Traiwit Chung | The University of New South Wales (original) (raw)
Papers by Traiwit Chung
Transport in Porous Media
Reliable quantitative analysis of digital rock images requires precise segmentation and identific... more Reliable quantitative analysis of digital rock images requires precise segmentation and identification of the macroporosity, sub-resolution porosity, and solid\mineral phases. This is highly emphasized in heterogeneous rocks with complex pore size distributions such as carbonates. Multi-label segmentation of carbonates using classic segmentation methods such as multi-thresholding is highly sensitive to user bias and often fails in identifying low-contrast sub-resolution porosity. In recent years, deep learning has introduced efficient and automated algorithms that are capable of handling hard tasks with precision comparable to human performance, with application to digital rocks super-resolution and segmentation emerging. Here, we present a framework for using convolutional neural networks (CNNs) to produce super-resolved segmentations of carbonates rock images for the objective of identifying sub-resolution porosity. The volumes used for training and testing are based on two differ...
Transport in Porous Media, 2021
Fluid mechanics simulation of steady state flow in complex geometries has many applications, from... more Fluid mechanics simulation of steady state flow in complex geometries has many applications, from the micro-scale (cell membranes, filters, rocks) to macro-scale (groundwater, hydrocarbon reservoirs, and geothermal) and beyond. Direct simulation of steady state flow in such porous media requires significant computational resources to solve within reasonable timeframes. This study outlines an integrated method combining predictions of fluid flow (fast, limited accuracy) with direct flow simulation (slow, high accuracy) is outlined that reduces computation time by an order of magnitude without loss of accuracy. A convolutional neural network (CNNs) is trained with various configurations on simulations in 2D and 3D porous media to estimate steady state velocity fields. Permeability estimation (as an average of the field) is accurate, but the velocity fields themselves are error prone, unsuitable for further transport studies. This estimate can either be used as an indicative prediction, or as initial conditions in direct simulation to reach a fully accurate result in a fraction of the compute time. Using Deep Learning predictions (or potentially any other approximation method) to accelerate flow simulation to steady state in complex structures shows promise as a technique to push the boundaries fluid flow modelling. Steady State velocity fields predicted in 2D and 3D using CNNs Permeability estimation with predicted fields over 95% accurate in most cases Fine scale velocity field prediction is error-prone, limited by CNN performance Fast, low accuracy CNN prediction is combined with slow, high accuracy simulation Accelerated technique produces fully accurate results in 10x less time Steady State velocity fields predicted in 2D and 3D using CNNs Permeability estimation with predicted fields over 95% accurate in most cases Fine scale velocity field prediction is error-prone, limited by CNN performance Fast, low accuracy CNN prediction is combined with slow, high accuracy simulation Accelerated technique produces fully accurate results in 10x less time
Digital rock analysis has become increasingly popular for studying the microscopic structure of r... more Digital rock analysis has become increasingly popular for studying the microscopic structure of reservoir rocks. Direct numerical flow simulations are a common approach to compute petrophysical properties of rocks by modelling fluid flow on rock micro-Computed Tomography (CT) images. However, they are computationally demanding and complicated to include additional flow mechanisms.In this Thesis, a Pore-scale Finite Volume Solver (PFVS) is proposed that solves an elliptic diffusion equation to obtain the spatial pressure distribution on the entire micro-CT image. The flow results have 11% error compared to other solvers such as Stokes solver and Lattice-Boltzmann method. However, the computation times of PFVS are typically 5 times less compared to other solvers. PFVS is also capable of resolving the flow within the microporosity of rocks that cannot be captured by the previous solvers. PFVS is equipped with voxel agglomeration to merge pore voxels locally reducing the number of voxel...
DeePore is a deep learning workflow for rapid estimation of a wide range of porous material prope... more DeePore is a deep learning workflow for rapid estimation of a wide range of porous material properties based on the binarized micro-tomography images. By combining naturally occurring porous textures we generated 17700 semi-real 3-D micro-structures of porous geo-materials with size of 256^3 voxels and 30 physical properties of each sample are calculated using physical simulations on the corresponding pore network models. Next, a designed feed-forward convolutional neural network (CNN) is trained based on the dataset to estimate several morphological, hydraulic, electrical, and mechanical characteristics of the porous material in a fraction of a second. In order to fine-tune the CNN design, we tested 9 different training scenarios and selected the one with the highest average coefficient of determination (R^2) equal to 0.885 for 1418 testing samples. Additionally, 3 independent synthetic images as well as 3 realistic tomography images have been tested using the proposed method and r...
