Sonic Logs Research Articles

An Integrated Machine Learning Workflow to Estimate In Situ Stresses Based on Downhole Sonic Logs and Laboratory Triaxial Ultrasonic Velocity Data

AbstractThe optimum performance of various subsurface operations such as stimulation treatments, wellbore drilling, horizontal well placement, underground mining, and tunneling rely on accurate estimation of the in situ stresses. This study presents an integrated machine learning (ML) workflow for the reliable determination of the in situ stress in subsurface rock formations. The study workflow was completed in three phases. In the first phase, six supervised ML regression techniques were employed to develop the stress prediction models using laboratory true triaxial ultrasonic velocity tests (labTUV) data of subsurface rock samples retrieved from well 16A(78)‐32 located at the Utah FORGE geothermal site. In the second phase, subsurface geological formations were classified into rock facies using an unsupervised K‐means clustering algorithm. Finally, in the third phase, the optimized ML models were employed for predicting the in situ stresses in the corresponding rock facies in the well 16A(78)‐32 using field sonic logs. A comparison of evaluation metrics revealed the superior performance of ANN models for horizontal and vertical stress predictions with root mean squared error of 1.5, 0.6, and 1.7 MPa, and determination coefficient (R2) of 0.98, 0.98, and 0.96, for the testing/validation phases, respectively. The generalization capability of ML models was explored by uncovering the underlying physics. The mathematical expressions of constitutive relations were extracted from three ANN models. Further, a total of five rock facies were identified in the subsurface geological formations. The novel workflow would be capable of delivering reliable in situ stress profiles in subsurface geological formations without performing expensive field tests.

Field test of internal visualization of deep-water pile foundation using ultrasonic waves

Early-age internal visualization is critical for timely assessing the quality of pile foundations, especially as super cross-sea bridges increase the diameter and length of pile foundations beyond the capabilities of traditional cross-hole sonic logging (CSL). This article introduces two innovations. First, ultrasonic instruments are specially developed to transmit waves through a concrete pile with a diameter of 4 m, a length of 92 m, and a water depth of 63 m, before the full hardening of concrete. Second, an imaging method is proposed to upgrade traditional CSL from 1D detection to 3D visualization. Both large-scale experiments and field tests are tested in the first 7 days after the concrete casting. The experiments successfully visualize typical construction defects, achieving a detection resolution of 18 cm for layered defects and 15 cm for void defects. During the field test, a low-pixel region is observed at the pile head, extending approximately 2.1 m deep, attributed to mud floating up during concrete casting into the steel casing. These developments in instrument technology and visualization methods provide essential tools for the quality evaluation of super pile foundations. Furthermore, the visualization data can support digital twin modeling in subsequent stages of construction and maintenance planning.

A comprehensive crosshole seismic experiment in glacial sediments at the ICDP DOVE site in the Tannwald Basin

Abstract. Glaciers have shaped the Alpine landscape by carving deep valleys and depositing sediments to form overdeepened basins. Understanding these processes provides information on the evolution of the climate and landscape. One such overdeepened structure is the Tannwald Basin (ICDP site 5068_1) north of Lake Constance, which was formed by the Rhine Glacier in several glacial cycles. In order to study these sediments and their seismic properties down to about 160 m depth, we conducted seismic crosshole experiments between three boreholes, obtaining compressional (P) wave data. The P-wave data are generated by a sparker source and recorded by a 24-station hydrophone string. We present the data acquisition and review our approach for future optimization, suggesting a finer time sampling interval and a separate registration of borehole and surface receivers. Travel-time tomography of the P-wave first-arrival picks under geostatistical constraints yields initial subsurface models. The tomograms correlate well with cased-hole sonic logs and the lithology derived from the core of one of the boreholes. These results will be further investigated in future research, which will include full-waveform inversion (FWI) to obtain high-resolution subsurface models.

