International Journal of Engineering Science 70 (2013) 124–134
Contents lists available at SciVerse ScienceDirect
International Journal of Engineering Science journal homepage: www.elsevier.com/locate/ijengsci
A general formulation of the reversible stress tensor for a nonlocal fluid Jiujiang Zhu a,⇑, John W. Crawford b, John W. Palfreyman a a b
SIMBIOS Centre, University of Abertay Dundee, Dundee DD1 1HG, UK Faculty of Agriculture, Food and Natural Resources, University of Sydney, NSW 2006, Australia
a r t i c l e
i n f o
Article history: Received 5 September 2011 Received in revised form 28 September 2012 Accepted 25 March 2013 Available online 29 May 2013 Keywords: Nonlocal fluid dynamics Generalized stress tensor Nonlocal functional variational principle Intermolecular interaction
a b s t r a c t The nonlocal stress tensor is an indispensable constitutive equation required to close the thermodynamic system of nonlocal fluid dynamics. A nonlocal functional variational principle is employed to derive a general expression for the thermodynamically reversible stress tensor for a two-phase, single component, nonlocal fluid. The Euler–Lagrange equation and Noether’s current are used to obtain the general form of the stress tensor, which is then used to derive a wide range of functional forms found in the literature. We also clarify some existing ambiguities. The general form of the nonlocal stress tensor is able to represent micro scale intermolecular interactions, and provides an efficient mesoscale numerical tool for multi-scale analysis using Lattice Boltzmann simulation. Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.
1. Introduction Nonlocal intermolecular interactions play a crucial role in the nucleation of phase transition and in the formation of interfaces (Drossinos & Kevrekidis, 2003; Llovell, Galindo, Blas, & Jackson, 2010; Roy, Rickman, Gunton, & Elder, 1998; Wertheim, 1976; Wu & Li, 2007). In the last few decades, these interactions have attracted significant attention not only from the point of view of theoretical studies aimed at understanding the mechanisms underlining phase transformation, but also in relation to a wide array of engineering problems, including fluid behavior in confining geometries (Tarazona, Marconi, & Evans, 1987), droplet and bubble formation (Khatavkar, Anderson, Duineveld, & Meijer, 2007a; van Giessen, Bukman, & Widom, 1997; Zhang & Kwok, 2005), contact line and wetting dynamics (Zhang & Kwok, 2004b; Zhang & Kwok, 2006), stability of capillary waves (Tarazona, Checa, & Chacon, 2007), and most recently, nanochannel flow and microfluidics (Li & Kwok, 2003). The challenges of dealing with nonlocal interactions are considerable and a wide range of different methods including continuum and particle-based approaches have been applied. the diffuse-interface model (Anderson, McFadden, & Wheeler, 1998; Anderson, Cermelli, Fried, Gurtin, & McFadden, 2007; Khatavkar et al., 2007a; Khatavkar, Anderson, & Meijer, 2007b); diffuse-interface model incorporating contact line motion (Ding & Spelt, 2007); Monte Carlo and molecular dynamics simulations (Galliero, 2010; Galliero et al., 2009; Yang, Fleming, & Gibbs, 1976); thermodynamic perturbation theory that incorporates molecular dynamics simulations (Zhou & Solana, 2009); density functional theory (Frink, Salinger, Sears, Weinhold, & Frischknecht, 2002; Li & Wu, 2006; Lovett & Baus, 1999; Patra, 2007; Reguera & Reiss, 2004; Tarazona et al., 2007; Tarazona et al., 1987; Wu & Li, 2007); Statistical associating fluid theory for variable range potential functions (Llovell et al., 2010); Van der Waals model dealing with Contact line and wetting dynamics (Sullivan, 1981); Dynamic van der Waals theory (Onuki, 2007); Integral equation method (Iatsevitch & Forstmann, 1997; Iatsevitch & Forstmann, 2000; Lovett, Mou, & Buff, 1976; Wertheim, 1976); Phase-field model based on order parameter (Chakraborty, 2007; Roy et al., 1998); Continuum ⇑ Corresponding author. Tel.: +01 2483828358. E-mail address:
[email protected] (J. Zhu). 0020-7225/$ - see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijengsci.2013.03.014
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
125
