=energy =global warming =technology
Here are most of the cheapest
grid energy storage systems, with costs per kWh output assuming input energy
costs $0.03/kWh, and approximate efficiencies. Note that these are wholesale
electricity prices; consumer prices are typically $0.06/kWh higher, with a
larger surcharge in some relatively corrupt locales such as California.
These estimates include the cost and inefficiency of electricity conversion,
and assume 8-hour constant discharge once per day.
If the goal is mitigating global warming, approaches can be compared
using
CO2 avoidance cost. Replacing natural gas, $100/ton CO2 is ~$0.03/kWh. In general, anything
above $100/ton isn't worth considering; reasonable approaches are typically
~$65/ton CO2. At $65, complete mitagation of US CO2 emissions would be about
10% of federal tax revenue, which is probably more than people are willing
to spend. As for China, it's willing to spend ~$10/ton.
As such,
energy output costs should be <= $0.09/kWh for a method to be worth
pursuing, and should be <= $0.07/kWh for a method to be a competitive
large-scale CO2 mitigation approach. Anything >$0.09/kWh is only relevant
for a small amount of peaking power (if it's cheap per power output) and
backup systems for buildings (if it works on that scale and isn't
location-dependent) - assuming that cheap natural gas is available.
What if natural gas isn't available? Coal emits ~3x the CO2 per kWh of
combined cycle natural gas generation, so we can increase the CO2 mitigation
value by $0.04/kWh. If the only alternative to solar + wind + storage is
nuclear power, relative costs are unclear: LCOE (levelized cost of energy)
for new nuclear plants has a very wide range of estimates, from $0.03 to
$0.20/kWh.
One of the more-accurate
evaluations of grid energy storage is a government report series funded by
the US Department of Energy, hereafter "DOE Reports".
Here's the 2020 version, and
here's the 2019 version. Some of the below estimates are extrapolated
from the DOE Reports.
--:-:-:-:-----------------------------------:-:-:-:--
pumped hydro: $0.09 / 80%
This is the main form of grid energy storage today, but the locations are
limited to where 2 reservoirs can be made with a substantial height
difference. Most good locations are already in use.
In theory, water
tanks could be built on mountains, which would make more pumped hydro
possible, but that would obviously increase costs significantly, and the
economics of pumped hydro are already questionable where cheap natural gas
is available.
diabatic compressed air energy storage (CAES):
$0.09 / 60%
In gas turbines, air is compressed,
heated by burning natural gas, and expanded. In diabatic CAES, the
compressed air is stored for a while first. The amount of electricity output
is about 40% more than just burning the natural gas would generate.
This is even cheaper per kWh stored than pumped hydro, but the efficiency is
lower.
The cheapest way to store the compressed air is in underground
caverns. There are many places where underground space has been made or can
be made by solution mining, so the availability of them isn't a major
problem, but they do leak a bit of air. Storing it in tanks is too
expensive. There was once a startup that got significant funding for CAES in
tanks on the basis that the raw materials needed for the tanks aren't too
expensive, but that was silly and that startup has failed.
adiabatic compressed air energy storage (CAES):
$0.15 / 60%
Compressing gas increases its
temperature. If that heat was stored, CAES could be done without burning
natural gas, but storing that heat would increase costs significantly. In
theory, the efficiency could be about 70%. In practice, every attempt at
doing that commercially has failed, and the cost estimates by proponents
have been inaccurate.
Of the companies trying to do this now,
Hydrostor seems to have the best approach.
electrically heated molten salt: $0.17 / 40%
It's possible to use electricity to heat molten salt, then use that heat
to run turbines. Malta Inc is a startup using this approach.
Malta
claimed to be able to reach 60% efficiency, which is a threshold
Breakthrough Energy Ventures set. That's the efficiency of combined cycle
gas turbines, which use far higher temperatures than are feasible to use
molten salts at, and they will not reach it.
Good efficiency requires
high temperatures, but molten salts are corrosive, which makes heat
exchangers for them expensive. Higher temperatures make this a bigger
problem.
The numbers I'm using here are with a slightly better
thermodynamic cycle design than Malta is using, and there's little room for
improvement.
gravitational energy storage: >$0.20 / 85%
There are some startups (such as Energy Vault) trying to store
electricity by
lifting solid blocks up and down with cranes. If grid energy storage
using concrete blocks lifted up 100m towers was done on a large enough scale
for 100% renewable energy for the world, the amount of concrete required
would be many times the world's annual concrete production. That's a
problem, but the bigger problem is that building towers for this is
constructing buildings, which is hundreds of times as expensive as the
concrete used for them. Estimating construction costs is hard, but I don't
expect this to be competitive with water tanks on mountains or LiFePO4.
hydrogen: ~$0.22 / 40%
This
is electrolysis of water to make hydrogen, storing compressed hydrogen in
underground caverns, and then using a gas turbine or fuel cell to generate
electricity from hydrogen.