Using the Conventional Method to Estimate Water Saturation of Khasib Formation in East Baghdad Oil Field Central Area

The water saturation level may change due to geological changes or human intervention, like earthquakes, excavation, injection, or extraction of fluids from underground reservoirs. This work helps the reader understand how to use electrical logging to acquire a suitable water saturation value in resistivity and dielectric studies. In the East Baghdad field, two wells that were drilled into the Khasib Formation had their petrophysical parameters assessed using IP software. This allowed for the determination of water saturation using the available well logs, including neutron, density, SP, sonic, gamma ray, and resistivity logs for wells EB-25, and EB-29. The limestone section in the Khasib formation represents calcite, the main mineral found. We counted the amount of shale in each well using gamma-ray logs. The porosity at the reservoir unit was determined using a sonic log. Archie's equation and resistivity data are also used to compute fluid saturation. Taking m=n=2, a=1, x=1.76, and Rw=0.033 for Khasib formation. The link between Archie's and Raiga's coefficients, a, m, n, and x, is primarily the basis for this discussion. There is high agreement when this equation is used. In clean formations (non-shaly). this formula can be regarded as a reliable indicator of water saturation in situations where porosity techniques other than acoustic logs are unavailable. It's important to note that this equation provides accurate and realistic answers for rocks with transit times between 40 and 150 As/ft. Porosity knowledge is not required for such an assessment.

Pattern recognition approach (PRA) for identifying oil reservoir lithology of Camaal oil field, Yemen

The accurate determination of reservoir lithology remains a challenge in petroleum engineering. There are some conventional techniques available to determine the lithology. However, the application of those techniques has been long and complex. So, the main goal of this study is to simplify the identification of reservoir lithology. This paper presents a Pattern Recognition Approach (PRA) to identify the reservoir lithology simply and accurately. It is type of artificial neural network. Four wells from the Camaal Field were chosen to develop this approach. Around 32400 data points from the previous wells were digitized. The PRA approach used depth, gamma ray, lithology, sonic, neutron, and density logs as inputs. The model classifies lithology into permeable and impermeable rocks, further categorizing them into clastic and carbonate rocks, and subsequently into specific types into sand, sandstone, dolomite and limestone. The results show that the proposed approach provides a suitable prediction of lithology with higher accuracy compared with actual lithology. The model demonstrates high accuracy rates in identifying various lithologies, with overall accuracies of 76.2% for permeable/impermeable rocks, 94.9 for clastic/carbonate rocks, 86.2% for sand/sandstone, and 92.8% for dolomite/limestone.

Three-Dimensional Coupled Temporal Geomechanical Model for Fault-Reactivation and Surface-Deformation Evaluation during Reservoir Depletion and CO2 Sequestration, Securing Long-Term Reservoir Sustainability

Assessing reservoir subsidence due to depletion involves understanding the geological and geophysical processes that lead to ground subsidence as a result of reservoir fluid extraction. Subsidence is a gradual sinking or settling of the Earth’s surface, and it can occur when hydrocarbons are extracted from underground reservoirs. In this study, a time-integrated 3D coupled geomechanical modeling incorporating the fourth dimension—time—into traditional 3D geomechanical models has been constructed utilizing seismic inversion volumes and a one-dimensional mechanical Earth model (1D MEM). The 3D geomechanical model was calibrated to the 1D MEM results. Geomechanical rock properties were derived from the density and sonic log data that was distributed with conditioning to the seismic inversion volumes obtained from running pre-stack inversion. The standard elastic parameter equations were used to generate estimates of the elastic moduli. These properties are dynamic but have been converted to static values using additional equations used in the 1D MEM study. This included estimating the Unconfined Compressive Strength. In situ stresses were matched using different minimum horizontal principal stress gradients and horizontal principal stress ratios. The match is good except where the weak carbonate faults are close to the wells, where the Shmin magnitudes tend to decrease. The SHmax orientations were assessed from image log data and indicated to be 110° in the reservoir section. A time-integrated 3D coupled simulation was created using the finite-element method (FEM). The effective stresses increase while there is depletion in all directions, especially in the Z direction. The predicted compaction in the reservoir and overburden was 350 mm. Most of the compaction occurs at the reservoir level and dissipates towards the surface (seabed). Furthermore, the case displayed no shear failure that might cause or fault reactivation in the reservoir interval (Kangan–Dalan Formations) located in the simulated area. In this study, we applied an integrated and comprehensive geomechanical approach to evaluate subsidence, fault reactivation and stress alteration, while reservoir depletion was assessed using seismic inversion, well logs, and experiment data. The deformation monitoring of geological reservoirs, whether for gas storage or hazardous gas disposal, is essential due to the economic value of the stored assets and the hazardous nature of the disposed materials. This monitoring is vital for ensuring the sustainability of the reservoir by maintaining operational success and detecting integrity issues.

