Bjorn’s Corner: The challenges of hydrogen. Part 25. Auxiliary power

By Bjorn Fehrm

February 19, 2021, ©. Leeham News: Last week we discussed hydrogen aircraft propulsion and found a shaft power producing gas turbine was considerably more weight-efficient than a fuel cell powering an electric motor. Both had the same 55% shaft power efficiency.

Will a gas turbine APU burning H2 be the best choice for auxiliary power as well?

Figure 1. The Honeywell 131-9 APU for the Airbus A320. Source: Honeywell.

APU or fuel cell for auxiliary power?

We found the fuel cell as a shaft power source for driving a fan/propfan/propeller ran into complications both for a 165 seater turbofan aircraft and the 80 seater turboprop airliner. The output from a fuel cell is electric power, and we need shaft power.

You must add electric motors, and these must be fed with the correct alternating current, forcing conversion electronics at the fuel cell and motors. It ended up three times heavier than a hydrogen-burning gas turbine core without offering higher efficiency.

For auxiliary power, to the aircraft’s systems, it’s the other way around. Most systems want electric power, and a shaft power-producing APU must add a starter-generator to produce electric power to consumers like the cabin and the aircraft systems running on electric power.

APUs also deliver bleed air for engine start and air to the ECS (Environmental Control System = aircraft air conditioning) at the gate. Bleed air is an inefficient form of power, bleed from the engine’s compressors during flight and the APU during stops. It’s too hot coming from the engines to distribute as-is and passes a pre-cooler in the pylons, ejecting about one-third of the energy to bypass air to get the air temperature down to around 200°C. Thus a third of the engine bleed air energy goes to waste before it’s distributed in the aircraft.

An APU also drives an electro-hydraulic pump through its generator to power brakes during towing and any movables needing maneuvering for service.

A more modern systems architecture

A modern aircraft design would change the architecture of the aircraft to a “more electric design”. Here the inefficient bleed air is replaced with electric power, and hydraulics is made local to where it’s needed (using electro-hydrostatic actuators). Such a design can reap the benefits of fuel cells replacing the APU.

In addition, the engine-based auxiliary gearbox can be skipped and replaced with starter-generators in the engines as described here. The ECS would use electric compressors and the wings electric de-icing like on the Boeing 787. Control surfaces that need hydraulic power use electric-powered self-contained actuators, as do the brakes.

We now have a design where a noisy and inefficient APU (about half the efficiency of an engine core) is replaced with quiet and efficient fuel cells. The APU sits at the aircraft’s tail for sound reasons. It’s very noisy and shall be placed as for from boarding and deplaning passengers as possible.

A fuel cell is quiet and can be placed where the primary consumers are, in the wing root area where the ESC, main landing gear, slat, flaps, spoilers, and the wings’ de-icing come together. This moves weights forward in the aircraft, which is a bonus for a tail tank architecture hydrogen aircraft.

Figure 2 shows a ballpark power and weight table for our 165 seater.

Figure 2. APU versus Fuel Cell auxiliary power comparison. Source: Leeham Co.

It’s not a detailed comparison as the figures are first-order approximations (there are a lot of “what counts to what” issues at play).

The purpose is not to make an exact comparison; it’s to check if we have a gross imbalance between today’s technology, the carbon or hydrogen-fueled APU, and an alternative based on fuel cells. In this spirit, we put the piping weight of air, hydraulics, and electrics for the APU alternative equal to a more elaborate electric distribution system for the fuel cell alternative.

We can see the fuel cell alternative is a bit heavier, but it’s also about twice as efficient as an APU. APUs are designed to be small and light, rather than efficient. With a dual fuel cell setup, we have enough redundancy to let the fuel cells be sole power to the aircraft’s systems, contributing to an efficient architecture, as we can then replace the engine auxiliary gearboxes with low power starter-generators. The efficiency is increased as fuel cells produce H20 and O2 that we can use as onboard water and emergency oxygen.

