Since the traditional rock block index(RBI) cannot represent the spatial characteristics of the rock structure，a quantitative system integrating the evaluation index of rock structure at the tunnel axis direction and the tunnel radial direction based on RBI was built. A two-dimensional fracture model of the tunnel work face in tunnel project was established first based on the digital imaging，and the method of sampling line was employed to obtain RBI of 19 scan lines which were set on the model using the computer. The rose diagram of the RBI was also proposed in which the radius in each direction and the area of the rose reflect the RBI of the scan line and the comprehensive evaluation index of rock structure(called Z-RBI) on the working face. The six color classification of the rock structure by Z-RBI was presented，and the prediction method of the rock structure at the tunnel axis direction was also proposed and verified.

Failure characteristics and brittle evaluation of reservoir sandstone are of great significance in oil and gas drilling engineering. The sandstone cores from the well of Shengli Oil Field were studied with the developed physical tests under condition of triaxial compression and numerical simulation. The results show that the failure mode of interlayer and reservoir cores under the effect of confining pressure is mainly the shear failure. Under the same level of confining pressure，the interlayer cores are more brittle than reservoir cores. The brittleness tendency can be evaluated by the brittleness index B effectively. Besides，the relevance between the main mechanical parameters of sandstone and the brittleness index B is analyzed as well. Under the condition of the same confining pressure，the peak strength，expansion point，elastic modulus，Poisson?s ratio，residual strength and residual strength coefficients of cores decrease with the increase of brittleness index B. The relationship between index B and parameters is linearly fitted well. The intrinsic factor of low brittleness is the high clay mineral content of the oil layer rock. The microscopic brittle characteristics of two kinds of sandstone are shown in the CT scans under uniaxial compression. In the case that other parameters are fixed，the numerical cores are more brittle with the increase of the elastic modulus，ratio of compressive strength to tensile strength and internal friction angle. However，the brittleness of cores reduces with the increase of Poisson's ratio，residual strength and residual strength coefficient. In addition，there is little relationship between the brittleness of cores and its strength. The modes of acoustic emission are dominant by the pre-main-after shocks and group shocks. The essential relationship between the index B and the acoustic emission modes is summarized based on the results of physical experiment and numerical calculation. Under the condition of same loading rate，the acoustic emission modes in the low brittleness zone，middle brittleness zone and high brittleness zone are swarm shocks，pre-main-after shocks and mass shock. Lastly，the mechanism of acoustic emission modes was verified with those of limestone and different types of coals.

In order to study the deformation of rock under cyclic loading and unloading，the tests to the 56 sandstone samples from Three Gorges Reservoir area were carried out. The concept of stress-strain hysteresis was modified according to the asynchronous phenomenon from the tests. The phenomena that the stress is ahead of or behind the strain exist at the same time，so it is more reasonable that the hysteresis phenomenon of stress-strain is named as the stress-strain asynchronous phenomenon. On the basis of the stress-strain asynchronous phenomenon，the asynchronous phenomenon of the stress rate-strain rate should be considered. After the analysis of the asynchronous mechanism of the cyclic loading and unloading test，four concepts，the stress-strain attraction point，the stress rate-strain rate attraction point，the stress-strain compensation mechanism and the stress rate-strain rate compensation mechanism，were proposed. Three findings were obtained. The stress-strain and stress-strain rate synchronization did not occur simultaneously. The asynchronous phenomenon is inevitable and synchronous phenomenon only exists at a certain point. The relationship of stress-strain is the process of spiral cycle progression. The asynchronous phenomenon of the stress rate-strain rate was found for the first time in the cyclic loading and unloading tests and the definition of three asynchronous stages was proposed. The predicted and actual deformation rate，the apparent elastic modulus and the instantaneous Poisson ratio of the samples with the time were analyzed and three methods were put forward to judge the asynchronous stages quantitatively. Besides，by analyzing the mechanism of stress-strain rate based on the same frequency and different peak strength tests，a conclusion is reached that the of the sample and the coincidence degree of  increases with the increase of the peak intensity.

Considering the three stages of methane flow in coalseam including desorption，migration and seepage，Darcy?s law was employed to describe gas migration in coal matrix and fracture，and the gas mass transfer flux between coal matrix and fracture was served as a coupling term to develop a mathematical model of methane flow in a dual-porosity，dual-permeability coal seam and its dimensionless model. The finite difference method was applied to develop a numerical solution for the model. The numerical results show that the decline rate of gas pressure and content in fracture is much faster than that in coal matrix. The gas pressure distribution and gas content in coal matrix are heterogeneous and unsteady over time. A higher permeability of fracture，a higher original gas pressure in coal，or a lower gas pressure outside coal wall，lead to a higher gas emission rate. The dual-porosity and dual-permeability model for methane flow in coal seam is validated to be correct by comparing the simulated results and the field data from Panyi mine.

