27
$\begingroup$

  • Why does the pilot make so many movements? I don't see that plane is moving all that much, but his yoke movements are strong and fast. For someone who is not a pilot, it almost looks like panicked flying.

  • Can airliners land with auto pilot during the day with strong winds and does night make landing even worse in these gusty conditions?

$\endgroup$
3
  • $\begingroup$ StephenS’ answer is so much better than my deleted comments. Michael Hall’s answer adds a lot to it to explain that large inputs at slow speeds don’t have as much of an effect as smaller inputs at much higher speeds. And, with such a large aircraft, there is a lot of inertia and momentum to counter. $\endgroup$
    – Dean F.
    Sep 21 '20 at 21:12
  • 3
    $\begingroup$ Watch a Formula 1 driver on cockpit cam along the straights. Their steering wheel action is just as chaotic, because they are doing the same thing, preempting the effect of minor aberrations from the course they need the vehicle to follow. $\endgroup$
    – Nij
    Sep 23 '20 at 5:24
  • $\begingroup$ Re -- "Why does the pilot make so many movements? I don't see that plane is moving all that much, but his yoke movements are strong and fast." Good thing he had such a heavy watch on his wrist, or his hand would have moved even more, making all the passengers sick. $\endgroup$ Sep 23 '20 at 20:26
43
$\begingroup$

To the first part of your question, the airplane is not moving BECAUSE his yoke movements are "strong and fast". Experienced pilots can feel a gust and respond with a control input to counter it before it has enough time to affect the aircraft. (You really can't appreciate this from a camera fixed to the airframe.) As mentioned in a comment, this is called "staying ahead of the airplane" or "flying by the seat of your pants". Quick and correct reactions are required to keep the approach on track when being buffeted around.

Another thing that is worth pointing out is that the magnitude of control inputs is higher at slower approach speeds because of the reduced effectiveness of the control surfaces. You wouldn't want to move the controls that much at cruise speed. Just like when driving on a bumpy road you move the steering wheel a lot more than when cruising on the freeway.

I have never flown an airplane with auto-land capability, so I cannot answer the second part. However, given the experience I do have with a couple different autopilots, my belief is that even the best systems would be stressed trying to keep up while landing in very gusty conditions. There are just some things you still need a trained and experienced human for...

$\endgroup$
8
  • 1
    $\begingroup$ For the auto-land I think that the autopilot might struggle with the drift induced by crosswind, and fail to correct the orientation of the plane just before or after touchdown. Otherwise he might be way better than us to react to gust wind given the proper sensor are available. $\endgroup$
    – MaximEck
    Sep 21 '20 at 20:42
  • 3
    $\begingroup$ Auto-land has a maximum wind component usually well below the max demonstrated crosswind. From googling it seems the 737 is limited to 15kts, 777 is 25kts. $\endgroup$
    – Ben
    Sep 21 '20 at 21:22
  • $\begingroup$ quick question, at 1:25 the pilot takes his hands off, has the copilot taken over at that point? $\endgroup$
    – Kai
    Sep 22 '20 at 14:25
  • 3
    $\begingroup$ @Kai The yoke isn't used for steering on the ground. That's done through the foot pedals and, for larger aircraft, a separate tiller control. $\endgroup$ Sep 22 '20 at 21:39
  • 1
    $\begingroup$ @Kai However, you need to keep using the yoke and rudder until the control surfaces stop working due to low airspeed. A common private pilot mistake is to let go of the yoke/rudder too soon while airflow still has effects. $\endgroup$ Sep 23 '20 at 6:38
21
$\begingroup$

Why pilot make so much input,I dont see that plane is moving even his yoke movements are huge and fast?

The plane isn’t moving much because as soon as the pilot senses either wing lifting, he is making a correction to counteract it. Since he is flying in a gusty crosswind, that means many such corrections are required to keep the plane stable.

For person who not fly,it looks panic.

To someone who does, it looks fairly normal. Gusty winds and crosswinds are something that pilots train for almost from day one. It does take practice to develop this skill, which is why student pilots aren’t allowed to fly alone in strong winds, but a licensed pilot will know what they (and their plane) can safely handle and will go elsewhere if needed.

Can airliners land with auto pilot on strong windy day

In theory, yes, but it will almost never be used. Pilots need to keep their skills sharp, so they will usually hand-fly the most challenging parts of a flight. The autopilot is there mainly for the boring parts or when the pilots have other important tasks to focus on.

and does night make landing even worse in this gusty condition?

Not really. If anything, night makes it easier to see the runway approach lights.

Notice the wide horizontal elements to the approach lights, which are there specifically to assist the pilot in keeping the plane wings-level during conditions like this.

