BOAC

I'll chuck this (2000) paper in too if it helps.

Fly-By-Wire A Primer for Aviation Accident Investigators

rudderrudderrat

Hi BOAC,

Your paper sums up in one sentence, what I've tried to say.
Once stall "protection", "limit" in Normal Law is lost - PLEASE return my "elevator feel" with Direct Law.

BOAC

It might be a good idea to return this thread 'to earth' with a brief summary of what we know.

Leaving aside what happened after the climb, what about the climb?

1) There is nothing in the BEA report to suggest that the aircraft caused it itself. The only 'query' here is the track record of system reaction to 'events', like the QF pitch anomaly and the 2001 North Atlantic altitude excursion with TC-JDN. It is worth noting that in this the crew did not trigger a climb. (There is some confusion (for me) in the AAIB report where the FDR traces suggest the TCAS RA occurred BEFORE the a/c left its cruise level). Had this crew been in the ITCZ at night and not in clear air in daylight it is worth contemplating what might have occurred.

2) There is nothing in the BEA report to suggest that the pilots caused the climb either.

Until the BEA provide the full CVR transcript AND RHSS trace from the FDR for the period 2:10:05 to 2:10:51 we are all guessing. Why they have not is a mystery to me.

Owain Glyndwr

If you ignore the slight complication of no-lift angles and wing body settings (which change the zero body angle downwash), then lift coefficient is just AoA times lift curve slope, so they move together. Almost all textbooks and reports use the AoA format as this makes the equations a bit easier to handle. I guess I just stuck with it from force of habit

The numbers I quoted are pretty good for an aircraft like the A330, at least up to stalling AoA. When lift curves go nonlinear then I would agree with you that lift coefficient would be a better handle. I don't think anyone knows exactly what the downwash would be at 60 deg AoA, but I was using it in an argument about THS stall and as the later bit of my argument took the THS away from stall I didn't worry too much about it

Thanks for the welcome!

syseng68k

Takata, #204

I'm afraid not. The voluminous fcom looks to me like a user guide,
rather than a serious technical manual and the logic that drives the
transitions between the various laws is not available anywhere, afaics.
Perhaps you could provide pointers ?. Sure, the fcom has bits and pieces
dotted around all over the place, but nowhere can I find a flowchart or
logic diagram that describes the whole process. The fact that, ergo, ab
seem to think that this is not important to crew is one of the most
worrying things about the whole episode. Fly the computers, not the a/c,
seem to be the modern mantra and you don't need to know about the
detail .

Are the personal, ad hominem attacks really necessary ?. If you don't
like a post, then don't read it. It's not as though you had anything
constructive to say in rebuttal anyway. It's not the way that it's done
here, so please, let's keep it civil...

Loerie

Sorry,but I find that takata`s outbursts are unpleasant and confrontational and not really required.Everyone is doing a fantastic job here with lots of team effort to try to discover what caused the death of so many innocent people,and that includes bearfoil.Some are more qualified than others and some have other agenda`s that are not easy to see.Sometimes those from without Looking In have better vision that those inside who may not see the Wood for the Trees......pull together guys and girls.Works much better.

syseng68k

BOAC, #222

Other than the sidestick input, yes ?.

Something that occurred to me is the fact that the sidestick transducer,
which translates physical input to electrical signal, could have failed
in such a way as to provide a bias in it's output. Don't have enough
info to know if it uses a synchro, resolver, potentiometer or optical
encoder, but any but the last of those could in theory degrade to
provide a bias away from zero, (ie: nose up) depending on failure mode.

Probably a long shot, but it is a possibility...

KBPsen

It is completely understandable that frustration surfaces when repeatedly confronted with fantasies, ignorance of facts, invention of facts, rewriting of facts and then attempting to mask it by consistent use of gobbledygook.

Loerie,

What you have here is a bunch of people running in circles and they have done so for quite a while. Some have even resorted to just spin around them selves.

Try this little experiment; Select at random five people who regularly posts on AF447 threads 1-5, select at random five posts from each, compare the five posts. If you can spot any difference let me know.

DozyWannabe

Maybe, but I'd say that they were fairly understandable.

Actually, I'd say that Bear has been deliberately trying to obfuscate and make things unclear in order to push his own agenda - the fact that he elected to do it so prolifically in the last few days does not help his case.

One of whom is Bear. I'm all for getting as many heads around the table as possible, but what he was doing was tantamount to spamming the thread - not just with questions, but with statements that were (based on the information given) demonstrably untrue.