Petrophysical identification and characterization of Cenozoic hydrocarbon reservoir zones at the R field, northern Rovuma Basin, Tanzania

ABSTRACT Cenozoic reservoirs of northern Rovuma Basin, Tanzania, have not yet been characterized in detail. Combined well log patterns and fracture indicators have not been presented before. This work used petrophysical analysis of density, neutron, gamma ray (GR), resistivity and sonic well logs to characterize reservoir zones at the study area. Twelve reservoir zones (A–L) have been identified based on lithological properties and resistivity anomalies. The presence of hydrocarbons in these zones is concluded based on acoustic parameters involving acoustic impedance and V p / V s ratio, and ∆TC and neutron porosity crossplots. Fractured sedimentary fills in different stratigraphic levels were assessed based on resistivity-logs crossplots. Results show that eleven zones (A–F and H–L) are gas-filled and zone G is devoid of hydrocarbon. We have shown that, in areas with resistivity anomaly, all GR-resistivity combined patterns may be used to indicate hydrocarbon bearing layers except when there is a co-occurrence of funnel-shaped pattern for both logs. Distribution of fractures and hydrocarbons suggest the existence of compartmentalized reservoirs. These reservoirs have been discussed based on a geoseismic model created to support our arguments. Our technique has allowed easier identification of hydrocarbon bearing zones through well Rx. This technique may be applied in other sedimentary basins.

Sonic P-wave attenuation based on fluid pressure differential between joint and matrix

When P-wave compresses in-situ rocks saturated with fluid, there is a fluid pressure differential between compliant joints and stiff matrix. The pressure differential automatically drives a mesoscale squirt between joint wall and matrix interior. This mechanism is likely to dominate sonic P-wave attenuation in the field. A new model of P-wave attenuation based on the mesoscale squirt is established, in which the first porosity is represented by matrix pores while the second porosity is represented by joint. Sonic log of a sandstone reservoir in Saudi Arabia is used to illustrate the model. Both estimated phase velocity and measured quality factor of sonic P-wave are accurately simulated. The results show that for the sandstone reservoir, the relative second porosity is 5%, inter-joint distance is 0.1 m, squirt permeability at very low frequencies is about half of Darcy permeability of the matrix, and joint gap is 800 μm (quadruple grain diameter of 200 μm). Surprisingly, the model of joint-matrix squirt is also capable of reproducing seismic P-wave attenuation measured in the shallow crust of southern California. Therefore, mesoscale squirt across the joint wall is probably the underlying mechanism of attenuation for sonic and seismic P-waves shallower than 10 km.

A Model-Based Systematic Innovative Design for Sonic Logging Instruments in Natural Gas Wells.

With the continued development of natural gas extraction technologies, the accurate determination of downhole temperature and pressure has become increasingly important. It is crucial for the optimization of gas well production and an important measure to prevent accidents. However, existing logging instruments have a series of deficiencies in measurement and cannot adequately monitor modern natural gas. In response to these problems, in this paper, we propose a new model-based systematic innovation design method for designing logging instruments and simulations using finite element software. Our research results confirm the theoretical and practical utility of this model-based design method and provide a novel approach to designing logging instruments.