hydrodynamic model (Qian, Wu, Lei, Wang, & Sheng, 2009); Experimental observation with laser scanning confocal microscopy (Aarts, Schmidt, & Lekkerkerker, 2004); Observation using grazing-incidence X-ray scattering experiments (Fradin et al., 2000); along with others. The advantage of micro scale models is that they can explicitly incorporate molecular scale interfacial details and study the larger-scale properties that emerge as a consequence. However micro scale models are computationally demanding and, as such, they can only model a limited range of spatial scales. Coarse graining is required to model larger scales. Traditional macro scale continuum models use a constitutive equation for the local stress based purely on heuristic functions of strain and their derivatives. This ignores micro scale information, such as intermolecular interactions within the interface. There are extended continuum models that can take the changes in bulk material behavior near material interface into consideration (Slattery, Oh, & Fu, 2004; Slattery, Sagis, & Oh, 2007). However in order to incorporate subscale effects in the continuum mechanics level, these models have to be extended either through adjustable parameters or correction for long-range intermolecular forces (Slattery et al., 2004). Recently the Lattice Boltzmann method has emerged as a potential compromise for the numerical modelling of mesoscale processes to deal with interface and contact line dynamics involving phase transitions (Chang & Alexander, 2005; Davies, Summers, & Wilson, 2006; Gonnella, Lamura, & Sofonea, 2007; Li & Tafti, 2007, 2009; Nourgaliev, Dinh, Theofanous, & Joseph, 2003; van der Graaf, Nisisako, Schroen, van der Sman, & Boom, 2006; Zhang & Kwok, 2005; Zhang & Kwok, 2006). The Lattice Boltzmann method has also been applied to the numerical simulation of nanochannel flow and microfluidics (Li & Kwok, 2003; Zhang, 2010). In order to carry out a Lattice Boltzmann simulation in a fluid including a phase interface, the stress tensor of fluid has to be explicitly set out. The pressure tensor of complex fluids has been previously investigated by a number of groups. For example, Maurits, Zvelindovsky, and Fraaije (1998), Todd, Evans, and Daivis (1995), Todd and Hansen (2008) and Zhang and Kwok (2004a) calculated the pressure tensor from first principles using methods of statistical mechanics. Varea and Robledo (1996) derived a smooth nonlocal stress tensor, which was related to the local second order density gradient term. Romero-Rochin and Percus (1996) and Percus (1996) proposed an expression for the stress tensor and proved that its divergence may correspond to intermolecular forces. However, as discussed in detail in Section 4.7, some formulations of the nonlocal stress tensor in the literature have important limitations (e.g. in the formulation of the nonlocal stress tensor by Zhang, Li, & Kwok (2004) and Zhang & Kwok (2004b) one term was missing). In order to explicitly connect the properties of intermolecular interactions within the phase interface to macroscopic flow properties using the Lattice Boltzmann method, a multiscale approach has to be employed. Ideally the stress tensor should be able to abstract microscopic nonlocal correlations and map these onto a mesoscale continuum model. Li and Tafti (2007) explicitly derived a nonlocal pressure equation using mean-field free energy theory, however the proposed formulation is only suitable for that case. A generalized expression for the stress tensor for a nonlocal fluid is still missing and this is an important impediment to practical engineering application. A nonlocal continuum field theory has been established by Eringen (1972) and Eringen (2002), who decomposed the constitutive dependent variables into two parts, namely a static (equilibrium or reversible) part and a dynamic (dissipative) part. However Eringen (2002) made an additional simplifying assumption that the static part of stress is a local tensor. Thus the mechanism for calculating the static part of stress tensor of a nonlocal fluid is still an open problem. The aim of this paper is to derive a general expression for the reversible part of the stress tensor corresponding to a nonlocal fluid