Habsheim

[DozyWannabe]

@Chris Scott:

I don't know for certain, but all the credible sources I've encountered state that flare mode was not a factor in the aircraft's behaviour. The oft-repeated claim that "the aircraft thought it was landing and wouldn't let the pilot pull up" was a complete fabrication of the press*.

Thinking about how the systems were designed to interact, I suspect that the engagement of flare mode would only happen if the descending "ramp" from the RA sensor was consistent over time, as any system responding instantly to transient changes would not be appropriate in an aircraft.

Airbus have changed code many times over the years (in fact the ability to fine tune aircraft qualities without requiring mechanical work was one of the factors in favour of civil airliner FBW), but even by 1988 they had a pretty good handle on this kind of basic behaviour - having tooled their modified A300 around for several years.

* - And also the opening salvo in terms of a lot of people misunderstanding FBW as a concept, by erroneously conflating it with autoflight. Given the usual short shrift accredited to journalists among pilots, it's surprising how long that particular rumour has lasted and how many were willing to believe it.

[rudderrudderrat]

Hi DW,

was a complete fabrication of the press

I agree with you there.
The baulked landing procedure (Select TOGA power and rotate to 10 degs pitch) allows you to fly away from any attempted landing - even after touch down.

Some pilots seem to forget that the only thing that makes an aircraft climb is thrust (when you have no more kinetic energy to swap for height).
Why on earth they were so close to Alpha Max AND Low Height AND Idle Thrust is beyond me.

[DozyWannabe]

Originally Posted by rudderrudderrat

The baulked landing procedure (Select TOGA power and rotate to 10 degs pitch) allows you to fly away from any attempted landing - even after touch down.

Indeed. A bit of background - The very first Software Engineering module lecture I attended in my first year at Uni had my late Prof. showing us the infamous video of the accident - he then took a show of hands to see how many of us thought it was an issue with the computer/software implementation - quite a few went up. He then proceeded to explain why it wasn't what so many of us believed, and provided an overview of how the systems were designed/implemented - finishing up by pointing out that it only takes a small mistake to cause a disaster, and just how much work and effort it takes to keep those mistakes to a minimum.

I've said it before, but I should reiterate that said Prof. was originally a dyed-in-the-wool sceptic when it came to digital FBW in civil aviation, and was well-known on the old Risks list for holding people's feet to the fire!

Some pilots seem to forget that the only thing that makes an aircraft climb is thrust (when you have no more kinetic energy to swap for height).
Why on earth they were so close to Alpha Max AND Low Height AND Idle Thrust is beyond me.

At least some of the answers are found in the lesser-known "Other factors" section of the report:

(Emphasis mine. Please excuse ropey Google translation - it was the best I could do in the time!)

3.2.2 . other factors

- Flight preparation was insufficient, particularly because of the brevity of the record provided late and information on the air show ;
- The distribution of tasks scheduled for overflight the crew was incomplete and has not been complied with;
- The party atmosphere for both passengers and spectators chip transmit the captain ;
-*A 320 has new features that could induce in the mind of the commander overconfidence
- These new characteristics of the aircraft were sometimes attacked and the captain wanted to defend ;
- neither pilot had experience such pre- flight presentations ;
- The late identification of the land has led some precipitation events and the realization of the descent and the entire passage above the track with full engine reduced and without stabilization of flight parameters ;
- Neither pilot had experience overshoot from flight idle in similar conditions.

In short - the airline and crew forgot the old "5 Ps" maxim*, and thus fell victim to the consequences.

* - i.e "Proper Planning Prevents P*ss-poor Performance".

** - @CONF iture - could you please point me to the page/section where you're getting the 2.5deg value from? Thanks.

[toffeez]

3.2.2 . other factors:

The pilot was known to his Air France colleagues as "Rambo".

[vilas]

r r r
"Why on earth they were so close to Alpha Max AND Low Height AND Idle Thrust is beyond me".

One guess is that the crew wanted to keep the aircraft in view for a longer time by taking advantage of high AOA protection to slow it down to minimum speed and maintain it/AOA by adding thrust. The computers adjust alpha max to stay below Vs1g stall, may be that slight adjustment was enough to cause a descent from 100ft to 50ft. When pilot decided to go round from twin evils of idle thrust and max angle of attack the aircraft performance was less than desired in the situation. One thing is certain unplanned manuevres at lower altitude is asking for trouble. Perpignan for example.

[rudderrudderrat]

vilas

The computers adjust alpha max to stay below Vs1g stall, may be that slight adjustment was enough to cause a descent from 100ft to 50ft.

