Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Cross-well tomography
Practical Implications of Nonlinear Inversion For Cross-well Electromagnetic Data Collected In Cased-wells
Gao, Guozhong (Schlumberger) | Alumbaugh, David (Schlumberger) | Zhang, Ping (Schlumberger) | Liu, Jianguo (Schlumberger) | Zhang, Hong (Schlumberger) | Levesque, Cyrille (Schlumberger) | Rosthal, Richard (Schlumberger) | Abubakar, Aria (Schlumberger) | Habashy, Tarek (Schlumberger)
Summary This paper first summarizes the theory of providing resistivity images from cross-well electromagnetic (EM) measurements in metal cased holes. Metallic casing significantly attenuates EM signals. As a result, low frequencies are required to acquire useful EM signals in cased holes, which reduces the resolution within the inversion images. Comparison of casing materials shows that chromium steel casing attenuates the EM signals much less than carbon steel casing. Although cross-well EM technology can be successfully applied in open/fiberglassto- open/fiberglass and open-to-metal cased wells irrespective of the type of the casing, for casing-to-casing scenarios, cross-well technology at this point is limited to chromium-to-chromium scenarios. Methods have been developed for the inversion of crosswell EM data in cased holes using data ratios to remove effects, or including the casing coefficient as an unknown in the inversion. This paper investigates the spatial and frequency-dependent sensitivity for different frequencies and inversion approaches for cross-well EM data acquired in cased holes. The applicability and limitations of these approaches are demonstrated on synthetic data simulating a water flood. Introduction open/fiberglassto- open/fiberglass, open-to-casing/ casing-to-open, and casing-to-casing. Cross-well Electromagnetics (EM) is an emerging deepsensing technology that has been shown considerable promise for interwell petroleum reservoir characterization and fluid monitoring. Initial experiments are captured in Wilt et al., 1995. In an active oil-field environment, one or both of the wells available to the cross well sensors will frequently be steel-cased. Because steel attenuates electromagnetic signals due to high conductivity and magnetic permeability, and because the attenuation increases with increasing frequency, measurements in steel casing require lower-than-optimal frequencies to be employed. As a result, cross-well EM surveys can be classified into three operating situations:
- Reservoir Description and Dynamics > Reservoir Characterization (1.00)
- Reservoir Description and Dynamics > Reservoir Fluid Dynamics > Flow in porous media (0.56)
- Reservoir Description and Dynamics > Formation Evaluation & Management > Cross-well tomography (0.47)
SUMMARY The growing use of the controlled-source electromagnetic method (CSEM) for exploration applications has been driving the technical development of data acquisition, as well as three-dimensional (3D) modeling and imaging techniques. However, targeting increasingly complex geological environments also further enhances the problems inherent in large-scale inversion, such as nonuniqueness and resolution issues. In this paper, we report on two techniques to mitigate these problems. We use 3D joint CSEM and MT inversion to improve the model resolution. Further, a hybrid model parameterization approach is presented, where traditional cell-based model parameters are used simultaneously within a parametric inversion. INTRODUCTION Large-scale inverse problems are usually under-determined, meaning that there are more unknowns, typically in the form of digitized model meshes, than data. This adds to the problem that errors are associated with every geophysical data. The resulting issue is referred to as the problem of non-uniqueness of inverse solutions. To mitigate this problem and to improve the resolution in an inversion, it is common to take advantage of complementary natures of different geophysical datasets. In electromagnetic imaging, magnetotelluric (MT) data, providing conductivity structure information on a gross scale, can be combined with CSEM data. With the latter method responding stronger to thin resistive targets, the joint CSEM and MT inversion has the potential of limiting ambiguities in the EM data interpretation relevant to many exploration scenarios. However, even with improved resolution capabilities, the solutions of 3D large-scale cell-based (or pixel-based) inversions with finely sampled models usually are still nonunique. Several strategies have been reported to limit the ambiguities for reconstructed targets and its conductivities. For cell-based problems, model-smoothing constraints are usually applied, limiting the solutions to a class of geologically more meaningful ones, i.e. avoiding too high conductivity contrasts. A different approach is to actually address the under-determinacy by casting the problem into a parametric problem. Usually, particular geometric shapes are assumed in parametric solutions, requiring a priori information. A model parameterization