Journal of Petroleum Science and Engineering, 2021
Resolution of micro-computed tomography (micro-CT) images of rocks affects flow simulation result... more Resolution of micro-computed tomography (micro-CT) images of rocks affects flow simulation results, especially on low resolution images where the pore-grain boundary is not well resolved. However, there are two occasions where conducting flow simulations on low resolution images may be beneficial. The first case is when computation costs need to be reduced, and second is when high-resolution scanned images are unavailable and segmentation on available low-resolution images is difficult due to uncertainty at the pore boundary. A novel analytical formulation, the concentric pipes method, is introduced in this paper to include the effect of sub-resolution features in flow simulation. For the first case, we compare porosity, permeability, pore connectivity, and the velocity field obtained from applying different approaches to solve for flow on images with sub-resolution features. We find that the permeabilities obtained from downsampled images using the concentric pipes method have 13.2...
ArXiv, 2020
Simulation of fluid flow in porous media has many applications, from the micro-scale (cell membra... more Simulation of fluid flow in porous media has many applications, from the micro-scale (cell membranes, filters, rocks) to macro-scale (groundwater, hydrocarbon reservoirs, and geothermal) and beyond. Direct simulation of flow in porous media requires significant computational resources to solve within reasonable timeframes. An integrated method combining predictions of fluid flow (fast, limited accuracy) with direct flow simulation (slow, high accuracy) is outlined. In the tortuous flow paths of porous media, Deep Learning techniques based on Convolutional Neural Networks (CNNs) are shown to give an accurate estimate of the steady state velocity fields (in all axes), and by extension, the macro-scale permeability. This estimate can be used as-is, or as initial conditions in direct simulation to reach a fully accurate result in a fraction of the compute time. A Gated U-Net Convolutional Neural Network is trained on a datasets of 2D and 3D porous media generated by correlated fields, w...
Advances in Water Resources
Journal of Petroleum Science and Engineering, 2020
Direct numerical methods are widely used to solve for flow on micro-computed tomography (micro-CT... more Direct numerical methods are widely used to solve for flow on micro-computed tomography (micro-CT) images of rocks. Generally, direct numerical methods are computationally demanding, especially on large micro-CT images (10003 voxels and more). We develop a fast, direct numerical approach to solve the elliptic flow equation and estimate the absolute permeability of micro-CT images of rocks. The approach involves voxel agglomeration, which is a common technique in computational fluid dynamics. By allowing specific locations of the pore voxels to be agglomerated, the number of pore voxels or active cells in a sparse matrix can be reduced by approximately 60%–80% depending on the type of rock and its pore size distribution. Nevertheless, the fine details obtained from the micro-CT scan are maintained. The results compared with the Pore-scale Finite Volume Solver (PFVS) are within 1.6% difference on level 1 agglomerated grids and 1.9% for level 2 agglomerated grids without loss of the fl...
Digital rock physics (DRP) is an eminent technology that facilitates and repeatable core analysis... more Digital rock physics (DRP) is an eminent technology that facilitates and repeatable core analysis and multi-physics simulation of rock properties. One of the challenges in this field is the scalability of the problem size, whereby large micro-CT images over the order of 10003 voxels incur a high computational demand on performance. We estimate the of permeability in large digital samples of rocks imaged by micro-CT by using a fast and efficient Dual Grid Domain Decomposition technique based on the Schwarz Alternating Method (slow, low memory) with Algebraic Multigrid (AMG) solvers (fast, high memory) to solve on an otherwise unfeasible shared-memory machine. The comparisons and differences to other methods commonly used have been added in the introduction. The method applies a scalable parallel simulation algorithm to solve pressure and velocity fields using the Semi Analytical Pore Scale Finite Volume Solver (PFVS) within real 3D pore-scale micro-CT images. The domain is split into...
SPE Journal
Summary X-ray imaging of porous media has revolutionized the interpretation of various microscale... more Summary X-ray imaging of porous media has revolutionized the interpretation of various microscale phenomena in subsurface systems. The volumetric images acquired from this technology, known as digital rocks (DR), make it a suitable candidate for machine learning and computer-vision applications. The current routine DR frameworks involving image processing and modeling are susceptible to user bias and expensive computation requirements, especially for large domains. In comparison, the inference with trained machine-learning models can be significantly cheaper and computationally faster. Here we apply two popular convolutional neural network (ConvNet) architectures [residual network (ResNet) and ResNext] to learn the geometry of the pore space in 3D porous media images in a supervised learning scheme for flow-based characterization. The virtual permeability of the images to train the models is computed through a numerical simulation solver. Multiple ResNet variants are then trained to...