From the link to the accident report I posted yesterday:
"The first officer informed the captain that the aircraft was reaching 100 feet at 14:45:14. The descent continued to 50 feet 8 seconds later and further to 30-35 feet. Go-around power was added at 14:45:35. The A.320 continued and touched trees at the end of the runway at 14:45:40 with a 14 deg. pitch attitude and engine speed being 83% N1. The plane sank slowly into the forest and a fire broke out."

I would suggest the aircraft sank below 100 feet because they didn't apply any power for 21 seconds, and hence they deliberately flew below 100 feet.

Villas, please explain why you think it was the computers that were not able to keep the aircraft airborne for over 26 seconds with idle power from 100ft at a speed close to the stall.

Hi DW,

@CONF iture - could you please point me to the page/section where you're getting the 2.5deg value from?

May be some confusion with 2.5%?
See FCB 17
"TOLERANCE OF FAC COMPUTED SPEEDS
Due to the data accuracy used to compute the characteristic speeds, and specifically the AOA accuracy, the precision of the computation is specified to be within ​2.5 %.
During acceptance flight, the tolerances are as following:

Clean aircraft
Green Dot ±​5 kt
VLS ±​4 kt
Vαprot ±​5 kt
Vαmax ±​5 kt

Conf full
VLS ±​3 kt
Vαprot ±​5 kt
Vαmax ±​5 kt"

[jcjeant]

vilas

One thing is certain unplanned manuevres at lower altitude is asking for trouble. Perpignan for example.

In the case of Perpignan this was a planned manoeuvre but at a wrong altitude
Methink if they had made the planned manoeuvre (it was a test) at a higher altitude the result was to be the same .. splash in water .... they were engaged in a deadly yoyo game

[captainsmiffy]

For DozyWannabes benefit.....Prior Planning Prevents P*ss Poor Performance is actually not the '5 Ps' as advertised but, actually 6!! Hopefully the performance will be better next time with a little prior planning.....?!!!

[vilas]

r r r
I do not have access to English version of the report. From what I read in the posts,I consider the following. The pilot wanted to fly as slow as possible by flying at Valpha Max by holding the stick back. Alpha floor gets activated between Valpha prot and Valpha Max and if it was instantly disconnected before speed increase beyond Valpha prot then it will be
deactivated till the speed has increased beyond Alpha prot before dropping below again. He did not seem aware of this that is why he was surprised by the lack of alpha floor response and moved thrust levers manually. Now he was at alpha Max and idle thrust at 100ft.Considering the response time of the engine and for the speed to recover from back side of the curve descent from 100ft and ground contact seems possible.

[rudderrudderrat]

villas

The French report, page 35:
12h 46' 26'' CDB "Bon Je vais bien la debrayer l'auto-manette." (Google translate = "Good I'm fine to disengage the auto-throttle"). Radio Altimeter shows 40 / 50 / 40 feet.

The Captain is aware the auto thrust is disconnected, yet he doesn't apply TOGA thrust for another 10 seconds with a plane full of passengers. Incredible!

He did not seem aware of this that is why he was surprised by the lack of alpha floor response and moved thrust levers manually.

I think you are right. It is the best explanation so far.

[CONF iture]

Originally Posted by dozy

The oft-repeated claim that "the aircraft thought it was landing and wouldn't let the pilot pull up" was a complete fabrication of the press*.

As was also the claim that : "had it been a 737 no one would have survived but thanks to the Airbus protections, almost all survived"

* - Given the usual short shrift accredited to journalists among pilots, it's surprising how long that particular rumour has lasted and how many were willing to believe it.

Not more surprising than most of you who have read or not the report are still pretending that the airplane was at alpha max.

Originally Posted by BEA

1.16.1.2. Lois de pilotage de l'Airbus A320
Cette loi de pilotage assure en particulier une protection automatique empêchant l'avion d'atteindre une incidence supérieure à 17.5 degrés, pour conserver une marge suffisante par rapport au décrochage, même si le pilote maintient sa demande au plein cabré.

[tdracer]

OK, I'm going to duck for cover after I post this, but....


There were reports that the DFDR and CVR didn't jibe - that there was a 3 second discrepancy. The implication was that the DFDR had been 'altered' to 'remove' a 3 second delay in the engines responding to the GA command . Basically, that there was a cover-up of some shortcoming in the aircraft and blame it on the pilot.


Based on an 'Air Disasters' program about the crash, the pilot is still claiming he was framed, pointing at the 3 second gap.


So, knowing relatively little about this accident (and most of what I do know is based on the press and TV reports), is there anything to the alleged gap in the data?

