https://www.businesslive.co.za/bd/natio ... -take-off/ (paywall)
https://simpleflying.com/dangerous-saa- ... stigation/ (free)

This seems like a mess. A cargo flight which was delayed due to concerns over the pilots' recent flight hours was finally allowed to go through. On takeoff, they nearly stalled it, and would have probably crashed had an autopilot fail-safe not taken control. This happened nearly a month ago but regulators have only recently been made aware of it.

http://avherald.com/h?article=4e4d2036&opt=0

same like
https://en.wikipedia.org/wiki/Emirates_Flight_407
?

The flight departed with 14 regulatory exemptions issued by the SACAA

'According to information The Aviation Herald received the flight crew created a manual flight plan for the trip to Brussels. It is being claimed that there was a confusion on the takeoff mass computation being about 90 tons below the actual takeoff mass'.

Surely the pilots should have some idea of aircraft weight. That is 20% of the difference between OEW and MTOW

https://en.wikipedia.org/wiki/Airbus_A3 ... ifications

'However, there are also reports of a well known software bug in the FMS, which reduces the takeoff mass by about 90 tons on the FMS (without affecting the manually entered V speeds) if the takeoff mass is entered before engine start. This could result in incorrect flaps monitoring speeds and such incorrect protection speeds while not affecting the actual takeoff performance'.

If it is that well known, how come they got it wrong.

'According to an A346 pilot, The Aviation Herald spoke to, this fault was actually personally observed by the pilot and caught by stringent standard operating procedures preparing for departure, which do require to compare the FMS takeoff mass again after engine start. This software issue is well known and the pilot had been informed about the bug some time prior to his observation'

If the pilot spotted this before take off, what action did he take.

'According to information The Aviation Herald received, the crew denied the occurrence and even stated: " The aircraft lied!". The crew also refused to participate in a meeting on Mar 16th 2021 intended to discuss the occurrence'.

Why are the crew refusing to participate. Do they believe that this will exonerate them. More likely to let the company of the hook and simply blame them. IMO, this is more likely to lead to dismissal..

SpectralK wrote:

On takeoff, they nearly stalled it, and would have probably crashed had an autopilot fail-safe not taken control.

It sounds to me like there is confusion regarding whether envelope protections may have activated erroneously and unnecessarily, possibly related to the FMS not understanding what the airplane's weight was, possibly related to some weird FMS software issue that has existed forever uncorrected.

bennett123 wrote:

'However, there are also reports of a well known software bug in the FMS, which reduces the takeoff mass by about 90 tons on the FMS (without affecting the manually entered V speeds) if the takeoff mass is entered before engine start. This could result in incorrect flaps monitoring speeds and such incorrect protection speeds while not affecting the actual takeoffte them. More likely to let the company of the hook and simply blame them. IMO, this is more likely to lead to dismissal..

A little off topic, but could this software fault have been a contributing factor in the EK407 A345 (A6-ERG) incident at MEL. I flew this aircraft after the accident (CPT-DXB) and have always been intrigued by that very near miss.

Sandyb123

bennett123 wrote:

IMO, this is more likely to lead to dismissal..

Hardly likely, the crew were part of SAA's new "mission" on flight crew and according to my information, among the most experienced which is very worrying.

The part about the bug is interesting and I'll await more info.

With that said, is it common to retract slats and flaps so soon after take-off ? Did they do that because they were light ?

IF the FMS activated envelope protections because it thought the aircraft was heavier than it actually was, the pilots are off the hook.

IF.

The plane is intact and the black boxes are available. We will know soon what really transpired. Until then, it might be premature to erect the gallows.

bennett123 wrote:

'According to an A346 pilot, The Aviation Herald spoke to, this fault was actually personally observed by the pilot and caught by stringent standard operating procedures preparing for departure, which do require to compare the FMS takeoff mass again after engine start. This software issue is well known and the pilot had been informed about the bug some time prior to his observation'

If the pilot spotted this before take off, what action did he take.

Pretty simple: the A346 pilot, I spoke to, put the correct takeoff mass back into the FMS when they caught the bug on the cross check after engine start.

My understanding of the occurrence, summarizing all the facts as currently known, is this:

There was no tail strike and no intervention by the FBW to reduce the possibility of a tail strike (the system actually can not prevent tail strike even though the system can reduce the elevator deflection), it is obvious therefore, that they used the correct V speeds for rotation and initial climb out. This is also the conclusion that the A346 pilot came to when we discussed the event, he was sure that had they used wrong V-speeds they'd not have been able to avoid a tail strike.

