BBC News: Molten metal batteries for the grid

[Mike O'Dell]

sigh

there area various liquid metal batteries already developed,
the most "famous" being NaS - Sodium-Sulphur - yes, both
liquid at about 600C. Tokyo Electric Power paid NKE several
billion dollars to develop it. it does have the problem
of having a *lot* of molten sodium which gets pretty nasty
if it gets loose. 


the BBC article describes one that's a bit less NIMBY.

The important point, however, is that just developing new
electrochemistry is NOT that difficult - it's been done
many times and there are others out there to be found.

The problem is moving anything like that from the lab
into something that's deployable at a scale which is
useful for grid-scale storage. as for the cost of the
cell materials, what is known as "balance of plant"
typically costs more than the storage cells so the
all-in system cost is the real issue. needing to 
operate continuously at 400C makes that rather harder
than a system which operates at room temperature
like, eg, a zinc-air cell with an aquesous electrolyte.

having evaluated a number of such inventions for the purpose
of investing in their development, it's *entirely* about
scalability. One of our West Coast guys has a PhD in
Materials from UCSB and worked for one of their
Nobel Laureates; he then went to Raychem and ended up
directing the development of the Polyfuse, so our view
of "scaling" is based on his experience which he saw
repeated multiple times at Raychem.

Getting from a single benchtop cell to
multiple cells for characterization is 5x the work.
to get from there to "batteries" for characterization
as a unit is 5x. from there to accelerated life testing
of large systems which can be used to build systems
of useful capacity is another 5x. etc, etc, etc.
All the comments about "balance of plant" above
apply when the systems start getting larger.

Very few technologies manage to hop off the bench
then out the door of the lab and into Manufacturing
and Product Engineering. Even fewer make that gauntlet.

There *are* grid-scale storage technologies in the
pipeline. Some of them may be successful at some scale.
But "grid-scale" means that it can *deliver* to its
output terminals line-synced power to the tune of
1-2 Megawatts for circa 8-4 HOURS. More importantly, 
they need to *charge* that fast, too! An inverter-charger
that can do that is non-trivial engineering all by itself,
especially if it can't use gold buss bars.

The cost bogey for grid-scale storage is roughly the same
as the LCE ("levelized cost of energy") of natgas-fueled
gas turbines. This was not easy in the first place and
has recently gotten harder since the price of natgas has
gone down so dramatically. At this point, cleaning up most
coal plants is rather more expensive than simply building
a few rows of gas turbines next to it. they are cleaner
than the coal plant post-retrofit and they are easy to
own and operate. they dispatch in a couple of minutes
max instead of half an hour at best (assuming the boilers
are kept pretty warm). Natgas-fueled Solid Oxide fuel cells
are competitors as well if you can used the heat output
of the cells - the total conversion effiency can approach
80% in that case. That's why they are increasingly being
considered for high-rel power even at the higher cost
because they can subsume massive UPSes in some cases.

There is indeed a bright potential future for grid-scale
storage if it can get to the capex and opex targets that
make the unit economics work.

whether this particular cell technology figures in that
is another question entirely.

     -mo