betaFeedback

WorksPhysics and Astronomy

Instability of a plane Poiseuille flow bounded between inhomogeneous anisotropic porous layers

Abstract

The linear stability analysis of a plane Poiseuille flow in a channel with anisotropic and inhomogeneous porous layers is performed. The effect of anisotropy and inhomogeneous permeability on the stability characteristics is addressed in detail. The stability characteristics of the anisotropy parameter (the ratio of permeability in the streamwise and the transverse direction) and the inhomogeneity function are presented in detail

Introduction

Stability analysis of the flow through multi-layer porous fluid systems is a subject that has been studied extensively for the past many decades. Various applications, ranging from geophysical, industrial, technological, and biological systems employ multi-layer porous flow systems [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17]. For instance, the permeable coastal sediment flows, flows in fractured geological strata [8, 9, 15, 16, 17], and gas flooding by injecting CO2\mathrm{CO}_2 or nitrogen to enhance oil recovery from oil reservoirs [1]. The manufacturing of cost-effective composite materials by resin film infusion (RFI), where resin flows, can be described by Darcy’s law. Such composite materials are used in aircraft or motor car production [2]. Another example of multi-layer porous-fluid systems is the blood flow through arteries, the gastrointestinal system, kidneys, and lungs [10, 13, 14]. For example, the stability analysis of blood flows in a large artery, by treating the large artery as a rigid channel with uniform width and the blood as an incompressible Newtonian fluid with variable viscosity due to transverse variation in hematocrit ratio, was performed by Makinde [10]. He observed that a transverse increase in the hematocrit ratio towards the arterial central region has a stabilizing effect.

Chang et al. [18] performed the foremost stability work on a pressure-driven fluid flow overlying a saturated porous region. They numerically solved the Navier–Stokes–Darcy model and identified three different instability modes. Liu et al. [19] solved the same problem using the Brinkman model and revealed two unstable modes in contrast to Chang et al. [18]. Hill and Straughan [20] adopted a more generalized model by considering three regions in which a Darcy–Brinkman porous media is introduced between the fluid and Darcy regions. The neutral stability curves are not found to be bimodal for specific parameter regimes. Detailed linear stability of a plane Poiseuille flow in a confined porous channel was analyzed by Tilton and Cortelezzi [21, 22] by considering the volume-averaged Navier–Stokes equations (VANS) equation [23] along with the shear stress conditions given by Ochoa-Tapia and Whitaker [24, 25]. Tilton and Cortelezzi [21, 22] suggested that inertial effects can be neglected if the ratio between the square root of permeability and channel half-height, i.e. the mean permeability, is minimal. They concluded that the small increment in wall-permeability decreases the critical Reynolds number compared to the plane Poiseuille flow between rigid walls. Some other studies reported instability mechanism of porous-fluid configuration in the presence or absence of heat transfer (see, e.g., Nield et al. [3], Straughan [26]).

In several practical situations and natural processes, a porous layer exhibits directional and spatial variation [27, 28]. Therefore, it is deemed necessary to consider the effects of anisotropy and inhomogeneity while exploring the stability characteristics of a porous-fluid system. Deepu et al. [29] analyzed the impact of anisotropic and inhomogeneous permeability on the linear stability results proposed by Chang et al. [18]. Their study reveals that a suitable wall-parallel or wall-normal permeability can control flow instability in porous channels. However, choosing the optimal model for governing porous-fluid channels remains unanswered [19].

In this regard, the main objective of the present study is to investigate the effect of anisotropic and inhomogeneous permeability on the flow instabilities in a confined porous channel in isothermal conditions by using the generalized Darcy’s equation along with the Beavers and Joseph [30] boundary condition. The paper follows the following structure. Section 2 describes the problem description and governing equations, the analytical expression of the base state solution, and the derivation of linear stability equations. The numerical method is explained in Sec. 3. The stability features for anisotropy and inhomogeneity are discussed in Sec. 4. Conclusions are in Sec. 5.

Mathematical formulation and governing equations

Consider a two-dimensional plane Poiseuille flow of an incompressible viscous fluid in a channel of height 2h2h bounded between two saturated porous layers of thickness 2h12h_1 and 2h22h_2. A constant pressure gradient drives the flow. The lengths are measured from the center line of the channel, as shown in Figure 1. The governing continuity and momentum balance equations in the

Image not extracted

Figure 1. Schematic of a two-dimensional plane Poiseuille flow in a channel bounded between two porous layers.

fluid layer spanning y∈(−h,h)y \in(-h,h) read

∇⋅u=0,(1)\nabla\cdot\boldsymbol{u}=0, \tag*{(1)}
ρ(∂u∂t+u⋅∇u)=−∇p+μ∇2u,(2)\rho\left(\frac{\partial\boldsymbol{u}}{\partial t}+\boldsymbol{u}\cdot\nabla\boldsymbol{u}\right)=-\nabla p+\mu\nabla^{2}\boldsymbol{u}, \tag*{(2)}

where tt denotes the time, xx and yy are the streamwise and transverse directions, respectively, u=(u,v)\boldsymbol{u}=(u,v), pp and μ\mu are the velocity vector, pressure and viscosity, respectively.

We consider highly dense porous layers, i.e., the permeability is low; hence the generalized Darcy’s law governs the momentum transport in the flow system [3]. The continuity and generalized Darcy equation in the porous layers spanning y∈(−h−2h1,−h)∪(h,h+2h2)y\in(-h-2h_{1},-h)\cup(h,h+2h_{2}) read

∇⋅uj=0,(3)\nabla\cdot\boldsymbol{u}_{j}=0, \tag*{(3)}
ρϵj∂uj∂t=−∇pj−μKuj,(4)\frac{\rho}{\epsilon_{j}}\frac{\partial\boldsymbol{u}_{j}}{\partial t}=-\nabla p_{j}-\frac{\mu}{\boldsymbol{K}}\boldsymbol{u}_{j}, \tag*{(4)}

where j=1j=1 and 22 denotes the lower and upper layers, respectively, ϵj\epsilon_{j} is the porosity, uj=(uj,vj)\boldsymbol{u}_{j}=(u_{j},v_{j}) is the volume averaged velocity vector, and pjp_{j} is the intrinsic averaged pressure. The permeability tensor K\boldsymbol{K} reads

K=(kjxηjx(y/h)00kjyηjy(y/h)),(5)\boldsymbol{K}=\begin{pmatrix} k_{jx}\eta_{jx}(y/h) & 0\\ 0 & k_{jy}\eta_{jy}(y/h) \end{pmatrix}, \tag*{(5)}

where kjxk_{jx} and kjyk_{jy} are the permeabilities in the xx- and yy-direction, respectively; ηjx\eta_{jx} and ηjy\eta_{jy} are the corresponding inhomogeneity functions depend only on yy. The anisotropy parameter ξj=kjx/kjy\xi_{j}=k_{jx}/k_{jy} is defined as the ratio of the permeabilities. Note that the inhomogeneity functions ηjx,ηjy\eta_{jx},\eta_{jy} can depend on xx and yy both; however, this gives rise to a two-dimensional base flow, thus making the problem complicated. A linear inhomogeneity function is used by Green and Freehill [31]. However, in the present analysis, an inhomogeneous exponential function [29, 32, 33] of the following form is employed,

ηjx=ηjy=eIj(1∓y/h),(6)\eta_{jx}=\eta_{jy}=e^{I_{j}(1\mp y/h)}, \tag*{(6)}

where IjI_{j} is the inhomogeneity factor. Note that at lower and upper porous-fluid interfaces, ηjx(∓h)=ηjy(∓h)=1\eta_{jx}(\mp h)=\eta_{jy}(\mp h)=1. And the permeabilities in the xx- and yy-directions increase (decrease) with negative (positive) values of IjI_{j} for both porous layers.

