Methods For Characterizing Geotechnical Spatial Variability Based On Cone Penetration Test And Surface Wave Exploration

Geotechnical site investigation is an important part of geotechnical analysis and design.However,the information obtained from geotechnical site investigation is usually very limited,and geotechnical materials are spatially variable,so the geotechnical parameters and soil layers determined based on the limited information are inevitably uncertain.In recent years,probabilistic site investigation methods of geotechnical engineering based on Bayesian Theory have been developed rapidly,which provides an effective tool for reasonably quantifying uncertainties in geotechnical parameters and soil layers based on limited site investigation information.In spite of this,there are still a few key problems in the applications of probabilistic site investigation methods,including:(1)Fast interpretation of in-situ test data(such as cone penetration test(CPT)data)considering the uncertainty of geotechnical materials is needed to improve the timeliness and rationality of the decision-making at site;(2)With the limited site investigation information,there is no probability model and identification method that can effectively describe the complex spatial characteristics of geotechnical materials(such as the non-stationary spatial variability of geotechnical parameters and two-dimensional soil stratification).To address above problems,this thesis develops probabilistic geotechnical investigation methods based on CPT and surface wave exploration,including a fast Bayesian approach for CPT-based soil stratification,a two-dimensional space model based on linear dynamic system(2DSM-LDS)and its parameter identification method,methods for characterizing spatial variability of soil parameters and identifying soil stratification based on surface wave exploration.The main work and conclusions are as follow:(1)Research background and significance of spatial variability characterization and soil layer identification based on site investigation data(including in-situ tests,surface wave exploration,etc.)were explained.The existing in-situ tests,geophysical exploration methods,and probabilistic site investigation methods have been summarized.The shortcomings of existing research methods of spatial variability characterization of geotechnical parameters and soil layer identification have been pointed out.Scientific problems to be solved and the technical route have been proposed.(2)This chapter introduced the probabilistic site investigation methods,and proposed a fast Bayesian approach for CPT-based soil stratification.The proposed method has simplified the calculation formula of the likelihood function of CPT data and developed a fast calculation strategy to solve the Bayesian equation.The validity and effectiveness of the proposed method were verified by 27 sets of CPT data.Results showed that the proposed method has improved,significantly,the computational efficiency of soil stratification and its uncertain characterization.The influence of soil layer thickness and model parameters of adjacent soil layers on the soil boundary identification uncertainty was explored.It is found that the soil layer thickness and the mean value of soil parameters are the most important factors.The uncertainty of the soil boundary decreases with the increase of the thickness of adjacent soil layers.The greater the statistical difference of adjacent soil parameter is(e.g.,the greater the difference in mean values is),the smaller the boundary uncertainty is identified.(3)A probabilistic model for characterizing spatial variability based on 2DSM-LDS has been established and its parameter identification method have been proposed.Based on the forward recursive algorithm,the likelihood function given 2DSM-LDS and the Bayesian equation has been derived to identify the 2DSM-LDS parameters.Given the optimal model parameters,the solution of the marginal posterior distribution of hidden variables based on backward recursive algorithm was presented.11 sets of simulation data have been used to verify the validity and effectiveness of the proposed model and its parameter identification method.The proposed model and parameter identification method have addressed the limitation of the linear dynamic system model in processing two-dimensional spatial data,and provided a new way for the two-dimensional spatial data processing and parameters inversion analysis,and expand the application scope of the linear dynamic system model.(4)A two-dimensional spatial variability characterization method for geotechnical parameters based on surface wave exploration was proposed.Based on the surface wave exploration data,multichannel analysis of surface waves(MASW)were conducted to obtain the two-dimensional profile of shear wave velocity.Given the two-dimensional data of shear wave velocity,the two-dimensional profile of soil parameters was inversed based on 2DSM-LDS,and the two-dimensional spatial variability of soil parameters also was quantified.Surface wave exploration were carried out on Majiagou slope in the Three Gorges Reservoir area,and the natural density of Majiagou slope has been inverted based on the two-dimensional shear wave velocity data.Results showed that,based on 2DSM-LDS,the spatial correlation in horizontal and vertical directions of soil parameters are reasonably reflected,and the horizontal correlation was significantly stronger than the vertical correlation.(5)A probabilistic inversion analysis method for two-dimensional soil layer profile based on surface wave exploration data was proposed.The two-dimensional CPT profile was inversed based on 2DSM-LDS and the transformation model between shear wave velocity and CPT data.The soil behaviour type index(I_c)was calculated based on the inverted CPT data to obtain the two-dimensional spatial distribution of soil types and to determine the two-dimensional soil layer profile.The two-dimensional soil layer profiles are simulated based on 2DSM-LDS.The random realization of soil layer profile not only contains the site investigation information but also is constrained by the 2DSM-LDS.Information entropy was used to quantify the uncertainty in two-dimensional profile of soil types based on the random simulation results.Based on the shear wave velocity data of Majiagou slope,the two-dimensional soil layer profile of the site was inverted,and the uncertainty of soil classification was quantified.Compared with the two-dimensional soil layer profile estimated by interpolation based on in-situ testing data,the uncertainty of the soil profile from two-dimensional surface wave data was reduced.The proposed method provides an effective way to improve the accuracy of geotechnical site investigation.(6)The stability of Majiagou slope was analyzed based on the spatial variability characterization results of geotechnical parameters and soil layer profile results identified from surface wave exploration data.The transformational model between shear wave velocity and effective friction angle was established,and the two-dimensional profile of effective friction angle of the Majiagou slope was inverted based on the 2DSM-LDS.Based on the optimal parameter of 2DSM-LDS,effective friction angle two-dimensional profiles were simulated.The stability of Majiagou slope was analyzed.The statistics and probability distribution of safety factors were calculated and it was found that the overall failure of the slope is less likely to occur.The local failure of the slope can be avoided through engineering reinforcement measures.