Some have advocated
storing energy (for utilities) as compressed air in large,
underground tanks or caverns and
have claimed efficiencies
above 75% are practical – close to what can rather easily
be achieved in pumped hydro storage. While 75% may be possible
with sufficient investment, there have been a few
demonstrations of CAES in the 50 to 300 MW range, and they have
probably seen efficiencies between 35% and 54%. The first CAES
facility was built in 1978 in Germany, and the second was built
in 1991 in McIntosh, AL.
This begs the question, “Why haven’t others been
built?” – as
everyone agrees energy storage would improve the growth of wind
and solar energy.
A major deficiency of all prior designs is the very limited range
in operating power flexibility during either charge or discharge
without seriously degrading efficiency. The power at which they
can operate efficiently is pretty much either off, or determined
by the cavern pressure at that point in time.
There have been rumors
circulating widely for more than five years that a company
in India, Tata, will soon
produce small cars powered
by compressed air, and several other companies have more
recently added to the hype. The uninformed green media
have distributed a lot of disinformation on the possible
performance and environmental impact of air cars. We’ll show below
that both of these applications for compressed-air energy storage are of
little value.
Utilities. The
Iowa Stored Energy Park, which has been under study since 2001,
has been projected to cost $220M. Reliable efficiency estimates
cannot be made from available information, but our estimates
are that it might achieve 35% efficiency – if it ever moves
from study to construction. Its peak turbine output power has
been claimed to be 268 MW, but
it is planned to be used in conjunction with a wind farm that
would have somewhere between 30 and 60 MW average output, so
the claimed
peak power number is perplexing. The air would be compressed
into a large, deep, aquifer. The amount of energy that might
be stored
is also apparently not being made public. A reasonable guess
might be 1000 MWhr. If so, that would make its cost $220/kWhrkWhr
if the total cost is simply ascribed to it energy storage capacity.
We realize the above estimate is two orders of magnitude higher
than some other
estimates that
have appeared in purportedly unbiased studies by big-name organizations.
(The only excuse for the studies from 6 years ago that came to
such unrealistic conclusions is that they were using optimistic
cost data from the 1980’s.) Our estimate
for cost at the 1000 MWhr level is
2 to 4 times what was estimated in a very recent paper (in JRSE)
for Advanced Adiabatic CAES (estimated to achieve 50-60% efficiency)
at the scale of 24 [1] GWhr. This difference is consistent with
would be expected for the typical 0.6-power scaling law.
The
small company Energy Storage and Power has shown a few details
on their website of their semi-adiabatic design,
which they claim would achieve 75% storage-cycle efficiency.
It appears they plan to use four oil storage tanks (somewhat
similar to what has been proposed in some CSP implementations)
to significantly
improve the efficiency over what can be achieved without the
oil
thermal storage loops. The oil thermal storage is essential
for reasonable efficiency. Compressing air from 1 bar to 90 bar,
assuming 93% polytropic efficiency with no intercool, would produce
pressurized
air at over 1050 K, which is far too hot for affordable storage.
Simple (dissipative) intercool would amount to throwing away
roughly 75% of the available energy in the compressed air, though
about a third of that can later be recovered from the atmosphere.
. (Not as impossible as it sounds. The compressed air is partially
expanded, and its temperature drops below atmosphere. It can
then be reheated to atmospheric temperature by heat transfer
from the atmosphere before final expansion.)Their plan to do
the compression in two stages with partial intercool
and
thermal storage is a reasonable approach.
However,
they only show half of the oil tanks their design requires
(one cannot mix
hot and
cold oil in a single tank
without
losing half the available energy), and there are major errors
elsewhere – even if one assumes
100% polytropic compression efficiencies (and 87% is more likely).
We suspect their estimate of about $750/kW for a relatively simple
and inefficient cycle when a nearly free cavern is available
is about right.
