X-ray computerized tomography (CT) is a nondestructive technique for obtaining quantitative information about the mechanisms governing the damage process. The detection of damage, that is the degree of damage and the size of the damage zone is a particular application where CT will provide new information. This information will contribute to the development of constitutive models and provide new insight on the interpretation of mechanical tests. For example, the detection of the transition from homogeneous to nonhomogeneous deformation in a triaxial compression tests can be determined. Also, issues such as, is strain softening a material property or not, can be addressed. Preliminary results are presented which show how CT information can be used to obtain a macroscopic measure of specimen damage and the extent of damage around a fracture.
Macroscopic failure of test specimens in the laboratory results from localization of damage along one or several discrete planes. In brittle materials where damage is the primary mechanism of inelastic deformation, localization of damage implies the localization of deformation. Therefore, it is important to determine when the deformation and damage begin to localize because at this point the strain field is no longer homogeneous. Constitutive models based on laboratory data usually assume that the stress and deformation are homogeneous right up to the point at which the specimen disintegrates. However, a careful examination of the test specimen will indicate that the assumption of homogeneous deformation is clearly violated as the specimen bulges or develops visible shear bands. Certainly by this time the homogeneity assumptions that are blithely used to convert load on the specimen to stress and dimensional changes in the length and diameter to strain are not valid. However, the point in the test at which the assumptions of homogeneity are violated is generally not known. Uncertainty in the point at which to stop regarding laboratory measurements of load and deformation as scaled values of the homogeneous stress and strain fields causes uncertainty in the amount of laboratory data that should be used to write constitutive models. Writing models that simulate specimen response after the damage and deformation localize causes structural behavior to be falsely modeled as material behavior, and leads to much controversy about the behavior of real materials. Especially contentious are the rate-independent strain-softening models that purport that the deformation in the post-peak regime is truly a manifestation of material response, and not that of the test specimen; a structure in which the deformation and damage has localized. Evidence that strain softening is a structural, rather than material phenomena, is contained in the vast amount of data analyzed by Read and Hegender (1984) that showed that the peak stress and stress-strain curve in the post-peak regime depended on specimen size, shape (e.g., L:D ratio) and lubrication at the platen-specimen interfaces. In addition, they have identified a weakness in past studies on strain softening in that they have provided little or no information on the physical condition of the specimen in the strain softening regime.
