A method to discretize non‐planar fractures for 3D subsurface flow and transport simulations

A method is presented to discretize inclined non‐planar 2D fractures within a 3D finite element grid for subsurface flow and transport simulations. Each 2D fracture is represented as a triangulated surface. Each triangle is then discretized by 2D fracture elements that can be horizontal, vertical or inclined and that can be triangular or rectangular. The 3D grid representing a porous rock formation consists of hexahedra and can be irregular to allow grid refinement. An inclined fracture was discretized by (a) inclined triangles and (b) orthogonal rectangles and flow/transport simulations were run to compare the results. The comparison showed that (i) inclined fracture elements must be used to simulate 2D transient flow, (ii) results of 2D/3D steady‐state and 3D transient flow simulations are identical for both discretization methods, (iii) inclined fracture elements must be used to simulate 2D/3D transport because orthogonal fracture elements significantly underestimate concentrations, and (iv) orthogonal elements can be used to simulate 2D/3D transport if fracture permeability is corrected and multiplied by the ratio of fracture surface areas (orthogonal to inclined). Groundwater flow at a potential site for long‐term disposal of spent nuclear fuel was simulated where a complex 3D fracture network was discretized with this technique. The large‐scale simulation demonstrates that the proposed discretization procedure offers new possibilities to simulate flow and transport in complex 3D fracture networks. The new procedure has the further advantage that the same grid can be used for different realizations of a fracture network model with no need to regenerate the grid. Copyright © 2007 John Wiley & Sons, Ltd.     

Numerical modelling of shear-dependent mass transfer in large arteries

A numerical scheme for the simulation of blood flow and transport processes in large arteries is presented. Blood flow is described by the unsteady 3D incompressible Navier–Stokes equations for Newtonian fluids; solute transport is modelled by the advection–diffusion equation. The resistance of the arterial wall to transmural transport is described by a shear-dependent wall permeability model. The finite element formulation of the Navier–Stokes equations is based on an operator-splitting method and implicit time discretization. The streamline upwind/Petrov–Galerkin (SUPG) method is applied for stabilization of the advective terms in the transport equation and in the flow equations. A numerical simulation is carried out for pulsatile mass transport in a 3D arterial bend to demonstrate the influence of arterial flow patterns on wall permeability characteristics and transmural mass transfer. The main result is a substantial wall flux reduction at the inner side of the curved region. © 1997 John Wiley & Sons, Ltd.     

Comparison of Flow and Transport Experiments on 3D Printed Micromodels with Direct Numerical Simulations

Understanding pore-scale flow and transport processes is important for understanding flow and transport within rocks on a larger scale. Flow experiments on small-scale micromodels can be used to experimentally investigate pore-scale flow. Current manufacturing methods of micromodels are costly and time consuming. 3D printing is an alternative method for the production of micromodels. We have been able to visualise small-scale, single-phase flow and transport processes within a 3D printed micromodel using a custom-built visualisation cell. Results have been compared with the same experiments run on a micromodel with the same geometry made from polymethyl methacrylate (PMMA, also known as Perspex). Numerical simulations of the experiments indicate that differences in experimental results between the 3D printed micromodel and the Perspex micromodel may be due to variability in print geometry and surface properties between the samples. 3D printing technology looks promising as a micromodel manufacturing method; however, further work is needed to improve the accuracy and quality of 3D printed models in terms of geometry and surface roughness.

     

A 3D smoothed particle hydrodynamics model for erosional dam-break floods

A mesh-less smoothed particle hydrodynamics (SPH) model for bed-load transport on erosional dam-break floods is presented. This mixture model describes both the liquid phase and the solid granular material. The model is validated on the results from several experiments on erosional dam breaks. A comparison between the present model and a 2-phase SPH model for geotechnical applications (Gadget Soil; TUHH) is performed. A demonstrative 3D erosional dam break on complex topography is investigated. The present 3D mixture model is characterised by: no tuning parameter for the mixture viscosity; consistency with the Kinetic Theory of Granular Flow; ability to reproduce the evolution of the free surface and the bed-load transport layer; applicability to practical problems in civil engineering. The numerical developments of this study are represented by a new SPH scheme for bed-load transport, which is implemented in the SPH code SPHERA v.8.0 (RSE SpA), distributed as FOSS on GitHub.     