Journal of Computational Physics
Abstract Computation of two-phase flow in porous media with low capillary numbers is challenging ... more Abstract Computation of two-phase flow in porous media with low capillary numbers is challenging due to slow convergence and the presence of spurious currents at the phase interfaces that are greater than the viscous flow. The relative permeability of such systems is a critical parameter for upscaling flow properties but requires steady state flow configurations at low capillary numbers; a computationally slow problem to calculate. By using a morphologically coupled multiphase Lattice Boltzmann Method (LBM), it is observed that phase distributions converge an order of magnitude faster (1,000-50,000 timesteps) than flow fields (250,000-350,000 timesteps) during capillary dominated regimes. The proposed method couples a quasi-static, morphological method with direct LBM simulation that combines the efficiency of morphological calculations with the accuracy of direct simulation. The system fluid distribution is initialised morphologically instead of using simulated forced primary drainage to reduce dynamic simulation time and remove saturation end effects. The approach preconditions the simulation towards steady state conditions and the LBM routine relaxes the phase distributions until phases are stable. The steady state velocity fields are obtained by solving for flow in each stable connected phase distribution with a fast Semi Analytical Laplace solver to overcome spurious currents. A morphological Shell Aggregation method is then applied, condensing the displacing phase as a shell over pre-existing phase distributions and allowed to again reach phase equilibrium. Results obtained from the simulations are consistent with experimental relative permeability curves and phase morphology obtained from Gildehauser sandstone. This method allows rapid computation of phase distributions and relative permeability for capillary dominated flows. Shell Aggregation typically reaches steady state within 50,000-150,000 LBM timesteps opposed to 250,000-350,000 by spinodal decomposition for the tested 500 cubed Bentheimer sandstone and 150 cubed sand pack samples. Solving for flow in each connected phase body after Shell Aggregation LBM reaches steady state is furthermore shown to require down to 1,000-10,000 LBM timesteps at the expense of some interfacial physics.
Advances in Water Resources
Transport in Porous Media
Transport in Porous Media
Reliable quantitative analysis of digital rock images requires precise segmentation and identific... more Reliable quantitative analysis of digital rock images requires precise segmentation and identification of the macroporosity, sub-resolution porosity, and solid\mineral phases. This is highly emphasized in heterogeneous rocks with complex pore size distributions such as carbonates. Multi-label segmentation of carbonates using classic segmentation methods such as multi-thresholding is highly sensitive to user bias and often fails in identifying low-contrast sub-resolution porosity. In recent years, deep learning has introduced efficient and automated algorithms that are capable of handling hard tasks with precision comparable to human performance, with application to digital rocks super-resolution and segmentation emerging. Here, we present a framework for using convolutional neural networks (CNNs) to produce super-resolved segmentations of carbonates rock images for the objective of identifying sub-resolution porosity. The volumes used for training and testing are based on two differ...
