June 11, 2021, ©. Leeham News: Last week, we scaled a nine-seat aircraft to a 19-seat aircraft and examined some of the pros and cons of such a change. The aircraft are certified to the 14 CFR Part 23 rules in the US, labeled “Normal Category Aircraft“.
This week we scale the aircraft up one step further to understand product certification and operation rules for the larger Transport Category Aircraft (14 CFR Part 25) class.
Having less than 19 passenger seats, as we have previously discussed, does not automatically mean that an aircraft falls under Part 23 (“Normal Category Airplanes” in FAA speak). Many business jets, such as the Citation XLS+ or the Global 7500, are Part 25 aircraft (“Transport Category Aircraft”) since they exceed the 19,000 lb maximum certified takeoff weight threshold. On the other hand, having more than 19 passenger seats always puts the aircraft in the Part 25 category. So, what are the major differences between a Part 23 and Part 25 aircraft?
As previously discussed, as the size and passenger count of an aircraft go up, the expected level of safety goes upwards as well. As a result, Part 25 contains a number of additional and more stringent rules. These differences can be about the capabilities of the airplane or the method in which it must be demonstrated. We give a couple of examples. For simplicity of comparison language, Part 23 texts are drawn from the prescriptive version instead of the new performance-based version.
Under 23.775 for Commuter Category Airplanes, you’ll find the following requirements for airplane integrity as a result of a bird strike: “Windshield panes directly in front of the pilot(s) in the normal conduct of their duties, and the supporting structures for these panes must withstand, without penetration, the impact of a 0.91 kg (2 lb) bird when the velocity of the aeroplane relative to the bird along the aeroplane’s flight path is equal to the aeroplane’s maximum approach flap speed.” (So in the speed bracket of 120-180kts, our comment).
In contrast, under EASA CS-25 (the EU Transport Category rules):
25.631: “The aeroplane must be designed to assure capability of continued safe flight and landing of the aeroplane after impact with a 4 lb bird when the velocity of the aeroplane (relative to the bird along the aeroplane’s flight path) is equal to Vc (Vc is the design cruise speed, our comment) at sea-level or 0·85 Vc at 2438 m (8000 ft), whichever is the more critical.“
25.775(b): “Windshield panes directly in front of the pilots in the normal conduct of their duties, and the supporting structures for these panes, must withstand, without penetration, the bird impact conditions specified in CS 25.631.”
And interestingly, this is even more constrained when we add in the US 14 CFR 25.631 from the FAA: “The empennage structure (the tail of the aircraft, our comment) must be designed to assure capability of continued safe flight and landing of the airplane after impact with an 8-pound bird when the velocity of the airplane (relative to the bird along the airplane’s flight path) is equal to Vc (design cruise speed) at sea level..“
So, under Part 23 we’re primarily concerned with a 2 lb bird impacting the windshield at landing speeds, while under Part 25 we must design for a 4 lb bird hitting a larger frontal area at up to cruise speed. In addition, this is an example of a non-harmonized regulation, where if we want to sell the aircraft in the US we must design for an 8 lb bird hitting the empennage structure at up to cruise speed. Part 25 rules add significant additional weight to an aircraft to achieve the needed impact resistance, more so for the US market than Europe. It also adds development and testing costs for the analysis on how to design for and later demonstrate compliance.
Next week we continue with more examples of the differences between Normal (Part 23) and Transport Category (Part 25) Aircraft.
“There are two kinds of birds in this world, Tuco. Those who fly and those who get thrown at planes…”
(I’m not sure what is worse; Getting eaten by a natural predator in the wild, getting shot out of the sky by some guy with a shotgun, ending up on a dinner table after being raised in a factory – or being smashed against an aircraft during testing. If reincarnation is a thing and you come back as an animal, I’m not sure that ‘bird’ would be my first choice…)
I guess the birds would dead before hand
EASA seems to have started the regulatory work of only one pilot for commercial airliners and no pilot for UAM’s with autoflight and digital communcation with ATC. Any news on how they think systems and aircaft certification will be effected? The cost benefit for UAM’s are enormous and alternate emergency landing spots quite large as they pretty easy can be built.
What disturbing me about large FAR 252 and 121 single pilot aircraft is the psychological state of the pilot will not be monitored by another human.
Hopefully the work on unmanned vehicles, drones will shake up the staid and slow practises of aircraft certification, regulation and safety improvement. It will be a radical change. When the famous “Sully” United Airlines Flight 1549 ditched in the Hudson in 2009 after both engines ingested birds I was puzzled as to why a computing system was not on board continuously computing alternate emergency runways to glide into and execute a dead stick landing, or even rivers and seas to ditch into etc. Instead this calculation has to be carried out manually thereby wasting crucial time that that was needed.
Drone engineers can’t afford the often false security of a pilot and to assume they can hand over a failing system rather than attempt to keep it working plausibly as much as possible. The software engineers who disabled the angle of attack sensors on AF447 when pitot-static tubes speed sensors froze over were writing to specifications that simply handed every problem over to a pilot and did not attempt to use the data they had on board (such as GPS, INS) to keep those alpha sensors active.
I am hoping that the development of certification standards for unmanned vehicles will lead to positive improvements in manned ones.
IMU you underestimate the scope expansion of possible errors from adding in INS and GPS … systems into the computation.
Boeing is all hip about synthetic data sources augmented by non aero positional information.
I do have some doubts about this having as low an error probability as Boeing believes ( or at least writes about ).
Look at the bugs exhibited by Honeywell’s ADIRU in those A330 dive cases.
KISS is a good design metric!
