Xuan Wang

We present a three-dimensional (3D) general-measure inversion scheme of crosswell electromagnetic (EM) data in the frequency domain with a direct forward solver. In the forward problem, we discretised the EM Helmholtz equation by the staggered-grid finite difference (SGFD) scheme and solved it using the Intel MKL PARDISO direct solver. By applying a direct solver, we simultaneously solved the multisource forward problems at a given frequency. In the inversion, we integrated a general measure of data misfit and model constraints with linearised least-squares inversion. We reconstructed a model with blocky features by selecting the appropriate measure parameters and model constraints. We used the adjoint equation method to explicitly calculate the Jacobian matrix, which facilitated the determination of an appropriate initial value for the regularisation coefficient in the objective function. We illustrated the inversion scheme using synthetic crosswell EM data with a general measure, the L₂ norm, and, specifically, two mixed norms.

The crosshole electromagnetic (EM) method is used in reservoir development monitoring and evaluation. It reflects the resistivity distribution between wells in the form of EM response changes and describes the spatial distribution of pore fluids in subsurface reservoirs based on the correlation between resistivity and pore fluids. This technique has been used to describe reservoirs, monitor steam and brine flooding to improve oil recovery, optimize the design of infilled wells, and verify the effectiveness of reservoir numerical simulations. However, due to the complexity of the electrical distribution of underground reservoirs and the limitations of the crosswell EM response range, there is still a lack of practical and effective crosswell EM imaging methods. This article presents a three-dimensional EM inversion scheme for interpreting crosswell data in the frequency domain with general measures. For the forward modeling, the equation for the secondary electric field was discretized using the staggered-grid finite difference scheme and was solved using the Krylov subspace iterative method. For the inversion problem, the general measure of the data misfit and regularization constraints were integrated using the Gauss–Newton minimization method to form an objective function. By selecting appropriate parameters and regularization constraints, the sharp boundaries and blocky features of the inverted target were recovered. The sensitivity was calculated using the reciprocity relationship, and the method was found to be beneficial to determining appropriate initial values of regularization coefficients. Two synthetic data sets and a field data inversion of a crosswell configuration demonstrate the feasibility of this scheme.

An inversion method of marine controlled source electromagnetic (CSEM) data is presented with general measures. In the inversion, the partial differential equation of secondary electric field is discretized by a staggered-grid finite difference method (SGFDM). The generalized productive bi-conjugate gradient (GPBiCG) solver is employed to solve linear system of equations derived from SGFD with a preconditioner of the incomplete Cholesky decomposition. In order to improve the convergence rate of the solver, a static correction is applied. Instead of using a traditional L2 norm, we apply Ekblom norm to formulate the total objective function. The Ekblom norm allows us to integrate a priori structural information with inversion. The inverse problem is solved by the Gauss-Newton method and the reciprocity relation is applied to obtain sensitivity matrix explicitly. A 2D synthetic example shows that the inversion with general measures is capable of recovering a relatively complicated model from a homogeneous layered model.

After Fan introduced the concept of synthetic aperture (SA) to marine controlled-source electromagnetics (CSEM), numerous of optimization methods are applied to SA weights selection for improving the detectability of deeply buried targets under seabed, but few are suitable for the seriously non-linear EM problems. This study presents an application of particle swarm optimization (PSO) to optimization of the phase shift and the amplitude compensation coefficient of SA for marine CSEM. Eigenstate analysis (EA), another non-linear SA weights optimization method, is also carried out in marine CSEM data processing for comparison. The 3D synthetic model reconstructed from Fan (2010) is used to better demonstrate the effects of the detectability with and without two optimization algorithms. In order to validate the effectiveness of PSO, we scan all the confined weights. The results of PSO converge to the global maximum in a robust and fast way. Presentation Date: Monday, October 15, 2018 Start Time: 1:50:00 PM Location: Poster Station 13 Presentation Type: Poster

Marine controlled-source electromagnetic (CSEM) processing techniques have been developed rapidly along with the proliferation of marine CSEM surveys since the early 21th century. A basic data pre-processing procedure has been summarized, including time to frequency domain transform, velocity corrections caused by the towed source, and normalizations of the source moment and source gesture. The basic time to frequency domain transform utilizes a linear least-squares method to solve for the approximate coefficients of the amplitude and phase at each frequency from the overdetermined equations, whereas the short-time Fourier transform (STFT) algorithm provides a full perspective view of the signal components in the spectrum domain of the time and frequency. Regular noise exposed in the STFT spectrogram is easily identified and attenuated by carrying out novel semi-automatic de-noising method proposed in the paper and the real data acquired in somewhere in the Atlantic Ocean has been processed. Presentation Date: Wednesday, September 27, 2017 Start Time: 2:15 PM Location: Exhibit Hall C, E-P Station 2 Presentation Type: EPOSTER

Marine controlled-source electromagnetics (MCSEM) has been used by industry as a geophysical tool for de-risking marine hydrocarbon exploration since early of this century. At present, MCSEM has been applied successfully to identify the presence of hydrocarbon reservoirs in deep water regions. But it is hard to apply MCSEM to shallow water regions, because the air wave interacts with the electromagnetic (EM) signal from reservoirs, which lead to a low signal-to-noise ratio. On the other hand, when it turns to deep water regions with deep-buried reservoirs (4km below sea floor), there is no air wave problem, but the signal of the reservoirs is so weak that one can barely distinguish it from MCSEM equipment noise. In order to mitigate the air wave effects and enhance the intensity of weak signal, we apply synthetic aperture source (SAS) to MCSEM. The concept of synthetic aperture was first introduced to MCSEM by Fan et al. (2010) and SAS construct an extended synthetic source by integrating a series of single sources in a specific configuration to steer the field to some specified offsets by properly weighting the single source fields. In other words, a long synthetic source is constructed by some weighted single sources and the field of long synthetic source is the interference of the fields from all single sources. There are amplitude and phase weights of the single source in SAS field according to Fan et al. (2012), which is the key to achieving significant detectability (reservoir anomalies) of reservoirs in certain model. The SAS field is not only beneficial to interpret the presence of the reservoir, but also help achieve fast and accurate MCSEM inversion. In this study, we design two types of models, which are deep-buried reservoirs in shallow and deep water regions. Then we adopt the mode decomposition method (David and Lucy, 2008) to deduce two potential functions, so the 1D electric and magnetic fields can be numerically computed by using the integral of the Bessel function. Besides, we use the finite element method (FEM) to calculate 2.5D electric and magnetic fields. We apply SAS technique to 1D and 2.5D data on two types of models and the numerical modellings verify the effectiveness of SAS in enhancing the detectability in MCSEM. In addition, we do some research in time-lapsed MCSEM, in order to test the feasibility of MCSEM in reservoir performance monitoring.