No penetration conditions are applied at the bottom y=−h−2h1y=-h-2h_{1} (j=1j=1) and top y=h+2h2y=h+2h_{2} (j=2j=2) of the porous layers,

vj=0.(7)v_{j}=0. \tag*{(7)}

The continuity of velocity and balance of normal and shear stresses are imposed at the interfaces, which, respectively, read

v=vj,p−2μ∂v∂y=pjand∂u∂y=±αjBJkjxηjx(∓h)(u−uj),(8)v=v_{j},\qquad p-2\mu\frac{\partial v}{\partial y}=p_{j}\qquad\text{and}\qquad\frac{\partial u}{\partial y}=\pm\frac{\alpha_{j}^{BJ}}{\sqrt{k_{jx}\eta_{jx}(\mp h)}}(u-u_{j}), \tag*{(8)}

where ηjx(∓h)=1\eta_{jx}(\mp h)=1, αjBJ\alpha_{j}^{BJ} denotes the Beaver–Joseph coefficient, which depends on the local structure of the porous materials near the fluid-porous interfaces [30, 34]. The positive and negative signs in (2.6c) stand for the interface of lower and upper porous layers, respectively. Note that Beavers and Joseph [30] condition (2.6c) is employed, along with the streamwise component of permeability [32].

Dimensionless governing equations and boundary conditions

The following scaling is used for non-dimensionalization of the flow variables

(x⋆,y⋆)=1h(x,y),t⋆=tUmh,u⋆=uUm,p⋆=pρUm2,uj⋆=ujUm,pj⋆=pjρUm2,j=1,2,(x^\star,y^\star)=\frac{1}{h}(x,y),\quad t^\star=\frac{tU_m}{h},\quad u^\star=\frac{u}{U_m},\quad p^\star=\frac{p}{\rho U_m^2},\quad u_j^\star=\frac{u_j}{U_m},\quad p_j^\star=\frac{p_j}{\rho U_m^2},\quad j=1,2,

where hh is the half height of the open channel layer and UmU_m is the mean channel velocity. After dropping ⋆\star, the non-dimensional form of (2.1)–(2.2) is

∂u∂t+(u⋅∇)u=−∇p+1Re∇2u,∇⋅u=0,(9)\frac{\partial u}{\partial t}+(u\cdot\nabla)u=-\nabla p+\frac{1}{Re}\nabla^2u,\quad\nabla\cdot u=0, \tag*{(9)}
\frac{\partial u_j}{\partial t}=-\epsilon_j\nabla p_j+\frac{1}{\sigma_j^2Re}\begin{pmatrix}\dfrac{1}{\eta_{jx}(y)}0 \tag*{(10)}
0\dfrac{\xi_j}{\eta_{jy}(y)}\end{pmatrix}\odot u_j,\quad\nabla\cdot u_j=0, \tag*{(11)}

where Re=ρUmh/μRe=\rho U_mh/\mu and σj=kjx/h\sigma_j=\sqrt{k_{jx}}/h are the Reynolds number and the mean permeability, respectively, and ξj=kjx/kjy\xi_j=k_{jx}/k_{jy} is the anisotropy parameter. Here, ⊙\odot represents the Hadamard product (or the element-wise product of two vectors), and the indices j=1j=1 and j=2j=2 refer to the lower −1−2d1<y<−1-1-2d_1<y<-1 and the upper 1<y<1+2d21<y<1+2d_2 porous layers, respectively. dj=hj/hd_j=h_j/h is the ratio of the thickness of the jthj^{\mathrm{th}} porous layer to the open channel layer. The non-dimensional interface conditions—continuity of velocity, normal and shear stresses conditions—at y=±1y=\pm1 become

v=vj,p−2Re∂v∂y=pjand∂u∂y=±αjBJσj1ηjx(∓1)(u−uj),(12)v=v_j,\quad p-\frac{2}{Re}\frac{\partial v}{\partial y}=p_j\quad\text{and}\quad\frac{\partial u}{\partial y}=\pm\frac{\alpha_j^{BJ}}{\sigma_j}\frac{1}{\sqrt{\eta_{jx}(\mp1)}}(u-u_j), \tag*{(12)}

and no-penetration conditions at y=±(1+2dj)y=\pm(1+2d_j) read as

vj=0.(13)v_j=0. \tag*{(13)}

Base flow

The undisturbed flow is assumed to be the steady [∂(⋅)/∂t][\partial(\cdot)/\partial t] and fully developed [∂(⋅)/∂x][\partial(\cdot)/\partial x] plane Poiseuille flow in both the channel and porous layers. Under these assumption solving the non-dimensionalized governing equations and boundary conditions (2.7)–(13), we get the base state velocity UB=[U2,U,U1]U_B=[U_2,U,U_1] and pressure [P2,P,P1][P_2,P,P_1] in each layer as

U2(y)=−2Aσ22η2x(y),1≤y≤1+2d2U_2(y)=-2A\sigma_2^2\eta_{2x}(y),\quad1\le y\le1+2d_2
U(y)=Ay2+By+C,−1≤y≤1U(y)=Ay^2+By+C,\quad-1\le y\le1
U1(y)=−2Aσ12η1x(y),−1−2d1≤y≤−1(14)U_1(y)=-2A\sigma_1^2\eta_{1x}(y),\quad-1-2d_1\le y\le-1 \tag*{(14)}
P(x)=P1(x)=P2(x)=1Re(11/6+C/A)x+contant(15)P(x)=P_1(x)=P_2(x)=\frac{1}{Re}\left(\frac{1}{1/6+C/A}\right)x+\text{contant} \tag*{(15)}

where

A=2(1/6−E/G),B=AF/G,C=AE/G,E=(α1BJ(χ12σ12+1)χ1σ1+1)(α2BJχ2σ2+1)+(α2BJ(χ22σ22+1)χ2σ2+1)(α1BJχ1σ1+1),F=α2BJσ2χ2−α1BJσ1χ1+α1BJα2BJ(χ1σ1χ2σ2−χ2σ2χ1σ1),G=α2BJσ2χ2(1+α1BJσ1χ1)+α1BJσ1χ1(1+α2BJσ2χ2),ηjx=ηjy=eIj(1+y),χ1=η1x(−1)=1, and χ2=η2x(1)=1.(16)\begin{aligned} &A=2(1/6-E/G),\quad B=AF/G,\quad C=AE/G,\\ &E=\left(\frac{\alpha_{1}^{BJ}\left(\chi_{1}^{2}\sigma_{1}^{2}+1\right)}{\chi_{1}\sigma_{1}}+1\right)\left(\frac{\alpha_{2}^{BJ}}{\chi_{2}\sigma_{2}}+1\right)+\left(\frac{\alpha_{2}^{BJ}\left(\chi_{2}^{2}\sigma_{2}^{2}+1\right)}{\chi_{2}\sigma_{2}}+1\right)\left(\frac{\alpha_{1}^{BJ}}{\chi_{1}\sigma_{1}}+1\right),\\ &F=\frac{\alpha_{2}^{BJ}}{\sigma_{2}\chi_{2}}-\frac{\alpha_{1}^{BJ}}{\sigma_{1}\chi_{1}}+\alpha_{1}^{BJ}\alpha_{2}^{BJ}\left(\frac{\chi_{1}\sigma_{1}}{\chi_{2}\sigma_{2}}-\frac{\chi_{2}\sigma_{2}}{\chi_{1}\sigma_{1}}\right),\\ &G=\frac{\alpha_{2}^{BJ}}{\sigma_{2}\chi_{2}}\left(1+\frac{\alpha_{1}^{BJ}}{\sigma_{1}\chi_{1}}\right)+\frac{\alpha_{1}^{BJ}}{\sigma_{1}\chi_{1}}\left(1+\frac{\alpha_{2}^{BJ}}{\sigma_{2}\chi_{2}}\right),\\ &\eta_{jx}=\eta_{jy}=e^{\mathcal{I}_{j}(1+y)},\\ &\chi_{1}=\sqrt{\eta_{1x}(-1)}=1,\text{ and }\chi_{2}=\sqrt{\eta_{2x}(1)}=1. \tag*{(16)} \end{aligned}

It is worth noticing that the base state velocity profile is independent of anisotropy parameter I\mathcal{I}. This is due to the fact that the yy-momentum equation for the porous layer is identically satisfied under the base flow assumptions.