Transport in Porous Media, 2021
Fluid mechanics simulation of steady state flow in complex geometries has many applications, from... more Fluid mechanics simulation of steady state flow in complex geometries has many applications, from the micro-scale (cell membranes, filters, rocks) to macro-scale (groundwater, hydrocarbon reservoirs, and geothermal) and beyond. Direct simulation of steady state flow in such porous media requires significant computational resources to solve within reasonable timeframes. This study outlines an integrated method combining predictions of fluid flow (fast, limited accuracy) with direct flow simulation (slow, high accuracy) is outlined that reduces computation time by an order of magnitude without loss of accuracy. A convolutional neural network (CNNs) is trained with various configurations on simulations in 2D and 3D porous media to estimate steady state velocity fields. Permeability estimation (as an average of the field) is accurate, but the velocity fields themselves are error prone, unsuitable for further transport studies. This estimate can either be used as an indicative prediction, or as initial conditions in direct simulation to reach a fully accurate result in a fraction of the compute time. Using Deep Learning predictions (or potentially any other approximation method) to accelerate flow simulation to steady state in complex structures shows promise as a technique to push the boundaries fluid flow modelling. Steady State velocity fields predicted in 2D and 3D using CNNs Permeability estimation with predicted fields over 95% accurate in most cases Fine scale velocity field prediction is error-prone, limited by CNN performance Fast, low accuracy CNN prediction is combined with slow, high accuracy simulation Accelerated technique produces fully accurate results in 10x less time Steady State velocity fields predicted in 2D and 3D using CNNs Permeability estimation with predicted fields over 95% accurate in most cases Fine scale velocity field prediction is error-prone, limited by CNN performance Fast, low accuracy CNN prediction is combined with slow, high accuracy simulation Accelerated technique produces fully accurate results in 10x less time
Digital rock analysis has become increasingly popular for studying the microscopic structure of r... more Digital rock analysis has become increasingly popular for studying the microscopic structure of reservoir rocks. Direct numerical flow simulations are a common approach to compute petrophysical properties of rocks by modelling fluid flow on rock micro-Computed Tomography (CT) images. However, they are computationally demanding and complicated to include additional flow mechanisms.In this Thesis, a Pore-scale Finite Volume Solver (PFVS) is proposed that solves an elliptic diffusion equation to obtain the spatial pressure distribution on the entire micro-CT image. The flow results have 11% error compared to other solvers such as Stokes solver and Lattice-Boltzmann method. However, the computation times of PFVS are typically 5 times less compared to other solvers. PFVS is also capable of resolving the flow within the microporosity of rocks that cannot be captured by the previous solvers. PFVS is equipped with voxel agglomeration to merge pore voxels locally reducing the number of voxel...
DeePore is a deep learning workflow for rapid estimation of a wide range of porous material prope... more DeePore is a deep learning workflow for rapid estimation of a wide range of porous material properties based on the binarized micro-tomography images. By combining naturally occurring porous textures we generated 17700 semi-real 3-D micro-structures of porous geo-materials with size of 256^3 voxels and 30 physical properties of each sample are calculated using physical simulations on the corresponding pore network models. Next, a designed feed-forward convolutional neural network (CNN) is trained based on the dataset to estimate several morphological, hydraulic, electrical, and mechanical characteristics of the porous material in a fraction of a second. In order to fine-tune the CNN design, we tested 9 different training scenarios and selected the one with the highest average coefficient of determination (R^2) equal to 0.885 for 1418 testing samples. Additionally, 3 independent synthetic images as well as 3 realistic tomography images have been tested using the proposed method and r...
Journal of Petroleum Science and Engineering, 2021
Resolution of micro-computed tomography (micro-CT) images of rocks affects flow simulation result... more Resolution of micro-computed tomography (micro-CT) images of rocks affects flow simulation results, especially on low resolution images where the pore-grain boundary is not well resolved. However, there are two occasions where conducting flow simulations on low resolution images may be beneficial. The first case is when computation costs need to be reduced, and second is when high-resolution scanned images are unavailable and segmentation on available low-resolution images is difficult due to uncertainty at the pore boundary. A novel analytical formulation, the concentric pipes method, is introduced in this paper to include the effect of sub-resolution features in flow simulation. For the first case, we compare porosity, permeability, pore connectivity, and the velocity field obtained from applying different approaches to solve for flow on images with sub-resolution features. We find that the permeabilities obtained from downsampled images using the concentric pipes method have 13.2...
ArXiv, 2020
Simulation of fluid flow in porous media has many applications, from the micro-scale (cell membra... more Simulation of fluid flow in porous media has many applications, from the micro-scale (cell membranes, filters, rocks) to macro-scale (groundwater, hydrocarbon reservoirs, and geothermal) and beyond. Direct simulation of flow in porous media requires significant computational resources to solve within reasonable timeframes. An integrated method combining predictions of fluid flow (fast, limited accuracy) with direct flow simulation (slow, high accuracy) is outlined. In the tortuous flow paths of porous media, Deep Learning techniques based on Convolutional Neural Networks (CNNs) are shown to give an accurate estimate of the steady state velocity fields (in all axes), and by extension, the macro-scale permeability. This estimate can be used as-is, or as initial conditions in direct simulation to reach a fully accurate result in a fraction of the compute time. A Gated U-Net Convolutional Neural Network is trained on a datasets of 2D and 3D porous media generated by correlated fields, w...