can for example be based on interfaces known from seismic reflection data. The 2D sharp-boundary inversion algorithm by Smith et al. (1999) features a parameterization with variable nodebased boundaries and greatly limits the number of unknowns. Parametric inversion algorithms have also been used for the simultaneous reconstruction of both geometry and conductivity of unknown regions (Commer, 2003; Zhang et al., 2007). The obvious drawback of such methods is the necessity of sufficient background information in order to find a suitable model parameterization. Here, we propose to use a hybrid approach, overlaying a cell-based inversion for a particular area of interest with a parametric inversion. This combines the advantages of cellbased and structure-based model parameters. We present two joint inversion examples using synthetic CSEM andMT data. The first example employs only cell-based model parameters, and simulates a survey in a marine environment. Second, we present an inversion study for a surface survey, using the hybrid parameterization approach. METHOD Our inversion algorithm’s underlying finite-difference (FD) forward modeling algorithm for EM field simulation solves a modified form of the vector Helmholtz equation for scattered or total electric fields.
- North America > United States (0.71)
- Africa > South Africa > Western Cape Province > Indian Ocean (0.24)
- Reservoir Description and Dynamics > Reservoir Characterization > Seismic processing and interpretation (0.77)
- Reservoir Description and Dynamics > Formation Evaluation & Management > Cross-well tomography (0.73)
Summary An inversion methodology for marine controlled-source electromagnetic (MCSEM) data with approximate Hessianbased optimization and a fast finite-difference time-domain forward operator is presented. Using data from a synthetic hydrocarbon reservoir, we demonstrate that models are reproduced with a spatial resolution determined by the skin depth of the frequencies included in the inversion. Both single and multiple resistive bodies can be resolved in the subsurface. Using reciprocal treatment and multiple frequencies at each receiver position, the comprehensive inversion sequence of a typical MCSEM survey, which should match the acquired data to within the measurement error, executes within ~100 iterations, with about 30 iterations per day, requiring at most a few hundred nodes on a parallel cluster. Introduction MCSEM surveys have been used as geophysical survey tools for several decades (e.g., Spies et al., 1980) and were revived after the pioneering studies for hydrocarbon exploration using Seabed Logging in this decade (e.g., Eidesmo et al., 2002). Continued evolution in operational accuracy and equipment have resulted in a vast improvement in data quality (see examples shown in Zach et al., 2008), which will allow the accurate measurement of both magnitude and phase for complex survey designs with arbitrary source-receiver orientations. This will further drive the development and application of 3D inversion algorithms. The inverse CSEM problem has been the subject of a number of studies. Newman and Alumbaugh (1996, 1999) used a finite-difference frequency-domain solver for a wide frequency band on a staggered grid, in an inversion scheme based on conjugate gradient update steps; see also the implementation study by Commer et al. (2008). Mackie and Watts (2007) and Bornatici et al. (2007) also use a staggered grid, finite-difference frequency-domain solver (e.g., Mackie, Madden, Wannamaker, 1993) in a preconditioned conjugate gradient loop for the joint inversion of both marine CSEM and MT data. 3D inversion based on integral equation Maxwell solvers is the subject of several studies from the University of Utah, where the MCSEM inverse problem is again formulated in regularized conjugate gradient steps, see Gribenko, Zhdanov, 2007. An efficient approach to 3D MCSEM inversion with a finite-volume forward solver, where the gradient is computed using the adjoint state method, was found by Plessix, 2006, and applied in Plessix, van der Sman, 2007. There, a quasi-Newton inversion scheme is used with a diagonal approximation to the inverse Hessian matrix. The forward solutions to all aforementioned approaches rest on the solution of Maxwell’s equations in the frequency domain, which requires separate solutions for each frequency mode in the source spectrum. With increasingly wide and complex frequency spectra in MCSEM surveys (Mittet, Schaug-Pettersen, 2007), this comprises a considerable numerical challenge. We present a method which is based on the finite-difference timedomain solution developed by Maaø, 2007, permitting the forward-solution of multiple frequencies simultaneously. Following Støren et al. (2008), the gradients are calculated from the difference field between real and synthetic data using the first Born-approximation. A quasi-Newton update is used for the optimization, where the Hessian matrix is approximated with BFGS matrices using the gradient projection method (Zhu et al., 1997, Byrd et al., 1995).