Achieving acceptable efficiency with simple adiabatic cycles requires extremely
large caverns that will tolerate high temperatures with large swings in pressure
and temperature. Very few caverns could tolerate the temperatures produced by
compression ratios above 11 without intercool (over 650 K for 85% efficiency,
which is above the critical temperature of water). The alternative is an advanced
adiabatic cycle (AA-CAES). It requires enormous reservoirs of a high-temperature
heat-storage oil, but it can achieve 52-64% efficiency.
The best published analysis of CAES is that by Packard et al [1]. Their theoretical
analysis of the simple adiabatic and isothermal cycles is sound, but their attempt
to analyze the AA-CAES cycle is of limited value. (Analytical approaches here
simply don’t work. Simulations are required.) It appears they assume ideal
isothermal expansion of the cavern pressure from 256 bar to 1 bar when estimating
required cavern volume. Such an approach would have completely unrealistic turbomachinery
requirements and completely unacceptable flexibility in power input or output
during the charge and discharge processes. They apparently would use 4 stages
(factor of 4 each). They estimated an oil cost of $80/kWhr for 55% efficiency.
We simulated an AA-CAES design that appears to permit 61% mean cycle efficiency
and an adequate operating power range, but it requires three large oil reservoirs,
three stages of re-heat near the beginning of the discharge cycle (using thermal
energy stored in oil), and high-performance turbomachinery. The cost of just
the oil for such a cycle would be $140/kWhr.
The AA design we analyzed required a cavern volume of 0.15 m3/kWhr with the pressure
ranging from 5 to 10 MPa and temperature swings of 310-370 K. The power-dependent
cost of the cycle was estimated to be $1200/kW when at the scale of 30 MW or
higher. Of this, about half is for the turbomachinery, about 25% for the heat
exchangers, and about 25% for the variable-speed motors, generators, and flexible
power conditioning.
The cost of boring large caverns in rock that is solid enough to withstand the
pressure cycles is about $2000/m3, which puts the cost of boring a cavern out
of solid rock at about $300/kWhr. Some large caverns with acceptable leak rates
are naturally available at very low cost, but not many. The cost of above-ground
steel tanks is somewhat higher. Solution-mined salt caverns may be a lower cost
option in some places, but they may also lead to increased equipment maintenance
costs. Storage in porous formations, like that being considered for the Iowa
Stored Energy Park, seems unlikely to permit acceptable efficiency. (We challenge
the design team that has been working on this project for the past 9 years to
provide credible evidence to the contrary.)
Some More Details
on an AA-CAES Design. Mass flow rates through
high-compression turbomachinery must increase at least as the
1.2 power of pressure ratio (while rotational rate increases)
to maintain high efficiency. This means that to handle a cavern
pressure ratio of a factor of 2 (e.g., 5 to 10 MPa) with good
efficiency, either switching of turbomachinery is required or
else the charge and discharge powers near the low end of the
pressure range must be only a third that near the upper end.
We opted to use three two-stage compressors (pressure ratio r=5
for each) in series (with recuperative intercool) at high pressure.
At low pressure, one section is bypassed, leaving two two-stage
compressors that wind up to higher speed for r=7 to 8 each. (An
alternative and possibly better approach would be a larger cavern
and a smaller pressure range.) The total output power is about
80 MW at 190 kg/s at low cavern pressures and about 45 MW at
100 kg/s during high pressure. (This is backwards from what might
have been expected, but it permits higher efficiency over a broader
range of conditions.) Mean compressor polytropic efficiencies
of 85% were assumed, and expander efficiencies were 88%.
The input air is assumed to be very dry and at 290 K. For the
low-pressure case, the output from the first compressor at ~590
K is cooled to 340 K by
counterflow exchange the storage oil before going to the next compressor. It
leaves there at 6 MPa, ~680 K, which means the high temperature heat storage
oil must be fairly expensive – about $7/kg. When using the compressed
air from the cavern, it is preheated by counterflow transfer from the oil (to
500-600 K, depending on the conditions) before partial expansion, during when
it cools to 300-360 K. It is then reheated by the oil before complete expansion
to atmosphere. At high cavern pressures, the process is similar except another
compressor (or expander) is needed, though all are operating at somewhat lower
speed, lower pressure ratios, and lower mass flows.
