Cracking mechanism and strength criteria evaluation of granite affected by intermediate principal stresses subjected to unloading stress state

Abstract

Extensive true-triaxial unloading tests are conducted on rectangular Miluo granite specimens to investigate their crack propagation behavior and peak unloading strength properties. Experimental results demonstrate that their failure mode transforms from shear to slabbing with the increase in intermediate principal stresses σ2. Under a lower value of σ2, the dominant failure mode represented by the macro shear band is composed of several meso-extensional cracks parallel to the maximum principal stress σ1, and the fracturing process indicates intensive dislocation and slide owing to the normal force acting on extensional cracks. However, the increase in σ2 exacerbates the extensional cracks and further propagates them toward the end of the specimen along the direction of σ1, if σ3 is unloaded to 0. This is primarily due to the enhanced ability to resist mutual slide and squeeze among extensional cracks in the shear band, and the promotion of crack initiation and propagation in the unloading direction (σ3) owing to the provided free surface. The transformation of the cracking mode is revealed via three-dimensional combined finite/discrete element modeling, scanning electron microscope observations, and acoustic emission signal feature (average frequency, rise angle, Acoustic emission (AE) counts, and accumulated AE energy) analysis. The peak unloading strength exhibits an initial increase followed by a decrease as a function of σ2. Four rock-strength criteria are employed to evaluate the true-triaxial data with emphasis on predictability in practical applications and accuracy between best-fitting functions (experimental data) and theoretical (predicted) solutions. It is demonstrated that modified the Wiebols–Cook criterion and modified Lade criterion cannot accurately estimate the unloading strength throughout the true-triaxial data. Although the Mogi 1971 criterion fits the experimental data (higher correlation coefficients) well, it cannot be correlated to strength parameters (cohesion and internal frictional angle). Hence, it fails to predict the real strength values if complex true-triaxial unloading tests are not performed. The linear Mogi criterion is recommended for predicting the true-triaxial unloading strength of granite.