Image not extracted

Figure 2. Base flow profile UB=[U2,U,U1]TU_B=[U_2,U,U_1]^{\mathsf{T}} in both the open channel and porous layers for three different values of inhomogeneity factor I1=I2=−2\mathcal{I}_1=\mathcal{I}_2=-2 (solid line), 00 (dashed line) and 22 (dotted line). The other parameters are fixed at σ1=σ2=0.001\sigma_1=\sigma_2=0.001, ϵ1=ϵ2=0.3\epsilon_1=\epsilon_2=0.3, α1BJ=α2BJ=0.1\alpha_1^{BJ}=\alpha_2^{BJ}=0.1. The inset shows the zoomed part in the upper porous region.

Figure 2 displays the base flow profile of identical porous layers (i.e., the mean permeability, porosity, Beavers–Joseph coefficient and inhomogeneity factor are the same in both the porous layers) for three different values of inhomogeneity factor I1=I2=I=−2,0\mathcal{I}_1=\mathcal{I}_2=\mathcal{I}=-2,0 and 22. Since all layers are of equal width d1=d2=1d_1=d_2=1, the velocity profile is symmetric about the center-line of the channel. A jump in the interface velocity exists for all considered inhomogeneity factors due to the choice of interface condition (2.15c) [30]. Note that for all values of inhomogeneity factor, the base flow profiles are parabolic and overlap in the open channel layer. However, the base velocity in the porous layers is uniform for the homogeneous porous layer (I=0\mathcal{I}=0) but varies for the inhomogeneity of the porous layers (I≠0\mathcal{I}\ne0). The base flow velocity profiles UjU_j in porous layers depend on the inhomogeneity factor, whereas velocity UU of the fluid layer is independent of inhomogeneity (I\mathcal{I}). This is due to the linear form of inhomogeneity function (2.4), as it has a unit function value at the interface y=±hy=\pm h, i.e., χ1=χ2=1\chi_1=\chi_2=1, see (2.12). As shall see in the following subsection that the coupled linear stability equations do not depend on the base state velocity UjU_j of the porous layers.

Linear stability analysis

The linear stability is examined by introducing an infinitesimal disturbance in the base flow (2.11). Only two-dimensional disturbances are considered due to the Squires theorem, according to which the most unstable perturbations are two-dimensional rather than three-dimensional [35]. A temporal linear stability analysis [36, 37] is performed by introducing a normal-mode ansatz

(u,v,uj,vj,p,pj)=(U,0,Uj,0,P,Pj)+(u^,v^,u^j,v^j,p^,p^j)eiα(x−ct),(17)(u,v,u_j,v_j,p,p_j)=(U,0,U_j,0,P,P_j)+(\hat{u},\hat{v},\hat{u}_j,\hat{v}_j,\hat{p},\hat{p}_j)e^{i\alpha(x-ct)}, \tag*{(17)}

where the hat variables are the amplitude of the disturbance; α\alpha is the real streamwise wavenumber, and cc is the complex phase-speed defined as c=cr+icic=c_r+ic_i. Note that a disturbance is unstable for αci>0\alpha c_i>0, stable for αci<0\alpha c_i<0, and neutrally stable for αci=0\alpha c_i=0. The disturbance equations are obtained by substituting (2.13) into (2.1)–(2.15) and retaining only the first-order perturbation terms in the resulting equations. Finally, the following set of linearized disturbance equations for vv and vjv_j are achieved after eliminating pressure and horizontal velocity from the disturbance equations

[(U−c)(D2−α2)−(D2U)]v^−1iαRe(D2−α2)2v^=0,(18)\left[(U-c)(D^2-\alpha^2)-(D^2U)\right]\hat{v}-\frac{1}{i\alpha Re}(D^2-\alpha^2)^2\hat{v}=0, \tag*{(18)}
ϵjiαRe[D2v^jηjx−ξjα2v^jηjy−DηjxDv^jηjy2]=cσj2(D2−α2)v^j,(19)\frac{\epsilon_j}{i\alpha Re}\left[\frac{D^2\hat{v}_j}{\eta_{jx}}-\xi_j\frac{\alpha^2\hat{v}_j}{\eta_{jy}}-\frac{D\eta_{jx}D\hat{v}_j}{\eta_{jy}^2}\right]=c\sigma_j^2(D^2-\alpha^2)\hat{v}_j, \tag*{(19)}

subject to the following boundary conditions

v^2=0at y=1+2d2,(20)\hat{v}_2=0 \qquad\text{at } y=1+2d_2, \tag*{(20)}
v^2=v^at y=1,(21)\hat{v}_2=\hat{v} \qquad\text{at } y=1, \tag*{(21)}
D2v^=α2BJ(Dv^−Dv^2)σ2η2x(1)at y=1,(22)D^2\hat{v}=\frac{\alpha_2^{BJ}(D\hat{v}-D\hat{v}_2)}{\sigma_2\sqrt{\eta_{2x}(1)}} \qquad\text{at } y=1, \tag*{(22)}
−D3v^Re+(3α2Re+iαU)Dv^−iα(DU)v^−1σ22Reη2x(1)Dv^2=iαcDv^−iαcϵ2Dv^2at y=1,(23)-\frac{D^3\hat{v}}{Re}+\left(\frac{3\alpha^2}{Re}+i\alpha U\right)D\hat{v}-i\alpha(DU)\hat{v}-\frac{1}{\sigma_2^2Re\eta_{2x}(1)}D\hat{v}_2=i\alpha cD\hat{v}-\frac{i\alpha c}{\epsilon_2}D\hat{v}_2 \qquad\text{at } y=1, \tag*{(23)}
−D3v^Re+(3α2Re+iαU)Dv^−iα(DU)v^−1σ12Reη1x(−1)Dv^1=iαcDv^−iαcϵ1Dv^1at y=−1,(24)-\frac{D^3\hat{v}}{Re}+\left(\frac{3\alpha^2}{Re}+i\alpha U\right)D\hat{v}-i\alpha(DU)\hat{v}-\frac{1}{\sigma_1^2Re\eta_{1x}(-1)}D\hat{v}_1=i\alpha cD\hat{v}-\frac{i\alpha c}{\epsilon_1}D\hat{v}_1 \qquad\text{at } y=-1, \tag*{(24)}
D2v^=−α1BJ(Dv^−Dv^1)σ1η1x(−1)at y=−1,(25)D^2\hat{v}=-\frac{\alpha_1^{BJ}(D\hat{v}-D\hat{v}_1)}{\sigma_1\sqrt{\eta_{1x}(-1)}} \qquad\text{at } y=-1, \tag*{(25)}
v^=v^1at y=−1.(26)\hat{v}=\hat{v}_1 \qquad\text{at } y=-1. \tag*{(26)}
v^1=0at y=−1−2d1(27)\hat{v}_1=0 \quad\text{at } y=-1-2d_1 \tag*{(27)}

