DozyWannabe

Lyman, my red line coincides with timestamp 02:10:11 - the A/P has been out for 6 seconds, and the PF sidestick trace has been registering significant nose-up inputs from the PF's sidestick since 02:10:05 (i.e. those same 6 seconds).

mm43

grity;

Back in AF447 - Thread No.5 Post #229 A33Zab posted a graphic of the SideStick and a brief description.

Here's the same graphic again.

Lyman

I think this thread is behaving worse than 447.


@2:10:08 "I have the controls". The climb started well before the PF had the stick. And VS SELECT was oscillating independently. Notwithstanding the aircraft captured 5000fpm instantaneously?

Pitch and VS (actual) track together? The graph is eating Wheaties, or was put together by Michael Mann.

grity

how works the mechanik of the rod-spring at the kink of 6.5 mm ??? (1/3 pull)
that is +/- the pitch aerea the PF hold the stick during the climb....


also from A33Zab:

mm43

Lyman;

I've expanded some of the FDR traces and in the case of the Pitch attitude one have added the Elevator and Stabilizer positions (inverted) for easy comparison. Immediately above is the Normal acceleration trace.

My summation is that the aircraft responded very well to the PF's inputs (as represented by the Elevator and Stabilizer positions) after the A/P hand-off, but just prior to, the turbulence had created a pitching moment that didn't correlate with the vertical speed. Hence the A/P disconnect.

The UAS was a secondary.

EDIT: Patently not true, as the RTLU was locked at 02:10:04.5 in response to the ADRs rejecting the two speeds recorded, or more precisely the Mach value of one of them.

Quote:

The two airspeeds recorded were still valid at this time; however, a false value point is present in the recording of the Mach. The low sampling frequency makes it impossible to determine the duration of the disturbance in the measured values; however, it is likely that it corresponds to when monitoring was triggered. (BEA Interim Report No.3)

The UAS was the primary reason for A/P disconnect.

EDIT #2: An interactive version of this graphic with vertical/horizontal Cross Hairs and additional time marks and scale is available here. You can also expand your browser page using Ctrl + and return to normal with Ctrl 0.

Lyman

Yes, I know. Thanks.

Machinbird

Grity, the way this is commonly done is two springs, one for the initial slope and a second that has a certain amount of freeplay to generate the steeper slope.
One of the guys with maintenance experience will have to tell us the specifics of the A330 feel system, but if you will note, there are two spring cartridges in the pitch channel. This may be a redundant design with two identical sets of spring cartridges, or a non-redundant design with one spring cartridge doing the first slope, and the second cartridge (with built in freeplay) doing the second slope. Each cartridge would contain at least two springs (nose up and nose down).

DozyWannabe

Are you disputing the BEA's findings?

According to the BEA, the turbulence was responsible for the roll, but the A/P disconnect was initiated by UAS - this makes sense as moderate turbulence should not be enough to cause A/P disconnect.

@Lyman - The graph comes from BEA Interim#3 page 111 - I just did a screengrab and moved the relevant traces next to each other (vertical movement only - no horizontal) in the same way I and airtren have done before so that the graphic does not take up more vertical space than it needs to. This is basic pixel-pushing of the kind I've been doing since I was 11 years old. You can print that page and draw the line with a ruler and it will tell you the same thing. I've done it again here with the sidestick trace included just so you'll know I'm not fiddling you:



The pitch up and climb was initiated by the PF's sidestick input - whether he manipulated the controls before, simultaneously or after he said "I have control", that stick was halfway back for 2 or 3 seconds before the aircraft started climbing.

Lyman

Doze. Check your RED VERTICAL LINE. You will notice it is at 2:10:10.

The initial PULL UP by the PF (F/O) is at 2:10:07. The START of the RED STICK TRACE. In the seconds before a/p disconnect, the Nose dropped 4 degrees, and the a/c was climbing at 1000fpm. It is this AoA increase that dropped her speeds (all three) beyond the 30 knot/second threshold, and caused the speeds loss: but after the disconnection. BEA re: ICE is "LIKELY" wrong.

Of Importance is what the a/c has done prior to 2:10:05. BEA wrote at first that the cg was at 37. They changed that later, in a follow on, to 28. If I had bet my life on their written data, I'd be dead.

DozyWannabe

I had it at 2:10:11, but whatever...

You can't keep bashing away at this, the data is there. The A/P commanded nothing untoward, the pitch-down and roll was induced by turbulence and the climb was as a result of sidestick input.