$\endgroup$
2
  • $\begingroup$ I thought airline crews were required to use autoland every now and then just for currency? And also required to use it below certain visibility conditions? $\endgroup$
    – reirab
    Sep 24 '20 at 21:39
  • $\begingroup$ @reirab For Cat II/III approaches, if the crew/plane/carrier are certified for that, yes. That’s the “almost” in “almost never”. $\endgroup$
    – StephenS
    Sep 24 '20 at 21:46
18
$\begingroup$

As already said in the other answers, the "huge" inputs to the yoke are to keep the aircraft stable on the approach path. There is no "panic" involved, it is simply necessary due to the gusty conditions.

Let me expand a bit more on your second question:

Can airliners land with auto pilot on strong windy day [...]?

The answer to this a clear no. First, there are the aircraft limitations to consider. An autoland has much stricter wind limitations than a normal landing:

Maximum allowable wind speeds, when conducting a dual channel Cat II or Cat III landing predicated on autoland operations, are:

  • Headwind 25 knots
  • Crosswind 20 knots

Cat IIIb:

  • Crosswind 25 knots
  • Tailwind 10 knots.

(Boeing 737NG FCOMv1 L.10.5 - Limitations - Operating Limitations - Autopilot/Flight Director System)

On a strong windy day these limits can easily be exceeded during a strong gust. The YouTube video description of the video you show says:

Wind reported 340°, 20 knots gusting 35 knots.

The approach shown in the video was to runway 26L at EGKK/LGW (London Gatwick), which has a runway heading of 258°. That means a wind gust of 35 knots from 340° results in: $$ \text{Headwind} = 35 \, \mathrm{kt} \times \cos(340^\circ - 258^\circ) \approx 4.9 \, \mathrm{kt} $$ $$ \text{Crosswind} = 35 \, \mathrm{kt} \times \sin(340^\circ - 258^\circ) \approx 34.7 \, \mathrm{kt} $$ This exceeds the limitations for autoland.

But there is a more important reason why you could not perform an autoland in the conditions shown in the video: you need a Cat II or Cat III ILS approach for autoland. Gatwick has an ILS up to Cat IIIb on runway 26L, but that does not mean the airport is operating in Cat II/III mode. This mode requires larger separation between aircraft as well as larger distance of aircraft and all vehicles on the ground to the runway. Furthermore, the ILS antenna need to be running on diesel generators. This results in a reduction of airport capacity and higher operating cost and will therefore only be done when absolutely necessary, meaning when the visibility is too poor for a normal Cat I ILS approach. The minima for Cat I for runway 26L are

DA(H) 396'(200') with RVR of 550m.

Such low visibility rarely occurs together with strong winds because the wind will cause fog to disperse or prevent it from forming in the first place.

$\endgroup$
3
  • 1
    $\begingroup$ Is there a specific reason why autoland is only allowed up to a certain wind speed? I would assume that a good control loop can react quicker and with less overshoot than a human. $\endgroup$
    – Michael
    Sep 22 '20 at 19:42
  • 3
    $\begingroup$ @Michael The autopilot can certainly react quicker, but for an autoland the most important aspects are accuracy and redundancy (in case of partial system failure). A sudden strong correction could be to counteract a wind gust, but also due to a disturbance of the ILS signal. The actual limits are probably derived from what was demonstrated during certification. I guess one could certify autoland for stronger winds, but there is really no point if the visibility is good enough for humans to land anyway. $\endgroup$
    – Bianfable
    Sep 22 '20 at 20:06
  • $\begingroup$ @Michael There's one more thing to consider: go-arounds. The goal of an auto-pilot is always to land, while a pilot can decide to go around. If there is an especially strong gust of crosswind, the autopilot will always correct, no matter how much this would upset the plane otherwise. An experienced pilot is able to think "Screw this, I'm unable to correct this in a safe manner, I'll apply full throttle and go around". $\endgroup$
    – Emil Bode
    Sep 24 '20 at 15:41
13
$\begingroup$

Why pilot make so much input,I dont see that plane is moving even his yoke movements are huge and fast? For person who not fly,it looks panic.

Look at the artificial horizon. It has a scale at the top that allows you to see even small variations in bank angle. When looking there, you'll see that the plane IS actually moving a lot. E.g. at 00:44 in your video you see that the wings are level and within that second that plane suddenly banks to the right without any obvious reason so the roll was probably induced by changes in wind. The pilot immediately makes a roll input to the left. If you pause the video to go through the situation really slowly, you can also see this quite well, also with the "real" horizon.

