FiveGirlKit

The left thrust lever has operated correctly; the problem is that in response to a thrust reduction command, when the climb mode of FLCH/VNAV - THR; changing to ALT CAP -SPD, that gives a thrust reduction. In the first instance, the left thrust lever comes back slightly, as the AT tries to maintain the commanded speed, which will probably be 250KCAS. As the RT TL is not being picked up by the clutch, the AT will increase the thrust reduction command, and effectively close the LH TL to idle or near that.

At 35 bank the GPWS Mode 6 "BANK ANGLE " sounds continuously. If the bank angle of 45 is correct, then sometime thereafter the APLT has disconnected, which can be determined how, and the aileron that has been applied has to be taken over by the pilot, which is probably not the case. Disconnect with a large aileron deflection often ends up with inadequate force to maintain the existing displacement and a return to a near-neutral aileron is highly likely. At that point, the roll rate is going to accelerate away from desired bank due to the yaw that still remains. The roll rate from the yaw is around 25/s so the aircraft is at 90 bank in a couple of seconds, and what the crew then do is going to be a major factor. the roll to the right is compromised by the more than 90-degree deflection needed on the yoke, but of all of the potential upsets, a captain in the left seat is going to get more right aileron on than the FO, If they are using their outboard hand only. If they grab the controls by both hands, then they have a reduced likelihood of getting full aileron on. With a bank of 90, more RH aileron is required as the natural tendency is then for the nose to fall and that increases the yaw and the subsequent roll rate, so it is not surprising that the crew get confused. (1994 Buff bingle Fairchild, "Budd" Holland and his unfortunate crew.... 2010 Elmendorf C-17 is slightly different, but kind of close) Approaching this point, the nose is going to start slicing down, and speed is going to start to increase, but so will the g loading, so initially the speed will not build rapidly. if the aircraft stalls (not necessary to have this event) the aircraft will rapidly roll inverted, and the nose will drop towards vertical promptly. Speed will increase rapidly but the plane steeply noses down, possibly still with the asymmetry complicating the recovery. Without a stall, speed/aoa stability will assure that the aircraft at the 90 bank or steeper will start loading up, and as soon as the vector is below horizontal, the g will build up rapidly, and the aircraft is going to have a steep spiral, It is possible the crew follow through with the roll at that point, it wouldn't be the first occasion of a crew doing that (B717 stall wild ride) It is conceivable that the aircraft recorder ends with the TLs still split, LH at idle, RH at climb thrust. There are surprisingly few ways to get from A to B in the time available; from the first report, a roll excursion was likely. reminiscent of Sharm Al Sheik.

Spot on Bleve. I have even recommended pilots consider such steps when obviously confused as to which engine has failed, quite often when manually or even automatically flying a holding pattern. Not recommened very close to the ground (!) but very effective in most other phases and can be completed in a few seconds and very effective in confirming the live engine(s) an so required rudder input.

Whilst not commenting on this accident the above "Done any aerobatics" reminds me of training we received at RR some 50+ years ago. We were sent up in a Dove, each seat kitted out with an instrument panel from which we were to note readings for instance at pre-stall and other conditions. One of those conditions involved the pilot putting both engines on full power then cutting one out completely and leaving the aircraft to do its thing for, well it felt like forever, whilst we were supposed to calmly note the various sensor readings.

It's a bit akin to getting into a skid in a car, everything happens very quickly unless you know what you are doing.

Indeed, the OM says in the Autothrottle System description:
So I would expect this to be common knowledge among 737 Classic drivers.

I wonder where the seemingly widespread idea originates that there is a single motor driving both levers.

It also seems that a "Thrust Split Monitor" is available as an option, described as follows:

How common is this thrust split monitor? It seems it might have been helpful in this case. The very fact that it was available indicates that someone thought it might be a good idea in some cases, but I know nothing about its history.

The aircraft was initially cleared to climb unrestricted to FL290. At 14:39:01, ATC instructed the aircraft to stop the climb at 11,000 ft due to traffic. The preliminary report states the L thrust lever started reducing at 14:38:42, ie BEFORE the aircraft was instructed to stop the climb. Does that not rule out a thrust reduction caused by an ALT CAP-SPD mode change?

That was the scenario I had in mind, when asking what a recovery with split thrust would look like.

The FDR has given investigators definitive information of an autothrottle defect where a split throttle situation occurred. To a competent crew it should be nothing more than a minor technical inconvenience. Yet in this case the incompetency of the crew to handle a split throttle situation must surely raise eyebrows.

