24
$\begingroup$

I am sure when getting their type rating for Airbus FBW aircraft, pilots learn how to fly in alternate and direct law. After that, how much opportunity do they get to practice this skill?

On Air France 447 and Indonesia AirAsia flight 8501, the A/P switched off unexpectedly and the control law degraded to alternate 2. It appears in both situations that the PF had trouble stabilizing the aircraft roll attitude. It's difficult to determine whether this contributed to the pitch up and eventual stall but the original pitch up command may have been inadvertent while struggling to level the wings.

Under normal law, which pilots are accustomed to flying, the FAC calculates the amount of aileron deflection required to turn the amount of lateral sidestick deflection into a rate of roll with a maximum of 15° per second. In alternate 2 law the computer cannot make these calculations so it becomes direct stick-to-aileron control.

This can mean that the aircraft responds very differently than what the pilot is used to. Sometimes the difference may be slight but at high speed and high altitude the same stick deflections can result in a significantly higher roll rate.

Additionally the stick does not provide tactile feedback as to how much force is required to move the control surfaces. With conventional flight controls "roll or pitch rate would more directly correlate with the force applied on the control wheel, and not the amount of deflection required."1

These things combined can result in the pilot over-controlling. The control inputs can get out of phase with aircraft movements resulting in pilot induced oscillation in the roll axis. This occurred in both of the above accidents.

It must be terribly disconcerting for a pilot when the A/P disconnects and they are handed control of an aircraft they fly every day, but doesn't respond the way they are used to. I'm wondering how much practice they get at flying in this configuration, either in a simulator or in the aircraft itself. Do most airlines flying FBW aircraft require this regularly? Does Airbus - or anyone else - recommend any amount of further training or practice? Have these recommendations changed in light of the accidents?


1Bill Palmer, Understanding Air France 447

$\endgroup$
9
  • 1
    $\begingroup$ The Roll wasn't the important part. The Pitch was. In both these accidents the problem wasn't an inability to control the aircraft, the problem was that one pilot in each case tried to pitch up (incorrectly) which cancelled out the other pilot trying to pitch down and prevented stall recovery. Without this, both flights would likely have recovered fine. $\endgroup$
    – Jon Story
    Jan 5 '16 at 11:54
  • 1
    $\begingroup$ @Jon Story I disagree that the problem wasn't in part inability to control the aircraft. An a/p disconnect in cruise should be a non-issue. All that was required was to grab the stick and maintain level flight. Neither of these two pilots was able to do that. In the process of trying to right the plane with suddenly unfamiliar controls both pilots introduced the pitch up that cascaded through a series of other errors into tragedy. $\endgroup$
    – TomMcW
    Jan 5 '16 at 20:13
  • 2
    $\begingroup$ @mins Of course it's easy to say in hindsight. I'm not pilot bashing, I'm saying the complexity of the aircraft worked against him. First order of business is to fly the plane but when he tried to do that he found it didn't fly like he was used to. He was giving full left and right stick movements to counter the roll oscillation even while the PNF was telling him to go light on it. They were in trouble before they even had time to do any troubleshooting. After the A/P disconnected it was only 5 seconds before he set off the stall warning and by 52 seconds they were in a full stall. cont... $\endgroup$
    – TomMcW
    Jan 7 '16 at 9:18
  • 1
    $\begingroup$ ...There's no indication he doubted the artificial horizon, he was just trying to right the plane but was struggling with unfamiliar controls. After they got into the stall certainly there was confusion and bad decisions, but had he been familiar enough with flying in alt law there would probably have been no upset at all. $\endgroup$
    – TomMcW
    Jan 7 '16 at 9:18
  • 1
    $\begingroup$ @TomMcW: The roll control may have been made difficult by the transfer function of the alternate law which the crew was not enough trained for. Agreed. The altimeter and FD false indications could alone explain the stall, their role may have been critical. The crew had not identified the pitot problem and the ECAM was not really helping in this identification. As mentioned in Wikipedia, no clear message stating the cause of the AP disconnection was offered to the pilots. They likely continued to use data that had been discarded as unreliable by the computers. $\endgroup$
    – mins
    Jan 7 '16 at 11:57
12
$\begingroup$

It appears in both situations that the PF had trouble stabilizing the aircraft roll attitude.

Note that roll control is secondary when recovering from a stall.

Restoring normal pitch and roll, he stresses, are "of secondary importance"

The situation required that the pilots heed the stall warnings, be aware of the stall and put the nose down to trade altitude for airspeed.

Have these recommendations changed in light of the accidents?

You can read the recommendations in the official accident reports. These usually explicitly state what such changes should be. I believe it is very rare (except when investigators disagree) for regulators, manufacturers and airlines to ignore any such recommendations.

