Afriqiyah Airbus 330 Crash

[Neptunus Rex]

Computers

It is not "a computer." On the FBW Airbuses, there are five Flight Control computers. They are designed and built by two independent companies, to avoid the duplication of any software glitches. For one computer to fail is rare; for two computers to fail on the same flight is rarer still. Even if all five were to fail, the aircraft can be flown manually, via the elevator trim for pitch and the rudder for roll. Sluggish, of course, but manageable and practised in the sim.

[GarageYears]

Sure let's blame the computers, they "did" it

I presume this was not the very first time an A330 had (attempted to) land(ed) at Tripoli? Assuming not, if the "computer" did it, how come these other flights managed to land OK. Did the computer just decide that this time it was going to screw up?

I have worked with computers for my entire life and find this whole line of discussion just plain wrong. The code running in the Airbus family is embedded, and quite different from the run of the mill home/office PC code we are so used to locking-up/crashing.

The code we're talking about here is much more akin the engine management system of your car (admittedly many times more complex) but (this important) the code is running as a single thread, dedicated to the computing platform itself, NOT parasitically balancing on top of a general purpose operating system like Windows or similar, that is designed for you to run Word, iTunes, your email, and any number of other applications all at once, and allow you, the user to switch them in and out at random.

An FMC runs FMC code. That's it. So the problem of testing it, debugging it, and error handling are in fact much simplified. Also the code itself is subject to REAL testing, unlike the majority of domestic/office code that is in the main left to trial by fire testing with beta testers...

Also, as another contributor has already pointed out, these systems are multi-redundant, and have at least 2 suppliers that developed code entirely independently.

I'm not saying these systems are infallible - but the problem is the USER. Despite all the protections built into these systems, if the pilot commands the aircraft descend, then descend it will.

Much more likely in this case is that the crew thought they were somewhere different than they really were and, due to a combination of sun, haze, and perhaps fatigue (other posters have already crafted similar and better written scenarios for this), got the aircraft into a position that was unrecoverable, no matter what automation was available to them.

- GY

[Tom355uk]

Ok, the IB A320 incident - that doesn't seem as though it was any sort of software/hardware failure. The aircraft was caught in an unusual and difficult situation just before landing, which would probably have resulted in a hard landing in ANY aircraft. However, if the CAPT priority button had been pushed, as per the FCOM, it seems as though it might not have happened quite so firmly. Still, nobody was hurt.

After forty years in professional airline flying, I rather think that this poor fellow has no idea about airline flying...nor much of anything else in aviation.
Sounds like a 9-5 desk jockey to me.

I should hope you are not directing those snotty comments at me.

I'm not pretending to know anything about airline flying. All I stated was the fact that aircraft do not crash themselves. You cannot, regardless of your experience (which I do respect, even if you do not respect me) tell me otherwise. I was even more disappointed that the original comment regarding soft/hardware failure came from someone who, if their profile is true, should know far better than to post such sensationalist drivel.

So what if I work 9-5 behind a desk? (It is 9-6, actually ) Just because I'm not fortunate enough to have had the chance to make a career as a pilot (yet), it doesn't mean I don't have a valid contribution to make to this thread - a contribution since vindicated by both Neptunus Rex and GarageYears.

As an aside, I always make a bloody thorough effort to make sure that I research my comments fully before posting, contrary to a majority of posters on this (and other) forum. I also happen to have a deep and lifelong passion for aviation - if you have decided that I know nothing, I hope that makes you feel better.

[C-SAR]

Out of Dodge

Today I got out of Dodge City...
While pushing back in our little 737-400 (as a passenger, of course) I was looking at the A330 parked next door. Particularly I was looking at the position of the horizontal stabilizers in relation to the fuselage axis and to the bottom of the fuselage and I was comparing that to my visit to the crash site, to the first print, to the cactus cuts.... Now, if the aircraft was leveled or pitch down there is no way the HT could cut the cactus where it did (the cuts are no more than 1 - 1.5 meters from the ground)
I am going to do some drawings and calculations when I get home, just to determine (if possible and always keeping in mind that these are empirical calculations) what sort of AOA the aircraft should have to produce a cut with the HT 1.5 meters from the ground. Also one of the pictures that shows the tail complex shows the angle at which the HT were (it looks like a healthy full pull up) Experts of A330 please look at that picture and express your thoughts if you think that indeed the position of the HT shows a pitch up command.

