HazelNuts39

In your graph, notice stick placement when "Pulled". The UP inputs are spikes, between them are lapses of linger in Nose Down.

It appears that his UP inputs are a severe reaction to something about the a/c behaviour. Quickly, after pulling up, he immediately reverses and pitches down; the stick does not remain in NU? What might be the cause of this actions? What is he getting from ND that he does not "like"?

Or Lonewolf?

Lyman, this is a guess response based upon a few things.

1. Stated sensitivity of this model acft at high altitudes by people who have flown it, as compared to control sensitivity in the terminal area and altitudes.
2. Change in power setting
3. Pilots initial overcontrol of roll previous to the graph

The above considered, what HN's graph excerpt shows me is that as with roll, the PF doesn't have a good "feel" for his pitch control during seconds 50-85, so there's a bit of over controlling going on. Part of what he's doing looks to me like a response to a pitch change that you'd expect from an addition of power to TOGA, and in part to correct for his own various pitch inputs, corrections, and counter corrections that he apparently isn't happy with. Why unhappy? I think he's aming for a particular pitch attitude and not hitting nor holding it. (Could be wrong).

I am uncertain what pitch attitude he is trying to establish and hold.

From 90-100 seconds, he makes what appears to be a substantial counter correction in the nose down direction with very small counter corrections ... he's milking the stick a bit, trying to avoid overcorrection during that phase, and seems to be making progress in establishing his "feel".

Not sure where the CVR sync's up, but I suspect this particular correction is a response to PNF's warning about 'you are climbing' and 'go down'

Not sure what other forces you are referring to. Seconds 120 to the end, he appears to be deliberately walking the nose down. The graph ends in what appears to me to be the next set of corrections / counter corrections about to start ... though why the SS at - angle induces nose up at the tail end there confuses me ... seeing the whole timeline/graph would probably resolve that confusion.

Around the time Captain re-enters, PNF takes control and inputs full left bank. So it seems that both pilots are acting with "Full Stick" ?

HazelNuts, I think I said PF was unaware of his climb, not the Stall Warn.

Lonewolf, not wanting to belabor it, but the graph above null is more reactionary than that below,,,,, it seems to ape the ROLL oscillations, but in PITCH.

The "UP" is spikes, the ND, more flat. So, he's lingering in ND, and spiking NU?

??

The zoom climb is not 'sustained'. In response to the PNF's "go down", the V/S is reduced from 7000 fpm at t=25s to 1100 fpm at t=49.

As I read his SS movement, he is trying to maintain 15° pitch but is overcontrolling, like he was earlier in roll. Each time the pitch increases through 15° his SS moves forward, and backwards again when pitch reduces through 15°. From 90-100 seconds, he says "I've lost control ..." and goes all the way to full nose-up to counter the nose-down pitch rate.

Lyman

I think you've got it about face again - +ve sidestick is ND - he's lingering in NU.
Check the DFDR traces for SS sign convention

Correct.

For what it's worth, I think he's chasing something on the ADI rather than the FD (although we'd need confirmation of what the FD would output in that circumstance to confirm or refute).

Dozy, Owain. Thanks for the correction. And thanks to HazelNuts39 for his establishment of a rationale in re: PF manual control of the a/c. Something is preventing his establishment of 15 degree Pitch NU, such that he must emphatically get the Nose down? Either way, the spikes to me are a sign of problems in Pitch, just as there were in Roll.

Can we entertain a chronic bias NU, similar to that seen in Right Roll? From the bank traces, just as here with Pitch, there appears (to me), to be just such a bias. Can we conclusively say that the a/c was unbiased in Pitch? Not to include the THS, at least at this point?

Thanks

Lyman,
I think we can conclusively say that the a/c was unbiased in Pitch and in Roll.

Hi Lyman,
I'd say the aircraft naturally wanted to lower it's nose as the speed washed off.
It was only due to the PF's mostly nose up ss demands, aided by full nose up stab trim, that he managed to prevent it from lowering.