We use the engines’ starter-generators as emergency backup power. The generators are not active during the flight, just for engine start. By it, we can skip the engine auxiliary gearboxes and bleed/auxiliary power off-take from the engines, thus increasing engine efficiency. The starter-generators chip in to help the core accelerate/decelerate when needed to allow a more aggressive compressor design, increase the efficiency of the core further.

In all, the fuel cell is an attractive alternative for the auxiliary power function of a future hydrogen airliner.

42 Comments on “Bjorn’s Corner: The challenges of hydrogen. Part 25. Auxiliary power

  1. just to make sure I am understanding correctly,

    you are saying that both on the ground and in flight all (non-emergency) electricity and formerly bleed air driven functions would be powered by a pair of fuel cells and that the turbines would only have “emergency” generator capability should a fuel cell fail and no bleed air, correct?

    what % efficiency gain does that deliver to the turbines (i.e. what % of engine power is normally diverted to bleed air and electric generation)?

    • Hi Bilbo,

      yes, this is correct. You zap around 0.3-0.5 MW from the engines cruise 6MW so around 5% to 8%. But this load is forcing the engine to run the aerodynamics more conservatively as it’s troubling when the engine is running in low power states, say 1-2MW flight idle. If you can skip the bleed and shaft power offtake from the engines you can increase the efficiency of the engine. Say you can gain one or two percent, which is a lot.

      • You have to compare the engine compressor efficiency with the electrical powered cabin compressors of usually Roots type screw compressors of lower efficiency than an engine compressor flr the same pressure/flowrate. Then the hot air is cooled and flows into regular air cycle machine before mixing with filtered recirculated air in a commercial jet (som biz jets just use fresh air). The airflow should be equal to the outflow valve flow.

        • One often annoying feature of travel is the aircraft air conditioning not being on during boarding or de boarding to save fuel and the time to unhook ground power. Hopefully that would end. The cryogenic latent heat of vaporisation would provide some cooling as well.

    • It is not only efficiency gains. The first H2 turbines will be kerosene based turbines modified for burning H2. They will be iterated fast into a truly H2 turbine but that iterative process is made easier if you only have to deal with two connectors, electricity and H2, in the pylon instead of three, electricity and H2 and bleed air.

      • “truly H2 turbines’? The turbine is where the energy is extracted, the burners are where combustion occurs, no reason to think the engine manufacturers with their vast resources cant come up with ‘truly H2 burners’ to work for existing cores .
        More likely changes are for the ‘clean burning’ of hydrogen and no need to account for carbon fuel emissions etc and a further jump in efficiency

        • Think Char ment that with the cold LH2 flow available you can design new engines that use the immense cooling possibility this opens up. Not only for turbine vanes, air heat exchanges but bearing compartments as well. So step by step this will make engines burning LH2 (does not need to be 100%) will evolve quickly in area after area using the cold LH2 in different ways.

          • Also, but that is more generation 3 stuff. Burning H2 is comparable to burning kerosene but not the same. Flame propagation is (much) faster. More energy per O2 used. No S or other contaminants. I, as amateur turbine designer, would think that more energy per O2 would lead to a higher bypass ratio. Higher flame propagation to a shorter distance between injecting the fuel and the first blades & fewer contaminants to a greater selection of (cheaper) materials the turbine can be made of. You could say do the optimizing before first flight but historically this has been done during production. It is also less risky.

          • The engine core is designed around the plane its destined for requirements. Example is the GTF which comes in a range of BPR and fan sizes. For the largest power version for the A320 series theres an existing choice of power outputs from 24,000lb to 32,000lb. Clearly it comes from adjusting the fuel flow.
            Like I said and Im sure you know the power for the front fan comes from the 2 stage turbines not the ‘burners’.
            The highly compressed air tapped off for internal cooling purposes seems to be more useful than low pressure cold H2 which is known to have piping issues. The main purpose of the H2 is fuel not really cooling and it puts the fuel supply at risk in micro passages for no real advantage when a simple inter-cooler will do.

          • @Dukeofurl

            Those single stage to orbit planes need the liquid H2 to liquefy the air but in this design the air is just cooled with a cooling loop that dumps its heat into the H2 going to the engine. Likely with an electricity extraction device and gas phased H2 injected into the turbine not much colder* than the compressed air.

