Abstract Geophysical well logging plays a crucial role in petroleum and natural gas exploration and production. In recent years, artificial intelligence—particularly machine learning—has driven remarkable progress in developing new logging methods and enhancing log data processing. However, research focused on logging tools themselves remains limited, primarily due to the constrained computing capabilities of downhole processors. This study reviews the current applications of machine learning in logging tools and embedded processors, examining potential implementations from three perspectives: the final log data produced by the tools, intermediate electronic signals, and other related data such as experimental measurements and noise information. Representative application scenarios include real-time prediction of logging parameters during drilling, adaptive control of acoustic excitation systems, and intelligent fault diagnosis of downhole electronic circuits. Finally, the challenges and future research directions are summarized in three key aspects: database construction, neural network model optimization, and embedded processor selection. This review systematically sorts out the research gap of machine learning application in downhole logging tool hardware deployment, and clarifies the core technical logic of matching intelligent algorithms with embedded processors for the field. This study provides a comprehensive reference for integrating intelligent algorithms into downhole logging technologies.

Borehole acoustic reflection imaging technology has been widely applied in hydrocarbon exploration.A detailed understanding of the radiation characteristics of borehole acoustic sources is crucial for imaging performance analysis and method optimization,particularly in formations with high porosity and permeability.This study,based on Biot-Rosenbaum theory,derives analytical expressions for the radiation directivity and wavefield energy flux of multipole sources in permeable porous media using the steepest descent method and the complex Poynting vector.A comprehensive parametric analysis is conducted to assess how poroelastic properties and source frequency influence the radiation behavior of different wave modes.Results reveal that a slow P-wave component emerges in permeable porous media and that increasing porosity and permeability substantially attenuate the radiation efficiency of multipole sources.Dipole radiation efficiency shifts toward lower frequencies,accompanied by broadened spectral responses.The SV-wave efficiency transitions from a dual-peak to a single low-frequency pattern.Although the SH-wave remains the dominant mode,its imaging capability degrades markedly under high-porosity and high-permeability conditions.A field reflection imaging case validates the theoretical predictions.These findings offer theoretical support for evaluating and interpreting the performance of acoustic imaging in permeable porous reservoirs.

Developing an effective and efficient method for extracting slowness from logging-while-drilling (LWD) dispersive guided-wave data has always been challenging. Conventional physics-driven methods estimate slowness by fitting theoretical dispersion curves to dispersion data extracted from waveforms. However, this approach is hampered by the inherent computational inefficiency of forward modeling and the instability arising from the extracted dispersion data quality. An efficient processing workflow addresses these limitations by employing a deep learning surrogate model for dispersion curve mapping, coupled with clustering and optimization modules, to determine LWD multipole slowness. The deep learning model employs a ResNet-LSTM architecture, where the residual network (ResNet) module extracts hierarchical features from formation parameters, and the long short-term memory (LSTM) layer models the sequential mapping to predict the dispersion curve. Subsequently, density clustering techniques ensure the quality and validity of the extracted dispersion data, while an optimization algorithm reduces the number of required forward calculations, thereby enhancing inversion efficiency. Comprehensive evaluations on synthetic and field data demonstrate that the workflow provides reliable slowness inversion results with enhanced computational efficiency in both single and joint inversion.

Acoustic logging-while-drilling (ALWD) has been widely used in recent years. This study demonstrates that the ALWD data can be used to estimate formation S-wave anisotropy, providing important information for the exploration and development of unconventional reservoirs. The inversion of the S-wave anisotropy is jointly done using the monopole Stoneley- and quadrupole-wave data, which, when applied in the ALWD condition, is hampered by the high cost of forward modeling and drilling noise. To achieve fast and stable inversion, a multi-component system which contains CNN-LSTM network, gradient-descent optimization, and joint inversion strategy is developed. It not only enhances the forward modeling accuracy and improves inversion efficiency but also reduces the nonuniqueness of the estimated anisotropy parameters. For the construction of this system, we first analyze the sensitivity of key parameters in transversely isotropic formations and construct a multipole LWD dispersion dataset. Then, a CNN-LSTM model is designed to build a nonlinear mapping between formation parameters and dispersion curves, which greatly enhances the forward modeling efficiency. The trained model is then introduced into the joint inversion objective function. A gradient-descent algorithm is further used to optimize the inversion process, attaining convergence in only a few dozen iterations. The efficient inversion method is applied to field ALWD data. The results show that the method is six orders of magnitude faster than conventional methods and the accuracy and robustness of the results are also improved, thus providing a fast and effective method for the ALWD applications.