The point is precisely to emit a logical hypothesis to your earlier question :
"If PF didn't notice stall warning, what caused him to select CLB, then TOGA, go to 15 degrees pitch and actively maintain that until the airplane dropped out of his hand after 40 seconds?"
If the V/S was at 1500 ft/min at the time the FDs reappeared, then the SEL V/S target reopened at that 1500 ft/min. But as the V/S soon was in reduction, the vertical command bar went up, and the PF was proceeding to follow the command and the trim was 'helping' him to do so.

To note that under stall warning the FDs were telling the pilots to pull ...

I'd say that he held it up there deliberately, but that's only nitpicking :D

I don't want to spoil the fun, but Clandestino was right when he cautioned against applying 'normal' rules to the stalled state (post #221). I think maybe you can apply some generalisations,i.e behaviour averaged over many seconds, but when you get down to specific time slots there are just too many unknowns.

Just for example; we know that stalling AoA depends on sideslip, so that pitch behaviour near the stall will also vary with slip. We know that the sideslip at high AoA varies kinematically with bank angle, as does AoA itself, so AoA, pitch and sideslip are linked to roll, and we know that the natural lateral behaviour at high AoA is a virtually undamped roll oscillation so that one can expect a sympathetic oscillation in the longitudinal axis. How the dickens can one unscramble that lot without access to some very good data and a powerful computer or simulator? And that is without considering any known possible effects from forebody vortex shedding or wingtip vortex movement over the wing or vortex bursting.

IMHO It just ain't possible - you are wasting your timehttp://images.ibsrv.net/ibsrv/res/sr...y_dog_eyes.gif

Thank you both. It may be obvious, I am trying to look at all the inputs, to find something, a clue, as to why: Stall. One needn't remind of the way it looks as a whole. His spike ND looks almost desperate, even angry. So what I seek is an opinion as to why these emphatic ND's? It is in this region of the accident that CRM seems conspicuously absent. If Someone could explain, for instance, why rather than the jerky motions, PF didn't settle on a nudge/ wait format. The aircraft in Pitch cannot move as his stick would, and since the trace of actual Pitch is far smoother than the inputs that command it, why the constant oscillation in stick/hand?

At the least, is it fair to think that the controls and procedures are more reflective of down low, than up high? From the stick trace, the ND inputs seem solitary, again, as if he is angry perhaps at his own over-control? An unfamiliar response from the a.c?

HazelNuts39
OK, I am working on just this bit of the graph.
From time = 0 to time = 25, zoom climb took x seconds.
As I understand it, that was due to more than a single input to the SS.
Am I wrong about that?
It is my understanding that he sustained the climb, but am happy to be corrected.

We enter your graph excerpt at time = 50, after the PNF has alerted him to climb. He's still climbing (sustains his climb), altitude increasing (until stall, and then until momentum peters out).
For this excerpt of the graph, "zoom" climb probably wasn't an apt description. Thanks for that. :)

By your graph, at time = 105 to about time = 115, he lowers, and keeps lowered, the nose ... or is that the aircraft doing it for him? (See rudderrat's comment). He's moving the SS to get the nose down, or so it seems from the lines. (Even though he says he's lost control, which apparently he has not, at least not in pitch)

As displayed on the graph, his pitch corrections follow the SS graph, out of phase or delayed by a few seconds as the plane takes X amount of time to respond. This ain't an F-16, eh? (Where's gums? :))
.
OK, I see where you get that, thanks. Strengthens the idea that he's executing the low altitude shaker stall avoidance procedure.
And then starts walking the SS down, 90-100 seconds, orange line.
And the nose follows.

rudderrat:
Thanks! That paints a slightly different picture for me, and makes aerodynamic sense.

Wasn't that the time he throttled back to idle? (not shown on HN39's graph)

Owain G,
Yes, that is correct. Reading from the graph on page 113, the thrust levers are moved to MCT at t=103, then to idle at about t=107, then at t=128, when the engines have spooled down to near-idle rpm, thrust levers back to CLB, and at t=153 to TOGA.

