In the NTSB report on the crash of American Airlines Flight 587, it is mentioned (in the context of the changes between the A300-B2/B4's rudder system, on the one hand, and the A310/A300-600's rudder system, on the other), that
Second, Airbus changed the rudder travel limiter system on the A300-600 from a variable ratio design, [...] which was used on the A300B2/B4, to a variable stop design. Airbus indicated that the variable stop design was chosen for the A300-600 over the variable ratio design because it was less complex and had less severe failure modes. [Page 38 of the PDF, numbered as page 24.]
- An aircraft's rudder travel limiter (RTL) is, as the name suggests, a device that, if working correctly, limits the amount of rudder travel available at high indicated airspeeds in order to help keep the pilot's control inputs from breaking the aircraft.1
- A variable-ratio RTL reduces the amount of rudder deflection per unit pedal travel as airspeed increases; it does not reduce the amount of pedal travel available, or the amount of pedal force needed to produce maximum rudder deflection.
- With a variable-stop RTL, the amount of rudder deflection per unit pedal travel stays constant; instead, the amount of pedal travel available to the pilot decreases as airspeed increases (in the case of the A310 and A300-600, from 4 inches at 135 KIAS to just 1.2 inches at 250 KIAS), as does the amount of pedal force required to move the rudder to its travel limit.
With this in mind, I'm wondering what they meant by "less severe failure modes" - what failure mode(s) were they scared of with the variable-ratio design? If anything, there's one severe failure mode that's present in the variable-stop design, but not in the variable-ratio design: if a pilot pushes the rudder to the stop while the aircraft's airspeed is rapidly increasing, this can negate the protections imposed by the RTL entirely, by physically preventing the RTL variable-stop arm from moving into position. This scenario has actually occurred at least once (during an inflight loss of control in a different American Airlines A300-600 in May 1997); to quote the aforementioned report:
[...] Testing of the rudder travel limiter determined that, if a pilot applied a sufficiently large pedal force when the pedal was at its travel limit, such a pedal force would further slow or stop the movement and, consequently, the effectiveness of the rudder travel limiter. The flight 903 event demonstrated that slowing or stopping the rudder travel limiter by application of large pedal forces could result in the rudder position substantially exceeding the designed travel limit. The Safety Board was concerned that such an increase in available rudder beyond its designed rudder travel limits could permit excessive rudder movements and possibly result in high loads on the vertical stabilizer. [Page 134, numbered 120.]
This failure mode would not be present with a variable-ratio RTL, the proper operation of which is not dependent on mechanically restricting the amount of rudder pedal travel.
What failure mode(s) were Airbus trying to eliminate/mitigate when they switched from a variable-ratio RTL to a variable-stop RTL for the A310 and A300-600?
1: Obviously, it was not successful in this regard as concerns AA587, as the pilot of that flight was able to break off his aircraft's vertical stabiliser without going beyond the limits imposed by the RTL; however, the RTL was never designed or intended to protect the aircraft from the extremely high aerodynamic loads produced by a rapid series of pilot-commanded rudder doublets,2 as occurred in AA587's case.
2: A rudder doublet is when the rudder is rapidly slammed from stop to stop - i.e., when the pilot sends the rudder to its travel limit in one direction, then immediately to its travel limit in the other direction.