This post from Rivan, a hydrogen startup, makes a few key points that I think are really interesting in the context of the research into fossil-free molecules you might need for aviation, shipping, all year round power generation, and a fossil free internet. I’ve mainly written for my future self as much as anyone else, but it may be of interest to you too.
For someone researching a fossil free internet, this post is relevant and useful because datacentres are typically designed to be able to run 24 / 7. Even if the power required by a datacentre might change at different times of day, there is still an expectation that power is always there, because the cost of kit (HVAC, and servers, etc.) in datacentres is so high that you need a minimum number of hours of availability to make them economically feasible to operate.
What does hydrogen have to do with this?
Anyway, hydrogen is important in this context, as it’s seen as one of the key inputs for creating fuels from non-fossil sources that can provide this minimum amount of availability – either in the form of hydrocarbon fuels like diesel, methane and so on, or in the form of hydrogen itself for fuel cells or direct combustion. Almost all hydrogen used right now comes from fossil sources, which means carbon emissions, so ideally you’d have some source of non-fossil, green hydrogen atoms for making these fuels.
So, what were the key insights about making green hydrogren that made it worth writing this post in the first place? Let’s have a look:
- the cost of energy is the key input for making green hydrogen at a price that could compete with energy from fossil molecules – you can’t just rely on electrolysers getting cheaper on their own to make green hydrogen, as they will always rely on electricity being generated, and each unit of green hydrogen needs lots of electricity.
- ultimately the efficiency of electrolysers is not as important as the their cost of capacity, because the most efficient electrolysers are also the most expensive. This pushes up the total cost so much that producing hydrogen at the required price point becomes impossible
- the cost of producing electricity from solar is falling, and in certain places, it is low enough that you can get away with relying on 4-8 hrs of daily hydrogen generation from electrolysis. This is because even that figure is a large enough number of annual hours to spread the capex over and make the economics work. You can use batteries and so on to increase the total hours of daily generation, but it increases the capex so much that the total cost of generation ends up too high.
A few helpful tables
This table halfway through that linked Rivan post is indicative of the second insight I’ve written above – that the cost of capex is one of the key important factors once you have cheap electricity.
| electrolyser technology | cost of capacity | h2 produced | energy cost for h2 | total h2 cost |
|---|---|---|---|---|
| Solid Oxide | £4000/kW | 300.00 kg | £0.40/kg | £13.73/kg |
| PEM | £1000/kW | 266.67 kg | £0.45/kg | £4.20/kg |
| AWE | £200/kW | 218.18 kg | £0.55/kg | £1.47/kg |
Here’s a key quote, when talking about how the cost of the electrolysers affects the cost
The outcome is that any capex reduction >2% for every 1 kWh performance loss improves £/kg H2, all the way to 10% where you slip into <£1/kg H2 with £72/kW capex and 75kWh/kg efficiency. These are just indicative numbers and can be way more aggressive for lower-efficiency architectures like AWE. This again shows that any further gains in electrolyser efficiency are actually counterproductive to cheaper H2 if accompanied by capex increases, and instead the focus should be entirely on using as much cheap energy as possible.
What does all this Solid Oxide, PEM and AWE stuff mean? Broadly speaking they refer to different kinds of electrolyser that you would use to turn water into hydrogen and oxygen. You can refer to this GenAi generated summary for a little more info, but the the short version is that Solid Oxide refers to Solid Oxide Electrolyser Cells, (which are expensive, efficient, slow to ramp up and down), PEM stands for Photon Exchange Membrane (which are less expensive, less efficient, easier to ramp up and down), and AWE stands for Alkaline Water Electrolyser, (which are much cheaper, less efficient, and less easy to ramp up and down than PEM).
Why doe the ramp up / ramp down behaviour matter?
The ramping matters because if you want hydrogen to be green, you need to be very very careful about how clean the electricity is that is used to generate it. This is because generating 1 kilogram of hydrogen might need around 40KWh of electricity, and carbon intensity of electricity changes based on what’s used generate it.
So if you’re using grid electricity, you only need a few kilowatt hours of even fairly low carbon intensity electricity before it’s worse, carbon emissions wise, than regular hydrogen from fossil gas. If you’re using clean energy like wind or solar, you don’t have direct control over the wind or the sun, so you can’t rely on it being constant for days on end either – so having something that ramps up and down well can be helpful.
At this point, you might ask yourself:
if you want power when it’s not windy or sunny, well… isn’t that what batteries are for?
