In order to study mechanical properties and failure laws of deep fractured rock mass under coupled static and dynamic loads，cylindrical granite specimens of 50 mm×50 mm were manufactured with a single artificial flaw. Coupled static and dynamic loads tests were carried out with a modified split Hopkinson pressure bar(SHPB) apparatus，and six typical levels of axial pre-stresses and three levels of impact pressures were designed. Three-dimensional digital image correlation(3D-DIC) was also applied to record and analyze the fracturing process and damage evolution of specimens. The test results show that the strength of the fractured specimen is obviously smaller than that of the intact specimen，and that，with increasing the axial pre-stress，the dynamic strength and dynamic elastic modulus generally increase first and then decrease，the combined strength generally increases while the dynamic strain generally decreases. Both the dynamic strength and the combined strength increase with increasing the impact pressure，showing that the granite has significant strain rate effect. The energy absorption ratio increases first and then decreases with increasing the axial pre-stress，but specimens will in turn release energy when the axial pre-stress ratio is 0.6–0.7. Besides，the greater the axial pre-stress is，the smaller the impact pressure is required to release energy，reflecting the characteristic of rockburst. The energy absorption ratio decreases with increasing the impact pressure. For intact specimens，high strain localization is concentrated at the end of the specimens and tensile cracks are formed，eventually causing splitting tensile failure of the specimens. For specimens with an artificial flaw，high strain localization is concentrated at the tip of the flaw or nearby，and wing cracks are formed，eventually causing combined tensile-shear failure of the specimens under coupled static and dynamic loads(axial pre-stress 0–30 MPa). However，two nearly parallel anti-wing cracks are formed at the tip of the flaw under coupled static and dynamic loads(axial pre-stress 50–70 MPa) due to the end effect，eventually causing shear failure of the specimens.

In order to investigate the mechanical properties and dynamic failure law of sandstone specimens containing weakly filled joints with various angles in deep roadways under impact loads，the impact tests of sandstone specimens containing 7 kinds of joint angles were carried out by utilizing split Hopkinson pressure bar(SHPB) with diameter of 50 mm. The crack propagation and dynamic failure processes were captured in real time with the help of a high-speed camera. The results indicated that the dynamic compressive strength and peak strain of jointed sandstone decrease first and then increase with increasing the joint angle. Moreover，the plasticity of jointed specimens decreases while brittleness increases with mounting the joint angle. The stress-time curves of specimens containing joint angles of 0°，15° and 30° perform obvious plastic platform section，which is contrary to the specimens containing joint angles of 45°，60°，75° and 90°. The stress-time curve of specimens containing joint angle of 45° shows an obvious“double peak”phenomenon. Furthermore，the joint angle controls the final failure mode of jointed specimens and the failure sequence of joint and rock matrix. The final failure mode of jointed specimens could be divided into three types，i.e.，split tensile failure，combined tensile-shear failure and shear failure. With increasing the joint angle，the failure sequence gradually changes from filled weak joint failure first to rock matrix failure first.

The mechanical properties and seepage characteristics of rock are closely related to stress，temperature and osmotic pressure. The Rock Top multi-field coupling tester was used to carry out temperature-stress-seepage coupling experiment of red sandstone under different hydrostatic pressures and seepage pressures at the high temperature of 100 ℃. The results show that，at high temperature，the total stress-strain of red sandstone goes through 5 stages including crack compaction，linear elastic deformation，stable development of microcracks，unsteady fracture development，and post-peak deformation and failure. Corresponding to rock stress-strain curve，the curve of the flow and the deviation stress involves inverse overflow，sharply reverse stable overflow，sharp rise and stable growth sections，and the permeability also show a four-stage behavior of decrease from the initial value，keeping level constant，rapid growth to the pseudo peak and declination，and stable growth to the true peak. The permeability of red sandstone under high temperature and high confining pressure shows an approximate linear decreasing trend with the constant gradient growth of the confining pressure. Under high confining pressure，the effect of the seepage pressure difference on the permeability is not obvious，and the permeability value tends to be stable. Both methods show that red sandstone belongs to typical low-permeability rocks. Under the same confining pressure and seepage pressure difference，the results obtained by the transient method and the steady-state method are similar. Under the confining pressure of 30–60 MPa，pressure condition is the main reason affecting the permeability.

