Integration of 3D Modeling and Real-time Processing for Enhancement of Anisotropic Formation Evaluation with Borehole Multicomponent Induction Measurements

Abstract

Abstract It is well known that around the world a number of oil and gas reservoirs consist of formations which are identified as resistivity/conductivity anisotropic by borehole induction tools, such as thinly laminated sand-shale or bedded sand-sand rock sequences. Therefore, resistivity-anisotropy formation properties are critical for accurately evaluating anisotropic reservoirs. For many years the logging industry has tried to use induction tools to measure both horizontal and vertical resistivities of reservoir formations. As one of the latest and most remarkable developments in the wireline induction logging domain, multicomponent induction (MCI) logging is now used to fill this requirement. Compared to conventional induction, this new logging technology is able to measure the formation anisotropy (vertical and horizontal resistivities, Rv and Rh, respectively), dip, and strike required to accurately evaluate different types of anisotropic reservoirs. When interpreting MCI data for the purpose of anisotropic formation evaluation, most cases theoretically require 3D electromagnetic (EM) forward modeling and inversion. However, experience has clearly shown that the current 3D forward modeling algorithms often fail to obtain accurate solutions in a reasonable amount of CPU processing time. Even for the most efficient algorithms, fully 3D inversion is impractical for the real-time or well-site delivery of inverted results from measurements. For fast and accurate 3D EM forward modeling, a practical 3DFD (finite difference) method based on an isotropic/transverse isotropic (TI) background is presented and used. This method has been tested by fast borehole-effect correction (BHC) and several independent 3D codes. Its practical application workflow is also proposed and tested. The time-consuming 3D inversion is generally partitioned into a few simple and fast data processes including resolution enhancement of MCI logs for reducing shoulder-bed effects and a few low-dimensional inversions such as radially one-dimensional (R1D) inversion, which makes possible the real-time delivery of formation anisotropy (Rh and Rv), dip, and strike information. Moreover, the R1D inversion is based on a fast and rigorous multistep inversion algorithm and a fast forward modeling engine which consists of the pre-calculated MCI-response library created by using the fast 3DFD method. This novel method of integrating 3DFD numerical modeling and real-time processing technologies has been proposed and implemented for enhancing anisotropic formation evaluation. To demonstrate its capability and effectiveness, we successfully validated the method on both synthetic data and field log data sets. Introduction Borehole multicomponent induction (MCI), also known as triaxial induction, logging was introduced to the oil and gas industry almost a decade ago. Its application has clearly shown that this relatively new logging technique is playing a vital role in improving the evaluation of different types of anisotropic formations. However, its successful application for this purpose is extremely dependent on the accuracy of the processing that yields near-wellbore information including, Rh, Rv, dip, and strike (or azimuth) from raw measurements. Various inversion and processing algorithms based on different forward models have been reported. For example, Wang et al. (2003) presented a fast and rigorous inversion method based on a vertically one-dimensional (V1D) layered formation model with no hole, which claims it is able to simultaneously invert Rh, Rv, dip, and azimuth as well as the bed boundary. But, its input data must be borehole-corrected and it assumes that the dip and azimuth are constant or have no abrupt change in the selected windows. In addition, the tool's transmitters and receivers are described by point magnetic dipoles. However, for MCI data processing the formation models are often three-dimensional (3D) and the dip and azimuth constantly show abrupt changes, which results in inversion errors and questionable data quality.