FMS capability (was: Re: AA965 proves Airbus bashers deadly wrong)

From: (Ed Hahn)
Organization: The MITRE Corporation, McLean, Va.
Date:         06 Feb 96 14:15:45 
References:   1 2 3 4 5
Followups:    1
Next article
View raw article
  or MIME structure (Luc Bauwens) writes:
> Isn't it true that

> 1. A good computer would not have taken liberties with procedures,
> check points etc., and it would not have set up on a short cut thru
> a mountain, wouldn't it?

Yes, assuming the computer had the information about terrain
available.  However, because of the (correctly) stringent nature of
certification rules, the FMS computers on board most equipped aircraft
have an upper limit of 1Mword of memory or less (i.e. 2MB), due to the
need to show that the electronics industry has experience with the
components to have REAL, demonstrated MTBF figures.

Given the storage necessary for both a good navigation data base,
there just isn't enough room available in the FMS right now for a good
high-resolution data base. (The CPUs, by the way, are generally of the
80286-class at best, again because of the need to have enough
experience to demonstrate adequate MTBF.  Avionics in the very
near future will use 486-class CPUs, however.)

However, avionics manufacturers aren't ignoring the terrain issue:
several are planning on making the next generation GPWS computer use
GPS-derived position and a terrain database.  They have been in
development for a few years already.  Of course, in this
implementation, it would exist as a separate check, rather than an
integrated database for the FMS to use/plan around.

> 2. And furthermore, a computer might have remembered about aerodynamic
> brakes when finally discovering the mountain?

As hinted above, the current aircraft avionics architecture does not
integrate the flight control computers / autopilots with GPWS.
Therefore, the GPWS would not have direct control over the
speedbrakes, which sounds like sound design to me, as false GPWS
warnings are not unheard of, and probably don't warrant direct control
over the aircraft control surfaces.

(BTW, I am personally avoiding any speculation as to the net effect of
the speedbrakes on the attempt to pull up until the NTSB releases
findings in the AA accident.  It may have been "too late".)

> Not to say that this really should be a strong argument in the
> debate...  (I.e., I am not really trying to argue in favor of the
> computers, joysticks etc.  But it seems to me this particular
> accident *really* can't be made into a argument against computers...)

I agree; avionics needs to continue to evolve to guard against the
holes remaining in the safety.  However, thoughtful design taking into
account the strengths and weakness of both the human AND the computers
ultimately is the best way to keep the skies safe.


--------   Ed Hahn | | (703) 883-5988   --------
The above comment reflects the opinions of the author, and does not
constitute endorsement or implied warranty by the MITRE Corporation.
Really, I wouldn't kid you about a thing like this.