In both cases, the total heat transfer rate from the oil is nearly identical
to the electrical power output. (This was initially surprising.) A cheap oil
($2/kg) is satisfactory for more than a third of the total energy storage,
but an expensive oil is needed for one third of the total. A practical temperature
difference is assumed between the oil and the air (~50 K) to keep the cost
of the exchangers reasonable. The total oil required (11,000 tons) could be
cut in half by backing off an estimated 5% on efficiency goals.
It is certainly possible that a more efficient or less costly AA-CAES cycle
could be devised. A clue comes from the fact that heat rejected during intercool
was over 40% more than needed for heating in our case. Still, our simulations
are probably not too far from an optimum design – and our time is limited.
Operating and maintenance costs will likely be much higher than the $2/MWhr
typically seen with battery storage. Simple estimates suggest $5/MWhr of delivered
energy from rock caverns, but much higher for salt caverns. Water will normally
condense in the compressed air as it is cooled, and some of this would be carried
into the cavern, where it will accumulate. So the compressed air there would
be near 100% relative humidity most of the time. At the flow rates involved,
it may be difficult to insure that salty droplets are not transported from
the cavern into the exchangers.
Compressed-Air
Cars. Compressed-air-powered
cars make even less sense than compressed-air storage for utilities,
but that hasn’t
stopped the rumor mill.
The amount of energy that can be
stored in the compressed air in an Air Car is about 2% of what
the average car owner is accustomed to, and getting even that
much requires four huge, high-pressure tanks.
Compared to electric cars, air cars would have
huge disadvantages.
1. Air cars are the dirtiest possibility because
they are the least efficient. Fossil-source-to-wheels efficiency
is
7%, compared to
about 24% for electric cars and a little higher for advanced
hybrids.
2. Air cars would have no luggage room and terrible performance
because the volume and mass of the air storage tanks are
huge. An initial version of the air car apparently would
store 6.7 kWhr of energy (similar
to the energy in 0.8 quarts of gasoline) in four 25-gal
tanks.
3. Its range under normal urban driving might be 25 to 40
miles.
4. Recharging the tanks would cost about $12.
5. Air cars would be more expensive than electric cars which
have much better performance.
Below, we present some of the support for these conclusions.
The R&D that has gone into hydrogen-powered
vehicles over the past decade has probably exceeded $10B globally,
but they haven’t succeeded – partly
because the amount of energy that can be stored in a practical hydrogen
tank is only about 10% of the energy that is easily stored in the common
gasoline
or diesel tank. Energy storage in batteries is much less than in hydrogen
tanks, and energy storage in air tanks would even be much less than in
batteries.
It’s hard to get real
numbers on an Air Car – it’s
been mostly hype so far, largely coming from the French inventors
at MDI. Another source of disinformation on air cars is Zero
Pollution Motors, http://zeropollutionmotors.us/ .
We’ll return with
a few more comments on this near the end, after analyzing
some of the information on the mythical Tata Air Car.
Some
reports indicate the air storage volume in the mythical Tata Air
Car will be 90 m3 (at STP) at ~260 bar (26 MPa, or
3800 psi), ~350 liters compressed. The four compressed-air cylinders
(in these
small cars) are each larger than the gasoline tanks in large
SUVs.
(The total volume of air stored is 13 times that stored in one
of the 80-lb steel cylinders commonly seen in welding shops and
chemistry laboratories.) Clearly, it won’t be a small car,
and it won’t have any luggage room or leg room.
A four-stage
expansion engine might be used. In this case, the 26 MPa gas could
be expanded first to about 6.4 MPa against the
first (very small) piston (which would get very cold). The high
pressure gas is then reheated using atmospheric heat, and the process
of expansion and reheating is repeated two more times. The fourth
expansion is to atmospheric pressure; and rather than reheat this
cold exhaust air, it is directed into the vehicle for air conditioning.
