Flexural-dipole sonic logging has been widely used as the primary method to measure formation shear slowness because it can be applied in both fast and slow formations and can resolve azimuthal anisotropy. The flexuraldipole waveforms are dispersive borehole-guided waves that are sensitive to borehole geometry, mud and formation properties, and therefore the processing techniques need to honor the physical dispersive signatures to obtain an accurate estimation of shear slowness. Traditional processing techniques are based on either a modeldependent method, in which an isotropic model is used as a reference to compensate for the dispersion effect, or a model-independent method, which optimizes nonphysical parameters to fit a simplified model to the field dispersion data extracted in the slowness-frequency domain. Many methods often require inputs such as mud slowness, frequency bandpass filter, or an initial guess of formation shear. Consequently, these methods often fail to interpret the dispersion signature properly when those inputs are inaccurate or when the waveform quality is poor due to downhole logging noises. The users must manually tune the processing parameters and/or choose different methods as a workaround, which causes extra time and effort to obtain the result hence imposes a significant challenge for automating sonic shear interpretation. We develop a physics-driven machine learning-based method for enhancing the interpretation of borehole sonic dipole data for both wireline logging and loggingwhile- drilling. Extensive synthetic databases (i.e., lookup tables) are generated from an anisotropic root-finding mode-search routine and used to train neural network models as accurate and efficient proxies. Those neural network proxies can be used for real-time sensitivity analysis and performing inversion to the measured sonic dipole dispersion data to estimate relevant model parameters with associated uncertainties. Alternatively, various machine learning methods can also be developed based on the generated training dataset and that can be used for inferring relevant model parameters with uncertainties from the field data directly. We introduce how these trained models can be used to enhance the labeling and extraction of different dispersion modes. We developed a new method that outperforms previous modeldependent and model-independent approaches because the new method introduces a mechanism to constrain the solution with physics that also has the capability to incorporate more complicated physical dispersion signatures. This new method needs neither prior information such as mud slowness and formation shear slowness, nor any tuning parameter to be played by the user. It also paves a way to automatically identify different anisotropy mechanisms such as intrinsic, layering, stress, or fractures. This leads to significant progress toward automated sonic interpretation. The algorithm and workflow have been tested on field data for challenging borehole and geological conditions and compared with traditional flexural-dipole processing techniques with great success.

ABSTRACT Electromagnetic formation evaluation currently relies on low-frequency resistivity and high-frequency dielectric measurements that are typically not interpreted jointly. In consideration that formation electromagnetic responses in different frequency ranges are controlled by different physical phenomena, analysis of a wideband electromagnetic response can provide new and complementary sensitivities to formation petrophysical parameters. The frequency-dependent complex conductivity of ion-conductive sediments in the range from millihertz to kilohertz exhibits the spectral induced polarization (SIP) effect, in which the impedance phase has a near-resonance peak at a characteristic frequency due to a strong polarization response. In this study, SIP spectra were measured on a collection of quarried clean sandstones saturated with brines. The influence of other factors on the SIP effect, such as the presence of clays, was minimized by carefully selecting samples. The dielectric dispersion was measured to characterize a subset of twin samples in the megahertz to gigahertz range. The combination of these methods provided core electromagnetic responses over 12 decades of frequency. We established a wideband rock model based on a differential effective medium approach that accounts for both the Maxwell-Wagner interfacial polarization related to the rock texture and the electrochemical polarization due to the presence of charged grains. The model is based on first principles and uses a minimal number of parameters to describe the essential electromagnetic properties of well-sorted clean sandstones in the millihertz to gigahertz range. We investigated the relationship between the SIP effect and the dominant grain size of our sandstone collection. The dominant grain size was determined using a digital image analysis of scanning electron microscope (SEM) images obtained on thin sections. SIP spectra were inverted with the rock model to obtain the dominant grain size. The model was shown to be capable of reproducing well the experimental SIP spectra, with the inverted dominant grain size comparing favorably with values determined from image analysis. We analyzed the wideband electromagnetic measurements by applying the rock model in the full frequency range. The wideband data inversion enabled the estimation of five formation parameters: water-filled porosity, water salinity, cation exchange capacity, dominant grain size, and cementation exponent. Our analysis also demonstrated that the use of only low- or only high-frequency data subsets is not sufficient to reliably invert for the full set of formation parameters.

ABSTRACT: Although the directional sensitivity and deep investigation depths of directional logging-whiledrilling (LWD) resistivity tools have led to their wide use in detecting bed boundaries for well placement, their multicomponent electromagnetic (EM) measurements also have important applications in formation evaluation. An area of current interest is the evaluation of resistivity anisotropy and formation dip in low-angle wells to better quantify hydrocarbon in place. We present a multistep inversion-based workflow for the interpretation of resistivity anisotropy and formation dip from the directional EM measurements. A 1D parametric inversion is used to construct a layer-cake, transversely isotropic formation model that has outputs of horizontal and vertical resistivities ( Rh and Rv ), formation dip, and layer thicknesses. The procedure takes advantage of predominant sensitivities of different groups of measurements to various formation parameters. After identifying boundary positions, we first invert Rh only from the standard resistivity logs and then add Rv and dip to the inversion model successively, each time incorporating a different group of measurements in the inversion. This significantly improves the robustness of the inversion. We also develop a method of assigning confidence levels to the inversion results for log quality control (LQC), using data misfits and the rate of dip change. The workflow can be used both to process recorded-mode data and to perform real-time interpretation, while drilling, from a subset of input channels. In addition, the algorithm can be applied to conventional LWD resistivity logs to improve their vertical resolution in low-angle wells and to extract Rh and Rv in high-angle wells. The inversion algorithm has been validated, first with synthetic cases constructed from field logs and then with field data in vertical and deviated wells. The examples show that resistivity anisotropy and dip information can be consistently determined from the input logs when the model is appropriate.

LWD invasion processing based on standard phase and/or attenuation resistivity values often leads to substantial underestimates of the true (uninvaded) formation resistivity Rt when dielectric effects are present on the data. This paper discusses using di

ABSTRACT We develop a novel algorithm for the integrated petrophysical evaluation of hydrocarbon-bearing formations using dual-physics measurement data. Specific data sets used in this paper are (a) timelapse electromagnetic (EM) measurements acquired with an array induction logging tool, and (b) pressure measurements acquired with a multi-probe wireline formation tester in a vertical borehole. Dynamic behavior of saturation and salt concentration distributions due to mud-filtrate invasion creates a two-phase three-component flow system. In this work, the inverse problem associated with dual-physics wireline measurements consists of the estimation of a two-dimensional (2D) axisymmetric petrophysical model described by layered parametric spatial distribution of both vertical and horizontal absolute permeabilities and porosities. We pose the inverse problem of estimating layer petrophysical parameters from discrete time-lapse EM induction and pressure measurements as an optimization problem. A weighted least-squares constrained optimization method is employed to solve the inverse problem. Tool responses within the framework of the iterative inversion scheme are numerically computed via simulating dynamic physics of two-phase three-component flow that takes place during mud-filtrate invasion and subsequent formation testing. Time-lapse EM induction measurements are simulated in a coupled mode to the fluid flow. Numerical examples have shown that the benefit of the joint inversion algorithm described in this paper is in the reduction of the nonuniqueness associated with the estimation process by simultaneously honoring two sets of complementary measurements that contain independent information about the underlying model.