Near-fault ground motion mechanisms have seldom been assessed in the seismic studies of underground rock caverns，especially the caverns controlled by large geological discontinuities. A nonlinear joint model was adopted to simulate the unfavorable geological discontinuities. The near-fault pulse-type，near-fault non-pulse-type and far-field ground motion records collected from the NGA-West2 database were used to analyze the influence of near-fault ground motions on unfavorable geological discontinuities. A damage potential index(DPI) was discussed and proposed for unfavorable geological discontinuities. The surge chamber #1 of the Baihetan hydropower plant，which is controlled by the interlayer shear weakness zone(ISWZ) C2，was used as a case study to investigate the differences between the pulse-type near-fault ground motion，non-pulse-type near-fault ground motion and far field ground motion. The significant velocity and displacement and the higher long-period response spectrum were found to associate with the pulse-type near-fault ground motion，while the non-pulse-type near-fault ground motions displayed the similar characteristics as the far field ground motions. The velocity pulse is responsible for the destructive capabilities of near fault ground motions，which may cause the unacceptable sliding along the discontinuity，leading to potential cavern failure. In comparison with several ground motion parameters，PGV was shown to be the most suitable DPI for large geological discontinuities under the earthquake excitation，which was true for both the near-fault and far-field ground motions. PGV was verified to be an effective DPI via the seismic analysis of the Baihetan surge chamber #1. The cavern becomes fragile when subjected to the near-fault ground motions. Therefore，the special seismic reinforcement measures are recommended. These findings may provide a reference for the seismic design of underground caverns.

In order to investigate the seismic response of shield tunnels in the longitudinal direction cross the soft-hard stratum junction，a set of shaking table tests with geometric similarity ratio 1∶40 were carried out based on the equivalent longitudinal stiffness model. The spectrum of the acceleration response for both the strata and the tunnel structure were studied as well as that for the tunnel strain. The test results were compared with the numerically calculated ones. The results show that the spectrum properties of the acceleration response for the soft-hard junction stratum and the tunnel coincide with each other well，and both exhibit the double predominant frequency. As the depth of the soft-hard junction stratum decreases，the acceleration does not increase monotonically when the excitation earthquake frequency approaches the predominant frequency of hard stratum. The strain of the tunnel in the soft-hard junction stratum increases significantly and the maximum strain occurs on the soft side of this stratum. The enlarged longitudinal strain area of the tunnel distributes within the scope of 2.5–3.5 times of the tunnel diameter on both sides of the strata interface. The strain of the tunnel in each cross section arranges in a descending order as crown arch，inverted arch and side-walls. The strain difference between the crown and inverted arch as well as between the side-walls increases obviously compared to that in the uniform stratum. The soft-hard stratum junction not only enlarges the longitudinal internal force of the tunnel but also alteres its longitudinal bending direction.

In order to evaluate the stability of rock blocks under various complicated conditions，a general method for stability evaluation of blocks using numerical approach was proposed based on the existed block identification method using element reconstruction and aggregation technique. The proposed method places the focus on the meshes that contain block information. Firstly，interface elements were introduced to simulate the slip and open behavior of rock block surfaces. Then，the method of determining the removability and kinematic patterns of blocks was proposed. Afterwards，the definition of the safety factor of block and its reinforcement method are proposed based on the strength reduction concept. The stability evaluation of blocks is thus realized using the numerical approach. Compared to the rigid body equilibrium method，the proposed method is able to take the influences of the initial geo-stress，rock deformation and anti-shear capacity of non-slipping interfaces into account in evaluating the block stability. Under certain conditions，the results of proposed method are equivalent to those of rigid body equilibrium. Finally，the reliability and effectiveness of the proposed method and its relationship with the rigid body equilibrium method are verified through several examples. The examples also indicate that the proposed methodology is able to reflect the influences of geometry，overburden depth and exposure height of blocks，is thus suitable for the stability evaluation of blocks during the excavation process under complicated conditions. The proposed method together with the existed block identification method form a complete system，including the identification，removability determination，kinematic pattern determination，stability evaluation and reinforcement method, which is parallel to the current methods for mechanical analysis of rock blocks (e.g. block theory).

