Solid propellant research has mainly been directed towards more accurate characterization of the propellant material nature and more reliable structural analysis of the grain. Internal star grain design is among the most popular grain shapes that are used in today's propulsion system. Due to its complex geometry, stress concentrations are inevitably present around the highly curved area. Furthermore, this geometric effect together with various loading conditions throughout the grain's service life actually causes numerous defects inside its body. However, little is known concerning the three-dimensional fracture mechanism of the surface cracks which are the most common defects detected in the real rocket motor grain. After a brief evaluation of the current status of solid propellant research, stress analysis of a star grain model under internal pressure was performed by both photoelastic experiments and finite element calculations. These results illustrated the stress concentration effect around the star finger tip in addition to the global stress distribution across the whole section. Meanwhile, the deformation of the grain's outer surface was also obtained from the finite element results. A series of photoelastic experiments was conducted on cracked specimens with surface flaws emanating both on and off the axis of symmetry starting from the star finger tip. For the symmetric crack problem, cracks with different depths were intensively studied and the three-dimensional stress intensity factor (SIF) distribution was obtained for each test. These experimental data were further used to construct three analytical models, the "equivalent" radius model, the weight function model and the notch-root crack model, to expand the application range of the experimental data base so that a symmetric crack's SIF distribution with an arbitrary depth can be predicted. Moreover, surface cracks initiated off the axis of symmetry were also investigated by considering two off-axis angles. The crack shape and propagation path were achieved through a series of experiments and two methods were developed to effectively predict the possible crack growth path under sufficient pressure. The SIF distribution around the crack border was obtained for different offaxis angles and the factors that might influence the distributions were addressed based on the comparisons between the symmetric and asymmetric cracks, and the asymmetric cracks with different geometries. / Ph. D.