PuraVidaTransport

When the plane reverted to Alternate Law (2B if I remember correctly), engaging the autopilot was not an option. Once latched into alternate law, the autopilot can not be reengaged until the system is reset on the ground.

My only questions still outstanding in this mystery (outside "What were they thinking") is why with several nose-down inputs to the control stick, some lasting several seconds, why the elevators never moved into a nose-down position. With old cable controls, if you put the stick all the way forward the elevators would instantly respond with nose-down movements however, in the Airbus fly by wire, the elevators never got into a nose-down configuration despite several full nose-down stick inputs. They began to move a bit but were still in a nose-up position the entire ride down. I've wondered if there is a cumulative effect with all the nose-up inputs that delayed the movement of the elevators. The THS I understand...since the elevators never moved to a nose-down position I wouldn't expect the THS to move that way either. I've asked that question a couple of times and no one seems to know

Edited to add: Looking back at the traces, there is around 15 seconds of nose-down stick from 2:13:45 to 2:14. The elevator was at 30 nose-up and in that 15 seconds of almost full nose-down stick, the elevator only moved to 15 nose-up. So even though the pilot commands nose-down, the fly-by-wire never moved the elevators to a nose-down position...for 15 seconds!!! As they only had a bit over 4 minutes from start to finish, I find that response time to pilot inputs awfully slow. Granted, by this point, it was to late to recover but if they had begun a recovery much sooner, such slow response to pilot stick input would have made recovery very difficult, especially considering the THS never moved from full nose-up. If it only moves 1 degree per second, that would mean almost a full minute of full nose-down stick would have been required to get the elevators to full nose-down with the THS beginning to move that way at some point in the process. So just how long would it have taken, with full nose-down stick, to get the elevators and THS into a nose-down position??

CONF iture

But as the CPT PITCH COMMAND is for ND the FO one is for NU, except for only 7 of those 15 seconds.
Also for that period the vertical acceleration is already slightly below 1.0 even with the elevators in a still UP position therefore the pitch law in force at the time is honored versus the request.

Direct Law would have been my choice too ... and also mainly for leaving the trim alone in the first place.

alogobotur

What is happening in that Airbus if the inputs from the pilots are opposite?
I saw comments like "dual inputs" in the reports. Does this mean that one pilot is giving nose-up inputs, the other one nose-down?

What is happening in that situation, who does the plain "listen"?

Lonewolf_50

The plane averages the inputs. If you are pulling nose up six degrees, and Im an pushing down six degrees, the plane will be about nose level.

Remember, the design and intention of this aircraft, a multi crew airline transport aircraft, was based in the law/rule (and common practice) that only ONE pilot is in control at a time. Rather than letting the sysetm go ape if two pilots "fight over the controls" the aircraft system enforces electronically, via the flight control system, a disincentive to have two people trying to fly at once. That's not that bad of an idea, actually.

Since so much of the flight control movement is via computer, electronics, and hydraulic linkages, two pilots puttting contradictory inputs in at the same time could well break something.

That would not be good, when it comes to flight controls.

Lonewolf_50

PuraVida:
Thank you for the more precise and more accurate statement. I had forgotten the latching in 2B.

algo, what that means in layman's terms is that not only will AP not function when certain data is unreliable, like airspeed, (the computer for the A/P has a logic function that detects conflicting data and at best avoids applying bad data to flight control inputs), but in the case that the degradation takes the plane into a control law called alternate 2B, you can't even put A/P on until you get to the next landing and maintenance facility and have the various features checked and reset.

Again, you don't actually need Auto Pilot to fly. You can fly the plane without it. (It's a bit more work for the crew to do so, however).

In other control laws and modes, you could reset it in the air once the airspeed data problem gets sorted out.

You are welcome to search Control Laws in the series of threads on this based on the link I gave to you for page one of this thread.