by means of the variational principle, so that the full constitutive equation of the nonlocal continuum field is addressed. Since the part the stress tensor corresponding to dissipation is not considered here, we can restrict ourselves to the study of isothermal thermodynamic equilibrium processes under Eringen’s nonlocal continuum field framework. In spite of the numerical efficiency of the Lattice Boltzmann approach, it still places significant computational demands. Because of this, attention has focused on models for single component multiphase flow, because it minimizes the demands on computer memory. The present paper considers the simple but important case, of single component two phase flow. Section 2 first provides a brief review of Eringen’s framework, and then the nonlocal variational principle is used to determine the extremum of the generalized action functional to derive the explicit form of the generalized stress tensor. Section 3 uses the generalized form to derive the special case corresponding to a local fluid. Section 4 then investigates a number of cases as validation, and demonstrates the application of the new formulation. 2. Nonlocal variational principle 2.1. Nonlocal fluid dynamics framework The framework for generalized nonlocal fluid dynamics can be found elsewhere in the literature, for example the book by Eringen (2002) includes details of the constitutive equations of memory-dependent nonlocal thermoviscous fluids. In summary the nonlocal stress tensor could be written as
TðxÞ ¼ R TðxÞ þ D TðxÞ
ð1Þ
in which RT is the reversible element of the stress and DT is the dissipation (or viscous) element of stress, the component form of DT may be written as (Eringen, 2002) D T kl ðxÞ ¼
Z
v
3
½kðjy xjÞdmm ðyÞdkl þ gðjy xjÞdkl ðyÞd y
ð2Þ
126
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
where the deformation-rate tensor
dkl ðyÞ ¼
1 1 @ v k ðyÞ @ v l ðyÞ ½v k;l þ v l;k ¼ þ 2 2 @yl @yk
ð3Þ
The viscosity coefficients k(jy xj) and g (jy xj) represent the nonlocal effects of the deformation-rate tensor at location y on the viscous stress tensor at location x. The total stress tensor at location x is then includes the integral of nonlocal contributions over the whole material space. If the stress at location x does not depend on any neighbourhood y but only on x, then the viscosity coefficients take a local form such that
kðjy xjÞ ¼ k0 dðy xÞ
ð4Þ
gðjy xjÞ ¼ g0 dðy xÞ in which d() is the Dirac delta function, Eq. (2) then becomes D T kl ðxÞ
¼ k0 dmm ðxÞdkl þ g0 dkl ðxÞ
ð5Þ
which is the constitution equation corresponding to a local fluid. Thus the viscous stress in Eringen’s framework was fully expressed in terms of nonlocal form. However in Eringen’s theory (Eringen, 2002; Eringen, 1972) the reversible part of the stress RT was assumed to take local form only. The aim here is to extend reversible stress to include nonlocal interaction. 2.2. Dual variational principle Consider the total nonlocal free energy or dual action functional
W½q ¼
Z Z v
v
3
3
L½y; x; qðyÞ; qðxÞ; ry qðyÞ; rx qðxÞd yd x
ð6Þ
In which, the dual grand potential L½y; x; qðyÞ; qðxÞ; ry qðyÞ; rx qðxÞ is the generalized nonlocal Lagrangian, where q (x) and q(y) are the density function evaluated at positions x and y respectively, and both x and y are integrated over whole body under consideration. The variation of the dual action functional Eq. (6) then reads as:
dW½q ¼
Z Z v
@L @L @L @L dqðyÞ þ ry dqðyÞ þ dqðxÞ þ rx dqðxÞ d3 yd3 x @ðry qðyÞÞ @ qðxÞ @ðrx qðxÞÞ v @ qðyÞ
ð7Þ
Eq. (7) may be rewritten in the form
dW½q ¼
Z Z " v
v
# @Ls @Ls dqðxÞ þ rx dqðxÞ d3 yd3 x @ qðxÞ @ðrx qÞ
ð8Þ
in which
Ls ¼ Ls ½x; y; qðxÞ; qðyÞ; rx qðxÞ; ry qðyÞ ¼ L½y; x; qðyÞ; qðxÞ; ry qðyÞ; rx qðxÞ þ L½x; y; qðxÞ; qðyÞ; rx qðxÞ; ry qðyÞ
ð9Þ
and is equal to two times the symmetrical part of L. For simplicity, in the remainder of the paper, Ls ½x; y; qðxÞ; qðyÞ; rx qðxÞ; ry qðyÞ is denoted as Ls only. Obviously Ls satisfies
Ls ½y; x; qðyÞ; qðxÞ; ry qðyÞ; rx qðxÞ ¼ Ls ½x; y; qðxÞ; qðyÞ; rx qðxÞ; ry qðyÞ
ð10Þ
Then the Lagrangian functional is defined as
L½x; qðxÞ; rx qðxÞ ¼
Z 1 3 Ls ½x; y; qðxÞ; qðyÞ; rx qðxÞ; ry qðyÞd y 2 v
ð11Þ
where the density function satisfies the mass conservation equation i.e.