Any help on doing the drawings and the calculation of the HT position is more than welcomed. Refer to the pictures and diagram already posted days ago.

Thanks

C-SAR

[SLFinAZ]

I couldn't disagree with statement any more strongly. The real question here isn't the quality of the software but the quality of the documentation and the motivation behind the software.

1) The primary purpose of the software appears to be engineering "pilot error" out of the equation and to allow for an overall deterioration of pilot quality. Instead of the software allowing for management of complex tasks in a fast moving environment it is being utilized to perform standard tasks in normal operating conditions. in effect the software is flying the plane and the "pilot flying" is monitoring (supposedly).

2) Not only do we have ongoing issues relating to "what is it doing now" but the software flies the plane...literally until it can't and then hands off a plane that by definition is beyond the capability of the AP and accordingly far beyond the capacity of many current airline captains to intuitively recover.

The simple reality is that current FBW software is contributing tremendously to the erosion of overall skills in the cockpit and a direct contributing factor in the increase in accidents directly attributable to the lack of hand flying ability being demonstrated again and again.

At some point either we go back to pilots actually piloting or we just turn automated airplanes over to bus and trolley drivers...

[GarageYears]

The primary purpose of the software appears to be engineering "pilot error" out of the equation and to allow for an overall deterioration of pilot quality. Instead of the software allowing for management of complex tasks in a fast moving environment it is being utilized to perform standard tasks in normal operating conditions

Am I reading this right? You believe the intent of the software is, as it's primary purpose, to allow for an overall deterioration of pilot quality? Bullst. Now that doesn't mean some pilots don't take excessive advantage of automation leading to a deterioration in piloting skills, but as many of the old hands on this forum have repeated - basic flying skills MUST still be practiced, and many SOPs encourage this (though true enough, others don't). But that was NOT what Airbus designed into the software.

(I don't disagree with your first point - engineering pilot error out of the equation - for example stall protections).

- GY

[fireflybob]

basic flying skills MUST still be practiced, and many SOPs encourage this

GarageYears, my experience is that most SOPS encourage full use of the automation and in fact discourage hand flying!

[mike-wsm]

Has anyone on this thread stalled this plane? Could you talk us through what happens, please? Yes, I know you can't stall it, but what actually happens when you fly it slower and slower? Does it sink, spin, drop a wing? What is the aoa? Does it recover or does it go on dropping? Sorry, not trying to be sensational, just trying to understand what happens when it flies slowly.

[gums]

A few thots from a "professional" pilot

Salute!

Compelled to post concerning this sorry story of a fatal crash. Won't comment on the "sorry" comments, however.


1) I am constantly amazed at all the autopilot and flight control "laws" for the AB. Scares me.

I realize that many "modes" and "functions" are intended to reduce crew workload and enhance overall safety. But somehow I have this feeling that the plane is just too "cosmic", and needs to be simplified.

2) I flew the first operational jets with a fully fly-by-wire system. My name ( callsign) is well-known in the community. And I did it years and years before the first FBW Airbus. Just establishing a small amount of credibility so nobody thinks I am a troll or computer flyer.

Our operational requirements were vastly different than for the commercial airliners. I understand all that.

What I have trouble with are all the fine points here that discuss such-and-such autopilot mode and such-and-such flight control mode and the various connections between them. It is personally disturbing to me. Someplace there must be a mode or law that allows the actual human being to command full power or full climb/dive command ( all with due respect to basic flight control computer laws) and eliminate all the "competition" between the various systems.

Continuuing...... The GD FLCS ( flight control system) folks refused to allow our autopilot to do much of anything. Sure. We had some basic functions like altitude or heading hold, but the FLCS had its own rules and at certain AoA's, the plane would descend even if altitude hold was engaged. It didn't even allow the AoA to reach normal limits that could be achieved if we had the autopilot turned off. One fatal due to that issue, BTW, and was simple pilot error by not monitoring altitude. 'nuff said.

