The
recent return in fashion of the idea of the "Hydrogen Based Economy" is
a classic illustration of what I call the "Seneca Effect." Expressed in
words, it says "Increases are of sluggish growth, but the way to ruin is rapid."
Applied to the hydrogen economy or to hydrogen cars, it says that our
attempt to keep the current standards and paradigms while just switching one fuel to another leads to sluggish growth and, eventually, to ruin.
I
can’t help but take a serious poke at the hydrogen fuel cell electric
vehicle, which some seem to be absolutely in love with. And what’s not
to love? You burn hydrogen, an absolutely clean fuel, to produce water
vapour and nothing else (no NOx, no particulates…). You do it in a
silent, compact machine which isn’t a heat engine and hence isn’t
subject to Carnot’s punishing efficiency limitations. Refuelling is a
snap- a little more complex than dumping gasoline into a tank, but it’s
still very quick. And if you’re an oil company, if demand for gasoline
and diesel starts to dry up, you’ve just found your next fuel to
sell! Salvation is at hand! Or maybe the ambitious homeowner is thinking
they can make it from water using the solar panels on their roof,
thereby solving both their transport fuel bill AND the storage problem
associated with solar generation, all with one magical technology. It
just sounds fantastic!
Regrettably, this guy is hiding in the details, and he’s not all that disguised to anyone who cares to look:
(and apparently he even has his own Twitter account, @devil which is where I got his picture!)
My last article dealt with the energy efficiency cycle chains for ICE and battery electric vehicles:
https://www.linkedin.com/pulse/energy-cycle-efficiency-vehicles-does-ev-really-win-paul-martin/
We’ll
refer to the results of that article when making the comparisons for
the fuel cell electric vehicle (FCEV), and we’ll use similar assumptions
and data sources, so reviewing the previous papers might make sense
before reading this one.
Full disclosure here: I am mentioned on a
couple Texaco patents which were picked up by Chevron when they
acquired Texaco, related to making hydrogen from natural gas to feed
proton exchange membrane (PEM) fuelcells. Hydrogen is an old friend from
all the way back to my university days, and every second project or so
during my decades at Zeton has involved either hydrogen or syngas. Let
me make this as clear as hydrogen itself: hydrogen is a wonderful idea- in theory. The
big problem with hydrogen is, well…the hydrogen molecule. And there’s
no fixing that, irrespective of how clever you are or what you invent.
Let’s
take the efficiency chain of a hydrogen fuelcell electric vehicle
(FCEV) apart, piece by piece, just like we did with ICE and battery EV
vehicles, and using the same assumptions.
Hydrogen Production
Hydrogen production is itself roughly 70% efficient-
regrettably, that’s at best. A recent conversation I had with
Hydrogenics, a major producer of both alkaline and PEM electrolyzers,
puts the efficiency of their cheaper alkaline units at around 60%, and
the efficiency of the PEM units at around 70%. From what I understand,
that’s scraping pretty near the ~ 77% ultimate efficiency for
electrolysis, in terms of LHV of H2 product out per unit electrical
energy input. As I mentioned, that in and of itself is a loss- it’s
acknowledging that you’ve put energy into the production of hydrogen
that you will not get back because you’re not recovering the heat of
condensation of the product water.
Note that most electrolysis
vendors state their efficiencies in HHV terms, i.e. including the heat
of condensation of the product water. On that basis. 70% LHV efficiency
(the figure I'm using) is about 83% HHV efficiency. That's achievable
in a PEM electrolyzer running at a low current density- you could likely
produce significantly more hydrogen from that same unit by simply
turning up the current and losing more energy due to reduced efficiency.
The
trouble with electrolysis is that some of the energy obviously goes
into making oxygen (though in Gibbs energy terms, you don't get any
credit for that). That might be valuable itself and hence worth a credit
if you’re doing it in large enough systems to make oxygen clean-up and
compression for sale a reasonable thing to do, or if you’re using the H2
not as fuel but as a process feed and that process also needs
oxygen. Regrettably, a vehicle refuelling station isn’t going to get any
benefit from the product oxygen- it’s going to vent it.
So, let’s
take 70% (of LHV) for the conversion of electricity, presumably
renewable electricity, to energy in the form of the LHV of hydrogen. To
be fair, we’ll have to throw in a 6% penalty for grid losses on the way
from the power plant to the electrolyzer.
