Originally Posted by
edmundronald
Gums,
Math jargon Sri’s , “may exhibit non-linear feedback” is the engineer’s equivalent of saying may become EXTREMELY unpredictable.
- one wants a system to be controllable - ie can be brought/forced to a desired eg. nice stable state starting from any state inside the envelope.
- if the system is “linear” then a branch of engineering maths called LINEAR control theory applies, the domain where it is controllable is known, and eveything is “copacetic”, automation can be easily used, and the system may be approved on the virtues of its design.
- if the system is “non-linear” and even worse if it exhibits non-linear feedback, then can as they say in the US it can “go postal”, and standard school maths can’t describe it nor control it. An auditor won’t certify such a system unless it is extensively tested -if the auditor be honest. And automation can be very unhappy about the surprises baked into such a unpredictable system.
The suspicion is that MCAS is there to keep the MAX centered in an envelope where it appears linear, but that without MCAS it might get into states where it would be really hard to control (except for Chuck Yeager and you). The unfortunate joke is that the extant MCAS took several airframes into states where they were hard to control. Of course one could argue one can always deviate into a non-linear behaviour eg. stall, but the suspicion here is that the volume of linearity is more limited than it should be, and that really ugly things happen in places where certified airframes should still be acting nice.
I’m sorry if the above is erroneous or unclear, my opinions are usually worth exactly what people are paying for them.
Merry Xmas
I consider this a bit too simplified approach. Basically every real world control loop is more or less non-linear. Yet, typically the control design is based on the linear analysis. However, the design rules includes stability margin. This stability margin is related to the Lyapunov stability analysis.
On the other hand, even the linear system might have uncontrollable and unobservable states. In a closed loop system, there tends to be observability problems that restricts the the usability of the closed loop data.
If we consider MAX case, there are two independent feedback controllers: the pilot and the MACS. So both of them have to operate with restricted information. I see this as the root cause of the problems. When the system starts to work in an unpredictable way the failure analysis might become unreliable.
I don’t understand why the MAX is still grounded. The control problems should be relatively easy to solve (with additional actuators) when the understanding of the underlying problem is present.
My guess is that MACS was used to fix something else that stick force problem (in the control engineering referenced as open loop gain). My guess is that the system was there to prevent to enter to a envelope where the the real problems begins. For the political reasons it was named and defined to system that makes the “feeling” like NG.
My guess is that the problem were solved if the politics, the "salesman's lies” and the lawyer advices were were forgotten and the raw engineering job from clean table were started. Some kind of spoiler in the wing root or somewhere near the pylon might be needed. There should be courage to think out of the box and forget the short term costs.