5 ms·
Nice to see the rest of the world isn't content to let Boeing/the FAA off the hook. I'm also seeing every indication I was confident would be found back in Marc
by salawat 7y ago
Nice to see the rest of the world isn't content to let Boeing/the FAA off the hook. I'm also seeing every indication I was confident would be found back in March...
Poignant points follow.
>The report found that while the F.A.A. had been made aware of MCAS, “the information and discussions about MCAS were so fragmented and were delivered to disconnected groups” that it “was difficult to recognize the impacts and implications of this system.”
>The task force said it believed that if F.A.A. technical staff had been fully aware of the details of MCAS, the agency would probably have required additional scrutiny of the system that might have identified its flaws.
This troubles the crap out of me. I was under the understanding aviation used a radically different development process specifically designed to avoid this type of thing. However, this rings of a bad attempt at courting with some sort of Lean or Agile process methodology. I certainly hope that's not the case, but I've seen those same exact symptoms when pressure is put on not properly documenting/analyzing things beyond the bare minimum required by law.
>A broad theme of the report is that the F.A.A. was too focused on the specifics of the new system and did not put sufficient effort into understanding its overall impact on the plane. In certification documents that Boeing submitted to the F.A.A., MCAS was not evaluated as “a complete and integrated function” on the new plane.
>The report also said Boeing had failed to inform the F.A.A. as the design of MCAS changed during the plane’s development. A New York Times investigation revealed that the system changed dramatically during that process, making MCAS riskier and more powerful, and that key F.A.A. officials were unaware of major changes to the system.
>The task force said the certification documents that Boeing provided to the F.A.A. “were not updated during the certification program to reflect the changes” made to MCAS. It added that two critical documents that describe the potential dangers of a system like MCAS, the system safety assessment and the functional hazard assessment, “were not consistently updated.”
These three points are consistent with a regulation scheme where the primary regulator is no longer the primary driver of the regulation process and is dependent on the regulated to raise awareness of "there may be issues here." Active participation of an adversarial regulator is absolutely essential.
>Boeing also failed to thoroughly stress-test the design of MCAS, according to the report, which found that “the design assumptions were not adequately reviewed, updated or validated.”
This is straight up unwillingness to test, "Steve says this'll never happen" cultural attitude. There is always a push against testers to minimize the cost of their testing as much as possible. This reinforces for me how strong that pressure really was. The single-event upset testing they did back around June-ish, while an admittedly highly unlikely occurrence is a textbook bog standard test case in any aerospace/safety critical design. To not have it accounted for, is tantamount to heresy from this humble tester's point of view. Lord only knows what other tests were shot down because everyone else in the management chain were more interested in getting the plane out the door.
>In addition, the report criticized Boeing for not adequately assessing the extra effort pilots might have to make to deal with MCAS, and it noted that Boeing had removed mention of MCAS from a draft of the pilot’s manual. As a result of that decision, some key F.A.A. officials were not fully aware of MCAS and were “not in a position to adequately assess training needs,” the report found.
This and the previous statements w.r.t tests undone, and documentation/awareness raising failures constitute for me sufficient evidence of a potentially deliberate regulatory hack. I'm glad we have Attorney's General conducting a criminal probe in this case. The convenient way the failures just happen to fall into line and that everything just happened to go off without any of the people who could have stopped the process of this error chain from propagating seems to me a bit far-fetched. Accidents happen, yes. This represented too many disparate failure modes in the process to not have been the result of a deliberate optimization decision made by someone; and that decision maker needs to be held to account.
>To address some of these shortcomings, the report recommends that the F.A.A. update the certification process to allow the agency to be more involved in the design process early on.
>Overall, the report found fault with the process for certifying a new plane based on an old design, saying that it “lacks an adequate assessment of how proposed design changes integrate with existing systems.”