The figure of 70% of
product LHV per unit feed LHV happens to match pretty closely to the
best estimate for the best available technology for hydrogen production
from natural gas, the large centralized steam methane reformer (SMR). An
SMR takes advantage of huge scale to provide benefits from heat
integration and thermal energy recovery, including burning the waste
gases produced when purifying hydrogen to the extreme levels required
for the long-term survival and efficiency of our “engine”, the PEM
fuelcell. These devices are really sensitive to carbon monoxide (CO),
which poisons the precious metal catalyst. Regrettably you get some CO
any time you reform a hydrocarbon fuel to make hydrogen. Even worse, the
catalyst itself can also make CO from CO2, so your hydrogen feed has to
be purified to remove both CO and CO2 to ppm levels. In fact, even
inerts like argon or nitrogen in the feed hydrogen have negative effects
on the efficiency of the PEM fuelcell by requiring more anode tail gas
venting, so in fact you need very pure
hydrogen to feed a fuelcell- something you’ll rapidly find out if you
get the purity spec requirements for a PEM fuelcell from someone like
Ballard, Plug Power etc.
Regrettably, the efficiency of SMR drops
like a rock the smaller you make the unit. Heat losses become more
profound, which matter a lot in a high temperature process like SMR. And
as the scale drops, so does the opportunity for beneficial heat
integration etc. Everything just gets worse at smaller scale for this
process, as anyone rapidly discovers when they try to design such a unit
for a small application like, say, a chemical pilot plant, or a vehicle
refuelling station…
Distribution from a natural gas well through a gas plant and pipeline to the SMR, and then distribution
of hydrogen from a centralized large SMR to fuelling stations, is
likely going to cost us a great deal more than 6% of the energy in the
product hydrogen, but let’s be generous and keep that loss at
6% total just so we have less math to do (spoiler alert- it won’t matter
in the end!). So whether we start with electricity or methane, we’re
down to 0.7*0.94 or about 66% of the feed energy by the time we’ve made
hydrogen- at best, without much room for improvement because we’re up
against the thermodynamic limits already.
Note also that it is
possible to achieve very high apparent efficiencies (even, somewhat
strangely, higher than 100%) if you electrolyze steam at high
temperature rather than starting with water (for instance by running a
solid oxide high temperature fuelcell in reverse). However, those
efficiencies are artifacts of the calculation- the energy used to
evaporate water to make steam and then superheat steam to high
temperature are not included in the calculation. Nobody uses steam
electrolysis to make hydrogen unless they have a use for either hot
hydrogen or hot oxygen or preferably both.
Hydrogen Storage
Now
we have to store the hydrogen, and the devil in that detail again
arises from the molecule itself. Though its energy density per unit mass
is quite impressive, hydrogen even as a cryogenic liquid (at 21 degrees
above absolute zero...) is only 75 kg/m3…so the only currently
practical means of storing hydrogen for small vehicle applications is as
a high pressure gas. Any means used to increase the storage density or
to reduce the storage pressure (things like metal hydrides, adsorbents,
organic hydrogen carriers etc.) either significantly increases the mass
of the tank, or increases the parasitic loss of hydrogen during storage,
or requires energy to recover the hydrogen, or a combination of those
things. So high pressure gaseous hydrogen it shall be, and I wouldn’t
count on some magical breakthrough to change that- we’ve had plenty of
time to consider the alternatives in the thirty years hydrogen has been a
serious contender as a vehicle fuel.
While much ado is made about
how dangerous hydrogen is, there will be no pictures of the Hindenburg
in my paper! In fact we’ve been handling hydrogen quite safely in
industrial settings for a long time- we know what it takes to keep it
safe. The wide flammability range is offset by its low density and high
diffusivity, making hydrogen explosions rather less likely in practice
than in the imagination of people doing HAZOP reviews. With proper
precautions during design and operation, high pressure hydrogen is quite
safe- in industrial settings that is! I don’t want my
neighbours to even think about making 6,000 or 9,000 psig hydrogen using
their home solar panels…that gives me nightmares on many levels.
The
trouble with high pressure hydrogen storage is that you have to
compress the gas from a modest ~300 psig exiting an SMR, or perhaps from
near atmospheric pressure exiting a PEM electrolyzer- a compression
ratio ranging from ~20:1 to over 400:1. That takes thermodynamic work,
which takes energy, typically electricity. And regrettably, the heat of
compression, although available, needs to be rejected at a rather low
temperature to protect the compressors’ components, and hence is rather
difficult to use in any meaningful way. Even worse is the fact that you
need a tank at, say, 6,000 psig pressure which can fall only to 5900
psig when filling the tank on the vehicle, so all the compression is
done at the highest compression ratio- and the tanks themselves at the
filling station need to be very large indeed.
When done on a massive scale with large compressor trains, high pressure hydrogen storage can be as good as 90% efficient
in terms of LHV of H2 stored per unit electrical energy used to run the
compressors, which is surprisingly good given all these
considerations. (Note that the polytropic efficiency of the compressors
themselves is a small fraction of that number- this is a very different
measure of efficiency). Regrettably though, when you reduce the size of
the compressors, the efficiency plummets. A single-vehicle multistage
diaphragm compressor may be as little as 50% efficient on that basis or
even less - this is something which, along with the unit capital
cost, gets much worse as the scale decreases. That’s a shame, because
distributing hydrogen over long distances is infeasible for exactly the
same reasons it’s hard to store- the properties of the molecule. All the
dreams about a “hydrogen economy” are predicated on small, distributed
hydrogen generation systems so the thing we need to move around from
place to place isn’t hydrogen, which leaves us in my view with only one
realistic option: electrolysis.
