Re: "Die Zeit" article on Airbus (longish excerpts)

From:         T.E.Thacker.Junior@lesueloc.com
Organization: Performance Systems Int'l
Date:         14 Dec 94 02:22:37 
References:   1
Next article
View raw article
  or MIME structure


In article <airliners.1994.1769@ohare.Chicago.COM>, 
<onat@turbine.kuee.kyoto-u.ac.jp> writes:

> Just before landing.
> Pilot is to set the descent rate to -700, but by mistake, sets
> the altitude to 700. The computer reacts by reducing power to 
> idle, showing the descent bar (I do not know what it is called
> in practice) way too low on the artificial horizon, and begins automatic
> descent to set altitude. Pilot, unaware of this pushes the stick further
> than necessary, following the descent bar on the screen. 
> 
> They realize the problem a couple of hunderd feet agl, but can not 
> figure out the way switch out of the mode properly.
> 
> And yes, the airpot altitude is actually 900 feet...
> 
> At least three accidents and one near miss took place with Airbus
> because the pilots misunderstood aborted landing sequence, resulting
> in the aircraft stalling at some hundereds of feet AGL, with the nose
> pointing up as much as 90 degrees. An East German (some time ago)pilot 
> escaped by using manual override for the horizontal stab trim, after 
> stalling four times.

  When I write rocket trajectory programs, I design in "reasonableness"
tests. I don't allow my elevation bug to descend below a certain point
no matter WHAT the iterative guidance commands. That way I don't get
a horizontally flying "land shark" doing mach five at the tree-tops!

  I remember the second Saturn V launch that had two engines out in the
S-II stage. The guidance wasn't designed to control that kind of loss of
thrust and expect a viable boost. It grossly overcompensated and severely
lofted the rocket much over the 100 nmi it was supposed to fly towards.

  When the third stage came alive it's guidance said the equivalent of
"Oh-My-God-I'm-Too-High" and proceeded to thrust directly at the ground.
At one point, it was actually thrusting BACKWARDS to attain it's target
condition.

  I design in constraints that tell the rocket "OK now, it's not reasonable
to go from +80 degress to -80 degrees in 2 seconds - let's constrain it to
0.25 deg-per-sec max rate". Or something like "I'm below 100,000 ft so I
must stay above 75 degrees up-elevation". Or, "My Dynamic Pressure is 1000
lbs per square foot now so I must keep less than 0.25 degrees angle-of
-attack to the wind".

  It seems to me they could include ground level info in each autopilot and
have the machine reject a request to "cruise" 200 feet below ground level.
If Flight Simulator can store every airport, VOR, and feature in the country
on two CD-ROMs and "simulate" a crash when you fly into the simulated ground,
then a real autopilot should be able to use the same info to keep a pilot
from killing himself accidentally. 

  It should also be able to say that 90 degrees nose-up is NOT a reasonable
attitude to shoot for.
----------------------------------
Tom "Cant Read/Write Email" Thacker, Jr
404-850-0445 Home
----------------------------------
   "Thou shalt maintain thy airspeed, lest the ground shall
    arise and smite thee".