Experimental investigation of crack evolution and failure behaviors in dual-curved flawed shale under hydraulic-mechanical loading

Investigating crack evolution and failure behaviors in flawed shale under hydro-mechanical loading holds critical significance for improving reservoir stimulation efficiency, and ensuring wellbore stability in shale gas development. While existing studies mainly focus on straight-flawed rock and rock-like under single stress state, the behavior of bedding shale with curved flaws under hydro-mechanical loading remains poorly understood. Additionally, digital image correlation (DIC) techniques, though effective in monitoring crack propagation under compression, have rarely been applied in hydro-mechanical conditions. This study developed a transparent enclosure apparatus capable of applying hydraulic pressure directly to crack surfaces in shale specimens. A series of systematic hydraulic-mechanical experiments, combined with DIC technique, were performed on bedding shale specimens with dual circular arc flaws, to investigate the effects of hydraulic pressure (P_H), bedding angle (α), and flaw curvature (κ) on crack propagation process and failure behaviors. Quantitative characterization of crack network complexity and connectivity was achieved through fractal dimension and topological analysis. Result shows that α has a greatest impact on the compaction stage, followed by P_H, and finally κ, with increasing P_H and α progressively suppressing compaction effects until near elimination. Nine crack types are identified, with α exerting a stronger influence than P_H and κ, particularly at larger α. κ predominantly controls non-tip crack initiation, followed by α, while P_H plays the minimal effect. Non-tip crack initiation usually appears in conjunction with the non-tip coalescence of rock bridges. The highest crack network complexity (D = 1.381) corresponds to moderate connectivity (normalized parameter ${\eta }_C^{\prime }$ = 0.667), whereas the greatest connectivity (D = 0.737) exhibits lower complexity (${\eta }_C^{\prime }$ = 1.363), indicating no direct correlation between the crack network complexity and connectivity. The complexity of the crack network does not exhibit a direct correlation with its connectivity, and greater complexity degree of crack network does not necessarily result in a larger connectivity. This study might provide valuable insights into the crack evolution and failure behavior of shale, contributing to the advancement of shale gas exploration strategies