[From Bill Powers (2003.03.19.0305 MST)]
NASA TV (C-band, Satellite W2, transponder 9) has been broadcasting
hearings conducted by the Columbia Accident Investigation Board. The
reports from investigators make use of control theory in a way that most
PCTers would easily understand, a variation on the Test for the Controlled
Variable. We could call it "Deducing the Disturbance." We've discussed it
before on CSGnet.
The Shuttle telemetry sends data to the ground from over 4000 sensors.
Among these are sensors that detect the angles of the various aerodynamic
control surfaces, the "elevons" (elevator-ailerons) and the trim
adjustments on them. These are controlled variables for the systems that
adjust the control surfaces which change the roll, pitch, and yaw forces
acting on the Shuttle. The reference settings for the control surface
deflections are in turn adjusted by higher-order control systems that
maintain a commanded pitch, roll, and yaw attitude of the body of the
Shuttle relative to an inertial frame of reference. And those control
systems receive their reference signals from the autopilot that has control
during re-entry, causing the shuttle to follow a sequence of S-curves in
space to the right, to the left, and back to the right, to slow it down as
it approaches the spaceport.
As Columbia re-entered the atmosphere, starting well off the California
coast the telemetry shows that the control surfaces began to deflect in a
way not typical of previous flights. From previous tests and design data,
the roll and yaw moments (moment = force exerted by the control surface
times its distance from the axis of twist) could be deduced. From those
forces, the corrective torques acting on the Shuttle could be deduced,
because the autopilot and its subsidiary control systems managed to keep
the Shuttle's attitude under almost perfect control right up to the last
few seconds before breakup. This means that the moments created by the
control surface deflections were perfectly balancing out whatever
disturbing forces were acting, so the control surface forces were direct
measures of the disturbances. This is the situation we have called the
"behavioral illusion" in PCT, but to the investigating engineers it is
simply a normal aspect of how control systems work, since they were never
under any illusions to the effect that disturbance stimuli were being
sensed and directly "causing" control surface deflections.
As re-entry continued, the records show that the yaw disturbance increased
at first to a plateau, then increased again to a new plateau. then
increased faster and again faster until the last seconds before breakup.
Toward the end, the reaction control jets began to turn on as the attitude
control systems activated more and more subsystems to keep the attitude of
the Shuttle body correct. At first two of the jets that acted against the
disturbance were turned on, then the other two, in addition to the
increasing deflections of the aerodynamic control surfaces. Just before the
end, all four yaw-control jets were firing, the control surfaces were at
maximum deflection, and the yaw and roll errors were just beginning to
become significant. The Shuttle body was still in nearly its correct
attitude for the banking maneuver that was under way, and the crew had no
idea that anything was wrong, as shown by voice communications right up to
the final Loss of Signal (LOS) less than a minute before total breakup, by
which time all of the crew had undoubtedly perished. Whatever they suffered
must have been over before they could even comprehend their situation.
A similar record exists for the roll moments (torque about the longitudinal
axis of the Shuttle body). The roll moments trended toward a left-wing-down
force, then reversed until an opposite force was acting. Again. maximum
corrections were being exerted just before breakup, with roll error still
nearly nominal.
With the magnitudes of the disturbing moments known all during re-entry,
and correlated with major debris-shedding events seen from the ground, the
investigators are now trying to deduce what changes in configuration of the
Shuttle parts could have produced those forces, as well as the off-nominal
temperature changes recorded on the skin of the fuselage, in the inside of
the left wing, and in the hydraulic fluids in the left main landing gear
wheel well. Models are being constructed and flown in the hypersonic wind
tunnels at the Langley research center, with notches in their left wings to
simulate missing panels from the leading edge, and various sizes of holes
(breaches) in the lower surface of the wing where tiles may have been
missing. Also, detailed simulations are being done computationally.
All these modeling efforts are aimed at finding just what damage to the
wing would be able to account for the aerodynamic forces and the
temperature changes now known to have occurred. It is clear that there was
at least one major breach of the left wing, perhaps at the wheel well,
because some of the recovered parts of that wing are coated with aluminum,
which was obviously melted. There is no exposed aluminum on the
undersurface or leading edge of the wing. The inside of the wing was most
probably subjected to a jet of air at somewhere between 3000 and 10000
degrees Fahrenheit, depending on the size of the breach -- like a cutting
torch, and possibly hotter, and much bigger.
It's possible that tiles on the underside of the left wing were damaged
during ascent, when three pieces of insulation were observed to break off
of the external tank and strike the bottom of the left wing. At the time
this happened, the relative air velocity was between 350 and 700 feet per
second, the latter being comparable to the muzzle velocity of a Colt .45
bullet. If that is what exposed the wing to further damage, the crew were
doomed starting 80 seconds after liftoff of the 16-day mission. This event
was seen and discussed, but I am very glad it was not considered a danger
to the mission, because there was absolutely nothing that could have been
done about it except perhaps to leave the crew in orbit until their air,
food, or power ran out (the fuel cells that generate power also generate
water). There was no robot arm on this mission, so no inspection was even
possible, and there was no possibility of repair or of rescue. Better, I
think, that the crew knew nothing until the last few seconds before the
control systems finally reached their limits, the shuttle yawed to the
right at the maximum rate the sensors could report, and the unprotected
sides and upper body were exposed to air moving by more than 100 times as
fast as a class 5 hurricane wind.
As of March 17, 19% by weight of Columbia has been recovered and
transported to KSC for cataloguing and examination. Links to more
information are at
http://spaceflight.nasa.gov/index.html
Look for the revised time-line, a 3.5 Meg PDF file containing maps which
show the re-entry flight path with all known events annotated in their
proper spatial and temporal locations.
Best,
Bill P.