- Reservoir Description and Dynamics > Reservoir Characterization > Seismic processing and interpretation (0.36)
- Reservoir Description and Dynamics > Formation Evaluation & Management > Cross-well tomography (0.34)
- Reservoir Description and Dynamics > Reservoir Characterization > Exploration, development, structural geology (0.34)
Summary While 3-D inversion is a particularly effective tool for interpreting marine CSEM data, 3-D data acquisition is needed to image the resistivity structure of an anisotropic subsurface. Tests on synthetic models show resistivity artifacts when anisotropic data are forced through isotropic inversion and when offline data are excluded from anisotropic inversions. Similar artifacts appear also when the inversion starting model differs too much from the actual subsurface. As a result, inadequate attention to anisotropy or inadequate data coverage can lead directly to misinterpretation of the subsurface resistivity structure. Anisotropic imaging with good data coverage, accurate receiver orientation, and good initial resistivity models are necessary to quantitatively image resistive anomalies in an anisotropic earth. Introduction Logs in horizontal and vertical wells have shown that subsurface resistivities can be highly anisotropic. This intrinsic, bed-scale anisotropy can be further magnified in a marine CSEM survey where the longer wavelengths average over different lithologies, each of which may itself be anisotropic. In any case, anisotropy must be included in any attempt to tie CSEM data to wells. The CSEM source must drive both horizontal and vertical currents through the subsurface in order to support anisotropic inversion. Jupp and Vozoff (1977) noted that magnetotelluric (MT) data generally probe the subsurface with horizontal currents in contrast to DC resistivity data where the currents are more nearly vertical. They showed how the joint data sets could be inverted to distinguish horizontal from vertical resistivity. More recently, Lu and Xia (2007) showed how CSEM source lines displaced from the receivers ("offline data") are more effective at probing horizontal resistivities, while the online or "flyover" CSEM data probe vertical resistivities more effectively. Where available, measurements of the vertical magnetic field can also prove useful to discriminate horizontal resistivity. The additional data required to analyze anisotropy renders “hand” interpretation by iterative modeling too cumbersome for general use. Inversion on parallel computers is a more practical and efficient approach to hydrocarbon exploration with CSEM data. CSEM inversion is normally formulated as a functional optimization with respect to the subsurface resistivity parameters (Newman et al., 1997). While inversion of CSEM data is already underdetermined and the addition of anisotropy at least doubles the number of unknowns, numerical experiments described in this paper showed that, given sufficient data coverage, the quality of the anisotropic inversion solution for an anisotropic earth is comparable to that of the isotropic inversion for an isotropic earth. Although each model update step is the solution to a linear problem, CSEM inversion remains a nonlinear problem. The final resistivity images from the inversion depend on the initial models. The best images are, of course, obtained when the initial models are as close as possible to the actual background resistivities. It is important to understand how both the horizontal and vertical resistivity structures respond to errors in the starting models. In this paper, we show the effects of anisotropy, data coverage, and initial models on CSEM inversions from the practical point of view by comparing exploration-scale inversions of 3-D synthetic data sets.
