Announcement

Collapse
No announcement yet.

777 Crash and Fire at SFO

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by Gabriel View Post
    But, as far as I understand the FLCH mode, if the selected altitude had been lower then it wouldn't have minded that the TL were idled by the AT or manually. Yes, the AT mode would have remained THR in the first case or changed to HLD in the second case.
    Actually, however the thrust levers get to IDLE, they will place the A/T in HOLD. In this scenario, IF the altitude had been set for descent instead of climb, with the target airspeed well below the current airspeed, I believe FLCH would bring the thrust directly to IDLE. Then, if the AP is disconnected and the FD's are not both cycled to off, the A/T is effectively off duty for the duration of manual flight.

    Comment


    • Originally posted by Evan View Post
      Actually, however the thrust levers get to IDLE, they will place the A/T in HOLD. In this scenario, IF the altitude had been set for descent instead of climb, with the target airspeed well below the current airspeed, I believe FLCH would bring the thrust directly to IDLE. Then, if the AP is disconnected and the FD's are not both cycled to off, the A/T is effectively off duty for the duration of manual flight.
      Okay, what happens in this scenario?

      An altitude different to the current one is selected.
      FLCH is selected.
      The FD remains on.
      The AT is not switched off.
      The airplane eventually reaches the selected altitude.

      I think that regardless of:
      whether the plane climbed or descended, whether the AP remained connected, was disconnected, reconnected or whatever, whether the thrust was advanced or reduced, whether it was done manually or the AT did it, whether it reached idle or not, whether the AT remained in THR or reverted to HLD,
      upon reaching the selected altitude, the FD will revert to ALT HLD, and that will make the AT to revert to SPD HLD and start to track the speed.

      --- Judge what is said by the merits of what is said, not by the credentials of who said it. ---
      --- Defend what you say with arguments, not by imposing your credentials ---

      Comment


      • MCM, would it be fair to describe FLCH as a way to manually select VNAV SPD in descent? Or is there a key difference there? As far as I can tell, they do the same thing.
        “To command effectively, the human operator must be involved and informed. Automated systems need to be predictable and capable of being monitored by human operators. Each element of the [cockpit] system must have knowledge of the other’s intent.”

        Billings, C. E. (1997).
        Aviation automation: The search for a human-centered approach.

        Comment


        • Originally posted by Gabriel View Post
          Okay, what happens in this scenario?

          An altitude different to the current one is selected.
          FLCH is selected.
          The FD remains on.
          The AT is not switched off.
          The airplane eventually reaches the selected altitude.
          If by different altitude you mean a lower one, than in this scenario upon reaching the selected altitude in FLCH the AFDS transitions pitch to ALT and A/T is back to controlling airspeed. (SPD/THR/THR REF).

          Comment


          • An altitude different to the current one is selected.
            FLCH is selected.
            The FD remains on.
            The AT is not switched off.
            The airplane eventually reaches the selected altitude.
            Your belief is correct. The altitude capture then alt hold mode is essentially 'armed' all the time and captures when within certain parameters (i.e. approaching the selected altitude). Once it has captured, the A/T mode would change to SPD.

            Evan, I'm not sure how 'clear' Boeing needs to make it.

            The operation of FLCH is clearly explained in the manuals. The 'non-use' of it for the approach is also clearly stated. If ground instructors aren't able to teach the systems properly, that is something for the airline to address. This is not a new aircraft type, nor is it new to the airline concerned. Is this a case of non-pilots (or at least non-experienced pilots) teaching people to fly an aircraft they themselves don't understand? Certainly wouldn't be unusual, sadly.

            It is certainly possible that this inexperienced pilot didn't yet fully understand the system - that is why you have other pilots/safety pilots. I am obviously concerned that a training department would allow a pilot to be operating on the line with such a large misunderstanding of basic auto flight modes.

            I'm also reluctant to immediately accept the statements that the trainers didn't know how it works when the airline was in the process of apportioning blame to system design.

            Comment


            • Originally posted by MCM View Post
              I'm also reluctant to immediately accept the statements that the trainers didn't know how it works when the airline was in the process of apportioning blame to system design.
              In the report, the trainers are apportioning blame to 'anomalies' in system design when the system is simply acting as designed.