Poromechanical Modelling and Inverse Approach of Drying Tests on Weakly Permeable Porous Rocks

The present paper deals with the determination of permeability in partially saturated conditions for weakly permeable porous rocks such as argillites or deep clayey formations. The level of permeability can be obtained via the measurements of transient weight loss of a sample submitted to a decrease in relative humidity imposed by saline solution in a hermetic chamber. An identification method based on simplified uncoupled 1D-linear and 1D-non-linear modelling was presented in a previous paper (Giraud et al. Trans Porous Media 69(2):259–280, 2006). The present paper takes into account generalized mass transfer phenomena such as Darcean advective transport of liquid and gas mixtures and Fickean diffusive transport of the vapour specie inside a gas mixture. Poromechanical coupling as well as 3D effects due to the geometry and finite dimensions of the tested samples are also covered by this approach. The coupled THM finite element computer code Code_Aster is then used to model the forward problem. The parameter identification procedure is based upon the solution of an inverse problem. The Levenberg–Marquardt algorithm was used for the problem of minimization. Comparisons between previous simplified 1D modelling and 2D-axisymmetrical coupled modelling show that the former method efficiently provides the correct order of magnitude of the level of permeability or the equivalent storage coefficients. Due to the boundary condition, the real 2D-axisymmetrical geometry of the sample must not be neglected if we are to obtain accurate results.     

Finite element solutions for nonhomogeneous nonlocal elastic problems

A finite element procedure for analysing nonhomogeneous nonlocal elastic 2D problems is presented and discussed. The procedure grounds on a variationally consistent approach known, in the relevant literature, as Nonlocal Finite Element Method. The latter is recast making use of a recently theorized phenomenological strain-difference-based nonhomogeneous nonlocal elastic model. The peculiarities of the numerical procedure together with the pertinent nonlocal operators are expounded and discussed. Two simple numerical 2D examples close the paper.     

页岩气和致密砂岩气藏微裂缝气体传输特性

页岩气和致密砂岩气藏发育微裂缝,其开度多在纳米级和微米级尺度且变化大,因此微裂缝气体传输机理异常复杂.本文基于滑脱流动和努森扩散模型,分别以分子之间碰撞频率和分子与壁面碰撞频率占总碰撞频率的比值作为滑脱流动和努森扩散的权重系数,耦合这两种传输机理,建立了微裂缝气体传输模型. 该模型考虑微裂缝形状和尺度对气体传输的影响. 模型可靠性用分子模拟数据验证.结果表明:(1)模型能够合理描述微裂缝中所有气体传输机理,包括连续流动,滑脱流动和过渡流动;(2)模型能够描述不同开发阶段,微裂缝中各气体传输机理对传输贡献的逐渐变化过程;(3)微裂缝形状和尺度影响气体传输,相同开度且宽度越大的微裂缝,气体传输能力越强,且在高压和微裂缝大开度的情况下表现更明显.      

