Rocks, because they are composed of different minerals, usually have a heterogeneous structure. This heterogeneous structure always affects the process of crack development and failure of rock samples. On the other hand, in most studies related to rock fracture mechanics, rock samples are considered homogeneous or the effects of sample heterogeneity are not considered. In this study, an attempt has been made to simulate the heterogeneity of mechanical properties in rock samples by changing the dispersion (variance) of rock mechanical properties and its effects on the failure mechanism of rocks are investigated. For this purpose, PFC software has been used to simulate rock samples. This software is calibrated based on the micromechanical parameters applied to the joints and comparing the mechanical properties of the simulated sample with the results of laboratory tests. The joint model used in the modeling is the planar joint model. In this model, each bond is a limited surface between two hypothetical surfaces, which is divided into several parts, each of which has an initial strength. Accordingly, one of the effective parameters that affects the mechanical behavior of the simulated samples is the adhesion strength between the bonds. In this study, considering different dispersion values ​​for bond adhesion, samples with different inhomogeneities have been simulated. Also, to examine the subject more comprehensively, an attempt has been made to use different rocks with different brittleness indices. For this purpose, three rock or quasi-rock samples, including gypsum, sandstone and granite with brittleness index ratios of 6.6, 13.7 and 22.4, have been selected. The results of the simulations show that the fracture shape of the sample changes with the change in the standard deviation of adhesion between the bonds. Therefore, to determine the standard deviation of adhesion in the simulation, the fracture shape of the simulated sample must match the fracture shape of the sample in the laboratory.