The shear strength of fractured rocks is of great importance for the stability prediction of rock mass. In order to investigate the relationship of the contact configuration from micro scope of double rough surfaces in contact with the shear strength of fractured rocks，the creep shear tests are carried out for fractured rocks induced by compressive shear and indirect tension. By means of laser scanning image procession and the geometric reconstruction of the double rough surfaces in fractured rocks， and through“slice-island method”，the contact configuration and contact area are analyzed. The study shows that the contact shape displays an irregular“coast line”style for compressive shear fractures and“scattered islands”style for tensile fractures in rocks. The contact area for tensile fractured rocks is significantly greater than that for compressive shear fractured rocks. Although the formation mechanism of the fractured rocks is different，or even the applied normal forces are at different levels，the creep shear strength is directly proportional to the contact area，implying that the contact area can be a candidate index in evaluation of the influence of surface roughness to the shear strength of fractured rocks. The results remain a referential value in construction of the contact constitutive model as well as of shear strength criterion for the fracture rocks.

Time-dependent dilation of rocks is directly related to the tensile expansion of numerous inner micro-cracks. Based on the fracture mechanics，the process of initiation and propagation of micro-fractures driven by localized tensile stresses in the region near the stress concentration was discussed in the paper. Further，the dispersion damage mechanics was used to describe the behavior of the crack system. Thus the connection between the micro-fracture and the macro dilation was established. Experimental analysis proved that the equation proposed in the paper described perfectly the dilation process of rocks during loading-unloading. This dilation equation together with the deformation equation existed formed a complete constitutive model.

The failure process of interlayered rocks at different confining pressures was numerical simulated. A modified rigid body spring method(RBSM) was proposed to simulate the initiation and propagation of micro cracks. Both tensile and shear failures were considered with a failure criterion combining Mohr-Coulumb criterion and the tensile strength criterion. The rock material was represented by an equivalent discrete assembly of rigid blocks based on the Voronoi diagram. In order to explicitly account for the structural planes，a multi-step node insertion scheme was proposed for mesh generation. The proposed method was applied to an anisotropic artificial rock containing interlayered structures. Various failure modes and strengths were obtained with regards to the different orientations of structural planes. Simulation results indicated that the proposed model had the capacity to describe the anisotropic behaviors of strength and the failure modes in both qualitative and quantitative manner.

Fisher distribution is the most commonly used function of probability density for discontinuity orientations. Before using Fisher distribution to describe the discontinuity orientations，a goodness-of-fit test for them should be performed. The traditional method of goodness-of-fit tests for fisher distribution to discontinuity orientations was firstly reviewed in this paper and an improved method was then proposed. The core idea of the improved method was to adjust those orientations extending beyond the edge of the upper hemisphere projection(OEBEUHP) during the test. Three examples were given. The data of examples 1 and 2 were generated using the Monte Carlo simulation technique，which theoretically followed Fisher distributions. A comparison of the test results from the proposed and the traditional methods was made to verify the proposed method. The data of Example 3 was the application of the proposed method to a mine slope of an open pit. The results showed that the test results from the traditional method might be wrong because of the influence of the OEBEUHP. The correct test results were always obtained with the proposed method since the OEBEUHP were adjusted(such as in examples 1 and 3). The proposed method was therefore suggested to be used while in preforming a goodness-of-fit test to Fisher distribution of discontinuity orientations.

To better achieve fracturing effect in coal and make up for the shortcomings of blasting and water injection，a method of coupled cracking were put forward. The coupled cracking was applied to a steeply dipping and extra-thick coal layer. The caving ability(U) of coal after coupled cracking was forecasted with the method of BP neural network and the differences of U obtained with the different methods were compared. The effect of pressure relief in rock was analyzed with the method of microseismic monitoring. The coupled-cracking technology was found to be an effective to improve the caving ability and to relieve stress concentration for steeply dipping and extra-thick coal layer. The effect of coupled-cracking was evaluated with the analysis process of coupled cracking. The prediction with BP neural network still had a certain degree of reliability for the cross regional of high stage and conventional stage coal mass and the assessment of coupled cracking effect for coal mass was realized. The number of low level microseismic event rose slightly during the water injection and after water injection and blasting，but the energy caused by the decreasing of high level event was more than the energy of new low level event. The released energy after injection water was only 12.5% of the released energy before injection water. The total energy after the ground blasting was reduced by 51% compared with the energy before blasting. Coupled-cracking effect was manifested in the conversion of the small number of microseismic events of high energy into more events of low energy，which reduced the frequency and extent of dynamic disaster and promoted the slow release of high pressure. The coupled-cracking shifted the dynamic disasters to slow static release.