These two points revolve around the grandfathered certification process in general, which has been a big motivator in terms of disincentivizing designing new planes. I'm fairly sure that there needs to be a greater threshold of proof tacked onto a grandfathered design after this point. I understand the intent to streamline by utilizing already proved design elements, but it is fairly obvious it invites in design complacency from the ground up as it were, and encourages the financial side of the house to feel emboldened in attempts to pressure engineering into skipping large portions of the work needing to be done.
>It recommended that the F.A.A. confirm that the Max is in fact compliant with regulations having to do with the plane’s flight guidance system, flight manual and stall demonstration.
This last point may doom the MAX depending on how accommodating regulators are willing to be.
Without MCAS, the airframe fails FAR 25.173 from the pilot's point of view. Period. That is beyond contention at this point. There will need to be a judgement call with regards to the delta in control forces as to whether or not a regular pilot, through training alone can be educated to safely compensate for those longitudinal stability prescriptive behavior divergences. In the end, it is left to the certifying authorities.
If it were me in that position, with the authority to make a binding proclamation on it; my answer would be no. Absolutely not.
I've dug into the history of aircraft certification enough to agree with some of the older school regulators from times before electronics were a given for flight control systems. We must make flyable planes first. If we let the viability be compromised by economic factors compensated for by gadget wizardry, then we're walking down the path where pilots are not "flying" the plane. This is unacceptable in my eyes. If we are going to put as much trust in pilots to the point that we lock the occupation behind years of education and draconian requirements for physiological conditioning and educational refresh, then the pilot should be as much a design consideration in the flying of the plane as possible. Which means at a minimum, the pilot needs to be able to fly the machine with minimum automation functional by virtue of passive flight characteristics. I'm not against fly-by-wire as a means of control signal translation to provide common interface for the pilot, mind, but the computer should not have to be resorted to to mimic passive longitudinal stability. That is a sin that cannot be pardoned in aerospace design for Civil transport, and yet we've had more than enough history of attempts to do so to make it clear it's a bad idea.
First the MD-11, now 737 MAX.
It needs to stop.
- PaulHoule 7y agoThere is a precedent for "flight envelope protection" for fly-by-wire airliners, which has been applied to all modern larger airliners except for the 737. (directly competitive A320, as well as A350, A380, 777, 787, 747-8, ...) Small problems with the fly-by-wire system on the A320 get found and fixed on a routine basis. Bigger problems have been found in the past, but how to manage the risks in this kind of system is well known and that knowledge base was entirely ignored in the development of the 737 because Boeing couldn't be bothered to bring the 737 into the 1980s, never mind the 21st century.
- salawat 7y agoGiven, but flight envelope protection has it's warts too, and I don't wish to (and probably chose my words poorly to convey my true meaning) turn this into a Boeing v. Airbus design philosophy flame war. Where your examples pass my test of acceptable fly-by-wire, is that even in their most automation crippled state, (direct law), the airframe demonstrates longitudinal stability, and the automation doesn't induce undesired behavior detrimental to flying the plane. Thereby it can be thought of as a Hardware Abstraction Layer between the inputs and the control surfaces that faithfully represent pilot inputs to output behavior. Boeing's MCAS fails this sniff test. If the input sensor breaks, it actually generates dangerous input that actively detracts from the flyability of the plane. The airframe itself necessitates this dangerous subsystem because uncorrected, no test pilot would be willing to sign off that it was consistent with the older 737's handling characteristics. Just wanted to be clear: I'm not anti-Airbus. I'm anti-"Hey let's automate this, but do it with minimum scrutiny and reckless levels of awareness building for the end operator." Which is what happened with the MAX, and seems like a page torn out of McDonnell Douglas playbook.
- PaulHoule 7y agoIt is not Boeing vs airbus, it is the 737 vs all other modern airliners. FBW on the 7x7 (for x>3) is not that different from Airbus airliners except Airbus has sidestick controls that are not mechanically connected whereas Boeings have jokes that are mechanically between the two pilots. If Boeing used the same flight controls that they use on all post 1970 planes it would be ok, Airbus has nothing to do with it.