The loading rate dependence of class I and class II rocks was investigated with the stress-feedback method. This method was executed on the servo-controlled testing machine with varied-resistance technology. The complete stress-strain curves were obtained with the stress-feedback method，and the testing machine was very stable at the post-failure region. The tests under the constant loading-rate，alternate loading-rate and loading-unloading- reloading were carried out with the stress-feedback method for class I rocks (Tage tuff，Ogino tuff) and class II rocks (Emochi andesite，Jingkou sandstone) and the elastic and inelastic strains were successfully separated from the loading-unloading-reloading test. The relationship between the inelastic strain and stress was obtained. Experimental results indicate the notable loading rate dependency of strength，failure strain，Young?s modulus and failure life span. Three methods were proposed to calculate the constant n of loading rate dependence at the pre-failure region.

In order to obtain the effective precursory information of instability destruction of rock，a multi- parameter comprehensive monitoring system for rock deformation and failure process was set up to investigate the abnormal signals of rock samples due to external loads，includeing microseismic，charge induction，self-potential and acoustic emission，etc. The monitoring system is used to synchronously monitor the deformation and fracture process under different loading rates. The results show that there are obvious precursory signals including of microseismic，charge induction，self-potential and acoustic emission before the sample instability. The signals of microseismic，charge induction，self-potential and acoustic emission are obviously different for different rocks，and the signals of granite porphyry are more abundant than the signals of marble. The acoustic emission monitoring data can better reflect microcracks of rock. The acoustic emission can be used for microcracks location，and reflect the precursor information of charge induction，self-potential better. The signals of charge induction and self-potential during the deformation and failure process occur much earlier than microseismic signals，and the interference is smaller. So the early signals of charge induction and self-potential can be used for early warning. It is also shown that the rock is about to enter the instability and failure phase when the various parameter signals of synchronization and larger amplitude appear frequently. The microseismic and acoustic emission can be used for microcracks location. The precursor information of rock samples? instability destruction can be obtained accurately based on the comparative analysis of signals of microseismic，charge induction，self-potential and acoustic emission.

The variations of the permeability and porosity of columnar jointed artificial rock under cyclic loading and unloading of the confining pressure were measured by using the inert gas permeability test device. In the first stage of confining pressure loading，the confining pressure has a great effect on the permeability and porosity，and the changes of the permeability and porosity are irreversible. In the subsequent stages of cyclic loading and unloading，the change of the confining pressure has a less effect on the permeability and porosity. The relationships between the permeability or porosity of columnar jointed artificial material and the confining pressure were fitted with the power functions during the confining pressure-loading stage. In the unloading stage of confining pressure，the relationship of the permeability and confining pressure can described as a power function，and the one of the porosity and confining pressure can described as an exponential function. The sensitivities of the permeability and porosity to confining pressure decreases with the confining pressure increasing，while the sensitivities of the permeability and porosity to confining pressure can be restored fully after confining pressure unloading. In the each stage of confining pressure loading and unloading，the relationship curves between the sensitivity of permeability to the change of confining pressure and column inclination exhibit a double peak pattern. The value of first peak is greater than the second one. The relationship between the porosity and permeability shows a power function in the loading stage of confining pressure，and a power-exponent function in the confining pressure unloading stage. The sensitivity of the permeability to the change of porosity increases with the porosity increasing. The relationship curves between the sensitivity of the permeability to the change of porosity and column angle exhibit a double peak pattern，and the first peak value is greater than the second one.

In this study，the alterations in the microstructures of three types of rocks (i.e. sandstone，granite and marble) were observed with the polarized light microscopy after thermal treatment at different temperatures. The variations in the physical-mechanical properties(including the longitudinal wave velocity，porosity，Young?s modulus，peak stress and the corresponding strain) of three types of rocks with the thermal treatment from room temperature to 800 ℃ were analyzed and compared and were correlated to the microstructural variations. The strong ductile behavior of the thermally cracked rocks was interpreted with a micromechanical model by taking into account the normal stiffness reduction of micro-cracks. A notable change in the carbonate cementation was observed in thermally cracked sandstone，in which the thermally induced cracking was not developed across the mineral particles. Both transgranular and intergranular cracking developed in thermally cracked granite with a maximum opening of 100 μm at 800 ℃，which was one order of magnitude larger than that at 400 ℃. The cracks in thermally treated marble were mainly intergranular，with the maximum width of about 20 μm at 600 ℃. Different from granite and marble，a dramatic decrease of elastic modulus occurred in thermally cracked sandstone when the treatment temperature was over 500 ℃，which was mainly ascribed to the phase transition of quartz. It was found that the physical-mechanical properties of the thermally cracked rocks depended much on the diagenetic processes，mineral compositions and cracking patterns. Moreover，the numerical simulation results are in good agreement with the experimental data，showing that the mechanical behaviors of the thermally cracked rocks are closely related to the density and stiffness of the thermally induced cracks.