Note, that the effectiveness of rudder deflection is reduced with reduced airspeed. The slower the airplane gets, the bigger the inputs got to be in order to achieve the same roll/jaw/pitch rate. In a traditional fly-by-cable airplane a bigger rudder deflection means more yoke/stick input is needed. However while flying slowly means that the rudder has less authority it also means less air resistance, which means that it requires very little force to move the control surfaces. In a real classic low-tech airplane (such as a Cessna 152) you'd actually feel the air resistance directly and when flying slowly, the yoke moves easily quite a lot with only a minimum force required. To apply the same yoke movement at high speed, you'd need a lot more force and the airplane would move way more erratically.

In contrast to a Cessna 152 the flight control systems a of 737 are somewhat more complex, involving hydraulics and an artificial feel system, but it is all built to feel similar to a very simple plane like the C152. This means that while the large inputs make it look like the pilot is going to have aching muscles after that landing, he won't because he needs only very little force and thus there is nothing "panic" about that.

Aside from that it is true that there are a number of pilots that make more inputs than necessary. From watching the video I cannot tell whether some of the inputs shortly before touchdown might have been unnecessary. However this is something that is often found in fly-by-wire airplanes (for example most Airbus planes, B777, B787,...). In contrast to classic airplanes the pilot (normally) does not command the control surface deflection by moving the yoke/stick, but rather commands a certain behavior (e.g. roll rate) and a computer will calculate and command the appropriate rudder deflection needed. That means that a fly-by-wire system "in theory" can fly with wings level continously even in gusty weather without pilot input. Of course in reality these systems are far from perfect and thus the pilot will still need to make inputs, however significantly less inputs are needed in contrast to fly-by-cable. Since most pilots received their original flying training on conventional planes this often results in overuse of yoke/stick.

Can airliners land with auto pilot on strong windy day

The auto land system is historically designed to make it possible to land the airplane in bad visibility, where a human pilot cannot land manually, simply because he cannot see anything outside. This usually means fog and these weather conditions usually come with calm weather as wind tends to clear the fog. While to auto land system can typically work with some crosswind and some gusts, limits are much lower compared to manual landing. This is not necessarily because it would be very hard to design a computer that could do this. Designing, implementing, testing, certifying such a system is expensive as hell and it is simply not needed: Performing an auto land means that the radio signal used for navigation (ILS) must be much more accurate than for a visual landing. However metal reflects radio waves and disturbs accuracy. Thus for an auto land it means that no other metal should be near the runway. That in turn means that no planes may cross the runway shortly before your landing and plane that landed previously must have cleared the runway area earlier than usual. This of course degrades the capacity of the airport and thus auto land is usually not used unless required due to weather. Since "required due to weather" and "lots of (gusty) wind" seldomly fall together, why design such a system?

What can actually happen when you use auto land anyway, although low visibility operation is not in progress at the airport and some other metal disturbs the radio signal? On 1st November 2011 Singapure Airlines demonstrated that for you in Munich. Her is the official report. It is in German, however look at the pictures on page 28-30. Additionally here is a video on YouTube covering the landing.

and does night make landing even worse in this gusty condition?

Landing at night may require more concentration, especially when the airport has only a very basic lighting system as is the case on many very small airports or in regions where low visibility is very rare and thus a basic lighting system is deemed sufficient. Since gusty conditions also requires extra brain cycles the combination might lead to a quicker depletion of brain cycles, however usually this is not a "problem".

$\endgroup$
11
$\begingroup$

I think it is interesting to look at this problem from a control engineering perspective.

Towards the runway, it is more important to track the centerline and glideslope. In control terms, better reference tracking is achieved by a higher control bandwidth ('crossover frequency').

As a crude first approximation, an airplane is a rotational inertia system and the control inputs act as moments on the airplane. This means that the transfer function from control input to attitude angle goes with $1/s^2$, with $s$ the (complex) frequency.

Since in the final stage of the landing, the pilot wants to control to a higher frequency. Due to the aircraft dynamics, this means they have to make quadratically larger inputs to achieve the same effect. This explains why you would see slow, small control inputs during cruise and large, quick control inputs during approach. (A non-pilot may compare flying to driving a car. However, in a car the steering angle directly controls the yaw rate of a car (until slip), so we have a transfer function of $1/s$ as opposed to the above $1/s^2$ so rapid steering inputs are undesirable)

Autoland limitations are usually more stringent than hand-flying in terms of allowable wind conditions. This is not because an autopilot would be incapable of handling the situation (in fact I daresay the oppositie might be true) but because the pilot must at all times be capable of monitoring the autopilot and initiate a go-around if necessary. A well-designed controller will manage to keep the aircraft on the glideslope until the physical limits of the aircraft are reached with disastrous consequences, whereas a human pilot will have initiated a go-around well before that point is reached.

$\endgroup$