Sulfate corrosion is one of the reasons why anchor cables undergo the degradation of long-term anchoring capacity. To simulate the sulfate corrosion to the host concrete of anchor cables and the anchoring behavior after corrosion under the restriction of rock，a kind of short-bonded specimen of cylindrical plain cement concrete wrapped by a steel pipe outside and permeated by a steel strand centrally was designed. Then a long-term sulfate corrosion to the specimens in cyclic wetting and drying conditions was conducted. When the corrosion test was completed，a static pull-out test was executed to observe the anchoring process，to obtain the bond-slip curves and to define the characteristic anchoring parameters. The results showed that the restriction of rock had a significant impact on corrosion process and anchoring behavior in pull-out process. The physical damage was prevented and the chemical damage was the main way of corrosion. In the pull-out test，the spiral ribs of concrete were planished gradually and the cable was pulled out finally. The whole bond-slip curve consisted of a sudden rising initially，then a jugged rising and a rapid falling stages finally. When the deformation modulus of rock increased，the allowable design slip，design strength，ultimate slip and ultimate strength increased accordingly. And when the corrosion of the concrete was aggravated，the design strength，ultimate slip and ultimate strength decreased accordingly，while the design slip increased due to the degradation of concrete modulus.

In the construction of large underground caverns，high geostress has become the key factor which affect the stability of surrounding rock after excavation. In order to ensure the stability of surrounding rock and to evaluate the effect of supporting design，back analysis of initial geostress field is needed. By studying the distribution and the calculation method of initial geostress field，it was found that the distribution of initial geostress field influenced by the spatial location and boundary conditions was highly consistent with the regionalized variable theory of variogram function，thus a statistical method of back analysis on initial geostress field was proposed based on the variogram function. The reasonable expression of regionalized variable was studied based on two essential factors of spatial distance and depth. The parameters of spherical model suitable for back analysis of initial geostress field in underground engineering was obtained by fitting analysis with the least square method. Applying the spherical model and data of measured points，the initial geostress field was interpolated，which reflected the influence of topography. Furthermore，with the theory of incremental and varied plasticity stiffness，the excavation process of underground powerhouse was simulated and the distribution of displacement and the plastic yield areas agreed with the measured value.

The concepts of the mesoscopic damage caused by freezing and thawing，the mesoscopic damage caused by loading and the macroscopic damage caused by joints were proposed for the jointed rock in cold regions. The compound damage variable of jointed rock considering the coupling of the macroscopic and mesoscopic flaws was deduced based on the strain equivalence hypothesis of Lemaitre under freezing-thawing and loading conditions. Defining the initial damage state of intact rock as the basic state，a constitutive model considering the freezing-thawing and loading damage for jointed rock was established. Finally，the experimental data was adopted to test the validity of this model. The mesoscopic damage caused by freezing-thawing was found to be a cyclical process of fatigue damage and to increase approximately linearly with the freezing-thawing cycles. The mesoscopic damage caused by the single freezing-thawing cycle was small，and was called the local damage. The mesoscopic rate of damage caused by loading varied unevenly in the process of strain increasing and reached the maximum at the peak strain. The macroscopic damage of rock with precast joints exhibited significant anisotropy and decreased with the freezing-thawing cycles. The joint properties and the freezing-thawing cycles were the critical factors for the weakening of anisotropy. The mechanical properties of jointed rock under loading in cold regions were determined by the mesoscopic damage of freezing-thawing，the mesoscopic damage of loading，macroscopic damage with joints and the coupling effects. The frost resistance of jointed rock was described well with the compound damage variable of freezing-thawing and loading.