where D=d/dyD=d/dy, D2=d2/dy2D^2=d^2/dy^2, D3=d3/dy3D^3=d^3/dy^3; v^\hat{v}, v^2\hat{v}_2, and v^1\hat{v}_1 are the amplitudes of the wall-normal velocity disturbances in the open channel layer, the upper porous and lower porous layers, respectively. For simplicity, we assume porous walls of identical heights and physical properties, i.e. d1=d2=dd_1=d_2=d, ϵ1=ϵ2=ϵ\epsilon_1=\epsilon_2=\epsilon, σ1=σ2=σ\sigma_1=\sigma_2=\sigma, α1BJ=α2BJ=αBJ\alpha_1^{BJ}=\alpha_2^{BJ}=\alpha^{BJ}, ξ1=ξ2=ξ\xi_1=\xi_2=\xi and I1=I2=II_1=I_2=I. The flow is controlled by the dimensionless parameters, namely, ReRe, dd, σ\sigma, ϵ\epsilon, αBJ\alpha^{BJ}, ξ\xi and II.

Numerical method

The coupled linear stability problem (2.14) along with the boundary conditions (20)–(27) is solved numerically using the Chebyshev spectral collocation method [36, 38]. We approximate the disturbance amplitudes v^2\hat{v}_2, v^\hat{v} and v^1\hat{v}_1 using the Chebyshev expansions in terms of NN Chebyshev polynomials as

v^2=∑n=0NanTn(y′),v^=∑n=0NbnTn(y′)andv^1=∑n=0NcnTn(y′),(28)\hat{v}_2=\sum_{n=0}^{N}a_nT_n(y'),\quad\hat{v}=\sum_{n=0}^{N}b_nT_n(y')\quad\text{and}\quad\hat{v}_1=\sum_{n=0}^{N}c_nT_n(y'), \tag*{(28)}

where ana_n, bnb_n and cnc_n are the unknown, called Chebyshev coefficients, to be determined, and Tn(y′)T_n(y') is the nn-th Chebyshev polynomial defined as

Tn(y′)=cos⁡(nθ),θ=cos⁡−1(y′),−1≤y′≤1.(29)T_n(y')=\cos(n\theta),\quad\theta=\cos^{-1}(y'),\quad-1\leq y'\leq1. \tag*{(29)}

As the Chebyshev polynomials Tn(y′)T_n(y') are defined over the domain [−1,1][-1,1], the following transformation is employed to transform the physical domain y∈[−3,3]y\in[-3,3] to the computational domain y′∈[−1,1]y'\in[-1,1]

y′={(y−d2−1)/d21≤y≤1+2d2y−1≤y≤1(y+d1+1)/d1−1−2d1≤y≤−1.(30)y'=\begin{cases} (y-d_2-1)/d_2 & 1\leq y\leq1+2d_2\\ y & -1\leq y\leq1\\ (y+d_1+1)/d_1 & -1-2d_1\leq y\leq-1. \end{cases} \tag*{(30)}

The expansions (28) are substituted into (2.14), and evaluated at the Gauss-Lobatto points yj=cos⁡(jπ/N)y_j=\cos(j\pi/N), which are extrema of the Chebyshev polynomials, where j=0,1,2,⋯ ,Nj=0,1,2,\cdots,N. The discretization of (2.14)–(2.15) leads to a generalized eigenvalue problem

AX=ωBX,(31)AX=\omega BX, \tag*{(31)}

where ω=αc\omega=\alpha c is the eigenvalue and XX is the associated eigenvector. Equation (31) represents 3N+33N+3 equations for the 3N+33N+3 unknown Chebyshev coefficients aja_j, bjb_j, cjc_j. To impose the boundary conditions, we follow Schmid and Henningson [36] and substitute the Chebyshev expansions (28) into (2.15). The eight boundary conditions are imposed by replacing the corresponding eight rows of AA and BB matrices of (31).

The corresponding disturbance amplitudes v^=[v^2,v^,v^1]T\hat{\mathbf{v}}=[\hat{v}_2,\hat{v},\hat{v}_1]^T is found using the Chebyshev coefficients and normalized by using the maximum fluid layer velocity (max⁡y∈[−1,1]∣v^∣)(\max_{y\in[-1,1]}|\hat{v}|). To check the convergence of the spectrum obtained numerically, a relative error is used as defined below

eN=∥ΩN+1−ΩN∥2∥ΩN∥2,(32)e_N=\frac{\lVert\Omega_{N+1}-\Omega_N\rVert_2}{\lVert\Omega_N\rVert_2}, \tag*{(32)}
Variation of relative error $e_N$ with the number $N$ of Chebyshev polynomials

Figure 3. The variation of relative error eNe_N with the number NN of Chebyshev polynomial in the log scale, when Re=2000Re = 2000, d=1d = 1, α=1\alpha= 1, σ=0.001\sigma= 0.001, ϵ=0.3\epsilon= 0.3, αBJ=0.1\alpha^{BJ} = 0.1, ξ=1\xi= 1 and I=0\mathcal{I} = 0.

where ∥⋅∥2\lVert\cdot\rVert_2 represents the L2L_2 norm, and ΩN\boldsymbol{\Omega}_N and ΩN+1\boldsymbol{\Omega}_{N+1} are vectors containing twenty least stable eigenvalues obtained using NN and N+1N+1 collocation points.

Figure 3 displays the variation of relative error eNe_N for isotropic ξ=1\xi= 1 and homogeneous I=0\mathcal{I} = 0 porous layers as a function of collocation points NN. It is seen that eNe_N saturates to O(10−6)O(10^{-6}) for N≥70N \ge70. For other values of ξ\xi and I\mathcal{I}, similar behavior of eNe_N is observed (figures not presented). In addition, at σ=0.001\sigma= 0.001, the thickness ratio dd has a negligible effect on the collocation points NN. Therefore, N≈80N \approx80 is adequate to obtain numerical accuracy at σ=0.001\sigma= 0.001.

Results and Discussion

The present numerical results are tested for validation with the available result of the classical plane Poiseuille flow [39]. Figure 4(a) depicts the neutral stability curve (ωi=0\omega_i = 0) in the (α,Re)(\alpha, Re)-plane for σ=10−5\sigma= 10^{-5}, with other parameters being set as ϵ=0.3\epsilon= 0.3, αBJ=0.1\alpha^{BJ} = 0.1, ξ=1\xi= 1 and I=0\mathcal{I} = 0. This set of parameters recovers stability results for classical plane Poiseuille flow. The flow is unstable (ωi>0\omega_i > 0) inside the neutral stability curve and stable (ωi<0\omega_i < 0) outside. The critical Reynolds number RecRe_c is found to be 3850, which occurs at wavenumber αc=1.02\alpha_c = 1.02. Note that the critical Reynolds number RecRe_c is defined as the lowest Reynolds number at which instability is triggered. Figure 4(b) depicts the eigenvalue and the eigenfunctions (inset) of the most unstable eigenvalue (marked with the red dot in the main panel) at the critical parameters (Rec,αc)=(3850,1.02)(Re_c,\alpha_c) = (3850, 1.02). The spectrum also shows the AA, PP and SS branches of the classical plane Poiseuille flow [36]. It is to be noted from the eigenfunction (see the inset) of the most unstable eigenvalue that the amplitude of the perturbation velocity vanishes at the interfaces and the porous regions; thus, we recover the TS wave observed in a channel flow with impermeable walls. These results show an excellent agreement between present results with mean permeability σ≈0\sigma\approx0 and the classical plane Poiseuille flow.

