|
|
- Analytic solutions and field-scale application for verification of coupled
thermo-hydro-mechanical processes in subsurface fractured media-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: We provide a suite of verification examples for coupled thermo-hydro-mechanical (THM) processes in subsurface flow. The considered scenarios are subsets of THM problems with and without fractures. The examples include classical problems presented by Terzaghi, Schiffman, Mandel, Wijesinghe, McNamee and Gibson, Noda and Lauwerier. For each example, we provide a description of the physical problem setup, the governing equations, the solution to the equation, and comparison of that solution with a new THM simulator. Python scripts for the solutions are available in a Git repository for other modeling groups to verify their own THM simulators. We demonstrate this new verified simulator’s full THM capabilities by modeling a hypothetical enhanced geothermal system to show how dynamically modifying fracture aperture impacts energy production over a period of ten years. PubDate: 2025-05-21
- Efficient multi-scale image reconstruction of heterogeneous rocks with
unresolved porosity using octree structures-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Identifying rock properties at the pore scale plays a crucial role in understanding larger-scale properties. For this purpose, the digital rock physics technique is used to model rock images at the pore scale. Achieving high-resolution (HR) images with a large field of view (FoV) is essential for pore-scale modeling of heterogeneous rock samples, which presents significant challenges due to their complex structures. However, because of the trade-off between resolution and FoV, it is not possible to acquire large HR images. Multi-scale image reconstruction methods enable modeling images at different resolutions and FoVs. Despite various approaches being introduced, a common limitation is the high computational cost. In this study, a novel approach based on Octree structures is introduced to minimize computational cost while maintaining accuracy. A Berea sandstone (BS) and an Edward Brown Carbonate (EBC) sample were scanned at both HR and low resolution (LR) using X-ray microtomography. Our method involves splitting the unresolved porosity in rock images into smaller sections of unresolved templates using the watershed algorithm and considering the optimized parameters. We then applied a cross-correlation based simulation technique to find the best match of each unresolved template. The novelty of our approach lies in the use of an Octree structure to perform calculations on LR images, significantly reducing computation time and memory consumption due to the fewer number of pixels in Octree LR structures. The accuracy of the images thus reconstructed using our approach was compared with those from previous methods by evaluating geometric properties and single- and two-phase flow properties. The results were promising, demonstrating that our approach achieved a permeability close to the real value, while the previous method had an error of approximately 4% for both BS and EBC rocks. More importantly, our approach was approximately three times faster and reduced memory usage by 20 to 130 times. The findings of this study facilitate dual- or multi-scale modeling and evaluate heterogeneous rock images at a significantly lower computational cost. In particular, for heterogeneous rocks, where multi-scale image reconstruction demands substantial memory and runtime, the use of the Octree technique enables accurate reconstruction with lower computational cost. PubDate: 2025-05-03
- Influence of fluid dynamics on flow and transport in natural fracture
networks-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: The flow velocity in metre-scale natural fracture networks readily exceeds centimetres per second, the threshold for non-stationary flow. However, despite widespread evidence of such dynamics, these are rarely considered in subsurface engineering applications, where steady-state simulation approaches dominate. Here, we compare Reynolds-averaged Navier-Stokes (RANS) and Detached-Eddy Simulation (DES) methods for the transient Navier-Stokes equation applied to fracture flow. These models are validated with experimental data of flow through fracture intersections. DES is then applied to a metre-scale fracture pattern with hundreds of discrete fractures, examining flow dynamics at velocities up to metres per second (m/s). DES accurately captures the temporal flow fluctuation and multiscale eddy formation, especially when a fine computational mesh is used in wake regions. By contrast, unsteady RANS fails to capture flow-field variations and produces results similar to steady RANS. DES reveals significant network flow periodicity ($$\sim $$40 Hz) at m/s velocities, unlike the low-frequency results ($$\sim $$0.4 Hz) from RANS. We also explore the impact of unsteady flow on particle transport by integrating mixture-multiphase and rheological models into DES. Corresponding results indicate that inertia alters the concentration of transported solids, mixture viscosity, and particle dynamics such as clustering. PubDate: 2025-04-28