Acoustic logging tools, deployed thousands of meters underground to detect geological structures and evaluate reservoir fluids, are essential for oil and gas exploration and development. These tools generate acoustic signals through piezoelectric ceramic transducers. The material properties of piezoelectric ceramics are significantly affected by the high-temperature downhole environment, leading to a failure in impedance matching between the transducer and its excitation circuit. This results in a substantial degradation of the tool’s performance. This paper experimentally obtains the electrical parameters and excitation energy of commonly used monopole transducers at different temperatures. Based on this data, the optimal matching inductance values at various temperatures are calculated. A temperature-adaptive transducer excitation circuit is then designed and implemented. This circuit can adjust the excitation frequency according to the measured temperature to compensate for resonant frequency drift and select the optimal inductor tap via a programmable multiplexer. Experimental results demonstrate that this circuit significantly enhances the transducer’s excitation energy at high temperatures. This technology is expected to markedly improve the operational stability of acoustic logging tools and facilitate the exploration and development of deep and ultra-deep oil and gas resources.

To evaluate cementing quality in the absence of a liquid medium inside the casing,Electromagnetic Acoustic Transducers(EMATs)can be employed to generate quasi-Lamb waves that propagate circumferentially along the casing.The attenuation characteristics of these waves can be analyzed to assess the bonding quality at the casing-cement interface.This paper investigates the dispersion and attenuation characteristics of quasi-flexural Lamb waves propagating circumferentially through both theoretical and numerical simulation methods,comparing the results with those of flexural Lamb waves propagating axially.The findings indicate that when the casing is coupled with fluid,the phase velocity of the quasi-flexural Lamb wave propagating circumferentially is slightly lower than that of the flexural Lamb wave propagating axially,but its attenuation is significantly higher.In terms of wave structure,the primary displacement during propagation along the casing is radial in both cases.While the displacement distribution between the inner and outer walls is uniform during axial propagation,the displacement amplitude of the outer wall is higher than that of the inner wall during circumferential propagation,resulting in energy leaking outward.When the casing is coupled with lightweight cement,the dispersion curve of the flexural Lamb waves propagating axially exhibits discontinuities,whereas the dispersion curve of the quasi-flexural Lamb wave propagating circumferentially remains continuous.Additionally,as the cement acoustic impedance increases,the attenuation of both axially and circumferentially propagating flexural Lamb waves initially rises and then declines.However,the attenuation of the quasi-flexural Lamb wave propagating circumferentially is significantly higher,and the cement impedance corresponding to the maximum attenuation is notably higher for the circumferentially propagating quasi-flexural Lamb wave compared to the axially propagating flexural Lamb waves.Using finite element simulation software,the well logging responses of the quasi-flexural Lamb wave propagating circumferentially and axially along the casing were simulated.The extracted attenuation values are consistent with the theoretical calculations.The study lays a theoretical foundation for further research on cementing quality evaluation using quasi-flexural Lamb wave that propagates circumferentially along the casing.