            * Doing a very ruff calculation without compression etc. i get 140K if you mix liquid H2 with -60C/210K air in a mix that would use all the oxygen if burned. That is cold but not have to use H2 in the cooling loop cold.

          • Regarding burner design you are right that the H2 flame velocity is much faster and hotter. The heat generates NOX as a function of temperature and time, hence you want to quench the flame as fast as possible using the compressor exit air not part of the combustion. The cooler it is the quicker you quench the flame and you want a pretty even temp distribution into the Nozzle Guide Vanes of the HP turbine.
            Simply the higher temps of H2+Air increases NOX but the higher flame velocity reduces the time NOX can form.

      • The combustion chambers of hydrogen powered gas turbines can be much shorter saving in weight. The fuel burns much easier hence vin Ohain made use of it when he ran the first jet engine,

        • No. von Ohain wasnt the first to ‘run’ a jet engine, it was Whittle with the first working jet engine in April 1937 , von Ohain with Heinkel had the first ‘flight’ in 1939.
          Whittle was much more hands on ( including being a pilot) and physically made many parts , in the pathway of the Wright Bros while the more academic von Ohain ( whos PhD was on sound recording for films) used skilled machinists and worked with closely with Heinkel.

          • von Ohain ran his engine 2,weeks before Whittle in Feb/March. The two men met and compared notebooks. Von Ohain however gave Whittle credit for the first jet engine because he said Whittle had solved the combustion issues first. Hence the two men got on. von Ohain’s Engine operated on hydrogen because the Germans wanted to separate turbine compressor design and matching from combustion hence the use of hydrogen.

  2. Bjorn, in this article you seem to mix the more-electric (M-E) aircraft benefits with those from a fuel cell. Couldn’t the M-E aircraft also be accomplished with a turbine APU running a similarly large generator?

    It would be interesting to see a comparison of all three options (current, M-E with turbine, M-E with fuel cell). I suspect the turbine APU would be competitive then, although as you point out there are other non-tangible benefits of the fuel cell, such as noise, vibration and location in the airframe.

    • Thanks Rob,

      I’ll complement up to a complete comparison as per your request next week.

      • Thank you, Bjorn. I was thinking the M-E aircraft development can precede hydrogen aircraft development, and then the fuel cell should fit right in when hydrogen arrives.

        • If all downstream consumers are electric (more than likely for the next all new aircraft) and the upstream fuel storage is H2, then it makes no sense to use a turbine to convert H2 to electricity in 2030. Fuel cells are much more efficient, and should be fully mature for aero applications by that time. And they bring many other advantages to the table : re-use of LH2 boil off, silent, no need to design against turbine blowing up and projecting fragments, and should have a much more constant performance vs altitude. A turbine will only make sense if you keep a high-pressure air system (bleed)

          • The point was that the M-E aircraft needn’t wait for a production aircraft fuel cell with LH2 fuel. It could be done now using existing turbine APU’s. Then the technology is already in place for the future, and the current generation of aircraft can benefit as well.

  3. One should consider that the turbine APU’s efficiency varies over the load profile. For example, the 777’s APU is only about 15% efficient over its normal operating profile and time in mode. The 787 MEA architecture’s APU is a few % more efficient. However, the fuel cell’s efficiency isn’t as sensitive to off-peak conditions.

    With regards to your previous post, a Solid Oxide Fuel Cell (SOFC) has been modeled to be about 70% efficient at altitude because it can use a power recovery turbine to harness the SOFC’s high grade exhaust heat vs the low grade waste heat of a PEM that is difficult to get rid of. Also, consider that a propulsive SOFC’s exhaust has much less volume and temperature than a turbine engine so one might be able to treat its exhaust in order to reduce or possibly eliminate contrails that contribute as much to global warming as the engine’s CO2.

  4. Hi Bjorn,

    If I remember well, for this size of aircraft, certification requests a three independent power sources (either hydraulic or electric) architecture, with no common point of failure.
    This were the 3 hyd pumps come from in your table.
    So, no matter fuel cell reliability or power, you would need not 2 but three of them to replace all engine driven pumps / generators, isn’t it?