There couldn’t be anything more absurd as a Prandtl probe (I prefer to call it after its inventor Ludwig Prandtl than a pitot static tube).
They’ve been failing and killing large number of passengers for decades. They freeze over, they get clogged from simultaneous bird strike, they develop wasp nests in them, they get taped over during washing and painting procedures. It seems folks haven’t accepted that the atmosphere is full of birds, insects and sometimes ash. That people think 3 is enough on an jet aircraft is perplexing. I would think 6 is enough several with self closing probe covers.
Synthetic Air Data in principle is simple. You use INS data of actual velocity and altitude, compare it with the heading of the aircraft to calculate cross winds, use the aerodynamic data of the aircraft in consideration of its flap and thrust settings and estimated weight to get air data. The INS accelerometers keep it responsive. It’s robust but not 100% accurate if say weight is out but its much better than nothing and its reliable and accurate enough to keep the aircraft flying in event of far less robust alpha and pitot static tube failure.
We will see how they solve it. Routes preprorammed, FIM (Failure investigation Manual) and som AI in the aircraft an the network to ATC and the aircraft OEM to decide when to let control to the pilot, then as flight is stabilized the computers take over again. The number of combinations of faults can be extremely large and hence hard to program correct reposnse to every possible combination of correct and false warnings. Just look at the MCAS with one bad alfa probe can set a unexpected chain of events in motion, imagine what a combination of up to 12 sometimes faulty probes together with some MEL’s and engine and APU failures in combination of fire onboard would require of the programmers.
The perfidy of the MAX crashes was that Boeing actively avoided looking at the bouquet of possible faults up front.
The cockpits are a Christmas tree of ‘automated warning lights’, to let the pilots know of some condition or other. The GPWS was put in to avoid the controlled flights into terrain, but of course the famous one where the warning sounded and the CVR later showed the pilots response was ‘Shut up Gringo’.
Some recent ‘runway crashes’ have shown some airlines now have mandated procedure to put the plane ‘in auto’ from wheels up to touchdown so they both become ‘pilots monitoring’ but that isnt working out as well as they might expect.
I think the Global Hawk and similar remotely piloted or automatic aircrafts in Military service will debug the system pretty well, then it comes to acceptable reliability, again I assume the military will fly soldiers and goods without pilot first for some years and pay for all the updates and fixes before the OEM’s show all the data to the FAA and then the US avionics manufacturers will be a step ahead. The EU knows this and might try hard to work on certifying UAM’s for auto flight together with the auto industry’s LIDAR and similar self driving efforts and together be ready to certify for commercial aircrafts. Still the US has a golden opportunity to corner this market thanks to its military programs.
US drones in general have an MTBF that you can “watch for action”.
And it does not seem to change much over time.
nobody beyond some heathen die so nobody in command cares as long as the budget is expanded.
Drones like the GH do have pilots, each time they fly it’s under pilot control, just not on board.
It’s a fundamental misunderstanding you have, and thus will achieve nothing by looking at what they do to proof test unpiloted flight
No doubt if approaches the issue of possible fault accumulation as one of permutations over the whole the problem looks intractably large. Perhaps the the approach of analysing all possible permutations of maintaining robustness can be approached using artificial intelligence.
My own view is that each individual subsystem should have 1 large reserves redundancy built in and 2 degrade gracefully rather than just shutdown and throw the mess at the pilot/s.
Triplication is no longer enough. We’ve seen deep space probes that have required quadruplex systems that within each system have duel triplex systems to achieve the 20 year operating life required.
A future generation of pilotless aircraft (with or without passengers) will need an approach that borrows from drones and deep space probes rather than piloted aircraft. In any case aircraft such as LiliumJet can not be flown manually (they directed manually is perhaps the term) and will be the safer for it.
In the case of AF447 let me first say that triplication is not enough.
The ADIRU on AF447 have data from the alpha sensors, prandle/pitot static tubes, barometric, Inertial updated by GPS and likely Loran etc.
All that is needed for these ADIRU to carry out a synthetic air data calculation is the supplementary data of thrust and flap settings of the aircraft as well as an estimate of weight in consideration of fuel burn.
However even if not all of the above is available an accurate enough calculation is possible.
However the issue with AF447 was that when all 3 pitot static tubes froze over not only did air speed data disappear from the pilots cockpit but all data was removed from the pilots and the alpha sensor data was ignored thereby disabling alpha protection. The ADIRU had INS data to inform the aircraft was at high altitude and at speed and this should have been enough to keep alpha protections from the alpha sensors alive and fully valid including stall warning.
We are even in the situation that GPWS is disabled on ADIRU fault and then can repeatedly come back and off again.
The attitude and architecture is completely wrong and seems absurd to those writing software for UAV. The tiniest fault leads to a hard shutdown of the entire subsystem rather than a graceful degradation.
Some of the same flaws that lead to the flawed MAX MCAS design were present in AF447: The 3 ADIRU were allocated to Pilot, Copilot and backup rather than being an integrated system.
I would hope that EASA/FAA make Airbus type passive side sticks illegal for new build aircraft (mandate active sticks) so that Airbus doesn’t try to ‘grand father’ its design forever like Boeing did because of minor training impacts.
Yes theres been a number of crashes with A320 where despite having 3 AOA sensors, when theres a ‘contested election’ but the winner is erroneous anyway it leads to the plane crashing…in conjunction with other errors from the crew.
Despite this nothing happens from Airbus….as they can blame the crew.
In 2014 another alpha sensor freezing fault: http://avherald.com/h?article=47d74074/0000
Fortunately not fatal this time.
The whole fault resolution Airbus expected the pilots to to go through to isolate the problem would be comedic were it not so harrowing.