MHD Turbulence at the Laboratory Scale: Established Ideas and New Challenges

The properties of MHD turbulence in the electrically conducting fluids available in the laboratory (where the magnetic Reynolds number is significantly smaller than unity) may be summarised as follows:(1) The Alfven waves, even under their degenerated form at this scale, are responsible for a tendency to two-dimensionality. Eddies tend to become aligned with the applied magnetic field and inertia tends to restore isotropy. The competition between these mechanisms results in a spectral law t^-2k^-3.(2) When insulating walls, perpendicular to the magnetic field, are present and close enough to each other, two-dimensionality can be established with a good approximation within the large scales, and the predominant mechanism is the inverse energy cascade.(3) These columnar eddies are nevertheless submitted to a dissipation within the Hartmann boundary layers present at their ends, whose time scale is independent of the wave number. When this damping effect is negligible, ordinary 2D turbulence is observed with k^-5/3 spectra. On the contrary when this (ohmic and viscous) damping is significant this 2D turbulence exhibits k^-3 spectra.Besides these homogeneous (except within the Hartmann layers) conditions, for instance in shear flows such as mixing layers, almost nothing is known except that two-dimensionality may be well established. The first results of a recent experimental investigation (still in development) are presented. Some challenging questions are raised, such as the interpretation of a surprising difference between the transport of momentum and the transport of a scalar quantity (heat) across that layer. A video was shown during the oral presentation of this paper, illustrating the energy transfer toward the large scales and the weakness of the dissipation suffered by this 2D velocity field.     

Convective drying analysis of three-dimensional porous solid by mass lumping finite element technique

A numerical analysis of convective drying of a 3D porous solid of brick material is carried out using the finite element method and mass lumping technique. The energy equation and moisture transport equations for the porous solid are derived based on continuum approach following Whitaker’s theory of drying. The governing equations are solved using the Galerkin’s weighted residual method, which convert the governing equations into discretized form of matrix equations. The resulting capacitance matrices are made diagonal matrices by following the classical row-sum mass lumping technique. Hence with the use of the Eulerian time marching scheme, the final equations are reduced to simple algebraic equations, which can be solved directly without using an equation solver. The proposed numerical scheme is initially validated with experimental results for 1D drying problem and then tested by application to convective drying of 3D porous solid of brick material for four different aspect ratios obtained by varying the cross section of the solid. The mass lumping technique could correctly predict the wet bulb temperature of the solid under evaporative drying conditions. A parametric study carried out for three different values of convective heat transfer coefficients, 15, 30 and 45 W/m² K shows an increased drying rate with increase in area of cross section and convective heat transfer coefficient. The proposed numerical scheme could correctly predict the drying behavior shown in the form of temperature and moisture evolutions.     

A new formulation for fictitious mass of viscous dynamic relaxation method

Abstract

In this article, a new relationship is proposed for the fictitious mass of viscous dynamic relaxation (DR) method. First, incremental equations are derived for DR steps. Using transformed Gershgörin theory, a new relationship is achieved for fictitious mass of viscous DR by formulating modified time step ratio. This procedure presents a new algorithm for the viscous DR method. To evaluate the numerical efficiency of the proposed method, some 2D and 3D truss and frame structures are analyzed with elastic linear and geometrically nonlinear behaviors. Results show that by using the proposed algorithm for fictitious mass, the convergence rate of the viscous DR method is improved so that the proposed algorithm presents the structural response with lower iterations in comparison with other common DR techniques.

Communicated by Joerg Fehr.     

Improvements on bed‐shear stress formulation for pier scour computation

This paper introduces an improved formula for the bed‐shear stress by applying the vorticity effect and its application in a 3D flow and sediment model to estimate scouring around bridge piers. Up to now, the sediment transport formulae used for computing pier scour were developed based on the general scouring in unobstructed flow. The capability for numerical models to predict local scour around bridge piers was severely restricted by the sediment transport formulae. The new formula introduced in this paper can take into account vortices that affect the local scour process by adding some terms into the classic bed‐shear stress equation. The 3D numerical model system used in this study consists of three modules: (a) an unsteady hydrodynamic module; (b) a sediment transport module; and (c) a Fation module. The hydrodynamic module is based on the 3D RANS equations. The sediment transport module is comprised of semi empirical models of suspended load and non‐equilibrium bed load. The bed‐deformation module is based on the mass balance for sediment. The model was used to simulate pier scour in tree different test cases: (1) a circular pier; (2) a square pier; and (3) a rectangular pier, by applying the ordinary sediment equation and the newly introduced sediment equation. Results of both numerical simulations were compared against laboratory measured data and also in case 1 with result of Olsen and Melaaen (J. Hydraul. Eng. 1993; 119 (9):1048–1054). Comparisons show that the new sediment formula could predict the scour more accurately than the ordinary one. Copyright © 2010 John Wiley & Sons, Ltd.     