    • Hi Alban, as in the article, we have two independent fuel cell systems producing electricity and then, additionally, two independent starter-generators in the engines. This adds to four independent sources of electric power on the aircraft.

      • Thanks Bjorn, I should have read you more carefully 😉
        That’s make indeed four, provided that each fuel cell is supplied by a dedicated H2 tank.

    • I think 2 is more a number. The Mirai is 100 kW and i think that standardizing on car fuel cells may be the better solution* so having 6 Mirai fuel cells is a possibility

      *) I don’t believe that hydrogen fueled cars will ever really be a market. But better to let somebody else pay for the development.

  5. How do fuel cell produce O2?

    I could see this done with liquefying the air with the liquid H2. But wont that be terrible complicated with H2O and CO2 . It would make the cabin compressor not necessary. You wont inject fresh air but only O2 and capture the CO2 and H2O inside the cabin air before recirculating the air. Add a tank for the captured CO2 and you have also a solution for the weight imbalance.

    • The fuel cell would not produce O2 itself, but would need to be fed O2 to react with the H2. O2 seperation systems already exist on aircraft to provide oxygen-depleted air for fuel tank inerting, but would need to be significantly upsized. The relatively small need for emergency O2 could be incorporated in there, provided all necessary redundancy measures are implemented in the design.
      So maybe a more exact proposition is that the “fuel cell system” produces O2.

      And while we’re at it, the bigger volume of oxygen-depleted air might be usable to assist with fire-fighting duties, in replacement of halon.

  6. Hydrocarbon powered Solid Oxide Fuel Cells with reformers are now close to 60% efficient. I wonder if they will be mature enough to incorporate into a next-gen single aisle as APUs to get the benefits of an “electric architecture” aircraft without waiting for a hydrogen based solution.

    Yes you could do that with APUs but fuel cells are nice in that you can get the capacity of a large one from two smaller ones without much of a weight penalty. The also start quickly and retain efficiency under lower loads.

    • This would work well with a PtL (power to liquids) carbon neutral jet fuel. Whether cryogenic hydrogen or PtL carbon neutral fuels will be expensive and aircraft efficiency will be critical. Unlike hydrogen there are commercial flights with small additions of PtL fuel already flying.

  7. Does the availability of a fuel cell make electric ground taxiing attractive enough to finally implement? I imagine the fuel cell will be on more often than a APU.

    • The problems with the electric taxi systems were economic, rather than having a sufficient source of power. The aircraft engines need to be warmed up before flight anyway, and the cost of tow vehicles is not that large in comparison to the fuel savings obtained. Plus one tow can service many aircraft with no weight penalty.

      There is an electric robotic tow vehicle development “TaxiBot” that connects to the aircraft, and can be controlled by the pilot. It’s certified but has not been widely adopted, for same economic reasons.

      If fuel prices rise substantially, the economics are more favorable for electric taxi.

      • Yes. Running engines at idle before push out is needed to check the various engine systems are OK, no point getting to the end of taxiway and finding a problem.

        • Yes and no.
          The electrical and hydraulic energy you need to energy to check most of the systems can be provided by APU or battery. So you can delay engine start.
          Of course, finding an engine related issue after leaving the gate is an increased burden, but, statistically the chances are low.

          • Any sort of synthetic carbon neutral fuel will be very expensive so the savings will be significant.

      • You can add to the list that designing an integrated electrical taxing system without affecting safety and reliability is quite challenging.
        For weight on wheel reason, only the main landing gear is the only suitable location and you can’t risk any interference (jammed wheel for example) during take off or landing…

        Safran Landing Systems is working a lot on it for nearly a decade now (see Electrical Green Taxing System) and still hasn’t a solution mature enough for flight testing…
        Here is a ling to a presentation from 2013:

        I share Rob opinion that electric tow robot (potentially remotelly commanded from the cockpit) is a more suitable option for aircraft flying efficiency.

