Certainly, you can send a high-voltage, low-frequency signal and over-extend the driver. But I don't think that's what most failures are the result of. From 35Hz up, heat buildup must be the primary cause of failure. I imagine it probably melts or fatigues the glue that holds the voice coil in place.
Tako Tamas has done excursion measurements of the LAB12 driver in the LABhorn and in Brad Litz's basshorn. You can see the
results here. His tests indicate that at 40VRMS, the driver stays well under excursion limits in the passband of the horn, above 35Hz.
I am currently running heat tests. At full power, the pole piece temperature is hot enough to boil water. The air surrounding it becomes superheated and venting becomes less effective. By conducting heat away and removing it from the box, the temperatures around the coil can be significantly reduced.
It will be interesting to see destructive tests at various frequencies, 20, 25, 30, 35, 40Hz for example. It would also be helpful to run tests using the LAB12 in various cabinets to see where the limits were, and if failure was caused by mechanical interference or voice coil heat. I don't think anyone has ever done this, so comments about limits and failure modes are largely speculative.
From what I've done so far, I believe that the thermal limit is the primary cause of failure in the LABhorn. Without the cooling system, that would probably be the primary mode of failure in the 12Pi basshorn as well, and it may still be, but hopefully at a much higher power level. We will see.