Spurious activation of Reverse Thrust
 

It looks like a spurious signal from the EEC drove the left Thrust Lever to idle to allow for an inadvertent deployment of Reverse Thrust (ala Lauda Air over Thailand). As Auto Thrust was engaged, the right engine will drive forward or remain in position to compensate for the loss of thrust from the left engine. It is this this assymetric condition that brought about the uncommanded bank resulting in the Auto Pilot tripping.

Teddy Robinson

I don't know where that summary came from but it is wrong in several important respects. Instead of re-quoting misinformation we should (and can) quote directly from the report.


There were three attempts to fix the autothrottle.

There was one occurrence of thrust lever decreasing, which continued for over a minute without apparent correction.

Except this crew apparently didn't recognise a thrust lever stagger, a big N1 difference, and the yokes a long way out of horizontal.

Knowing that their auto-thrust had been faulty, PF must have had their hand on the thrust levers, one would hope? Something else must have been going on, hence the CVR is needed.

Obviously we should regularly look at our FMA and N1/ EPR, (and PFD of course), on any aircraft, because they tell us the result of our inputs. The moving thrust levers and yoke movements are the inputs, not the outputs. But you don't need to be glued to them - just part of your normal scan which includes looking out of the window. I have never had to watch Airbus FBW "like a hawk" - it works very well and reliably. I've certainly never felt the need to hand-fly every departure up to cruising level before putting the automatics in, as someone suggested.

But we are getting off piste here.

I mentioned before on this thread about the possibility of a single thrust lever reducing/not advancing and that is now confirmed by the preliminary report. It's actually unbelievable that it went unnoticed for over a minute, but still very possible knowing the training/standards in indo.

I think we would all like to know that.

Derjodel in his #552 post clearly states the issue of the era. This is a time of transition in the airline industry, as too much of the work required to maintain good flying skills has been automated. Unlike earlier days, the pilot’s essential tactile and cognitive connection with the airplane when several things are happening at once is nowhere near what it needs to be when things go wrong. No amount of training or improved simulation can make up for this, and further automation can only make it worse. You cannot make better pilots in airplanes designed to increasingly diminish the need for them.
Another issue also mentioned before is the covid crisis that does not end with the last of the pandemic. A complex but smoothly functioning airline industry has been completely derailed and the problem confronting the flight crews now afflicts the entire industry. Benched for too long, the need for rusty but essential skills will be greatest when they are anything but sharp.

This can and does happen anywhere in the world, as standards are incredibly low for manual flying on instruments, let alone upset recovery training which - if it exists at your company - is probably a joke. Having it beaten into their heads to use automation as much as possible or giving into that habit by themselves, crews tend to respond to abnormal situations by trying to diagnose the problem with the automation instead of disconnecting AP & AT and flying out of it.

At this point, I'm afraid the CVR won't have much to reveal. The crew was so caught up with figuring out what was happening with the autothrottle that they allowed the situation to progress to the point where the AP couldn't hold off the roll anymore and it left them in a steep dive to the left, as the malfunctioning autothrottle increased the thrust asymmetry for a few more seconds before disconnecting.

To their credit, I think they were pulling out of the dive when they hit the water, they just weren't aggressive enough on the recovery. fdr was putting up some scenarios earlier on what might have happened. 10900' should be enough for any 737 to recover from a 1-2 turn spin or rolling inverted. In this case, I think the roll at AP disconnect just rolled them to 80-90 degrees of bank and the nose sliced on down as they were wondering what the hell was happening. Perhaps they didn't trust their instruments at first, but they would have felt the initial acceleration downhill, and as the aircraft picked up speed they would have heard it too. If they were pulling up straight at the end, they probably addressed the thrust asymmetry, but I don't see any of those details in the last 20 seconds of FDR recording.

To the incredulous, this is definitely not the first time this happens. This is basically a copycat of the TAROM A310 at Otopeni in the 90s, where the First Officer allowed the malfunctioning AT to retard one lever to idle and push the other one to TO thrust, resulting in a wingover in IMC (also to the left, if I'm remembering it right) and a basically vertical dive into the ground. In a sinister coincidence, the Captain was having a heart attack or seizure while this was happening.