Following previous studies [19, 40, 41], we now fix porosity, mean permeability, and Beaver-Joseph coefficient to ϵ=0.3\epsilon= 0.3, σ=0.001\sigma= 0.001, and αBJ=0.1\alpha^{BJ} = 0.1, respectively. In Secs. 4.1 and 4.2, we analyze the effect of anisotropy ξ\xi and inhomogeneity I\mathcal{I} with respect to the depth ratio dd, respectively.

Image not extracted

Figure 4. (a) The neutral stability curve in the (α,Re)(\alpha,Re)-plane for σ=10−5\sigma= 10^{-5}; (b) Spectrum at the critical parameters Rec=3850Re_c = 3850, αc=1.02\alpha_c = 1.02. The inset shows the eigenfunction, ∣η∣|\eta| (Solid line), ℜ(v)\Re(\mathcal{v}) (dotted line), ℑ(v)\Im(\mathcal{v}) (dashed line), of the marked eigenvalue. The other parameters are fixed at ϵ=0.3\epsilon= 0.3, αBJ=0.1\alpha^{BJ} = 0.1, ξ=1\xi= 1, I=0\mathcal{I} = 0.

Effect of anisotropy

This subsection studies the effect of anisotropy ξ\xi with respect to the depth ratio dd when the porous layers are homogenous (i.e., ηx=ηy=1\eta_x = \eta_y = 1). Figure 5 displays the anisotropy effects on

Image not extracted

Figure 5. Variation of the growth rate ωi\omega_i with the wavenumber for ξ=0.001\xi= 0.001 (solid line), 1 (dash-dotted line) and 10 (dashed line) at d=1d = 1. Left and right panels represent results for (a) Re=4000Re = 4000, and (b) Re=6000Re = 6000.

the growth rate ωi\omega_i of the most unstable mode when d=1d = 1 for two values of Reynolds number Re=4000Re = 4000 [panel (a)] and Re=6000Re = 6000 [panel (b)]. For both these cases, the growth rates decrease as ξ\xi increases for a fixed wavenumber α\alpha. The configuration remains stable when Re=4000Re = 4000 and ξ≥1\xi\ge1 [panel a]. Moreover, at this ReRe, when the relative magnitude of the cross-stream permeability is larger (i.e., ξ=0.001\xi= 0.001), the growth rates become positive at wavenumber α≈O(1)\alpha\approx O(1), and the system becomes unstable. As soon as the Reynolds number ReRe increases to 6000, the growth rates gradually increase and become positive at each ξ\xi [panel b]. Thus, depending on the anisotropy parameter and Reynolds number, the system destabilizes in a bandwidth of unstable wavenumbers. It is also seen from Figure 5 that at fixed ReRe, decreasing and increasing the anisotropy parameter ξ\xi enhances and suppresses the growth rate of the most unstable mode and, therefore, increases and decreases the bandwidth of unstable wavenumbers.

To examine the effect of depth ratio dd, Figure 6 shows the neutral stability curves for four values of dd, with the anisotropy parameter being fixed at ξ=0.001\xi= 0.001 [panel a], 1 [panel b], 10 [panel c]. Note that the unstable regions become larger and the range of unstable wavenumbers increases as dd increases at any fixed ξ\xi. When ξ≥1\xi\ge1, the neutral stability curves almost coincide, and thereby the corresponding critical Reynolds numbers are practically insensitive to the depth ratio [panels b and c]. Furthermore, at each fixed depth ratio dd, the unstable region widens when the relative magnitude of the streamwise permeability is small [panel a] relative to the cross-stream permeability as against the case when the relative magnitude of the streamwise permeability are larger [panel c] relative to the cross-stream permeability. It is also evident from Figure 6 that a single mode characterizes the instability domain for all ξ\xi and dd considered here. Therefore, the neutral stability curves are unimodal, similar to the classical plane Poiseuille flow [36].

Image not extracted

Figure 6. Neutral stability curves in (α,Re)(\alpha, Re)-plane for various depth ratios dd with ξ\xi as (a) 0.001, (b) 1, and (c) 10.

For more understanding of criticality, the critical Reynolds number RecRe_c and critical wavenumber αc\alpha_c are obtained for various ξ\xi, and dd values, and illustrated in Table 1. The table indicates that RecRe_c reduces by 0.052% at ξ=0.001\xi= 0.001 compared to its value at ξ=1\xi= 1 for a small depth ratio d=1/8d = 1/8. The same trend is observed at higher d=4d = 4 (reduces 32.799% for ξ=0.001\xi= 0.001 compared to ξ=1\xi= 1) when the thickness of porous layers is significantly greater than the fluid layer. However, when the relative magnitude of the streamwise permeability is greater than the relative magnitude of the cross-stream permeability, i.e., when ξ>1\xi> 1, RecRe_c increases compared to the isotropic case (ξ=1\xi= 1) for all values of depth ratio dd. Particularly, for ξ=10\xi= 10, there is an increase of 0.378% and 6.788% at d=1/8d = 1/8 and d=4d = 4, respectively.

ddξ=0.001\xi= 0.001ξ=1\xi= 1ξ=10\xi= 10
RecRe_cαc\alpha_cRecRe_cαc\alpha_cRecRe_cαc\alpha_c
1/81/84509.380.98644511.710.98644528.780.9862
13860.961.01094205.730.99504476.480.9870
23384.051.03404192.150.99604476.480.9860
42816.991.05704191.890.99604476.430.9860

Table 1. The effect of anisotropy parameter ξ\xi and depth ratio dd on the critical Reynolds number and critical wavenumber.

Observe that the anisotropy parameter ξ\xi and the depth ratio dd offer a viable control technique for instability in a confined porous–fluid–porous channel. To suppress (enhance) the instability in such a system, one must design porous walls of various widths with the cross-stream permeability being smaller (more) than the streamwise permeability.

It is crucial to study the dominant mode (mode with maximum growth over all wavenumbers) of instability. In this regard, the dominant instability mode at (αc,Rec)=(1.0109,3860.96)(\alpha_c, Re_c) = (1.0109, 3860.96) is chosen for ξ=0.001\xi= 0.001 and d=1d = 1. For these parameters, the normalized eigenfunction [depicted in Figure 7(a)] displays the significant variation in the open channel layer; therefore, this dominant mode is a fluid-layer mode. The corresponding contour of stream function perturbation in the (x,y)(x, y)-plane is shown in Figure 7(b). Here x∈[0,3λ]x \in[0, 3\lambda] and y∈[−1−2d,1+2d]y \in[-1-2d, 1+2d], where λ\lambda represents the unit wavelength. Note that the variation of normalized eigenfunction generates a series of identically shaped vortices at the center of the open channel layer, which may be responsible for the primary instability. This fact fully supports the priority of the dominant fluid-layer mode on the primary instability, as predicted in Figure 6. Similarly, for the other values of ξ\xi and dd, the dominant mode of instability has an enormous variation in the open channel-layer.

Image not extracted

Figure 7. (a) The normalized eigenfunction, v^\hat{v} (Solid line), ℜ(v^)\Re(\hat{v}) (dotted line), ℑ(v^)\Im(\hat{v}) (dashed line), when ξ=0.001\xi= 0.001, Rec=3860.96Re_c = 3860.96 and αc=1.0109\alpha_c = 1.0109. (b) The contour plot of associated stream function in the (x,y)(x, y)-plane. Here λ=2π/αc\lambda= 2\pi/\alpha_c.