The anisotropic permeability of oil shale affects the migration velocity and extent of pyrolysis oil or convective heating fluids in oil shale formations. Therefore，studying the anisotropic permeability of oil shale at high temperatures is of great significance for in situ thermal recovery of oil shale. The real-time high-temperature triaxial stress permeability testing equipment was used to study the evolution of the anisotropic permeability of oil shale at different temperatures. The results show that the permeability in the direction perpendicular to bedding is below 10－20 m2 at the temperature form 20 ℃ to 450 ℃，which is beyond the test range of the steady-state permeability testing equipment and is in the ultra-low permeability stage, and that the permeability is 2.5×10－19 to 1.17×10－17 m2 when the temperature exceeds 450 ℃ defined as the threshold temperature for vertical bedding permeability. The permeability in the direction parallel to bedding at the temperature ranging from 20 ℃ to 400 ℃ is in the lower permeability stage with a value of 2.3×10－19 to 2.9×10－18 m2，while the permeability increases sharply in the range of 10－16 and 10－15 When the temperature is higher than 400 ℃ called as the threshold temperature for parallel bedding permeability evolution. The high-temperature triaxial pulsed-decay permeability test system was used to measure the permeability of the oil shale in the direction perpendicular to bedding at the temperature of 20 ℃ to 200 ℃，and the results show that the slippage effect dominates the permeability at this temperature range. The microstructure of the sample was characterized in detail using X-ray computed tomography，and it is found that thermal cracks mainly occur in the direction parallel to rather than perpendicular to bedding. Combined with permeability and failure laws of oil shale，the reasons for the anisotropy of the permeability in different bedding directions of oil shale at different temperatures were discussed. Finally，the theoretical guiding significance of the research results in oil shale in situ mining was discussed.

The particle discrete element method(DEM) is generally used in the macro-micro investigations of rock and soil samples. In order to simulate engineering problems，it is necessary to use larger elements to represent the rock and soil masses and to assign corresponding mechanical properties. However，there are three problems in the application of the DEM：(1) it is difficult to directly get the packing model with specific mechanical properties，i.e. the modeling is difficult，(2) large-scale engineering applications are limited by the huge computational cost and (3) the multi-field and fluid-solid coupling theory and software are not perfect. On the basis of analyzing the essence and research progress of these problems，macro-micro conversion formulas and an automatic training method are proposed for fast modeling. High performance numerical calculations can be completed based on the matrix discrete element method and the pore density flow method is introduced to realize fast multi-field and fluid-solid coupling simulations. Based on these solutions，a high-performance discrete element software MatDEM was developed from scratch，and large deformation，multi-field and fluid-solid coupling numerical simulations were realized. Finally，key science and technology problems in the future researches were discussed.

Three-dimensional morphology of rock joints is the main factor affecting the shear mechanical properties. Aiming at the change of the morphological characteristics of rock joints with different sampling intervals，a morphological classification method based on the limit sampling interval is proposed，the two-order roughness of the macroscopic undulation component and asperities of joint surfaces is obtained，and further the size and interval effects of the two-order roughness of four types of natural joints are studied. The results show that the total roughness of joints is close to the second-order roughness at a small sampling interval while close to the first-order roughness at a large sampling interval，which verifies the feasibility of morphology classification using the sampling interval. The first-order roughness has obvious negative size effect while the size effect of the second-order roughness is not obvious，indicating the size effect of the joint roughness is mainly reflected by the first-order roughness. The analysis of the geometric mechanism of two-order roughness shows that the larger undulation wavelength of the macroscopic undulation component compared to asperities is the key factor that causes the size and interval effects of the two-order roughness. Through the mechanical analysis of the shear model test，it is found that the two-order roughness contributes differently at different shearing stages. The study of the joint shear strength should comprehensively consider the two-order roughness characteristics at different scales.

In order to observe the evolution process of loaded coal fractures，CT real-time scanning test of coal samples during the uniaxial compression process through using a CT scanning system of the loaded coal-rock industry was carried out，and the CT scanning images were imported into VG Studio MAX image analysis software to reconstruct 3D digital model of coal and further to obtain the spatial distributions of fractures within coal samples. The gray histogram of coal sample internal structure and fracture fractal dimension are obtained by analysis and calculation with MATLAB software，and the dynamic evolution characteristics and fractal law of fractures in coal samples are analyzed comprehensively. The results show that the analysis results of gray-scale histogram and VG Studio MAX software can be used to qualitatively and quantitatively describe the dynamic development of fractures in coal samples. In the total stress-strain process，the fracture area，peak area of fracture spectrum and three-dimensional fracture volume of the loaded coal sample decrease first and then increase，which truly and objectively reflects the dynamic evolution law of the fracture structure in coal samples and fully reflects the control effect of external load and deformation on the fracture development. The fracture fractal dimension mainly experiences four development stages including slow decrease，steady increase，large sudden increase and slow increase，which is consistent with the fracture dynamic change law，can effectively depicts the evolution and development process of internal fractures of the loaded coal sample，and hence can be used to predict the instability and failure of loaded coal samples.