The industry are treating these horrendous accidents as statistical blips, but the reality is that some sort of IMC mess-up resulting in a steep dive into the ground happens every year or two, sometimes more than one in the same year. It usually involves a 737, as there is a lot of them out there, there are much more short & medium range flights than long haul, and the automation is pretty crude compared to something like the A320. One would think that this would trigger some alarm bells and that someone would recognize that there is a need for crews who can confidently handfly these aircraft in IMC and recover from severe upsets on instruments.

Btw, handling-related question: for those who got the chance to do the rudder hardover on the 737 sim, did unloading & reducing angle of attack give you enough aileron authority to arrest the roll? How did you fly afterwards? Did you need to offset the thrust to compensate for the yaw from the stuck rudder?

See posts #89 and #147.

The Preliminary Report is silent regarding the final 20 seconds of the flight.

there you have it. more aircraft flights of aging technology = more cock ups
so do we need to scrap out old fleets or below avg pilots?

I used the descriptor of below avg pilots because of what I read on here most seemed to know what to do

Incorrect summary posted again and again and again... This thread is moderated and so many posts are being deleted. Why aren't posts like these deleted, which simply copy & paste false information all over again? Don't just copy stuff from the internet - read the actual report, guys! The report does not report any A/T failure during takeoff. It reports a normal takeoff.

Thrust reverser is a mechanical mechanism, if the translation sleeve moves the warning light illuminates and crew action would be to shut the engine down. The auto throttle system will only command the climb power selected by LVL Change or VNAV, if the crew had selected VS mode in the last 1000 feet, the thrust adjusts according to the MCP speed.

Spurious activation of Reverse Thrust
 

Well the spurious part is right.

EEC ? 737-500 ? Hmmmm.

It looks like an entirely evidence-free proposition.

There is no EEC on the 737 Classic, it has a hydromechanical fuel control system.

The thrust reduction on the left engine appears to be very gradual. Should reverser deploy in-flight on the classic 737, the thrust lever is rapidly pulled back to idle by the restow mechanism to prevent thrust in excess of idle. This would have definitely been noticed by the crew.

DaveReidUK

No EEC, purely mechanical T/R thrust pullback, not to mention the EEC on Lauda had zilch to do with the T/R deployment.

So, great post aside from everything in the post being wrong :ugh:

Quite the contrary - the crew failed to address a straight forward problem - one that had reportedly occurred on previous flight. ALL they needed to do was take manual control of the thrust levers to save the situation. The question is why they didn't take the relatively simple steps necessary to save themselves (along with the aircraft and passengers).
Only the CVR can possibly reveal that answer.

FlyingStone

That is my recollection of the FCU on the -3 engine as well. There is mechanical feedback to the throttle to ensure the lever is and remains at idle as the FCU schedules into reverse. That would suggest that a TR problem will give a rapid and complete closure of the thrust lever to idle. If it is working anywhere near correctly.. If the cable itself has a failure (other types I fly have repetitive inspections of the TR follow-up cables due to failures), a failure in the cable itself could be an issue. Cannot recall if the TR cable acts under tension to close or not, which could discount that as a failure mode. The other issue that can cause both thrust levers to reduce is a TAT probe failure, which signals for limit cases to the AT system calculation, or of the PT2 on an engine going bad, as that has a limiter input to the FCU, but that should not move the thrust lever. The main memory of the FCU was it was quite expensive to buy the replacement.

The FCU was properly known as a MEC and had the PMC as a trimming system which was a form of limited EEC/DEEC. The MEC took the PS12 and T2 for the governing system input along with PLA. The only other items that it did was to give flight idle, variable bleed valves, VBV and the variable stator valves, VSVs. The latter was IIRC fluidic control system but with mechanical feedback to the MEC. Expensive when they went bad. The PMC trimmed fuel electrically and didn't drive the PLA. Hazy.

TD Racer On the CVR, I'm not sure it will reveal anything of great note in this case. That the crew didn't correct the TL split is indicated in the FDR data reportedly. There are not too many logical reasons why a crew member on seeing a split would not remove the split, yet that appears not to have happened. The highest likelihood is the crew were not aware of the split through distraction with other tasks, and that can be assessed in a simulator with a cohort of drivers as to the likelihood of detection. To get in before the objections, there is a history of extended time to detect a condition or a complete failure to detect in our sorry past. Take a B747 that has a gear red light for 2 minutes and lands with no nosewheel... and that was a "top-of-the-line" IP. Expectations of immediate and correct crew responses is akin to "they said artificial sweeteners were safe, and WMD's were in Iraq, and Anna Nicole married for love!". If it's recovered, great, but the event shows that the crew likely never recognized the condition, otherwise they would have got to Ponkianak for their Kopi Luwak.

