Iron-air batteries, thermal storage, deep geothermal, modular nuclear
My current thinking on the power/energy future.
While wind and solar adoption will increase over the next 10-20 years, their future role is overplayed because the downside of their variability (i.e the cost of storage to allow for power on demand) has not fully be internalised by markets (i.e. the intermittency is being subsidised) or the general public.
The value/potential of firm (non-variable) sources of power (e.g. deep geothermal, nuclear) is under-appreciated, but will become more important over time (e.g. after 10-20 years from now). I see these as under-appreciated for the same reason intermittent renewables are over-appreciated.
Since the downsides of renewables are not yet internalised, the importance of long term energy storage is somewhat under-appreciated. Implicit in wind/solar being overplayed is an under-appreciation for the challenge of achieving cheap long-term energy storage.
I don’t justify these claims here, but you can gain some insight through this piece. Rather, I focus on some emerging(ish) long-term storage and firm power source technologies. The list is very incomplete. It’s just based on some recent reading I’ve been doing.
Every power generation technology requires convenient energy storage, it’s just that some have it built-in
The gas for natural gas power plants is conveniently stored in tankers. The nuclear fuel for nuclear plants requires sophistication to store, but the storage is energy dense and cost effective. For wind and solar we have to come up with things like batteries to provide convenient storage needed to have power available on demand.
In moving from fossil fuels to renewables (or indeed to nuclear/geothermal), we’re moving from a situation where we take storage for granted to a situation where we cannot. We have power on demand. We’re moving to power on supply.
Ok, enough high level stuff.
Back of the Envelope Energy Storage Cost Target
To be cost effective for long-term energy storage, the storage needs to be really cheap. To ground things, let’s consider how much a battery storage technology needs to cost to hit the following assumptions:
Add no more than $0.1/kWh to the levelised cost of electricity
Battery life of 10 years or 10,000 cycles, whichever is reached first
Assume the battery is 100% efficient for charging and discharging
One way to phrase the above is: “Let’s say we had renewable power available at $0.05/kWh in levelised cost of energy, what would energy storage technology have to cost in capital if it were to add no more than $0.10/kWh to the overall levelised cost of the power.” Note that, in scenarios like this, the cost of storage – not the cost of renewables – dominates the cost of power delivered.
Here’s the graph (spreadsheet here):

What the graph says is:
If you want to store power for up to one day, you need your battery storage to cost less than a few hundred dollars per kWh. This is why lithium ion batteries, often costing roughly $200-400/kWh, are commonly used for short duration storage.
As you come up to storing power for a month, your form of storage needs to drop below $10/kWh.
For storage of many months – approaching a year – capital costs need to move towards a dollar per kWh of energy stored.
All of the above assumes perfectly efficient batteries, so the target costs for storage to be viable (assuming a $0.1 LCOE target for storage) are lower.
As a side note, large scale Tesla Megapack batteries retail at $400+ per kWh of storage. If you see a project using Megapacks for any more than daily storage, you know they are adding a lot more than $0.10 to their levelised cost of power. (Granted, that’s at list price, although I wouldn’t expect all that much of a negotiated reduction.)
Iron-air Batteries for Long Term Energy Storage
Form Energy claims to achieve (or achieve in the future) less than $10/kWh capital costs for energy storage. This looks like a reasonable target for medium term storage based on the graph above.
The idea is to oxidize iron using a voltage, and then reverse the reaction to deliver power when needed. Iron and air (for oxidation) are cheap.
Very back of the envelope, if iron-air batteries have a theoretical energy density of around 1 kWh/kg, and the cost of iron is $200 per ton, that’s about $0.2 per kWh of storage. Obviously there will be more costs involved in building balance of the system and the actual energy density may only be a fraction of theoretical. Still, it seems plausible the technology can come in under the $10/kWh claimed.
I don’t know the charge/discharge efficiency of these batteries, or their actual energy density in practise. I don’t believe a large scale (say 10+ MWh) project has been deployed, although it seems over $300M in funding has been raised. Still, all considered, the technical direction seems worth driving to a conclusion of viability or not.
Thermal Batteries for Energy Storage
Not just the thumbnail, but the general feel of this video makes you feel you’re watching off-road dirt-bike racing!
The idea here is to heat up a massive bunker of sand and use that as a form of energy storage. Critically:
Heat might be generated using off-peak renewable electricity to resistively heat the sand.
