Computations of permeability of large rock images by dual grid domain decomposition

Wang, Ying Da; Chung, Traiwit; Armstrong, Ryan T.; McClure, James E.; Mostaghimi, Peyman

doi:10.1016/j.advwatres.2019.02.002

Cited by 55 publications

(26 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the number of grids in this study was small and the gradation parameters were not considered. There are also some researches focusing on unsaturated seepage (Dou et al, 2013;Wang et al, 2016Wang et al, , 2019 or rock seepage (Zhang et al, 2016;Akai et al, 2018;Li and Berkowitz, 2018) based on LBM available in literature. However, the morphological characteristics of particle were not considered in unsaturated seepage studies, and the pore structure of rock is very different from that of particle-packed porous medium.…”

Section: Accepted Manuscriptmentioning

confidence: 99%

Pore scale study of the influence of particle geometry on soil permeability

Liu

Jeng

2019

Advances in Water Resources

View full text Add to dashboard Cite

The permeability of a saturated porous medium is an important parameter in the field of water resources and geotechnical engineering. The geometric characteristics of a porous medium are key factors in the prediction of its permeability. In this paper, particles of different shapes are constructed by Cellular Automata (CA) random growth model, and particles with different surface characteristics are constructed by the spherical harmonic function. Then, porous media of different porosities, shapes, surface features, and particle size distributions are generated on the basis of the constructed particles. Three-dimensional Lattice Boltzmann Method is used for the pore-scale simulation of the seepage flow in a porous medium. The numerical results show that the effects of the particle shape and surface characteristics on the permeability are too obvious to be ignored. Using strict univariate analysis, the sensitivity of the various factors to the permeability, ordered from large to small, is as follows: porosity>particle size distribution>particle surface>particle shape. Based on numerical studies, a modified Kozeny-Carman (KC) formula is proposed by considering all the geometrical factors. All the parameters (the Wadell sphericity S w , Cox roundness R c , coefficient of non-uniformity C u , the curve coefficient of curvature C c , and effective particle size d 10 ) in it are easily obtained in engineering practice and the accuracy of the formula is verified. Although It has been proven that the KC formula is applicable to multi-dispersed spherical particles and non-spherical particles whose surfaces are not very rough, its applicability to rough particles is limited. The modified KC formula does not have this limitation; therefore, it has a wider scope of application than the conventional KC formula.

show abstract

Section: Accepted Manuscriptmentioning

confidence: 99%

Pore scale study of the influence of particle geometry on soil permeability

Liu

Jeng

2019

Advances in Water Resources

View full text Add to dashboard Cite

show abstract

“…The main limitation of direct methods is the high computational cost [1], which could be a major obstacle when we are dealing with large-sized and high-resolution volumetric images of porous material. As a solution to this size and time limitation, domain decomposition and parallel computation have been comprehensively hired to increase the models efficiency and scalability [32,33,34,35,36]. Additionally, as another solution to deal with computational limitations, machine learning can be employed to mimic the behaviour of the complex solid/fluid systems.…”

Section: Direct Simulation Methodsmentioning

confidence: 99%

Hybrid pore-network and lattice-Boltzmann permeability modelling accelerated by machine learning

Rabbani

Babaei

2019

Advances in Water Resources

125

View full text Add to dashboard Cite

show abstract

“…Considering that data analysis accelerated by Graphic Processing Units (GPU) is becoming very critical in data science, a distributed file architecture helps the data to fit in limited memory of the GPUs (Bauer et al, 2011). Wang, Chung, et al (2019) and are two examples of application of distributed architecture for analysis of the big tomography data, both focusing on predicting the permeability of the porous material via Lattice Boltzmann Method and Pore Network Modeling, respectively. used a domain decomposition technique to analyze large size images of porous sandstones and extract separated pore network models for each of the domains.…”

Section: Distributed Architecturementioning

confidence: 99%