Unsteady three-dimensional MHD flow of the micropolar fluid over an oscillatory disk with Cattaneo-Christov double diffusion

Cattaneo-Christov heat and mass flux models are considered rather than Fourier and Fick laws due to the presence of thermal and concentration transport hyperbolic phenomena. The generalized form of the Navier-Stokes model is considered in hydromagnetic flow. Three-dimensional(3 D) unsteady fluid motion is generated by the periodic oscillations of a rotating disk. Similarity transformations are used to obtain the normalized fluid flow model. The successive over relaxation(SOR) method with finite difference schemes are accomplished for the numerical solution of the obtained partial differential non-linear system. The flow features of the velocity, microrotation, temperature,and concentration fields are discussed in pictorial forms for various physical flow parameters. The couple stresses and heat and mass transfer rates for different physical quantities are explained via tabular forms. For better insight of the physical fluid model, 3 D fluid phenomena and two-dimensional(2 D) contours are also plotted. The results show that the micropolar fluids contain microstructure having non-symmetric stress tensor and are useful in lubrication theory. Moreover, the thermal and concentration waves in CattaneoChristov models have a significance role in the laser heating and enhancement in thermal conductivity.     

2D and 3D Turbulent Fluctuations in Open Channel Flow with Re τ = 590 Studied by Large Eddy Simulation

Well-resolved 3D Large Eddy Simulations (LES) are presented for open channel flow at a Reynolds number Re _τ  = 590 based on friction velocity u _τ and water depth h. The results are depth-averaged and thereby information is obtained on the 2D horizontal fluctuations in the channel. The total turbulence is decomposed into 2D and 3D fluctuations and the energy content of these as well as their spectral distribution is studied. It is found that only 15% of the fluctuating energy is contained in the 2D fluctuations and that these are mostly of scales larger than the water depth while the 3D fluctuations are restricted by the limited vertical extent of the water body and have scales smaller than the water depth. Information is obtained on the dispersion terms arising from the depth-averaging procedure, and scalar transport due to a vertical line source of tracer is studied thereby investigating the contribution of the 2D and 3D fluctuations to the transverse mixing of the scalar.     

An efficient method for discretizing 3D fractured media for subsurface flow and transport simulations

We introduce a new method to discretize inclined non‐planar two‐dimensional (2D) fractures in three‐dimensional (3D) fractured media for subsurface flow and transport simulations. The 2D fractures are represented by ellipsoids. We first discretize the fractures and generate a 2D finite element mesh for each fracture. Then, the mesh of fractures is analyzed by searching and treating critical geometric configurations. Based on that search, the method generates a quality mesh and allows for including finer grids. A solute transport problem in fractured porous media is solved to test the method. The results show that the method (i) adequately represents the fractured domain by maintaining the geometric integrity of input surfaces and geologic data, (ii) provides accurate results for both simple and complex fractured domains, (iii) is insensitive to spatial discretization, and (iv) is computationally very efficient. For inclined and vertical fractures, analytical and numerical solutions are shown to be in good agreement. The method is therefore suitable to discretize fracture networks for flow and transport simulations in fractured porous media. Copyright © 2010 John Wiley & Sons, Ltd.     