Drilling data-calibrated shale compaction models for pore pressure evaluation from geophysical well logs in the North Alpine Foreland Basin, SE Germany

This paper combines quality-controlled pore pressure measurements and indicators with geophysical well logs from 310 deep wells to evaluate the pore pressure distribution in the overpressured North Alpine Foreland Basin in SE Germany. Previous studies relied chiefly on low-resolution check shots and vertical seismic profiles calibrated to only a few pore pressure measurements. The current study, based on sonic and electrical resistivity logs, indicates that more than one shale normal compaction trend is required to appropriately assess pore pressure from geophysical well logs. Thereby, the necessary adjustments of the derived compaction trends apply to both sonic and resistivity data, which possibly reflect a gradual change in the mineralogical composition with decreasing clay content or progressing cementation from north to south or an increase in horizontal loading-driven compaction towards the North Alpine Thrust Front. The newly derived compaction-based pore pressure distribution is combined with drilling data-based pore pressure evaluations and offers a comprehensive update of previous pore pressure investigations in the North Alpine Foreland Basin. Our study is therefore of key importance for optimized well planning and cost-effective drilling in the course of the recent expansion of deep geothermal energy exploration in the overpressured North Alpine Foreland Basin.

Estimation of Porosity for Sedimentary Rocks and Volcanic Rocks from Sonic Log Data in the Futagawa Fault Drilling

Estimation of Porosity for Sedimentary Rocks and Volcanic Rocks from Sonic Log Data in the Futagawa Fault Drilling

Advancing Reservoir Evaluation: Machine Learning Approaches for Predicting Porosity Curves

Porosity assessment is a vital component for reservoir evaluation in the oil and gas sector, and with technological advancement, reliance on conventional methods has decreased. In this regard, this research aims to reduce reliance on well logging, purposing successive machine learning (ML) techniques for precise porosity measurement. So, this research examines the prediction of the porosity curves in the Sui main and Sui upper limestone reservoir, utilizing ML approaches such as an artificial neural networks (ANN) and fuzzy logic (FL). Thus, the input dataset of this research includes gamma ray (GR), neutron porosity (NPHI), density (RHOB), and sonic (DT) logs amongst five drilled wells located in the Qadirpur gas field. The ANN model was trained using the backpropagation algorithm. For the FL model, ten bins were utilized, and Gaussian-shaped membership functions were chosen for ideal correspondence with the geophysical log dataset. The closeness of fit (C-fit) values for the ANN ranged from 91% to 98%, while the FL model exhibited variability from 90% to 95% throughout the wells. In addition, a similar dataset was used to evaluate multiple linear regression (MLR) for comparative analysis. The ANN and FL models achieved robust performance as compared to MLR, with R2 values of 0.955 (FL) and 0.988 (ANN) compared to 0.94 (MLR). The outcomes indicate that FL and ANN exceed MLR in predicting the porosity curve. Moreover, the significant R2 values and lowest root mean square error (RMSE) values support the potency of these advanced approaches. This research emphasizes the authenticity of FL and ANN in predicting the porosity curve. Thus, these techniques not only enhance natural resource exploitation within the region but also hold broader potential for worldwide applications in reservoir assessment.

Implication of fast shear wave azimuth and critically stressed fractures in a coalbed methane reservoir

The orientation of the fractures and stresses within coal seams plays a critical role in gas production from coalbed methane reservoirs. In this study, we used resistivity images and sonic logs to investigate these parameters, aiming to (1) establish the relationship between fracture orientations and the polarization angles of fast shear waves and (2) detect active fractures in the coal seam. Shear waves in anisotropic formations are split into fast and slow components, with Alford’s rotation method used to determine the polarization angles of the fast shear wave. We found that the fast shear wave aligns with the direction of higher fracture intensity. Subsequently, we incorporated the poroelastic strain model to estimate the vertical ([Formula: see text]), maximum horizontal ([Formula: see text]), and minimum horizontal ([Formula: see text]) stresses in the wellbore. These stress magnitudes aided in identifying the faulting regime and were corroborated by vertical opening mode fractures. Validation of [Formula: see text] and [Formula: see text] involved comparisons with the breakout width in image logs and the closure pressure observed during hydraulic fracturing treatments. Applying Mohr’s Coulomb criteria, the stress model discerned the state of the fractures, transforming the stress magnitudes into shear and effective normal stress on each fracture plane. Our observations indicated that the identified fractures existed in a noncritically stressed condition, suggesting a lack of interconnectivity among them. These findings correspond to the absence of gas production to date, providing insights into the dynamics of fractures and their impact on production behavior.