Described in Post 310 of this thread is the utter confusion I witnessed among the two pilots in the simulator - in particular the very experienced captain - when an autothrottle clutch motor was failed while both thrust levers were at idle as part of speed control. As power was increased with flap and gear extension and only one throttle opened up, the stunned crew did nothing and watched the aircraft roll into a spiral dive to oblivion.

At the time I found it hard to believe such a simple defect could quickly turn into a major loss of control disaster. Now it has happened in real life in the form of the Sriwijaya 737. One wonders how many other experienced airline captains are out there that simply can't fly.

This preliminary report is eerily similar to what Centaurus wrote in this post.

https://www.pprune.org/tech-log/6369...l#post10935471

Centaurus

I think I must have been writing my post and searching for the link, as you typed yours. Thank you for your words of wisdom. It really underscores the need for less automation. Especially after covid.

And yet managers insist, especially as they know we are all a bit rusty, to use automation as much as possible, and of course most pilots do.

@ FlyingStone and bieseker:

Thank you both, my mistake re number of motors, (corrected). Only flew it for one season but I should have checked before posting. General hypothesis regarding lever asymmetry remains.

Some flying rules for flying Boeing's that have served me well:

1. Never totally rely on the automatics to do the right thing when you need or expect them to (be that the auto-pilot and / or the auto-throttle). Assume that those auto-bastards are out to get you!
2. Any time that the thrust is changing keep your hand on the thrust levers and if you get a mismatch / throttle stagger or the levers don't move as & when expected, do something about it (disconnecting the automatics as required). The same applies if the control yoke suggests that the aircraft is not in trim [with typical reasons for that being: asymmetric thrust (why?), and / or asymmetric fuel load between the wing tanks (fuel leak maybe?), and / or that the rudder trim is not set correctly (why would that be?), and / or that the silly sod in the other seat is resting their foot on a rudder pedal (****!)... use DODAR]... figure out why, do something about it if so required and keep your colleague in the loop.
3. Airmanship applies ! (though I'm well aware that the 'A' word is a dirty word in some airlines).

bsieker... fwiw, almost every time I go to work (in a B737 Classic) I've got matched N1's but with thrust levers that are spilt (see pic)... and yes I know all about the throttle stagger limits, and yes I do write it in the TechLog (and take pics of it too), and yes the engineers fiddle about and try to fix it, but hey ho what to do eh?! (see the rules above)

https://cimg5.ibsrv.net/gimg/pprune....f9cf6d12ed.jpg

Old King Coal
Is that your pic? Does it happen the levers are that much split with the same N1?

Check Airman

Or highlights the need for better automation. I am surprised that something as benign as a large engine asymmetry (however caused), can potentially result in a loss of control. Why isn't the rudder controlled by the auto-pilot?

BuzzBox

I believe you are correct. Probably the problem wasn’t the Auto Throttle but the engines settings/readings. Something similar to Aeroflot-Nord B737-500 crash in Perm in 2008. Have a look at this throttle stagger:

Crash: Aeroflot-Nord B735 at Perm on Sep 14th 2008, impacted ground while on approach to Perm

Anyway, I absolutely agree with Old King Coal, assume that those auto-bastards are out to get you!

That looks a little beyond limits that our engineers would accept when I was flying these. Maximum was the handle width so still the best part of an inch on some older -300s and -500s

Auto-throttle malfunction during take-off, and again during the climb. And it was re-instated twice, and not monitored. Surely, disconnect the darn thing, and fly the sector with manual throttle.. Maybe I'm an old dinosaur.

The crew seemed hell bent on using a defective A/T system, the roll at 4.5 degrees nose up could have easily been due to them yanking back the thrust levers and stalling the AC, in any event, Boeing recommend disconnecting the A/T, once again this is a case of Pilot Error. Basic airmanship seems to have been lost and without the CVR it's just speculation as to what was going on. This was, according to their "hours" an experienced crew, which again can be a recipe for disaster if the cockpit gradient is slight, Captain is described as "kind Pilot", usually this ex Military guys are assertive, the plot thickens.

One thing's sure and that is that with the kind of leadership as illustrated in the report, nothing is going to get better. Any of their pilots that accept this kind of tosh will, I hope, never fly me.