- Damage evolution characteristics of natural fractures during hydraulic
fracturing process: A peridynamic hydro-mechanical coupling model-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Hydraulic fracturing is widely employed in the exploitation of underground resources, and the variation in natural fracture width plays an important role in the effectiveness of hydraulic fracturing. In this work, a Hydro-Mechanical Coupling Model (HM) was primarily established to explore the influence of fracture width, fluid viscosities, and fracture angle on natural fracture transformation based on the Peridynamics (PD) method. Then, the accuracy of the PD model was validated by comparing the numerical simulation results with analytical solutions. Ultimately, a series of numerical models were established to numerically investigate the influence of fracture widths, fluid viscosity, and fracture angles on the development of natural fracture. Results show that a larger initial natural fracture width or a decrease in fluid viscosity can increase the pore water pressure in the natural fracture, which in turn leads to a better natural fracture modification. The initial width of the natural fracture and the fluid viscosity directly influence the rate of variation in natural fracture width. The angle of natural fracture has a relatively small impact on the pore water pressure in fractures, while a greater impact on the displacement field surrounding the fracture. PubDate: 2025-04-28
- Lightweight permeability prediction of digital rocks by merging 3D
depthwise separable convolution with efficient multiscale attention-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Permeability is a crucial parameter in geology and petroleum engineering. Traditional methods for measuring the permeability of rocks primarily rely on physical experiments, which can easily damage the integrity of rocks, thus affecting subsequent analyses. Current approaches are mainly based on digital rock images using direct numerical simulation methods or deep learning methods to obtain permeability. Although these methods avoid destroying rocks, there are challenges of high computational complexity and capturing complex features on the microstructure for different rocks. To address these challenges, we design a novel network for the lightweight and efficient prediction of permeability in 3D porous media. We made our novel design in two steps: First, to reduce the computational burden, we introduced three-dimensional depthwise separable convolution into the network. Second, to enhance the capture of micro-features in digital rock, we incorporate an efficient multi-scale attention mechanism into the network. Finally, we propose the 3D-EmaSepNet. Experimental results demonstrate that 3D-EmaSepNet performs well on different rock types, such as sandstones and carbonates. Specifically, compared to traditional three-dimensional convolution, the computational cost of 3D-EmaSepNet using three-dimensional depthwise separable convolution is reduced by a factor of 9.6. Our 3D-EmaSepNet achieves coefficient of determination (R2 score) of 94.57% and mean squared error (MSE) loss of 0.10 on the validation dataset for sandstone, while reaching R2 score of 93.91% and MSE loss of 0.31 on the validation dataset for carbonate. These results show the potential of 3D-EmaSepNet for practical applications in geology, petroleum engineering, and other fields. PubDate: 2025-04-26
- Frequency-domain vector finite element forward modeling of 3D GPR data
using exact PML absorbing boundary conditions-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: We present a vector finite element method for 3D ground-penetrating radar forward modeling in the frequency domain using exact perfectly matched layer (EPML) absorbing boundary conditions. The corresponding edge-based finite element solution, for both hexahedral and tetrahedral meshes, is derived in detail, and the attenuation characteristics of the EPMLs are compared to their standard uniaxial counterparts. In doing so, we demonstrate the superiority of EPMLs in eliminating the non-physical reflection at the boundaries of the computational domain. Further, we demonstrate that the use of EPML absorbing boundaries effectively improves the efficiency of 3D ground-penetrating frequency-domain simulations, as they require significantly fewer layers and are less parameter-dependent than conventional uniaxial perfectly matched layers. The practical viability of the proposed simulation approach is demonstrated through its application to complex 3D models, involving pronounced topography along the air–soil interface and strong heterogeneity in the probed surficial region. PubDate: 2025-04-25
- Multi-scale channel enhanced transformer for rock thin sections