In order to explore the mechanisms that govern the self-weakening of strain hardening of salt rock，the tests under uniaxial compression unloading，standing still and second time loading were performed with the variation of indices of strength and strain obtained in this study. The micro-mechanism of the phenomenon of self-weakening of strain hardening was explained with the dislocation theory. After self-weakening，the peak strength and elastic modulus of rock salt in the second loading period exhibited a weakening trend with the interval time between two times loading tests. When the interval time was less than 4 h，the strength indices decreased sharply. While for the case of more than 4 h，no evident decrease was observed. When the interval time was approximately 4 h，Poisson?s ratio and isometric deformation rate of salt rock jumped upward and downward. The self-weakening effect was resulted from the behavior of crystal dislocations that was removed away from original dislocation accumulation zone due to the drive of residual stress(mainly friction and unevenly elastic dislocation force).

The internal force distribution in lattice beams of prestressed lattice anchorage system under the action of prestress，in the process of landslide and at the limit state of the landslide was investigated through large-scale physical modelling test. The test results were compared with the ones calculated with the reverse beam method. It was found that the strains in lattice beam increased gradually in the process of landslide development with two extreme values，of which the positive maxima occur at the mid-spans and the negative maxima appear at the nodes. The abstract values of the latter were greater than the former. The mid-span of the beam was in lateral tension while the node part was in lateral compression. The strains in horizontal beam were symmetrical while the strains in vertical beam were increased from top to bottom. In the prestressing stage，the forces in each horizontal beam and each vertical beam were the same. But in the limit state，the forces in horizontal beams increased from top to bottom. The internal forces in bottom beam was significantly greater than that in top beam. The forces in horizontal and vertical beam were different. The bending moments at the nodes of horizontal beam were greater than that of vertical beam. The internal forces of the test values and theoretical values with the reverse beam method had basically the same trend，which were relatively close at two ends but were different at the middle part. The method of reversing beam cannot reflect the exact state of forces in the middle of beams.

A layered structural model for site selection of gas storage in aquifer was established. Based on the results regarding the underground gas storage(UGS) and the worldwide UGS databank，and a standard of index classification was put forward. The qualitative and quantitative methods with the extensible elements and extensible dependent function were used in the model. Finally，this method was applied to evaluate the suitability of building gas storage in aquifer in oilfield of Huabei. The results of evaluation showed that the sites Da 5 and Ge 2 were suitable to build gas storage in aquifer，and that the developed approach was reasonable.

The elastic and elasto-plastic stages of rock are very difficult to be separated precisely and few of data in failure stage has been measured. To establish a model，fitting the P-S curve of rock bolts at large and affected little by the data at failure and the elastic stages，is therefore very important for the prediction of the ultimate bearing capacity of rock bolts with higher accuracy. After analysis of the hyperbolic，index and the power function，a general model was put forward to simulate the load and displacement curve of rock bolts(shorted for P-S curve). According to the actual P-S curve(through the origin，non-negative bounded，monotone increasing，infinite convergence and convex)，a mathematical function for P-S curve of rock bolts based on the general model was established. The exponential function model was revised and the hyperbola model were adjusted so that，a mixed function model was established. The application of the mixed function model simulated the P-S curve and the ultimate bearing capacity of rock bolts with higher accuracy. Key words：geotechnical engineering；rock bolt P-S curve；mixed function model of improved exponential and power function；the ultimate bearing capacity of rock bolt

S-wave phase picking is a fundamental and important part in microseismic monitoring technology. The efficiency of S-wave picking affects the analysis of microseismic source location，seismic source mechanism，seismic activity prediction and rock stress redistribution directly. Currently，the automatic S-wave identification in microseismic monitoring system is not precise enough，time-consuming and lowly efficient and the manual identification need be carried out again. By analyzing three time-domain characteristic functions of the microseismic signals in mines，including the average cross zero ratio，the average amplitude and the short time energy over short time，four S-wave identification indicators were obtained. Then，an algorithm with four indicators of three functions for S-wave identification was established. The empirical thresholds to detect S-phase were obtained according to the actual microseismic events. And a code was written on the basis of the empirical thresholds for automatic analysis of examples. Comparisons of the results from the automatic and manual identification showed that the accuracy rate of automatic identification reached within 88% and 74% within 25 ms and 12.5 ms respectively . The recognition effect was the best for the short time energy，followed by the short time average cross zero ratio and the short time average amplitude.