The Chief Pilot of Sriwijaya Air issued the following notice to pilots on 20 January 2021: This notice reach you as a call toward the safe flight. With recent tragedy, we urge all pilots to raise awareness and keep the highest professionalism and discipline on your duty. This can be fulfilled with many guidance that we had: • Follow Operating Experience guidance. 27 • Review Training Aid. • Awareness of aircraft position, attitude, aircraft systems by active monitoring the state of aircraft on every phase of flight. • Awareness of aircraft configuration, thrust lever position/power setting and flight control system modes, anytime airplane deviate from its intended state must be corrected immediately. • Cockpit crew is responsible for entering clear and accurate write-ups of any discrepancies, including any incident or anomaly observation in AML, use of FRM (737NG) and/or describe discrepancy information comprehensively.

I suppose it is nice of him to remind them that professionalism and discipline on duty but it is hardly a strict set of orders

So this had been written up in the ATL, cleared with "cleaned contacts tested found satis report further" or similar .... then ...
it retards a TL during the takeoff, and again at 8000 feet ....

I really have to ask why the AT was not disengaged by the second rollback at the latest.
Is the B737-5 really that difficult to fly without it ?

The answer has to be no.

In the preceding discussion I see two issues in blaming it on the pilots.

1. is the issue the lack of normal skills in recovering from asymmetrical thrust?
2. Or is the issue the failure to recognize that thrust asymmetry is happening until too late?

the corrective actions the industry must address are key to identifying some of the links in this accident without regard to blame or demonizing the crew.

Uplinker

Disagree. Less automation. A “better” system will fail in a new and improved way.

Autothrust predates me, but I’m sure when it was introduced, there was a laundry list of things it helped with. Now we’re seeing problems it causes when it doesn’t work exactly as advertised.

Why wasn’t the rudder controlled by the pilot?

The issue is the inability of pilots to fly the airplane up to cruise altitude, trimming everything up, and THEN, turning on the automatics to see if they actually work. If the pilots cannot do that, then they belong back in seat 28D and 28E instead of 0A and 0B.

This was mentioned multiple times now - but is not backed up by the report. The report first mentions the TLs to diverge while climbing through 8150ft. Nothing unusal is mentioned for the takeoff.

Back in the old day, having staggered thrust levers wasn’t unusual, particularly on 4 engined types. Whilst this is an older generation of aircraft, the accident lends weight to the Airbus philosophy of fixed lever positions and not having the levers move by themselves.

Crew competence seems to be the main factor here, the ability to recognise that a system wasn’t performing correctly and take appropriate action was lacking.

On the contrary. While it looks like the A/T malfunctioned in a fairly strange way, the thrust lever angle seemed to correspond to engine thrust output and the tactile/visual feedback in form of the moving thrust lever was still there.

In an Airbus, the levers would have been in the climb detent and would not have moved of their own accord.

The whole think is an example of the Swiss cheese model or accident chain:

1. If the system had been repaired correctly.
2. If the crew had recognised the problem and taken appropriate action.
3. If the crew been able to cope with the unusual attitude the aircraft got into.

A layer would have stopped the hole forming or broken the chain and it wouldn’t have happened.

If only they would have identified the asymmetric flight the aircraft was in, than a gentle rudder input would have given them all the time in the world to figure out what was the problem, as autopilot would have remained engaged.

I do hope CVR is found, so it can shed some light on the crew's perception of the events.

FlyingStone

Hmmm, conflicted on the need for the CVR to shed light.

Res ipsa loquitur.

Plane goes from sunny side up to zero in ~20 seconds. Required flight path has a roll. FDR is indicated to show a throttle split. The physics of what needed to happen is straightforward. The CVR would arguably add insight into the human interactions, as well as adding additional physics data points, automated alerts etc. On the other hand, the CVR will likely show the level of confusion that arises in a crew's comprehension of the events, but this tune has been played before, repeatedly, AF 447, Adam 574, AS214, Perpignan (XL888T), CAL140, CAL676, IA605 (Bangalore) etc. In all of these sorts of cases, the crews were dealing with trying to catch up to what the aircraft was actually doing, vs their expectation of state. Other cases such as AI855, KAL 8509, and similar are cases where the information to the crew has failed and the crew don't react in time to resolve the differences. This is not a TE901 type investigation as to WITWHH. The facts speak for themselves. CVR's have a paradoxical impact, that it becomes easy to dismiss the crew as being exceptions, rather than the generic run of the mill mix of aluminum tube importers/exporters. The recommendations will almost certainly add a need for better training, but of what? Of being human, of the time to recognize that things just ain't so? To open up a NNCL that takes 20 seconds to find, 30 seconds to read the index, 10 seconds to open the page, and then adds...not much help. The main problem is the crew taking a finite time to recognize what the issue is, and then to work out what is good, and what can work next to sort out their issue. The Jet upset, flight with unreliable airspeed, etc is all well and good, and great guidance, when the crew are aware that it now applies to their suddenly unhappy world. the upset training we do in the sim is vanilla flavored; 'OK, blogs, and bloggette, today, we are going to do UA. hand over the aircraft to the other pilot, and put your head down.... now....... wait for it.... wait for it..... OK.... recover.... " Just like the real world. So in the sim, having been prebriefed of the event, and recovery, and then at the point of the session that it is undertaken, then being a party to the process of the setup, and preparing for when they open up their eyes, we end up missing the bit that happens in the real world, the time it takes to recognize that the world just ain't so, and that all that training in the sim needs to be applied in an eye-pleasing manner. In summary, all the training in the world on upset technique will matter nowt, if the crew don't recognize that they have a problem in time to apply their techniques so assiduously gained in the sim sessions. Funny thing is, its not that hard to set up, we just don't have the inclination or enthusiasm to do it. There is a downside to such SA training, as that is what it is, and the lack of SA training is what keeps killing people. The problem is (and was noted in an SA training workshop some 27 years ago), that a competent crew member losing SA in a training session has a huge hit to their ego, and confidence needs to be restored thereafter. Canned UA training is a step above lipstick on the pig, and alone will not improve the outcome. The training and recurrent training programs we have are a bureaucratic box-ticking exercise, not risk reduction oriented, and that is a missed opportunity. AQP was supposed to improve that and ended up being hijacked by beancounters and compliance managers.

If SA becomes recognized as the fundamental issue we have in aviation, then that epiphany gives a direction towards solutions that may be meaningful. Or we can apply more regulatory bandaids, as the FAA did post Colgan. WTF!

No! definitely better automation. Airbus already has it. The computers restrict asymmetric effect and also the AP trimms rudder as well. Moreover thrust when it fails fails at present value. It's basic demonstration for EFATO. Whether throttles move is not the issue. Even with static thrust levers Engine failure at cruise has same effect. No ATHR means long winter holiday, no CAT3.

Bingo! The real problem here is the assumption that airplane makers make about humans: that they can be programmed like machines.

Nothing could be further from the truth. For starters, we now know that human consciousness does not perceive reality directly, but actually as an indirect hallucination based on sensory inputs. Homo sapiens do not experience reality as is, there is a layer in between. To make it worse, we believe that our perception of reality is reality.

No amount of training can change that. No amount of blaming the pilots can change that. And yes, it can happen to you. You don't know when it hits you. Your perception departs from the true reality, and when the two don't match, you are caught in cognitive dissonance which consumes 110% of your brain power, losing time to recover from salvageable situation.

The only solution to this is to completely redesign the industry around this simple fact (airplane-human interface, SOPs, CRM,...). Now what are the chances of doing that, eh?

So we will continue to blame the pilots, call it "pilot error", where in reality it's just bad design. Until the AI takes over....

fdr derjodel and others above

We have the same discussion after practically every accident and incident that involves loss of control, cfit, etc. The fundamental problem is that humans (some of them) are good at doing one thing, while automation (all of it) is good at something else. Automation, as long as it works, is good at performing precisely and flawlessly mundane repetitive tasks, precisely those where humans are likely to make mistakes because of boredom, complacency, tiredness, distraction. How many times in living memory did it happen that an autopilot failed to level off in ALT mode at precisely the preset altitude? Exacly none... How many times did it happen that those twitchy fingers turned that knob to the wrong altitude... ? On the other hand automation is incapable of performing anything outside the limits of what it was designed to do, including recognising its own failure. Humans on the other hand are quite capable of thinking creatively and out of the box, and can process and act upon a vast amount of visual information very rapidly, especially in situations they have been preconditioned to.

The fundamental problem is that humans are not equal (despite all the wishful thinking by some). Some have better skills than others in different areas. It is true that piloting an aircraft 50-60 years ago required above average skills in many areas. However the regular appearance of smoking holes around the world showed that even these above average individuals were prone to making mistakes, and sometimes simply the situation turned so complex that the human brain could not catch up with it within the time frame available. With the onset, development and improved reliability of automation, these situations have been vastly reduced, and piloting an aircraft no longer requires exceptional skills, anyone with a good average skill set can do it safely as millions of takeoffs and landings prove around the world. However the advance of automation brings about one aspect that is very difficult to deal with. It now requires way above average skills to recognise and anticipate the various situations that may arise if one or more components of the surrounding comfortable automation fails, and at least average or better skills to recover from an unanticipated situation. By definition, at least half will fail with the latter, but recognising those who will in advance is very difficult because training by nature conditions for anticipated situations.

derjodel

Autonomous flight over populated areas, or with a large number of pax is a loooooooong way off I wager. The reliability doesn't exist at present to make it possible. The car case has 2-D dynamics with predictable interactions. The aircraft has 3-D dynamics and has interactions with weather, traffic, technical faults that all need to operate perfectly. As this sort of flight indicates, there is a technical issue that triggers the outcome, and the system has to get it right, every time, without external input. Humans remain the most effective safety device for making decisions under conditions of uncertainty at this time. I don't think removing the human from the loop is the solution, I do think that enhancement in SA training is 1/2 a century overdue, and would be cost-effective and "impactful". At least bringing in UA training that are not canned exercises in training sessions would be illuminating and might change the views of some observers to the "only happens to them, not me" brigade.

What is needed is better automatics and better integration with all systems and the pilots. I am type rated on, and have flown both Airbus FBW and B737 on the line, and I have to say the Airbus FBW is excellent but the B737 automatics are more primitive and have to be watched carefully.

If an aircraft like this B737 can go from normal flight on automatics to loss of control unless PF takes immediate action on the rudder pedals and ailerons, then is it a truly stable aircraft? Why doesn't the autopilot control the rudder?

And is it realistic for the B737 PF to keep their hands and feet on the controls at all times, including during the cruise, just in case an engine suddenly fails and the A/P drops out?
.

Any conventional aircraft will do just the same. Remember the light twin from the multi engine training? Try flying that without rudder and ailerons with an engine out.

It is not and it is not even required. At higher speed, the engine failure with regards to controllability isn't a problem, as significant margin to Vmca exists. Asymmetric thrust can be easily recognised by progressively increasing deflection of the control column towards the live engine, and the autopilot will put in quite a lot of force/deflection before it gives up.

I've got experience of both and both need to be watched like a hawk. The only difference is that with one you can rest your hands gently on the control column and thrust levers and look out the window, and with the other one you need to have your eyes glued to the FMA and N1/EPR to see if something untoward is going on.

Uplinker, I normally find your posts quite good and balanced, That one though is ridiculous. What are the pilots doing through all of this, may I ask?

We have a good idea they was comunicating with ATC several times, they was configuring the aircraft for cruise, they was manipulating the AP for heading AND level change, they certainly have been discussing weather as they requested heading change for weather.
In my airmchair pilot experience there was actually quite a bit going on in the cockpit through all of this.

Uplinker

As an alternative, you could do better selection and training. Not to say that there aren’t flaws, but how much better should the integration be? Do we need a neural link, with the plane plugged into our brains?

I’m not rated on the 737, but I assume that the control column would have been fully deflected as the AP applied aileron to correct. Is that right? We’ll really need the cvr to see what was going on at the time, but I fear Centaurus has already given us a preview.

Quote: "I’m not rated on the 737, but I assume that the control column would have been fully deflected as the AP applied aileron to correct. Is that right? "

Not really, the 737 AP will disconnect when unable to hold a course or heading selected in the MCP, even in LNAV under certain conditions. The initial roll of 45 degrees should have been managed by the crew..
It seems they are not the only crew bitten in the arse by a 735:
https://www.flightglobal.com/safety/...140072.article
As an observation, It appears the transition from AP to manual flight, coupled with lack of diagnosis and CRM is an "ongoing trend"
When we teach Upset recovery, normally the PF is told to close their eyes and then open when the condition has established, and depending on our degree of nastiness, I've never seen a crew that couldn't eventually recover, some make a mess, but having said that, sitting in the comfort of a sim is a world apart as psychologically you know there is a safety net. The guys that do mess up always say " I would never do that in the REAL aircraft" nonsense, that's exactly what they would do.
I just got two things out of my draw, my "bottom dollar" to bet its a CRM issue and "my hat" to eat if it's not!

Really ?
 

QDM360

From Report..
1. Auto Throttle broken. Repaired twice.
2. During take off AT pulls back left engine. AP engaged around 2000 feet.
3. At around 8000 feet, AT again pulls left engine down. Right Engine stays. Continue climb.
4. At 10600 feet, AT again pulls left engine. Right engine stays.
5. Altitude drop. Pitch nose up. Roll started (to left). Then AP disengage. Then nose pitch down.
6. FDR stops recording after 20 seconds. No detail data released.