Engineering rock mass in cold regions has obvious freeze–thaw damage as a result of extreme differences in temperature, rainfall, snow, and other factors, which is one of the main causes of frequent geological disasters. Therefore, it is important to investigate the deterioration mechanism and evolution laws of rock mass freeze–thaw damage. Considering a hydropower station’s left bank slope in a cold region, model testing and numerical testing of slope rock mass failure under freeze–thaw conditions are here carried out by developing a generalized model. The results reveal that heat is transmitted from the outside to the inside of the slope and that the rate of temperature change varies with depth; the frost-heave force causes tensile cracks in the rock mass, with crack propagation taking the form of a circular arc; the presence of an original structural plane influences the propagation direction of the frost-heave crack, whereas the freezing rate of the fissure water influences the amplitude and growth rate of the frost-heave force. Additionally, a novel method of measuring frost-heave force is proposed. The largest frost-heave force caused by water–ice phase change is 19.3 kPa, which is equivalent to 3.17 MPa of the actual slope.

A comprehensive investigation to the characteristics of valley deformation in Xiluodu project is briefly reviewed. Extensions for hydro-mechanical coupling simulation with DDA method for reservoir impoundment has been firstly derived. The DDA numerical model by considering interactional aspects of original geological cross section, geo-stress state, and the reservoir impoundment, etc., is then established. Simulations for the given numerical model with different initial geo-stress state after the water impoundment are performed, and the results are compared with that of typical valley deformation measurements, and at last, the deformation mechanism in causes of changing of effective stress in the particular geological condition is discussed.

The discontinuous deformation analysis (DDA) method provides unique advantages in simulating the mechanical properties of coarse granular materials. However, the DDA method cannot automatically generate numerical samples of coarse granular materials that can strictly meet the gradation requirements. Therefore, to solve this problem, a new generation method is proposed for generating a block system of coarse granular materials. First, a particle is randomly selected from the basic morphology database and randomly scaled and rotated to match the particle size requirements of the current gradation, based on the relevant requirements for laboratory tests of coarse granular materials. The accumulated particle area will enter the next gradation when it meets the area proportion requirements of the gradation, and this step is repeated gradually from the large particle size group to the small particle size group. Second, the random drop area, which has the same diameter as the coarse granular sample but a height several times that of the sample, is defined. The particles are randomly placed in this area in order of large to small particle size groups, and the requirements of nonoverlapping and noncrossing between particles or between particles and boundaries must be met. When the efficiency of random dropping is very low or even fails, it is important to keep increasing the sample height multiplier. Third, a DDA-based compaction operation is performed to obtain a stable block system with close particle contact. Fourth, a coarse granular material sample that fully meets the gradation requirements is obtained by gradually adjusting the height of the prepared sample. Finally, an example is used to demonstrate the feasibility of the proposed algorithm.