The exhaustion theory of logarithmic creep was reconsidered for the case of crystalline materials in which the obstacles to dislocations were small, localized, and had maximum activation energies of not more than about 1eV. When the theory was generalized so as to include the regeneration, as well as exhaustion, of weakly obstructed dislocation segments, and to take account of thermal fluctuations which were larger than the minimum activation energies of weakly obstructed dislocations, it no longer predicted the occurrence of logarithmic creep via the process that was originally assumed. In the absence of strain hardening, it predicted the occurrence of steady-state creep. Andrade creep was predicted to occur when the strain hardening was weak, whereas logarithmic creep was predicted to occur when it was strong. It was suggested that the original form of the exhaustion theory might be applicable in the case of materials which were hardened by larger obstacles (precipitation hardening), where the maximum activation energies were too high to be reached by thermal fluctuations.

Logarithmic and Andrade Creep. A.H.Cottrell: Philosophical Magazine Letters, 1997, 75[5], 301-7