Originally Posted by
G0ULI
The tail and rudder are pretty substantial structures and should have had sufficient authority to level the wings. However the low airspeed during the initial climbing turn must have prevented full rudder authority being available to level the aircraft. ...
An incident of this type requires immediate intervention and rapid full control inputs, something that airline pilots are trained not to do for the safety and comfort of their passengers. Survivable in a test pilot flight test scenario but not in day to day commecial operations.
I discussed some pages back what I call "bevel flying", a sort of poor man's knife-edge flying. For any speed below which knife edge flying is possible, there's some maximum bank angle at which the rudder is sufficient to counter the bank. The closer to that bank angle, the less authority the rudder has.
When right on that angle, it's like balancing on top of an edge, and the phenomenon can roll off either way - increasing bank or decreasing. For an Electra at about 165 kts per my sim runs that angle is about 57 to 60 degrees, somewhere in that neighborhood.
This suggests that left rudder near that angle would be able to counter the bank but it would do so very slowly. As the plane started to level, the relative authority of the rudder would increase.
In my flight sim video I discussed this as if they were simply slow to go to full left rudder. This response at less than the critical 57-60 this gives the same apparent response. Either way, they were well into the zone where rudder authority is degraded by the bank angle. This is why I said the bank was an uncredited culprit in the affair. If they hadn't been already banked when they discovered the problem, I think it would have turned out fine.
As I mentioned a few pages back, modeling this critical angle is probably not possible as part of the sim of the crash, though it's easy to model in free flight. It's like rolling a bowling ball along the crest of a steeply cambered road for quite some distance and having it fall off to hit a specific target. The road crest is an error magnifier and the accuracy drops the longer the ball is on it. But it's much simpler to model a slow command input for left rudder, so that's how I did it.
So I suggest they may have been too slow to call for full left rudder, or they were very close to this critical angle and it took some time for the left rudder to tell.