However, due to the software bug the FMS computes and displays a wrong flaps schedule possibly causing premature flaps retraction at too low speeds, which in turn can result in high angles of attack and Alpha Floor Activation.

I agree with you regarding being puzzled why the crew refused to participate in that meeting to discuss the occurrence. However, I can, to some extent, understand their emotion that the aircraft lied on them.

The checklist and cross check after engine start should have caught that bug. As far as I understand this cross check became mandatory as mitigation of that bug (and therefore it was not seen as a necessity to address the bug with any kind of urgency).

Hence in my personal opinion: if the bug occurred - after they had entered the correct takeoff mass and correct V-Speeds into the FMS initially - and reduced the takeoff mass by about 90 tons during engine start and they missed this change in the takeoff mass in the after engine start checklists, then they would feel indeed, that the aircraft lied on them and produced an invalid Alpha Floor protection. However, if they followed the flaps schedule displayed on the PFD and began to retract the flaps at too low speeds, they would have encountered high AoA and proper Alpha Floor Activation. So overall, my personal opinion is, that the scenario as we currently know it, is entirely consistent with that software bug and non-detection of the bug.

I do not think, that the occurrence can be compared with the EK-407 in Melbourne: that was an overrun of the runway because the aircraft did not unstick followed by a tail strike when they forced the aircraft into the air. This one was a computational error resulting in a computed takeoff mass 100 tons less than their actual takeoff mass as well as in wrong V speeds based on that 100 tons less takeoff mass than they actually were taking off with.

There are a lot of other noises in the background of SA-4272 about who the pilots were, the conflict about contracts with the pilots, the race etc. I don't think, that plays any role in assessing this occurrence and getting to the bottom of it. Such arguments are just distractions. What may become important however in establishing the root causes of the occurrence might be the statement by South Africa's CAA initially denying permission for the flight claiming the crew was not current. The question therefore stands big in the room: were they really current after the additional training received between Feb 18th and 24th 2021?

Now, is the noise abatement violation on subsequent departure from Brussels perhaps a consequence of the departure out of JNB? Did the crew perhaps not realize what went wrong in JNB and tried to avoid the issue by just accelerating the aircraft as quickly as they could after departure?

So in short from my personal perspective based on what we currently know about the occurrence, this serious incident most likely was the result of an all too human blunder by overlooking the cross check of the FMS takeoff mass after engine start as well as the mere (continued) existence of that bug.

Servus, Simon (yup, the one of The Aviation Herald)

AustrianSimon wrote:

then they would feel indeed, that the aircraft lied on them and produced an invalid Alpha Floor protection.

@Simon thanks for taking the time to post to Anet and thanks for your excellent work reporting these incidents.

A possible invalid activation of envelope protections intrigues me. Does anyone know how often this might occur? I don't think I've ever heard of an instance before.

AustrianSimon wrote:

So in short from my personal perspective based on what we currently know about the occurrence, this serious incident most likely was the result of an all too human blunder by overlooking the cross check of the FMS takeoff mass after engine start as well as the mere (continued) existence of that bug.

Servus, Simon (yup, the one of The Aviation Herald)

I find the occurrence interesting as they have been operating the type for close to 20 years.

AustrianSimon wrote:

bennett123 wrote:

'According to an A346 pilot, The Aviation Herald spoke to, this fault was actually personally observed by the pilot and caught by stringent standard operating procedures preparing for departure, which do require to compare the FMS takeoff mass again after engine start. This software issue is well known and the pilot had been informed about the bug some time prior to his observation'

If the pilot spotted this before take off, what action did he take.

Pretty simple: the A346 pilot, I spoke to, put the correct takeoff mass back into the FMS when they caught the bug on the cross check after engine start.

My understanding of the occurrence, summarizing all the facts as currently known, is this:

There was no tail strike and no intervention by the FBW to reduce the possibility of a tail strike (the system actually can not prevent tail strike even though the system can reduce the elevator deflection), it is obvious therefore, that they used the correct V speeds for rotation and initial climb out. This is also the conclusion that the A346 pilot came to when we discussed the event, he was sure that had they used wrong V-speeds they'd not have been able to avoid a tail strike.

However, due to the software bug the FMS computes and displays a wrong flaps schedule possibly causing premature flaps retraction at too low speeds, which in turn can result in high angles of attack and Alpha Floor Activation.

I agree with you regarding being puzzled why the crew refused to participate in that meeting to discuss the occurrence. However, I can, to some extent, understand their emotion that the aircraft lied on them.