QDM360

I have quoted what is claimed to be a translation of the report. Nothing more, nothing less.

I think it has more to do with the degradation of piloting skills all over the world. This has been the result of airlines, manufacturers and regulators facilitating a rapid increase in world wide airline travel without the impediment of ensuring a high standard of training. You can blame it on the nationality and culture of the individual pilots but how do you explain the Turkish Airlines accident in Amsterdam and AF447? Certainly Indonesia seems to have more of the headlines but I think it is a symptom of the world wide problem not just because it was a B team crew at the controls.

Fdr
Not aware of the TL split...not aware of increasing yaw to the left? Assuming inputs to requested heading change accounted for the yaw? We need that CVR to try to determine what human conditions may have led to such a very early lack of attention (or inability to respond) by PF to clear warning signals. I’d also be interested to have been a fly on the wall during the previous 12|24 hours to know if anything about the preflight period may have impacted on crew, especially PF. These are things that can never be discovered in a report unfortunately, but they do fan the flames of “more automation” calls.

tdracer

I forgot to mention that on the TAROM accident flight, the FO kept trying to re-engage the autopilot as the aircraft rolled over on its side. Forget about disconnecting the autothrottle, forget about flying manually.
If people are wary to handfly on a normal day, how can they be expected to suddenly become confident in a crisis? In weird situations, many people revert to instinct. If the instincts are wrong (automatics are going to save the day), because the training was all wrong (automatics do it so much better than you, don't even try), accidents will happen. I'm only bringing up the TAROM flight, because it was exactly the same type of upset caused by the exact same thing. But there are many examples where crews have allowed automation to kill them, because they either lacked the confidence to disconnect or because they didn't even notice what was happening.

So I wouldn't find it so surprising if this particular crew was actually aware of the autothrottle problem and kept trying to figure it out, instead of disconnecting it, until they found themselves rolled over and in a dive. In the context of what's been happening over the years, it's really not that strange a scenario. There are people who would literally rather die than take manual control of their aircraft outside of takeoff and landing.

fdr, if the mechanical T/R thrust reduction comes into play, it rapidly moves the throttle to idle (less than a second - known as a 'throttle snatcher'). It's rapid enough that you don't want your hand in the way. It also makes a very audible noise when the lever hits the stop. Oh, and it would also have been obvious on the FDR. All in all, the T/R is a non-sequitur in this discussion.

I maintain it's going to very difficult to understand why the crew failed to notice or react to the throttle split without the CVR. For example, the key to the similar TAROM crash was the the left seat pilot suffered some sort of medical emergency just as the uncommanded throttle movement occurred. It was the distraction of whatever happened to the left seat pilot that distracted the PF enough that he let the situation get out of hand. A simulator will never give that sort of insight into the why...

No argument, If anything, that would have ensured some detection by the crew of the problem.


tdracer

It's always interesting to go over a CVR, at least for the spectral data, and the level of confusion that arises. IOn this case, however, much like CI140 or even GF172, or Flash 604, what happened leads to a relatively small number of causations. Once instrumentation error such as AI855, KAL 8509, etc has been discounted, and where a flight control has not been compromised (maybe arguably XL888T... through the flight test procedure followed by the crew) possibly still Flash 604 [according to the official report... according to the flight data, nope, it was not a technical defect, full flight control authority existed in the "recovery" which report disregards], thereafter, UAs result from a loss of SA. There are myriad manners to lose SA, and we are all susceptible to loss of SA, generally being human, and we have inertia in our processes that make it time-consuming to recover SA if we actually do recognize a loss.

I do hope they find the bits of the CVR, but I suspect that the insight it provides will be confirmatory rather than a revellation. Would prefer that the industry not undertake canned UA training, that it be done in conjunction with normal flight profile training, to address the deficiencies we have in SA in our community. To that end, if TM's or CP's want to PM me, I am happy to propose various scenarios that can be introduced into a normal flight to give real-world SA training value to the crews. Patently, UA training as done today doesn't do a great job., in fact, paradoxically the current training may give some adverse levels of confidence to the system, "...we have ticked that box so it's done and dusted, let's get on with the next agenda item..." The MAY in that is just a curiosity, a study into the efficacy of existing UA training would be hard to quantify.