identification and sequence consistency optimization-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: The identification of rock thin sections plays a pivotal role in geological exploration, as it provides critical insights into the fundamental properties and composition of rocks. However, the accurate identification of mineral particles presents significant challenges due to three primary factors: the inherent imbalance in data distribution, misclassification caused by feature similarity, and substantial feature variations observed under different cross-polarized angles. These complexities render conventional deep-learning models with single-structure architectures inadequate for precise mineral particle identification. Therefore, this study proposes a novel rock thin section image classification methodology that combines a Multi-Scale Channel Enhanced Transformer (MSCET) with a Sequence Consistency Optimization (SCO) strategy. This integrated approach is designed to effectively extract distinctive features of mineral particles while fully exploiting the influence of polarization angle variations. The MSCET architecture synergistically combines Convolutional Neural Networks (CNN), Squeeze-and-Excitation Networks (SENet), and Transformer mechanisms to enhance the network’s feature representation capabilities. Specifically, it employs distinct convolutional operations to extract both coarse- and fine-grained features of mineral particles. The SENet and Transformer structures are then utilized to aggregate global information across both channel and spatial dimensions. Furthermore, we introduce the SCO strategy to refine low-confidence predictions, thereby mitigating the impact of feature variations in multi-angle cross-polarized images. Comprehensive experimental evaluations demonstrate the efficacy of our proposed method, achieving a classification accuracy of 92.35% on the test set. The method also shows significant improvements in key performance metrics, including recall, precision, and F1 score, substantiating its potential for robust rock thin section identification in geological applications. PubDate: 2025-04-24
- Identification of moving sources in stochastic flow fields: A bayesian
inferential approach with application to marine traffic in the mediterranean sea-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: A Bayesian inference approach for inferring the source of marine pollution released from a moving source in an uncertain flow field is proposed. A Markov Chain Monte Carlo (MCMC) algorithm is developed and applied for inferring single and multiple release events from vessels moving at known velocity along a predefined path in the Mediterranean Sea. The likelihood is based on a logistic regression cost function that measures the discrepancy between the modeled spill distribution and a binary representation of the observed images. We assess the performance of the proposed methodology using a synthetic release scenario employing realistic ocean currents to drive a stochastic Lagrangian Particle Tracking (LPT) algorithm to generate a probabilistic representation of the spill distribution. The MCMC algorithm employs an adaptive scheme to robustly ensure convergence and well-mixed chains. The proposed Bayesian framework is tested by inferring the location, or injection time, and relative contributions of single and multiple moving sources, contributing to separate and common observation patches, with a focus on various scenarios that demonstrate the efficiency of our sampling algorithm. The performance of the proposed framework was further assessed by comparing the model predictions with the most probable release parameters predicted by a global optimization algorithm. PubDate: 2025-04-10
- Multiple data assimilation as an approximate maximum a posteriori
estimator-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Data assimilation is the process of integrating observational data into a time series of state estimates. For linear-Gaussian systems, an optimal state estimate always exists. This optimal solution both minimizes mean-square error and maximizes the Bayesian posterior distribution at every time step. If the observation model is nonlinear, then the posterior distribution is no longer Gaussian; the minimum mean-square error state estimate and the maximum a posteriori (MAP) state estimate may not coincide. Still, the MAP estimate may be sought using optimization or other means. Multiple Data Assimilation (MDA) is a recursive approach that has been found empirically to produce similar state estimates to optimization-based approaches. However, convergence has been difficult to prove analytically. In this work, we shed new light on MDA by deriving it from a homotopy function. From this viewpoint, MDA is not an optimization but an approximate numerical solution to an ordinary differential equation (ODE). Given this new ODE formulation, we propose a new adaptive-step MDA algorithm inspired by the Runge Kutta integration method. The new algorithm outperforms standard approaches on a chaotic system with highly nonlinear measurements. We conclude that MDA is not an exact MAP estimator, but it can still provide a good approximation of the MAP given reasonable local linearity of the observation model. PubDate: 2025-04-08