HN39, thanks. As usual, it isn't just stick/yoke that influences attitude.

We return you to your regularly scheduled program, which today is a live performance of The Pitch and Power Chorus singing their cover of REM's Feeling Gravity's Pull.


...
What the hell is it doing now, ami?
Pitch and power and out of synch here
Pull up, pull up, pull up the sky is open-armed
When the nose stays up, I feel gravity's pull
...

With apologies to Buck, Stipe, Mills, et al ...

@HN39; CONF iture:
 

@2 h 10 min 47: Selected V/S equals V/S ~ 1500ft/min as CONF correctly stated.

The first 10s the FD bars will be flashing, just before STALL it would have been steady.
During the STALL (negative V/S) the FD order would have been SS UP (in reference to +1500ft/min), Max Guidance is hard limited at 6000ft/min.

However, soft limits are build in and that will make it difficult to determine the exact FD Pitch Bar position.

- FD bar order will never guide into Vmo/Mmo or below VLS (VLS-5 if VLS = target)
was VLS available? and if so was airspeed already below VLS? and if so will it guide to reach VLS?

FCOM section 1.21.00 states:

Quote:

---

Note :

When flying with FD bars only (AP OFF), the FMGS adjusts the pitch bar so that VLS is maintained.

However, no triple click is generated and the V/S target display on the FMA remains unchanged.

---

No pilot would use FD command bars in a situation like this unless they were right out of FO school. Experienced pilots do what they have to and ignore FD commands doing what they are trained to do. Everybody knows that.

was VLS available? and if so was airspeed already below VLS? and if so will it guide to reach VLS?. From...above, @ A33Zab

Isn't that dependent on ADR's? HazelNuts?

Weren't FD selected to acquire FPV?

b44:

Any chance of you coming out of retirement?

I'm a big fan of people who know what they're talking about...as a result, from here on in, without you, I'm taking Amtrak, Greyhound or some Italian cruise line.

OK465, Thanks, because I thought I would get beat up on my post. We old timers used a FD as a guide if it was going where we wanted to go. If not just disregard and go the way you want to go. Worked fine for years. Pretend you are a 727 and fly as you want, it works great.

I even did it with a check airman out of SJO in a 757 on my initial international captain check. He had it so screwed up after our new clearance I was in a 30 degree right climbing turn and he was in a descending dive on the FD. I ignored it and eventually he caught up. I did it right, he was trying to make automation do it.

Interesting.

For the FDs to reappear, if I am correct, they need 2 airspeed indications in agreement. As both recorded airspeed indications were different, probably the unrecorded one was in agreement with one of the recorded. At that time, both recorded airspeeds were below VLS, one slightly above 200, the other one around 100, and the VLS was around 240 if I get it right from the QRH.

Now, the automatic speed protection you mention applies if VLS is approached from above, but would it apply if the airspeed is already and suddenly well below VLS ?

Also, such automatic speed protection applies for a climb scenario. Would the FDs command a temporary descent in order to get back to VLS ?

@CONF:
 

Correct, FD needs 2 airspeeds. (FE for speedtargets calculations only 1)

@ 02:10:42 FD was transitonary available; IMO ADR2 was increasing and crossing the ADR 3 airspeed (~transitonary).
@ 02:10:47 it was consistent with ADR1 and FD returned.

I interpolated Vs1G (205t, FL350) ~ 185kts, VLS = 1.23Vs1g ~ 225kts.
anyway both airspeeds below VLS.

BEA didn't have an answer in IR#3 either:

Quote:

---

4.3 Recommendations relating to Flight Recorders

---

Quote:

---

Analysis of the FDR parameters and audition of the CVR provide
information that is essential to an understanding of the event.
However, it is difficult to reconstruct the indications that were available to
the crew on their instrument panel, especially the instructions given by
the Flight Director crossbars when they reappear.

---

They will know about now!

The latest FCPC mod 2K?* / software may be a result of this.

Ref. AD 2011-0199R1 and AD2010-0271

Agreed.

