Rudder for final runway alignment (?)

RK Henry writes:

Don't underestimate the capabilities of a trained, experienced crew to
cope with equipment failure. Case in point, United flight 232 at Sioux
City, Iowa. Despite complete hydraulic loss and concomitant loss of
flight controls the crew was able to bring the aircraft to what turned
out to be a survivable crash for most of the occupants.

They were also flying an aircraft that did not have fly-by-wire
systems.

Fly-by-wire means that no command bypasses the computer. If the
computer malfunctions, or if it decides to ignore your command, you're
out of luck, and no amount of skill will help you. If flight 232 had
been a fly-by-wire aircraft, everyone aboard would have died.

The problem with fly-by-wire is that digital systems have typically
catastrophic failure modes, which is the dark flip side of their
superlative performance within the envelope. If flight remains within
the envelope foreseen by the developers of the system (and assuming
there are no bugs in the software), FBW aircraft fly better and more
easily than non-FBW aircraft. However, if flight ventures outside the
programmed envelope, failures in the system _will_ occur--and failures
in digital systems are often catastrophic failures, because of the way
digitization separates control from any constraining physical
parameters.

This issue is not limited to FBW aircraft, but it is much more
critical in FBW because the results of a malfunction are usually
fatal.

Everything is different in more conventional aircraft. You might lose
the hydraulic assist on control surfaces, but you can still move them
to some extent, and they won't snap into implausible positions that
exceed the physical limits of the system.

As an example, if you have a purely analog throttle, if you push it
beyond the maximum or below idle, the worst you're likely to get is no
effect at all, i.e., you'll still have full throttle or idle,
respectively. In a poorly-designed FADEC, however, your throttle will
just be providing a number to a computer. If the computer has
throttle settings from 00 to 99, and you push the throttle to a point
that sends the internal computer setting beyond 99, it may roll over
to 00, setting the engines abruptly to idle. Thus, the FBW throttle
has a catastrophic failure mode that is completely absent in the
conventional throttle.

Add to that the fact that many FBW systems are not ergonomically
designed and may have features that were conceived by engineers or
project analysts rather than pilots, and you multiply the chances of
problems.

I think every FBW should have a button that says "do exactly what all
the control inputs tell you to do," but many engineers apparently
disagree, and most people (including some engineers) don't know enough
about computers to realize the danger in this.

--
Transpose mxsmanic and gmail to reach me by e-mail.

Thomas Borchert writes:

BS! Point me to where the FBW does something the pilot doesn't.

Habsheim. The pilot apparently put the throttles all the way forward,
but the FADEC ignored this, because the aircraft was not in the right
"mode." We don't know for sure because of tampering with the flight
recorders, but this is one of several plausible scenarios.

Normally, a pilot expects engines to spool up when he advances the
throttle. But FBW systems do what they want, not what the pilot
wants.

In Habsheim, the pilot did something the plane wasn't designed nor
able to do, namely extremely low flight in a wooded area. His last
words were probably along the lines of "Hey, watch this!".

I read the transcript, and those were not his last words, although
I've forgotten what they were (and it's not certain that the published
transcript actually reflects what was originally on the voice
recorder).

FBW had nothing to do with it. The accident report reflects this.

FBW had everything to do with it. There were too many FBW systems
aboard, and they were too complicated, and they worked in ways that
were counterintuitive for an experienced pilot. When the pilot
followed his instincts in an emergency situation, the FBW refused to
follow.

Yes. That reason is called "Boeing Marketing Department".

The real reason why I prefer Boeing is the ten-day disappearance of
those flight recorders. If the aircraft were so reliable, there would
be no reason to sneak away with the hard data. Boeing has a long
history of concern for safety and experience therewith that its main
competitor does not share.

--
Transpose mxsmanic and gmail to reach me by e-mail.