- Reservoir Description and Dynamics > Reservoir Characterization > Exploration, development, structural geology (0.48)
- Reservoir Description and Dynamics > Formation Evaluation & Management > Cross-well tomography (0.48)
Possible Source Effects Observed In a Magnetotelluric Monitoring Site In Southern Italy
Siniscalchi, Agata (University of Bari) | Balasco, Marianna (Institute of Methodologies for Environmental Analysis) | Telesca, Luciano (Institute of Methodologies for Environmental Analysis) | Lapenna, Vincenzo (Institute of Methodologies for Environmental Analysis) | Romano, Gerardo (Institute of Methodologies for Environmental Analysis)
Summary Since 2003 continuously operating magnetotelluric (MT) systems have been installed in Agri Valley (Southern Italy) by the Institute of Methodologies for the Environmental Analysis (IMAA). The valley is a NW-SE intermontane basin filled by middle Pleistocene alluvial deposits. The pre-quaternary bedrock is composed by Mesozoic- Cenozoic carbonates, mainly outcropping along the west site of the basin, while toward east and south-east, the bedrock is formed by Tertiary siliciclastic sediments. The valley of the Agri River is affected by a coherent fault system N120 trending left-lateral strike-slip fault. In recent years the Agri Valley has been interested by intensive oil extraction. In the present study the stability of the apparent resistivity values in several sites of Agri Valley is performed in order to find a site sufficiently “quiet” to be used as remote reference in a MT network. Introduction At present in the investigated area of Agri Valley two MT stations are operating: Tramutola (LAT. 40.297 LONG. 15.805, 700 m a.l.s.) in south-eastern flank of the Valley and Villa d''Agri (LAT. 40.193 LONG. 15.495, 587 m a.l.s.) in the middle of the Valley. Each MT station is equipped with a receiver MT24LF (Magnetotelluric 24-bit A/D Low Frequency system) which records the magnetic field by means of two induction coils (EMI Inc., BF4), and the electric field by means of electrical dipoles. In Villa d''Agri site the electrical field is measured in 0°N direction by means of 70m and 50m length electrical dipoles, Ex1 and Ex2 respectively, and along 90° N direction by a 50m length electrical dipole. In Tramutola site 50m length electrical dipoles are one in 0°N direction (Ex,) and two in 90°N direction (Ey1 and Ey2). The magnetic field, in each station, is measured along 0°N and 90°N directions (Hx and Hy respectively). Methodology In this study we analyse the data measured by Tramutola station. In particular, the analysis of the data of apparent resistivity between July 2007 and February 2008, with few data missing, is focused. The frequency of data recording was set to 6.25Hz (low frequency) in continuous mode; monthly data at high frequency sampling (250Hz) have been acquired for 30 minutes at 1:00 GMT, in order to have the best signal-to-noise ratio. The continuous data stream at low frequency was subdivided into subsets corresponding to intervals of different length hours and processed separately. No significant modifications in apparent resistivity and phase estimates have been observed considering sets of 2, 4, 6, 8 and 24 hours until a maximum period of 84.56 s. To obtain a sufficiently long sample for reliable statistics (Eisel and Egbert, 2001) and a sufficiently short time window to appreciate possible variations in the investigated series for the further analysis the 8 hour length subsets were used. Therefore, the data were analyzed using the Detrended Fluctuation Analysis (DFA). This method highlights the presence of a scaling behavior in the resistivity time series. The DFA was originally proposed by Peng et al. DFA as a method for determining the scaling behavior of data in the presence of possible trends without knowing their origin and shape.
- Geology > Rock Type > Sedimentary Rock > Clastic Rock (0.54)
- Geology > Structural Geology > Fault (0.54)
- Reservoir Description and Dynamics > Reservoir Characterization > Faults and fracture characterization (0.88)
- Reservoir Description and Dynamics > Formation Evaluation & Management > Cross-well tomography (0.83)