In order to study the influence on determining rock dynamic fracture toughness accurately with center hole，some 80 mm diameter marble specimens of Holed-cracked flattened Brazilian disc(HCFBD) with different diameter of center hole were performed in the split Hopkinson pressure bar system. The results showed that the average rock dynamic fracture toughness was 4.57 MPa•m1/2 as the dimensionless diameter of center hole r0/R∈(0.10，0.30). There was not obvious change for the test results of dynamic fracture toughness with different diameter of center hole，but there were a certain differences about specimen fracture patterns. When the center hole diameter was small，there were more secondary cracks in the process of the main crack extension. However，there were more obvious macro transfixion fracture and less secondary cracks with the increase of the center hole diameter. Which provides an important significance to promote the method of determining rock dynamic fracture toughness with Holed-cracked flattened Brazilian disc and to reveal the rock fracture characteristics under dynamic impact.

A mathematical model describing the temperature field during tunnel construction with the method of horizontal freezing is presented considering various initial and boundary conditions such as ground temperature，surface convection and latent heat of phase change，etc. The heave rate of soil frosting was defined as the instantaneous volume strain and the deformation of frozen soil was considered orthotropic. The frost heave was assumed to occur mainly along the direction of heat flow(temperature gradient) and a coefficient to describe the deformation characteristic was introduced. The transient thermal strain components(frost heave strain components) were derived after the soil temperature reached the freezing temperature. A coupled thermal-stress elasto-plastic mathematical model for the frost heave of ground was presented. The model was implanted into the finite element software ABAQUS. A shallow subway tunnel of large section applying horizontal freezing was simulated numerically and the numerical results were compared with the field measured data，which verified the reliability of the numerical model and the necessity of considering the orthotropic frost heave deformation of frozen soil in the analysis of ground displacement.

The linear deforming joint was adopted to describe the deformation behaviour of the joint. Two measuring points were arranged in the line normal to the joint and were at two sides of the joint. The waveforms from the incident side and transmitted side were measured respectively. The spectrum of waveform at the measuring point of the incident side was calculated using the spectrum of waveform at transmitted side and the transmission and reflection coefficients of the normally stress wave propagating across the linear deforming joint. The method for measuring the joint stiffness was deduced assuming the spectrum of waveform of incident side equal to the calculated one. The rationality of the stiffness measurement was verified through the numerical simulation with FLAC3D. The joint stiffness and elastic modulus of rock were shown to affect the precision of the measurement. The greater the joint stiffness，the smaller the measuring error. The larger the rock elastic modulus，the bigger the test error. The peak frequency had little effect on the testing results. Experiments were carried out to a typical joint in a slope at Dexing Copper Mine. 9 tests were performed totally. The test results were all similar with the variance of normal stiffness of 0.72 and the variance of shear stiffness of 0.56.

The deformation of mudstone foundation under the second line of Lanzhou to Xinjiang Railway was studied with in-situ tests. Overloadings of six levels such as 9.75，21.44，38.99，58.48，65.31 and 77.99 kPa were applied to three pits and the corresponding expansions with the water soaking depths of 0.7，1.1 and 1.5 m respectively were measured. The process of mudstone expansion due to water soaking was found to have four stages：brief softening and subsidence，surging expansion，slow expansion and  stable expansion. A nonlinear relationships between the water socking depths and soil expansion was obtained. When the water soaking depth was 1.5 m and the over load was 9.75 kPa，the largest soil expansion value was 41 mm. The maximum moisture content of mudstone was about 14% and would not increase with water recharging. When the moisture content of mudstone was increased around a range of values，with a bit of a lag，the expansion deformation of mudstone foundation was showed by inner accumulated energy. The relationship between the over load and expansion deformation was obtained. The deformation was mainly plastic and the rebounding value due to unloading was 6.39%–7.00% of the total deformation.

