1

The ill fated MAX of recent news had 2 angle of attack vanes, with the MCAS computer picking the worst one.

It seems a "software patch" may not be the only improvement option available. Sensible engineering may suggest that a minimum of three, possibly 5, or even 7 vanes would enable writing a subroutine to reject "outliers" and use an Average AOA within an acceptable Standard Deviation.

Could more AOA sensor vanes, at various locations on the aircraft, be added to this design?

Robert DiGiovanni
  • 20,216
  • 2
  • 24
  • 73
  • 4
    The question in the title is sensible, understandable etc and (potentially) answerable from regulation. The majority of the body is rambling about something else, and then the question(s) towards the end are completely different to what is asked in the subject. Could you clarify this all a bit and perhaps remove the information which is unrelated to the question. – Jamiec Nov 29 '19 at 13:49
  • The question in the title is sensible, but redundant with https://aviation.stackexchange.com/q/61144/17780 – bogl Nov 29 '19 at 13:55
  • 4
    Its just a distraction. If I'm asking how to bake a cake, should I specify what I had for tea, what I'm doing a week next tuesday or what time I'll be home from work? No. Probably not. Similarly, if youre asking about AoA vanes, we dont need to know your opinion on how you think MCAS worked (your opinion, mostly wrong) how computers work or how humans react to changing environments! – Jamiec Nov 29 '19 at 14:16
  • 1
    The description of your question MUST focus on the specific problem asked. As is, except for the second paragraph, it is just a comment to convey your opinion about the MCAS, which is not relevant and blur your real question. Thus I'm tempted to tag this question as unclear. – Manu H Nov 29 '19 at 15:04
  • Ok, Ive gone ahead and put this on hold this as unclear after 3 other votes. Please specify what you are actually asking, and remove the of-topic bumpff and flag for mod attention if you want it re-opened. – Jamiec Nov 29 '19 at 15:19
  • Wow! A simple flow diagram shows the AOA sensor vane inputs connected to MCAS logic. The logic flow could be programmed in BASIC, including the math calculations, rejection of outliers and response. Unfortunately, non-constructive emotional responses are preventing proper examination of a very relevant issue. The question is: ARE 2 ANGLE OF ATTACK VANES ENOUGH. The remainder is some rational why 2 may not be enough. WHAT DO YOU THINK!. – Robert DiGiovanni Nov 29 '19 at 15:28
  • 2
    The problem was that MCAS only took the signal from the left. They did this because someone in their risk assessment organization classified the failure mode of loss of input as "major" instead of "hazardous" to justify one input. The motive would be to improve dispatch reliability (two is less reliable than one if you must have both for normal function), and ease cert. – John K Nov 29 '19 at 16:11
  • @John K I want to make it clear this is not to BLAME, only to move forward with a better aircraft. Putting a few more AOA sensors on there should not be a major problem, as automated flight is analogous to flying blindfolded with limited instructions on what to do. – Robert DiGiovanni Nov 29 '19 at 16:19
  • @bogl your answer 17780 was very good. Notice 2 sensors with pilot intervention does not quite apply here. Also, amazingly, adding more sensors and averaging is one of the easiest fixes! – Robert DiGiovanni Nov 29 '19 at 16:35
  • 1
    They didn't need to add more sensors. All they had to do is take a signal off both existing ones, and configure the sytem disable MCAS, with a fault message, where there was a miscompare or loss of signal. Notwithstanding all the other creepy crawlies that have emerged from under the rock in the investigation, had they done that none of this whole mess would've happened in the first place. – John K Nov 29 '19 at 16:44
  • @John K well, flipping over rocks are what researchers do, starting with that wing, and those engines. This is solvable, and the company can do it. I would recommend the engineers start by looking at what light R/C Extra 300s with outsized control surfaces can do, and realize a heavy aircraft with a finicky wing is on the other side of the spectrum. If human pilots cannot respond in time to prevent stalling, then then automation must be better. Another creepy crawlie would be controlled airspace routes these planes are required to fly, but a regional could be more robust. – Robert DiGiovanni Nov 29 '19 at 17:52
  • @RobertDiGiovanni That level of redesign would have killed the project. And it wasn't really necessary. If they had gone with dual AOA input sources,a simple wiring architecture/software issue, where a miscomp or loss triggered a fault message, disabled MCAS with maybe an appropriate flight envelope restriction via QRH,none of this would've happened and MCAS would've been merrily doing its thing,and AOA problems would've been a no more than a maintenance/dispatch reliability headache.Of course a dedicated QRH or flight restriction would've probably impacted the 737 type status and round we go. – John K Nov 30 '19 at 17:34
  • MCAS, with further work, may have a future. Would I love to be on that engineering team (its been a while)!! I would start at 5 degrees, where the supercrit L/D peaks, rather than at 12 degrees near stall. The computers can really help, as long as the sensors are reliable and the software is on the same page. Best of luck returning soon for this one. – Robert DiGiovanni Nov 30 '19 at 19:01

1 Answers1

6

This depends very much on what the AOA sensors are used for.

Traditional airplanes

Traditionally, AOA sensors are only used for stall warning and stall barrier (i.e. stick pusher). In this case, you can get away with only two AOA sensors. A sensible design would trigger the pilot/copilot side stall warning with the onside (or opposite side) AOA sensor, which would confer the largest availability on stall warning. This is sensible because traditional stall warning does not effect flight controls and a nuissance stall warning can be inhibited manually by the flight crew.

Stick pusher, on the other hand, should require more prudence. By regulation, the activation of the stick pusher must be beyond the stall warning activation; and the airplane is designed to be stall-free if appropriate actions are taken from the recognition of stall warning activation. Therefore, the use of stick pusher is beyond the operational envelope of the airplane. A sensible design would be to require both AOA sensors to be functional and agree for the stick pusher to activate.

Fly-by-wire airplanes

For FBW or even partial FBW airplanes, AOA information is often directly in the loop of flight controls for gain scheduling and envelope protection (e.g. MCAS). This becomes a completely different problem. By regulation, continued safe flight and landing must be possible without exceptional pilot skills following any single point failure in any flight control related system.

This would require at least three AOA sensors for these airplanes. It's not enough to know that the AOA measurements disagree, but it is also necessary to detect which measurement is erroneous. With three sensors, if a single one fails and disagrees with the other two, you can shut off the failed sensor and continue flying as if nothing happened. Of course, the airplane would not be dispatchable until that failed sensor has been serviced once landed.

Some airplanes (e.g. A220) have four primary AOA sensors. But this is more for dispatch availability (MMEL) than for safety. Note that the A220 has two redundant AOA vanes on-top of the primary AOA sensors (SmartProbes) for Direct Mode operation.

JZYL
  • 11,066
  • 2
  • 15
  • 49
  • thanks! I'm seeing at least 4 locations (possibly 5), 4 sides of forward fuse and tip of vertical stabilizer. Any area experiencing turbulence due to relative wind would be the outlier, and programming could accept the closest 3 and average them. – Robert DiGiovanni Nov 29 '19 at 15:34