The *very* start of the PF's sidestick trace is unfortunately obscured by the data points of the PNF's sidestick trace. The "zipper" could be many things - I suspect it's wonky data (it looks too uniform to me to be anything else), but, for the sake of argument it could alternatively be pitch corrections made by the autopilot in MANAGED mode to maintain assigned altitude in turbulence. In any case, the vertical speed and altitude traces show that there is no major deviation from assigned altitude during the time it is happening, and even *if* the autoflight was commanding one of these corrections when it tripped out (which doesn't tie in with the trace - at 2:10:05, it's commanding 0 ft/sec), then a small pitch correction is all that is needed to bring the nose back up. But this is not what the PF does - he pulls back and holds it there for several seconds and commands a climb.

The CG readings given in the interim report #3 (which was the first issued with actual data) are 27.5, 28.7 and 29.1% MAC respectively. Where do you get the other number from? If it's from an earlier report it would have been an estimate because they had no data - hardly their fault.

Lyman

2:10:11 or "WHATEVER"? Are you kidding? Your line is off by six seconds, and your time index is completely WRONG. Check mm43's graph. Do you dispute that the a/c had dropped 4 degrees in PITCH as she climbed (ACTUAL) at 1000fpm? Look a little closer. Please remember that to climb with nose low requires more vertical than the actual rate of the airframe. This is what duffed the three speeds simultaneously, and by the same value, imho. A computer computation of the discrepancy would result in the descent rate due PITCH to be overlimit. This bounces the A/P. The speeds follow.

DozyWannabe

mm43's graphs are from a different page than mine, and have had more editing done to them, but for whatever reason, unless i'm reading the graph completely wrong, the pitch attitude only dropped below 0 degrees at 02:10:03 and was on it's way back up at 02:10:05 (AP disconnect - though mm43 has it at approx 02:10:04.7 - perhaps he's privy to better information than me). This has all the hallmarks of a minor turbulence encounter, but the climb proper is started between 02:10:10 and 02:10:11 *as a result of the pitch up ordered by the PF's sidestick*. On this the graphs do not differ.

mm43

Dozey;

1.. The RTLU was locked at 02:10:04.5, and
2.. The aural autopilot disconnection warning (cavalry charge) was heard at 2:10:04.6.

We are "splitting hairs", but they're all that's left to split!

DozyWannabe

@mm43

That's fair enough, but I don't see how that contradicts what the BEA are saying about UAS being the reason for autopilot disconnect as opposed to turbulence. Even if that is disregarded completely it doesn't support the hypothesis that an autopilot that disconnected between 02:10:04.5 and 02:10:05 ordered a climb that did not begin until 02:10:11 (not that said hypothesis is one of yours).

As I said, it looks to me like the autopilot disconnected in the middle of one of it's turbulence encounter compensation inputs when the nose was on the way back up (i.e. not on the way down as Lyman hypothesises), but the real beginnings of the zoom climb are clearly ordered by the PF's sidestick input.

On autopilot disconnect, the aircraft locks thrust at the last setting (as you say) and holds the pitch where it was being commanded at point of disconnect (in the middle of the correction) allowing time for elevators to return to neutral, which is borne out by the graph settling at 2 degrees nose up for a couple of seconds (just short of the 3.5 degree pitch it was holding before) before the nose-up commands from the sidestick begin to take over and progressively increase pitch attitude to 11.5 degrees nose up.

Machinbird

Just my personal opinion, but if we really want to get to the core of the nose up control input, I think we have to look at PF's control problem.
Something along the lines of what he did versus what he should have done. If Grity's chart of control motions is accurate (and not overly interpolated/smoothed) then PF used an inappropriate control strategy.

Having never flown an Airbus, I'd like to hear from those that have, regarding control strategies used in Normal Law, and Direct Law.

For example, from the videos recently posted, it appears that the majority of Normal Law control inputs are pulses away from neutral, but sometimes in two axes simultaneously.

Is direct law also done with pulses, or is it a continuous input? (I have to assume that Alt2 Law experience with mixed control modes is very unusual/scarce.)

As I mentioned in an earlier posting, Grity's chart shows a lot of diagonal control inputs along the NE-SW axis. I really think PF was having difficulty separating his lateral inputs from his pitch inputs, probably due to an improper stick grasp.

I'm posting grity's image again below for convenient reference.

rudderrudderrat

Hi Machinbird,
In ALT Law, it has to be flown like any conventional aircraft in roll. The control input has to be held against any roll tendency until the load is trimmed out. Since there is no aileron trim, you have to use rudder in order to be stick free. It would appear that PF was attempting to fly the aircraft using large pulses when he was in roll direct.

Most of my ALT LAW practice in the sim was at low level.

TTex600

Machinbird, after taking a second look at grity's graph, and assuming that the data contained therein is correct; I make this conclusion - Something is lying to us.

The graph shows an initial move out of center, followed by three complete left/right reversals including two up/down movements in 11 seconds. That amount of movement in that time frame is IMO almost impossible based upon my time spent handling the A320 SS. In other words, the Bus SS is not a cheap video gamer's joystick. The spring loading and damping would make that amount of movement quite difficult.