              WIthout a doubt, Boeing explains the modal interactions of FLCH SPD in the FCOM and makes it clear that this is not an appropriate mode for descent beyond the FAF. I'm not saying that Boeing has failed to do that. But here is an airline that clearly doesn't get it. It leads me to think that Boeing needs to explain these modes more intuitively in some way. I wonder if it has something to do with a cultural gap in understanding.

              The quote I posted from Aviation automation: The search for a human-centered approach is actually from a NASA article exploring ways to construct a more intuitive VNAV interface. VNAV, they argue, by transitioning 'silently' between speed-on-thrust and speed-on-elevator, can at times behave in a way that is counter-intuitive to the crew (despite the clear FMA indications), especially under stress. With FLCH, which is always speed-on-elevator, a properly trained pilot should know intuitively what to expect. I'm not saying there is anything wrong with FLCH. I'm saying there is something missing with pilot intuition, something not getting across, at least over there at Asiana, and I suspect in other places as well...

              Could Boeing make their training more focused on building autoflight intuition?

              Interestingly, in the report it is evident that at least some Asiana ground instructors make it clear to their pilots not to use FLCH on final, but even they don't seem to understand where the danger originates.

              Comment


              • Originally posted by Evan View Post
                Could Boeing make their training more focused on building autoflight intuition?
                Maybe, but that's a secondary issue IMHO.

                If you ask me, the proper course of action beginning somewhere between 4000 and 3000 feet would be (in the circumstances of this particular flight):
                1) Press "off" switch on AP and AT
                2) Place one hand on yoke, other hand on throttles, feet on rudder pedals
                3) Fly the airplane

                Or... go around. Although I'm not sure that would be such a good recommendation in this case, as attempting to go around in the midst of "automation mode confusion" could have caused an even worse outcome.
                Be alert! America needs more lerts.

                Eric Law

                Comment


                • Originally posted by elaw View Post
                  Maybe, but that's a secondary issue IMHO.

                  If you ask me, the proper course of action beginning somewhere between 4000 and 3000 feet would be (in the circumstances of this particular flight):
                  1) Press "off" switch on AP and AT
                  2) Place one hand on yoke, other hand on throttles, feet on rudder pedals
                  3) Fly the airplane

                  Or... go around. Although I'm not sure that would be such a good recommendation in this case, as attempting to go around in the midst of "automation mode confusion" could have caused an even worse outcome.
                  EXACTLY RIGHT....which takes us back to the "Children of the Magenta" video.
                  If it 'ain't broken........ Don't try to mend it !

                  Comment


                  • In the report, the trainers are apportioning blame to 'anomalies' in system design when the system is simply acting as designed.
                    If you were trying to shift blame away from your airline and towards the manufacturer, it would be essential to show that there was mass confusion, not just from one individual. Food for thought.

                    Interestingly, in the report it is evident that at least some Asiana ground instructors make it clear to their pilots not to use FLCH on final, but even they don't seem to understand where the danger originates.
                    I think that says a hell of a lot more about Asiana and its ground instructors than it does about the aeroplane. People who don't fly aeroplanes trying to teach others how to perhaps? Any 777 pilot with any experience would be able to confidently and thoroughly explain FLCH operation. And if he can't, you'd have to seriously question their abilities. Understanding the auto flight system is the most basic of skills.

                    I honestly don't know how FLCH could be any MORE intuitive to a pilot. It is basic, simple, and does what it says on the tin. Aircraft pitches for speed, and uses thrust if you want to climb, and no thrust if you want to descend (if the A/T is engaged). How much more f@#$ing simple and intuitive can it be? That said, it still has to be trained properly.

                    Regardless of all that - the fact they sat there with the thrust at idle while the speed decayed is just shocking. Remember Turkish in AMS? There's lots of reasons the Autoflight system can behave not as you expect, and the pilots MUST be able to deal with that.

                    The VNAV silent switch is another issue entirely - that one does crop up as it is definitely subtle. But FMA awareness solves the problem.

