We have seen three B737 fly into the ground/water at steep angles in recent years, and there appears to be a common thread.
- All were significantly out of trim nose down,
- All experienced zero g or negative g during the pitch over,
- All did not significantly reduce their angle of descent once they started down.
As pilots, we have to ask, how did this happen? How did these crews get themselves so badly out of trim and why couldn’t they at least try to pull out of their dives?
The answer is beginning to emerge that they could no longer control their aircraft in pitch! I suppose it is not surprising that an aircraft that is trimmed heavily nose down might be hard to keep in level flight, but there are several effects that do not appear to be widely understood that are making this situation much worse than it initially appears! My education began with a PM question to FCeng84 on control jam prevention measures implemented in the 737. FCeng84 has agreed to the sharing of his replies.; This is not his or my complete response, however. There is more detail available.
Originally Posted by Machinbird
What keeps the PCU input torque tube from back driving the jammed control column side and therefore jamming the free side. How is the necessary free motion generated?
Originally Posted by FCeng84
One of the characteristics of the 737 elevator control system that must be taken into consideration is the compliance (stretch) of the mechanical cables that run the length of the airplane from the control columns in the flight deck to the elevator input torque tube at the back end. When considering the behavior of the overall system, these cables are modeled as springs. That design came together well before my time so I don’t know if those cables were specified to have the spring constant (i.e., stiffness) that they do or if that was just a fallout of the materials used. Either way, the stretch of those cables contributes to both the handling qualities of the nominal system and the ability to cope with the consequences of a column jam.
For nominal operation (no failures) when the flight crew applies force to the column that force is transmitted through the mechanical cables to the elevator input torque tube. The motion of the torque tube (and thus the elevators) is less than it would be if these cables were infinitely stiff. Some of the column motion is essentially lost to the compliance of the mechanical cables. The amount of force it takes to move the columns (and similarly the amount of “lost motion”) is related to the stiffness of the feel and centering unit located at the back end of the airplane. At low speeds when the centering unit is softest (low forces) most of the motion of the column is transmitted to the elevator input torque tube. As speed increases and the centering unit stiffens, more and more of the column motion is lost to mechanical cable stretch and not reflected in elevator motion. The net result is that increasing airspeed causes the column stiffness to increase and the column to elevator gearing ratio to decrease. These two characteristics combine to result in a significant variation in elevator per pound of column force. This elevator per force gain variation is far greater than it would be if the mechanical cables were infinitely stiff.
An interesting side note is that when Boeing designed the 777 it was decided that the elevator control system mechanical gain variation described above should be preserved even though the 777 has no mechanical elevator control cables running the length of the fuselage. The 777 column linkages, variable feel mechanism, and the column position sensor pickoff point were designed to provide via mechanical means the same variation in column force to elevator gearing that was historically provided by the arrangement on the 737 (and subsequently 747, 757, and 767). There are specific elements in this design specified to have the proper amount of compliance to emulate the cable runs of the earlier non-FBW models.
Now back to the 737NG and the question of column jam mitigation. With one column jammed, the other column can be moved separately if sufficient force is applied to cause the cross cockpit column-to-column linkage to break free. In that event, motion of the column that is not jammed causes the mechanical cables between that column and the elevator input torque tube and the parallel cables running from the torque tube back up to the jammed column to stretch. The jammed column does not move, but the elevator input torque tube will move as it sits between the two sets of cables that are both being stretched. The forces will be significantly higher than nominal, but analysis and testing has shown that sufficient elevator motion can be commanded in this manner to provide for continued safe flight and landing with one column jammed.
Note that this system does not provide mitigation for a jam of the elevator input torque tube itself. Design and analysis has shown the probability of that event to be sufficiently remote that provisions for continued operation with that failure were not required.
To the question of elevator authority, a characteristic that has not had much (if any) play within PPRuNe recently is that of elevator blow down. The elevator actuators do not have the ability to generate enough hinge moment to push the elevators to their mechanical limits when operating at very high speeds. When this happens, max hydraulic pressure will extend the elevators as far as it can, but will be force limited. It is possible that toward the very end of the Lion Air 610 flight speed increase resulted in elevator blow down that further compromised the crew's ability to counter the stabilizer pitching moment via the columns/elevators.
After some back and forth discussion: FCeng84 had the following observation:
Either way, it seems clear that the flight crew were not able to generate enough nose up pitching moment using the elevators via the column to counter the mis trimmed horizontal stabilizer. Any time that a pilot flying such a plane with such a configuration (elevator and all moving horizontal stabilizer) senses that they are at risk of running out of pitch control authority they should get on the pitch trim in that direction first and ask questions later.