- Three-dimensional numerical modeling of natural convection in underground
cavities connected to the surface by a well-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: We present two- and three-dimensional numerical simulations of the natural convection flow within an artificial underground cavity. We consider a realistic geometry reconstructed from measurements (3D point cloud obtained with a photogrammetric method) of the Barcq cavity existing in Normandy region, France. The numerical model solves the Navier-Stokes equations coupled with the unsteady heat equation using a finite-element method for the space discretization. Time integration is based on a semi-implicit method and the Newton method to solve the resulting non-linear equation. To implement the method, we use the free software FreeFem++ offering efficient mesh refinement/adaptivity tools. The parallelization of the simulations relies upon the PETSc software library, to which FreeFem++ offers an easy-to-use interface. We analyze the onset of the natural convection and characterize the convective cells developing within the Barcq cavity for four Rayleigh (Ra) numbers (from $$\varvec{10}^{\varvec{5}}$$ to $$\varvec{10}^{\varvec{8}}$$). Dimensionless temperatures, velocities and Nusselt numbers are presented for each Ra number. We discuss how these results can be used to understand the thermal behavior of small artificial cavities. We address in particular the influence of the value of the Rayleigh number on the flow inside the well, which is essential for designing detection methods based on surface temperature observations. The FreeFem++ numerical code used for this study is distributed as supplemental material. PubDate: 2025-04-03
- A post-processing solution to restore numerical consistency for classical
flow routing algorithms-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Water flow routing algorithms are a cornerstone of landscape evolution models (LEMs), enabling efficient simulations of complex water flow networks across varying spatial and temporal scales. Among these, multiple flow direction (MFD) and single flow direction (SFD) algorithms are widely used to compute local drainage areas, yet they suffer from mesh dependency issues that compromise their consistency. This long-standing problem has motivated various empirical corrections. Despite these efforts, the lack of a robust mathematical framework has hindered a complete resolution of these deficiencies. Building on recent findings based on general Gauckler-Manning-Strickler (GMS) water mass conservation models, this paper unifies existing MFD/SFD methodologies, including node-to-node variants. We demonstrate that all such algorithms can be corrected through a simple post-processing step, effectively eliminating anomalous grid dependency while preserving the diversity of approaches in the literature. The proposed framework bridges the gap between traditional MFD algorithms and more modern definition of the specific catchment area. It also enables a classification of classical MFD/SFD algorithms based on their flow-sharing formulas. Numerical examples illustrate the versatility and effectiveness of this correction. PubDate: 2025-04-03
- High-precision geosteering via reinforcement learning and particle filters
-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Geosteering, a key component of drilling operations, traditionally involves manual interpretation of various data sources such as well-log data. This introduces subjective biases and inconsistent procedures. Academic attempts to solve geosteering decision optimization with greedy optimization and approximate dynamic programming (ADP) showed promise but lacked adaptivity to realistic diverse scenarios. Reinforcement learning (RL) offers a solution to these challenges, facilitating optimal decision-making through reward-based iterative learning. State estimation methods, e.g., particle filter (PF), provide a complementary strategy for geosteering decision-making based on online information. We introduce RL-Estimation, a method that integrates an RL-based geosteering framework with PF to address realistic geosteering scenarios. RL-Estimation deploys PF to process real-time well-log data to estimate the location of the well relative to the stratigraphic layers, which then informs the RL-based decision-making process. Our findings indicate that RL-Estimation achieves at least 20% higher reservoir contact compared to using RL or PF alone. Additionally, RL-Estimation performs within 2% of the theoretically optimal benchmark, which assumes access to the true state instead of relying on estimates from PF. These results demonstrate the synergy between RL and PF, highlighting the method’s effectiveness and near-optimal performance in realistic geosteering scenarios. PubDate: 2025-04-01