  8. How much energy does an APU consume relative to main engines?

    Boeing tested a fuel cell APU years ago, using a retired WestJet B737 in the hanger of BCIT at YVR.

    APU has to be same fuel as main engines, especially in case of getting stuck overnight. I was getting a reputation for sending engineering staff to the High Arctic to get stuck overnight. They got up with the flight crew at least once in the night to run APU for a while to keep water from freezing. I never had that when I went north. 😉

    [Well perhaps I spent more time with the Hercs than the Boeing’s, definitely overnighted at YRB as routine for the Herc operation as it was operating Up There whereas the Boeing’s were in and out. First morning up and walking to the Herc I heard an odd airplane engine, the Bristol Freighter piston engine, mechanics had to get up early to warm it up. The Bristol was useful to take in a small bulldozer to open a runway for the Herc, for smaller loads the turbine Twin Otter was a revolution in dependability. The Herc took modification by Pacific Western, seals on brakes and prop blades were trouble – I suspect the NY ANG never shut theirs down in Antarctica.]

    APUs in those days were crude affairs, simple pneumatic control system, probably not fuel efficient. Engineer Terry Nord designed a modification to replace the C-130’s APU with the B727’s. I presume they are now much better, three generations I suppose (747 I don’t recall, L1011 had PT6 Twin Pack as its base, a new consortium arose because Airesearch precursor of Honeywell products was not responsive to reliability concerns).

  9. So this makes hydrogen fuel less theoretical – I just walked by a sign in my neighbourhood announcing planning for a hydrogen fuelling station (I guess we should call them “gas” stations). Public consultation on the 25th, construction to start July 2021. Anyone interested can sign up for the public consultation session at

  10. I’m uncertain where the LH2 boils in this design. Some comments advocate doing it at the engines to cool the combustion and reduce NOx emissions.
    This involves piping LH2 from the rear tanks to the wings (or making a rear engined plane like a VC10 or B727, but who wants all that weight back there?). Boiling the gas by the tanks makes it much harder to utilize that coldness. What am I missing?

    • The boiling discussed is the unavoidable boil-off in the LH2 tanks in the rear. You pipe this off the thermos flask tanks to avoid pressure build-up higher than the slight overpressure you run in the tanks. Go back in the series, we discussed it in Part 4-6. When LH2 boils to H2 it consumes a lot of energy and this keeps the remaining LH2 at -253°C in the tank. The boil-off you can use in an APU or Fuel Cell or just went it off the aircraft (look at the Airbus ZEROe turbofan airliner, it has a boil-off chimney at the top of the vertical stabilizer). To the turbofan engines, you route LH2 that is turned to H2 in a heat exchanger before entering the combustors. The Airbus Cryoplane study suggested a winding pipe heat exchanger around the turbofan’s core nozzle. You could also make smarter use of this strong cooling effect in a recuperated compressor design and/or cool the turbines.

      • Thanks Bjorn – Yes, I understand about why it has to boil and the latent heat required. My question really is how difficult is it to pipe LH2 from the tanks at the back to the turbines under the wings (as opposed to warmer gaseous H2)?

        • Two issues i see immediately with LH2

          The LH2 pipe is cooled by moving the H2. If it is standing still it will turn into gaseous H2 that has a much bigger volume. But gaseous H2 has the almost as big a problem going from 20K to 40K

          LH2 is cold enough to liquefy air. During that process the air is separated into liquid O2 and gaseous N2. O2 is great to create fires and only N2 makes breathing a short time thing. But gasified H2 is likely also cold enough for this to be a problem.

          Gaseous H2 has also problems

          Needs much bigger tubing

          a way to heat the LH2

          VP/T= constant. T is in K so small delta T will have a big influence on VP. Makes getting a constant mass flow hard on the other end of the tubes.

          still cold enough for O2 problems

  11. Were you aware of an initiative by the Swiss materials and testing agency, EMPA, to research the feasibility and economics of synthesising kerosene from hydrogen and carbon dioxide? Not quite on topic, of course, but an interesting parallel approach to aviation fuels of the future

Leave a Reply

Your email address will not be published. Required fields are marked *