Numerical simulation of borehole acoustic reflection imaging of far-borehole fractures requires consideration of the borehole size and the fine structure of the fractures. The 3D discretization for fine structure modeling results in a huge number of grid nodes and extremely long computational time, which severely limits the efficiency of forward modeling of far-borehole reflection wave fields in fractured reservoirs. To address this problem, an analytical method was proposed for quickly calculating snapshots of acoustic reflection imaging wave fields on both sides of a fracture using the far-field asymptotic solution of dipole radiation shear wave displacement. This method combines the angular spectrum method with the slip interface theory. The waveforms obtained are consistent with those obtained by 3D finite-difference time-domain elastic wave simulations. For the same fracture model, the computational efficiency of the analytical method is two orders of magnitude faster than that of the finite-difference numerical method. Using this method, the effect of shear stiffness changes on the real-time propagation of dipole SH wave fields in a double-fracture system was analyzed. These findings can be used to characterize and interpret multi-fracture systems in single-well reflection imaging using field data. The results show that shear stiffness changes caused by near-fracture filling materials are a prerequisite for the effective identification of far-fractures in multi-fracture systems. The analysis results provide a theoretical basis for time domain simulation and imaging analysis of dipole acoustic reflection imaging wave fields in multi-fracture systems.

Cementing plays a critical role in petroleum well completion, ensuring well integrity and zonal isolation. Conventional acoustic impedance-based logging tools face limitations in evaluating low-density cement bonding; however, these challenges can be effectively addressed using circumferential SH-waves. SH-waves, polarized axially along the casing and propagating circumferentially, offer distinct advantages for assessing the cement bond. In this study, the dispersion and attenuation characteristics of SH-waves under casing and steel plate models are investigated using analytical methods. The radial or through-thickness variations in displacement amplitude and energy density are quantified. Key findings reveal distinct attenuation behaviors between casing and steel plate geometries due to curvature effects. While the steel plate exhibits a slight decrease in zeroth-order SH (SH0) mode attenuation with increasing frequency, the casing demonstrates a pronounced increase in SH0 attenuation under identical conditions. Particle displacement amplitude progressively intensifies from the inner to the outer casing wall as frequency increases, with SH-wave energy distribution simultaneously concentrating toward the outer wall. Due to the symmetry of the steel plate's upper and lower surfaces, displacement amplitudes and energy distributions are symmetric with respect to the middle plane. Frequency-dependent analysis reveals an inverse relationship between attenuation and frequency for SH1/SH2 modes in casing structures, accompanied by pronounced energy localization at the inner wall. Consequently, these modes demonstrate limited discriminative capability for cement impedance assessment in thick-walled casing applications. The SH0 mode attenuation demonstrates superior differentiation of cement impedance in thick-walled casing. These results are significant for applying SH-waves in cementing quality evaluation under complex well conditions.

The electroacoustic response of high-voltage burst-driven piezoelectric transmitter systems is jointly influenced by the drive circuit, load coupling, dynamic impedance, and transient nonlinearities, making it difficult for conventional theoretical models to accurately describe the mapping from excitation conditions to acoustic output. This study establishes a synchronous electroacoustic measurement system and an experimental database containing control parameters, excitation-voltage waveforms, excitation-current waveforms, and hydrophone responses. The proposed framework is demonstrated using a monopole piezoelectric transmitting transducer for acoustic logging. Four data-driven prediction frameworks with different levels of input information are developed and compared, including a PCA--MLP model based on control parameters, a conditional temporal convolutional network (TCN) driven by theoretical control waveforms, a two-stage cascade model using predicted electrical waveforms as intermediate states, and a TCN driven by measured excitation voltage and current. The measured-electrical-input TCN achieves the highest prediction accuracy, with a coefficient of determination of 0.9984, a mean waveform correlation coefficient of 0.9988, and a mean peak-to-peak relative error of 1.172 %. The results demonstrate that excitation voltage and current constitute key intermediate state variables linking excitation control to acoustic output. Compared with theoretical control information alone, measured electrical-input waveforms provide a substantially more complete representation of the operating state of the transmitter. The proposed hierarchical data-driven framework enables virtual sensing of acoustic output and provides an approach for transmitter-state evaluation, operating-frequency selection, and excitation-parameter optimization. Although validated using an acoustic logging transmitting transducer, the proposed methodology is readily extendable to other high-voltage burst-driven piezoelectric transmitter systems after appropriate retraining.

