| With the rapid development of hydropower engineering in China,more and more hydropower projects are located in mountain and ravine region with complex geological conditions,and the construction contexts are more complicated.A large number of fractures and joints exist in the rock mass of dam foundation.These geological structures have various shapes and fully random distributions and are often difficult to measure.These characteristics will directly affect the grouting,seepage and stability analysis of dam foundation,and then affect the safety of the project.Building accurate 3-D fracture network model in the rock mass of dam foundations by using probability and statistics theory is one of the key problems to be solved urgently.In order to improve the accuracy of the fracture network model,the rationality of fracture dominant partitioning,the accuracy of fracture geometric parameter estimation and the accuracy of fracture network model were discussed and researched based on optimization algorithm theories in this paper.The achievements are listed as follows:1.Considering the problem that the fracture parameter for dominant partitioning is single and the parameter quantification is not reasonable,the multivariate parameter dominant partitioning method of fractures based on PSOBFO hybrid algorithm is proposed:Current fracture set partitioning methods have the limitations of a single type of parameter,unreasonable quantification of fracture characteristics and imperfect optimization results,and these limitations will determine the rationality of the fracture dominant partitioning and further influence the accuracy of fracture network simulation:1.1 A multivariate parameter dominant partitioning mathematical model of rock mass fractures is proposed,which builds the theory foundation of multi-parameter fracture set partitioning and optimization.1.2 The concept of fractal dimension is introduced to quantify the fracture roughness more objectively and accurately,and a comprehensive analysis of multivariate parameters including occurrence,trace length and roughness is realized.1.3 The PSO-BFO hybrid optimization algorithm is applied to the multi-parameter fracture clustering,and the validation is performed with computer simulated data sets.1.4 A case study of the multi-parameter dominant partitioning of a dam foundation fractures in the southwest China is carried out,and a reasonable result is acquired.And this can provide a reliable data support for the further fracture simulation,grouting,seepage and stability analysis.2.Considering the problem that the fracture size and volume density estimation methods are numerous with the lack of research on the applicability and accuracy of different methods,the error contrast analysis method of different estimation methods are studied:There are various methods to realize the parameter estimation of the fracture geometries,including the trace length,occurrence,location and density,but the applicability and accuracy of different methods need to be further researched and clarified.Based on a comparative analysis of the accuracy of different fracture size and volume density estimation methods,the conclusions are draw as follows:2.1 A comparative analysis of the fracture volume density estimation error of the four methods proposed by Masanobu Oda,Kulatilake,Mauldon and Song is made in this paper,and the conclusion shows that the end-point estimator proposed by Mauldon has higher accuracy in fracture volume density estimation.2.2 A comparative analysis of the fracture size distribution estimation error and applicability of the three methods proposed by Einstein,Song and Tonon is made in this paper,and the conclusion shows that the sample histogram method proposed by Song has higher accuracy and better applicability in fracture size distribution estimation.2.3 The fracture volume density estimation method proposed by Mauldon and the fracture size distribution estimation method proposed by Song is adopted in the fracture parameter estimation process,which are benefit for the guarantee of the fracture network model accuracy and provide a reliable initial model for the dynamic simulation scheme.3.Considering the influence of the estimation accuracy of fracture geometry parameters on the model accuracy,a 3-D fracture network dynamic modeling method in rock mass of dam foundation is proposed:Current studies are faced with the problems that the model precision is limited by the parameter estimation accuracy but the quantification of model errors is not reasonable enough.Meanwhile,there lacks the model examining and error reducing methods.A 3D fracture network dynamic modeling scheme is rock mass of dam foundation is proposed this paper is with the aim of reducing the model deviation from the measured data and improving the simulation accuracy:3.1 The research framework and mathematical model of 3-D fracture network dynamic simulation is proposed,and the theoretical foundation and technical support are established for optimization-based accurate 3-D fracture network model.3.2 The error index,as precise expression of the model error,representing the deviation between fracture network model and measured data is established with integrated use of fractal dimension and Relative Absolute Error(RAE).3.3 A 3-D fracture network dynamic simulation method is proposed,in which the minimum of the aforementioned error index is used as the objective function and the fracture geometric parameters are used as the decision variables.The downhill simplex method is applied to the dynamical adjustment of the initial values of the fracture density and size parameters to calibrate the model error.Finally,a 3-D fracture network model could be obtained which meets the requirements in statistical sense.3.4 The proposed method is validated in a 3-D fractures simulation in a hydropower project in China,and the results showed that the model has high fitting degree with the measured data and the proposed method can provide a reliable model foundation for rock mass mechanics and hydraulics analysis of hydropower engineering. |
PDF Full Text Request