Sorry for nitpicking, but just to ensure that we work from the same basis:
At 02:10:49 ISIS altitude= 37500 ft, ADR1 airspeed=216 kCAS, M= 0.68. Then Vs1g=200 kCAS and Vls=1.23 Vs1g= 246 kCAS.

@HN39:
 

Not considered as nitpicking, thx for rectifying my wrong interpolation.

246kts accepted as VLS....

Sorry PJ2 - it's just that having been told I'm misinformed and know nothing of the protections, I felt we'd better make things as exact as possible!

Flight Directors
 

CONF: This is as you say definitely interesting.

The specific failure mode that put you into alternate law will determine whether VLS is computed and displayed or not.

From simulations (using manual deselection & reselection of the FD's), if the failure mode does not preclude computation and display of VLS on the PFD with an operative FD then the following appears to apply:

1. If above VLS and decelerating when the FD's are reselected, the FMA vertical mode will default to VS and display the current VS and the FD pitch bar will command that specific VS (i.e. +1500) until such time that VLS is reached and then command nose down (decrease in pitch) to honor VLS. The FMA annuciation will remain VS with the displayed command (i.e. +1500) that was in effect when the FD was reselected even though the pitch bar is honoring VLS.

2. If below VLS when the FD's are reselected, and VLS computations available, the pitch bar will immediately command nose down (decrease in pitch) to honor VLS, but the FMA will still default to VS and display the specific rate in effect at the time the FD's were reselected (i.e. +400, etc.)

***************************************

If the failure mode responsible for alternate law precludes computation and display of VLS (i.e. VSW only zipper with flap position sensor problems and VLS dashes on the PERF page), then the following appears to apply:

1. Regardless of whether above or below the actual but uncomputed and undisplayed VLS, when the FD is reselected, the system will default to VS and the current rate (i.e. +1500) and the pitch bar will command that rate right into the stall, if followed.

*******************************************

I would guess that recovery from actual dual or triple ADR problems that result in 'latched' ALT 2 would result in no VLS computation or display but I'm not positive. Maybe you or A33Zab would know.

Simulations that allow for dual or triple ADR failure are directed at the failure remaining in effect until landing so that if the ADR's are 'revived' the aircraft returns to normal, not latched ALT 2.

Possibly a new simulation requirement? I'm always in the market for good tech info. :)

"Also, such automatic speed protection applies for a climb scenario. Would the FDs command a temporary descent in order to get back to VLS ?"....

tantalizing.....

I see "Zipper" (5000fpm descent), " FD commanded descent" (PF's pull to defeat), and FD commanded ascent......

What does this mean to a layman?

Lyman/Layman:

With A/P disengaged, FD's do not cause the aircraft to 'do' anything, don't require 'defeating', and should be deselected when unreliable.

Different 'zipper'. :)

More to the point, after a UAS/autopilot disconnect incident the priority should be to just keep the thing stable. In cruise the FD, being a visual guide as to how autoflight would follow the FMS programming, is only really useful for navigation and to a lesser degree maintaining flight level. Like autopilot, the FD cannot aid a recovery.

In short, Aviate, Navigate, Communicate - in that order. The FD only really applies to the second of these.

CONF iture:

Same graph, with vertical speed and FD availability added.

http://i.imgur.com/zJssX.gif?1

The failure mode that caused reversion to Alternate2 law is detailed in IR#1 pp. 55-56 and IR#3 p. 40.

FCOM 1.27.30 associates failure of Vs1g computation with "loss of weight, or slat/flap position".

Can we conclude that Vs1g computation was available, and therefore VLS was computed and displayed?

OK465
Lyman/Layman:

With A/P disengaged, FD's do not cause the aircraft to 'do' anything, don't require 'defeating', and should be deselected when unreliable.

Different 'zipper'.

Thank you . I recall the Flight Directors were not turned off after UAS, as they should have been according to UAS drill, AF. (Did this drill predate 447?). So any way, if the flight directors indicated "Low", would the pilots command a climb? Were they working/accurate at the loss of auto?