The evaluation of cementing quality through tubing is widely needed in the abandonment of offshore oil and gas well,so it is of great significance to study the evaluation method of cementing quality in double casing well.Aiming at the defect that the high frequency acoustic wave method such as Cement Bond Logging(CBL)and Variable Density Logging(VDL)have small radial penetration and cannot be used to evaluate the cementing quality through steel tubing,this paper has carried out theoretical and experimental research on low frequency dipole acoustic logging.Firstly,the dispersion curve of double casing well under dipole source is analyzed theoretically,and it is found that there are additional high order casing flexural modes when the outer casing is poorly cemented.On this basis,the high order casing flexural modes is extracted from the waveform and a dipole cementing quality evaluation method is formed.Finally,the experimental research on cementing quality evaluation of double casing well is carried out based on the laboratory model well,which further verifies the validity and reliability of this method.The theoretical and experimental results show that the extra high order casing flexural modes can reflect the cement bonding condition of the outer casing,which provides a feasible method for the evaluation of cementing quality of double casing well.

Fractures are the main seepage channels and storage spaces of fractured-vuggy carbonate rocks, and their accurate description has great significance to the classification and evaluation of reservoirs. The well logs response characteristics of fractures under different well types are different, and the evaluation methods are different as well. In vertical wells, continuous core and logging data are obtained to accurate calculation for parameters such as fracture length, width and thin section porosity after calibration, by drilling ground-coring scientific exploration wells. The combination of array acoustic logs and micro-resistivity image logs are used to evaluate the effectiveness of fractures and connectivity of pore systems. For horizontal well reservoir fracture evaluation, through the establishment of fracture logs response characteristics chart, the relationship between fractures, formation and wellbore is analyzed, thus the technical support is provided for the construction of horizontal wells in the long open-hole section. For evaluation of borehole side fractures, optimization of remote detection acoustic logging data processing algorithm, clear image of borehole side fracture reflectors, and establishment of borehole side fractures description and evaluation methods. The application of the above logging technology has realized the fine description and evaluation of fractures, from the borehole to borehole side under the different well types. It provides an important basis for the optimization of oil testing schemes for fractured-vuggy carbonate reservoirs.

Rock fracturing is widespread in the upper crust of the earth. Nearly all in-situ rocks are subject to subsurface pressures. Understanding the influence of effective pressure (difference between confining stress and pore pressure) on the elastic properties of fractured rocks is crucial for estimating in-situ seismic properties. Wave-Induced Fluid Flow (WIFF) and fracture Elastic Scattering (ES) attenuation mechanisms have been widely studied. However, the effects of effective pressure on WIFF and ES mechanisms in fractured rocks are relatively unexplored. To investigate the pressure influence on WIFF and ES in P and SV wave propagation, we developed a pressure-dependent dynamic model to incorporate Multi-shaped Microcracks' Squirt Flow (MMSF), Fracture-Background WIFF (FB-WIFF), Biot Flow, and ES mechanisms with non-linear elastic and hyper-elastic deformation stages. The results show that both the WIFF and ES mechanisms are affected by effective pressure, except for the ES mechanism of normal incident SV waves. The WIFF is more susceptible to effective pressure than the ES mechanism. In addition, the closure of microcracks (soft pores) during effective pressure loading continuously decouples the MMSF mechanism from the FB-WIFF, Biot Flow, and ES mechanisms, causing the increase in the velocity of the P- and SV- waves and the variations in the attenuation processes. These wave propagation characteristics help to understand the hydraulic properties of in-situ fractured rocks. By comparing model predictions with ultrasonic laboratory velocity data under varying effective pressure loading, we validated our model.

In acoustic logging, multipole waves (such as dipole waves and quadrupole waves) generated by the sound source are dispersive mode waves, and their dispersion effects can lead to inaccurate wave velocity measurement. Traditional physics-driven methods for forward modeling dispersion curves are computationally inefficient, affecting the reliability and timeliness of wave velocity applications in pore pressure prediction and wellbore stability analysis. A neural network-driven method for forward modeling the dispersion of mode wave is proposed to address this issue. This method combines equivalent tool theory with borehole acoustic field propagation theory to forward model dispersion datasets. Using a multi-layer fully connected neural network architecture and dispersion calculation model training, it can rapidly output theoretical dispersion curves. Applications show that using the proposed neural network model reduces computational costs by four orders of magnitude. Theoretical simulations and field data validation confirm the stability and effectiveness of the method. This study develops a novel approach for forward modeling the theoretical dispersion of multipole mode waves and inverting formation parameters.