Comparative analysis of bulk and velocity-derived density data in deep wells in the Adriatic Region (Italy)

This work is based on a systematic comparison between two datasets related to density data of geological formations crossed in 13 deep wells. The wells are located in the Adriatic area and reach depths of over 5000 meters. The main lithologies involved include sandstones, marls, clays, evaporite, and carbonate rocks. The first dataset concerns density values obtained from the analysis of sonic logs recorded along the wells, by applying the Gardner relation. The second dataset, on the other hand, refers to actually acquired density log measurements. Differences among the values in the two datasets have been calculated. The aim behind this work is to assess the reliability of rock densities estimated using the Gardner equation by comparing them to measurements obtained through density logs, despite many factors influencing the log density. This comparison of the densities obtained from various lithologies and geological formations leads us to draw, in this region, some initial considerations regarding the applicability and accuracy of Gardner formula, usually considered as standard reference, since the density log is generally available only within the reservoir. In the area we analyzed it is observed that the density values estimated from sonic velocities are underestimated for the Plio-Quaternary formations characterized by clayey-sandy lithologies by at least 0.1 g/cm3; whereas, densities of the carbonate sequences are overestimated by the same extent. Noteworthy, the density estimates deviate from the real values especially for gypsum units, overestimating by a factor of approximately 0.3 g/cm3. The results we obtained emphasize differences in density values when using the Gardner formula and suggest the need for taking into account the possible errors in the specific geological context or instead lithologies, such as those explored in this study.

Estimation of pore-type distribution utilizing petrophysical data and rock physics modeling on an Iranian carbonate reservoir

Pore types in carbonate reservoirs are more complex than their sandstone counterparts due to the wide spectrum of their depositional environments and their more complicated post-depositional processes. This means that a good knowledge of pore types is vital in determining carbonate formation’s elastic and reservoir properties. This study aims to develop a multi-physics approach to determine pore-type variations in a carbonate reservoir using well log information from one of the oilfields of the Abadan Plain in southwest Iran. Firstly, we determined lithology, porosity, and fluid content by interpreting conventional well logs (gamma ray, resistivity, density, neutron, photoelectric, and P-wave sonic). Then, nuclear magnetic resonance data were used to determine different pore types within the Main Ilam and Sarvak formations. We distinguished between clay, micro, meso, and macro pore types. We confirmed our interpretation results using thin sections, scanning electron microscopy photographs, and pore-size distribution on the available core plugs. Finally, a carbonate rock physics model was employed to model sonic velocities using petrophysical interpretation along with pore-type determination results. A good match between modeled and measured sonic velocities confirmed that using nuclear magnetic resonance data for pore-type determination can reasonably estimate pore-type variations needed for rock physics modeling. The standard industry procedure for carbonate rock physics modeling uses sonic logs, core data, or thin sections to determine pore types. We offer a substitute approach with reasonable accuracy for pore-type modeling needed for carbonate rock physics modeling. We modeled pore types independently from sonic velocity and used them to predict P-wave velocity with a correlation coefficient of 92 and 64 percent accuracy in the Main Ilam and Sarvak formations.Graphical abstract

Inversion of shear wave slowness and attenuation from dipole acoustic logging data based on an equivalent tool theory