Z
qðxÞd3 x ¼ const
ð12Þ
v
Finally, the total Lagrangian functional L is defined as
LT ½x; qðxÞ; rx qðxÞ ¼ L½x; qðxÞ; rx qðxÞ lqðxÞ
ð13Þ
in which l is the Lagrange multiplier relating to Eq. (12), and the physical interpretation of l is the chemical potential. The nonlocal variational principle then reads
Z 3 d LT ½x; qðxÞ; rx qðxÞd x ¼ 0
ð14Þ
v
Note that rxdq(x) and dq (x) are not independent variables. Through manipulation of the partial integration, Eq. (14) leads to the Euler–Lagrange equation
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
rx
dL½x; qðxÞ; rx qðxÞ dL½x; qðxÞ; rx qðxÞ þl¼0 dðrx qðxÞÞ dqðxÞ
127
ð15Þ
in which,
Z dL½x; qðxÞ; rx qðxÞ @Ls ½x; y; qðxÞ; qðyÞ; rx qðxÞ; ry qðyÞ 3 d y ¼ dðrx qðxÞÞ @ðrx qðxÞÞ v
ð16Þ
Z dL½x; qðxÞ; rx qðxÞ @Ls ½x; y; qðxÞ; qðyÞ; rx qðxÞ; ry qðyÞ 3 d y ¼ dqðxÞ @ qðxÞ v
ð17Þ
and
For a process in thermodynamic equilibrium, Noether’s theorem (Noether, 1971; Tavel, 1971) may be extended to the nonlocal case. Noether’s current then gives the nonlocal reversible stress tensor RT
¼ LT I rx qðxÞ
@LT @ðrx qðxÞÞ
ð18Þ
Substituting Eq. (15) into Eq. (13)
LT ½x; qðxÞ; rx qðxÞ ¼ L þ qðxÞ rx
dL
dðrx qðxÞÞ
dL dqðxÞ
ð19Þ
Substituting Eq. (19) into Eq. (18)
T ¼ LI þ q ðxÞ r R x
dL
dðrx qðxÞÞ
dL @L I rx qðxÞ dqðxÞ @ðrx qðxÞÞ
ð20Þ
Substituting Eqs. (16), (17) and (11) into Eq. (20) R TðxÞ
¼
1 2
Z v
Z 3 Ls d y I þ qðxÞ rx
Z Z @Ls @Ls 3 1 @ 3 3 d y d y I rx qðxÞ Ls d y 2 @ðrx qðxÞÞ v v @ðrx qðxÞÞ v @ qðxÞ
ð21Þ
It should be emphasized again that the brief notation Ls in Eq. (21) stands for Ls ½x; y; qðxÞ; qðyÞ; rx qðxÞ; ry qðyÞ, defined in Eq. (9). 2.3. Commentary To investigate a smooth attenuated neighbourhood fluid, it may be of interested to extend the constitutive variables set to include up to the nth order of the density gradient, i.e. fy; x; qðyÞ; qðxÞ; ry qðyÞ; rx qðxÞ; ; rny qðyÞ; rnx qðxÞg. From a mathematical point of view, this extension does not introduce additional complexity, since the density and its gradients are not constitutive independent variables in the variation process. The nth order density gradient can be converted to a density through n partial integration steps, which leads to the nth order Euler–Lagrange equation. A special case of this kind of fluid is the so-called smooth attenuated nonlocal fluid, or more simply, the smooth nonlocal fluid, where the free energy at location x depends on fx; qðxÞ; rx qðxÞ; ; rnx qðxÞg only. Eq. (21) sets out a general formulation of the reversible part of the stress tensor for nonlocal fluids. This is as same as the dissipation component of the nonlocal stress tensor (or viscous stress) in Eringen’s constitutive theory, where the stress tensor at each point may interact globally with all points of the body. This nonlocal interaction is expressed as an integration over whole body. The great advantage of this integral constitutive equation is that the integration is carried out in real physical space, which is convenient for practical engineering applications. Comparing Eq. (21) with earlier forms of the integral constitutive equations for nonlocal fluids (Romero-Rochin & Percus, 1996; Percus, 1996), the previous stress tensors were expressed in terms of the integration of a perturbation parameter k over an interval (0,1). The physical interpretation of the perturbation parameter k is unclear, which presents challenges for practical applications, and in addition the interpretation of the divergence of the previous stress tensor in terms of intermolecular force may not be unique (; Romero-Rochin & Percus, 1996). Section 4 will demonstrate that all stress tensors found in the current literature are special cases of the general formula Eq. (21). In contrast to the constitutive equations proposed by Romero-Rochin and Percus (1996) and Percus (1996) which apply only to smooth attenuate nonlocal fluids, the form in Eq. (21) applies to local, smooth attenuate nonlocal (refer to the next section) and real nonlocal fluids. 