The classical elastic wave theory can model the propagation of the acoustic field in a cased well as a structure with cylindrical layers. A thin fluid annulus between the interface is often used to simulate uncemented conditions. However, the traditional cased borehole models and their associated techniques are idealized approximations. In reality, irregular and rough interfaces, along with poor cement bonding, lead to partial contact between the interfacial surfaces. This results in shear coupling as acoustic waves propagate through the interface. Therefore, the fluid annulus model is not applicable to the partially cemented interface. To address this limitation, we propose a casing coupling model based on slip interface theory, which describes the gradual transition of the interface from well cemented bonding to complete debonding. Simulation results show that combining the casing coupling model with the classical model allows for modeling the amplitude changes in the acoustic-wave measurement from well bonded to free casing conditions. Based on the theoretical modeling, we introduce a method to quantitatively evaluate cementing quality at the casing-cement interface using first-arrival wave data. To validate the effectiveness of slip interface theory for assessing cementing quality, we conducted verification tests on two experimental wells with varying uncemented sector angles (USA) and cement densities. The results reveal that the shear coupling stiffness curve is sensitive to small USA but relatively unaffected by changes in cement density. Finally, we applied slip interface theory to interpret field acoustic data. Compared to conventional methods, the shear coupling stiffness curve from the slip interface model is more effective in identifying small USA or channeling issues.

ObjectiveDuring the exploration and development of fractured-vuggy carbonate reservoirs, acoustic reflection imaging logging technology enables the detection of fractured-vuggy bodies around wells. However, this technology suffers from unclear response characteristics and a strong multiplicity of solutions. Focusing on the Tarim Oilfield, this study aims to establish a tailored interpretation method by integrating geological and geophysical characteristics. The purpose is to enhance the identification accuracy and reliability of fractured-vuggy bodies in these reservoirs. MethodsBased on the distribution characteristics of fractured-vuggy carbonate reservoirs in the Tarim Oilfield, this study established the numerical simulation models of three typical fractured-vuggy bodies: inter-breccia porous, fault-cavity, and tectonic-fracture types. Then, the imaging results of acoustic reflection imaging logging data were optimized using Hilbert transform-based envelope extraction and vertical constraint-based data reconstruction technique. Using the finite-difference numerical algorithm, this study simulated the theoretical reflection wavefields of various fractured-vuggy bodies. Then, using the optimized procedure for the imaging results of acoustic reflection imaging logging data, this study processed the migration imaging results of different types of fractured-vuggy bodies and summarized their typical characteristics. Based on typical theoretical response characteristics, the imaging results of acoustic reflection imaging logging data can be interpreted. Results and ConclusionsThe imaging results of acoustic reflection imaging logging data, obtained using the optimized procedure, exhibited clearer features. Compared to the imaging response characteristics of forward modeling, the imaging results of acoustic reflection imaging logging data revealed the presence of three sets of reflectors in the target interval. These reflectors were characterized by clear and distinct imaging results, mutual independence, and the occurrence of arc-like pseudomorph caused by adjacent reflectors. These characteristics aligned with the imaging response characteristics of fault-cavity-type fractures, demonstrating that these extra-well reflectors at this well interval are of the fault-cavity type. The interpretation results corresponded well with conclusions from well tests, confirming the presence of high-productivity fault-cavity-type fractured-vuggy reservoirs at this interval. The processing results of actual data validated the reliability of the identification of fractured-vuggy carbonate bodies based on theoretical response characteristics. This method provides robust technical support for the precise identification and interpretation of near-well fractured-vuggy carbonate bodies in the Tarim Basin.