[jcjeant]

One dedicated site for this accident (in french ..from a journalist who has followed the case since the first minute ... because he was a passenger on that plane)
Crashdehabsheim.net
So it is an analysis among others, and which are often contradictory

[Chris Scott]

BEA Final Report - Extracts

Quote from Conf_iture:
Originally Posted by BEA [see its Habsheim final report of 1990/04/24]
1.16.1.2. Lois de pilotage de l'Airbus A320
Cette loi de pilotage assure en particulier une protection automatique empêchant l'avion d'atteindre une incidence supérieure à 17.5 degrés, pour conserver une marge suffisante par rapport au décrochage, même si le pilote maintient sa demande au plein cabré.

In case it helps anyone, this is my translation into received English:
This flight law provides a special automatic protection preventing the aeroplane from reaching an incidence [angle of attack] greater than 17.5 degrees, to conserve a sufficient margin with respect to the stall, even if the pilot maintains a full climb [pitch-up] demand.

Note that this BEA description does not specify that an alpha of 17.5 deg will be achieved if the pilot maintains full back-stick.

An earlier part of 1.16.1.2 sheds light on the question of whether any engagement of Landing Mode might have at any stage affected the ability of the pilot to achieve Alpha Max. Many of you will prefer to read the original in French, but for those who don't and to save space I'll simply offer this translation:

During all flight-phases prior to the moment the machine reaches a height of 50 feet, measured by the radio altimeter, the flight law is the usual one, called C*.... [there follows a description of the latter] ...
During the phase of descent between 50 and 30 feet, the flight law is modified progressively to take into account (*), instead of the term of load factor [normal G] , a term in pitch (difference between the pitch recorded passing 50 feet, called the reference pitch, and the instantaneous actual pitch);
when passing 30 feet, a supplementary order intended to simulate an effect of conventional landing is introduced to complement [modify] the previous pitch law: this order progressively reduces the the reference pitch (established at 50 feet) to -2 degrees in 8 seconds, creating a dive [pitch-down] moment that the pilot must counteract ("derotation").
At all times, if the angle of attack reaches 14.5 degrees, the flight law is modified and the term of load factor or the term of pitch (whether modified or not by the derotation order) is replaced by a term of incidence [Alpha Prot] (difference between the current angle of attack and 14.5 degrees). This flight law provides a special automatic protection preventing the aeroplane from reaching an angle of attack greater than 17.5 degrees, to conserve a sufficient margin with respect to the stall, even if the pilot maintains a full climb [pitch-up] demand.
This latter flight law, by the name of Alpha Prot (angle-of-attack protection), is a priority mode once the angle of attack reaches 14.5 deg. It does not constitute a degraded mode and cannot be de-activated by the crew.
[The section continues with a description of Alpha Floor.]

(*) This transition is progressive: the change of law is effected in one second.

The reference to Alpha Prot being a priority mode suggests that it would take priority over all others, including Landing mode. The report states that the modes/laws of engagement were specifically recorded by the DFDR as follows, where the time base is in seconds relative to the time of "impact", notated as "t" (see also my previous post). These are listed as follows:
Before [and including?] t - 22, Normal C*;
between t - 21 and t - 20, in transition to landing mode;
between t - 20 and t - 18, progressive return to C*;
between t - 18 and t - 04, landing mode (a reference pitch of +6 deg having been measured passing 50 ft);
at t - 4, switch to Alpha Prot, the angle of attack having exceeded 14.5 degrees; this law maintained thereafter.

Confit and others,
Please let me know soonest of any anomalies in my translations...

[balsa model]

Originally Posted by CONF iture

...pretending that the airplane was at alpha max

Please excuse (or ignore) my intrusion into this discussion - I'm the least qualified to comment here, but I do have a question:
Would the extra couple of degrees of AoA, that are presumed to be held back by the s/w (is that your point, CONF iture?), have changed anything for the better in the outcome?
I understand that the lift coef curve is mostly flat near stall as it is. In addition, IF the extra increment in lift extended the flight long enough for the engines to spool up, wouldn't the go-around thrust-pitch couple just push the airplane into a stall? The crash would then be tail-first, perhaps.

[awblain]

If the angle of attack had been greater, and the slowing with the same power had thus been more alarming, then perhaps the power would have been added earlier, and the trees would have been avoided. The trees would also have been further away in time, allowing the engines to spool up more before they turned into chippers.

Crashing on the airfield might also have been more benign than crashing in the forest. There'd probably have been less chance of smashing open all the fuel tanks, and it would have been easier for whatever rescuers were available to get to the wreckage.

As it was, it was close. Perhaps with three fewer feet of treetops in the engines… it might have flown away. The aircraft wasn't dragged to a dead stop - it was moving through the forest for five seconds before it stopped and burned.