From what I recall, the major problems in these two incidents were not that the pilots were simply unfamiliar with how control inputs affect flight in alternate law. major problems were that the pilots did not seem to be aware that they were in alternate law or even aware that the aircraft was stalled. In both cases the pilots were confused, failed to follow procedures and failed to coordinate their actions effectively (i.e. CRM failures)

Airbus Flight Crew Training Manuals do cover alternate law.


Recommendations included

5.3 Aircraft Manufacturer
1. The KNKT recommends that Airbus to consider in developing a means for flight crews to effectively manage multiple and repetitive Master Caution alarms to reduce distraction.
2. The KNKT recommends that Airbus to consider and review the FCTM concerning the Standard Call-Outs in all phases of flight.

The Indonesian report lists in some detail the actions taken by the airline and includes an Airbus article "What is a stall? How a pilot should react in front of a stall situation". It also includes some guidance issued by Airbus which includes recommendations concerning upset training in various control laws

enter image description here

After that how much opportunity do they get to practice this skill?

This depends on airline (and relevant regulators) more than on manufacturer. Upset recovery is probably something that all pilots should be trained in, and maintain familiarity with, regardless of who manufactured the aircraft.

$\endgroup$
3
  • $\begingroup$ I understand that the roll was secondary to the accident but the roll was the first upset and it was during the process of controlling it that both pilots inadvertently induced the pitch up. Your last section answers my question, though. Was the paragraph in the image included in the Indonesian report or was it issued by Airbus separately? $\endgroup$
    – TomMcW
    Jan 5 '16 at 19:53
  • $\begingroup$ @TomMcW: The image is from the Indonesian report. It is also part of a separate Airbus document. $\endgroup$ Jan 7 '16 at 9:43
  • 1
    $\begingroup$ good answer and I feel adding that the problem of too much "autopilot" time (understandably, an A/P consumes way less fuel than a human pilot) is well known and discussed in the industry, to the point that a few airlines were discussing (idk how it ended up) about providing free/mandatory flying hours in gliders and GA planes to keep stick and rudders skill sharp. After all roll rates of a K13 aren`t that different from those of a 747 :p $\endgroup$ Sep 21 '17 at 7:22
8
$\begingroup$

Air France 477 was in a fully developed stall. In a normal flight attitude. All the way from 40,000 feet to the surface of the ocean. The aircraft was falling like a leaf from a tree, with an angle of attack of 90 degrees. No buffeting. Yes they rolled back and forth, again like a leaf in the wind.

Before AF 477, stall training focused on recognising the onset to stall, and how to prevent getting into one. The F/O who got the plane in a stall in the first place was partially guided there by the flight director: the BAE found in it accident report that Airbus avionics did not present vital flight information that the computers were aware of. Direct AoA indication to the pilots may have saved them, because they never realised they were in a fully developed stall. They were startled, alarms were going off, and they could not find a visual clue in their instruments what was happening.

Pitching the stick down would have initiated recovery. But the air data computer stops sounding the stall warning at AoA over 70 degrees because that can't make sense, right? at under 60 knots of forward speed. When they lowered the nose, the airflow came within the awareness region again and the alarm sounded. Pulling the stick back stopped the alarm sound. What would you have done, had you been the pilot?

They fell all the way into the ocean, not being aware that they were in a fully developed stall, because that was not incorporated in their flight training. Or in anyone elses - it has only recently been added to the training requirements. Alaska Airlines is one of the first to have added this facility to their simulator.

Your question is a valid one, but is based on only a small part of the chain of events that led to the crash:

  • Blocked pitot. Automation detects this, switches off, and sounds a tone. But where was the clear and immediate indication that the pitot data was invalid?
  • Confusing information on the Flight Director
  • Flying in alternate law, and therefore being able to actually enter a fully developed stall. The basis of your question.
  • Not recognising that they had entered a fully developed stall at cruise speed, cruise altitude. Stall prevention training had always focused on low airspeeds, not on Coffins Corner in the flight envelope.
  • Once they had entered the stall: no visual indication, only many sounds that there was something wrong. Pilot training 101: push the stick forward to recover from a stall. These guys had received those instructions during training in the little Cessnas! They knew that.

The captain was asleep when this happened, and locked out of the cockpit. He could have saved the situation, who knows? One of the blessings of middle age is less fight-or-flight and more of a clear head. Or just a fresh pair of eyes coming in and assessing the situation anew.

The Air Asia captain did realise they were in a fully developed stall, and immediately pitched the stick forward. He was an ex military pilot. The issue there was:

  • Language. An Indonesian national using English to instruct the French co-pilot to Push Back.
  • No direct force cue that the copilot was pitching back while the captain was correctly pitching forward. Airbus averages out conflicting stick inputs if no pilot has taken exclusive control by pressing the override button > 20 seconds.

