Monday, May 30, 2011

My Little Runaway

Since I am a college professor, I say "It's more complicated than it appears" and "Wait until we have all the facts before we draw a conclusion" often. That's why I haven't said anything about AF447, except for one tweet about this post when it became clear that the aircraft had stalled. I felt that I was in no position to judge the flight crew or its actions. "It's more complicated than it appears," I thought, "Wait until we have all the facts."

I'm glad I waited. An article today in the German Der Spiegel raises some questions about the Airbus's automated flight control system. It appears that both of the Airbus's side sticks were commanding "nose down," as expected. But the nose didn't come down, and the horizontal stabilizer moved in the nose up direction. The captain evidently took a seat with a comment from one of the co-pilots like "Here, you try it!"

They did the right thing.

The airplane didn't.

I am hesitant to draw a conclusion about French aircraft and stabilizer trim, but it is tempting to see a relationship between the recent grounding of the Falcon 7X fleet and this report. A Falcon 7X had a pitch trim runaway, so the fleet was grounded. Good move.

Those of us who fly smaller airplanes might wonder why the flight controls didn't overpower the stabilizer trim. In some heavier airplanes (starting with the Lear 25), the stabilizer trim is much more powerful than the flight controls, especially in the situation where the whole stabilizer moves, rather than a tab (tabs are bad because in some situations at high airspeed they can cause control reversal). United lost a DC-8 in 1983 due to a mis-set stabilizer trim (NTSB data here). Alaska lost an MD-80 to a stabilizer trim problem. Next time you're around airliners look at the stabilizer: there are usually marks on the tail showing the maximum and minimum stabilizer position, so it can be verified during the walk-around.

I've had two pitch trim problems, neither one of which caused a crash. The first was in my Taylorcraft: I was a little too enthusiastic with the trim crank and pulled the cable off the pulley, leading to a full nose-up trim condition. The T-craft uses a single trim tab on the left elevator, and while it was annoying I had no trouble pushing the nose down and completing the pattern. After that I learned my lesson: I trimmed slowly (which is the right way), and told riders "That's the trim crank; don't touch it!" No problem.

The other was in a Seneca. The electric pitch trim wouldn't respond. I played with it for a while and, consistent with my philosophy that circuit breakers are for pulling but not for pushing, I used manual trim for the rest of the flight. Good thing: there was a short in the system, and there were sparks behind the panel.

I've also had an autopilot runaway, in a Cessna 414A. Wait, I've had two autopilot runaways, but the second was caused by a pilot with an unpleasant personality who wanted to see how I'd react. (He didn't last long with the company.) The real one was on the final leg of an air ambulance flight. It was about 0400 and the final leg was less than 50NM in night VFR conditions. But it was 0400 and I was tired. I engaged the autopilot and the nose shot up into the darkness. Huh? Whoa! Phew!. I hit the autopilot disconnect and lowered the nose.

But it was 0400, so I engaged the autopilot again. Huh? Whoa! I disconnected it and hand-flew the rest of the way home.



At May 30, 2011 at 8:42 PM , Anonymous Dave Starr said...

Well thought out and timely. Some of the larger aviation-oriented blogs and websites have been a bit too quick to leap at what they "think" the 4 page report really says.

In my view that report itself is ill-advised, but hey, no one asked me.

I worked for some years as a maintainer/procedures trainer for the USAF in the (then) front line C-141 simulator program. Lockheed built in three redundant (and often recalcitrant) pitch trim systems.

We nearly lot several aircraft over runaway pitch trim scenarios (several different systems and causes).

We trained on runaway pitch trim recovery relentlessly in the sim, because much of what we did was way too risky to demonstrate in the airplane.

In one specific failure mode, which was eerily similar in outcome of AF447's accident 9although totally different causal factor), the aircrew had about 20 or 30 seconds to roll into a 60 degree or steeper bank, because beyond a certain point even both pilots on the controls could not get the angle of attack down enough to prevent a stagnated, deep stall situation which is essentially unrecoverable.

Bad design? In my book. IF 447 proves to be another case of too little control authority/too much errant trim, it would be sad indeed, since my C-141 experiences date back almost 40 years now.

Are there no engineering classes under the general area of "lessons learned"?

At May 30, 2011 at 9:09 PM , Blogger Dr.ATP said...

Thanks for your comments, Dave; it's always nice to hear from you.

I have a friend who was a KC-135 sim instructor at Castle and he described the same recovery maneuver for nose high attitudes, but in that case it was to keep the fuel from unporting. The difference here is that the Airbus has envelope protection and might not have allowed the maneuver!

At June 13, 2011 at 9:28 PM , Anonymous Dave Starr said...

Indeed, when I said 'too little control" authority I was using an imprecise way to refer to the envelope protection algorithms. It's indeed a hard decision for a designer to make, but limiting the pilot's control authority seldom seems to be the best solution.

Fred Weick thought he had it nailed down back as early as 1934 (the W-1, direct ancestor of the Ercoupe), but although much of his work was brilliant, limiting the pilot's control authority didn't produce the great leap forward in safety that was envisioned.

I never worked in the KC-135, but the fuel unporting issue certainly seems logical ... the entire KC-135 and 707 family needs very careful management in that area, but as I recall, our instructor pilots drilled our sim students continually on the first step of unusual attitude recovery being to note if the attitude indicator 'ball' showed a lot of brown or a lot of blue ... more brown than blue roll towards wings level, more blue than brown, roll toward the down wing and only then start troubleshooting further.

Pretty simplistic approach, but it worked. There is a lot to learn from AF447, that's for sure. Keep writing.


Post a Comment

Subscribe to Post Comments [Atom]

Links to this post:

Create a Link

<< Home