Mortar is widely used in rock engineering，such as mortar jet supporting tunnel and landslide grouting reinforcement. Due to the change of temperature，the interface transition zone between rock and mortar is easy to form the weak area of micro cracks. At the same time，the underground rock structure often bears the dynamic load such as blasting，impact and earthquake. Therefore，it is of great significance to study the strength and failure characteristics of the interface transition zone between rock and mortar. In this paper，a rock-mortar bimaterial Brazilian disk configuration is proposed，and the impact experiments of bimaterial Brazilian disk with two kinds of interface roughness(smoothness and roughness) are carried out by using the split Hopkinson compression bar system. At the same time，in order to compare the difference of failure performance between single material and bimaterial，Brazilian disc specimens of single granite，sandstone and mortar were also prepared. In order to overcome the problem of uneven stress distribution of the bimaterial Brazilian disc，the central tensile strength of the disc was determined by the experimental numerical method. In addition，two kinds of numerical models are built with Python language to simulate the failure process of bimaterial ITZ. The results show that，when the dynamic loading rate is the same，the interface roughness has little effect on the strain rate of the specimen but has a greater impact on the tensile strength of the specimen，and that，compared with the granite-mortar bimaterial specimen，the interface roughness has a greater impact on the dynamic initial tensile strength of the sandstone-mortar bimaterial specimen. It is also shown that，based on the numerical model，the shear stress exists in the center of the disk model and is far less than the tensile stress，and that the failure phenomenon of the numerical model embedded with cohesive element are close to that of the test specimen，which verifies the effectiveness of the method.

In order to study the effect of test system compliance on the determination of hydraulic fracturing parameters and principal stresses in detail，hydraulic fracturing field tests were performed by using a recently improved single-cycle hydraulic fracturing equipment which can monitor the pressure of wellhead，test interval and packers during the entire test process. The results show that the values of the HF parameters Ps，Pr，Pb and SH derived from the wellhead pressure data are generally larger than those determined from the test interval pressure data due to the effect of the test sytem compliance. The absolute difference of Ps，Pr，Pb and SH between the wellhead and test interval results are 0–0.375，0–0.75，0.125–0.875 and －0.25–0.875 MPa，respectively，and the relative differences mainly range within 0–5%，0–6%，0–4% and 0–5%，respectively. The absolute difference value of Pb is considered to be controlled not only by the test system compliance but also by the mechanical properties of the strata and in situ stress state. The affecting mechanism of test system compliance on the Pr absolute difference magnitude is relatively complex，and test interval flow rate data should be supplemented. In addition，a new method referred as shut-in test was proposed to obtain the shut-in pressure more accurately during HF tests. The results relating to shut-in tests indicate that the method can effectively eliminate the influence of the test system compliance and the pipeline frictional resistance. It is suggested that，when only the wellhead pressure data is used，Ps，Pr，Pb and SH should be revised according to a maximum proportion of 5%，6%，4% and 5%.

Digital core establishment，as an ideal model to study the physical and mechanical properties of rock, provides an undifferentiated numerical simulation. However，high level of accurate and efficient modeling restricts the promotion of digital core reconstruction technology. The traditional methods are time-consuming and laborious in processing CT slice based scanning data，due to limited number of scanning layers and the pore-fracture recognition depending on the traditional threshold segmentation algorithm. Taking coal as an example，the artificial intelligence recognition is introduced to realize the recognition of four micro phase states of pore，fracture，high-density mineral and coal matrix，and the fractal reconstruction is carried out for filling in information gaps. Data sets of four micro phase states are established and enhanced based on micro CT scanning，and a labelling software is developed for effectively distinguishing four kinds of micro-phases of materials. Especially for improving the efficiency and precision of identification，the FCN architecture is optimized and the Crack-FCN network structure is proposed，which has few network layers and low error rate. Moreover，the Potrace algorithm is introduced to quantitatively calculate fracture area，length and width，and the centerline extraction algorithm is introduced to effectively determine the complex topology. Considering the fractal similarity of fractured surface and to solve the problem of missing information between two adjacent CT slices，a fractal reconstruction algorithm is developed dependent on the local self-similar property and then optimized to improve the computational efficiency. Compared to the line interpolation and cubic spline interpolation，the fractal interpolation is more effective to describe the local roughness，and more importantly，the accuracy of intelligent recognition will continue to be improved with the continuous enhancement of data-set. This paper breaks through the traditional view and introduces the FCN into construct digital core of rock，and provides new technical support for the efficient and accurate establishment of numerical modelling.

