HYPPS: A hybrid geostatistical modeling algorithm for subsurface modeling

Tahmasebi, Pejman

doi:10.1002/2017wr021078

Cited by 56 publications

(16 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Subsurface flow and transport modeling studies often rely on inversion methodologies to derive subsurface structures that are consistent with both available prior information and indirect measurements of one or more state variables, such as hydraulic head or concentration. When it is unrealistic to assume that the subsurface is multi‐Gaussian (e.g., Gómez‐Hernández & Wen, ; Journel & Zhang, ), one popular solution is to resort to multiple‐point statistics (MPS) simulation (e.g., Li et al, ; Mariethoz et al, ; Strebelle, ; Tahmasebi, ; Tahmasebi & Sahimi, ). MPS techniques generate model realizations that honor a prior model which is determined by a training image (TI).…”

Section: Introductionmentioning

confidence: 99%

Training‐Image Based Geostatistical Inversion Using a Spatial Generative Adversarial Neural Network

Laloy

Hérault

Jacques

et al. 2018

Water Resources Research

325

257

View full text Add to dashboard Cite

Probabilistic inversion within a multiple‐point statistics framework is often computationally prohibitive for high‐dimensional problems. To partly address this, we introduce and evaluate a new training‐image based inversion approach for complex geologic media. Our approach relies on a deep neural network of the generative adversarial network (GAN) type. After training using a training image (TI), our proposed spatial GAN (SGAN) can quickly generate 2‐D and 3‐D unconditional realizations. A key characteristic of our SGAN is that it defines a (very) low‐dimensional parameterization, thereby allowing for efficient probabilistic inversion using state‐of‐the‐art Markov chain Monte Carlo (MCMC) methods. In addition, available direct conditioning data can be incorporated within the inversion. Several 2‐D and 3‐D categorical TIs are first used to analyze the performance of our SGAN for unconditional geostatistical simulation. Training our deep network can take several hours. After training, realizations containing a few millions of pixels/voxels can be produced in a matter of seconds. This makes it especially useful for simulating many thousands of realizations (e.g., for MCMC inversion) as the relative cost of the training per realization diminishes with the considered number of realizations. Synthetic inversion case studies involving 2‐D steady state flow and 3‐D transient hydraulic tomography with and without direct conditioning data are used to illustrate the effectiveness of our proposed SGAN‐based inversion. For the 2‐D case, the inversion rapidly explores the posterior model distribution. For the 3‐D case, the inversion recovers model realizations that fit the data close to the target level and visually resemble the true model well.

show abstract

Section: Introductionmentioning

confidence: 99%

Training‐Image Based Geostatistical Inversion Using a Spatial Generative Adversarial Neural Network

Laloy

Hérault

Jacques

et al. 2018

Water Resources Research

325

257

View full text Add to dashboard Cite

show abstract

“…; Karimpouli et al . , ; Karimpouli, Tahmasebi and Saenger ; Tahmasebi, Sahimi and Andrade ; Tahmasebi ,b; Tahmasebi ,b,c). Therefore, six real 3D digital samples, three sandstones (Bentheimer, Clashach and Doddington) and three carbonates (Estaillades, Ketton and Portland), are used in this study.…”

Section: Introductionmentioning

confidence: 99%

3D multi‐fractal analysis of porous media using 3D digital images: considerations for heterogeneity evaluation

Karimpouli

Tahmasebi

2018

Geophysical Prospecting

Self Cite

View full text Add to dashboard Cite

Pore structure heterogeneity is a critical parameter controlling mechanical, electrical and flow transport behaviour of rock. Multi‐fractal analysis method was used for a heterogeneity comparison of three‐dimensional rock samples with different lithology. Six real digital samples, containing three sandstones and three carbonates, were used. Based on the mercury injection capillary pressure test on these samples, we found that the carbonate samples are more heterogeneous than sandstones, but primary results of multi‐fractal behaviours for all samples were similar. We show that if multi‐fractal is used to evaluate and compare heterogeneity of different samples, one needs to follow some considerations such as (1) all samples must have the same size in pixel, (2) samples volume must be bigger than representative volume element, (3) multi‐fractal dimensions should be firstly normalized to a determined porosity value and (4) multi‐fractal results should be interpreted based on resolution of the imaging tool (effects of fine scale sub‐resolution pores are missed). Results revealed that using normalized fractal dimensions, the real samples were divided to less and high heterogeneous groups. Moreover, the study of scale effect also showed that porous structures of these samples are scale invariant in a wide range of scales (from one to eight times bigger).

show abstract

“…Our work involved using two data sets, Berea sandstone and a synthetic sample, namely a polydisperse sphere-pack; see Figure 1. These data are taken from the available data on Pore-Scale Modeling group at Imperial College University (Dong, 2008 Water Resources Research be modeled through the stochastic algorithms by which one can generate various realizations [Tahmasebi et al, 2017b;Karimpouli and Tahmasebi, 2016;Tahmasebi, 2018c;Tahmasebi, 2017].…”

Section: Discussionmentioning

confidence: 99%

“…The sandstone consisted of 2,074,984 tetrahedral elements, 18,459 edge elements, and 4,313 vertices, while the sphere‐pack has 1,104,097 tetrahedral elements, 11,107 edge elements, and 2,349 vertices. It should be noted that the uncertainty of the initial model can be modeled through the stochastic algorithms by which one can generate various realizations [ Tahmasebi et al, ; Karimpouli and Tahmasebi , ; Tahmasebi , ; Tahmasebi , ].…”

Section: Porous Medium Pore‐scale Imaging and Meshingmentioning

confidence: 99%

Interaction Between Fluid and Porous Media with Complex Geometries: A Direct Pore‐Scale Study

Fagbemi

Tahmasebi

Piri

2018

Water Resources Research

Self Cite

View full text Add to dashboard Cite

Effect of boundary stress on permeability in porous media at pore‐scale is investigated by means of numerical methods for computing deformation of rock material in the presence of pore fluid. We implement a partitioned, strongly coupled solver for transient fluid‐solid simulations with independent conformed interface meshes for the fluid and solid domains. Fluid flow and solid computations were performed using the finite element method. External load is prescribed along the solid boundary in gradual steps to examine the extent of pore‐scale deformation along internal grain walls while applying a fixed constraint at the opposite end to ensure uniform stress distribution across the media. Hydrodynamic interactions between the pore walls and the matrix leading to interface displacements by fluid viscous forces are also investigated by examining fluid flow with different combinations of Reynolds number. Coupling is achieved via traction and displacements boundary conditions at the interface with respect to Lagrangian coordinates. The role of particle shape in the matrix deformation process is also considered in this paper. In order to examine the how particle shape affects local affects local permeability/stress, we examine fluid movement based on varying Reynolds numbers in Berea sandstone obtained from Micro‐CT image scans at high resolution, and artificially constructed random polydisperse 3‐D sphere packs. It was found that permeability alteration under the influence of uniaxial stress was more in the sandstone sample than in the sphere‐pack and that the sphere‐pack deformed less than the actual sample due to pore geometry, surface smoothness, and homogeneity.

show abstract

HYPPS: A hybrid geostatistical modeling algorithm for subsurface modeling

Cited by 56 publications

References 64 publications

Training‐Image Based Geostatistical Inversion Using a Spatial Generative Adversarial Neural Network

Training‐Image Based Geostatistical Inversion Using a Spatial Generative Adversarial Neural Network

3D multi‐fractal analysis of porous media using 3D digital images: considerations for heterogeneity evaluation

Interaction Between Fluid and Porous Media with Complex Geometries: A Direct Pore‐Scale Study

Contact Info

Product

Resources

About