The presence of a sonic logging tool affects the propagation characteristics of dipole-flexural waves in a fluid-filled borehole. As a result, the tool effect should be considered during the processing and inversion of the dipole-flexural waves for the acoustic properties of a shear wave (S wave). We show that the presence of the tool not only changes the slowness dispersion of the flexural wave but also results in a shift of its frequency-dependent attenuation toward lower frequencies. The equivalent tool theory (ETT) can realistically model the effect of an acoustic tool on guided wave propagation. With the ETT framework, we develop a unified workflow for estimating the S-wave slowness and attenuation from dipole array acoustic waveforms. The workflow is an extension of the model-based dispersive processing of dipole logging data. Our workflow involves two inversion procedures. By matching the actual dispersion data with the theoretical dispersion predicted by the ETT, we can invert the S-wave slowness and tool parameters. With the estimated S-wave slowness and other borehole parameters, we further calculate the partition coefficients of flexural waves using the ETT. Based on the concept of energy partitioning, we formulate a linear optimization problem and invert for the S-wave attenuation from the frequency-dependent dipole attenuation. We use synthetic and field data to demonstrate the effectiveness and applicability of the developed method. The results show that it is important to incorporate the tool effect when processing dipole acoustic logging data for S-wave slowness and attenuation, especially in a fast formation.

The spatio-temporal variation of exhumation and its impacts on the resource distribution in the Northern Carnarvon Basin, Australia

Deformation and rock fracture during exhumation can enhance rock porosity and permeability, thereby improving fluid migration through reservoirs and destroying seal integrity in caprocks. Therefore, analysing the spatio-temporal distribution of exhumation in connection to fluid migration models can be crucial for energy resource exploration and storage. Geological and geophysical evidence suggests that parts of the Northern Carnarvon Basin (NCB) were exhumed. However, the spatio-temporal distribution of this exhumation and its impacts on the region's resource distribution is yet to be established. We analysed porosity, sonic logs and paleotemperature data from 210 boreholes using compaction and vitrinite reflectance analysis techniques to constrain the spatio-temporal distribution of exhumation and its correlation to the region's resource distribution. We discovered that multiple exhumation phases occurred since the Late Triassic. The highest exhumation occurred in the Mid Jurassic, where approximately 1000 m of sediment thickness was eroded from the western and southwestern parts of the Exmouth Plateau. Exhumation at this time spatially correlates with magmatic intrusions in the NCB, and paleotemperature profiles indicate that sediments in the exhumed regions were partly heated by increased paleo-basal heat flow, suggesting that magmatic bodies through underplating and crustal intrusions were the primary causes of this exhumation. Another significant exhumation phase occurred in the Early Cretaceous, removing ∼800 m of sediments from the Exmouth, Barrow and Dampier Sub-basins. Depth-dependent extension events and magmatic intrusions possibly caused exhumation at this time. Significant exhumation ceased in the Early Cretaceous, and our study has established that the exhumation episodes in the NCB were part of large-scale regional events that affected the Australian Northwest Shelf. Combining results from different estimation techniques increased the spatio-temporal coverage of the analysis and improved the precision and accuracy of exhumation results.