The increasing use of cluster wells in recent oil and gas drilling has raised concerns about well collisions. Although the successful application of single-well acoustic reflection imaging for nearby well detection has made it a potential ranging technique for collision prevention, further analyses are needed to demonstrate its utility for detecting multiple target wells around the measurement well. We develop a logging data to azimuthally scan multiple nearby wells and determine their distances and directions. To this aim, we develop a fast analytical method to model elastic wave radiation from the measurement well and wave scattering from multiple nearby wells and analyze the sensitivity and effectiveness of the 4C dipole tool in the generation of the transmitted wave and detection of the scattered waves. By means of 4C rotation and migration, the scattered wave data, as measured by the tool, are used to image the targeted wells and determine their distance and azimuth relative to the demonstrate the effectiveness of this multiple-well acoustic ranging technique.

Fractures are widely distributed in upper crustal rocks and significantly affect rock elasticity. Experiments and field studies indicate that pressure influences the rock's elastic properties. Therefore, it is critical to understand the pressure dependence of rock elastic properties. For this purpose, a theoretical model is developed that considers both pressure-dependent background elasticity and fracture deformation within the hyper-elasticity stage. Using the model, the dynamic (frequency-dependent) attenuation mechanisms of fracture-background wave-induced fluid flow (FB-WIFF), multishaped microcracks' squirt flows (MMSF), fracture elastic scattering and their coupling effects under different effective pressures are investigated. The results indicate effective pressure can greatly reduce fracture normal and shear compliances. The stiffness coefficients increase with the increasing effective pressure and the MMSF mechanism gradually disappears due to the almost completely closed microcracks. Effective pressure has a stronger effect on wave-induced fluid flow (WIFF) mechanisms (including FB-WIFF and MMSF) than the elastic scattering mechanism. The P-wave dynamic anisotropy is modulated by FB-WIFF, elastic scattering and their coupling effects, while the S-wave anisotropy is modulated only by elastic scattering. Compared to P-wave anisotropy, the S-wave anisotropy Thomsen coefficient ${\gamma _{\rm TH}}$ is almost independent of effective pressure. In addition, P-wave attenuation anisotropy is more sensitive to effective pressure than P-wave velocity anisotropy. The predicted velocities at ultrasonic frequencies were compared with previous laboratory ultrasonic velocity data under effective pressure loading to validate the model.

Due to overburden and tectonic stresses, many deep formation rocks contain vertically oriented fractures, resulting in significant seismic S-wave anisotropy and attenuation, as commonly observed from borehole acoustic anisotropy measurements in deep wells. Understanding the fracture-induced phenomena is of theoretical interest and practical importance. By applying a fracture scattering theory to a fluid-saturated rock containing vertical fractures, we investigate the fracture-induced effects on dynamic S-wave moduli, anisotropy, and attenuation anisotropy in the rock medium. This includes the effects of fracture-background wave-induced fluid flow and elastic scattering on the fractured rock's stiffness. The results indicate that fracture elastic scattering has a significant influence on the S-wave moduli and attenuation anisotropy. Compared with the velocity response of elastic scattering, Rayleigh scattering increases the S-wave anisotropy, whereas Mie scattering decreases it. The elastic scattering-dominated S-wave anisotropy and attenuation anisotropy are significantly affected by fracture size (diameter) and density. We use the theoretical results to interpret field acoustic anisotropy and attenuation anisotropy data acquired in a deep well, wherein the theory and measured data are in good agreement.

Fluid-filled dual-string pipe systems are widely utilized in the petroleum industry. Taking the example of a concentric structure composed of production tubing and cemented casing strings, the bonding integrity of the outer string plays a critical role in ensuring safe extraction operations. This study conducts theoretical modeling and acoustic simulations of flexural mode guided waves in dual-cased wells, with a focus on comparing modal responses under varying media conditions in the B annulus of the outer casing. Theoretical and experimental results demonstrate that when the B annulus is fluid-filled, significantly more flexural modes (particularly higher-order modes) are excited compared to cemented scenarios. These higher-order flexural modes can be effectively utilized as acoustic indicators to evaluate annular bonding quality. The findings provide a theoretical foundation for non-invasive integrity evaluation in multi-string well systems.