A great deal of very educational discussion was posted here regarding how A330 Control Laws work. These aren't the AutoPilot per se, but the compbination of mechanical, electric, hydraulic, and computer driven inputs to the positioning of the flight controls as a response to pilot action, or level of AutoPilot pilot assistance selected for a given phase of flight.

jcjeant

I'm certain that the pilots of the AF447 had read and learn it time ago before it was posted here
However it does not have served them a lot to regain control of the aircraft
I think they had to learn other things .. or they had already learned it .. but forget it just when they need it so much ..

Lonewolf_50

jc, your opinions aren't helping explain this to a layman.

The dual problems of training and proficiency have been beaten to death. I am trying to make neutral and helpful replies. Where I fail to do so, please help explain to this layman.

Our friend algobotur initiated his inquiry with a seeming belief in AutoPilot as a tool and a fail safe. I have tried to explain to him how that is an invalid approach to take in understanding this tragic accident.

The core flying principle involved in my explanation to a layman is that you fly the airplane, and use various AutoPilot features help you in doing so, or in managing your task load. I suppose we should explain that AutoPilot is required for use as much as possible by many companies since they find fuel (and thus cost) savings from modern automated flying tools.

I suppose we can then explain that if the pilots, like our departed friends in AF 447, spent a great deal of their flying time monitoring the function of AutoPilot rather than flying themselves, the sharpness of their hand flying skills may have eroded. I think it's fair to say that many pilots believe that industry wide situation contributed to this accident. But that's also a conclusion with significant discussion and debate remaining.

It ought to be a major point of concern to the airline industry, since the pilot is expected to fly the plane regardless of how many sub systems are acting up.

Whether or not Air France, and its pilot training and quality programs, are about the same as, worse than, or better than the industry norm is to me unknown, and may be unknowable.

EMIT

A couple of posts ago, somebody asked about inputs by both pilots at the same time.

Actually, the reaction to such a situation by the aircraft is not so different in a Fly by Wire Airbus than in any other aircraft.

If in a yoke equipped aircraft, one pilot pulls the yoke with a force of 30 Newtons and the other pilot pushes with a force of 15 Newtons, guess what, the strongest force wins and the yoke moves in the pull direction.

In an Airbus, this happens electronically: 30 N pull minus 15 N push is equal to 15 N pull.

The difference is - electronic addition is not felt by the pilots, they do not feel that the other guy is also handling the controls. With a yoke, you can feel whether the other guy is interfering with the controls.

Busserday

One sidestick at the forward stop and the other at the rearward stop equates to both neutral sticks regardless of the force (pressure/torque) applied, does it not?

PJ2

"One sidestick at the forward stop and the other at the rearward stop equates to both neutral sticks regardless of the force (pressure/torque) applied, does it not? "

Yes, that's correct. I don't think the example using control columns is quite correct but I know what the poster is getting at. In the example, the CC would move in the direction of the greater force and the airplane would respond according to the extent of the movement which would, in the example, continue until the CC hit the stops (commanding full-up control deflection), while in the Airbus the SS movement (in degrees) is algebraically summed and the output to the flight controls is that sum. In the case where the sticks are at full deflection but in the opposing directions, the output commanded is neutral and no control deflection occurs.

The fact that the other pilot cannot know what input is being made and therefore cannot make correct assessments as to aircraft handling is the reason why dual inputs are prohibited. The correct procedure is to press the takeover button which temporarily locks out the other sidestick. Keeping the button pressed for (IIRC) 45 seconds locks out the other sidestick until landing.

jcjeant

Even this definition is not strictly accurate (pedantic .. lol)
If it was really the sum of the degree of displacement of the joysticks .. when both joysticks are in the same extreme position the output (deflection) should be double that obtained with a single joystick ... which is not the case .. so it might be simply that this is a algebraic sum if both joysticks move in the opposite direction ..

fizz57

No jc the inputs are always summed even if they are in the same direction, however the output is limited to full deflection.

