Study On Back Analysis Method In Underground Engineering Based On Parallel Computing And Optimization Algorithm

With the building of more and more large-scale underground engineering, stability assessment of surrounding rock becomes a hot topic in geotecnical engineering. As the complexity and uncertainty of rock mass media, it’s very difficult to obtain accurate information such as geological conditions and occurrence of rock mass before constructing of underground engineering. So there are some problems to analyze stability of surrounding rock utilizing empirical analogue methods, numerical analysis mehod or other traditional analysis methods. As a result, feedback analysis of insitu measured information becomes necessary and has been broadly applied in underground engineering. More reliable engineering information can be obtained by insitu monitoring and geological data in excavation, by which we can analyze stability of surrounding rock with real-time feedback. Rock mass parameters are back-analyzed to predict the following excavation, so as to achieve goals of dynamic feedback and information-based construction.In this thesis, several back analysis methods are studied based on summarizing the results of previous studies. The contents spread out around issues of feedback analysis of insitu measured information, failure characteristics and computational method of layered rock, displacement back analysis of parameters field in underground engineering based on excavation damaged zone (EDZ) of surrounding rock, numerical simulation of drainage hole, back analysis of seepage field, parallel optimization algorithm and so on. The proposed methods are further illustrated with their application to some underground chambers. The thesis mainly contains following contributions:(1) The features and back analysis methods on the stability of surrounding rock in underground engineering are reviewed in detail. Several key practical problems are presented in back analysis and dynamic optimization design.(2) Particle Swarm Optimization (PSO) algorithm is used in parameter optimization and back analysis and its convergence is modified to obtain good global search capability. As back analysis of geotechnical engineering involves enormous amounts of computation, the back analysis process is improved by introducing message passing interface (MPI)-based master-slave parallel framework. The parallel computation can be conducted using computer cluster networks, thus considerably reducing the cost and enhancing the efficiency of computation. The modified parallel PSO is used to back-analyze parameters in several projects. The results are favorable and prove that the proposed method is effective and applicable. And the parallel computation has considerably reduced the cost and enhancing the efficiency of computation.(3) The implicit composite element method of drainage hole in 3D seepage field in underground engineering is presented. Drainage holes are implicit in model and their data are gained according to pre-process. This method can largely reduce the computational workload and cost. Its feasibility and reliability have been verified through a numerical example and some practical projects. The calculation result shows that the proposed method is feasible to simulate drainage holes. To determine accurate permeability parameters, back analysis of seepage field is conducted by observing water level and flow.(4) The real-time feedback analysis and dynamic optimization design are used to back-analyze the process of excavation and support in large underground caverns. The insitu monitoring data is deeply analyzed according to its pre-processing and tracking analysis. Cluster Computing is adopted to back-analyze parameters and evaluate stability of surrounding rock. Synthetical prediction methods with back-analysis and normal computation and grey system theory are used to forecast the following excavation. According to the results of feedback analysis of insitu measured information, back analysis and synthetical prediction, timely and effective change programs and preventive measures are worked out to ensure that underground caverns construction and operation of safe and stable operation.(5) The failure characteristics and the calculation method of layered rock are expounded according to its physical and mechanical characteristics. Using 3D nonlinear anisotropic elastic-plastic finite element method, the influence of rock angle on stability of underground houses is analyzed in a hydroelectric power station. After comparison with engineering experiences, the reasonable layout of underground houses in layered rock is proposed. As the anisotropy of layered rock, MPI-based parallel Particle Swarm Optimization algorithm is adopted to back analyze parameters of layered rock.(6) Due to excavation blasting, the surrounding rock of underground caverns is loosened and its strength lowers. A continuous three-dimensional data field of surrounding rock parameters is employed to describe the different damage degrees of surrounding rock. According to the formation mechanism of excavation damaged zone (EDZ) of surrounding rock, the method to calculate the EDZ induced by excavation blasting in underground caverns is analyzed. By taking the range of EDZ into consideration, a back analysis method based on incremental displacements is put forward. By employing this method, the fields of surrounding rock parameters can be back analyzed. The back analysis results are favorable and prove the reliability and rationality of the proposed method. Based on the surrounding rock parameters field obtained by back analysis, the current anchor support scheme and the stability of the underground cavern are assessed. Also, the stability condition in subsequent excavation is predicted, and the rational recommendations are made for the engineering design and construction, providing the parameters back analysis of underground engineering with a new method.Finally, some researches to be explored in future are put forward after summarizing research achievements of this thesis.The research achievements in this thesis work from practical engineering problems. Some back analysis methods are systematically studied in undergrounding engineering. In order to solve practical engineering problems, several simulation methods has been improved in many aspects, such as optimization algorithm, master-slave parallel back analysis method, numerical simulation of drainage hole, failure characteristics and computational method of layered rock, displacement back analysis of parameters field in underground engineering based on excavation damaged zone of surrounding rock and so on. These research achievements provide some technical references for design and construction of underground engineering.