An ejection of a control rod could conceivably give rise to very rapid rates of reactivity increase. The accident might be feasible if the control assemblies were only restrained by their own weight and if some ejection mechanism could be envisaged.

However, in fast reactor power systems, the control rods are generally driven into the core and their followers and drive shafts remain above the rods. Only in the case in which a rod was in the act of dropping might its upward travel be unimpeded. In other cases physical restraint would not allow more than a few thousandths of an inch of upward travel.

Suggested mechanisms for rod ejection have included coolant flashing immediately below the rod tip or simple hydraulic forces. However hy­draulic forces are not sufficient to move a large control rod and coolant flashing would indicate that some other accident was already in progress and it would be most unlikely that the required force could be obtained in this way at exactly the instant of rod disengagement from its drive. Indeed, if a credible rod ejection process could be postulated, it would be relatively simple to include an anti-ejection device in the drive mechanism. However, credible ejection mechanisms cannot be envisaged, and a rod ejection cannot be considered a CDA initiator in a large fast power reactor.