The linear modeling of concrete creep with aging has matured, and it is now appropriate to develop a nonlinear theory. This research will establish such a theory, based on micromechanics analysis of the solidification process of portland cement and nonlinearities due to strain-hardening and development of backstress. The model will first be developed for basic creep at constant humidity and temperature, and is then generalized to include the effects of variable humidity and temperature. This model is based on the hypothesis that the rate of bond ruptures (which is the source of creep) depends on the microdiffusion flux of water molecules in cement gel. The model will be extended to include probabilistic aspects of materials parameter uncertainty, including correlation of random parameters, and fluctuation of environmental conditions. Random effects of environment are handled by a new sampling scheme based on spectral approximations and approximate solutions from diffusion theory. A numerical algorithm for step-by-step integration in time in finite element programs will also be developed. Creep experiments will be conducted with a novel large triaxial-torsional testing machine with temperature control. Hollow test cylinders will be subjected to triaxial loading at various temperatures with confining pressure which prevent cracking or simultaneous strain- softening due to drying or temperature change. The results will improve creep predictions for radioactive waste storage, large span bridges nuclear reactor accidents, concrete shells and tanks, buildings and other concrete structures.