OK, so we’re at 70% (H2 production) x 94% (grid/distribution loss) x 90% (high pressure storage) = 59% from energy source to tank,
compared with 80% for gasoline. Clearly we’re not going to be feeding
that hydrogen into a lossy ICE as a replacement for gasoline, especially
if the source of the H2 is fossil- we'd be far better off feeding the
ICE directly with whatever fossil we started with. And if we care about
GHG emissions, we certainly can’t make that H2 from fossil sources- we’d
be better off with the Prius for sure. Electrolysis from renewable
electricity is our only hope.
The Proton Exchange Membrane (PEM) Fuelcell
Sadly, we’re not done losing energy yet- next is the loss in the PEM fuelcell. Despite the fact that it is not a heat engine, it still has its own limiting thermodynamics. PEM
fuelcells are achieving efficiencies of about 50-60%, and that is not
far off the ultimate thermodynamic limit of about 83% for an ideal fuelcell.
https://www.princeton.edu/~humcomp/sophlab/ther_58.htm
So
let’s be generous and take 60% as the fuelcell efficiency- that will
get us from the well or power plant all the way to the output of the
fuelcell.
The FCEV From Energy Source to Wheels
Now we have the electric drivetrain (inverter and motor) and its 90% efficiency- so “well to wheels”, or “power plant to wheels”, we’re now at 94%x70%x90%x60%x90% = 32%. I’ll
remind you that on a well to wheels basis, the Prius achieved about 30%
on gasoline- so we’re doing better than the Prius, and with no tailpipe
emissions! And rapid refuelling. Hurray! Right? Right?...

I remind you that my home-made electric vehicle, the E-Fire, on the same basis, was achieving 76.5%...and
it had no tailpipe emissions either. And despite having a very small
pack by OEM EV standards- only 18.5 kWh- it has an adequate range for my
commute. We’re just crossing 12,000 nearly fossil-free miles driven so
far, and I’ve never waited around for it to charge- I just plug it in
once at night, and once in the morning at work. It doesn’t replace
everything a gasoline car can do, doesn’t try to, and doesn’t HAVE to in
order to serve a very valuable purpose- getting me to work and back
with acceleration that makes my neck sore.
The FCEV Loses on Cycle Efficiency- But We're Not Done Losing Yet!
So
when we talk about the FCEV, if we’re honest, we’re talking about a
technology which has a best case energy source to wheels efficiency of
roughly 2.4x inferior to that of an existing alternative technology (the battery EV). What
do we get in return for that huge efficiency hit? Faster refuelling,
and possibly modestly greater range between refuellings- that’s it.
Seems like too much of a price to pay? Wait- we’re not done yet! We haven’t even started talking about cost…
Hydrogen is a very expensive fuel, regardless how you slice it.
The
2.4x inferior efficiency- again at best- means we’ll have to build out
at least 2.4x as much renewable infrastructure as if we used it to
recharge EVs. That alone should give hydrogen promoters significant food
for thought.
Then there’s the hydrogen distribution
infrastructure. You’re not going to be refuelling at home, folks, unless
the local fire marshal falls asleep at the wheel of his diesel fire
truck. So that means businesses are going to have to build out that
infrastructure, and they’re going to want a return on that
investment. They’re not going to MAKE that investment, because they know
that a return is impossible.
And as if that weren’t enough, now
let’s talk about what else is in a FCEV. There’s of course a hydrogen
storage tank and a PEM fuelcell. Oh, and every other part of the battery EV, including the battery! The
battery will be smaller- closer to that used in a Prius than to that
used in a BEV, but it’s still needed to capture regenerative braking
energy, to manage the power demand on the fuelcell stack to keep its
costs down, and to manage the start-up and shutdown process of the
fuelcell system. So the FCEV will be a hybrid.
Furthermore, we’ve
had a long time to drive down the costs of the PEM fuelcell, and the
costs are still very high. Though that would certainly drop further as a
result of the “learning curve” with mass adoption and mass production,
just as it continues to do for Li-ion batteries, there’s a nagging
limiter to dropping that price too far: platinum group metals (PGMs)
such as platinum and palladium that are used in the fuelcell as
catalysts. Reduce the PGM content and the fuelcell becomes even more
susceptible to hydrogen impurities, and the efficiency drops too I
suspect. Replace the PGMs with cheaper metals like nickel and most of
the benefits of the PGM fuelcell go away- you’ll need to operate it at
higher temperature etc.
