Reliability Analysis Of Foundation Settlement Of Large Thickness Fill Site Based On Random Finite Element

Under the background of the "The Belt and Road" strategy,infrastructure construction in China’s western region has ushered in a new round of development upsurge.Coexisting with infrastructure construction is the increasingly prominent problem of shortage of construction land,and urban development is increasingly limited by the topographic and geomorphological conditions of loess ravines.In order to meet the needs of urban development,it is imperative to open mountains and fill in mountains and artificial land reclamation,but due to the special topography and engineering nature of the loess area,the excessive settlement of the high fill loess foundation has caused a large number of engineering failures or damage.The superposition of two soil masses with different geological genesis and stress history determines the inherent uncertainty of the filled foundation.It is of great significance to combine the randomness of soil parameters with spatial location,and consider the strata variability between the original soil layer and the fill layer,and carry out the study of foundation settlement characteristics.The research results can provide a theoretical basis for the foundation design calculation and specification revision of high-fill loess sites.The main research contents and results of this paper are as follows:(1)Based on the field and indoor test data of large-thickness fill site in Lanzhou New District,the random field model of soil parameters and geostatistical theory were used to calculate the parameters of the random field model of the fill,namely the fluctuation range and the spatial coefficient of variation.The spatial variability of the parameters was evaluated with the change of soil depth.The results show that when the semivariogram method is used to calculate the fluctuation range of soil parameters,the spherical model can be used as the optimal fitting function.The compressive modulus of the fill body was moderately variable,and the natural density,cohesion and internal friction angle were weak variation.(2)According to Wlather’s phase order law,the relationship between the vertical counting matrix and the horizontal counting matrix is established,that is,the main diagonal element of the horizontal counting matrix is K times the main diagonal element of the vertical counting matrix,and the remaining elements are equal,and then the K value is determined by trial calculation to obtain the horizontal counting matrix.Based on the drilling data,the coupled Markov chain model is used to calculate the state transition probability matrix,and the influence of different drilling schemes on the transfer probability matrix estimation is analyzed.When estimating the state transition probability matrix,try to select more drilling data with uniform distribution,which is conducive to the accuracy of the estimation results.When the diagonal element values in the matrix are large,the transfer probability matrix estimation is less sensitive to the drilling arrangement scheme.(3)The random field of discrete fill parameters by covariance matrix decomposition method was used to simulate the spatial variability of fill parameters.The coupled Markov chain Monte Carlo simulation method is used to obtain the random distribution and transfer state of the original soil layer and the fill layer in the area,and realize the simulation of stratigraphic variability.At the same time,considering the spatial variability of strata and soil parameters,the distribution of spatial variation soil is mapped to the foundation model,and the settlement at different depths of the foundation is calculated by combining random finite elements,revealing the change law of the reliable index of settlement of deep fill foundation,and proposing a reliability analysis method for settlement prediction.The results show that the effects of soil parameter variability and stratum variability on basal settlement are larger and uneven settlement,respectively,so it is necessary to consider formation variability and parameter variability when calculating settlement.Formation variability and parameter variability have an impact on the reliability index of foundation settlement.When considering only stratigraphic variation,the higher the number of boreholes used for soil state prediction,the more reliable it is.After adding a random field,the reliability decreases,and the greater the spatial variability of the parameters,the lower the reliability.(4)Taking the settlement determination solution without considering the uncertainty of the compressive modulus as the starting point,the functional relationship between random settlement and random field local average is established,and the basic settlement expression that can reflect the spatial variation characteristics of soil parameters is proposed.The direct probability analysis of foundation settlement is carried out by the second-order moment method,and the confidence interval of settlement under a specific probability is predicted,and the reliability index is calculated.Comparing the reliability index obtained by stochastic finite element method and direct probability analysis,it can be seen that the random finite element method is better than the direct probability analysis method in reliability analysis.