A coupled computational and experimental investigation was undertaken to assess the feasibility of a procedure for subsurface crack identification based on inspection and/or inversion of surface displacements. The study began with the linear problem of generating contour maps of the surface deformations produce by buried fractures of known geometry and loading. An indirect boundary element formulation using the fundamental solution for tensile and shear multipoles near a half-space provided an efficient mathematical representation of the 3-D fracture. These preliminary results offered evidence for the existence of unique correspondences between crack grometry (and loading) and the resulting uplift at the free surface. The inverse problem of crack identification was then addressed beginnign with the development of a hybrid of the Marquardt-Levenberg algorithm. Numerical and physical experiments were conducted to assess robustnes sof the proposed inversion methodology. The experimental medium was a cube of transparent brittle material in which a fracture was hydraulically pressurized. Displacements induced at the surface of the specimen were measured by laser interferometry and compared to numerical results.
展开▼