The headgate of 11061 longwall face in Shoushan No.1 coal mine had the problems of low strength of thick mudstone compound roof, unstable bearing structure and the big deformation of roof. By monitoring on propagation of failure and separation of surrounding rock and counting failure model of the headgate，the failure mechanism of roadway with thick mudstone compound roof and complex stress distribution were gotten. Based on the compressive arch theory and in-site supporting and failure condition, an improved supporting method which is pre-stressed bolts and cables as the main bearing structure combined with grouting reinforcement in fissile zone was proposed. Pre-stressed cables and bolts built the primary and secondary bear structure. The separation of compound roof seam would be limited and the plastic range developing around the roof corner would be decreased. The detail parameters of cables and bolts were decided by the plastic ring theory and numerical method. The comparing results between the improved supporting method and former supporting method showed that: the deformation decreased 50% in the improved supporting method, the stability of the headgate?s surrounding rock enhanced obviously and the large deformation and failure in thick mudstone compound roof were well controlled.

To overcome the deficiency of the existing test devices and methods for rock burst modeling，a gas-liquid composite loading test device was developed to conduct the large size rock burst modeling under different stress gradients. In this paper, the occurrence，development processes and failure mechanism of the rock burst were studied through the analysis of the macroscopic failure features，interior deformation characteristics and the acoustic emission characteristic parameters of the sample. The results show that the occurrence of rock burst is closely related to the stress gradient，the energy concentration degree and the energy release rate during the rock burst processes. The differences of load gradient difference in the model test lead to the different concentration degree of internal stresses. These differences also influence the energy dissipation and agglomeration of the specimen in the early stage and the energy release rate at the moment of rock burst. If the difference of the vertical fracturing load gradient becomes greater，the degree of the local stress concentration becomes higher and the energy release rate will be faster. At the same time，the energy converted into the kinetic energy will increase and the probability of rock burst will be larger. When the specimen is broken，the ratio of the maximum vertical load  on the top to the uniaxial compressive strength R is  . Under the gradient loading condition，the static and dynamic composite loading is the main reason for the reduction of the failure load capacity of the specimen.

Statistical analysis of crack network in natural rock is the foundation for the studies of rock mechanical properties. The geometry of cracks is closely related to the stress state of rock. Based on the relativity of length and direction of the cracks network，a new parameter which takes into consideration of the length and direction of cracks was proposed and named as the geometry directional coefficient(GDC) . The coefficient can be used to calculate accurately the preferential angle of crack network and then to deduce the random ratio(RR)  describing the random degree of cracks networks. The crack network in rhyolite from the tunnel site at Niba Mountain was analyzed. The results indicate that(GDC) and (RR)   have clear physical meaning. Based on ，the accurate preferential angle of crack network(PA)   can be obtained. Meanwhile，κ can be used to evaluate the random degree of crack network quantitatively. The smaller the ? is，the greater its random degree is. Actually，there is quantitative relationship among(PA) ，(GDC) and (RR) . The results provide an effective method to statistically analyse the complex natural cracks network and the basis to the study of mechanical properties of naturally cracked rock.

The physical simulation about the coal-bed methane(CBM) exploitation under 3D stress state was conducted with a multi-field coupling test system developed in-house. The development of gas pressure under the different borehole lengths was analyzed. The results show that the gas pressure decreases rapidly near the center of the borehole under the different borehole lengths. The rate of pressure drop is high at the beginning and then decreases gradually. With the increase of the borehole length，the area with high pressure gradient increases significantly，which leads to the rise of gas desorption rate and productivity. On the cross sectional plane of borehole，the pressure contours are ring shaped with the center located at the borehole，indicating that the gas flows from the coal to the borehole. On the horizontal plane，the pressure contours present the funnel shaped distribution around the symmetry axis of the borehole. Although the length of the borehole has limited influence on the occurrence time of the danger elimination area of gas outburst，the time needed to prevent the gas outburst is greatly reduced with the increase of the borehole length. Because of the high stress and the low rate of gas desorption in the coal seam area of stress concentration，more time is needed for the gas production to eliminate the outburst danger than in the other areas.