[BOAC]

balsa - we did this topic to death on a previous Habsheim thread, but the view is that had the a/c been a 'traditional' non-fbw type then the trees might well have been cleared with the extra degree or two of 'alpha', but conversely if the crew had over-rotated in their anxiety they could well have stalled and possibly dropped a wing which would almost certainly have caused far greater loss of life. The jury, as they say, remains out.

[Chris Scott]

Given that they were where they were...

Hi balsa model and awblain,

Reading your posts, a couple of points spring to mind.

(1) Pitch rose through +10 deg at about t -13. No particular significance in +10, but my guess is that, assuming the pilot's eye height was adjusted (using the seat-height adjustment) to the recommended level, the trees would not have been visible through the windshields after that. Also, not many pilots are used to the perspective at that kind of attitude - except on T/O. At t -7 pitch was about +14. Unfortunately I don't have an A320 AWOPS guide to hand, which might help with the relevant geometry.

At about t -8 the co-pilot warned the captain of some pylons ahead. They were very much further away than the trees, so one can assume that either the co-pilot could not see the trees, or that the steep deck-angle caused him to think that the a/c was higher than the treetops.

(2) Some readers may be surprised to learn that at t = 0 (defined as impact) the stick was still not fully back. IIRC, full deflection is 20 degrees. The last recorded stick position was -17 deg (17 deg nose-up command), rising rapidly from -5 deg at about t -3. What happened after that is not recorded, because the power supplies to both the DFDR and CVR were simultaneously cut (by severing of shared conduit).

Between t -9 and t -4, the stick command was between -6 deg and -7 deg, but this was relaxed slightly at t -4, falling to -5 deg.

ALPHA PROTECTION

As mentioned in the BEA report (see previous posts), the flight-control law changes from Normal Law to Alpha Protection (in effect a sub-law of Normal Law) if the angle of attack rises to an arbitrary figure known as alpha-prot. Alpha Prot "law" tries to maintain that angle of attack if the stick is neutral. A higher angle of attack, known as alpha-max, can be achieved with sustained, full back-stick. Alpha-max provides a small margin from the stall (the latter usually defined as alpha-CL-max), and cannot be exceeded. Intermediate angles of attack are commanded acording to the amount of back-stick applied by the pilot.

The BEA report states that, in their configuration (Flaps 3 and L/G):
alpha-prot was 14.5 deg;
alpha-max was 17.5 deg.

At t = 0 ("impact") both pitch and alpha were about +15. The latter represents the apparent discrepancy of about 2.5 deg that Conf_iture has called into question. As stated above, alpha-max can only be achieved by sustained, full back-stick.

[awblain]

The trees being out of sight under the nose is interesting, which might have contributed to a lack of urgency - although the approach of the line of the trees should have been perceptible in peripheral vision, even if they were below the sill directly ahead. From the video, the whole of the flyby appears to have been done at a substantial deck angle, so any visual impression of height throughout the whole pass would have had to have been from the side. It's interesting that the first officer was able to see the pylons 8-s from impact though - distant 30-m-high pylons (which look like they're 3km away on Google Earth) - and nearby treetops are not likely to be at very different elevation angles from 8-s (~350m?) away from the trees. Looking forward, it would be trees all the way from the base of the pylons to the sill, and with the ground whizzing by at 40 feet or so to the side and the tree line approaching, it seems to be no place for a confidence-building visual illusion.

Isn't it all navel-gazing anyway? - the plan for the show presumably didn't have any tree surgery in it, and the arrival at the treetops was presumably at the end of a substantial deviation from the intended plan.

Prompt cutting of the power to the recorders implies a substantial impact - rather more than from twigs hitting the rear fuselage - perhaps a few extra feet wouldn't have made any difference. With more angle of attack, the rear might have been thumped into that conduit-slicer even harder (3 degrees being about a meter at the tail from the center of mass), helping to pitch back down and further into trouble.

Would you agree that the prime human factor is that no-one would ever have dreamed of trying this in a non-FBW airliner? Perhaps the same sort of overconfidence factor that has been claimed to account for the accident statistics when ABS systems were first fitted as options to cars, and the serious accident rate was higher in the otherwise identical cars with the ABS than those without.

[RoyHudd]

I fly Airbus types, inc. 320, and I used to fly out of Habsheim in a previous job. (Not A320's) It is a small airfield surrounded by forestation on 3 sides.

The CM1 was taking an enormous risk in flying too low and slow in an aircraft he clearly did not understand well. (Parallels with the Costa Concordia accident exist in Human Factors terms)

5 dead are on this plonker's conscience. Forget nationality, or background. On the day, he showed off, screwed up, and killed people.