In order to study dynamic compressive properties of composite layers of rock and steel fiber reinforced concrete(SFRC) under impact load, dynamic compressive experiments of rock，SFRC and rock-SFRC composite layer were conducted using split Hopkinson pressure bar with six strain rates，and dynamic compressive strengths of specimens were compared with their static compressive strengths. The results show that SFRC enhances the static and dynamic compressive strengths of the composite layer. The dynamic compressive strength，energy absorption and dynamic strength increase factor(DIF) of rock-SFRC composite layers were sensitive to the strain rate. The dynamic compressive strength and dissipated energy of composite layers increase with increasing steel fiber content at a similar strain rate. With steel fiber content of 80 kg/m3，the dynamic compressive strength and dissipated energy of the composite layers respectively show a maximum increase by 30.1% and 53.9% compared with those without steel fiber. Steel fiber within content of 60 kg/m3 can effectively enhance DIF. With increasing the strain rate，the damage modes of composite layer specimens are divided into four types Including peripheral tensile strain failure of the concrete layer，crack penetrating the composite layer，core-retaining and overall fragment. The dynamic mechanical properties of rock-SFRC composite layers can provide a basis for the study on the mechanical mechanisms of supporting structure under impact loads.

The discontinuous deformation analysis(DDA) method is widely used in geotechnical engineering. Different from the two-dimensional DDA (2D-DDA)，the three-dimensional DDA(3D-DDA) has more remarkable capability to analyze practical deformation and stability problems of jointed rock masses. For 3D-DDA，however，due to the complexity of block contact，the increase of unknowns and the management of data and memory in the program demand a more stable and efficient algorithm for solving global equilibrium equations. For the successive over-relaxation(SOR) algorithm used in the original DDA program，the improper selection of the SOR factor will make the solution of the global equilibrium equations unable to converge. In the present work，based on the compute unified device architecture of GPU，the parallel Jacobi-preconditioned conjugate gradient(JPCG) algorithm is developed to solve the 3D-DDA global equilibrium equations. Simulation examples are given to demonstrate the acceleration effect of the JPCG algorithm combined with the GPU technology. Compared with the original serial SOR algorithm，the parallel JPCG algorithm not only avoids the influence of the SOR factor on the convergence but also improves solution efficiency，which creates favorable conditions for the 3D-DDA to be used in practical rock mechanics and rock engineering problems.

In order to understand the dynamic strength change law of compacted loess caused by wetting-drying cycles，a series of wetting-drying cycles tests，soil dynamic triaxial tests and scanning electron microscope tests were carried out under different wetting-drying cycle paths，and the dynamic strength and microstructure images of compacted loess were obtained. The results indicate that the dynamic strength and dynamic strength index of compacted loess decrease first and then increase with increasing the number of wetting-drying cycles，and that the turning points of the ，cd-n and  curves are obviously different due to the influence of the wetting-drying amplitude. The degradation degree of the dynamic strength increases first and then decreases with increasing the number of wetting-drying cycles，and even has a negative value. The critical wetting-drying cycles number nc is the limit of the whole structure damage and the whole structure strengthening of compacted loess，and changes with the wetting-drying amplitude. The macro and micro structure characteristics of samples in the process of wetting-drying cycles indicate that the increasing number of pores in the humidifying process and the cracks on the surface and in the inside of samples during dehumidification lead to the structural damage and strength reduction of compacted loess. The main reasons for structure strengthening and strength increasing are that soil particle structure develops to mosaic structure，pore structure develops to uniform small pore and matrix suction makes soil sample tend to be dense.