Geopolymer Implemented in Primary Casing Cementing

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper OTC 32218, “First Global Implementation of Geopolymer in Primary Casing Cementing,” by Mark Meade, SPE, Yeukayi Nenjerama, SPE, and Chris Parton, SPE, SLB, et al. The paper has not been peer reviewed. Copyright 2023 Offshore Technology Conference. _ Portland cements are integral components in the oilfield well-construction process; however, the confluence of various business drivers have created the need to find sustainable alternative materials. The complete paper presents the evaluation of geopolymer cementing for use in oil and gas wells, specifically the primary cementing of liner strings in the Permian Basin. Geopolymer cementing offers a unique opportunity for the oilfield industry to decrease CO2 emissions related to well construction and reduce dependence on the constrained supply of Portland cements. Introduction Manufacturing Portland cement emits approximately 1 ton of CO2 for every ton of cement produced. Additionally, the manufacturing of Portland cement is a complex process requiring specialized equipment. Thus, during spikes in demand of Portland cement, the supply of cement cannot be rapidly increased to meet demand. Geopolymers are slurries that set to become a hard, durable solid that can withstand stresses and strains and are resistant to corrosion. Moreover, geopolymers can be made from a broad range of aluminosilicate raw materials sourced from waste products from other industries, mined materials, and biowaste. The low manufacturing and processing requirements reduce the carbon footprint of geopolymers to a fraction of that of manufacturing Portland cement. Whereas Portland cement chemistry is driven by hydration of calcium aluminates and calcium silicates, geopolymer chemistry is based on the polymerization of aluminosilicates initiated by activators. The creation of geopolymers starts with dissolution of the aluminosilicate raw material into monomers by an increase in the pH of the fluid from the activator. Then, the activator provides a site for covalently bonded chains to form, and these chains undergo polycondensation to form the set 3D network of polymers. Implementation of this new material with such a different chemistry into the complex oilfield-construction process presents challenges that must be overcome. Challenges Existing geopolymer formulations typically are characterized as either two-step or one-step geopolymers. In two-step geopolymers, the activators are added into the mix fluid, whereas, in one-step geopolymers, the activators are dry blended with the aluminosilicate source. Implementation of two-step geopolymers would come with operational and logistical constraints impairing their scalability, along with increased service quality and health, safety, and environmental (HSE) risks. On the other hand, common existing one-step geopolymers are formulated with activator types that dissolve rapidly and disperse in the slurry. Therefore, development of novel chemistries for one-step geopolymers are required to introduce geopolymers into oilfield well construction. Such applications expose geopolymers to temperatures and pressures significantly above ambient for extended periods of time during slurry placement within the wellbore. To ensure quality assurance of job placement, hydraulic simulators should be reliable predictors of the top of the placement, and geopolymers set within the wellbore must be detectable using established sonic logging tools.

Volve Oil Field S-Wave Log Data Prediction Using GBR and MLPR

The shear wave sonic (S-wave) log data is essential for identifying the reservoir's geomechanical properties, which is an important factor for the drilling, completion, and optimization processes, where obtaining S-wave requires capital investment and reduces the cost. In this paper, we used the multi-layer perceptron (MLPR) and the gradient boosting regression (GBR) to predict the missing S-Wave log data for the Volve Oil Field in the North Sea. Prescriptive and predictive analysis were carried out in series to achieve a high blind data accuracy rate of 0.943 and 0.982 for the multi-layer perceptron regression and the gradient boosting regression, respectively. It was observed that the gradient-boosting algorithm achieved higher accuracy than the MLPR algorithm for the limited dataset. It was also found that hierarchical clustering can reveal information regarding the feature's importance similar to the relevant AI tools, which makes hierarchical clustering a faster tool in eliminating the nonimportant inputs from the dataset while the use of artificial intelligence tools showed a significant effect in predicting the missing values of the sonic wave log and the neutron porosity log in an efficient way by selecting the relevant important features.

Basin‐scale prediction of S‐wave Sonic Logs using Machine Learning techniques from conventional logs

AbstractS‐wave velocity plays a crucial role in various applications but often remains unavailable in vintage wells. To address this practical challenge, we propose a machine learning framework utilizing an enhanced bidirectional long short‐term memory algorithm for estimating S‐wave sonic logs from conventional logs, including P‐wave sonic, gamma ray, total porosity, and bulk density. These input logs are selected based on traditional rock physics models, integrating geological and geophysical relations existing in the data. Our study, encompassing 34 wells across diverse formations in the Delaware Basin, Texas, demonstrates the superiority of machine learning models over traditional methods like Greenberg–Castagna equations, without prior geological and geophysical information. Among these machine learning models, the enhanced bidirectional long short‐term memory model with self‐attention yields the highest performance, achieving an R‐squared value of 0.81. Blind tests on five wells without prior geologic information validate the reliability of our approach. The estimated S‐wave velocity values enable the creation of a basin‐scale S‐wave velocity model through interpolation and extrapolation of these prediction models. Additionally, the bidirectional long short‐term memory model excels not only in predicting S‐wave velocity but also in estimating S‐wave reflectivity for seismic amplitude variation with offset applications in exploration seismology. In conclusion, these S‐wave velocity estimates facilitate the prediction of further elastic properties, aiding in the comprehension of petrophysical and geomechanical property variations within the basin and enhancing earthquake hypocentral depth estimation.