On Thu, 28 Sep 2006 05:56:24 +0200, Mxsmanic
wrote:

RK Henry writes:

Don't underestimate the capabilities of a trained, experienced crew to
cope with equipment failure. Case in point, United flight 232 at Sioux
City, Iowa. Despite complete hydraulic loss and concomitant loss of
flight controls the crew was able to bring the aircraft to what turned
out to be a survivable crash for most of the occupants.

They were also flying an aircraft that did not have fly-by-wire
systems.

Fly-by-wire means that no command bypasses the computer. If the
computer malfunctions, or if it decides to ignore your command, you're
out of luck, and no amount of skill will help you. If flight 232 had
been a fly-by-wire aircraft, everyone aboard would have died.

The similarity is that in both cases there's no hard connection
between the control surfaces and the flight controls. The DC-10 had no
cables between the controls and the flight surfaces, just pipes, so
with no hydraulic power there was no control action at all. But the
crew was left with control of the engines and that made it possible to
control the airplane. My assertion is that there is insufficient
evidence to conclude that everyone would have died had the airplane
been FBW. Uncertainty remains as to that outcome.

However, much as it may disturb some other, less tolerant, members of
this group, I tend to agree with your qualms about FBW. FBW makes it
possible to implement exotic airframe designs by defining their flight
characteristics in software instead of in hardware. That's fine when
the pilot sits in an ejection seat, but when this technology is
extended to airplanes whose occupants don't have the option of pulling
a handle when things go badly then we must be very careful. I'm sure
the problems can be solved, probably with multiple redundancy, but a
century of experience with aviation shows that we must always consider
the possibility of failure.

I've been similarly apprehensive about glass cockpits in light
aircraft. A single point of failure in an inadequately designed and
tested system could conceivably leave a pilot deaf, dumb, and blind.
Show me the test plan.

RK Henry

RK Henry writes:

The similarity is that in both cases there's no hard connection
between the control surfaces and the flight controls. The DC-10 had no
cables between the controls and the flight surfaces, just pipes, so
with no hydraulic power there was no control action at all. But the
crew was left with control of the engines and that made it possible to
control the airplane.

True, but at least the control system didn't have a mind of its own.
It may stop working, but it isn't likely to start doing things that
the pilot doesn't want it to do.

My assertion is that there is insufficient
evidence to conclude that everyone would have died had the airplane
been FBW. Uncertainty remains as to that outcome.

From what I've read, they weren't even supposed to have survived the
accident as it was. I don't know if anyone ever managed to duplicate
their landing feat in a simulator (I've read that it has never been
successfully done in simulation).

However, much as it may disturb some other, less tolerant, members of
this group, I tend to agree with your qualms about FBW. FBW makes it
possible to implement exotic airframe designs by defining their flight
characteristics in software instead of in hardware. That's fine when
the pilot sits in an ejection seat, but when this technology is
extended to airplanes whose occupants don't have the option of pulling
a handle when things go badly then we must be very careful.

I agree. Furthermore, I just don't see a need for it. Just because
you can do it doesn't mean that you must or you should; it doesn't
even mean that you have anything to gain from it.

Military fighters need the best possible performance (or at least they
did, until they hit the obstacle of keeping pilots alive). FBW can
achieve that in certain situations, but at the expense of higher risk
for all flight in general. I don't see how any of this would be
applicable for general or commercial aviation.

I'm sure
the problems can be solved, probably with multiple redundancy, but a
century of experience with aviation shows that we must always consider
the possibility of failure.

Perhaps the problems can be solved, but if there is nothing to be
gained by FBW in the first place, why bother? Does anyone who flies a
plane for pleasure dream of being able to fly by just pushing one
button, or having a plane that does what it thinks is best, instead of
what the pilot tells it to do?

When planes are for pure transportation, perhaps that might argue in
favor of such systems, but in that case why bother keeping a pilot in
the cockpit at all? If FBW is completely reliable, you don't need a
pilot. If you need a pilot, then FBW is not completely reliable, and
the pilot needs a way to override it.

I've been similarly apprehensive about glass cockpits in light
aircraft. A single point of failure in an inadequately designed and
tested system could conceivably leave a pilot deaf, dumb, and blind.