The location accuracy of traditional methods based on the time difference is always influenced by the iterative method，the initial value and the pre-measured velocity. Solutions are not unique due to the square root calculations. In order to resolve the above two problems，the nonlinear location equations for the time difference of arrivals were simplified to linear equations and an analytical location method without the square root calculations was developed. A set of analytical solutions for AE / microseismic source location (ASS) were proposed for the unknown velocity systems where N = 6. Only the coordinates of the sensors and P-wave arrival times are required by the ASS method to accurately solve the source location parameters in real-time situations. The log-logistic distribution was used to determine the microseismic source locations. The blasting tests in the Dongguashan mine were used to verify the proposed analytical method for N≥6. Results show that the location accuracy of the proposed analytical solution is higher than the traditional method. The method highlights four outstanding advantages: no iterative solution; no pre-measured velocity; no initially evaluated source coordinates and no square root calculations.

In this research，the simulation of grouting for water rich and broken sandstones was performed. The ordinary Portland cement #42.5(PO.42.5)，the Sulphatealuminium cement #42.5(SAC.42.5) and the self-developed cement-based composite grouting material(CGM) were applied. Comparison of the uniaxial strengths of samples of broken limestone and sandstone after grouting shows that when the broken degree of rock is higher，the strength improving effect is more obvious，and that the larger the porosity，the better the strengthen improvement. The toughness enhancement of grouted stone is significant. The strengths of CGM strengthened stones are significantly higher than those of PO and SAC materials. The performance improvement of CGM on the porous sandstone is the best. The slippage probability on the surface of CGM and rock is the least. The effect of SAC grouts is better than PO grouts. Scanning electron microscopy was applied to analyze the mineral species and distribution patterns in the interface transition zone of rocks and grouts，which reveals the essence of different grouting effects. The bonding types of the interface between grouts and rocks are mainly determined by the types of rocks. The high-density limestone rocks belong to the Ollivier-Grandet model，and the porous sandstone rocks belong to the Zimbelinan model. The results of X-ray diffraction and scanning elemental analysis show that the grouting can not only improve the mechanical properties of rocks，but also change the chemical composition of rocks. The generated minerals by grouts in the interface zones can improve the bonding strength between the interface of rocks and grouts.

Spherical and cylindrical cavity expansion/contraction analyses are employed for estimating the pile-tunnel interaction. Firstly，a cylindrical cavity contraction model was developed to simulate the process of tunnel excavation based on the Mohr-Coulomb yield criterion，and the elasto-plastic horizontal displacement of adjacent single pile was calculated based on Pasternak foundation model. Then，an elasto-plastic method for evaluating the pile load capacity affected by tunneling was proposed. The total load capacity of pile was determined as the summation of the contribution from the ultimate end-bearing capacity and the ultimate shaft friction. A spherical cavity expansion model in the infinite medium was used to account for the ultimate cavity pressure at pile tip，and then the ultimate end-bearing capacity of the pile was obtained. The equivalent average shear stress along the pile length was estimated using the modified ?-method，and the ultimate pile shaft friction was obtained. The modified method considers the reduction of the shear stress of the pile shaft due to tunneling. On this basis，the elasto-plastic effect of tunneling on stress in surrounding soil was calculated. The distance effect of pile to tunnel on the pile load capacity was analyzed. Defining that the pile failure occurs when the pile capacity is reduced to 85%，the relationship between the distance of tunnel to pile tip and critical tunnel volume loss is analyzed. The parametric studies were performed to discuss the influence of various factors on this relationship，including the soil cohesion，internal friction angle，density，soil modulus and the pile diameter. Results show that，the elasto-plastic cylindrical cavity contraction model is capable in predicting the lateral displacement and bending moment of pile due to adjacent tunneling. A plastic zone is formed in a certain range after the tunneling. The effective soil stress reduction factor Rp in this area is less than 1，which indicates that the pile bearing capacity is reduced，while the pile capacity outside the plastic zone is not influenced by the tunneling. The distance of tunnel to pile has a significant effect on the load capacity of pile. If the tunnel-pile distance is a constant，the pile end-bearing capacity reduction factor Rqb will gradually approach 1 with the increase of the tunnel depth，which indicates that a deeper tunnel is more favorable for pile bearing capacity. There is a linear relationship between the critical tunnel volume loss at which pile failure occurs and the square of tunnel-pile tip distance. The load capacity of pile is sensitive to the soil modulus.