- Ensemble-regularized Kernel density estimation with applications to the
ensemble Gaussian mixture filter-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: The ensemble Gaussian mixture filter (EnGMF) is a non-linear filter suited to data assimilation of highly non-Gaussian and non-linear models that has practical utility in the case of a small number of samples, and theoretical convergence to full Bayesian inference in the ensemble limit. We aim to increase the utility of the EnGMF by introducing an ensemble-local notion of covariance into the kernel density estimation (KDE) step for the prior distribution, regularizing the local covariances by the ensemble that generated it. We prove that in the Gaussian case, our new ensemble-regularized KDE technique is exactly the same as more traditional KDE techniques. We also show an example of a non-Gaussian distribution that can fail to be approximated by canonical KDE methods, but can be approximated well by our new KDE technique. We showcase our new KDE technique on two simple bivariate problem, showing that it has nice qualitative and quantitative properties, and improves the estimate of the prior and posterior distributions across a broad range of possibilities. We additionally show the utility of the proposed methodology for sequential filtering for the Lorenz ’63 equations, achieving a significant reduction in error, and less conservative behavior in the uncertainty estimate with respect to traditional techniques. Additional experiments on the Lorenz ’96 equations show that EnGMF type filters can converge for low amounts of samples, though without significant improvement for the approach presented in this work. PubDate: 2025-03-29
- Fluid flow analysis of discrete fracture networks using a locally
conservative stable mixed finite element method-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: This paper reports on analyses of fluid exchange between fractures in the simulation of flow through fractured porous media. Flow in the porous media is modeled with traditional Darcy’s equations and the coupling between flow in the porous media and fractures is based on the conceptual Discrete-Fracture-Matrix (DFM) representation, where the fractures are idealized as lower-dimensional elements at the interface of matrix elements. The DFNMesh algorithm is adopted to generate the Finite Element meshes and a novel methodology to handle overlapping fractures in the context of Mixed Finite Element methods is explored. Flux approximation with H(div)-conforming spaces are adopted which are particularly appealing for this analysis because of its inherent properties such as local mass conservation and strong divergence-free enforcement for incompressible flows. The analyses are carried out using simple two-fracture domains and a benchmark problem with eight fractures. PubDate: 2025-03-29
- A fully-implicit solving approach to an adaptive multi-scale model -
coupling a vertical-equilibrium and full-dimensional model for compressible, multi-phase flow in porous media-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Vertical equilibrium models have proven to be well suited for simulating fluid flow in subsurface porous media such as saline aquifers with caprocks. However, in most cases the dimensionally reduced model lacks the accuracy to capture the dynamics of a system. While conventional full-dimensional models have the ability to represent dynamics, they come at the cost of high computational effort. We aim to combine the efficiency of the vertical equilibrium model and the accuracy of the full-dimensional model by coupling the two models adaptively in a unified framework and solving the emerging system of equations in a monolithic, fully-implicit approach. The model domains are coupled via mass-conserving fluxes while the model adaptivity is ruled by adaptive criteria. Overall, the adaptive model shows an excellent behaviour both in terms of accuracy as well as efficiency, especially for elongated geometries of storage systems with large aspect ratios. PubDate: 2025-03-29
- Downscaling using CDAnet under observational and model noise: the
Rayleigh-Bénard convection paradigm-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Efficient downscaling of large ensembles of coarse-scale information is crucial in several applications, such as oceanic and atmospheric modeling. The determining form map is a theoretical lifting function from the low-resolution solution trajectories of a dissipative dynamical system to their corresponding high-resolution counterparts. Recently, a physics-informed deep neural network (“CDAnet”) was introduced, providing a surrogate of the determining form map for efficient downscaling. CDAnet was demonstrated to efficiently downscale noise-free coarse-scale data in a deterministic setting. Herein, the performance of well-trained CDAnet models is analyzed in a stochastic setting involving (i) observational noise, (ii) model