Within the community, we have a problem across all countries, languages, and religions, that crews don't pick up the cues of anomalies developing. dumping a 777 on its belly, an engine fire in reverse following a FOHE failure... 263 kt approaches (not an F-104) parking in the water short of the runway, pulling the att system platforms in IMC, basically all of the stuff we do frequently and apparently irreducibly. The G/As that end up with some 55-60 deg nose up, at low speed and a prompt youtube image thereafter. We spend our time in the sim ticking matrix boxes on instrument approaches we do every day and fail to train on SA loss, recognition, and recovery. One can argue that the engineering and certification of the MAX was an SA failure, it was just in the world of MEDA, and DERs, and ODA/DOAs. Certainly in the cockpit as well, irrespective of the fact that the crew had no information on the matter at hand or not. The human goes one way, the world goes the other and it ends in tears. How about we start training SA properly.

On APLT disconnect, automatic disconnect requires reaching the limit torque of the servo for 10 seconds to trigger the automatic disconnect.

Correction
 

QDM360

I previously referred to post 519 as a credible source for a translated version of the initial report.

Having now read the English version of the official report, THERE IS NO REFERENCE WHATSOEVER to an un-commanded A/T activation during the takeoff.

My apologies.

tdracer

I'm not sure why "throttle snatcher" got into this discussion. The implication to me is that it would have been obvious to any pilot.
But wan't Tarom which you referred to as an example a slipping or slow reduction while in a high workload (and the degree of pilot incapacitation challenged).

For the people saying that "it can happen everywhere", while true, it's not statistically relevant. Having operated in those region for a few years I can tell you that there is a reason this events keep happening there, but for political correctness I can't explain the why's and how's, but if you haven't operated there, you just don't know and you can't generalize it as a global problem, because the statistics in Indonesia paint a very clear picture and people refusing to see it are part of the problem.

Irrespective of what part of the world you are talking about, the quality and style of the basic training at flight school sows the seeds for the embryonic pilot.

Whilst instrument flying is a major part in what is taught prior to multi-crew flying, being comfortable “turning an aircraft onto its back” is part of the development needed to expand the SA capability of student pilots.

Limiting the basic non IF flying training to steep turns and stalling, together with the EASA 5 hour UPRT package, to “tick the box” is not the solution to prevent this type of tragic accident in the future.

Whilst EASA recognise the need to improve SA, it must be only a temporary measure until a more robust training syllabus is adopted.

And yes, it will cost more, but compare that to the price of this style of accident, not only in momentary terms for the beancounters, but the human suffering which inevitably occurs as well.

parkfell, I agree - getting used to being upside down and then having the bottom fall out of your world - all on limited panel instruments - is the good basic training that lives with you for the rest of your days. It helps with the 'startle effect'. Then, after that, you need regular top-ups of basic instrument flying to maintain the scan and situational awareness.

fdr, In an earler post, you mentioned something about the kinds of SA training you would like to see rather than the current ticks in boxes. Rather than a PM, could you share your ideas with us all please.

Even though this tread is creeping I'm helping the centipede with a two penny worth.. when we "teach JU" we are really just teaching recovery FROM unusual attitudes, not UPSET per se. The Unusual attitude assumes we have somehow mismanaged the aircraft and ended up exceeding the acceptable limits, thus we are now "recognising upset" and recovering. What is more difficult to teach or indeed demonstrate is the Upset caused by External factors such as flying through the bottom of a CB, Mountain Waves etc etc or indeed an Upset induced by poor handling of a system failure.
The brain has limited diagnosis capacity and tends to try to put the aircraft in a dynamically stable sate before recognising why it is unstable to start with, our teachings say nose low..do this... nose high...do this.. high bank... do this... Which is all well and good if you have recognised the influence that put you there in the first place.
We have limited time in the sim and it's basically about threat and error management within the training budget and capacity of the crew, an old lag Captain may be cool as cucumber but is newly typed FO is sitting like a rabbit with eyes in headlights and in this scenario "advanced training" becomes of limited value as the new FO has no capacity left for CRM or fault diagnosis.

