The joints within the rock mass control the stability of rock engineering. Most of the existing peak shear strength criteria are empirical models

and the sampling effect and anisotropy of shear strength of rock joints have been rarely reported. A total of 5 groups of matched resin molds were prepared based on 3D laser scanning technology and 3D printing technology. A batch of matched mortar joints was duplicated using these molds. A total of 25 groups of the direct shear test were conducted under five constant normal load conditions. A novel theoretical peak shear strength criterion incorporating the newly developed roughness parameters was proposed based on the modified relationship between contact area ratio and threshold apparent dip angles. The proposed criterion was simple and no additional fitting constant was introduced in addition to the roughness parameters. Comparisons between the new criterion and the existing empirical criteria in the literature showed that the new criterion had better accuracy in predicting the shear strength of rock joints. In addition

it was found that the new model was slightly affected by the sampling interval and could capture the anisotropic characteristics of the shear strength of rock joints.