Study On Reliability Analysis Method And Engineering Application Of Deep Underground Research Laboratory Structure

As a low-carbon and efficient energy,nuclear power has attracted more and more attention,but how to safely dispose of a large amount of high-level radioactive waste(HLW)accumulated in its production process is the key to realize its sustainable development.At present,deep geological disposal is an internationally recognized safe disposal method of HLW.However,due to the lack of practical disposal experience,many countries have proposed to build an underground research laboratory(URL)in the deep,as a platform for research on geological disposal of HLW.URL is usually composed of shafts,tunnels,spiral ramps and other caverns.Compared with ordinary underground projects,URL has large buried depth,complex structure layout and higher safety level requirements.Therefore,it is necessary to carry out effective analysis and research on the safety and stability of its cavern group structure.Considering that the URL is built in natural stratum,the physical and mechanical parameters of surrounding rock have significant uncertainty,which is an important factor affecting the safety and stability of cavern structure.However,the traditional design method uses a general safety factor to summarize all uncertain factors,which can not really reflect the reliability of the structure.Therefore,it is necessary to adopt the reliability analysis method based on probability theory and mathematical statistics to effectively analyze and evaluate the safety and stability of URL structure.However,the existing reliability analysis methods still have some shortcomings when the performance function of deep tunnel structure is strongly nonlinear implicit.In this regard,taking China’s first URL for deep buried geological disposal of HLW in Beishan,Gansu Province as the research background project,this paper established the reliability analysis method of the cavern structure of the deep URL.Obtained the spatial distribution law of the reliability index of the cavern structure after excavation of the URL,and puts forward the research conclusion with engineering guiding significance.The specific research results are as follows:(1)The probability distribution model of mechanical parameters of deep surrounding rock in Beishan URL is established by using the maximum likelihood method and K-S test.Among them,the elastic modulus obeys the extreme value I-type distribution,the Poisson’s ratio obeys the normal distribution,the cohesion and internal friction angle obey the lognormal distribution,and there is a strong negative correlation between cohesion and internal friction angle.(2)An explicit analysis method of implicit performance function of deep cavern structure based on Support Vector Regression(SVR)is proposed.Firstly,the implicit performance function of surrounding rock based on Morh-Coulomb criterion and ultimate tensile strain criterion is constructed.Then,through Latin hypercube experimental design and FLAC3D numerical simulation,high-quality sample data of implicit performance function of deep cavern are obtained.Finally,according to the sample data of implicit performance function,the explicit model of implicit performance function is established by SVR.In this process,in order to ensure the modeling accuracy and efficiency of support vector regression,a Particle Swarm Optimization algorithm with improved dynamic inertia weight is proposed to optimize the parameters in SVR.Finally,the rapid analysis of the explicit performance function of each part of the deep cavern structure is realized.(3)A calculation method of reliability index of deep cavern structure based on adaptive importance sampling is proposed.Firstly,according to the probability distribution model of surrounding rock mechanical parameters,the explicit performance function constructed by SVR is transformed into independent standard normal space;Then,the distribution samples in the failure domain of the performance function are obtained by Markov chain simulation method.In this process,an ndimensional annular uniform distribution function is proposed as the proposed distribution,which effectively improves the adaptability of Markov chain simulation when there are discontinuous intervals in the failure domain of performance function.Finally,according to the distributed samples in the failure domain of the performance function,the importance sampling density function close to the theoretical optimal model is established through Kernel Density Estimation,so that the rapid calculation of the reliability index of deep cavern structure is realized.(4)According to the established reliability analysis method,relying on FLAC3D and MATLAB platform,the reliability calculation program of deep cavern structure is developed and applied in engineering,and the spatial distribution law of excavation reliability index of Beishan URL is obtained.The calculation results show that:1)After the excavation of the URL,the overall structure of the tunnel group is safe and stable.Through the comparison with the 3D geomechanical model test results,the reliability of the reliability calculation results is effectively verified.2)At the intersection of URL caverns,especially the intersection of large section caverns and small angle intersection caverns,the reliability index is reduced compared with the non-intersection part.It is suggested that the intersection of caverns should avoid acute angles,adopt the design of large angle intersection,and carry out systematic shotcrete anchor mesh support for the intersection of caverns.The research results provide scientific guidance for the optimization of the overall construction scheme of URL.