3) I fully understand a scenario where sun in the windshield could obscure the landing environment. I simply don't understand a decision to either manually fly or allow the gizmos to fly you below the MDA. Maybe I am too old-fashioned. And I manually landed in many situations that required a level off for a NDB (ADF) or VOR or TACAN or localizer-only approach. Granted, I could easily pull up and almost instantly climb if I had set the altimiter wrong if I saw something that didn't "look right", as I didn't have to fight all the control laws and autopilot modes and even throttle modes. No "autopilot-coupled" modes in the Viper, although there was an excellent one for ILS in the VooDoo.

So I would wait to see what the CVR says and find out if the crew was talking about what a ****ty view they had on the approach.

4) I am amazed by the small pieces of the debris field. Sheesh.

5) For C-SAR, there should be decent geometry to "guess" whether the main gear hit first or at the same time as the tail plane. The actual aircraft pitch attitude should be used and not AoA limited by the gizmos, as we must add AoA to angle of descent. Sadly, it appears that the most serious aero/physical crapola happened downstream of the recorders. Nevertheless, we can still get valuable data.
++++++++++++++++

Finally, I want to award C-SAR a few dozen "attaboys" for his relentless pursuit of the truth. Outstanding work, and most of us really admire your efforts.

Gums sends....

[infrequentflyer789]

Originally Posted by SLFinAZ

The real question here isn't the quality of the software but the quality of the documentation and the motivation behind the software.
[...]
in effect the software is flying the plane and the "pilot flying" is monitoring (supposedly).
[...]
The simple reality is that current FBW software is contributing tremendously to the erosion of overall skills in the cockpit and a direct contributing factor in the increase in accidents directly attributable to the lack of hand flying ability being demonstrated again and again.

It's all very well talking about the documentation and design quality, but you can make that as high quality as you like, and there will still be those who fail to read or understand even the basics of it ...such as what the FCC and the autopilot bits actually do, and which is "FBW".

[Jetjock330]

Mike-wsm,

Has anyone on this thread stalled this plane?

Yes, it does stall, outside of normal law. It is a simulator exercise we do in alternate law and it falls like a brick in a deep stall, nose down in the vertical. Don't expect much out of 14000ft!

[HundredPercentPlease]

Originally Posted by Bluestar51

100PP,

Is it SOP for the Airbus to fly NPAs using a constant rate of descent? Is the NPA flown in MAP mode (not sure about the Airbus term) with the raw data being monitored?

BS

There's a lot of guff on this thread, so I'll take the opportunity to explain in simple terms how you fly an NDB in an A320 (the A330 I understand is identical in this respect - if any 330 guys want to point out any differences then that would be good). Small details have been omitted.

The bus has a small button which flips between HDG/VS and track/FPA (flight path angle). In track/FPA the heading window now displays a track and the VS window displays FPA in degrees. The flight director changes and a little "bird" appears, which is a flight path vector.

You can see that this is a very powerful set of controls. At the FAF you can just select a track that is the same as the approach track, and a descent angle (typically 3 degrees), taking you accurately down the approach. As the wind shifts direction and speed, the aircraft will maintain the track. As you slow down, or the wind changes, you will maintain your FPA. No need to worry about V/S!

Now, you also have a choice whether to use this "pilot selected" method, or let the FM system steer you laterally (NAV) and/or vertically (FINAL APP) - "managed" modes. In this case the airline required/allowed managed lateral and selected vertical, so we shall look at how that is flown.

You would always use the autopilot and autothrust.

Before the approach, you prepare the FMC by checking the approach, missed approach point, MDA (normally +50 on published) missed approach and auto-selected navaids. You tell it that you want Vapp at the FAF, and you put your ND into map mode, with the NDB needle displayed. You plan to fully configure the aircraft just before the FAF, so you fly the final approach fully stabilised. This minimises the workload so you can concentrate on the 2 or 3 important jobs on the way down.

