DozyWannabe

I'd have thought that type conversion training would take care of that tendency. More to the point, if I recall my Chippy days correctly you only need apply backpressure when rolling into a turn. I don't think it should be necessary just for levelling the wings.

Machinbird

Dozy,
Remember that whatever Bonin was doing, it was logical to him to be doing that at that time. He was not a stupid person. You don't make it into a cockpit if you are stupid. (foolish is different). The problem is that our hind sight bias blinds us to understanding these perfectly logical steps that Bonin took.

The art of hand flying an aircraft has certain elements to it that are common across a wide range of aircraft. Sometimes by looking in great detail at the problems an accident pilot faced, we can begin to understand why, at the detail level, he did what he did. This can be the key to understanding the larger context. Is it a perfect technique? No, of course not, but when your analysis of what was done begins to make sense, then you have to be getting close to what must have happened.

Please do not trot out that over-repeated phrase of Bonin's that he had been at maximum nose up for some time. The flight conditions had changed considerably from when he first took contol and he must have made 200 to 500 small decisions about handling details of his control problems before he made that statement. To try to apply that statement to the beginning of the manual control period is reeeeaaaally a long stretch. You can just about see when he implemented that decision, and it was much later from what I can see.

_Phoenix_

DozyWannabe, interesting suppositions about the back pressure reasons

However, I would not stick with only one possibility

1. initial climb was deliberate, as stated in CVR, his intention was to climb over the weather, but in the same time he had difficulties in controlling the roll, he over-controlled, because of surprise effect and direct law for roll, at high speed. The travel of sidestick in roll is larger compared to pitch direction, therefore by applying large movements on lateral he might applied more back pressure than intended

2. He followed the FD. At 2:10:50, stall alarm goes off, but aircraft was still in climb for next 20 seconds, Bonin pitch command looks erratically, mostly NU input, but if graph is correlated with FD order, it seems he was actually chasing the horizontal line (Final report, figure 28 correlated with figure 69)

3. Inadvertent pull, at 2:11:30 roll command is at max left, but the aircraft continued to roll towards right. He was fighting with roll control, for next 10 seconds, slowly and inadvertently, the sidestick moved from 9 o'clock position to 7 o'clock (dead stop for both roll&pitch)
Bonin might lost the reference of sidestick fwd-aft and left-right. He exits from 7 o'clock corner directly in the opposite corner full right-ND, the opposite combined stop. Then he might re-position his hand, but already was too late... the irreparable damage for THS position was done.

4. overspeed fear - I would just add the aerodynamic noise combined with misleading stall alarm, Bonin might believed that stall warning comes on because the airflow is disrupted when aircraft is approaching the sound barrier

vilas

Dozy
This is off topic my apologies for that. I would like to hear about Airbus elevator integrator function from you. Thanks.

DozyWannabe

No, they'll be internal designations used within Airbus for certain configuration types. The FCTM does mention that in some Alternate modes, roll will be direct, and either some or all "soft" protections will be missing.

@vilas - I'm not the person to ask about that really, but I'll try to find out. What specifically did you want to know?

vilas

Dozy
I have an article that explains it as a circuit in FBW flight control. I have some idea of its function but it say terms like "feed forward gain" and response error over time. As I understand it moves the control first till auto trim catches up. I thought you could give better explanation.

Owain Glyndwr

@vilas


Could you post a URL for the article?

CONF iture

Therefore I strongly suggest you read again the CVR ...
The difference between the crash and a crew who knows what's going on and how to deal with the situation more elegantly ...
http://www.pprune.org/7450265-post576.html

jcjeant

Hi,

Tragic animation indeed .... show more than all graphics and curves !

_Phoenix_

Winnerhofer, Thank you! I heard about this simulation

Indeed, I'm amazed how fast the time elapsed. We zoom in graphics, but actually, in reality, in about 90 seconds (at 1:30 in video) the THS is near max NU - there, they were already doomed. More than first half of this period of time, Bonin learned to fly the plane at high altitude, he was absorbed by roll control.
As I read somewhere (sorry I don`t remember the article, please correct me or confirm), for both PFDs of simulation, the inputs are from captain's DFDR data, we will never know what Bonin saw on his display.
We can see better from his point of view if hide the imagine of stalled plane (i.e. a notepad window) and try to ignore THS and AOA indications.

By the way, if aircraft has been equipped with those instruments, AF447 would still fly today.

Edit:
An interesting result for combination of simulation video with sound of the reconstruction in link below, start at time 3:24
https://www.youtube.com/watch?v=dKT3dd_ko8E#t=204

I watched them in parallel, then I realize that my post #649 makes a lot of sense.

DozyWannabe

Not necessarily - if the sim experiments are anything to go by, then there was at least a minute after the THS hit the nose-up stop in which full nose-down could have remedied things.

@CONF - I see nothing that relates to any reset of the system. Where do you see that happening?

_Phoenix_

DozyWannabe,
Please revise posts #528 and #528, page 27

RetiredF4

A theoretical assumption.

In the animation video
at 2h:11m:55s the pitch is -10°, AOA is 38°
at 2h:12m:08s the pitch is -10°, AOA is 41°
at 2h:12m:52s the pitch is - 8°, AOA is 36°

The decrease in pitch did not do that much in decreasing the AOA.

Now take your guess how far the nose has to be stuffed down to reduce the AOA to below stall AOA, how long it will take, and how much time and altitude a cautious recovery to level flight will chew up without risking secondary stall.