Do they show the 'BIRD', indicating FlightPath?

If the STALL speed was displayed, does that mean ADR'S are back on line?

HN 39 & Lyman:

Once again an interesting question that goes to the crux of whether the FD might have coerced a momentary nose up input out of him a couple of knots above stall speed that put him 'over the edge' as the pitch attitude was nearly under control.

I don't know, but of course would like to. There's not a lot of discussion of intermittent availability of FD's as a result of intermittent ADR data available to the FMGEC in the documentation I have, nor the 'latched' ALT 2 display on the airspeed tape following recovery of an ADR.

On the surface, after recovering an ADR, it would appear that the capability to compute VLS is available, but would it be displayed, possibly...

I defer to CONF or A33Zab on that. :)

@HN39 and OK465:
 

I would say YES, but VLS information is somewhat conflicting and at least not clear.

FMGEC supplies GW/Cg to FCPC for calculation of Vs1G, FCPC returns this to FMGEC for VLS calculation.

FMGEC was ready to calculate VLS but was FCPC - in ALT 2 Latched -able to produce Vs1G?

BUSS (NOT on F-GZCP!) when 3 ADR switched off & FCPC revert to ALT (x) latched) is based upon VLS so why not when 2 ADRs/FMGECs returned...

I need some time to sort this out.

Back in AF447 Thread No.6 I constructed a graphic incorporating BEA Interim Report No.3 FDR data in a similar format previously used when HazelNuts39 provided simulated data derived from the BEA's Interim Report No.2.

I have now placed additional data on the graphic, and it can be viewed here.

With reference to the FD, it is interesting to note that the THS (cyan trace) takes off as the PF appears to try and hold the pitch attitude at 15° NU during the period the FD was apparently available. In the same period the thrust has played a considerable part in maintaining the NU attitude.

Yes, it is remarkable how, just before the FD reappear, the situation was improving, or at least stopped deteriorating. It is more noticeable in the graph linked by mm43, just above : The pitch was reducing at 6 degrees and the AoA was getting stable around 5.

But then, the FDs are back !

The question to know if the VLS is computed and displayed is very technical, but I'm not sure it has a big influence on what the FDs are commanding at that moment, or I don't think the resulted command was to put the nose down :

What we can guess is, that the PF, from that moment, is looking for to maintain 15 degrees up, and he's pretty good at it. We know the FDs are displayed, so we may assume the FDs are suggesting that 15 degrees up attitude. I don't see the PF purposely disregarding the FD commands given to him and being that focused and successful. The rule is still : If you don't follow the FD, turn them off.

In that respect, the pertinence of both ADs posted by A33Zab just earlier is also remarkable ... We were aware of the first one dealing with the operational procedure as an OEB was published, but I was ignoring the second one. You produce a lot of relevant information. Thank you.

Whoa there, hoss.

The ADs suggest that the FDs should have extra logic to force and latch them off in the case of UAS, but it's a stretch to say that this is what happened to AF447. More likely they identified it as an issue while running the sim checks and want to close that particular potential hole in the cheese.

I still don't buy the idea of a pilot following FD in the middle of (what started as) a minor emergency situation such as this unless they have a serious scan breakdown - as I said earlier, the FD is for navigation - not aviation. The other reason is that the FD would have been taking it's information from the selected values, correct? According to the trace, none of the "Selected V/S" values during HN39's excerpt even come close to requiring a 15 degree NU climb.

When PF says: "I've lost control of the aircraft" ............

Does he instead mean: "I've lost command of the aircraft"




??

Hi DozyWannabe,
I think with 2 or more ADRs disagreeing, the FDs bars will be removed, and all previous FD/AP modes are "forgotten".
When 2 or more ADRs agree, the FDs will reappear (because they were never turned off) and are "synched" to present aircraft attitude in HDG and VS (whatever the aircraft was doing at that moment). The FD bars will now command that VS achieved on re-engagement.

If the aircraft increased its ROD, then presumably the FD bars will command more nose up.

Probably an excellent reason to turn them off iaw the UAS procedure.