(I hope this won't be construed as being "personal", this is just my observations from the last day or so)

@rudderrudderrat : There has been no such thing as real "elevator feel" since the '60s, and Direct Law does not provide it. The only difference between Normal/Alternate and Direct Law when it comes to tail surfaces is that the trim movement becomes manual-only in the latter, and the elevator command directly affects deflection rather than commanding a rate.

@KBPsen - some of us have been effectively forced to repeat ourselves because of the drip-feeding of misinformation that you are describing - in fact quite a few genuinely knowledgeable people have got fed up and only pop in occasionally, if at all. That's probably a major source of the frustration right there, because if some of us don't fight to keep the discussion to the facts then we could end up with some of the misinformation quoted in a newspaper somewhere, and that's how fallacies like "The plane thought it was trying to land and overrode the pilot" get spread around.

Diversification

BOAC posted an intressant link; Fly-By-Wire A Primer for Aviation Accident Investigators
I have enjoyed reading the short description of how FBW systems operate and about C, C*, and G*U. In principle it looks almost like normal PI controllers and not PID ones. What strikes me is that all is written with the assumption that the controllers are analog and not digital. From my own experience - writing well performing digital controllers and their real-time performance is a very different story.

Regards

A33Zab

2 Xdcrs(contains 4x triple potentiometers per axis).

FCPC1 & FCSC1 driven by separate linkage. (Dual Channel output 4 potentiometers, 2 unused) per axis.
FCPC2 & 3; FCSC2 driven by the other linkage (Dual Channel output 6 potentiometers) per axis.

The sidestick is placed in a polycarbonate container to prevent entrance of foreign matter.
The sidestick design is such that rupture, disconnection or jamming of any of the parts can't cause the loss of total A/F loads on one axis or block free movement of a set of a potentiometer group related to one axis.
Transducer output is monitored and compared, if any difference F/CTL L(R) SIDESTICK FAULT ECAM message is triggered.

Schematic:

airtren

Thanks for your clarifications

Then it would seem that naming of that curve "SW threshold" would make the graph clearer/self explanatory. A "SW" curve (could be non-contiguous) could be the curve representing on the time scale the intervals when the Stall Warning was (should have been) active.

As I looked again at the graph, I thought it would be helpful to see the corresponding curves/data that were available to the "a/c systems" and "a/c pilots" in the context of "loss of data", and showing the discontinuity in the SWT (stall warning threshold) curve, due to NCD - was there such a graph already made available?

The Stall Warning ON->OFF->ON->OFF transitioning during the Stall is one of the several elements that seem to be in the most troubling category. Would this qualify the SW during the event in a different type than "lack of info"? Would it be, the "FALSE info" type?

Is the calculation/equation used for determining the providing of the Stall Warning relying on too few parameters?

"Vertical acceleration", or "vertical speed", perhaps "no lift"? Is any of the existing sensors, and parameters available on the "a/c" able to provide info, so that a Stall Warning could have been clearly brought to the knowledge of the pilots for the entire duration of the Stall?

PJ2

syseng68k;

Re your post #224, good observation. The FCOM is a need-to-know user's manual and indeed does not explain the aircraft. I think this is one of the underlying issues. FBW and C-star law are not in and of themselves problematic but unlike complex hydraulic or electrical systems, the differences in "the how" of control are materially different and require understanding, as do the more bread-and-butter areas of high altitude, high Mach number flight and even some jet transport aerodynamics. I did not know and did not understand until a lengthy and patient exchange with HN39 that the stall AoA for my aircraft was much lower at high Mach numbers in cruise flight than what Davies had expressed in his wonderful book, which was the approach and landing case with high-lift devices extended. Big, big difference and, even in retirement, I learned some fundamentals.

These are not taught very thoroughly in initial ground schools at least in my experience, and if one wasn't in the air force one's knowledge was increased largely through one's own efforts.

The frustrations of not knowing and not being able to find things out easily have been endemic - while easy to fly and a joy to hand-fly, "automation" has become as much a marketing tool as it has a way of solving the problems of flight. The resistance to knowing more than the "NTK". need-to-know, ground-school curriculum requires, comes first from how expensive it is to train well, and next from a lack of knowledge in those who must do the teaching, always of course, with wonderful exceptions from those memorable instructors who's passion takes them, and their students beyond NTK.

Learning is expensive and NTK and automation are assumed to "solve" that "problem" for a cash-strapped, (de-regulated) industry.