To summarize effect of anisotropy and depth ratio on the onset of instability, the variation of the critical Reynolds number RecRe_c and critical wavenumber αc\alpha_c with the anisotropy parameter ξ\xi for various depth ratio dd are displayed in Figure 8(a) and Figure 8(b), respectively. While RecRe_c increases as ξ\xi increases for all depth ratios considered here, the critical wavenumber αc\alpha_c decreases with the increase in ξ\xi. It is also seen from Figure 8(a) that at a fixed depth ratio dd, there is a cutoff value

Variation of the critical Reynolds number and critical wavenumber with the anisotropy parameter $\xi$ for various depth ratio $d$

Figure 8. Variation of the (a) critical Reynolds number and (b) critical wavenumber with the anisotropy parameter ξ\xi for various depth ratio dd.

ξm\xi_m of anisotropy parameter below which RecRe_c remains unchanged and above which it increases rapidly with increasing anisotropy parameter ξ\xi; therefore, the neutral stability curves for various ξ≤ξm\xi\le\xi_m values merge (figure not shown). Thus, the destabilizing effects of anisotropy parameter ξ\xi are insignificant with further reduction in ξ\xi from ξm\xi_m. Recall that a reduction in the anisotropy parameter ξ\xi is associated with either a relative increase in the permeability in the wall-normal direction (kyk_y) or a relative decrease in the permeability in the wall-parallel direction (kxk_x), which reduces the flow resistance, enhances the disturbance growth rate, and destabilizes the flow. The above observations reveal the significant destabilization of the system with anisotropic walls relative to the channel with porous walls as the relative magnitude of cross-stream permeability increases. Therefore, anisotropic permeability relative to isotropic permeability provides a potential strategy to control flow instability. In addition, the critical Reynolds number RecRe_c (the critical wavenumber αc\alpha_c) decreases (increases) for each fixed ξ\xi as dd increases. As the porous layer thickness increases (decreases) relative to the fluid layer thickness, the system becomes unstable (stable), and this is true for the Poiseuille flow in a channel with either isotropic or anisotropic walls.

Effect of inhomogeneity

This subsection studies the effect of inhomogeneity I\mathcal{I} concerning the depth ratio dd while keeping the anisotropy parameter ξ\xi fixed. It is worth noticing that changing I\mathcal{I} is equivalent to changing the overall permeability of the porous medium. It can be seen from (19) that in contrast to the anisotropy, the effect of inhomogeneity on the stability of the flow is more unpredictable since it generally appears in a complex way, as shown in this section.

Figure 9 shows the effect of the inhomogeneity factor on the growth rate of the most unstable mode when d=1d=1 for two values of the anisotropy parameter ξ=1\xi=1 [panel a] and ξ=0.001\xi=0.001 [panel b]. It is seen from Figure 9(a) that the growth rates enhance as I\mathcal{I} decrease for a fixed wavenumber α\alpha and thereby destabilizing the flow. This destabilization effect is striking for ξ<1\xi<1 when inhomogeneity factor decreases, see Figure 9(b). Therefore, for any fixed ξ\xi when compared to both homogeneous (I=0\mathcal{I} = 0) and inhomogeneous (I≠0\mathcal{I} \ne0) porous walls, decreasing and increasing the inhomogeneity factor I\mathcal{I} enhances and suppresses the growth rate of the most unstable mode, which increases and decreases the bandwidth of unstable wavenumbers.

Temporal growth rate as a function of wavenumber for various inhomogeneity parameters

Figure 9. The temporal growth rate ωi\omega_i as a function of the wavenumber α\alpha for various values of inhomogeneity parameter I\mathcal{I}; (a) ξ=1\xi= 1 and (b) ξ=0.001\xi= 0.001. The other parameters are fixed as Re=6000Re = 6000, d=1d = 1.

Since the linear stability results of anisotropy parameter ξ\xi are shown in section 4.1, we thus focus on the effect of inhomogeneity factor I\mathcal{I} when ξ=1\xi= 1 to recover inhomogeneous but isotropic porous medium. Figure 10 shows the neutral stability curves in the (α,Re)(\alpha, Re)-plane for three values

Neutral stability curves in the alpha-Re plane for different depth ratios

Figure 10. Neutral stability curves in the (α,Re)(\alpha, Re)–plane for different values of depth ratio with I=\mathcal{I} = (a) −1-1, (b) 00, and 11.

of inhomogeneity factor I∈{−1,0,1}\mathcal{I} \in\{-1, 0, 1\} and various values of the depth ratio dd. Similar to the anisotropy, the destabilizing effect of the depth ratio continues to exist as the unstable zones expand as dd increases. When d≥1d \ge1, the neutral stability curves almost coincide at each I\mathcal{I}, and thereby the associated critical Reynolds numbers are nearly insensitive to the depth ratio dd. It is also seen from Figure 10 that for each dd, the unstable regions become larger, and the range of unstable wavenumbers increases as the inhomogeneity factor I\mathcal{I} decreases. Similar behaviour of stability boundaries are verified for other values of anisotropy parameter ξ\xi.

The critical Reynolds number RecRe_c and critical wavenumber αc\alpha_c are obtained for various (I,d)(\mathcal{I},d) and listed in Table 2. The table indicates that RecRe_c reduces by 0.938% at I=−1\mathcal{I}=-1 compared to its value at I=0\mathcal{I}=0 (homogeneous) for a small depth ratio d=1/8d=1/8. The same trend is observed at higher d=4d=4 (reduces 5.026% for I=−1\mathcal{I}=-1 compared to I=0\mathcal{I}=0), when thickness of each porous layer is fourth times to the thickness fluid layer. However, for larger inhomogeneity factor I=1\mathcal{I}=1, there is an increase of 0.291% and 3.632% at d=1/8d=1/8 and d=4d=4, respectively.

ddI=−1\mathcal{I}=-1I=0\mathcal{I}=0I=1\mathcal{I}=1
RecRe_cαc\alpha_cRecRe_cαc\alpha_cRecRe_cαc\alpha_c
1/81/84469.390.98644511.710.98644524.850.9864
114001.391.00154205.730.99504348.260.9915
223981.371.00154192.150.99604344.090.9915
443981.191.00154191.890.99604344.150.9915

Table 2. Effect of inhomogeneity factor I\mathcal{I} and depth ratio dd on the critical parameters.

In the end, the variation of critical parameters RecRe_c and αc\alpha_c with inhomogeneity factor I\mathcal{I} for various depth ratios dd are displayed in Figure 11(a) and Figure 11(b), respectively. RecRe_c (αc\alpha_c) is a linear function of I\mathcal{I} with a very modest slope when d=1/8d=1/8, i.e., for very thin porous walls. It is seen that the critical Reynolds number RecRe_c (critical wavenumber αc\alpha_c) increases (decreases) as I\mathcal{I} increases for all depth ratios considered in the present study. The above-presented results are summarized as follows. A reduction in I\mathcal{I} translates to an increase in the overall permeability of the porous layers. Consequently, the flow rate increases, which reduces flow resistance, and hence flow becomes more unstable.

Variation of the critical Reynolds number and critical wavenumber with the inhomogeneity factor for various depth ratios

Figure 11. Variation of the (a) critical Reynolds number and (b) critical wavenumber with the inhomogeneity factor I\mathcal{I} for various depth ratio dd.

