A carbonate reservoir which has extremely low permeability hold more than 60% of the world’s remaining oil and 40% of gas. Since oil or gas mainly pass through fracture or fault zone in the carbonate reservoir, an accurate and effective tool for the...
A carbonate reservoir which has extremely low permeability hold more than 60% of the world’s remaining oil and 40% of gas. Since oil or gas mainly pass through fracture or fault zone in the carbonate reservoir, an accurate and effective tool for the simulation of the seismic responses from the fracture or fault zone is indispensible. However, it is highly inefficient to deal with an individual fracture in the numerical simulation because of its thin width. Thus Coates and Schoenberg (1995) suggested the equivalent medium substituting the individual fracture.
In this research, I have modified the 3D elastic forward modeling algorithm developed by Han et al. (2011) in order to successfully handle the fractures represented by the anisotropic equivalent medium using variable grid strategy. In spite of substituting individual fractures to the equivalent anisotropic medium, the medium is thinner than the ordinary anomalous targets. The computation cost for the simulation of this very thin object using the equi-spaced grid is quite expensive. Thus I compose a variable grid model which has the minimum grid size for the fracture zone, gradually increasing grid size for the transition zone and the maximum grid size for the carbonate reservoir. Since the increasing rate of the grid size in the transition zone determines the stability of this forward simulation, I have tested several decline functions in order to choose proper one. Through numerical experiments, I confirmed a gridding method, which uses steep slope for the coarse grid zone and gentle slope for the fine grid zone, shows the minimum error.
The wave propagation in the complex fractured media which is one of most wayward models can be successfully simulated by this efficient 3D seismic modeling algorithm using the variable grid approach. In addition, this method efficiently suppress artificial boundary’s reflections by implementing convolutional perfectly matched layer (CPML) and dramatically reduces the computational time by incorporating MPI (Message Passing Interface) technique in z axis.