We present a 2.5-D finite difference (FD) algorithm to simulate the logging-while-drilling (LWD) electromagnetic (EM) measurements with arbitrarily oriented magnetic dipole sources in 2-D anisotropic media. The real 3-D modeling in the spatial domain is converted to a series of independent 2-D equations in the spectral domain by implementing the Fourier transform (FT). By replacing the dipoles with incident fields, the scattered fields are solved directly to avoid the point source singularity as well as improve the convergence of inverse FT (IFT). The efficiency of the IFT is further enhanced with a modified Gauss-Hermite (GHM) quadrature-based numerical integration scheme. Numerical tests demonstrate that the proposed 2.5-D algorithm is fast and accurate in modeling the LWD EM measurements in complex scenarios. Discussions are also made to compare the performance of the proposed methods with 1-D and 3-D algorithms.
Imaging hydraulic fractures is of paramount importance to subsurface resource extraction, geologic storage, and hazardous waste disposal. The use of electrically conductive proppants and current energized steel casing provides a promising approach to monitor the distribution of fractures. In this article, a borehole-to-surface system is employed to energize the steel casing and measure electric and magnetic fields on the ground. A convolutional neural network (CNN) is then trained to learn the relationship between the measured field pattern and the parameterized fracture, namely, the lateral extent and direction. To accelerate the generation of training data with limited accuracy loss, an approximate hollow casing is modeled by the impedance transition boundary condition with a tenfold magnified radius and reduced conductivity. Two training strategies are then presented with a grid search of the network's hyperparameters. The well-trained CNN shows good generalization to unseen fracture conductivity, the true casing model, as well as white Gaussian noise. Finally, we apply the CNN to image irregular fractures and obtain reliable results even under strong noise, indicating a promising imaging technique for more complicated fractures.
4 October 2018 in Petroleum Exploration and Development
Azimuthal electromagnetic logging while drilling (LWD), which is capable of providing accurate position information approaching bed boundary, has been widely applied in real-time geosteering. For the inversion speed and precision of azimuthal electromagnetic LWD data, the key lies in the selection of proper inversion model and corresponding optimization algorithm. In this study, we first simplified the complex three-dimensional (3D) inversion of data into a series of one-dimensional (1D) inversion problems by using the dimensionality reduction scheme. Then, the feasibility and inversion performance of various 1D inversion models and different optimization methods were investigated, and the best combination between the inversion model and inversion algorithm was also obtained. Numerical simulation results show that the selection of 1D inversion model is dominated by the thickness of targeted beds, whereas the determination of inversion algorithm relies on the total layers amount of the inversion model. For the formation with thickness larger than 4.0 m, the single boundary inversion model and gradient optimization method are recommended. When the bed thickness is between 1.0 m and 4.0 m, the two-boundary inversion model instead of the single-boundary one is needed to estimate upper and lower boundaries around the borehole. For the inversion of azimuthal electromagnetic LWD data of thin layers, the multiple-boundary inversion model and the Bayesian algorithm should be employed.
This letter extends the discontinuous Galerkin frequency domain (DGFD)-based domain decomposition method (DDM) to model the open borehole environment accurately with a newly developed hierarchical sudoku mesh. Both the borehole and mud invasion effects can be modeled by this framework efficiently. Furthermore, deep reading measurements can be modeled conveniently by inserting arbitrarily shaped objects into the hierarchical sudoku mesh; the capability to distinguish a cavity saturated with either oil or water with different borehole-object distances are then studied for a deep reading tool in an open borehole carbonate environment. The DGFD with the hierarchical sudoku mesh shows six times faster speed than the traditional finite-element method (FEM) for the deep reading case.