The checklist and cross check after engine start should have caught that bug. As far as I understand this cross check became mandatory as mitigation of that bug (and therefore it was not seen as a necessity to address the bug with any kind of urgency).

Hence in my personal opinion: if the bug occurred - after they had entered the correct takeoff mass and correct V-Speeds into the FMS initially - and reduced the takeoff mass by about 90 tons during engine start and they missed this change in the takeoff mass in the after engine start checklists, then they would feel indeed, that the aircraft lied on them and produced an invalid Alpha Floor protection. However, if they followed the flaps schedule displayed on the PFD and began to retract the flaps at too low speeds, they would have encountered high AoA and proper Alpha Floor Activation. So overall, my personal opinion is, that the scenario as we currently know it, is entirely consistent with that software bug and non-detection of the bug.

I do not think, that the occurrence can be compared with the EK-407 in Melbourne: that was an overrun of the runway because the aircraft did not unstick followed by a tail strike when they forced the aircraft into the air. This one was a computational error resulting in a computed takeoff mass 100 tons less than their actual takeoff mass as well as in wrong V speeds based on that 100 tons less takeoff mass than they actually were taking off with.

There are a lot of other noises in the background of SA-4272 about who the pilots were, the conflict about contracts with the pilots, the race etc. I don't think, that plays any role in assessing this occurrence and getting to the bottom of it. Such arguments are just distractions. What may become important however in establishing the root causes of the occurrence might be the statement by South Africa's CAA initially denying permission for the flight claiming the crew was not current. The question therefore stands big in the room: were they really current after the additional training received between Feb 18th and 24th 2021?

Now, is the noise abatement violation on subsequent departure from Brussels perhaps a consequence of the departure out of JNB? Did the crew perhaps not realize what went wrong in JNB and tried to avoid the issue by just accelerating the aircraft as quickly as they could after departure?

So in short from my personal perspective based on what we currently know about the occurrence, this serious incident most likely was the result of an all too human blunder by overlooking the cross check of the FMS takeoff mass after engine start as well as the mere (continued) existence of that bug.

Servus, Simon (yup, the one of The Aviation Herald)

IMO, they would be better served by being in the room and fighting their corner than letting SAA make it's decision without their input.

This well known bug - is is documented anywhere?

No bug is involved in the EK407 event (A340-500), see the ATSB final accident report. It is a mistake of the FO entering 2xx instead of 3xx on the EFB, subsequently undetected / uncorrected. Amzingly this is the same mistake the the AF 777 event in CDG.

Human factor studies show that when typing a 3 digit figure, we tend to put more attention on the right hand digit (the unit) rather than on the left one (the hundred). Silly.

If launched (there is no damage), the Annex 13 investigation will tell if anything happened in the SAA flight, and why. The case is now political in South Africa, good luck.

Someone need to fix the spelling mistake on the thread's title....Sorry, as a perfectionist, I can't handle that

hivue wrote:

A possible invalid activation of envelope protections intrigues me. Does anyone know how often this might occur? I don't think I've ever heard of an instance before.

When talking about low speed protections (as we are here) weights entered into the FMS will have no effect.

V alpha Prot and V alpha Max are generated by AOA. That is why they change during wing loading, angle of bank for example.

Alpha floor is activated by exceeding a safe angle of attack.

While incorrect weights and thus flap (retract and extend) speeds are predicated on what is in the FMS, the activation of Alpha Floor is not. So retracting the flaps too early because the retract speed was incorrect may, and in this case sounds like it did, cause a higher AOA which will activate Alpha Floor.

In thousands of hours on the A320 series and the A330, I have only seen it activated once. ATC had us slow to Approach speed, then S turned us as we were vectored in too close to the preceding traffic. In the lee of the Rocky Mountains it was very very bumpy, at V App in a 30° banked attitude, Alpha Floor was activated during one of the bumps.

Good thing I didn’t try to pretend it didn’t happen like these gentlemen. While I filed a safety report that evening ... I got several calls from various departments, all who received automatic messages.

bennett123 wrote:

'According to information The Aviation Herald received the flight crew created a manual flight plan for the trip to Brussels. It is being claimed that there was a confusion on the takeoff mass computation being about 90 tons below the actual takeoff mass'.

Surely the pilots should have some idea of aircraft weight. That is 20% of the difference between OEW and MTOW

https://en.wikipedia.org/wiki/Airbus_A3 ... ifications

380 t MTOW - 174 t OEW = 206 t difference
90 t / 206 t = 0,44

How does one make a 90 t mistake? Is Marijuana legal in South Africa?

