output gas and whether Germany is (as likely) going to have to pay the market price for renewable ammonia. The two main German companies involved (Uniper and E.ON) have an appalling record of adapting to strategic change in the energy market, and the German government an even worse track record - I used to work for E.ON before and during the Uniper demerger, I'm very familiar with this. The German government had to nationalise Uniper to avoid its collapse. and quite remarkably, the same German government needed to make a €7.5bn bailout of the Siemens Gamesa wind energy business (how?). I'm guessing Germany and its energy sector are committing to something akin to a take or pay at market price deal, completely de-risking the investment for Everwind, whilst yet again loading vast long term risks on to German shoulders.
"Personally I'm not convinced that ammonia is a great idea as a synthetic energy material."
There are advantages to Ammonia in some use cases. Ammonia production is currently one of the biggest users of electricity in the world. Besides it's use in fertilizers needed to be able to use soil to turn petroleum into food, it's a handy precursor for many other chemicals.
Ammonia can be made into a "fuel", but it needs to be thought of as a storage medium since it takes a fair whack of energy to create. That's not to say it's worthless. It may be better to use an intermittent power source such as wind to make Ammonia and be able to remove that amount of production from grid sourced power.
If between onsite PV and grid supplied "excess" wind/solar (which is often cheap during e.g. sunny but not hot days and occasionally even priced negative) you can make enough for 24x7 operation you're good. You have a natgas feed available as a backup you're caught short or the grid is temporarily offline. If both the grid and the natgas are offline at the same time you've probably had an earthquake and that's why datacenters have redundancy.
Before someone says "but what happens when there's no longer an excess of green energy on the grid" having consumers for the excess will result in more being supplied.
The idea is that natural gas will be phased out and replaced by ammonia. There will be no natural gas available to act as a backup.
Natural gas requires significant capital investment in exploration, production, processing, transport, and distribution. Once natural gas consumption falls below a certain level, that whole supply chain goes away and it won't come back. For example, a single LNG export plant typically costs $40 billion. Nobody is going to make those sorts of investments in order for them to sit idle just in case they are needed now and again. The same goes for pipelines.
You also can't have LNG sitting in a tank indefinitely, as it continually boils off. You either need to be using it all the time, or it will be lost anyway (and methane leaks are worse than C02 emissions in terms of greenhouse gas effects).
You could try storing very large amounts of liquid ammonia, but that is a very dangerous material to keep around in quantity. It's rated as "VERY TOXIC, can cause death" by Canadian authorities. People die from ammonia leaks on a regular basis, and that's just with small scale use such as in refrigeration. A medium size ammonia leak at a food processing plant in Senegal killed over 100 people and injured over 1,000.
If you are going to keep ammonia in tanks at a data centre as a backup source, then the hazardous material aspects of this will be interesting, to say the least. I can't see those data centres being too popular with their neighbours, assuming that they get permission to build at all.
If you are dealing with ammonia on an industrial scale, then the risks can be mitigated by siting the plant in a remote area with fewer people around and a large exclusion area around it. That however is not a great solution for data centres.
The realistic answer for data centres is to locate somewhere which has very reliable utility power. For most countries that implies nuclear power once natural gas and coal plants have shut down.
I'm reading that ammonia burns, but not as well as hydrocarbons. The burn isn't as fast, so it's harder to sustain. The burn isn't as hot (about a fifth), so it's less useful for power generation. Ammonia combustion doesn't produce CO2, but it does produce NOx, so it's just replacing a pollutant that is effectively plant food (yes, it's a GHG, but let enough plants and trees grow and it's a self-solving problem) with a pollutatant known to cause acid rain and breathing problems.
All in all, it doesn't sound any better. At best, just different. In many ways, worse.
Cost of downtime * amount of downtime - cost of installing and maintaining backup generation = decision.
Even with amount of downtime measured in the 0.00001% range it still favours having the additional backup over grid.
Of course, not all grids are created equal and have five-nines reliability stats, so, YMMV.
No shortage of reciprocating engines available to run on a range of fuels, just as turbines can be run on a bunch of fuels.
MTU / Rolls Royce catalogue alone offers up dozens of options. The advantages of piston and rotary engines are well understood and so the choice boils down to application. But I dispute the efficiency claim (and, more to the point, so does 150+ years worth of maths). A piston literally throws away energy on every stroke as work has to be done to flip the piston's direction again-and-again.
There are still cases where piston makes more sense than turbine of course. Turbines in cars did not work terribly well because they are more about providing continuous power rather than variable power on tap. See the crazy Rover experiment in the 1950's...
Hydrogen makes zero sense. It's inefficient to create it, and there's very little infrastructure for it. It seems to be the go-to buzzword for those who love fossil fuels but need some green-washing for PR. The same niche filled by "clean coal" about 20 years ago.
In the US at least, the practical option would be natural gas turbines for electrical generation. with absorption chillers running off the waste heat:
https://www.youtube.com/watch?v=Q0tsDsVC21o
The only downside being the lack of green-washing that "hydrogen" buys you, but you could do a lot more actual good by complimenting the gas turbines with as much solar PV as you can reasonably install, to reduce the need for natural gas.
>It's inefficient to create it
Gasoline powered cars achieve an overall well to wheel efficiency of at best 15%.
And that is 2 or 3 times better than back when "the system" was developed and deployed at scale.
Low efficiency has simply not stopped energy systems being deployed at scale in the past, and there is no particular reason to think that is has to in the future.
Hydrogen has lots and lots of issues, but efficiency is not a show stopping one in it's own right.
Hydrogen (green) is also likely to happen at scale as it is currently the most obvious option for multiple large scale chemical processes. This would mean that it becomes available as a byproduct of chemical industries i.e. it becomes available for use as an energy carrier without the energy industry having to fund and develop the manufacturing infrastructure,
steam engines (locomotives not modern steam turbine plants) were even less efficient. But coal was cheap and there were no suitable alternatives so we used them. That's very different from taking expansively acquired energy and storing it in probably the most impractical and inefficient means available. Hydrogen is indeed a good candidate for some industrial processes, such a steel making, but it's a fuel (or feedstock). It goes in but it's not coming out, at least not for free. (What comes out is usually good old water which is why though processes are desirable.)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
