Static And Seismic Reliability Analysis Of Geosynthetic Reinforced Soil Slopes

Geosynthetic reinforced soil slopes(GRSS)are consisted of layered geosynthetic materials which help to share the tension force soils cannot sustain,and thus improve the stability of slopes,especially the resistance to degradation under the cyclic force induced by strong earthquakes.GRSS has been widely employed in the construction of transportation,hydraulic projects and architecture.However,the inherent uncertainty of soils,reinforcement and earthquake generally leads to failure of some of GRSS though satisfying the requirement of corresponding specifications,and causes a threat to people’s life and property safety.Compared to the sufficient deterministic analysis of GRSS,the relevant reliability analysis is limited,and mostly focuses on the uncertainty of slope parameters such as soil and reinforcement but neglects the uncertainty of seismic time history parameters.The available system reliability analysis of GRSS is also performed by assuming a certain correlation of failure modes,while less attention is paid on the influence of local failure of reinforcement and the correlation of failure modes on the system reliability.Additionally,the study of analyzing multi-source monitoring information for reducing the uncertainties of GRSS is still lacking.In this thesis,uncertainties of slope soil,reinforcement and seismic time history parameters are fully considered.The local failure of reinforcement and correlation of failure modes are both incorporated in the system reliability analysis.The reliability analysis of GRSS under stochastic seismic action,and the back analysis as well as the alert level based on integrating multi-source information are performed,respectively.The main research content and results are summarized as follows.(1)The concept of redundancy of GRSS is proposed and its allowable redundancy is also defined through analyzing the influence of reinforcement failure on slope stability.After quantitatively assessing the contribution of reinforcement at different locations to the overall stability of GRSS,the progressive failure pattern can be summarized.The influence of soil properties and the number of layers of failed reinforcement on the redundancy of GRSS is evaluated.The results show that the reinforcement with low redundancy fails first,causing the potential sliding surface to move towards the location of the failed reinforcement and another significant increase in the tensile force of adjacent reinforcement,which gradually exceeds the ultimate tensile strength and finally triggering the progressive failure of GRSS.(2)A reliability analysis method for GRSS considering local failure of reinforcement is proposed based on the total probability formula,after decomposing the overall failure of GRSS into mutually exclusive events of stepwise-layer failure of reinforcements.The effect of local failure of reinforcement on the reliability is analyzed.It is shown that the proposed method provides an accurate solution for the system reliability due to avoiding the assumption of certain correlations of failure modes.Through analyzing the development of tension force of remaining reinforcements and the relevant failure modes of slopes since the failure of prescribed reinforcements,the most likely progressive failure mode and its failure probability are obtained.(3)The reliability analysis method of GRSS considering multiple failure modes as well as their correlations under seismic loading is proposed on the basis of introducing common source random variables and assuming different failure modes as a series system.The effects of soil parameter variability,horizontal and vertical seismic coefficients,reinforcement length and strength,and spacing of reinforcement on the system reliability are analyzed subsequently.It is found that the proposed method can calculate the correlation of failure modes automatically.A strong correlation between different failure modes is observed given the same random parameters,leading the system failure probability to be closer to the maximum failure probability of a single failure mode.The failure mode of GRSS is highly related to the reinforcement strength and length.As far as the reinforcement strength is lower than the critical reinforcement strength and the reinforcement length is greater than the critical reinforcement length at the same time,the internal stability failure mode of GRSS happens,otherwise the composite stability failure and even the external stability failure will occur.(4)The reliability analysis method of GRSS under random earthquakes is proposed through the probability density evolution method where random earthquake samples are generated from the generalized F deviation method and the spectral representation random function method.The effects of slope angle,reinforcement spacing,reinforcement length and stiffness on the failure probability are also analyzed.It is found that the proposed method provides an efficient and accurate prediction of the reliability of GRSS as far as the randomness of the seismic spectrum is concerned.The failure probability of GRSS is highly dependent on the selected earthquake samples,indicating that ignoring the randomness of seismic spectrum will result in large errors in predicting the failure probability.The PDF curve fluctuates intensively with time,which denotes propagation of uncertainty during the steady phase of acceleration time history.As the reinforcement length increases,the failure mode of GRSS experiences the external-to-composite-to-internal transformation.(5)The dynamic back analysis method of GRSS is proposed through Bayesian network integrating physical and mechanical principles and multi-source monitoring information.The influence of single-source monitoring information on the dynamic back analysis results is compared to multi-source monitoring information.The alert level of monitoring information is therefore provided by establishing the correspondence of monitoring information and safety factor with failure probability.The influence of expert experience on the dynamic back analysis results was quantitatively evaluated by quantifying the expert experience through Bayesian network.It is found that the integrated multi-source monitoring information can balance the influence of different monitoring information on the stability of GRSS and improve the accuracy of dynamic assessment results.