The analysis of lateral dynamic responses of high cap straight piles in saturated sand liquefaction sites is a hot and difficult point for geotechnical engineering. A 2×2 group straight pile model in saturated sand liquefaction sites was designed，and the centrifuge vibration test was carried out to analyze the dynamic responses of piles and soil in liquefaction sites. The sand liquefaction large deformation constitutive model was introduced into ABAQUS finite element software platform and the finite element mesh adaptive adjustment technique was used to overcome the large deformation distortion problem. A 2D finite element model for simulating the static and dynamic coupled nonlinear interaction of pile foundations in liquefiable sites was established and compared with the test. The test results show that，in the case of the sine wave with a peak acceleration of 0.3 g，the liquefaction rate of the saturated sand foundation is very fast. The peak accelerations of both the pile cap and the soil will not exceed the input wave peak，and the acceleration of the pile cap begins to attenuate after the ground liquefies. The development of the excess pore water pressure in saturated sand soil directly affects the acceleration response，and the liquefaction of soil directly leads to acceleration attenuation. The acceleration dynamic response behavior by numerical simulation is similar to that from the experimental test，and the scaled-down results of numerical simulation are approximately equal to the experimental results. The excess pore water pressure and the excess pore pressure ratio obtained from simulation and test are consistent. The simulation results also indicate that the shallow soil liquefies more obviously than the deep soil and that the displacement of the pile cap is smaller than that of the test.

In-situ and laboratory test data are often quite limited，by which it is hard to determine the statistical characteristics of geomechanical parameters. Fortunately，the stochastic back analysis method provides an approach to overcome the shortcoming. In this paper，three stochastic back analysis methods(i.e.，DREAM(zs)，BUS and aBUS) of geomechanical parameters accounting for the effect of spatial variation are developed，and the basic principles of the three methods are compared from the aspects of generation of random samples，convergence criterion，model evidence and estimation of posterior probability of failure. Two slope examples are investigated to further compare these three methods systematically on the convergence，computational accuracy and efficiency. Based on these，the DREAM(zs)，BUS and aBUS methods are respectively recommended to give priority to tackle different stochastic back analysis problems. The results indicate that the DREAM(zs) method has good computational accuracy and efficiency only for dealing with low-dimensional problems，that the BUS method，in which the value of likelihood function multiplier has to be determined before the operation of subset simulation，is preferable to solve high-dimensional problems involving the spatial variability of mechanical parameters and intensive computations likelihood function，and that the aBUS method，which does not rely on the likelihood function multiplier and has good computational accuracy，is fairly suitable for analyzing high-dimensional problems involving the spatial variability of mechanical parameters and less computation of likelihood function，although it is time consuming to quantitatively determine whether the computations converge to the accurate results.

According to the secondary development platform provided by FLAC3D，the anisotropic structured clay model(ASCM) is implemented using the coding language of C++. The model is first verified by simulating the undrained triaxial tests on K0 consolidated Shanghai clay. Numerical simulation of ground loss during tunnel construction is then performed，and the influences of clay structure and adhesive stress on ground surface settlement are analysed. The results show that the user-defined ASCM constitutive model can be used for numerical simulation of 3D tunnels and that the calculated ground surface settlement curves due to tunnel construction agree well with measurements of the centrifugal model tests. The profiles of the ground settlement curve like a “V” with the peak point located near the centre of the two tunnels with a spacing of 1.5D. With increasing the spacing distance of twin-tunnel，the settlement curve shape changes from “V” to “W” with two peak points located above the tunnel axis. Finally，the influence of clay structure on tunnel construction is numerically analysed. It is shown that the influence of clay structure on the ground deformation shape under the disturbance of ground loss in tunnel construction is not significant while that the ground settlement at the tunnel axis increases significantly due to structural deterioration. The influence of the clay adhesive stress on the ground settlement and deformation shape under the disturbance of ground loss in tunnel construction is not significant.

The particle breakage of granular geomaterials under loading alters the particle composition and further affects their mechanical properties. In order to reveal the effect of particle breakage on shear strength characteristics，a series of true triaxial shearing tests with different stress paths on sandy gravel were carried out by using a large true triaxial apparatus，and the breakage mechanisms of different particle breakage patterns were described. It is revealed that particle breakage is an important factor for the nonlinearity of the shear strength. A power function fitting strength formula was established and the non-linear shear strength indexes were calculated. The influence of particle breakage on the critical state line in meridian plane was described by introducing the strength factor，which is defined by the ratio of the area enclosed by the non-linear strength curve and the linear critical state line to the area under the linear critical state line. The larger the strength factor，the more significant the nonlinear characteristic of the critical state line in meridian plane，and the larger the mean spherical stress，the greater the impact of particle breakage on the shear strength. The sieving results show that the particle breakage rate by Marsal is exponentially related to the strength factor. The variation rules of the internal friction angle and interlocking strength with the index of fine-grained breakage were sorted out. It is shown that the particle breakage rates of particles larger than and less than 1 mm respectively increase and decrease with increasing the mean principal stress. The internal friction angle and occlusion strength show a certain correlation with the change of the breakage rate.