I must do a bending analysis and i'im looking forward to force displacement instead of using a force. I've looked at different work where they add a spring to take the force back during the buckling, but it's not really applicable here.
I have never used the fully non-linear Nastran analysis options but, in linear analysis one could use the Large Mass approach, i.e. simply apply a Force = Mass*Acceleration to a mass that is, say 1000x heavier than your model.
To good approximation, this will enforce whichever time history you wish to insert to the object.