Each expansion is at one-fourth the pressure of the previous but
at four times the volume, and thus each generates the same power.
If the expansions are adiabatic (no heat transfer) with 85% polytropic
efficiency, the expanded gas would exit at about 210 K. With 10%
additional energy losses
(leakage and friction), a flow-rate of 25 g/s would produce about 2 kW (~3
hp) per expansion, or about 8 kW (~12 hp) total. Of course, there will be some
heat transfer during the expansion, a point we return to later.
At 25 g/s, the first storage tank would be 75% depleted in 12 minutes. Total
output power is then down by 25% (the last three expansions are still delivering
full power), and it’s time to switch to the next tank if full power
is needed. (Using 4 separate cylinders makes it easier to have high power
capability
near the end of the total charge.)
By the time all four tanks are down to 25% of their full pressure,
the total amount of energy produced is about 6 MJ, or 80% of that
in one cup of gasoline.
One nice thing about high-pressure gas-energy storage is that
when the tank pressure is down to 25%, only 25% of the total available
energy in the tanks
has been used. (That is because the work done is the product of the pressure
and volume change. Expanding compressed gas from 64 bar to 1 bar gives
16 times the volume change in going from 260 bar to 64 bar.)
The
total energy produced by the four, huge, 3800 psi, air tanks
could be up to 24 MJ (~6.7 kWhr), or about the same as in 0.8
quarts of gasoline.
(One can calculate that the
theoretical limit for isothermal rather than adiabatic expansion
is 45 MJ, but
this requires a very
large and slow motor, which would have high leakage losses.
When including practical losses and averaging over the tank-discharge
cycle, the 24 MJ calculated by the nearly adiabatic method
is probably generous for the practical semi-isothermal case.
Piston
leakage
and other losses could easily be twice what was assumed in
this
calculation.)
Advocates claim the
energy storage is equal to 2 quarts of gasoline. Undoubtedly,
they can achieve higher
tank-to-wheels
efficiency
than a gasoline engine – perhaps by nearly a factor of
two. So comparing the effective energy storage in the air car
to 2 quarts
of gasoline is not much of an exaggeration.
How far a mid-sized
car will go on 2 quarts of gasoline depends mostly on how it
is driven and its load.
The Toyota Prius (hybrid)
will go 20 to 25 miles.
The Air Car is not a hybrid, so it has no breaking-energy recovery. Still,
perhaps 30-40 miles can be expected from a mid-sized Air Car under
average urban conditions, since it will be a little lighter than
the Prius.
The claim of a range of 125
miles may be possible – downhill, at 20 mph, with no stops, and a
strong tailwind.
The environmental and
cost claims by Air Car advocates are even more exaggerated.
Air compressors, even up
to 100 times the size
that any local station would
consider purchasing, are extremely inefficient. In principle, it may be
possible to compress air to 260 bar at about 75% efficiency,
but 12% to 30% is more
typical. The small (5.5 hp) compressors that individuals might purchase
for charging their tanks overnight at home will consume about
60 kWhr of energy
to fill the tanks with about 9 kWhr of (gross) energy. That’s about
15% efficiency. At $0.12/kWhr (expected US mean price in 2010), consumers
would be paying $7.20 to fill their tanks. That’s like paying $13/gal
to fill a Prius.
Filling compressed air tanks at the local station will cost even
more. A very large compressor getting 30% efficiency and capable
of servicing 100
cars per
day would require five to ten times the electrical power that most gas
stations have available. Hence, the more common power option would
be a large diesel
engine driving the compressor.
In either case, a reasonable
estimate is that upgrading the gas station to pressurize Air
Cars would cost over
$1.5M – about
four times what it would cost for the capability of refilling
a similar number of CNG vehicles.
Challenges include silencing the enormous compressors to a tolerable
level and drying, cleaning, and cooling the compressed air.
The biggest challenge
will be addressing the safety issues associated with tanks and lines
at 3800 psi. (Tire-pressure tanks are seldom above 75 psi.)