                    Comment


                    • Originally posted by MCM View Post
                      I honestly don't know how FLCH could be any MORE intuitive to a pilot. It is basic, simple, and does what it says on the tin. Aircraft pitches for speed, and uses thrust if you want to climb, and no thrust if you want to descend (if the A/T is engaged). How much more f@#$ing simple and intuitive can it be? That said, it still has to be trained properly.
                      I agree that FLCH is simple to understand, aside from the lingering effect on the AT after you disconnect AP, nor is that hard to understand if it is explained by someone who, in fact, understands it. What I'm referring to is an intuition for knowing when to use it and how to select it. For instance, because FLCH puts the AT into THR for climb and IDLE for descent, a pilot using it for descent should intuitively set the descent altitude, then push the button, then select the airpeed (right?).

                      Here's what I think happened: I think he was very stressed about losing face by losing a straight-in visual approach on a perfectly clear day. He desperately had to get back down on the glidepath and he wanted flaps badly, so intent was he on getting to flaps 30 that he called for them above the extension speed. WIth distance running out, he was desperate to lose speed and in his haste he intuitively hit the FLCH button to get the thrust to IDLE, so he could get flaps 30 and rapidly descend at a V/S exceeding 1000fpm until he could get back on glidepath. The problem as I see it is that his intuition was wrong. His learned intuition, I suspect, was that FLCH was the button you pushed to slow down and lose altitude in a hurry, until you capture the GS and all is well. A proper intuition would be to first select FLCH in descent rather than climb because it does very different things with the thrust in climb. One is the brakes and the other is the gas. Same button.

                      I think, ultimately, he balled it up because he wasn't monitoring the FMA, the AT or the V/S, and as you say, was not flying the aeroplane, but if he hadn't stabbed at the FLCH button intuitively, he could have stayed in automation, the AT would have not been holding at IDLE and he might have never heard another word about it.

                      Until one day, as a flight instructor, he mentioned that "wierd anomaly" in the autothrottle...

                      Aside from this, wouldn't speedbrakes have been a better idea at such a low altitude?

                      Comment


                      • Originally posted by Evan View Post
                        Could Boeing make their training more focused on building autoflight intuition?
                        What exactly is "autoflight intuition"?

                        Comment


                        • The initial actions were sound. If they'd have left the V/S at -1500fpm, and not brought it back to 1000fpm still about 500ft above the profile, we wouldn't be discussing this.

                          The PF clearly couldn't fly to a visual aim point, and the PM didn't continue to suggest he was high.

                          Speedbrake is probably not ideal down low, but could be used earlier if necessary. They had more than enough drag available out to solve the problem. Pressing FLCH was ridiculous at that stage of flight, even if he was trying to get down onto a glideslope, let alone a visual aim point.

                          This sort of vertical offset is not unusual. The V/S -1500 would have worked, as would have extending the gear if the speed was getting too high at that V/S setting. Intercepting the visual path, you simply reduce the V/S to something more appropriate (such as -800) and continue on your merry way.

                          By the time he pressed FLCH, it was time to either disconnect all the automatics, or go around.

                          Comment


                          • Originally posted by ATLcrew View Post
                            What exactly is "autoflight intuition"?
                            See MCM's post above

                            Comment


                            • I will give my vision, that doesn't match neither Evan's nor MCM.

                              Let's do a timeline based on the NTSB's animation.
                              Credit: National Transportation Safety Board (NTSB) The National Transportation Safety Board determines that the probable cause of this accident was the flig...