noise, and (iii) a combination of observational and model noise. The analysis is performed employing the Rayleigh-Bénard convection paradigm, under three training conditions, namely, training with perfect, noisy, or downscaled data. Furthermore, the effects of noise, Rayleigh number, and spatial and temporal resolutions of the input coarse-scale information on the downscaled fields are examined. The results suggest that the expected $$\ell _2$$-error of CDAnet behaves quadratically in terms of the standard deviations of the observational and model noise. The results also suggest that CDAnet responds to uncertainties similar to the theorized and numerically-validated CDA behavior with an additional error overhead due to CDAnet being a surrogate of the determining form map. PubDate: 2025-02-06
- Theoretical results on a block preconditioner used in ice-sheet modeling:
eigenvalue bounds for singular power-law fluids-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: The properties of a block preconditioner that has been successfully used in finite element simulations of large scale ice-sheet flow is examined. The type of preconditioner, based on approximating the Schur complement with the mass matrix scaled by the variable viscosity, is well-known in the context of Stokes flow and has previously been analyzed for other types of non-Newtonian fluids. We adapt the theory to hold for the regularized constitutive (power-law) equation for ice and derive eigenvalue bounds of the preconditioned system for both Picard and Newton linearization using inf-sup stable finite elements. The eigenvalue bounds show that viscosity-scaled preconditioning clusters the eigenvalues well with only a weak dependence on the regularization parameter, while the eigenvalue bounds for the traditional non-viscosity-scaled mass-matrix preconditioner are very sensitive to the same regularization parameter. The results are verified numerically in two experiments using a manufactured solution with low regularity and a simulation of glacier flow. The numerical results further show that the computed eigenvalue bounds for the viscosity-scaled preconditioner are nearly independent of the regularization parameter. Experiments are performed using both Taylor-Hood and MINI elements, which are the common choices for inf-sup stable elements in ice-sheet models. Both elements conform well to the theoretical eigenvalue bounds, with MINI elements being more sensitive to the quality of the meshes used in glacier simulations. PubDate: 2025-02-04
- Enhancing 3D seismic facies interpretation through a modified patched deep
learning approach leveraging spatio-temporal dependencies-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Seismic facies interpretation plays a vital role in oil and gas exploration and production. However, traditional methods, such as trace inversion and manual interpretation, are often time-consuming and labor-intensive. In recent years, deep learning algorithms have emerged as promising and efficient tools for facies identification with 3D seismic data. As a rapidly developing field, deep learning models with various network structures rise up all the time. Some of them are employed by researchers in the case studies of facies interpretation and are claimed as the better methods. However, the influence of the input features especially their inherent data structure, has attracted few discussions so far. Furthermore, most current studies using artificial intelligence for seismic interpretation primarily rely on two major branches of deep learning algorithms: convolutional neural networks (CNNs) which are skilled in capturing spatial patterns, and recurrent neural networks (RNNs) which are effective at modeling temporal dependencies. As a result, these networks and their variants fail to simultaneously leverage both spatial and temporal coupling of the multidimensional data. In this paper, we replace the matrix multiplications inside the memory cell of the general long short-term memory unit with a convolution operation, which is a basic module of the deep learning framework, to attach the capability of capturing the spatial dependencies with temporal dynamic behavior to the recurrent architecture. A patched deep learning model based on this theoretically rational and programming feasible RNN variant is implemented in three experiments of the 3D seismic facies interpretation. Our study firstly highlights the importance of the input seismic attributes in providing valuable information for making accurate predictions. The results from the first experiment demonstrate that the selection of seismic attributes based on their correlation with the interpretation target greatly enhances the model performance. Furthermore, by comparing the predictions from the proposed model with the ones from the model that just utilizes the spatial dependencies, our study emphasizes the significance of incorporating spatio-temporal dependencies within the chosen seismic attributes during the interpretation, as it leads to improved predictions, especially in boundary detection. Last but not least, our experiments demonstrate that the contribution of spatial dependencies to 3D seismic interpretation diminishes as the spatial distance increases. Therefore, selectively augmenting the training data with samples that have weaker spatial correlations can significantly enhance the model’s performance. Based on our findings, we prefer to conclude that interpreters that consider spatio-temporal dependencies inside the full covering optimized attributes can improve the quality of 3D seismic facies interpretation. This conclusion can serve as an outline for the workflow of Deep Learning-assisted 3D seismic interpretation. PubDate: 2025-01-16