The strain localization in granular material due to its internal fabric under external load is a popular topic in the field of geotechnical engineering. However，current studies have focused on the analysis of statistical parameters，while the intrinsic link between the macroscopic phenomena and the mesoscopic mechanism has been ignored. The macroscopic and mesoscopic mechanical characteristics and the mechanism of deformation failure in the direct shear tests were systematically analyzed with the discrete element method(DEM). According to the development of shear stress ratio under the different normal stresses during the shearing process，anti-friction properties of granular material were discussed from the perspective of particle movement. A synchronic development between the anisotropic parameters of fabric and the macro-scale stress ratio was found. Through the statistical analysis of particle rotation，the friction between particles was revealed to be an important factor to maintain the relative stability of the mesoscopic structure. The morphological analysis of the networks of force chains revealed an agreement between the direction of principle stress and the principle direction of anisotropy.  The decrease of the force chain intensity and the increase of porosity during the shearing tests were discovered. Two main mechanical models of force chain structures under different normal stresses were proposed. The fluctuation of macro-scale mechanical parameters was explained with the stability of system and the cumulative release of energy，showing the microscopic mechanism of deformation localization and dilatancy.

True triaxial strength criteria and three-dimensional stress state predictions for unsaturated soils are important problems in engineering applications. Based on the two stress state variables of unsaturated soils and the characteristics of suction strength，a new twin-shear element model suitable for unsaturated soils was proposed and then a piecewise linear unified strength theory(UST) for unsaturated soils was derived. The model was validated with the data from literature about the rigid and flexible true triaxial tests on unsaturated silty sand. Some special cases and the limit line on the deviatoric plane and the limit plane in the principal stress space for the UST were analyzed. It was shown that the limit loci on a deviatoric plane for the UST covered all the convex regions. The UST of unsaturated soils not only included the UST of saturated soils and the Mohr-Coulomb criterion of unsaturated soils，but also included many new strength criteria. All the true triaxial test data of unsaturated silty sand were in the range of limit loci on a deviatoric plane for the UST of unsaturated soils and agreed well with the predictions when the parameter of the unified strength theory b = 1/2. The circumscribed Drucker-Prager criterion did not reflect the true strength characteristics of unsaturated soils and the nonlinear spatially mobilized plane(SMP) criterion was linearly approximated with the UST of unsaturated soils when b = 1/2.

The minimum buried depth of shield tunnel below the riverbed of Yellow River for Metro Line 1 in Lanzhou was studied considering the weak cemented gravel layer with rich water and under high pressure. Two methods for calculating the minimum buried depth were put forward according to the stability of cutting face and the stability of counter-floating of the segments at the rear of shield tail respectively. The upward buoyancy acted on segments was divided into two parts，namely the static buoyancy depending on surrounding liquid grout，and the dynamic buoyancy due to the vertical unbalanced force generated by formation stresses. The decisive role of upward buoyancy on the minimum buried depth was made clear，and the corresponding minimum depth was obviously larger than 1.0 D. The minimum buried depth 1.0 D in the code for design of metro(GB50517—2003) has to be improved. The safety and feasibility of the tunnel alignment design was proved with considering influence of bridge piers near the project site. The influences of local erosion of existed structures on new structures was made clear.

In order to study the influence of freezing-thawing cycles on the yielding and strength properties of silty sand from Qinghai—Tibet，the consolidated drained triaxial shear tests under different confining pressures and freezing-thawing cycles were conducted. The freezing temperature was chosen to be －5 ℃ during freezing- thawing cycles. It was found that the silty sand exhibited shear contraction and strain hardening during shearing. The types of volumetric curves and stress-strain curves were not changed after freezing-thawing cycles. The volumetric yield surface of unfrozen silty sand was expressed with an elliptic curve，while the yield surface for shearing was represented with a linear function passing the origin. The shapes of both volumetric yield surface and shear yield surface were not changed after freezing-thawing cycles. Both the volumetric yield function and the shear yield function were expressed as the power function of the plastic strain and numbers of freezing-thawing cycles. The shear strength of silty sand increased with the increasing of the normal stress，but decreased and then rised with increasing the number of freezing-thawing cycles. The cohesion of the silty sand was changed from 4.82 kPa to 2.07 kPa and the angle of internal friction was varied from 27.11°to 22.93°after freezing-thawing cycles. The linear Mohr-Coulomb criterion was adopted to describe the strength properties according to the strength variation of silty sand to the normal stress under freezing-thawing cycles.