(picture of the Toyota Mirai FCEV, courtesy of www.automobilemag.com)
Does this mean that hydrogen is dead for personal transport applications? In a word, and in my opinion, YES.
Elon Musk and I agree violently on that topic. Erm, I'd qualify that by
saying except in a world where electricity somehow costs nothing, or
even worse, its price goes negative, because renewable generation
basically becomes so cheap that it costs basically no capital to
install. But I’m betting that a) that will never happen, and b) even if
we were to come close to that puzzling economic outcome, the capital
cost and other practical problems with the electrolyzers, compressor
trains, storage tanks and fuelcells would kill the idea deader than a
doornail anyway.
A comparison of two real vehicles you can buy (in
California at least) makes it clear that my estimates are optimistic in
favour of hydrogen. For cars of similar features and the same EPA
range, the hydrogen car uses 3.2x as much energy (i.e. much more than
the 2.4x minimum I've calculated above) and costs 5.4x as much per mile
driven:
https://www.linkedin.com/pulse/mirai-fcev-vs-model-3-bev-paul-martin/
Of
course BOTH can improve from where they are now- but the calculations
in my paper above set the limits. You can't overcome thermodynamics
with invention or wishful thinking.
Does this mean there is no use
for PEM fuelcells? Absolutely not! There ARE established markets where
PEM fuelcells make good sense- but they’re all applications where energy
efficiency matters a lot less than something else- rapid refuelling as
an example. Hence, Plug Power is finding a niche market powering
warehouse forklift trucks, particularly in refrigerated warehouses.
(FC forklifts, from www.plugpower.com)
The
same goes for the so-called “power to gas” or P2G schemes that some are
developing. These are an entirely different model: they use “excess”
renewable electricity to make hydrogen which is then dumped at low
pressure into the natural gas grid where it is ultimately used to make
heat- often in devices which actually finally recover the heat of
condensation of the water of combustion. As a means of storing
electricity, P2G schemes are so ridiculously inefficient that they’re
not even worth talking about, but they’re also very low in capital
investment AND they reduce GHG emissions when the H2 displaces
methane. That’s not all bad.
Other Transport Uses for Hydrogen
Batteries
are either marginally feasible or infeasible for some forms of
transport right now: aircraft, long-distance transport, trains and
ships. The real question for these applications is simply this: how much
do we care about toxic tailpipe pollution? If we care about this most
of all, then hydrogen is the only game in town. But if our primary focus
is fossil GHG emissions, we have the option of using biofuels for these
applications as an alternative to hydrogen. For aircraft, biofuels are
likely the only practical solution until something which is as much
better than Li ion than Li ion was better than lead-acid batteries is
invented- perhaps a rechargeable metal-air battery. And while the total
replacement of diesel and gasoline with biofuels is infeasible even if
economics are thrown aside completely (see www.withouthotair.com for the
figures on that), if we were to do 90% of the car miles with
electricity, we should have enough biofuels production potential to
handle the remaining 10% PLUS all these other applications for which
batteries are currently infeasible. Toxic tailpipe emissions matter a
lot less when they're emitted between cities.
The option of using
hydrogen or electrochemical means to reduce CO2 to produce liquid
hydrocarbons is, obviously, significantly less efficient than hydrogen.
The same with ammonia, which is trotted out as a way to overcome some of
hydrogen's deficiencies. Ammonia is a toxic gas- and making it is again
less efficient than making hydrogen. The thought of using ammonia as a
vehicle fuel absolutely terrifies me, given the scale of ammonia-related
deaths from its existing uses as a refrigerant and agricultural
chemical.
The Real Future of Renewable Hydrogen
Right now,
over 96% of the hydrogen produced in the world is produced from fossil
fuels either deliberately (coal gasification or natural gas in SMR or
ATR units), or as a byproduct of petroleum manufacture. We're going to
need to become very, very good at making hydrogen from renewable
resources like PV solar and wind electricity, not to waste it as an
inefficient vehicle fuel, but to use it to make things like ammonia and
urea which are used in fertilizers. We'll need to replace this enormous
fossil hydrogen generation infrastructure so that we can get the fossil
monkey off our backs without starving.
More on the future of renewable hydrogen from renewable electricity: is it the future? Or a greenwash?
https://www.linkedin.com/pulse/hydrogen-from-renewable-electricity-our-future-paul-martin/
My next article will address the bogeyman of embodied energy, particularly in batteries.
Disclaimer: everything
in this series of articles is my own opinion, for which I try to state
sources and references whenever possible. It’s highly likely that
something I’ve said is just plain wrong- for which I apologize in
advance. Whenever you can show that I’m wrong with a good reference,
I’ll go back and edit the text with the correct information. My
employer, Zeton Inc., is in an entirely different business, and doesn’t
endorse or even have an opinion on these issues. We design and build
pilot plants- that’s it- and we love doing it!