From: (Christoph Bernhardt)
Organization: Eurecom, Sophia Antipolis, FRANCE
Date:         21 Jun 95 02:56:53 
References:   1 2 3 4
Followups:    1 2
Next article
View raw article
  or MIME structure

>>>>> "Reeves" == B Douglas Reeves <> writes:

    Reeves> While both Boeing and Airbus are producing fly-by-wire
    Reeves> aircraft, the fact that the computer has ultimate control
    Reeves> in the Airbus systems versus the Boeing system where the
    Reeves> software plays more of an advisory role makes the Airbus
    Reeves> system more vulnerable to software errors.

I don't think that your above statement applies as it is. As far as I
know the 777 is like the 320 completely FBW. So even with a
conventional interface there is still software in the loop for things
as simple as translating input at the interface (yoke) to movements of
the control surfaces. What might be interesting is a comparison of the
complexity of the software that is involved in certain "simple" but
crucial tasks in the 320 and in the 777.

    Reeves> An excellent example of this is the Airbus A330 crash.  My
    Reeves> understanding of the reports is that the pilots were
    Reeves> making an engine out takeoff and that the combination of
    Reeves> (1) the use of TOGA vs Flex49 power setting and (2) the
    Reeves> nose-down input from the co-pilot made the autopilot
    Reeves> engage (1) late and (2) in altitude acquisition mode.  The
    Reeves> autopilot then kept increasing the pitch until the pilot
    Reeves> realized what was going on and three seconds later took
    Reeves> control and pitched the aircraft down.  By then the speed
    Reeves> had decayed to ~100kt.  The aircraft was unrecoverable.
    Reeves> Apparently the software killed them.  Good thing it wasn't
    Reeves> a revenue flight.  I guess Airbus will be fixing that
    Reeves> piece of the software, but how many other "bugs" like this
    Reeves> are there?

As already mentioned before in this thread I don't think the 330 crash
fits into the discussion here. If you want you can even go so far to
say, it happened in the debugging phase of the aircraft (that's why
they do these tests). And there were other more human factors involved
in the fatal outcome (nothing to to with man machine interface) that
we can only speculate on (e.g. differences between engineers and the
test pilot).

As a further note or question: How many of the Airbus crashes were
traced to "real" software errors and not to interface problems? Not
that it makes a difference for the severity of the problem, I just

Christoph BERNHARDT	        | "Every time I think I know where
Institut EURECOM		|  it's at, they move it."
2229 Routes des Cretes		|
B.P. 193			|	voice : +33
06904 Sophia Antipolis Cedex	|	  FAX : +33
FRANCE				|	email :