Announcement

Collapse
No announcement yet.

At about 50 feet AGL you'll receive some clips around the ears...

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

  • At about 50 feet AGL you'll receive some clips around the ears...

    Almost there... alllmost there... http://avherald.com/h?article=4b054ccf&opt=0
    Attached Files

  • #2
    I think I like some operational limits on crosswinds...

    Edit:...and maximum wind speeds and gusts.

    It seems lots of folks were going around, and the guy who landed...it was just almost pushing that NEAR TOTAL DISASTER!!! sort of deal...sort of like it was simply too much wind for folks to consider it safe to land...at least that's what it looks like from my keyboard.
    Les règles de l'aviation de base découragent de longues périodes de dur tirer vers le haut.

    Comment


    • #3
      Well the good news is the impact with the ground did level the wings nicely!
      Be alert! America needs more lerts.

      Eric Law

      Comment


      • #4
        so this is a serious question: is auto land better capable of landing in high cross wind/gusty conditions than the pilot with a heartbeat?

        Comment


        • #5
          My understanding is the answer to that question is almost always "no". When the going gets tough, in general the autopilot gives up and hands control back to the pilot.
          Be alert! America needs more lerts.

          Eric Law

          Comment


          • #6
            Originally posted by TeeVee View Post
            so this is a serious question: is auto land better capable of landing in high cross wind/gusty conditions than the pilot with a heartbeat?
            Adding: I THINK autoland is designed for zero visibility which usually equates to very calm conditions AND generally should not be used in windy conditions as it really can't properly handle them.

            Hopefully outsider engineers or insiders can confirm and clarify.
            Les règles de l'aviation de base découragent de longues périodes de dur tirer vers le haut.

            Comment


            • #7
              Originally posted by TeeVee View Post
              so this is a serious question: is auto land better capable of landing in high cross wind/gusty conditions than the pilot with a heartbeat?
              Autoland requires ILS. This was a non-precision instrument approach. That means APP mode (required for both autopilots to be engaged), and thus autoland, is not available.

              Also, the crosswind limits for autoland on most NG's is 25kts. Unlike more sophisticated jets, the NG autoland cannot decrab and execute rollout after touchdown thus it is more limited in crosswinds (newer or upgraded NG's MIGHT have this capablity).

              This was either a manual approach or a single-channel autopilot approach transitioning to manual before flare followed by a manual go-around (FD is available). It looks like it was perfectly executed however. The engine spool-up noise comes a moment after impact and the flaps begin retracting within a few seconds.

              Comment


              • #8
                F.A.O.: Gabriel

                Can an autoland (and I'm talking a somewhat different system) be designed to deal with nasty, gusty winds and out do a human?

                Part of me sees that 'instant response to a deviation' might cause a computer to win out.

                ANOTHER part of me sees that whomever is driving cannot see the wind gust before they are encountered, and the low-silicon autopilots are often using near-full control inputs...can the computer truly out do them?

                ...and note that there was a Domestic, Super-Genius, Pilot of the Millennium who stated (I'm paraphrasing) that you don't want to aggressively kill every deviation as the wind Gods frequently tend to undo what they just did...(seems like a bad job for HAL).

                Back to tee vee's comment, I am reminded that a lot of current autoland systems don't really do crossed control inputs (or maybe they don't do rudder inputs) which you need typically need/want to compensate for crosswinds...(I THINK).
                Les règles de l'aviation de base découragent de longues périodes de dur tirer vers le haut.

                Comment


                • #9
                  Originally posted by elaw View Post
                  Well the good news is the impact with the ground did level the wings nicely!
                  So, it would appear to be a great landing?
                  Les règles de l'aviation de base découragent de longues périodes de dur tirer vers le haut.

                  Comment


                  • #10
                    Originally posted by 3WE View Post
                    Adding: I THINK autoland is designed for zero visibility which usually equates to very calm conditions AND generally should not be used in windy conditions as it really can't properly handle them.

                    Hopefully outsider engineers or insiders can confirm and clarify.
                    Parlour-talking outsider says you are correct. Though TeeVee is certainly asking a reasonable question. Autoland is not intended for use with gusts and crosswinds but theoretically if ILS is present (along with a fail-operational triplex-autopilot not commonly found on the 737) the AP should be better than the human pilot at compensating for upsets without overcontrol. We just aren't there yet...

                    Anyway, I'm not sure ANY high- or low-silicon pilot would successfully land in these conditions. Hopefully, the AP of the future would auto-go-around.

                    Comment


                    • #11
                      Originally posted by Evan View Post
                      ***the AP should be better than the human pilot at compensating for upsets without overcontrol.***
                      Better for 'instantly' responding to deviations...not sure on the over control and read F.A.O. Gabriel above where I paraphrase a Super Genius insider who made an awfully valid comment.