Again, the short answer is yes, but the more storage you have, the more expensive it gets. The chart below from that post shows the cost of generating capacity for generating capacity from solar and storage. It starts low with solar, because there is usually a set number of hours of sunlight per day, but from the 8 hour mark, you need to rely on more and more stored energy to provide a steady amount of power, and while the cost of batteries is falling, it is still comparatively high. The different colour charts who how the cost is expected to change over the next decade.
What is this hydrogen being used for?
In the context of a fossil-free internet, I can see why the founder of Rivan the startup might have created the company to produce methane from fossil free means. There’s a massive industry that relies on methane gas for heat in industrial processes, and as a feedstock for making chemicals:
At Rivan Industries, we produce millions of litres of extremely cheap green H2 that we convert almost immediately to a larger, more utilitarian molecule in CH4 that is distributed in the grid to decarbonise heavy industries. We wrote about the last 12 months of R&D here on our quest to produce the cheapest H2 and capture the cheapest CO2 on the planet, and are hiring amazing engineers to make it happen.
Source: The quest for £1/kg H2 – Rivan Industries by Rivan
How does this fit in into research on fossil free power for datacentres?
In the world of datacentres, there’s a depressing trend towards relying on gas engines and gas turbines to provide power for them then there is not sufficient capacity from the grid.
So on the face of it, technological solutions from groups like Rivan might represent a path away from fossil fuels for facilities that are relying on fossil methane to supplement power from the grid.
Technically, if methane isn’t coming from fossil sources, and the carbon is coming from captured carbon in the atmosphere or biomass, then you’re not actually putting new fossil carbon in the atmosphere – you’re just cycling the same carbon that’s already around us.
There are still a few problems here though, that I don’t see a solution to.
The Fugitive Emissions problem
One of the problems though is that methane, even when from non-fossil sources like this is. still a massive driver of global warming because it’s such a powerful greenhouse gas.
It’s tens of times worse in terms of global warming potential than carbon dioxide, and you only need a small amount of unburned methane to leak from gas turbines and gas engines during operation to be worse than coal, climate-wise. This post here from the Gas Outlook is a good example of why this matters.
Looking upstream, because methane is a gas, it can escape from the transmission infrastructure like pipes, and so on – at this point, we only need a little bit of leakage to be back in “worse than coal” territory again.
This is one of the reasons that even carbon captured methane is not accepted in the climate group’s own technical guidance on 24 /7 carbon free energy.
The Local Harms problem
There’s also a non climate issue here – burning methane for power generation still harms local communities, because it can make the surrounding air quality much worse, and bring all kinds of respiratory harms on people living nearby.
One of the most vivid examples of this right now is likely the X.ai Memphis datacentre, and this story in Politico of community backlash is instructive.
What else is out there?
There’s a good post from Michael Liebreich about how the cost of decarbonising power changes the closer you get to complete decarbonisation of the grid, and one of the charts I’ve added below gives a good idea of the problem we face once we are making good progress in decarbonisation.
Decarbonising power is comparatively cheap to begin with, but the cost of different forms of generation gets more expensive quickly as you go above 90%, because you need to account for the variability of renewables.
Right now, this is seen one of the reasons people use to justify investments in nuclear – i.e. if it’s going to be that expensive towards the end, why not just have one form of power that gets us all the way there, even if it’s more expensive than wind and solar?
Well, one of the reasons is because you’re only needing to power that last 10%, you don’t need as much of the expensive stuff, as long as you’re prepared to have power from multiple sources.
This is one of the reasons almost all transition modelling shows wind and solar doing most of the work to decarbonise the power sector, waaay ahead of other forms of generation.
This also leaves the door open for somewhat more niche options, that would not be feasible if you had to use them for all 100% of the power demand.
I’ve seen hydrogen proposed here a few times, and even written a few posts too, but one recent paper from Tom Brown and friends on the Minimal Methanol Economy has recently caught my eye too.
It proposes using non-fossil methanol to help deliver these last few percent, whilst relying on more conventional renewables and electrification to decarbonise everything else – as opposed to seeing hydrogen like some direct replacement for all fossil fuel usage we see now. Even methanol is expensive, it has a few advantages over hydrogen and methane.
Unlike methane or hydrogen, fugitive emissions are not such a problem, because it’s easier to move around and store, and it’s a liquid at room temperature. It’s also cleaner burning than other fuels like diesel. That said, I’m less clear on the air quality story if used in some equivalent to an onsite engine or turbine – that’s something I’m reading up on to understand better.
That’s another post though, once I know more.
But for now, I hope this provides some extra context to discussions around alternative fuels for datacentres than the default we’re seeing of methane, or in some cases just running diesel for months on end.
Leave a Reply