As a simple ppl I have a question that baffles me, The FDR shows a bank up to 45 degrees after AP disengage but no corrective aileron applied, If I fly through a lumpy bit of air and one wing suddenly drops it is automatic or instinctive to immediately apply opposite aileron, no need to think about it it just happens so what stopped the pilot in this incident doing what should be automatic?

I would like to think that any pilot worth their salt would immediately grab the controls to level the aircraft. As you say, this should be instinctive and automatic - no thinking required.

I wonder if PF got too closely involved with avoiding CBs and just did not notice the increasing thrust lever and N1 split and the increasing bank on the yokes? If both PF and PM were concentrating on the weather radar and looking out to avoid and find a path between CBs; maybe the increasing asymmetric situation was simply not noticed. Then with no warning, the auto-pilot suddenly drops out and the aircraft rapidly banks over..............

From translations of a briefing conducted Wed 2021.02.10 by Captain Nurcahyo Utomo, Chairman of the KNKT Aviation Accident Investigation SubCommittee.
.

jodeliste

In the absence (until the final report) of an FDR trace, we have no knowledge of the bank angle or control inputs after it had passed 45°.

Jodeliste

I share your concern as to why no manual corrective action was taken following A/P disconnect. No.1 task : fly the aircraft (aviate)
I am assuming pilot incapacitation was not a factor.

Unfortunately my memory of this isn't very good (I never had much involvement with the 737, so didn't pay as close of attention to the discussions as I might have). But after the China Southern 3943 crash (one thrust lever stalled at idle, resultant thrust asymmetry resulted in a loss of control and crash), Boeing did come up with a warning that there was excessive thrust asymmetry with the autothrottle active (I was on the Boeing Propulsion Safety Committee and it was discussed during at least one of our meetings). I don't recall the detains of how it worked or what it did, but I had assumed that - having contributed to a crash - it would be fitted fleet wide (by AD if necessary). For whatever reason, it apparently wasn't...

I always thought that on the B737 Boeing kept the big clunky control column , clattering trim wheel and oversized moving throttles for a reason.
That is that you don’t need a warning system. Its in your face.
In most of the UAs , SA problems described above crews have watched on while things get clearly out of shape.
How you train to mitigate against that in already full training schedules I don’t know.
But I do know that when I have seen these scenarios repeated in the sim. the instructors input is usually
“ Don’t disconnect now , let it fully develop”
How do you manufacture airmanship ?

It's an unfortunate game we have to play in the sim. Your immediate reaction is to fix it, but if you fix it too early, you miss the "good bit" of the training exercise. I certainly hope we don't have people flying around who leave the sim not knowing which parts were done purely for demonstration/training.

jodeliste

You’d think so, would’t you? When I first started flying my first multi-crew airliner (an ATR), having come from a single pilot IFR operation a basic aircraft, we were in the hold and hit what I think was a bit of wake turbulence (on autopilot.) We had a sharp wing-drop. As First Officer, Pilot Monitoring, I was a little bemused when the Captain’s response was to grab the overhead grab-handle and stare wild-eyed at the PFD! I quietly just hovered near the control column, but then we recovered to wings level.

Hi Jodeliste,
But as far as we guess it already had full autopilot opposite aileron. Thus aileron alone cannot be manually made to go even more.

[As our learned friends have been explaining, rudder and throttle plus elevator and believing the instruments when upset, are one's tools.]

With all those poor folk dead it's not a light discusssion here.

The autopilot has limited aileron authority compared to manual input. The following AAIB report states:

AAIB Bulletin No: 2/97 Ref: EW/C96/9/3 Category: 1.1

Very interesting.
We have heard here that the crew should have noticed the large control wheel movement before the AP disconnected.
The aircraft was supposed to turn right in the timeframe just before the upset. Therefore the crew would expect some amount of right control wheel movement.
I don't know the speed at which the AP put in right aileron, ie how long time passed between say 10 degrees right control wheel position and reaching 20 degrees limit position?
Should a 10 degree right control wheel position raise any alarm from the pilots when the pilots expected some level of right control wheel movement due to right turn put into the AP? If no, how much movement should raise an alarm? How long time would pass from the "alarm threshold" position until AP disconnect?