                      Would HAL be too quick to fix X only to have the wind Gods hit you with -X?...where the Cowboy improviser might let X happen on a really good hunch (which worthless 100 hour PPL can confirm) that -X (+/- some error) is coming within the next 3 seconds to largely cancel it out...
                      Les règles de l'aviation de base découragent de longues périodes de dur tirer vers le haut.

                      Comment


                      • #12
                        Originally posted by 3WE View Post
                        Can an autoland (and I'm talking a somewhat different system) be designed to deal with nasty, gusty winds and out do a human?

                        Part of me sees that 'instant response to a deviation' might cause a computer to win out.

                        ANOTHER part of me sees that whomever is driving cannot see the wind gust before they are encountered, and the low-silicon autopilots are often using near-full control inputs...can the computer truly out do them?

                        ...and note that there was a Domestic, Super-Genius, Pilot of the Millennium who stated (I'm paraphrasing) that you don't want to aggressively kill every deviation as the wind Gods frequently tend to undo what they just did...(seems like a bad job for HAL).

                        Back to tee vee's comment, I am reminded that a lot of current autoland systems don't really do crossed control inputs (or maybe they don't do rudder inputs) which you need typically need/want to compensate for crosswinds...(I THINK).
                        In principle and in theory, it is possible to design an AP that performs better than a human pilot at the technical stuff of actuating stick and rudder (and throttles). All the inputs that the human pilot gets (linear and angular accelerations, velocities, and positions, changes in dynamic pressure and static pressure, radalt, GPS and navaids read outs, engine parameters, etc) can be fed in the AP and the AP can recognize them, read them, and react to them much faster and accurately than a human pilot, the model of the engines'and airplane's response can be fed in the AP who would then have a much better "understanding" of the expected response to a given input, and in turn of the input needed to get an expected response, and the AP can much more quickly and accurately asses the airplane's actual response to its inputs and compare it against the expected response, asses the difference and adjust the model and the inputs (adaptive control). Even if you don't want to fight against the random variation of every jolt, you can also program that into the AP to adjust how much it compensates for deviations (P for proportional), trend (D for derivation), and accumulation (I for integration). This exists since way long ago, with analogic controls (and digital ones too of course) called PID. The formula for an old PID control is just A=-a*(X-Xo)-b*(X'-X'o)-c*integral/-to/0/((X-Xo)dt), meaning that the amplitude of the input A (for example aileron deflection) is in part proportional and opposite to the difference between the real bank (X) and the target bank, in part proportional and opposite to the difference between the actual roll rate (X'=dX/dt) and the desired roll rate, and in part proportional to "how much and how long" has the bank been cumulatively different from the target bank in the last to (tee zero) seconds. Adjusting the proportionality constants a, b and c, you can adjust how much the AP takes into account each of these deviations from targets. Modern controls, especially those that are purpose-built for a given known dynamic model, can be much more complex, taking higher order of derivation and integration, higher powers of the dynamic variables (like X^2), thresholds and caps, and adaptive variation of the target values, and all that can be in function of things like airplane weight and moment of inertia (that are variable but known for each moment of the flight), air density (so calibrated and true airspeed differences are taken into account), and so on.

                        However, there are things that an AP can't do (and neither a human pilot can): Basically, and AP cannot violate the laws of Physics. And of control theory: there is so much you can do when there are delays between a variable changing and the AP becoming "aware" of the change, when there are delays between the AP commands and the actual response of the controlled variables -elevators don't move instantly, engines have to spool up, etc), and limitations in the resolution, accuracy, and noise-to-signal ratio of all the measured variables (collectively known as uncertainty of measurement). Going to the Physics side, if you receive a negative-performance windshear that leaves the plane moving at an indicated airspeed that is below the "official" stall speed, the AP (as well as the human pilot) has the choice (or the programmer had) between trying to keep the vertical speed at the target, stalling and increasing the sink rate, or lowering the nose and increase the sink rate without stalling. In any event, the sink rate will increase no matter how good the human or auto pilot.

                        Another thing that I feel computers are not so good at, yet, is at looking at the big picture to make a judgement and define a strategy. "Hmmm, these clouds next to the approach path look threatening, the rain is increasing, I saw some lightning bolts nearby some seconds ago, the previous plane reported moderate turbulence and wind shear, and I remember the lessons learned of Delta 191, I better enter a hold until things clear out a bit".