                              The first 20 seconds the video is frozen. I'll take this point as "time = zero seconds".
                              At t=0, when the animation starts, the plane is at 3460ft, 187kts, 8.6NM from the threshold, flaps 5 and in SPD:172 / VS:-1000, and the pilots select gear down.
                              The TLs are not fully retarded, I am not sure why. I know that in this mode the AT aims for a rate of speed gain/loss that is function of the difference between the selected speed and the actual speed, but the rate of deceleration looks too slow IMO for a 15kts gap. In fact, the plane gains 2 knots in 5 seconds before it starts to slow down again, and this slow down I'm sure is the result of added drag of the landing gear extending, to the point that the AT does't "like" it and starts to advance the TL to reduce the rate of deceleration.
                              The PM says "It looks a little high" and the PF replies "I will descend more".
                              By this time (t=15) the plane is still at 185kts and the PF starts to increase the selected VS to -1500.
                              The TLs move back, once again close to idle but not fully there.
                              The plane keeps slowly loosing speed, and as it does so the AT slowly start to move the TLs forward again (consistent with reducing the deceleration when the speed approaches the selected one).
                              At t=42 the pilot selects the speed back to -1000, maybe because he wanted to slow down faster in order to add more flaps, they are at 178 kts (just 6kts above the selected speed) and still with 5. I don't know what's the limit for 20, but here we have a clear contradiction: It's impossible that 180kts is too fast for flaps 20 but 172 is the lowest you can select with flaps 5. 8 kts of overlap between flaps settings is too little. So the pilot should have been able to select either a lower speed (that would have increased the gap and hence the deceleration aim of the AT) or a higher flaps setting to add drag earlier.
                              At t=55 the pilot selects 3000ft for the GA altitude. The plane is at 176kts and the TLs are more forward. The selected speed is still 172kts.
                              At t=74 the speed has been stabilized at 175 kts for a bunch of seconds, and the pilots select flaps 20. The selected speed was still 172kts but immediately after selecting flaps 20 the PF selects 152kts. Almost immediately
                              after that the pilot calls for flaps 30, but the PM replays that they are still too fast.
                              The new speed gap makes the TL move back again close but not reaching idle. The rate of loss of speed increases, it seems like more than what the AT liked because it advances the TLs a bit.
                              At t=97, with the plane at 171kts, the pilot selects FLCH. The AP pitches up a bit to slow down and the AT advances the TLs a bit "thinking" in a sustainable climb to 3000ft.
                              Almost at the same time (t=99) the PM extends the flaps to 30.
                              One second later (t=100) the PF disconnects the AP, but leaves the FD on, so the pitch mode remains in FLCH. The plane is now at 168kts (selected still 152), so the combination of very slight pitch up and flaps extension made it lose 3kts in 3 seconds despite the TLs moving forward (probably the engines were still spooling up, althoug there was not much to spool up there because the TL's, even if more advanced, are still in the "close to idle" zone).
                              At t=102 the pilot manually retards the throttles. The speed is 168kts and the altitude 1500ft.

                              Stop. We know the rest.
                              This is the first time in all the animation that the TLs are at idle.

                              And this is, perhaps, the most obvious and intuitive opportunity that the pilot lost. I've personally seen several times (not a lot, but only because I wasn't in a cockpit a lot of times) that the pilots briefly override the thrust levers.

                              My problem is that I find it hard to slow down and descend? I think that the most natural move for a pilot is grab the throttles and check that it's fully back, and if if it's not then bring and hold it back, yes, with the AT still on.

                              The more that I see it, the more convinced I am that they lost lots of opportunities to loss speed.
                              - They extended the landing gear much later than what they could have.
                              - During many second they kept a small gap between the selected speed and the actual speed, without neither reducing the selected speed not adding more flaps, what made the AT add thrust.
                              - During 102 seconds (in the animation, it surely started earlier than the animation) they held a thrust above idle.
                              - They increased the descent rate while still to fast. While they needed both descend and slow down, when you need to dirty the plane but are too fast for that, then FIRST slow down, THEN dirty up, THEN descend. Every time you spend clean you are losing the opportunity to get rid of energy. When dirtier you can achieve a higher sink rate without increasing the speed, a quicker deceleration at one gives sink rate and, not many take this into account, a steeper glide slope at one given sink rate. For example, at 180kts that they had at the beginning, 1000fpm equals to 3.1°; at 140kts that was about the flaps 30 Vapp, the smae 1000fpm equals to 4.0°, making it much faster to descend to the correct glide path.
                              - Finally, yes, they could have used some spoilers, especially during the first 74 seconds that they were with flaps 5 (I found a comment in the internet that the FCTM recommends against significant speedbrake extension with flaps in excess of Flap 5). But maybe the company's SOP was not to use spoilers with flaps at all.

                              Now, I think that this would have been quite easy, intuitive, and would have removed workload.