Secondly, that amount of movement would be on a scale never before ordered by the PF prior in his career. If I have followed the timeline correctly, the aircraft was relatively stable when the a/p and a/t dropped off line. This graph indicates that the PF began this series of SS movements within two seconds of a/p a/t disconnect. The aircraft does not appear to have been in an upset at that time, so what caused the PF to almost immediately begin a series of spastic SS movements? These movements are not compatible with normal hand flown inputs.

We are left with resolving a series of over exaggerated control inputs with a relatively stable aircraft. Either the data is corrupt, or the PF made completely inappropriate control inputs, almost impossible control inputs. ...........OR, the PF was just hanging on. Or, the aircraft was upset more than I realized from reading the data.

Regarding control strategies, I find that small deflections, held for longer time spans work better than short higher magnitude deflections.

FWIW, the 320 requires very minor control inputs when flown a/p, a/t, f/d's off.

gums

Salute!

I believe the stick inputs being awry due to spring failure or degradation is a red herring. The pilot commanded and the system responded best it could with screwed up speed and disregard for AoA limits.

The chart of the breakout forces was a surprise to me and I had missed that if posted before. The breakout force is lots lower than the Viper. Ours was 1 lb for roll and 1.75 for pitch. Secondly, the gradient for max command is lots steeper. For example, ours was about 7 pounds to command 2 gees. Then the curve got steep and a 9 gee command was about 31 or 32 pounds.

Secondly, my understanding from looking at the roll laws is the roll command works the same in "alternate" as in "normal". The only difference I see is the max roll angle allowed in "normal", not the command, but actual body-referenced roll angle. So what I see is what I was used to - let go of the stick and the roll command is zero deg/sec. Am I missing something?

As the Bus drivers are now finding out, there are "hidden" aspects of the control laws. Same as we had in the Viper. For example, no stick pressure in roll and zero roll rate command. Not so fast! Turns out that there are flight dynamics which are not completely corrected for. Look at the pic on my profile bio. With that LEF up I had to apply almost full left stick to stay wings level ( it was a control surface limit, and the engineers told me that I only had about a pound or two of authority or would have had to bail). The zero roll rate "law" didn't work. We also saw this when dropping a heavy store from one wing. We had to apply aileron pressure to stay wings level on the pullout and we had thot the jet would maintain zero roll rate. Same deal with allowable control surface deflection.

Only thing that puzzles me is the pilot having to apply roll in the first place. Could be a spoiler or aileron wasn't moving correctly, and a determined manual input was required to override the deflection limits. Same as I had to do with the LEF failure. And remember, control surface movement rates and such use dynamic pressure for "gains". What does the system do when dynamic pressure is deemed unreliable by the system?

later,

Organfreak

@TTex600:
I'm going to repeat my possibly stupid questions that I asked the other day:
1. What about a broken longitudinal transducer?
2. Were all four transducers found?
3. Have they been verified as working properly at the time of the accident?

If there were to be such an (outrageous) finding, that one was broken, this might explain the unexplainable.

Lyman

Simply put, The Pitch was below cruise for nine seconds from 2:10:00.

It reached its lowest value (4.5 degrees ND) just as the a/p tripped out.

As it increased, the aircraft was still in descent.

At 2:10:07 PF input back stick, after the a/c had started to climb.

The airframe was out of phase v/v ascent/Pitch for nine seconds.

The PF inherited an airframe in phugoid oscillation.

How many times had PF been handed an a/c out of autopilot? As many as he had landings logged, most likely. His stick work is a copy of the two vids grity has posted, perfectly acceptable, when approaching to land.

His elevators were following the airframe at first, from his display.

The autopilot could not keep up with the turbulence, and tripped out. Out of this trip came airflow that sensed airmass inconsistent with the airspeed prior to its disturbance, and so the airspeeds were lost.

In two seconds, the handling pilot was given an impossible task. True to form, just as the Airbus got into hot water, it changed the "rules", and took a powder. So now, the Pilot has a very long list of things to learn, or die. Not the least of which is to acclimate to a Stick that has morphed into a goblin in ROLL, and a PUSSYCAT in PITCH. Has he trained for this?

I think NO. As someone has said here, put a single engine pilot in a twin, and pull an engine at liftoff. What are his chances? Not good.

At 2:10:00, the a/c started a phugoid. Was it speeds? No, I think airmass related. Had it been speeds, I think the engines would have been throttled, and they remained at cruise level. And if not speeds, then ICE is not a player.

For good or ill, in my opinion, this flight was out of control when the Pilot's input doubled PITCH, at 2:10:11. One can insist it is complex, and it may be, but had PF had a current Attitude display, and listened to PNF......