This isn't "THE" problem, because clearly the aircraft and the design's record is no worse in terms of fatal accident rates than conventional types - in short, the airplane and the design work extremely well but one should never be in a position to not understand and not anticipate what his airplane is doing.

Below is a nuts-and-bolts schematic of the pitch-basic loop. I hope it is of some service in understanding the pitch control of the A330.

Meikleour

Takata: When the g load is less than 0.5g then the THS is frozen. Perhaps this is not known to Bearfoil?

bearfoil

Meikleour

It was not known to me, and I appreciate rudderrat's rephrase. I'll be brief with my next question. Of course it may (is) in the FCOM, but may I see any answer in the thread?

If frozen, the THS would be unavailable for Trimming, (or Pitch changes), until what point? When does it become available again?

So the original question remains: If less than one g, does the THS move independently of Pilot's inputs? Does it try to regain 1g or satisfy itself that anything greater than -1g is acceptable?

Does it "Unfreeze" if the g becomes <-1g? And then input NU?

The limits are intriguing to me, until I have a better picture of what the THS may have been doing in <1g, who can say what the a/c's behaviour was?

To expand on BOAC's #2, BEA have not assigned the climb to the pilots, that is true, and they affirm that by specifically stating, "ONE" (a) Noseup, left roll input. That apparently happened prior to the "climb", as I read it.

takata

Hi BOAC,
There is another way to change pitch attitude than pulling up on the sidestick as simply applying thrust could do the job if the ammount is large enough.
While I agree that QF event remains a mystery, TC-JDN case is showing that, right after autothrust disconnection, manual thrust was applied and increased from N1 below 70% to 100%. Sidestick (ND?!) imputs were not applied until the top of the climb 30 seconds later, hence nothing contraried TC-JDN to change its level flight. Someone also posted a revised chart of DFDR control imputs but, as far as I remember, it did not include any thrust track.

There is some more confusion (for me) in the AAIB report and in the cockpit. Had this crew been in the ITCZ (or anywhere else at night) it would be certainly dangerous, but they still were in NORMAL LAW and the flight envelope, contrary to AF447, was still fully protected as airspeed was valid all the time.

Well, I should have read another "report" as in mine, a nose up order was (almost immdediately) given by the pilot... the later nose down orders that were applied only reduced the climb rate from 7,000 ft to 700 ft/mn (which is not enough to start a descent). So here, we effectively need to see the tracks in order to understand what the PF really wanted to do. It is not like this initial climb looked like uncommanded, but the report doesn't tell much about how it was controled and what could have prevented this high climb rate control by the PF.

HazelNuts39

To start with just a reminder of BOAC's justified concern that these curves should not be mistaken for real DFDR data. What you are asking is problematic because the curves assume still air, and one should be aware that turbulence would superimpose "grass"-like fluctuations on most parameters. IMO that is the reason that the two brief activations of S/W after 2:10:05 are not shown as threshold exceedances on mm43's graph. May I suggest that you just make a note that the AoA was 'invalid' between 2:11:40 plus 'a few seconds' and 2:12:17?

The systems consider the signals coming from the AoA vanes invalid, there is no discontinuity in the SWT.

You pose more valid questions, but I regret not being in a position to answer them.

JD-EE

HeavyMetallist, "Why some people persist in believing that stalled == totally ineffective is a mystery."

gums has posted some graphs that show an actual inversion of control for a totally different kind of aircraft when in a bad enough AoA configuration. Somebody else posted a graph I believed was for the A330 produced by calculations that indicated an AoA of 61 was sufficient to put the A330 into a realm where the tail was not particularly effective.

That said, it appears from the release, in a part I'd forgotten (take your own advice, Joanne), that there still was some authority from the tail surface. Unless the PF pulled up again after the stall warning the tail surface did not have enough authority to pull them out in the remaining 2 minutes. So it was basically ineffective or marginally effective.

Diversification

A33Zab
Thanks for a nice and illustrative picture of the stick mechanism.
I wonder if a very strong force on the stick to the left could force some undesired nose-up if the enclosing structure is not sturdy enough.
I am also a little surprised that potentiometers are used - even with two (or more) in parallell for each axis.

Regards

JD-EE

bearfoil, you'd best concentrate on something you know somewhat better than law. I can think of several instances off the top of my head wherein people who are not parties to a contract or any other record do not have access to the data.

One is when a settlement is reached and both parties insist it be kept secret as a part of the settlement.