Shale Anisotropy Estimation From Logs in Vertical Wells

Abstract

Anisotropic elastic parameters for shales are widely needed in seismic imaging, reservoir characterization, and carbon sequestration monitoring. Unlike other elastic parameters such as vertical P and S wave velocities, anisotropy parameters are not measured directly from the acoustic well logs due to the single-directional nature of a well. We assume that shale anisotropy is induced by thin cracks that are filled with liquid in a background isotropic medium, whose bulk and shear moduli are obtained from the vertically measured P and S wave velocities, density, and porosity from corresponding well logs through a formalized inversion scheme. We show that the estimated anisotropy using the proposed method is consistent with the mineralogy and agrees with the published laboratory measurements. This framework allows us to quantify the uncertainties in the anisotropy parameters estimated from the inversion, which can be used as a measure to evaluate the validity of the chosen rock physics model. Plain Language SummaryThe differences in the wave speeds as they propagate in different angles, defined as anisotropy, are widely needed for imaging the subsurface and understanding the tectonic processes. However, measurements made in a vertical borehole cannot provide this directional information. In this study, we assume a rock physics model for shales that adds anisotropy-inducing thin inclusions in an isotropic background, the elastic and fluid properties of which are inverted formally from the measured well logs. The proposed method generates anisotropy estimates that agree with published laboratory measurements and that are consistent with the mineralogy log. Moreover, the proposed method quantifies the uncertainties in the estimated anisotropy, which can be used to evaluate the applicability of the chosen model on a particular rock formation.

Shale Anisotropy Model Building Based on Deep Neural Networks

“…However, it is impossible to collect massive labeled well log data in the field due to the lack of direct measurements of seismic anisotropy in well logging. The Hudson-Cheng model has been demonstrated to be the simplest applicable model to shale anisotropy estimation from well log data in our previous research (Li et al, 2019). Therefore, we employ the Hudson-Cheng model to generate paired training data of features (i.e., the inputs to the neural network) and labels (i.e., true values for the outputs from the neural network).…”

“…Therefore, we employ the Hudson-Cheng model to generate paired training data of features (i.e., the inputs to the neural network) and labels (i.e., true values for the outputs from the neural network). As shown in Figure 1, the Hudson-Cheng model decomposes an intact rock into two sections: (1) a homogeneous, isotropic background composed of a rock matrix and a group of randomly distributed spherical pores, also known as stiff pores, and (2) a set of aligned ellipsoidal cracks, also known as soft pores, that account for VTI anisotropy (Li et al, 2019). The effective elastic moduli for the entire rock are given by…”

“…Among all anisotropic rock physics models, the ellipsoidal crack model of Hudson (1981) and Cheng (1993) is the simplest and most efficient Eshelby inclusion-based model, with only four physically meaningful and bounded parameters. Li et al (2019) formalized an inversion workflow to estimate anisotropy from vertical well logs based on this simple model, yielding lithologically consistent results. Despite the consistency and simplicity offered by the inversion workflow and the Hudson-Cheng model, significant amounts of human interpretation and subjective parameter selection are still needed.…”

Crack‐Induced Shear‐Wave Orthorhombic Anisotropy: Modeling and Inversion of Shear‐Wave Borehole Measurements

“…where the C Θ 13 and C Θ 11 are calculated by the ANNIE approximation (Schoenberg et al, 1996), and subscript Θ represent "parallel" in (B2) or "perpendicular" in (B3), because borehole measurements are insensitive to these two parameters (Li et al, 2019;Xu et al, 2017).…”

“…The transversely isotropic model with a vertical axis of symmetry (VTI) is often used to describe sedimentary layered formations, especially shales (e.g., Barbosa et al, 2017; Choens et al, 2019; Li et al, 2019). As for acoustic logging in a vertical borehole, the properties for a VTI formation have been studied by, among others, White and Tongtaow (1981) and Tongtaow (1982) for monopole logging and Ellefsen (1990) and Schmitt (1989) for multipole logging.…”