A simple and efficient BEM implementation of quasistatic linear visco-elasticity

A simple yet efficient procedure to solve quasistatic problems of special linear visco-elastic solids at small strains with equal rheological response in all tensorial components, utilizing boundary element method (BEM), is introduced. This procedure is based on the implicit discretisation in time (the so-called Rothe method) combined with a simple “algebraic” transformation of variables, leading to a numerically stable procedure (proved explicitly by discrete energy estimates), which can be easily implemented in a BEM code to solve initial-boundary value visco-elastic problems by using the Kelvin elastostatic fundamental solution only. It is worth mentioning that no inverse Laplace transform is required here. The formulation is straightforward for both 2D and 3D problems involving unilateral frictionless contact. Although the focus is to the simplest Kelvin–Voigt rheology, a generalization to Maxwell, Boltzmann, Jeffreys, and Burgers rheologies is proposed, discussed, and implemented in the BEM code too. A few 2D and 3D initial-boundary value problems, one of them with unilateral frictionless contact, are solved numerically.     

Geometric and Hydrodynamic Characteristics of Three-dimensional Saturated Prefractal Porous Media Determined with Lattice Boltzmann Modeling

Fractal and prefractal geometric models have substantial potential for contributing to the analysis of flow and transport in porous media such as soils and reservoir rocks. In this study, geometric and hydrodynamic parameters of saturated 3D mass and pore–solid prefractal porous media were characterized using the lattice Boltzmann model (LBM). The percolation thresholds of the 3D prefractal porous media were inversely correlated with the fraction of micro-pore clusters and estimated as 0.36 and 0.30 for mass and pore–solid prefractal porous media, respectively. The intrinsic permeability and the dispersivity of the 3D pore–solid prefractals were larger than those of the 3D mass prefractals, presumably because of the occurrence of larger solid and pore cluster sizes in the former. The intrinsic permeability and dispersivity of both types of structure increased with increasing porosity, indicating a positive relationship between permeability and dispersivity, which is at odds with laboratory data and current theory. This discrepancy may be related to limitations of the convection dispersion equation at the relatively high porosity values employed in the present study.     

Comparison Between Equivalent Continuum and Discrete Crack Models of Transient Radon Transport at the Soil–Building Foundation Crack Interface Using TOUGH2/EOS7Rn

In the framework of radon risk management in France, it is necessary to enhance knowledge on radon transfer from its source to exposure areas (e.g., buildings) by developing simple, accurate, numerical models for transient radon transport in three-dimensional (3D) unsaturated porous materials. The equivalent continuum model (ECM) of flow and transport at the interface between the soil and cracks (fissures) in a building foundation (e.g., slab on grade, basement) is attractive, since equivalent (effective) continuum properties assigned to model cells can represent the combined effect of individual cracks and solid matrix of the cracked concrete of the foundation (slab and blocks walls). Although the ECM approach based on the volume averaging method has been used to model flow and transport through cracks at the soil–building interface, it has never been verified numerically. Thus, the goal of the present work is to develop an ECM using this averaging method and to quantify its uncertainties based on its comparison to an accurate numerical discrete crack model (DCM) for flow and transport in the crack. As a first step, the DCM implemented in the TOUGH2/EOS7Rn module has been verified numerically through a comparison to a reference 3D steady-state numerical solution for radon transport into a house with basement under constant negative pressure. Then, 3D results of the DCM and ECM approaches were compared, under time-dependent indoor–outdoor pressure differentials conditions, for two crack line configurations in the basement slab floor and two different soil configurations with different soil permeability and radium \(^{226}\)Ra mass content values. Results of this comparison show that, for a homogeneous soil configuration, discrepancies between ECM and DCM simulated indoor radon activity concentrations decrease with the increase in soil permeability, regardless crack line configuration in the slab floor and soil radium mass content. However, ECM uncertainties were not within the range of absolute errors on measured radon concentration for the higher soil permeability \((1\times 10^{-9}, 1\times 10 ^{-8} \hbox { m}^{2})\) and the higher \(^{226}\hbox {Ra}\) mass content values (4500 \(\hbox {Bq\;kg}^{-1})\), especially for high radon pics induced by sudden increase in indoor air pressure drop. Regardless soil \(^{226}\hbox {Ra}\) mass content and crack line configuration in the slab floor, the ECM showed to be conservative for the two-layered soil configuration with the presence of aggregates beneath the slab foundation, generally practiced in buildings constructions.