Thus 1/2 NU+ 1/2 NU= full NU, any further NU deflection of either sidestick causes no further effective input.

Source: smartcockpit.

Try actually doing some reading before expressing an opinion, will you? You're reminding me of anothe very opinionated guy who had to "refer to FCOM" to find out how the airbus AT behaves with the lever out of climb...

alogobotur

Thx for the AP answers. So, AP switching on was not an option at all. It was also technicaly impossible.

The other thing, what was the situation with the AF447 radar? They were obviously flying thru the storm, why they didn't go around it? As the transcripts say, Robert said to Bonin something like "don't you want turn to the left" and then he turned to the left, but obviously didn't avoid the storm.

In one video I hurd something that this radar also could be misfunctional?

I know all of these questions are probably answered in the Final report, but this is much more easier way for me to have the answers

Lonewolf_50

They did in fact make a course deviation, but the freezing up of the pitot tubes was not due to being in a "storm" per se, it was due to the character of frozen water droplets at altitude and a pitot tube that wasn't as able to handle that problem as other pitot tubes one could equip an aircraft with.

You will need to go back to threads 1, 2, and 3 and read about the in depth discussion on the radars and pitot tubes to help flesh out this answer.

Remember this: passenger aircraft flying at similar altitudes carrying hundreds of passengers cross through the ITCZ every day and don't crash.

"They flew into a storm and crashed" is a red herring.
If you bother to read the history of this investigation and this flight, a few things become apparent:

1. The pitot tubes (all three froze up) were scheduled to be replaced in Air France aircraft, and some already had been, but had not yet been replaced on this particular jet. There was an airworthiness bulletin to that effect published a some months (years?) before this accident.

2. The pilots had already slightly reduced speed to account for flying through/near a storm and expected turbulence before they ran into that pitot icing problem.

3. The pitot tubes being rendered unreliable for a short time due to a particular kind of high altitude ice crystals made for unreliable airspeed.

4. The procedure for dealing with that at high altitude appears not to have been followed. That said, the UAS procedure has come under review since that accident. Many discussion on how to train for that malfunction have been recorded in the threads, on this topic. Use the search function on page on to find them.

5. Even with a less than stellar response to the UAS problem, a great many pilots who have years and years of experience flying that route will tell you that setting the correct pitch and power for the altitude and speed desired will keep the plane mostly level, at cruise, until the crew can sort through the malfunction and get the various systems back on line.

Point 5 is a pilot thing, not a machine thing.

While there was a machine malfunction, the pitot tubes taking time off due to ice crystals, most machine malfunctions have a series of procedures and remedy that the pilots apply to mitigate their impact, and to restore performance. The UAS procedure is one such remedy procedure.

The core pilot problem was in not applying the simple measure of flying an acceptable pitch and power while the NON-flying pilot would work through the systems to get them back on line. A contributor to that problem in this case seems to be the lack of hand flying time a lot of air transport pilots get at altitude in many fly by wire jets. If you don't practice something, your skills will be rusty when you need them. This is, based on the inputs of many professional pilots who post here, an industry wide problem.

The core machine problem began as all pitot tubes failing to function (even though there are three to provide redundancy), in part due to an already identified sub-par set of tubes being installed and not yet replaced by better tubes. Airspeed indication and input into the flight system is a crucial bit of information for both pilots and computers to use when flying the aircraft.

The radar don't really enter into it. The judgment call on how far to deviate from a given weather system is just that, a judgment call. Feel free to read the hundreds of posts on that element of this accident. The opinions among those who fly big jets for a living varies a bit on that point.

With that in mind, many experienced pilots note that some of the other intercontinental flights that night had deviated further around the weather system than AF 447 did. The analysis of the FDR data, once it was recovered, didn't show a significant amount of storm associated turbulence to have been a factor in the flight departing controlled flight. (Stall is a departure from controlled flight).

