Dynamic Creep Instability of Initially Straight, Centrally Loaded Columns

An examination is made of the assumptions underlying the Rabotnov-Shesterikov theory of dynamic creep instability [1] to determine a reason for the reported poor agreement between theoretical predictions and experimental data on creep-buckling times of columns [2]. It is found that improved agreement is achieved when the mechanical equation of state is modified to account for the temporary halt in creep during unloading (more precisely, when stress and stress rate differ in sign), an effect which has been reported by several authors [3, 4, 5]. The column instability times predicted by the modified theory are, like those predicted by the Rabotnov-Shesterikov theory, generally conservative estimates of the experimentally measured column collapse times. The difference between the theoretical and experimental results is believed to represent the time which elapses between the onset of instability and the completion of collapse. A detailed examination of the creep vibration of a loaded column with an idealized I-section shows that the neutral axis moves from the center of the section to one flange, back to the center, and over to the other flange during one cycle. During each cycle a small amount of permanent creep strain is developed due to the fact that the creep strain is not reduced during unloading.