Pit excavation changes the ground stress state which affects the existing metro tunnels. An analytical method considering the shearing effect of tunnels was proposed to assess the longitudinal responses of tunnels induced by the adjacent excavation. The tunnel was assumed as a Timoshenko beam resting on the Winkler foundation. A two-stage approach was used to analyze the longitudinal deformation of the tunnel due to excavation. Firstly，equilibrium differential equations considering the effect of tunnel shearing were established based on the Timoshenko beam theory. Secondly，additional unloading pressure on the existing tunnel due to excavation was calculated using Mindlin′s elastic solution. Finally，the longitudinal deformation of the tunnel due to the unloading pressure was solved numerically using the finite difference method. The measured results from three well-documented published cases were selected to compare with the predicted results calculated with the proposed method as well as the Euler-Bernoulli beam method. Fairly good agreements are obtained between both two methods and the measurements. However，the Euler-Bernoulli method overestimates remarkably both the bending moment and shear force induced in the tunnel when compared to the results given by the proposed method. Because the proposed method is able to effectively model the shearing effect of the tunnel，the dislocation between the adjacent segmental linings subjected to the excavation can be further obtained by the proposed method.

Dynamic triaxial tests are performed for typical soft clays from Binhai New District in Tianjin to study the microstructure characteristics of structural clays under dynamic loads. Soil samples after the specific numbers of loading cycles are sliced and scanned with SEM. The research focuses on the effects of different cyclic stress ratios，the number of cycles and the cell pressures on the microstructure variation of soft clays. The test results show that the numbers of particles and pores increase with the increase of cyclic stress ratio，the number of cycles and the cell pressure，and on the contrary，the equivalent diameter and the shape ratio of particles and pores decrease. The microstructure tends to be stable with the increase of the number of cycles when the cyclic stress amplitude is less than a critical value. However，the microstructure changes constantly with the increase of the number of cycles when the cyclic stress amplitude is greater than or equal to the critical amplitude. Microstructure parameter values have obvious changes during the cyclic loading when the cell pressure is around the consolidation yield stress. Different microstructure parameter values change with the number of cycles logarithmically with good correlation.

Based on the Rayleigh-Love rod model，the vertical vibration of a large diameter pile embedded in inhomogeneous soil is investigated considering the transverse inertia effect. First，the reaction of the soil on the pile tip is simulated by virtual soil pile model，and the radial inhomogeneity of the surrounding soil caused by the construction disturbance effect is considered by using the vertical shear complex stiffness transfer model. Then，the pile and virtual soil pile are assumed to be Rayleigh-Love rod，and the vertical vibration equation is built. The analytical solution for dynamic response at the pile head is obtained in the frequency domain by virtue of Laplace transform and impedance function transfer method，and the semi-analytical solution in the time domain is derived by inverse Fourier transform. Based on the present solutions，the influence of the transverse inertia effect on the dynamic response at the pile head and its relationship to the pile-soil parameters are discussed. Finally，the calculated curves are compared with the measured results to demonstrate the rationality of the present solution.

A novel method for measuring the albedo of crushed-rock layer was proposed in this paper. The albedo of the crushed limestone layers with single-size aggregates of 1，2，3 and 5 cm sizes was measured respectively to reveal the main factors controlling the albedo of the crushed-rock layer. The layer was painted with the high-reflective pigment to increase the albedo of the crushed-rock layer and then the albedo of the painted layer was measured. The albedo of the crushed-rock layer was tested and compared against the albedo measured by the ASTM E1918A. It was found that when the variation of the incident solar intensity was less than 20W/m2 (tolerant errors stated by ASTM E1918A)，the ASTM E1918A either underestimate or overestimate the albedo of the crushed-rock layer up to 0.00–0.03. The reflectivity of the crushed-rock layer decreases with the aggregate size，and the albedo of the crushed-rock layer is 0.10–0.25 lower than that of the fresh flat crushed-rock. This difference is because photons leaving the crushed-rock layer have a probability to return to the layer，reducing the albedo of the crushed-rock layer. Painting the crushed-rock layer with high reflective pigment raised the albedo of the layer from 0.262 to 0.433. Raising the reflectivity of the crushed-rock layer on slope of embankment reduces the temperature of side slope and protects the permafrost stratum.