Why would a plane descend rapidly in uncontrolled manner after a steep turn?

Question

Here's some footage from 2010 Red Bull Air Race. At 0:17 the plane is passing through a gate and then makes a steep turn to the left and that requires the plane to have left wing pointing almost downwards and right wing pointing almost upwards - this is how such turns are typically done. Right after the turn ends - at about 0:18 - the plane just starts falling down. It's only the pilot's skill which prevents it from entering water with left wing pointing down - the pilot promptly levels the airplane such that wings are now horizontal and this mostly stops the incident from developing.

How is this turn different from "good" turns which don't end up with plane falling down?

Answer 1

Markus Voelter interviewed Matt Hall for an episode of his omega tau podcast. They discuss this incident in more detail at 1:13:00.

The aircraft experienced a high speed g stall, also called an accelerated stall. Stall is determined by the angle of attack, which is the angle of the airflow hitting the wing. Stall is most often thought of at low speed, because as the plane goes slower, it takes higher angle of attack to provide enough lift, until it reaches the stall angle of attack and stalls. But this can also happen at high speeds in an aircraft performing a high g turn. Much higher lift is required to make the airplane turn, meaning much higher angle of attack is needed. Trying to turn the aircraft too tightly can increase the angle of attack too far, and the airplane stalls at a high speed.

The other factor is that the airplane will tend to roll left due to torque from the engine and P-factor of the propeller. Since the plane was already in a left hand turn, this plus the stall made the airplane roll past 90 degrees and head down towards the water.

Another aggravating factor was the airplane's stability, which is discussed at 1:31:00. The CG was fairly aft, making the plane less stable. The elevators were also over-balanced, meaning they would tend to make the airplane pitch up even further as the airplane pulled more g's. These changes had helped to make the plane a bit faster but were corrected after the incident to make it safer.

On the human factors side, Matt also mentions that he had a head cold at the time, and made the decision to take a phone call right before his flight, meaning he was distracted and not at his best.

Elsewhere in the interview he mentions that when he flies in air shows, he is more conservative in his margin of safety than other pilots might be.

You don't need the money if you're dead. So it's just a show.

Answer 2

I haven't watched the video, but keep in mind that to a first order approximation, lift is perpendicular to the airfoil.

In ordinary level flight or close to it, this causes the lift vector to have a significant vertical upwards component, which nicely offsets the force of gravity acting downwards on the aircraft's weight.

However, if the airfoil (in this case, wings) are fully vertical, then the lift generated by that airfoil becomes fully horizontal. The wings don't care about their angle to the ground; only about the movement through the surrounding air.

However, gravity still applies that same vertical force downwards.

Level flight (in any attitude) requires that there is a upwards vertical component to the lift vector, and that this component is the same as the downwards vertical force of gravity. If the two are unequal, the aircraft will climb or, in this case, descend.

With gravity still pulling the aircraft down, and lift acting horizontally (sideways), there is nothing left to maintain the aircraft's altitude, and the aircraft will do the proverbial drop from the sky. To recover you'd better hope you're able to level the aircraft at least somewhat, to get a bit of vertical lift out of the wings.

This, too, is why if we don't apply additional upwards lift in a turn, we'll lose altitude; the lift vector has a smaller vertical component (even though the absolute value of the lift vector is unchanged), but gravity doesn't relent, so gravity wins until we reestablish a balance between gravity and lift, typically by raising the nose a little.

Answer 3

What this comes down to is that generally, wings create lift in whatever direction the top surface is facing.

In normal flight, most of the force holding the airplane up is upward lift from the wings. When a plane is level, all of the lift force from the wings is directed straight up, and so the plane is able to maintain altitude. Even in an ordinary (non-aerobatic) banked turn, the tops of the wings are still facing mostly up, so the pilot can maintain altitude by just increasing the angle of attack a bit.

In this video, the turn the pilot made was no ordinary banked turn. It looks like he banked the plane by about 90 degrees. At a 90 degree bank angle, all of the lift from the wings is pointing sideways (into the turn), and none of it is holding the plane up. So unless there's something else holding the plane up, that plane is going to start falling.

There are two ways to keep from falling in an extreme banked turn. One is to keep the turn brief, so that you don't fall very much. The other is to stay up using thrust and fuselage lift, as detailed in the answers to this question: Can you fly an airplane at a 90° roll angle without losing altitude?

Answer 4

What he does is called a "G stall". Basically, he he pulled the aircraft into a harder turn (which is what he describes about 0:57 in the video). What that did was increase the centripetal force on the aircraft (measures in terms of Earth gravity, or "Gs"). This acrobatic pilot describes the effect

The stall speed of an aircraft increases as the wing loading increases. To be precise the stall speed increases with the square root of loading so if an aircraft stalls at, say, 60mph in 1G level flight in a particular configuration it'll stall at 120mph at 4G, 180mph at 9G etc. (This is one reason why "stall speed" can be a misleading term - a wing stalls at a given angle of attack.)

This is also sometimes called an accelerated stall, which can be quite common in these types of events, due to the rapid inputs and stresses the pilots are putting on the aircraft

Accelerated stalls are often caused by abrupt or excessive control inputs made during steep turns or pull-ups. If you’re in a dive and pull back with enough suddenness and force to load the airplane to a typical design load factor of 3.8 G’s, you’ll enter an accelerated stall if the airspeed drops below 1.95 times the stall speed at 1 G loading (the square root of 3.8 is approximately 1.95).

It then describes the crash of Midwest Express Airlines Flight 105, which offers the following illustration (source)

Answer 5

What happened was an accelerated stall "snap roll", which Matt Hall explains at 50 seconds into the video, he "g stalled" the plane "which rolled me slightly past the knife edge", so the plane was literally slightly upside down turning downwards towards the water, as opposed to falling.

An accelerated stall snap roll occurs in high g turns when one wing experiences a high speed stall before or more than the other. The plane rolls towards the stalled wing increasing it's angle of attack further still, while reducing the angle of attack on the other wing, making the situation worse. Note that an accelerated stall snap roll only requires excessive elevator at high speed, no aileron or rudder inputs are required.

In this case, the pilot quickly eased off the elevator input, which stopped the stall and allowed the ailerons to function to recover from the snap roll.

This was somewhat unlucky, as the snap roll could have rolled the other way, so the plane would have been turning upwards away from the water instead of downwards into the water. It's pretty much a 50% - %50% chance which which direction a plane will roll in a snap roll, unless ailerons are used to initiate the roll before applying excessive elevator.

True elevator only snap rolls are done on model aerobatic airplanes, either on purpose as an aerobatic maneuver, or unintended in the case of model pylon racing, which is the same circumstance as Matt's incident (without the risk to the model pilots life). I've even done this on a small radio control glider in a vertical dive (required to maintain speed since drag is high in a snap roll). Pull too far back on the transmitters elevator control and the glider just rolled, without a hint of upwards pitch. Eased off the elevator and the model recovered from the snap roll and responded to non-excessive elevator inputs in a normal fashion.