Sokes wrote:

bennett123 wrote:

'According to information The Aviation Herald received the flight crew created a manual flight plan for the trip to Brussels. It is being claimed that there was a confusion on the takeoff mass computation being about 90 tons below the actual takeoff mass'.

Surely the pilots should have some idea of aircraft weight. That is 20% of the difference between OEW and MTOW

https://en.wikipedia.org/wiki/Airbus_A3 ... ifications

380 t MTOW - 174 t OEW = 206 t difference
90 t / 206 t = 0,44

How does one make a 90 t mistake? Is Marijuana legal in South Africa?

It’s probably pretty easy, not checking thoroughly. or SQ286 a 744 ex AKL in 2003 suffered a severe tail strike, crew entered a weight 100t less than the actual weight.

Glad there weren’t more serious consequences in either incident.

https://aviation-safety.net/database/re ... 20030312-0

hivue wrote:

A possible invalid activation of envelope protections intrigues me. Does anyone know how often this might occur? I don't think I've ever heard of an instance before.

There was an OEB about the incorrect activation of the AOA protection. The reason for the publication of this OEB were two events of incorrect AOA protection. One was involving an A330 one an Lufthansa A321. I don't know the details about the A330 event, but on the A321 event the aircraft was parked overnight in rainy conditions. Apparently the seals of the AOA sensors were not fully tight, which led to water ingress into the sensors. When the temperature dropped below freezing during climb to cruise level the water froze. Due to the increasing mach number and other factors the AOA required for AOA protection activation increases during climb. The protection activated and could not be overcome by the crew by normal control inputs. In consultation with the maintenance department, the aircraft was put into alternate law, which deactivated the faulty protection and restored full control.
Since this event improvements were made to the AOA sensors and the software of the flight control computers was improved to enable a better recognition of a similar event by the flight control computers. For this reason the OEB was removed from the manuals.
The AOA protection is independent from data entries by the crew and uses the AOA sensors. Therefore it's highly unlikely to get an incorrect AOA protection, as shown by the fact that it happened only two times in the entry history of Airbus aircraft.

Well there was a crash also, XL Airways Germany Flight 888T. Not exactly the same thing but related (the cause being mainly pilots playing with fire).

I thought that SAA was currently basically-shut-down due to failure of its turnaround plan, then exacerbated by Covid, with the majority of its fleet returned to lessors.

wjcandee wrote:

I thought that SAA was currently basically-shut-down due to failure of its turnaround plan, then exacerbated by Covid, with the majority of its fleet returned to lessors.

The remaining fleet including the A340-600 is owned, I believe. Though the pilots lacked recent experience with the model, which likely played a role in this incident.

AustrianSimon wrote:

Now, is the noise abatement violation on subsequent departure from Brussels perhaps a consequence of the departure out of JNB? Did the crew perhaps not realize what went wrong in JNB and tried to avoid the issue by just accelerating the aircraft as quickly as they could after departure?

Call me conspiratorial....but maybe the crew trying to get the airline a big fine for noise infringements? Probably not out of the question for a crew who may have a "chip on their shoulder" with their management, knows they're likely to be fired anyway and knows that even if they're not fired now, their airline likely will be wound up completely in the coming months. Maybe I'm cynical but crazier things have happened.

armagnac2010 wrote:

Amzingly this is the same mistake the the AF 777 event in CDG.

When was this?

When was this?

May 2015, both crew made the same mistake on the hundred digit, 100 tonnes difference, slow acceleration during take off, TOGA selected, no damage.

What is also noticeable in the 777 case is that the FMS detected something was out of range, but did not raise any flag or warning.

Full report on the BEA web site:

https://www.bea.aero/fileadmin/uploads/tx_elydbrapports/BEA2015-0225.pdf

zkojq wrote:

AustrianSimon wrote:

Now, is the noise abatement violation on subsequent departure from Brussels perhaps a consequence of the departure out of JNB? Did the crew perhaps not realize what went wrong in JNB and tried to avoid the issue by just accelerating the aircraft as quickly as they could after departure?

Call me conspiratorial....but maybe the crew trying to get the airline a big fine for noise infringements? Probably not out of the question for a crew who may have a "chip on their shoulder" with their management, knows they're likely to be fired anyway and knows that even if they're not fired now, their airline likely will be wound up completely in the coming months. Maybe I'm cynical but crazier things have happened.

armagnac2010 wrote:

Amzingly this is the same mistake the the AF 777 event in CDG.

When was this?

Apart from anything else, they may want to fly with another carrier post SAA.

If this was proven, who would touch them?.