Sand is cheap. Plus, unlike water, sand can be heated to temperatures much higher than 100 Celsius (maybe up to 400 – 600 C) at atmospheric pressure without boiling. This allows for more energy storage (without creating steam or requiring pressure).
The idea is (primarily) not to convert the heat back to electricity, but rather to use the heat for district heating systems.
This is exactly the kind of system my Dad would thermodynamically hate (because of the energetic (well, exergetic) degradation involved in converting electricity to heat, and then the use of high temperature heat for low temperature uses, e.g. district heating). However, and maybe because this video was recommended to me by his brother 😂, he may appreciate the practicality of the solution.
Here’s the rough envelope: A ton of sand is maybe $20 and holds about 730 Joules per kg per degree C. Heating a ton of sand by 400 deg C would allow for storage of 400 * 730 * 1000 = 292 MJ of energy stored (thermally), which is about 81 kWh.
Piping and bunkering and infrastructure costs would add significantly, but that’s about $0.25/kWh of energy storage as a baseline cost. So, it looks plausible that the $1/kWh target for 1 year of energy storage in the graph above might be within reach.
For places that have or are planning district heating, and have excess power nearby, this seems like a decent idea.
Very Deep Geothermal Power
Geothermal is where you put some pipes underground to capture heat. Unfortunately, the heat is relatively low temperature unless you go very deep.
Advanced geothermal, or what I call “very deep geothermal”, is where you drill down below rock, to a point where you can’t use normal drills, and have to use high power millimeter waves instead. The benefit is temperatures high enough to generate high pressure steam and run conventional turbines for power generation. Another benefit is heat available on demand rather than intermittent.
There’s an interesting point in common between natural gas, nuclear and very deep geothermal power plants: they all use steam turbines to generate power. The technologies are heat sources to which a steam turbine is tacked on. If we could generate very cheap heat from nuclear or geothermal, the price of power would asymptote towards the amortised and operating costs of steam turbines. Interestingly, steam turbine manufacturers like GE are very incentivised to see technologies like deep geothermal and nuclear gain traction as it allows for continuing/growing business as fossil fuels are phased out.
If you think about getting to a world where we have $0.03/kWh electricity available on-demand (which is what I would like to see), the options I see are:
a) Directly making on-demand power cheaply with nuclear or deep geothermal or something else.
b) Making renewable power really cheaply and making storage really cheap. For example, you could generate renewable power at $0.01/kWh (for variable power) and seek storage for $0.02/kWh on a levelised cost basis, which, for monthlong storage would come out to needing a storage technology (with perfect charge/discharge efficiency) hitting less than about $2.5/kWh in capital costs.
My guess for the next ten years is that we see more of option b) approach (although not hitting a $0.03 all-in price target), while on a 20+ year horizon we find that option a) is the one that achieves $0.03/kWh power all-in.
Modular Nuclear Power Plants
This startup aims to take nuclear waste and uses it to make heat available either for industrial or domestic heating OR for electricity generation. The approach is “heat as a service”.
The design is modular and the units are small (40 foot container sized, it seems).
Let’s contrast this approach with standard nuclear fission at large scale:
Source of fuel – as with many energy technologies (fossil fuel, batteries included), there are geopolitical risks with fuel/materials sourcing. At least here, there should be ample fuel from existing nuclear plants.
Nuclear waste handling – I’d like to better understand the waste management specifics. Apparently the isotope half life is on the order of 100 years, rather than multiples more, so this seems an advantage.
Permitting – adoption would require a country with a somewhat dynamic regulator willing to consider new approaches. At a high level, installing/permitting something small seems easier than something big. The usual questions apply around what happens if there is a process failure or sabotage/attack on the nuclear plant.
Bitcoin Mining with Wind/Solar Power
I’m throwing this in here because a) I think it’s a bad idea and b) some crypto-heads suggest it as a good idea and/or justification for why Bitcoin is good.
It’s a bad idea because mining computers undergo thermal cycling when operated intermittently. Plus, the payback period on the mining computers is really bad when you can only run them a fraction of the time (which would be the case if you’re using an intermittent source of power). I wrote a detailled piece on this on my crypto blog here:
Alright folks, that’s it for now, let me know of any other technologies I should be reading about OR mistakes in the above.