Fossil-source-to-wheels
efficiency in either case (diesel or
electric air compression) would be about the same – about
7%.
(In principle,
the station could recover about half of the compression energy
lost when pressurizing their storage tank while filling
yours; but in practice, it will be too expensive to recover
any appreciable amount of this energy – at least without
taking an hour to fill your tanks.)
How much
would the station charge to refill your Air Car if there
were thousands of customers in
the area? Probably about $12 – nothing
close to the $0.25 claimed by advocates. (Indeed, 7 kWhr of coal
energy costs less than $0.25, but that is an insignificant portion
of the total cost of filling the air tanks.) And you would only
go 25 miles on that $12 charge in normal city driving – not
120 miles.
What about local air pollution? The diesel engines driving
the compressors at the filling stations will consume much more
fuel
and emit much more pollution (possibly 10 times more) than would
be the case with consumers using compact hybrids instead of Air
Cars.
Finally, how much might the
Air Car cost? The cost of 4 huge 3800 psi air tanks alone
might
be $9,000 if made of
a light-weight composite, though aluminum tanks might cost only
$2000. The engine and
transmission
will not be much cheaper than those in conventional small cars.
Perhaps if the air storage volume were doubled (a big tank on
top of the car, and one filling the trunk) and pressurized
to just 50 bar (730 psi), the Air
Car would be close to being competitive in some urban settings (as the
tanks would cost only half as much). However, the huge air tank
on the top would
probably limit top speed to 30 mph. So even if nuclear-generated electricity
is cheap and abundant, there is no way the Air Car will ever make sense.
We return now to the disinformation released
on air cars by companies such as Zero Pollution Motors. It appears
that our efforts to expose this disinformation has led MDI to
be more careful about the misrepresentations on their website
http://www.mdi.lu/english/produits.php ,
but they still abound – particularly
with respect to performance, cost per mile, and pollution. Unless
the energy on the electric grid is over 80% clean (wind, nuclear,
hydro...), a mid-sized air car results in the release of more
pollution than the Prius burning conventional gasoline. In most
countries, the grid is less than 30% clean.
Bottom line: The
Air Car is in all respects a fundamentally flawed concept.
It doesn’t compete
with electric cars even with 15-year-old battery technology. It
will never compete with the
electric car of tomorrow in any market, and we don’t think
the electric car will compete successfully with the hybrid in most
markets for at least the next four decades – but that is
another story.
References:
Recent article in JRSE on some theoretical limitations of CAES:
WF Pickard, NJ
Hansing, and AQ Shen, “Can large-scale advanced-adiabatic
compressed air energy storage be justified economically in an
age of sustainable energy?”, JRSE 1, 033102-1-10, 2009. http://jrse.aip.org/
http://scitation.aip.org/getpdf/servlet/GetPDFServlet?filetype=pdf&id=JRSEBH000001000003033102000001&idtype=cvips
One of the few, sound discussions of how an ultracapacitor works:
http://www.mpoweruk.com/supercaps.htm
Energy Storage Association. Best overview of batteries:
http://www.electricitystorage.org/site/technologies/
advanced lead-carbon batteries
http://www.greentechmedia.com/articles/read/axions-lead-carbon-batteries-sweet-spot-for-micro-hybrid-vehicles/
http://en.wikipedia.org/wiki/Grid_energy_storage
http://en.wikipedia.org/wiki/CAES
http://www.sandia.gov/ess/About/docs/haug.pdf
http://www.bine.info/pdf/publikation/projekt0507englinternetx.pdf
http://www.geocities.com/CapeCanaveral/Lab/8679/battery.html
http://www.eere.energy.gov/de/cs_energy_storage.html
http://www.seco.cpa.state.tx.us/re_wind-reserve.htm
Air Car hype:
http://www.mdi.lu/english/produits.php
http://www.autobloggreen.com/2008/01/11/tata-motors-mdis-air-car-requires-nearly-two-years-of-work/
http://www.espcinc.com/