On an intercept heading you would select track/fpa. You would push the heading knob to arm NAV, so that the aircraft will intercept and follow the FMC inbound track. Configure the aircraft, and just before the FAF you preselect your FPA (-2.6 degrees here) and when you want to start down you pull the knob and you will get 2.6 degrees down.

Exactly when you pull the knob is a complex subject and will vary by type, available aids and maybe some more accurate points that you can generate yourself (more accurate than a marker beacon).

You can vary your FPA on the way down to keep in line with your height/distance checks. On this NDB there are none, so you have to just maintain the rate. You cross check the raw data (NDB needle) to make sure the NAV mode is working correctly. In real life, NAV is deadly accurate.

At MDA +50 you decide and either:

• Disconnect the AP and FDs, and land.
• Move the thrust levers all the way forward. This gives you TOGA thrust and the AP will pitch up and fly the go around. You will follow the track you are on when you move the levers forward until you push the heading knob to engage NAV, whereafter you will fly the missed approach in the FMC.

If you descend below the database (or pilot entered) transition altitude on STD, then "STD" is boxed and flashed on the PFD. You can't miss it.

When the radalt becomes alive, you both cross check the reading with your calculated height (by subtracting approximate terrain altitude from your indicated altitude).

Hope this helps.

Some pics from the FCOM (it all looks nicer in real life!):

[GarageYears]

Thank you 100%

Excellent and well written explanation. Now stand back and wait for the clockwork crew to chime in. As for the Luddites, well, 'nuff said.

- GY

[Bluestar51]

100PP,

There's a lot of guff on this thread, so I'll take the opportunity to explain in simple terms how you fly an NDB in an A320 (the A330 I understand is identical in this respect - if any 330 guys want to point out any differences then that would be good). Small details have been omitted.

Thanks for taking the time to explain the A320 magic. It certainly appears to be a very good way to fly an NPA. I'm a little "old school" so I've always prefered "dive and drive" for NPAs as opposed to constant rate of descent. Kinda like I don't do circling approaches where base is downwind.

BS

[HundredPercentPlease]

Bluestar,

Don't forget that because the bus flies a constant angle, the ROD may vary. But it's the angle that matters!

I know of no airline that operates larger machinery on a "dive and drive". It's all about stable approaches now, so you simply "bounce off" MDA or land. Levelling off at MDA is only for light a/c.

[aterpster]

HundredPercentPlease:

Don't forget that because the bus flies a constant angle, the ROD may vary. But it's the angle that matters!

Amen. But, I would add that target KIAS maintained to the flare also matters a whole lot.

[411A]

I know of no airline that operates larger machinery on a "dive and drive".

I do, I fly for one.

Levelling off at MDA is only for light a/c.

Nonsense, ours is a quite heavy widebody three engine jet.

[Fly747]

Dinosaur

Now we all know that you really are a dinosaur 411A if you think that "dive and drive" is the way to do it in a big jet. If not then you should be doing all you can to change your airline's SOPs.

[411A]

If not then you should be doing all you can to change your airline's SOPs.

Negative, for one very important reason....we circle to land on a regular basis, so dive/drive keeps our guys in practise.
It would appear with this accident, that the crew was woefully unprepared for that 'stabilized approach' that seems to be favored so much, these days.

Considering this...dive/drive might actually be safer.
Shock, horror.

[heavy.airbourne]

Let me add, that for serious circling approaches like e.g. SAH, ADD, or even SSG, there is a data base that lets you fly those all managed, which is available for all straight-in NPA, as well: No need to fiddle with FPA, inbd track, V/S, whatever. You must check that the data provided in the database is correct, though, as database providers like LIDO employ too many airheads to be reliable. Something I wanted to point out for a long time; now, i feel better.
The hardest part is to transition from instruments to visual clues, that is where so many accidents happened, and still seem to happen. After a long night of flying (in Africa), it is very difficult to switch to G/A mode. It is easy to say: "If you cannot make sense of what you see, just go around!", but all mental energy to make this seemingly simple decision has been drained from you hour after hour...They call it fatigue, I reckon, and target fixation.
If you ask to much (from your flight crews), you will get too little, eventually. We're only human, after all.