Conclusions and Outlook

Linear stability of a plane Poiseuille flow in a porous–fluid–porous configuration system with anisotropic and inhomogeneous permeability has been analyzed. The generalized Darcy’s law has been used to model the dynamics in the porous layers and Navier–Stokes equation to model the dynamics of the open fluid-layer. The base velocity has been found analytically. The linear stability of the base flow to two-dimensional infinitesimal disturbances has been analyzed. The resulting Orr–Sommerfeld eigenvalue problem has been solved numerically using the Chebyshev spectral collocation method and the QZ–algorithm. Our preliminary computations have revealed that the primary instability is triggered by the fluid-layer mode for a range of parametric values considered. Furthermore, the stability results of anisotropic and inhomogeneous permeability variation have been compared with the isotropic and homogeneous permeability. It has been found that the directional and spatial variation of permeability, i.e., anisotropy and inhomogeneity, respectively, affect the flow instability significantly. By suitably manipulating the porous walls, one can modify and control the stability characteristics of the confined porous channel flow.

The critical Reynolds number RecRe_c is higher (lower) with increasing (decreasing) the anisotropy parameter. Note that an increase in ξ\xi corresponds to either a relative increase in permeability in the wall-parallel direction (kxk_x) or a relative decrease in permeability in the wall-normal direction (kyk_y). Due to a rise in ξ\xi, the flow resistance becomes sufficiently higher, delaying the onset of instability. On the other hand, a relative increase in wall-normal permeability, i.e., decreasing ξ\xi reduces the flow resistance, thereby advancing the onset of instability compared to a confined isotropic porous channel. It has also been found that a decrease (increase) in inhomogeneity factor I\mathcal{I}, decreases (increases) the critical Reynolds number RecRe_c and destabilizes the flow. An inhomogeneous variation of permeability increases (decreases) the porous medium’s overall permeability and reduces (increases) the flow resistance, destabilizing (stabilizing) effects on the flow. Furthermore, present study has revealed that the channel with thicker porous walls is more unstable than the thinner porous walls.

Apart from the anisotropic and inhomogeneous permeability, and depth ratio of the porous wall in controlling the instability of the flow in a confined porous-fluid-porous, as demonstrated in the present study, there is at least one instance wherein the magnetic field acts as a control parameter. Such effects are commonly studied under magnetohydrodynamics. Several studies have demonstrated the stability of hydromagnetic steady flow between two parallel plates [42, 43, 44, 45]; those are either plane Poiseuille flow, generalized Couette–Poiseuille flow, or a flow through a permeable bed. The magnetic field intensity and a decrease in porous layer permeability while flowing through a porous bed has a stabilizing effect on the fluid flow. The inclusion of the effects of the magnetic field for the present case will be considered in the future.

The present investigation motivates the experimental study to examine the instability in a pressure–driven flow in confined porous channels. However, the nonlinear analysis generally gives more significant results and will be presented in the future. The present linear stability predictions can be utilized to avoid instability in practical applications such as manufacturing composite materials. Furthermore, the results from the present study can be used to manufacture and develop smaller types of equipment required for heat exchanges where instability is desirable.

Linear stability results for asymmetric velocity profiles

This appendix presents the linear stability results for asymmetric velocity profiles by considering the upper porous layer as impermeable and the lower porous layer as permeable. Figures

Image not extracted

Figure A.12. Comparison of (a) the critical Reynolds number RecRe_c as a function of anisotropy parameter ξ\xi when I=0\mathcal{I}=0 and (b) the critical Reynolds number RecRe_c as a function of inhomogeneity factor I\mathcal{I} when ξ=1\xi=1 in a channel with two porous walls (solid line, σ=0.001\sigma=0.001, d=1d=1) and a channel with only one porous wall (dashed line, σ1=0.001\sigma_1=0.001, d1=1d_1=1).

A.12 (a) and A.12 (b) show the stability results of a channel with an impermeable upper wall and a permeable lower wall characterized by the parameters ϵ1=0.3\epsilon_1=0.3, α1BL=0.1\alpha_1^{BL}=0.1, σ1=0.001\sigma_1=0.001 and d1=1d_1=1 (dashed line), and allow for a comparison presented in figures 8 and 11 for a channel with two porous walls (solid line). Compared to a channel with two permeable homogeneous (I=0\mathcal{I}=0) porous walls, the critical Reynolds number RecRe_c and the critical wavenumber αc\alpha_c for a homogeneous lower permeable wall are less and higher, respectively, as shown in figure A.12(a). Now, we set ξ=1\xi=1 and find RecRe_c and αc\alpha_c as a function of inhomogeneity factor I\mathcal{I} for a channel with two permeable porous walls and a single lower porous wall. Although the critical Reynolds number RecRe_c (critical wavenumber αc\alpha_c) grows (decreases) with inhomogeneity factor I\mathcal{I}, the critical Reynolds number RecRe_c (critical wavenumber αc\alpha_c) for a channel with a single porous wall is smaller (larger) than a channel with two porous walls for each I\mathcal{I}. From the above, it is concluded that the anisotropic and inhomogeneous permeability has less effect on the critical Reynolds number and critical wavenumber in a channel with a single porous wall than in a channel with two porous walls.