Advances in Water Resources
Journal of Petroleum Science and Engineering, 2020
Direct numerical methods are widely used to solve for flow on micro-computed tomography (micro-CT... more Direct numerical methods are widely used to solve for flow on micro-computed tomography (micro-CT) images of rocks. Generally, direct numerical methods are computationally demanding, especially on large micro-CT images (10003 voxels and more). We develop a fast, direct numerical approach to solve the elliptic flow equation and estimate the absolute permeability of micro-CT images of rocks. The approach involves voxel agglomeration, which is a common technique in computational fluid dynamics. By allowing specific locations of the pore voxels to be agglomerated, the number of pore voxels or active cells in a sparse matrix can be reduced by approximately 60%–80% depending on the type of rock and its pore size distribution. Nevertheless, the fine details obtained from the micro-CT scan are maintained. The results compared with the Pore-scale Finite Volume Solver (PFVS) are within 1.6% difference on level 1 agglomerated grids and 1.9% for level 2 agglomerated grids without loss of the fl...
Digital rock physics (DRP) is an eminent technology that facilitates and repeatable core analysis... more Digital rock physics (DRP) is an eminent technology that facilitates and repeatable core analysis and multi-physics simulation of rock properties. One of the challenges in this field is the scalability of the problem size, whereby large micro-CT images over the order of 10003 voxels incur a high computational demand on performance. We estimate the of permeability in large digital samples of rocks imaged by micro-CT by using a fast and efficient Dual Grid Domain Decomposition technique based on the Schwarz Alternating Method (slow, low memory) with Algebraic Multigrid (AMG) solvers (fast, high memory) to solve on an otherwise unfeasible shared-memory machine. The comparisons and differences to other methods commonly used have been added in the introduction. The method applies a scalable parallel simulation algorithm to solve pressure and velocity fields using the Semi Analytical Pore Scale Finite Volume Solver (PFVS) within real 3D pore-scale micro-CT images. The domain is split into...
SPE Journal
Summary X-ray imaging of porous media has revolutionized the interpretation of various microscale... more Summary X-ray imaging of porous media has revolutionized the interpretation of various microscale phenomena in subsurface systems. The volumetric images acquired from this technology, known as digital rocks (DR), make it a suitable candidate for machine learning and computer-vision applications. The current routine DR frameworks involving image processing and modeling are susceptible to user bias and expensive computation requirements, especially for large domains. In comparison, the inference with trained machine-learning models can be significantly cheaper and computationally faster. Here we apply two popular convolutional neural network (ConvNet) architectures [residual network (ResNet) and ResNext] to learn the geometry of the pore space in 3D porous media images in a supervised learning scheme for flow-based characterization. The virtual permeability of the images to train the models is computed through a numerical simulation solver. Multiple ResNet variants are then trained to...
Journal of Computational Physics
Abstract Computation of two-phase flow in porous media with low capillary numbers is challenging ... more Abstract Computation of two-phase flow in porous media with low capillary numbers is challenging due to slow convergence and the presence of spurious currents at the phase interfaces that are greater than the viscous flow. The relative permeability of such systems is a critical parameter for upscaling flow properties but requires steady state flow configurations at low capillary numbers; a computationally slow problem to calculate. By using a morphologically coupled multiphase Lattice Boltzmann Method (LBM), it is observed that phase distributions converge an order of magnitude faster (1,000-50,000 timesteps) than flow fields (250,000-350,000 timesteps) during capillary dominated regimes. The proposed method couples a quasi-static, morphological method with direct LBM simulation that combines the efficiency of morphological calculations with the accuracy of direct simulation. The system fluid distribution is initialised morphologically instead of using simulated forced primary drainage to reduce dynamic simulation time and remove saturation end effects. The approach preconditions the simulation towards steady state conditions and the LBM routine relaxes the phase distributions until phases are stable. The steady state velocity fields are obtained by solving for flow in each stable connected phase distribution with a fast Semi Analytical Laplace solver to overcome spurious currents. A morphological Shell Aggregation method is then applied, condensing the displacing phase as a shell over pre-existing phase distributions and allowed to again reach phase equilibrium. Results obtained from the simulations are consistent with experimental relative permeability curves and phase morphology obtained from Gildehauser sandstone. This method allows rapid computation of phase distributions and relative permeability for capillary dominated flows. Shell Aggregation typically reaches steady state within 50,000-150,000 LBM timesteps opposed to 250,000-350,000 by spinodal decomposition for the tested 500 cubed Bentheimer sandstone and 150 cubed sand pack samples. Solving for flow in each connected phase body after Shell Aggregation LBM reaches steady state is furthermore shown to require down to 1,000-10,000 LBM timesteps at the expense of some interfacial physics.
Advances in Water Resources
Transport in Porous Media