Numerical Test System For The Equivalent Hydraulic And Mechanical Properties Of Fractured Rock Masses

The hydraulic and mechanical properties of fractured rock masses are significant for the safety of structures, but commonly they are very complicated and difficult to be obtained. The existence of fractures increases and heterogenizes the compliance and permeability of rocks significantly. Moreover, the seepage/stress coupling behavior in rock masses leads this problem further.The implicit approach of numerical methods is popular in rock engineering applications. This approach makes the function of fractures integrated into constitutive relation and treats a rock mass as a continuum. In this way, the modeling of fractures can be greatly simplified and the large-scale analyses for quantities of fractures can be enabled. It is worthwhile to point out that the key issue to use the implicit approach is how to efficiently obtain the equivalent hydraulic and mechanical properties of fractured rock masses.In order to identify the equivalent properties, a "Numerical Test System (NTS)" is established. In the NTS, the seepage and stress/strain fields in rock masses are simulated using numerical methods firstly, and then the equivalent properties are inversely calculated from simulated results. According to the investigation for fractures, the NTS falls into two categories:(1) If only dominating fracture sets are measured, the permeability tensor is obtained by a revised analytical solution, whose unknown parameters are back analyzed from packer tests using Artificial Neural Network (ANN); (2) If all fractures are measured clearly, the permeability tensor and the elastic compliance tensor are calculated from numerical sample tests, using Composite Element Method (CEM).The NTS involves several advanced mathematical and mechanical tools:Finite Element Method (FEM) and Composite Element Method (CEM) are used for seepage/stress simulations; Artificial Neural Network (ANN) is used for back analyzing and database modeling; Monte Carlo Method (MCM) is to generate stochastic Discrete Fracture Networks (DFNs). These algorithms are verified by analytical solutions and used in complicated applications.In addition, a systematic framework to analyze fractured rock masses is established, which combines both of implicit and explicit approaches. That enables the fractures, even having greatly different dimensions in one project, can still be successfully simulated. The equivalent properties needed by the implicit approach are provided by the NTS. This framework is successfully used to deal with a project of dam foundation.