References

  1. [1]M. B. Allen, Collocation techniques for modeling compositional flows in oil reservoirs, Springer (1984). doi:10.1007/978-3-642-82213-1.DOI
  2. [2]D. C. Blest, S. McKee, A. K. Zulkifle, P. Marshall, Curing simulation by autoclave resin infusion, Compos. Sci. Technol. (1999). doi:10.1016/S0266-3538(99)00084-6.DOI
  3. [3]D. A. Nield, A. Bejan, et al., Convection in porous media, volume 3, Springer, 2006.
  4. [4]J. Hussong, W.-P. Breugem, J. Westerweel, A continuum model for flow induced by metachronal coordination between beating cilia, J. Fluid Mech. 684 (2011) 137–162. doi:10.1017/jfm.2011.282.DOI
  5. [5]R. E. Ewing, S. Weekes, Numerical methods for contaminant transport in porous media, Comput. Math. 202 (1998) 75–95.
  6. [6]F. Chen, C. F. Chen, Onset of finger convection in a horizontal porous layer underlying a fluid layer, Journal of Heat Transfer 110 (1988) 403–409. doi:10.1115/1.3250499.DOI
  7. [7]F. Chen, L. H. Hsu, Onset of thermal convection in an anisotropic and inhomogeneous porous layer underlying a fluid layer, J. Appl. Phys. 69 (1991) 6289–6301. doi:10.1063/1.348827.DOI
  8. [8]A. Goharzadeh, A. Khalili, B. J. Børgersen, Transition layer thickness at a fluid-porous interface, Physics of Fluids 17 (2005) 057102. doi:https://doi.org/10.1063/1.1894796.
  9. [9]B. Berkowitz, Characterizing flow and transport in fractured geological media: A review, Advances in water resources 25 (2002) 861–884. doi:https://doi.org/10.1016/S0309-1708(02)00042-8.
  10. [10]O. Makinde, Chebyshev collocation approach to stability of blood flows in a large artery, African Journal of Biotechnology 11 (2012) 9881–9887. doi:https://doi.org/10.5897/AJB11.3906.
  11. [11]N. Dukhan, Özer Baǧcı, M. Özdemir, Metal foam hydrodynamics: Flow regimes from pre-darcy to turbulent, International Journal of Heat and Mass Transfer 77 (2014) 114–123. doi:https://doi.org/10.1016/j.ijheatmasstransfer.2014.05.017.
  12. [12]Z. Xiao, Y. Zhao, Heat transfer coefficient of porous copper with homogeneous and hybrid structures in active cooling, Journal of Materials Research 28 (2013) 2545–2553. doi:10.1557/jmr.2013.190.DOI
  13. [13]J. Majdalani, C. Zhou, C. A. Dawson, Two-dimensional viscous flow between slowly expanding or contracting walls with weak permeability, Journal of Biomechanics 35 (2002) 1399–1403. doi:https://doi.org/10.1016/S0021-9290(02)00186-0.
  14. [14]H. N. Chang, J. S. Ha, J. K. Park, I. H. Kim, H. D. Shin, Velocity field of pulsatile flow in a porous tube, Journal of Biomechanics 22 (1989) 1257–1262. doi:https://doi.org/10.1016/0021-9290(89)90228-5.
  15. [15]J. J. Voermans, M. Ghisalberti, G. N. Ivey, The variation of flow and turbulence across the sediment–water interface, Journal of Fluid Mechanics 824 (2017) 413–437. doi:10.1017/jfm.2017.345.DOI
  16. [16]Y. Kuwata, K. Suga, Direct numerical simulation of turbulence over anisotropic porous media, Journal of Fluid Mechanics 831 (2017) 41–71. doi:10.1017/jfm.2017.619.DOI
  17. [17]M. E. Rosti, L. Brandt, A. Pinelli, Turbulent channel flow over an anisotropic porous wall – drag increase and reduction, Journal of Fluid Mechanics 842 (2018) 381–394. doi:10.1017/jfm.2018.152.DOI
  18. [18]Chang, Min-Hsing, Chen, Falin, S. Brian, Instability of poiseuille flow in a fluid overlying a porous layer, J. Fluid Mech. 564 (2006) 287–303. doi:10.1017/S0022112006001583.DOI
  19. [19]R. Liu, Q. S. Liu, S. C. Zhao, Instability of plane poiseuille flow in a fluid-porous system, Phys. Fluid 20 (2008) 104105. doi:10.1063/1.3000643.DOI
  20. [20]A. A. Hill, B. Straughan, Poiseuille flow in a fluid overlying a porous medium, J. Fluid Mech. 603 (2008) 137–149. doi:10.1017/S0022112008000852.DOI
  21. [21]N. Tilton, L. Cortelezzi, The destabilizing effects of wall permeability in channel flows: A linear stability analysis, Phys. Fluid 18 (2006) 051702. doi:10.1063/1.2202649.DOI
  22. [22]N. Tilton, L. Cortelezzi, Linear stability analysis of pressure-driven flows in channels with porous walls, J. Fluid Mech. 604 (2008) 411–445. doi:10.1017/S0022112008001341.DOI
  23. [23]S. Whitaker, The forchheimer equation: a theoretical development, Trans. Porous media 25 (1996) 27–61. doi:10.1007/BF00141261.DOI
  24. [24]J. A. Ochoa-Tapia, S. Whitaker, Momentum transfer at the boundary between a porous medium and a homogeneous fluid—i. theoretical development, Int. J. Heat Mass Transfer 38 (1995) 2635–2646. doi:10.1016/0017-9310(94)00346-W.DOI
  25. [25]J. A. Ochoa-Tapia, S. Whitaker, Momentum transfer at the boundary between a porous medium and a homogeneous fluid-ii. comparison with experiment, Int. J. Heat Mass Transfer 38 (1995) 2647–2655. doi:10.1016/0017-9310(94)00347-X.DOI
  26. [26]B. Straughan, Stability and wave motion in porous media, volume 165, Springer Science & Business Media, 2008. doi:https://doi.org/10.1007/978-0-387-76543-3.
  27. [27]B. Straughan, The onset of convection in porous media, Proc. R. Soc. Lond. Series A 452 (1996) 97–115. doi:10.1098/rspa.1996.0006.
  28. [28]M. Malashevich, M. Swamy, The onset of convection in a binary fluid saturated anisotropic porous layer, Int. J. Thermal Sc. 49 (2010) 867–878. doi:10.1016/j.ijthermalsci.2009.12.008.
  29. [29]P. Deepu, P. Anand, S. Basu, Stability of poiseuille flow in a fluid overlying an anisotropic and inhomogeneous porous layer, Phys. Rev. E 92 (2015) 023009. doi:10.1103/PhysRevE.92.023009.DOI
  30. [30]G. S. Beavers, D. D. Joseph, Boundary conditions at a naturally permeable wall, J. Fluid Mech. 30 (1967) 197–207. doi:10.1017/S0022112067001375. 1759–1762. doi:https://doi.org/10.1063/1.1657844.
  31. [32]F. Chen, Salt-finger instability in an anisotropic and inhomogeneous porous substrate underlying a fluid layer, Journal of applied physics 71 (1992) 5222–5236. doi:https://doi.org/10.1063/1.350579.
  32. [33]P. Deepu, S. Dawande, S. Basu, Instabilities in a fluid overlying an inclined anisotropic and inhomogeneous porous layer, Journal of Fluid Mechanics 762 (2015) R2. doi:10.1017/jfm.2014.644.DOI
  33. [34]B. Goyeau, D. Lhuillier, D. Gobin, M. Velarde, Momentum transport at a fluid–porous interface, Int. J. Heat Mass Transfer 46 (2003) 4071–4081. doi:10.1016/S0017–9310(03)00241–2.
  34. [35]H. B. Squire, On the stability for three-dimensional disturbances of viscous fluid flow between parallel walls, Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character 142 (1933) 621–628. doi:10.1098/rspa.1933.0193.DOI
  35. [36]P. J. Schmid, D. S. Henningson, Stability and transition in shear flows, Springer (2001). doi:10.1007/978–1–4613–0185–1.DOI
  36. [37]P. G. Drazin, W. H. Reid, Hydrodynamic stability, Cambridge university press, 2004.
  37. [38]C. Canuto, M. Hussaini, A. M. Quarteroni, T. A. Zang, Spectral Methods in Fluid Dynamics, Springer-Verlag, Berlin, 1988.
  38. [39]S. A. Orszag, Accurate solution of the orr–sommerfeld stability equation, Journal of Fluid Mechanics 50 (1971) 689–703. doi:10.1017/S002211207002842.DOI
  39. [40]M.-H. Chang, F. Chen, B. Straughan, Instability of poiseuille flow in a fluid overlying a porous layer, Journal of Fluid Mechanics 564 (2006) 287–303. doi:10.1017/S002211206001583.DOI
  40. [41]M.-H. Chang, Thermal convection in superposed fluid and porous layers subjected to a plane poiseuille flow, Physics of Fluids 18 (2006) 035104. doi:10.1063/1.2182004.DOI
  41. [42]O. Makinde, Magneto-hydrodynamic stability of plane-poiseuille flow using multi-deck asymptotic technique, Mathematical and Computer Modelling 37 (2003) 251–259. doi:https://doi.org/10.1016/S0895–7177(03)00004–9.
  42. [43]O. Makinde, P. Mhone, Temporal stability of small disturbances in mhd jeffery–hamel flows, Computers & Mathematics with Applications 53 (2007) 128–136. doi:https://doi.org/10.1016/j.camwa.2006.06.014.
  43. [44]O. D. Makinde, P. Mhone, On temporal stability analysis for hydromagnetic flow in a channel filled with a saturated porous medium, Flow, Turbulence and Combustion 83 (2009) 21–32. doi:https://doi.org/10.1007/s10494–008–9187–6.
  44. [45]O. Makinde, S. Motsa, Hydromagnetic stability of generalized plane couette flow, Far East J. Appl. Math 6 (2002) 77–88.

Paper details

vlnc:W_net9nkpx
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents