Well i'm looking forward to a few more G/A's in the Sim. That said, i can't help looking at this from a User Interface design question. There are myriad buttons & knobs within arms reach, but only 3 big chunky ones. Yoke/Le Sidestick & Thrust levers & Gear I suggest this is for a reason. The law of Primacy. This was a high workload incident, granted, with a tired crew, but those big levers should be high on the list of things to be all over. TOGA triggers et al are important, but "Go around-Set Thrust-Flaps-x" and push and hold the things forward, old school. A low energy state near the ground is not the time to be wondering 'what it's doing now...'
There but for the grace....

Rat 5
"If every GA needed manual thrust", I suspect we'd just experience different types of accidents. In fact, likely more accidents, because we'd be changing the 99% of our GAs that are 'easy' and mostly work well, for the sake of the 1% of GAs that are badly described in the manuals, are hardly ever trained or briefed, but which will generally catch pilots by surprise!

More training of this manouvre, better manuals and knowledge, and regular pre-briefing are what's required to prevent this accident occurring again.

In fact, had they simply recently pre-briefed this manoeuvre, I could pretty much guarantee this accident would not have happened!

That's all.

Any other ideas are absolutely fine, but are simply solving problems that frankly, didn't occur here.

You don't check the thrust to retract LG, you confirm a sustained positive climb.
But you should check the thrust output before you rotate ...

CONF iture
Sure! I am not linking the two but indicating wrong priorities. In a bounced landing GA there is no need to do anything other than thrust and attitude. Within two seconds flaps and gear was retracted indicates the crew wasn't treating this GA any differently. If briefing for this difference is not an option then perhaps following options can be considered:
1. Changing nothing because it didn't happen before and may not happen again.
2. Boeing activates the TOGA switches for all occasions and take chances if someone accidently triggers after landing or during flare.
3. Boeing deactivates TOGA auto throttle/FD mode for Go around and makes it a manual manoeuvre triggering even the FD GA mode at TOGA position.

From Airbus FCTM:

GO-AROUND NEAR THE GROUND


If the flight crew performs a go-around near the ground, they should take into account the

following:


‐ The PF should avoid excessive rotation rate, in order to prevent a tail strike.

- A temporary landing gear contact with the runway is acceptable.


‐ In the case of bounce, the flight crew must consider delaying flap retraction


‐ The PF should order landing gear retraction when the aircraft reaches and maintains positive


climb with no possibility of subsequent touchdown..


I don't think it can be different in any aircraft

Rat 5 @ 1636

"Make all GA's manual thrust = system redesign"
"Every GA is the same motor action. No choice, no excisions, no doubts. JUST DO IT."

You mean just as they were doing it on the DH121, AKA Trident over 50 years ago.
The Autopilot would fly the G/A from a DH of 12 ft. The activation of this mode was achieved by fully advancing the power levers manually.

I can't vouch for what would have happened if this was attempted after touchdown as in the 20 or so sim sessions I did I cannot recall this exercise. Nor for that matter on any other types I flew after that.

Is this or something similar that happens on the wonderbus? Power lever position or a TOGA button?

TOGA on Airbus FBW can only be selected* by pushing the thrust levers fully forward. This is instinctive and logical.

Two big bad crashes have happened relatively recently to modern Boeings when the moving thrust levers did not move, and the crews apparently did not realise it and take appropriate action?

Perhaps the Airbus designers got it right after all when they made the thrust levers non-moving?


(*TOGA is automatically set if the aircraft slows into the Alpha prot region.)

I'm not going into details, just trying to have a discussion about avoiding mis-manipulation of a/c systems in stressful/startle scenarios.
It has been mentioned many time that one common screw up is an all engine GA. People have said that a manoeuvre brief will solve many problems. Hm? Briefing something you do rarely, 30mins before you might need to do it, is not a guarantee of success. It might help, but.......What can help is if the manoeuvre is simple and familiar to another common manoeuvre. This was the case with a normal GA & takeoff of B757/767. We executed both the same. Now, on B737 CL/NG I was astonished to discover that the GA is very different to takeoff: you accelerate during flap retraction via the flap lever. This is the only manoeuvre where this happens. I am still searching for a good reason why?
To make things worse there are some operators who emphasise that flap retraction must start at 400'. So they have the GA call, initial action of thrust, flaps, gear, roll mode all coming in very close succession, and then, just to increase the workload even more they have to accelerate the a/c and retract the flaps using a method they rarely practice. What's the rush, and why design the system to be so different to a normal takeoff?
Every other time we take off, be it normal or with engine failure, we accelerate at 1000'. A SE GA is 1000': in all cases the a/c is accelerated via MCP speed bug. Why make the all eng GA so different?

What happens on other Boeings? Is this now a common company technique? I realise this is not connected directly with DXB accident, but it does seem this can have human factors elements to the root cause.

On B737NG the AT disengages 2secs after touchdown. I assume this is 2secs WOW and not a bounce. It doesn't say if it's both wheels or one. On B737NG the only time you'd land with AT engaged is an autoland. It would be possible to still make a GA, but TL's would need to be pushed manually. In all my various Boeing airlines B575/676/737 the GA SOP was PF to follow through TL's and PM verify thrust, even on a GA with AT engaged. This would happen from an autopilot approach. The autopilot might disengage, or not, but AT 'should' advance to a suitable GA thrust. Perhaps because we flew more manual approaches, certainly in the latter stages, manual thrust was in our 'instincts & muscle memory'. It also improved, sharpened and widened our scan. We flew 1 handed on each control and were aware of the thrust setting.
I admit I had forgotten this performance, or lack of, of the AT after touchdown. It had never been trained nor experienced. However, because the technique was for PF to always follow through the TL's, just as on takeoff or any other GA, I would not expect it to have been a problem. Thrust would have been advanced.
I wonder if Boeing FBW AT pilots cover TL's or just trust, and only place hands on for TOGA selection? (I hear an AB GA is initiated by manually fire walling TL's, but do you cover them on approaches?)
I just wonder if this automated system is the best it can be, is a real safety improvement or needs tweaking.

Vilas, you are right. It is no different in any aircraft. "PAFU" was a mnemonic drilled into us old boilers at basic flying training school. Power, Attitude, Flaps, Undercarriage. Even in liitle old Cessnas where the undercarriage was fixed, we still had to call it out. After CPL/IR and 200 hrs, I was one of the fortunate old school cadets that went straight into a heavy airliner. "PARFU" was the drill, again; "R" added for Radio in order to set the Flight system. It has worked for me in every aircraft and once, I bounced and decided to GA. PAFU had been well rehearsed, well practiced and on the day, proved to be a non-event. well, apart from Tea & biks with the Fleet Skipper who wanted to know why we bounced. Good grief. Biks were rubbish too.

There are constant references on these posts to this being a 'bounced landing'. It wasn't. This was a decision to go around from a smooth but long landing. It was not the standard G/A, neither was it the standard 'bounce' that may have been previously practiced. This was a case of treating it like a touch and go but delaying the flaps until a safe rotation was made with sufficient thrust to safely climb away. The only concern would be ignoring the config warning as you climb away with landing flap, something the aircraft will happily do if done correctly.

No Airline can effectively teach for every possible scenario. They can, however, start by employing the best pilots and retain them by offering industry leading terms and conditions. You then recruit the best within this pool to become trainers who have enough experience and knowledge to encourage resilience and airmanship within a framework of robust SOPs, all the time being supported by a Company ethos that's not overly restrictive or prescriptive. It needs to be a genuine 'Just Culture' with effective support from Senior management, the CP's and their deputies.

Nobody for certain will ever know the reason why the Captain initiated a straightforward G/A procedure, other than the Captain himself. However, it would not take a genius to reason that many of the above mentioned factors may well have played a role, one way or another.

Harry

What he said!

Good post harry. Pilot errors, not wanting to excuse them, can also be very much 'automatics' or 'management culture' induced.

Good post Harry. Much to agree with.

Except:
You only get the config warning whilst on the ground. Once you're airborne, you are no longer 'taking off', and hence don't get the take off config warning.

As a point of interest, because you generally don't have much accelerating to do before reaching the bug and rotating, config warning only occurs for a very few seconds.

1/ well in this case training is available for "rejected landings" after touchdown if the Airline Flight Ops Dept chose to do it. Ours does and has done for over 10 years. I last did it about 12 months ago in a Sim session in RJTT.

2/ Once you are Airborne the Config Flaps warning stops, then push TOGA and commence the SOP GO Around calls for your type.

Good post Harry.

I sure don't expect a guy at EK to answer, here, the following, but I can still ask :

1. What happen to that crew if they disregard the audio caution and taxi to the gate ?
2. How is it seen by the management to disconnect AP or AT ?

Smooth landing:sad: One pilot witness was recorded as saying he watched that 777 do a very heavy landing and the crew went around from that heavy landing. That wasn't mentioned in the official initial report. But then again, it wouldn't be, would it? Local culture affects accident reports in some countries.

If the second main landing gear touches 3 sec after the first one did, I wouldn't think it is a "very heavy landing" ...

Possible thread drift but to be accurate there is one 777 config warning which is available when airborne (but only on GE-powered models), all others are inhibited after take-off.

That "pilot witness" is clearly contradicted by several other eyewitness accounts and the preliminary report from the Aviation Authority which all describe a smooth and gentle landing. It is essentially impossible to have a "very heavy landing" (BTW, that should be "hard" not "heavy") when there are THREE seconds between the right and left main landing gear touching down. And then it takes another TWO seconds after that for the RAAS to make a "Long Landing" alert and FOUR seconds after that (a total of NINE seconds after initial touchdown) that the aircraft became airborne again. The timing of that sequence is impossible for a "hard landing" that results in a "bounce." The facts are quite clear. The aircraft made a smooth and gentle touchdown, and before the nose gear touched down the pilots initiated a Go Around/Rejected Landing. This was NOT a "bounce" in any sense of the term.

This crash (thankfully no body bags, nice job on the evac) seems to hinge upon three core elements:


1. Decision to go around (SOP based or condition based?) - possible "supervisory error" finding
2. Go around actions taken (pitch and power?) - possible "training/CRM" finding
3. Decision to raise gear (when?) - possible "training/CRM" finding.


Some comments regarding corporate culture and their influence (sometimes insidious) have been made here, but whether or not the final report addresses that remains to be seen.

Someone pointed out earlier that had the plane settled with the mains still down, there may still have been some damage to the aircraft (due to a firmer landing/thump) but significantly less damage than with the landing gear up and the aircraft sliding down the pavement. (For example, less likely that engine integrity / mount being compromised ... fire ... )

Off course you follow them in any aircraft. Using automation the scan remains same but it is for monitoring of automation effecting changes. During approach pilot has to keep his hand on throttles/ thrust levers. Even in airbus the last point is 1000ft AAL although you can't follow thrust levers but like this accident things can happen so fast it's not a good idea to go searching for them

Ken,
I agree with your post, but check the sequence again, there are only 6 sec between initial touch down and airborne status.

The Raas alert is supposed to occur before touchdown to give time to GA. Did not happen here. Was that due to operator's preference? On reflection although presumably it is possible to bounce 300 tonnes or so if you really try, it must be rare for it to happen. 300 tonnes usually arrives on terra firma so sweetly you dont notice or with such force it rattles your teeth, but always sticks. As SLF these days I stand to be corrected.

Up linker said

One of those crashes involved a Capt converting from an Airbus. He didn't even bother doing anything with the thrust levers. Admittedly Autothrottle makes you lazy if you let it.

I think an issue we have these days is pilots who rely on protections or the automation to get them out of trouble. Most people I know that had flown aircraft that were totally manual or had only basic automation never get anywhere near a flight regime that will trigger a protection unless they have severely cocked it up. Recent accidents and incidents suggest that pilots who have only flown fly-by wire aircraft are allowing themselves to get into undesirable states knowing the protection is there to save them..... Until it doesn't.

At the end of the day the automation may or may not be intuitive, but the pilot certainly needs to be.

This was a case of treating it like a touch and go but delaying the flaps until a safe rotation was made with sufficient thrust to safely climb away.

Harry; I assume you meant they should have treated it like a touch & go---but they didn't. Is it the B777 technique to make a touch & go with landing flap? On my types it is: select GA flap, thrust, trim, accelerate, rotate.

Off course you follow them in any aircraft TL's.

Vilas: I agree with you, of course. The reason I ask, and it's no guarantee to be true, but the Air Crash Investigation of B777 at SFX showed PF flying 2 handed during the visual with nothing on TL's.

However. I have watched cadets in the sim flying A/P A/T approaches hands off everything; until my rolled up newspaper made contact with their dome. Those telescopic pointer thingies are great too.

0 sec: R/H main gear touches down
3 sec: L/H Main gear touches down
5 sec: RAAS alert
9 sec: Aircraft airborne

Depends entirely on the programming, and that is up to the customer. RAAS can be programmed to be inhibited only after all three gear are on the ground. The nosegear never touched the ground in this incident.

I agree. Plenty of airline SOP disagree.

Clearly we disagree. I believe we can prevent no thrust go arounds just as we prevent no thrust take offs. Both are impossible. However, a no thrust GA is much much more dangerous than a no thrust take off. I don't know the SOP for autothrottle takeoffs of all airlines, but I personally cannot conceive of a pilot doing a take off on autothrust with his thrust hand not on the thrust levers in the event of a rejected takeoff.

On the other hand I cannot conceive of a pilot doing an approach/landing on authothrust with his hand not on the thrust levers in the event of a go around or rejected landing. It simply flies in the face of all the training I have ever had. Yet the pilots of some (many?) airlines apparently do the latter routinely. Do those same pilots also do takeoffs with one hand in their lap? I really don't know.

Nice post 4468.

I beg to agree and disagree - because the chimp inside us can be very different, depending upon our upbringing. Having served my apprentiship on basic aircraft that have no FMC, my 'chimp' never forgets the thrust levers, but it does have a tendancy to forget the toga button. So I am a reverse chimp, and need to practice other aspects of this scenario.

There is also a problem in that some will vehemently disagree that there is any trace of a chimp in any of us, for cultural reasons. And that leads to what one might call 'overconfidence'.

Not necessarily. If the pilot had manually pushed the thrust levers forward and established a climb rate and then pushed the TOGA buttons, once away from the runway the automatics may have kicked in and given him FD guidance and authothrottle to maintain an optimum climb rate.

So from my perspective, doing the manual stuff instinctively will keep you alive and then you rationally ask the automatics to help you out to make you look like a good pilot. But instinctively asking the automatics to help you look good AND keep you alive seems like a recipe for disaster.

Now I'm getting confused Ken.

I clearly said "he shouldn't have pushed the TOGA buttons on the ground AT ALL!"

I thought your entire point was that he pushed the buttons and expected the automatics to do the rest?:rolleyes:

What you have just described is precisely what we all know he SHOULD have done, but didn't! Because he wasn't sufficiently familiar with this particular manouvre to instantly select the correct actions! (They would NOT have been the correct actions for most other GAs!) This has no bearing whatsoever on how many manual approaches he chooses to fly. Which in any event, would ALWAYS have been flown with auto thrust in! As is the wisdom of 'the manufacturer'!

You are of course correct that shortly after airborne, pushing TOGA THEN produces valid FD and auto thrust modes

All aside the initial factor in this accident was the touch down 1100 meters beyond the threshold. Two seconds after that came the second factor, the smart cockpit RAAS annunciator with its "LONG LANDING" call. The well drilled crew immediately reacted to it.
To me this is yet another case of unquestioning submission to and total dependence on automation and abandonment of an important human sense of intuition and instinct. Not even flying by numbers but flying by on/off electronic signals.
Would this accident have occurred if there was no RAAS.

ken, conf iture has gone to the bathroom....
37:17 rh gear down
:19 raas alert
:20 lh gear down
:23 airborne.

Sorry I was not clear. Let my try this approach:
1. A pilot can be trained to perform different "correct" actions for different scenarios, based on system logic of the specific aircraft he's flying.
2. A pilot can be trained to perform the same basic action regardless of the scenario and independent of system logic of any aircraft he's flying.

In this case the pilot was expected to know the details of the RAAS activation logic and the details of the TOGA switch inhibition logic and take the "correct" action based on his knowledge of the automatic system's logic. He failed.

I believe that pilots should be trained that EVERY time he wants increased airspeed and/or altitude that he ALWAYS pushes the thrust levers forward and manually establish the acceleration/climb and THEN pushes the TOGA button to assist him to establish the optimum acceleration/climb rate. Thus in an unusual situation such as this one, he instinctively performs the correct action (manually push the levers forward) and he would not have to know (and under great stress remember) the details of either the RAAS or the TOGA logic to take the correct action. The correct action (push the levers forward) is always instinctive and the automatics are only used to finesse the pilot's already correct action. To me, the worst thing that would result from such training is a short term overtemp or overspeed of the engine (and with modern fadec engines that is unlikely) resulting from moving the levers too aggressively. And this instinctive action will get him (and his passengers) safely away from the ground independent of system logic. Once he is away from the ground and his airspeed is up he can figure out the system logic that's preventing him from getting FD bars and/or autothrottle. Expecting him to figure that out very close to the ground and at very low airspeed is in my mind a recipe for disaster.

According to the timeline in the preliminary report in post #1574, three seconds are missing from the above timeline. In that report, there were 3 seconds between main gear touchdowns, then 2 seconds to RAAS alert, then 4 seconds to get airborne. 3+2+4=9. Apparently I'm misinterpreting the wording of the preliminary report. My bad.

RAT 5

That's probably the same on most, if not all commercial jet aircraft, including the B777. However, doing 'circuits' is one thing, recovering from a baulked landing is quite another. In the case of the latter, keeping landing flap is preferable as it's unplanned and the priority is to get the aircraft airborne again. This is in effect an undesired state whereas a touch and go is a stable approach and landing, with a planned ground flap retraction and trim. The config warning will be present, on the ground, but the flap can be selected to the G/A setting once safely established in the climb. Higher than usual control forces may be required as the aircraft is trimmed for landing, not take off.

Then can begin the sequence of a 'normal' go around. Hope that clarifies?

Harry

ken v, you're in the clear. You were just misled by post 1574 who was reporting on the report and got it wrong. Always go for the original. Cheers.

Ken
I'm afraid we will have to disagree.

On a Boeing, simply pushing the thrust levers forward is absolutely NOT always correct, or advisable! (Though it is on an Airbus, that so many seem to decry!)

Try doing that on a coupled approach, with a decision height below 50R and see what happens.

That's because the TOGA switches don't merely control the thrust levers. They also signal the autopilot and/or flight directors to immediately commence pitch and roll manoeuvres. Or do you advocate simoultaneously selecting the autopilot out too, and looking through the flight directors on every go-around?

Oh and you'll have to continuously push the thrust levers too, because if you don't, auto thrust will be trying to close them! Unless you take that out as well!

As I said in an earlier post, you need to think what other traps your 'solutions' create, and understand why it's not how we do things in modern machinery.

The solution here is more familiarity with this simple manouvre. That's all.

Not simple, this flying business, is it?

The Crew made a massive cockup BOTH OF THEM this is not in question.They crashed/destroyed their lovely 777 because they completely screwed up from an unstable condition on or near the ground (failed to ensure enough power was applied regardless of automation status) this is without question.There is not much doubt that the cabin crew did a great job with the evacuation but the PRIMARY reason for this and a lack of injuries was the very fortunate fact that due to the gear being retracted, the doors were very very close to the ground, you could basically just walk off the 777 before it exploded.What a stroke of luck, which got EK off the casualty hook.Peter.

in the face of a probable visit to the office and bullocking for a/landing long and b/ignoring the RAAS callout?

The other major factor for survivability seems to be the robustness of the 777 airframe. In three major accidents (LHR, SFO, DXB) the airframe remained intact long enough for all the passengers to evacuate. The evacuation slides do, however, seem suspect. They failed to properly deploy in SFO as well as here.