(My apologies if this should go in Aviation.SE instead - I'm not quite certain where questions regarding the behaviour of the shuttle's flight controls during reentry should go.)
For flight in atmosphere, the space shuttle orbiter had a suite of aerodynamic flight control surfaces (elevons on the trailing edges of the wings for pitch and roll control, and a split rudder mounted on the vertical tail for yaw control and to act as an airbrake). During reentry, the elevons would become active for pitch and roll trim (with primary pitch and roll control still being provided by the aft RCS jets1) once the dynamic pressure (q-bar) on the vehicle exceeded 0.5 force-pounds per square foot (psf), and became the primary means of pitch and roll control once q-bar climbed through 2 psf, with the RCS roll jets being disabled entirely at 10 psf and the pitch jets switching off at 40 psf.2
Between then and landing, there would be a major switch in the elevons' behaviour.
- At high angles of attack (high alpha), used early in reentry to decelerate the vehicle and limit structural heating, the elevons would operate with a negative aileron gain (GALR), or adverse aileron; in the high-alpha flight regime, the elevon's drag coefficient rises steeply with increased alpha while its lift coefficient rises only a little, causing an asymmetric elevon deflection to strongly yaw and roll the orbiter towards the wing with the down-deflected elevon (although there would be a slight initial roll towards the up-deflected elevon), the opposite of what aileron deflections do during normal flight in most aircraft.
- In contrast, at low alpha (used later in the descent for optimal ranging capability), the elevons would operate with a positive GALR (proverse aileron); when alpha is low, the elevon's lift-coefficient curve rises more steeply with increased alpha than its drag-coefficient curve, causing an asymmetric elevon deflection to strongly roll and yaw the orbiter towards the wing with the up-deflected elevon (although there would be a slight initial yaw towards the down-deflected elevon) in the normal aircraft manner.
Partway through entry, GALR would switch from negative to positive as the vehicle slowed down and its alpha decreased; this occurred according to a set schedule which adjusted GALR as a function of the vehicle's mach number. All well and good so far; however, if, for some reason, the orbiter's angle of attack differed significantly from that assumed by the flight-control computers' canned mach-alpha reentry profile (i.e., if, at a given mach number, the vehicle's alpha was significantly higher or lower than the nominal normal-reentry alpha at said mach number), this could cause GALR to have the wrong sign (commanding proverse aileron when adverse would actually be needed, or vice versa), which could rapidly lead to loss of lateral/directional control of the orbiter. An example of this (although, thankfully, one that never happened on an actual mission) would be certain contingency abort scenarios involving unusually-steep entries (where a higher-than-mach-scheduled alpha would be required to achieve a successful pullout without exceeding the orbiter's limits for equivalent airspeed or vertical Gs):
3.3 Adverse Aileron Trim
Adverse aileron trim leading to a Loss of Control (LOC) starts as a consequence of excessive beta ( β) during entry with Mach < 5. This is particularly true if the MM 602 transition occurs below Mach 5. The rudder is active in flight control on entry below Mach 5 and the rudder is commanded to trim to counter β. However, at high angles of attack (> 50°) the rudder aerosurface is not effective. As a result the RUD trim will be saturated (Left or Right 6°).
Thus the aileron trim is then commanded in an attempt to counter the existing β. However the aileron trim is a function of a flight control gain (GALR) which was designed as a function of Mach, assuming a nominal entry alpha to mach profile. The GALR is intended to determine the sign of the primary aileron control term, and is used to transition from the early entry adverse aileron to the late entry proverse aileron which Mach < 4. However, the Contingency Abort entry profiles (particularly at low mach) follow a much different alpha to mach profile. Thus at low Mach, with high alpha, the GALR gain is incorrect, causing adverse AIL trim which is opposite the trim actually required to null the existing β. Trimming the wrong direction with the aileron exacerbates β even further, quickly leading to an LOC. [NASA Contingency Aborts 21007/31007 Student Handout, pages 34-35; section heading bolded in original, other emphasis mine.]
Given that the need to switch from adverse to proverse aileron gain is based entirely on the change in the orbiter's angle of attack, I don't see why they didn't simply base GALR directly on alpha (which would eliminate the potential for loss of control from wrong-sign GALR in the event of the vehicle's actual alpha being significantly different from that in the canned mach-alpha profile), instead of using mach number as a proxy for alpha; does anyone know why the shuttle's designers did this?
1: The orbiter's forward RCS jets were not used during reentry, and all forward-RCS propellant was dumped prior to entry interface (to eliminate a potential explosion hazard in the event of the forward RCS propellant tanks being fractured by an overly-hard nose slapdown during landing derotation) unless it was necessary to retain some propellant in the forward-RCS tanks in order to keep the orbiter's center of mass from lying beyond its aft limit (which could cause a loss of longitudinal control during entry).
2: The RCS yaw jets remained active down to mach 1, as the orbiter's vertical tail provided no significant degree of directional stability or control at the high angles of attack used at hypersonic speeds during reentry (the vertical tail was almost completely blanked out by the wings and fuselage at high alpha, and, due to the tail's high degree of sweepback, the rudder's hinge line would have been nearly parallel to the relative wind under these conditions, providing little to no control authority even if it weren't blanked), and, even at the lower alphas used below mach 5 (the point at which the rudder came alive), it would not have been sufficient on its own without assistance from the RCS yaw jets.