- Multi-parameter post-stack seismic inversion based on the cycle loop –
semi-supervised learning-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Seismic inversion is used to evaluate hydrocarbon reservoirs by inferring subsurface physical properties from seismic data. The most prevalent seismic inversion method is post-stack seismic inversion, as most seismic data is available in post-stack format. However, this method only transforms impedance parameters from seismic data. Therefore, to extract additional parameters such as density and P wave velocity/Vp from post-stack seismic data, we developed Cycle Loop – Semi-Supervised Learning for multi-parameter post-stack seismic inversion (Cycle-MPInv). This method combines a deep learning-based inversion network and forward modeling to perform seismic inversion and forward modeling simultaneously (semi-supervised learning). The forward model provides geophysical constraint, while deep learning employs convolutional neural networks (CNNs) and bidirectional gated recurrent unit (Bi-GRU) to extract both high- and low-frequency features. In Cycle-MPInv, we also process labeled parameter data and calculate the loss to enhance the learning process and improve accuracy, so there are two loops that we call the cycle loop process. The proposed method was tested on synthetic data (Marmoussi II model) and data from the Netherlands offshore F3 block. Results from both datasets demonstrate that Cycle-MPInv effectively leverages both low- and high-frequency features of multi-parameter properties (density, Vp, and impedance) with limited labeled data. Furthermore, Cycle-MPInv achieves superior accuracy and robustness compared to other deep learning methods, even when handling noisy seismic data. These findings suggest that this method can offer valuable additional parameter insights for hydrocarbon reservoir evaluation using post-stack seismic data. PubDate: 2025-01-11
- Minimum acceptance criteria for subsurface scenario-based uncertainty
models from single image generative adversarial networks (SinGAN)-
Free pre-print version: Loading...
Rate this result:
What is this?
Please help us test our new pre-print finding feature by giving the pre-print link a rating. A 5 star rating indicates the linked pre-print has the exact same content as the published article.
Abstract: Evaluating and checking subsurface models is essential before their use to support optimum subsurface development decision making. Conventional geostatistical modeling workflows (e.g., two-point variogram-based geostatistics and multiple-point statistics) may fail to reproduce complex realistic geological patterns (e.g., channels), or be constrained by the limited training images and computational cost. Deep learning, specifically generative adversarial network (GAN), has been applied for subsurface modeling due to its ability to reproduce spatial and geological patterns, but may fail to reproduce commonly observed nonstationary subsurface patterns and often rely on many training images with the inability to explore realizations around specific geological scenarios. We propose an enhanced model checking workflow demonstrated by evaluating the performance of single image GAN (SinGAN)-based 2D image realizations for the case of channelized subsurface reservoirs to support robust uncertainty around geological scenarios. The SinGAN is able to generate nonstationary realizations from a single training image. Our minimum acceptance criteria expand on the work of Leuangthong, Boisvert, and others tailored to the nonstationary, single training image approach of SinGAN by evaluating the facies proportion, spatial continuity, and multiple-point statistics through histogram, semivariogram, and n-point histogram, along with evaluating the nonstationarity reproduction through multiple distribution checks ranging from local scale pixel distribution to multiscale local distribution. Additionally, our workflow incorporates reduced-dimensionality analysis through self-attention, providing a flexible approach for deep learning-based enhanced model realization to single training image comparison. With our proposed workflows, the robust application of SinGAN is possible to explore uncertainty around geological scenarios. PubDate: 2025-01-10
|