                        Finally, given that a) the auto-land is intended for very low visibility conditions, b) that these conditions typically are not accompanied by strong gusting winds, and c) that, for the reasons stated above, an approach can go wrong even with the best autopilot that is even theoretically possible (and even if that best-theoretically-possible AP and all related systems are working 100% correctly) it is better for the airplane manufacturer to get rid of possible liability issues. While this perfect AP would be better than a human, a human can be judged for poor decision making, poor judgement, bad training, bad skills, startling and confusion, etc, etc, etc... while an AP can only be blamed for bad design, which would be the manufacturer's fault.

                        And of course, let's not forget that that most perfect theoretically possible AP which is also infinitely reliable doesn't exist. We are just there yet with the realization of the full potential of what we theoretically know can be done.

                        --- 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


                        • #13
                          Originally posted by Gabriel View Post
                          Another thing that I feel computers are not so good at, yet, is at looking at the big picture to make a judgement and define a strategy. "Hmmm, these clouds next to the approach path look threatening, the rain is increasing, I saw some lightning bolts nearby some seconds ago, the previous plane reported moderate turbulence and wind shear, and I remember the lessons learned of Delta 191, I better enter a hold until things clear out a bit".
                          I don't think you are keeping up with AI. All of this 'learned experience', complex situational analysis and predictive 'thinking' is within the realm of possibility and I think soon we will be talking about 'artificial intuition' and 'artificial anxiety'. But as I said, we just aren't there yet, and the 737 is barely in the 21st century.

                          Finally, given that a) the auto-land is intended for very low visibility conditions, b) that these conditions typically are not accompanied by strong gusting winds, and c) that, for the reasons stated above, an approach can go wrong even with the best autopilot that is even theoretically possible (and even if that best-theoretically-possible AP and all related systems are working 100% correctly) it is better for the airplane manufacturer to get rid of possible liability issues. While this perfect AP would be better than a human, a human can be judged for poor decision making, poor judgement, bad training, bad skills, startling and confusion, etc, etc, etc... while an AP can only be blamed for bad design, which would be the manufacturer's fault.
                          Very true. In the end, it is the lawyers that drag us down.

                          Comment


                          • #14
                            Originally posted by Evan View Post
                            I don't think you are keeping up with AI. All of this 'learned experience', complex situational analysis and predictive 'thinking' is within the realm of possibility and I think soon we will be talking about 'artificial intuition' and 'artificial anxiety'. But as I said, we just aren't there yet, and the 737 is barely in the 21st century.
                            Modern AI is not designed for big picture strategy. It is only as good as the dataset it is given ahead of time. You have to train the AI with a lot of real world examples and then have a human tell it which datasets were good and which ones were bad. There are newer AI's that can make decisions based on smaller sets of data, but these are very limited right now.

                            Humans will be landing planes in difficult conditions for a long while yet.

                            Comment


                            • #15
                              Originally posted by 3WE View Post
                              Adding: I THINK autoland is designed for zero visibility which usually equates to very calm conditions AND generally should not be used in windy conditions as it really can't properly handle them.

                              Hopefully outsider engineers or insiders can confirm and clarify.
                              Every a/c type has its own limits concerning cross wind conditions,
                              completely independent from the frequencies that the airport infront of you or your favorite airport is able to provide.

                              That's the reason why Hurricane Irma in September 2017 led to the closure of not only
                              Daytona Beach DAB - why I know this airport, with a rater high percentage of General Aviation, we'll see later..
                              or
                              Tampa TPA.

                              Even a 747 has limits. Thus, LH 464 was not only cancelled on Monday September 11th 2017 by LH. Also our host for LH 464,
                              Orlando MCO,
                              cancelled ALL flights for that day.

                              So, you can almost guess why I also know DAB, TPA, and MIA. All those strips are long enough for my avatar. But that Monday, Florida was completely closed.
                              Thus, LH 464 did not take off that day.

                              Now, what are these limits. Me personally regularly switches off the a/t with e.g. a wind of 350 @ 18, or 20 in gusts, if you fly 010 or 020. Why. Because in these conditions, the pfd spd cursor tends to jump like a monkey with a coffeine overdose. And if you leave a/t on, the computer and all of your four engines exactly behave like that monkey, it tries to compensate the gusts instead of holding the N numbers constant.

                              But, as 3WE said, this is only valid for me, an outsider who only sometimes operates a pmdg LH-B744 fsx with real flight numbers and real weather across the pond.

                              Don't know what the answer of an insider would be.
                              The German long haul is alive, 65 years and still kicking.
                              The Gold Member in the 747 club, 50 years since the first LH 747.
                              And constantly advanced, 744 and 748 /w upper and lower EICAS.
                              This is Lohausen International airport speaking, echo delta delta lima.

                              Comment

                              Working...
                              X