                              PM: "We are a little high."
                              Me: (I note that yes, we are high and, not only that, we are fast too).
                              Me: Pull back on the TLs all the way to idle, and keep them there "fighting" against the will of the AT that wants to move them forward. I win.
                              Me: Select the minimum selectable speed for the current flaps 5.
                              Me: I leave the AP in VS -1000 and the AT in SPD.
                              Me: without abandoning the instruments scan, I pay more attention to mainly 2 things: the airspeed and the VASI/PAPI.
                              Me: As soon as we've slowed down to a speed suitable for flaps 15, I call "Flaps 15" (yes, that is not a flaps setting normally used for approach, so what).
                              PM: "Speed check, flaps 15."
                              Me: Select the minimum selectable speed for flaps 15.
                              Me: As soon as we've slowed down to a speed suitable for flaps 20, I call "Flaps 20".
                              PM: "Speed check. Flaps 20."
                              Me: Select the minimum selectable speed for flaps 20.
                              Me: As soon as we've slowed down to a speed suitable for flaps 30, I call "Flaps 30"
                              PM: "Speed check. Flaps 30".
                              Me: I select Vapp (137 kts).

                              If during this sequence, I reach the "2 red / 2 white" condition (unlikely given the initial condition), I select VS -700 fpm and keep pulling back on the TLs until the speed is approaching Vapp.

                              If not (most likely), upon approaching Vapp I'll release the TL's and leave them back to the AT (checking that it's still in SPD, as it surely will) and judge the altitude situation:
                              - If we are not so bad ("1 red, 3 white), I'll likely leave it at 1000 for a while more (now it's 4°, not 3.1° as before).
                              - If not ("4 white"), I would go around if too horrible or, if it looks "salvageable", I might increase it a bit for a few seconds (-1500ft will give me 6.1° at this slow speed!!!, and the speed will not increase because I'm very dirty, although the thrust will be very likely below the minimum for the stabilized approach, but we'll care about that later), when I have "1 red, 3 white" I'll re-set to -1000, and when we have "2 red, 2 white" I set -700.
                              - If we rach 500' and the approach is not stabilized yet (including that the speed is STABLE at about Vapp and the descend is STABLE at "2 red. 2 white" and the thrust is STABLE at an appropiate value (which I don't know what it is but certainly excludes anything close to idle), I go around.

                              I know the above looks too complex, but it's not. I'm concentrating on just two tparameters (without forgetting the rest), airspeed and PAPI/VASI, and I touch only with 3 controls, the thrust levers (that I just keep back until close to Vapp), the vertical speed (that I keep I either keep at -1000 until "2 red / 2 white", or I can briefly increase a bit beyond -1000 if I've already slowed down and I'm still way too high), and the MCP speed which is the only control that I play several times during the approach with a loop "Slow enough? ==> More Flaps ==> Select Slower"). I don't play changing automation modes (not to mention disconnecting them), I don't relay on a protection or spooky "wake up" for the AT to save the day (the AT is always in SPD and will track the selected speed as soon as I let go the TLs), and it will give me the easiest transition at any moment that I want to do "click click, clack clack, I've got the airplane".

                              I've done it dozens of times!!! (in Microsoft Flight Simulator )

                              --- Judge what is said by the merits of what is said, not by the credentials of who said it. ---
                              --- Defend what you say with arguments, not by imposing your credentials ---

                              Comment


                              • Originally posted by Evan, MCM & Gabriel
                                They weren't trained, No, they were overloaded, Well my version is...many pages later..."
                                Dittos on Kickers popcorn- but I'm enjoying the debate. Parlour talkers trying to figure out how in the hell do three 777 pilots wind up out of energy, altitude, airspeed and ideas?

                                I originally thought it was tiredness- based on the fact that running out of those four things does happen when you don't pay attention and it's caused more than one crash before (but usually not top of the line airliners).

                                And monitoring airspeed seems so basic, but if the autothrottles have 'always' been there for you...

                                But MCM has poured over the details and this discussion leaves me with two comments:

                                1) Perhaps the a lack of training is a key reason they got overloaded.

                                2) And amazingly, they should not have been overloaded. I understand that landing a plane is critical- but one hand on the yoke, one hand on the throttles and a good scan of the front window and VASI, ASI, VSI, altimiter, AI, heading (or the data on the flat screen in the same place) on a beautiful sunny evening with light winds should not be an overload.

                                Even if you are high and fast, it's the same- monitor the view out the window, and the speed and the vertical speed and aim at the runway and remember when to call it quits- when (except for Boeing Bobby) it is not a major deal to go around.

                                It is interesting that switching off the automation will reduce your workload in certain situations.
                                Les règles de l'aviation de base découragent de longues périodes de dur tirer vers le haut.

                                Comment

                                Working...
                                X