The cost assumptions for electrolysis and
fuel cells in the DOE Reports are a bit too low. Economically, for large
applications, a combined cycle gas turbine is preferable to a hydrogen fuel
cell. A better methodology is to look at the cost of hydrogen from
electrolysis, adjust that for part-time electrolyzer usage, and consider the
efficiency and cost of combined-cycle gas turbines.
lithium-ion batteries: ~$0.26 / 85%
The battery cost estimates in the DOE Reports are too low, because
they're using BloombergNEF survey results, which are heavily weighted
towards subsidized Chinese batteries which are only for use in China.
The actual batteries are <50% of the initial purchase cost according to
the DOE Reports, which reduces
the impact of this inaccuracy, but the batteries need replacement more often
than other items.
The cycle life of LiFePO4 batteries is
pretty good, but the negative side is the same as in other lithium-ion
batteries, and the electrolyte slowly decomposes at the negative side,
forming a solid SEI layer that gradually increases in thickness. So, the
degradation over time is similar to that of other lithium-ion batteries.
Part of the reason why the cost estimate for this in the DOE Reports is
higher than some other sources is less-optimistic battery life estimates.
One proposed approach is using the batteries of electric cars to store
energy. One problem with this is that people don't want to charge their cars
according to schedules set by electric power availability. Proposals for
things like scheduling dryer usage have the same problem. Another problem is
that the current battery technologies with relatively high specific energy
are degraded by charging them and discharging them, and the cost of that
degradation is greater than the value of the electricity stored. If LiFePO4
batteries are used, then the cycle life is good enough to use them for grid
energy storage, but those are also much heavier for the same capacity, and
carrying more mass around in cars is bad.
sodium-sulfur batteries: ~$0.35 / 75%
These are high-temperature batteries with a ceramic (sodium
beta-aluminate) between liquid sodium and liquid sulfur. The main problem is
that making lots of thin special ceramic with no cracks is expensive;
solid-state lithion-ion batteries have the same problem, but even more so.
Sodium, sulfur, and beta-aluminate are probably all optimal, so the
prospects for better materials for these are poor, but theoretically the
manufacturing costs could be reduced significantly.
vanadium flow batteries: ~$0.37 / 60%
These used to be a viable backup power solution, because vanadium was
cheaper and lithium-ion batteries were more expensive. Regardless of current
vanadium prices, there's simply not enough vanadium to use these for grid
energy storage. Even if vanadium was cheap, the membranes are also too
expensive.
The membranes for these are already made on a large scale
for NaOH production, so simply scaling up production isn't enough.
Some other flow battery chemistries have been tried. EnerVault was a company
that tried to commercialize iron-chromium flow batteries and went bankrupt.
There's also been
a relatively large amount of university research on flow batteries,
mostly because there are many possible variations of membranes and organic
solutes, which makes it easy to find a novel thing to publish. Scientists
are incentivized to pretend their ideas are useful industrially when
publishing, so it's not easy for non-experts to judge designs based on the
scientific literature.
salinity gradient energy storage: ~$0.40 / 60%
Reverse osmosis uses pressure to remove salt from water. It's possible
to store energy that way. The equivalent of 10km of height is possible.
If you look at normal reverse osmosis, the efficiency of this looks bad.
However, with a closed system, you can choose salts that cross over
membranes much less than NaCl does. With an infinite amount of membrane, the
efficiency would then be similar to that of pumped hydro. Unfortunately, the
membrane isn't free.
In general, a bit less than half the cost of
reverse osmosis is electricity. If you only generate power from it 1/3 of
the time, then the equipment costs >3x the value of the electricity used,
and less electricity is output than is input. Reverse osmosis is typically
50% efficient; higher efficiency requires even more membrane.
A
closed-loop system does reduce the need for pre-filtering water, but still,
this doesn't seem economically viable due to the membrane costs. However, if
there was a breakthrough in reverse osmosis membrane design or
manufacturing, then this could theoretically be viable, but reverse osmosis
has already been researched a lot.
--:-:-:-:-----------------------------------:-:-:-:--
Looking at the above list, there are multiple conclusions you might come to:
A) Grid
energy storage is hopeless, so invest in natural gas production. This is the
petrochemical companies' conclusion.
B) Compressed air systems are the
closest to viability, so invest in those. This is the DOE's main conclusion.
C) Radically new technology is needed, so invest in something that hasn't
been tried before. This is the Silicon Valley investors' conclusion.
All of those conclusions are wrong. The correct conclusion is that somewhere in the pile of flow battery research, there are pieces that can be put together and extended into a viable system. Otoro Energy has part of a solution (although their ligand selection is still slightly suboptimal) and I've talked with them, but better membranes are also needed, and no startups are currently working on a good approach for that.