6. There were 32 incidents in the years previous to this mishap of unusual aircraft behavior and upset that were somewhat similar to, though not identical to, the malfunction the AF 447 crew ran into.
None of the others crashed, though some of them had bizarre altitude excursions that they had to deal with. A Qantas flignt (QF 32?) is one you might want to read up on for its similarity to the AF 447 accident.

Once again, please read the report.

Chris Scott

Quote from Lonewolf_50:
"Once again, please read the report."

Good advice....

You could read it slowly ten times over, and analyse the DFDR traces at length, in the time it would take to read all the AF447 contributions - good, bad, irrelevant, bigoted, open-minded, pig-headed, inquisitive, perceptive, indifferent, and even brilliant - on this forum.

And that short list of adjectives could no doubt be amplified.

CONF iture

... And also by how much he's interfering with the controls. It is direct first class information.

Electronic addition can take place without aural warning too ...

CONF iture

Not necessarily until landing as a deactivated sidestick can be reactivated at any time, by momentarily pressing either takeover pushbutton on either stick.

A33Zab

In fact its a little bit more complicated.

The AP(AUTOMATIC FLIGHT) can be engaged in both NORMAL as ALTERNATE LAW!.

The AP has its own 'law' built in and 'bypasses' the MANUAL control laws inside the FCPC(aka PRIM).

The AP commands direct flight surface deflection orders to the FCPC, the role of FCPC in automatic flight is to limit those orders when it 'senses' to be required so and to transfer the surface commands to the slaved FCPCs and FCSCs and their respective servos.

The AP however needs reliable airspeed parameter (to set the gains) and VLS(lowest selectable speed) to be available for operation and that was absent in this case.

Apart from the explained SS algebraic addition:

Due to missing airspeed parameters, the filtergains of the flight control law were set to a fixed high speed value (IIRC 330KT).
This combined with the rather small integral(in time) part of the SS order resulted in this 'sluggish' elevator reaction but as one can extract from the FDR traces the reaction of the aircaft was a negative pitch and a positive Nz as commanded.

CONF iture

Would filter gains apply to Direct Law too ?
In other words, is Direct Law that much 'direct' after all ?

Where ?

Lyman

CONFiture

"Where?"
At Court, to the Judge.

Autopilot is available to the flightcrew after UAS, Airbus (OEB) directed crews to not select it, as it "may" cause an "uncommanded climb" creating "risk".

It is BUSS that disqualifies A/P, until a reboot, on the deck...imo. Also I think NORMAL LAW is not reselectable post BUSS.

Captain DuBois, in the CVR, made it a point to check the A/P switch, late in the game. Why? Did he cycle it? The CVR reports (CAM) a noise of switch movement? Was he concerned about the "sluggishness" of the elevators? Whether due Bonin's pulling, Captain was obviously concerned about the lack of ND? I cannot find any command from Captain: "Putz!, NOSE DOWN" Can You?
Instead, the command is "Try Climb" "But I have aft stick for awhile..." Elevators were the topic of the day, yet all were at a loss as to continued descent.....

The instruments were alive, and accurate, we think, wouldn't the commander be distressed at the NU along with the 10000fpm descent?

WITHOUT QUESTION, the Captain must have assumed the controls were INOP, hence no NOSE DOWN response from the A/C. Where is evidence of his dilemma as to no ELEVATOR control? If he considers Bonin competent to fly, he can have no other conclusion, nor can Robert, who has been privy to the a/c PITCH migrations, from the outset.

Captain not once looks at Bonin's stick, just to make sure? Baloney. Something is fishy here. Robert is satisfied that the a/c is at fault, he must be, he agrees with Bonin "We have lost all control". Not a peep about the climb when DuBois returns?

As to CVR. Failing a release of the (complete) record, perhaps the BEA would consider releasing the audio portions that were transcribed.

Why not? The words are released, why not the voices along with them?

Any guesses?