Recommendations were made for Airbus to change flight information presentation. From Cockpit Standard 1.7 onwards, direct AoA information is presented when the automation switches off. See below under Update. Recovery from a fully developed stall will be part of the curriculum soon, and this is a challenge by itself. You've pointed the nose down, and you're diving straight down at mach 1. Please now gently pull the nose back up while keeping lower than N = 2.5 so that the wings don't snap off.

Update

Asked the question to an A380 pilot who upgraded from A330/A340 recently. According to him:

  • Training now includes flying in alternate law and basic law in multiple flight states, including cruise.
  • Proper procedure in the Air Asia crash would have been to command the F/O to not touch the stick.
  • The direct AoA I mentioned above is actually a backup speed tape which seems to work (is my understanding) on basis of AoA. Keep the needle in the green by providing pitch stick input.
$\endgroup$
1
  • $\begingroup$ I saw the YouTube video. My understanding is they were in a rather severe pitchup. I would have gone to full throttle and levelled the nose. I think even in a full stall that will get the horizontal speed up to something the avionics could understand, and quite possibly that is enough to recover. $\endgroup$
    – Joshua
    Jan 16 '19 at 1:58
4
$\begingroup$

When I did my initial type rating on the Airbus it was rare to find the aircraft in normal law. The thing is most of the complex failures such as dual hydraulic failures, emergency electrical configuration, multiple ADR failures, dual radio altimeters etc always has an effect of downgrading aircraft control laws. The first time I flew in alternate law was in FFS (Full flight simulator) session 3. This is where the instructor forces the aircraft into alternate law to show you how it works. So, for 2 hours you fly in alternate law (no autopilot or flight director). It is pure raw data flying, doing ILS approaches and visual approaches. The whole point of the session is to get familiar with the reconfiguration control laws.

The sessions that followed also contained many scenarios where the aircraft went into alternate law. As I said complex failures always equal flying in alternate law and in the A320 the aircraft will always reconfigure to direct law when the gear is down (if in already in alternate law). That is the story of the one and a half month long conversion course. So, how about after that? As airline pilots we go back to simulator training once every 6 months so it is very frequent. My very first recurrent after the initial did include a complex failure. The dual radio altimeter failure. In this failure the aircraft goes into direct law once the gear is down and you have to fly a raw data approach in instrument conditions. So, do we fly alternate law every 6 months? The answer to that is no. We have so many failures that we need to practice so we cannot ram in failures that make the aircraft go into alternate law every time we come for recurrent. But the training is so frequent that it hardly poses a problem. There will always be alternate law here and there. And we do upset training which is now compulsory. It is always done in alternate law.

enter image description here

enter image description here

Here are Airbus approved A320 type program FFS 5 and FFS 6. I have marked in red those failures that put aircraft in a reconfiguration law.

I also want to comment on one of your statements.

This can mean that the aircraft responds very differently than what the pilot is used to. Sometimes the difference may be slight but at high speed and high altitude the same stick deflections can result in a significantly higher roll rate.

Before I came to Airbus I flew in my airline's regional operations. I flew Dash 8s into small airports with short runways with little to know approach guidance facilities. So, we did a lot of visual approaches and a lot of hand flying. For me, personally, flying the Airbus is the same. I have never felt anything that different. And flying in alternate law is a non event. I have not found the response of the aircraft any difficult to handle. You should not make a statement based on what is written. You have to have first hand experience before you say something like that. One of the most important skills of a pilot is his ability to adapt. I mean in the Dash 8s, sometimes the control cable rigging makes aircraft controls heavy and sometimes weird which makes each air frame unique. But as a pilot you handle the situation and fly the aircraft. Flying in alternate law should not be a problem for a trained pilot. And we cannot really use one accident or actions of one set of pilots and use it to define the whole piloting community.

$\endgroup$
2
  • $\begingroup$ Thanks for your response. I'm glad you hear that you get upset training. I think that might have been made compulsory in response to the two accidents I mentioned in my question. There are many questions that are probably unanswerable in the AF447 crash. But the one I keep coming back to is this: why could a trained pilot not seem to be able to fly his aircraft straight and level for 30' seconds in cruise flight without stalling it. From the data and CVR he appeared to be struggling with pitch oscillations, which I simply wouldn't expect from a trained pilot. $\endgroup$
    – TomMcW
    Dec 13 '20 at 23:37
  • $\begingroup$ I'm conjecturing, really that the change in control laws without any force feedback threw him off enough that he struggled to contain the oscillations. I'm sure that roll inputs feel very different in the coffin corner than they do low and slow and usually the computer accounts for that. With the Dash 8, for example, you can immediately feel the difference between stiff or sloppy controls. With a sidestick there's no such feedback. The good thing is that the industry learns from these things and today you will hopefully be prepared in these situations better than the AF447 was. $\endgroup$
    – TomMcW
    Dec 13 '20 at 23:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.