In hindsight with the FDR data and with this animation we know how high the AOA really was, the crew did not and therefore did not react in an apropriate way.

A successful stall recovery after 2h:12m without the knowledge part of the flying community gained from AF44 is out of my imagination.

Bpalmer

My own A330 simulator trials of recovery after the stab trim was full nose up and the stall fully developed was that a prolonged nose down input did help—but only initially. However, as the angle of attack reduced and the speed increased, apparently the full nose up stab was more than the elevator could overcome and the airplane pitched up regardless of the nose-down stick position.

It's not clear if the autopitch trim was operational with questionable airspeed inputs, but despite high bank angles in an attempt to keep the nose low, recovery was not achieved until I reduced the stab trim manually. Once trim was reduced manually, pitch control resumed and recovery was possible.

I cover this in more detail in my book "Understanding Air France 447", including what would have been necessary to recover from each point in the scenario. Essentially, the less skilled they were getting in to the stall, the smarter they had to be to get out of it. As we all know, had Bonin simply attempted to maintain a normal 2.5–3° cruise pitch attitude for 40 seconds (the total pitot tube incapacitation time), they would have survived just like the other crews that encountered the same phenomenon and made it OK.

The appalling thing is that the stall warning was going on for over 50 consecutive seconds (plus significant aerodynamic buffet), and the only thing done to recover from it was to apply TOGA power (which at that altitude gives very little additional power over normal cruise setting and zero additional power over climb thrust). I attribute this to training stall recoveries only at low altitude with an emphasis on "minimal altitude loss." This amounts to applying TOGA thrust and reducing the pitch only slightly—what AF447's crew did. But when the airplane was falling at 10–20 thousand feet per minute about the only thing Bonin could think of was pulling back to make the airplane stop descending. For that is where his training of sticking a low-time pilot in the seat and tell him to just follow the flight director did all 228 a great injustice.

Machinbird

The PRIMs have AOA inputs. With NCD AOA info, do the PRIMs still work?
We have had discussions about this before but I no longer remember the outcome.

Bpalmer

Actually the Elevators each have four computer control possibilities and two hydraulic drive possibilities:


Control priority for each elevator and associated hydraulic system for each respective actuator is as follows:
Left Elevator:
PRIM 1: green
PRIM 2: blue
SEC 1: green
SEC 2: blue

Right Elevator
PRIM 1: green
PRIM 2: yellow
SEC 1: green
SEC 2: yellow

Stab trim is driven by blue and yellow hydraulic driven actuators (usually simultaneously), If one is out, then the other can drive it independently.
Hydraulic power to the two motors can be controlled by any one of three electric motors or mechanically through cables from the pitch trim wheels. The three electric motors are each controlled by a different PRIM.

MrSnuggles

This is a fascinating accident indeed. So many levels of wrong in such a short period of time.

What I fail to comprehend is how a trained pilot can completely disregard the information "nose up input" and "spinning altimeter" and "STALL STALL". I read some posts ago that maybe Bonin was afraid of an overspeed situation. While that is a serious concern I can not see it was in any way applicable here.

Could someone please enlighten me as to why Bonin might have thought it was an overspeed? I really don't get it. He should have been aware of the stall buffet waaaay before starting at Air France.

MrSnuggles

-deleted stuff-

MrSnuggles

Mods: Feel free to put this post in the bin.

RetiredF4

First I'm cautious concerning the extent of reality of present flight sims concerning stall behaviour and reaction of the sim on flight control inputs in this untested flight regime.

But looking from a logical point, why should the THS trim start running nose down right away after applying SS nose down?

The position of the SS commands a loadfactor (g) demand, not a pitch demand. With SS neutral the value is known to be 1 g. SS Nose down demands less than 1 g, SS nose up demands more than 1 g. At low speeds the command is a pitch rate change. AFAIK we still do not know, what the preset rates for those changes are per unit SS deflection, and how the speed changes the preset gains. In case of lost speeds, what value of speed is used by the FBW system to transfer a SS command to the appropriate elevator deflection? And does it change again when the speeds are valid again? Was the FBW system using low speed gains during the stall, high speed gains or preset gains, did the SS stick command a loadfactor or a pitchrate change during stall, or was it a mix with changing gains?

Looking at the FDR and the simulation video i would assume the result was more loadfactor than pitch rate driven. Look at the g values during the stall, due to the increasing sinkrate the value was less than one g most of the time, despite the full nose up elevators and despite the nose up driving THS trim. Both inputs could not arrest the descent and thus could not deliver 1 g or more with full deflection.

SS nose down would command a value of less than one g, or if low speed gains would be used a pitchrate change to nose down. I am pretty sure that the system is set up very sensible for nose down inputs to not spill coffee and bisquits all over the place during normal operation, and it would not know that the aircraft is stalled and therefore a quick change is necessary. As the loadfactor was already below 1 g most of the time, small elevator deflection to less than full nose up elevator would be sufficient to decrease the loadfactor within the allowed gains and values. That's what happened when the SS was kept forward and the elevators went from full nose up to about 1/2 nose up. The THS trim would only react with a Nose down travel to remove a prolonged nose down elevator position, which never happened. It would happen in the end if the SS is held down long enough, but imho that time was very quickly running out.

To expedite the process of reducing the AOA in the available time would only be to use manual nose down THS trim, thus avoiding the comfort gains of the FBW system. How cautious this has to be done in order not to overstress the airframe in the low g or even negative g region i dont't know.