And he would already be broke from paying for the avionics.

Wouldn't this also be a risk for any other aircraft?

Show me the test plan.

That's the part that worries me most. I know what passes for
"testing" in the world of computers, and it's an accident waiting to
happen. Yes, avionics are tested more thoroughly--but not thoroughly
enough. The failure modes multiply exponentially as gadgets and
features are added, and nobody is testing all the possible scenarios.

--
Transpose mxsmanic and gmail to reach me by e-mail.

On Thu, 28 Sep 2006 19:23:39 +0200, Mxsmanic
wrote:

Military fighters need the best possible performance (or at least they
did, until they hit the obstacle of keeping pilots alive). FBW can
achieve that in certain situations, but at the expense of higher risk
for all flight in general. I don't see how any of this would be
applicable for general or commercial aviation.

There is one potential advantage to computer enhanced control for
civil aviation: fuel economy. We can build more efficient airframes,
but they're often unstable. We can compensate with software. FBW
offers the promise of allowing airplanes to use less fuel. That's too
important a benefit to ignore, but we can't sacrifice safety in the
process.

RK Henry

Mxsmanic,

If flight 232 had
been a fly-by-wire aircraft, everyone aboard would have died.


Oh, are we clairvoyant now, too? You are spreading more clueless BS.

--
Thomas Borchert (EDDH)

Mxsmanic,

but this is one of several plausible scenarios.


A flat earth is one of several plausible scenarios to many people, too.

--
Thomas Borchert (EDDH)

RK Henry writes:

There is one potential advantage to computer enhanced control for
civil aviation: fuel economy.

Yes, but you could still build a system that the pilot can override
and yet preserve fuel economy. Autopilots and flight computers are
existing examples of this. Overriding them requires only the press of
a button, or (in some cases) merely a movement of the controls by the
pilot.

We can build more efficient airframes, but they're often unstable.

Why would a more efficient airframe be unstable? Something that
requires constant computer surveillance to stay in the air likely has
some aerodynamic instabilities, which implies sources of drag or lift
that would sap energy from the powerplant. Gliders require very
little energy.

FBW offers the promise of allowing airplanes to use less fuel. That's too
important a benefit to ignore, but we can't sacrifice safety in the
process.

Corporations don't care about safety.

--
Transpose mxsmanic and gmail to reach me by e-mail.

RK,

but when this technology is
extended to airplanes whose occupants don't have the option of pulling
a handle when things go badly then we must be very careful. I'm sure
the problems can be solved, probably with multiple redundancy, but a
century of experience with aviation shows that we must always consider
the possibility of failure.


And your worries show in the accident statistics of the Boeing 777 and
the Airbus 32x and up exactly where, after over 20 years of service?

A single point of failure in an inadequately designed and
tested system could conceivably leave a pilot deaf, dumb, and blind.


There's a ton of single points of failure in any aircraft ever designed.

--
Thomas Borchert (EDDH)

On Sat, 30 Sep 2006 17:01:37 +0100, Thomas Borchert
wrote:

RK,

but when this technology is
extended to airplanes whose occupants don't have the option of pulling
a handle when things go badly then we must be very careful. I'm sure
the problems can be solved, probably with multiple redundancy, but a
century of experience with aviation shows that we must always consider
the possibility of failure.


And your worries show in the accident statistics of the Boeing 777 and
the Airbus 32x and up exactly where, after over 20 years of service?

Umm...20 years of service? They've got a way to go before they rack up
that much service experience. However, I have few doubts. The
engineers share my concerns, that's why they've provided multiple
redundancy.

My Warrior has non-redundant cables operating the control surfaces. As
long as metal continues to obey physical laws and they get looked at
once a year, I have few worries. However, there are people who suggest
FBW is in the future for light aircraft. That worries me. How is a 777
flight control system going to scale to a Warrior? Is there going to
be some kind of cost-saving breakthrough or are they going to cut
corners?

RK Henry