Jan 16, 2008

Rockin' and Rollin'

So, if the A319 which experienced a sudden flight upset earlier this week was not struck by clear air turbulence, what other factors could account for the roll oscillations reported? Here are a couple that come to mind:

Modern Jet Transport autopilots provide inputs on all three aircraft axes. But normal roll or steering inputs would never command the steep bank angles reported. If there was an automation-related failure, it would have to be something very unusual.

One key part of the autoflight system that may be associated with sudden roll inputs is called a yaw damper. How does yaw interact with roll? By design all modern aircraft will immediately begin to roll when they experience any significant yaw. And this effect is magnified by swept wing configurations. In fact, on most aircraft and certainly on swept wing aircraft, pilots will generally notice and react to the rolling caused by unexpected yaw much sooner than we'll notice the yaw itself. In flight simulators, where we practice engine failures during takeoff, we have to condition ourselves to apply necessary rudder inputs to stop the yaw rather than becoming distracted by the noticeable roll that results as well.

A failure in the yaw damper system could cause large and sudden roll movements, but again, this would require a pretty rare and drastic failure mode. Notably, the Boeing 737 has had some serious accidents and incidents attributed to the rudder suddenly moving "hard over:"
http://www.airlinesafety.com/faq/B-737Rudder.htm

Another potential cause could be an inadvertant overspeed of the aircraft causing shock waves to form on the wings. Any uneveness between the left and right wing could cause some sudden yawing-rolling motions. But again, this is unlikely due to the over-speed protection features built into the Airbus flight envelope protection system.
http://en.wikipedia.org/wiki/Transonic

One other source of sudden, inappropriate control inputs could be the pilots themselves, either a foot slipping and hitting the rudder pedal (see the notes above about yaw and roll), or accidentally leaning against the Airbus joystick - which apparently has happened before according to this report from an aviation forum:

When my company first got their A320's we had similar incident. The cause was put down to the F/O sitting cross legged and his Knee pressing on to the side stick. If I remember correctly they thought either the pressure gradually increased until the autopilot let go or that he moved and knocked the side stick hard over. Neither pilot could remember this happening but the FDR showed the side stick movement.

Again if I remember rightly the a/c went 60 deg one way then 60 the other as the other pilot grabbed his side stick to correct. Then a bit of dual stick input (it sums the inputs) until things settled down. Luckily no one was hurt.

So, you can see that even if clear air turbulence is ruled out, there will still be a lot of other factors to study before the real cause will be known. The investigation process may take a while yet. In the meantime, Airbus Inc. and airlines who operate large fleets of A319s will be watching closely. As will stockholders and of course pilot associations.

Time will tell.

For detailed discussions of aircraft control and flight dynamics :
http://en.wikipedia.org/wiki/Flight_dynamics
http://en.wikipedia.org/wiki/Flight_control_surfaces

5 comments:

MathFox said...

The autopilot has (at least) two controls that you didn't mention: engine power and airbrakes that control the airspeed. Furthermore there are the flaps and the landing gear that the pilots can employ to influence the speed.

The analysis gets more complicated because of the interactions between the various controlled variables: pitch influences speed; roll and yaw are coupled too. However, there are other non-linearities in control: the effect of control plane deflection increases with the air speed (as long as flow is laminar, turbulent flow and shock waves may lead to a sudden loss of control.)

One does not need to have a plane that is stable on all axis, in all attitudes, etc. as long as the control system (pilot or autopilot) can compensate for those instabilities. It is just easier to fly a stable plane.
To stabilise an unstable system the control system needs specific inputs. If these inputs are missing, the system will go its unstable way (VFR pilots losing track of the horizon...) If the inputs are plain wrong all kind of bad things can happen. It's good that a pilot knows to ignore defective instruments.

Soaring Student said...

One of the networks ran a piece last night, perhaps yet more speculation, that wake turbulance might be a cause. Apparently there was a United 747 ahead and higher of the 319.

Note the cultural difference: The aviation community explores the possibilities, gets the facts, correlates and determines (at least, to the extent that the information is available). The news community needs to fill the time, say stuff so they get watched rather than one of the 450 other channels, supported by a number of wannabee experts that want to get on TV.

I remember an American "aviation expert" who got in TV after the Air France incident in Toronto - who stated categorically that the problem had to be fuel starvation because the pilot went through with the landing (implying some urgency/desperation), and because it wasn't a huge fireball (therefore, of course, the tanks were empty). The wannabee was, of course, completely wrong.

Anonymous said...

Thanks for adding other insights. The list of possiblities is long. We can rest assured that the investigators will be looking into it all and will eventually come up with a reliable report as to the cause.

Of course this will not stop some people from coming up with "alternate" conspiracy theories and wild speculations, including government and corporate cover-ups.

But ususally the truth is the most amazing story of all.

Teller said...

Does the bus have a rudder boost system, in addition to the yaw damper? I know we practice unplanned rudder boost activation, in the same way we practice runaway trim...it just happens randomly, we click it off (once we figure out that it's not an engine failure), then right the plane. Seems like (if installed) this would have a pretty decent chance of causing un-commanded roll?

Aluwings said...

mathfox - yes, so many considerations. Happily the Airbus Flight Data Recorder tracks hundreds of parameters so there are no secrets. I'm sure the story will emerge once investigators do their work.

soaring student - wake turbulence hadn't even occurred to me. Could be an interesting avenue of investigation, especially with the relatively new One Thousand foot vertical spacings now in use.

teller - I'm not sure what the rudder boost system is exactly, but the auto-flight system does have the ability to apply significant inputs to the rudder. Some recent aircraft (I believe) have an automatic rudder response to an engine failure, but not the A319 that I know of. We do sometimes practice recovering from Dutch Roll in case of yaw damper failures - but this is pretty rare and I think it was covered once during intial conversion training, then never reviewed again.