The main purpose of this study is to investigate the crack growth process and the coalescence mechanisms of two parallel pre-existing 3-D surface flaws under uniaxial compression in real rocks. In this study, the flaw angle, flaw length and the distance between two surface flaws (bridge length) are fixed. The bridge angle (the relative inclination between two flaws) is varied from 45º to 90º. Two observation systems (CCD camera and acoustic emission (AE) system) were used to study the propagation of cracks in the specimen. It was observed that petal cracks initiated along the interior surface of the flaw front. The propagation of the petal cracks is in three-dimensional curve shape towards to the surface of the specimen and extended to the bridge area. The coalescence mechanism depends on the bridge angle and bridge length. When the bridge angle is 90º with the bridge length equal to the flaw length, coalescence occurred. The coalescence crack is formed by the mixed mode of tensile cracks and petal cracks. But when the bridge angle is 45º with the same bridge length, no coalescence occurred. Further experimental study is required for coalescence mechanism of variables not fully investigated in the current study.
Natural discontinuities such as faults, joints or cracks are the geological fractures exist in the form of enechelon array in parallel discrete segments. A number of theoretical and experimental studies has been carried out by modelling the natural discontinues as penetrative (2-D) fractures (or flaws) to investigate its failure process and the coalescence mechanisms [1-5]. In nature, pre-existing fractures exist in threedimensional (3-D) type either fully embedding in rock mass (defined as 3-D internal flaws) or semiembedding in rock (defined as 3-D surface flaws). The crack growth and coalescence mechanisms of 3-D flaws are more complicated. A number of theoretical and experimental studies has been done by Dyskin and his co-authors [6-8] on 3-D internal flaws using both PMMA and cement material. They reported that unlike 2-D flaws, wing cracks (mode ?) sprouted from the two tips of the initial flaw and then wrapped around it. Wing cracks grew to about the length of the initial flaw and then stopped. The stress for wing crack initiation was affected by the angle of initial flaw where the shallow incline angle required lesser stress for wing crack initiation than that of the steeper angles. For the study of growth and interaction of multiple 3-D flaws, Dyskin et al  created specimens with two to several aligned inclined flaws. They reported that if the distance between two flaws was less than four times of the radii, a third large tensile fracture appeared suddenly and tended to split the sample. The results from Dyskin and his co-authors on the 3-D internal flaw are useful initial studies. They used three point bending method to produce a 3-D surface flaw. The created 3-D surface flaw extended into the specimen at about 73% of the thickness of the specimen (Fig. 1a).