3. Local fluid Here, a special case that of the local fluid with density gradient effects, the smooth nonlocal fluid, is considered. For smooth nonlocal fluids, the dual grand potential is assumed to take the form
L½y; x; qðyÞ; qðxÞ; ry qðyÞ; rx qðxÞ ¼ L½x; qðxÞ; rx qðxÞdðy xÞ
ð22Þ
128
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
in which d() is the delta function, Eq. (11) then reduces to
L½x; qðxÞ; rqðxÞ ¼ L½x; qðxÞ; rqðxÞ
ð23Þ
i.e. L½x; qðxÞ; rx qðxÞ is the local Lagrangian. Also Eq. (6) can be reduced to a local action function of the form
Z
W½q ¼
v
3
L½x; qðxÞ; rqðxÞd x
ð24Þ
The Euler–Lagrange equation Eq. (15) then becomes
r
dL½x; qðxÞ; rqðxÞ dL½x; qðxÞ; rqðxÞ þk¼0 dðrqðxÞÞ dqðxÞ
ð25Þ
and Eq. (21) becomes
@L½x; qðxÞ; rqðxÞ @L½x; qðxÞ; rqðxÞ 1 @L½x; qðxÞ; rqðxÞ I rqðxÞ T ¼ L½x; q ðxÞ; r q ðxÞI þ q ðxÞ r R @ðrqðxÞÞ @ qðxÞ 2 @ðrqðxÞÞ ð26Þ 4. Some special cases In this section, the general stress tensor formulation obtained in section 2 is used to study certain special cases found in the literature. 4.1. Ideal gas An ideal gas is a theoretical gas composed of a set of randomly-moving, non-interacting point particles. The ideal gas obeys the ideal gas law, a simplified equation of state. For ideal gas, assume that
L ¼ Lid ½qdðy xÞ ¼ qðxÞfid ½qðxÞdðy xÞ
ð27Þ
Lid ½q ¼ qkT ln q
ð28Þ
fid ½qðxÞ ¼ kT ln q
ð29Þ
where
and
And k is the specific gas constant, such that
k¼
R R kB ¼ ¼ M mNA m
ð30Þ
in which R is the universal gas constant, Na is Avogadro’s constant, kB is Boltzmann’s constant, and m is the mass of single molecule. Substituting Eq. (28) into Eq. (26) gives
R TðxÞ
¼
Lid ½qðxÞ qðxÞ
@Lid ½qðxÞ I @ qðxÞ
ð31Þ
Denoting
Pid ¼ q
@Lid ½q @fid ½q Lid ½q ¼ q2 ¼ qkT @q @q
ð32Þ
Introducing the number of moles n ¼ mN , where N is total number of molecules, and noting that q ¼ mN , then using Eq. (30), M V Eq. (32) may be rewritten as
Pid V ¼ nRT
ð33Þ
Thus RT
¼ Pid I ¼ qkTI
ð34Þ
4.2. Van der Waal’s Gas A Van der Waal’s Gas is assumed to satisfy Van der Waal’s Gas equation, which is based on a modification of the ideal gas law and approximates the behavior of real fluids, taking into account the nonzero size of molecules and the attraction between them. For Van der Waal’s Gas taking
L ¼ Lv dw ½qdðy xÞ ¼ qðxÞfv dw ½qðxÞdðy xÞ
ð35Þ
129
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
where
Lv dw ½q ¼ qkT ln
q 1 bq
aq2
ð36Þ
aq
ð37Þ
and
fv dw ½qðxÞ ¼ kT ln
q 1 bq
where a and b are Van der Waal’s constants. Then substituting Eq. (35) into Eq. (26) gives RT
¼ Pv dw I
Pv dw ¼ kT
ð38Þ
q 1 bq
aq2
ð39Þ
4.3. Carnahan–Starling fluid The Van der Waal’s equation Eq. (37) may be generalized as
f ½q ¼ kT ln q þ bkT
Z
vðqÞdq aq
ð40Þ
which gives rise to the relationship
(
P ¼ phs ðqÞ aq2
ð41Þ
phs ðqÞ ¼ qkT½1 þ bqvðqÞ where phs(q) is the pressure of a hard sphere reference fluid. For example, for a Carnahan–Starling fluid
v¼
1 bq=8
phs ðqÞ ¼ qkT (
ð42Þ
ð1 bq=4Þ3 " # 1 þ h þ h2 h3
ð43Þ
ð1 hÞ3
h ¼ bq=4
ð44Þ
3
b ¼ 2pd =3m
in which d and m are the diameter and mass of the molecules, respectively. The free energy corresponding to Eq. (42) or Eq. (43) is consistent with the result given by van Giessen et al. (1997)
(
L ¼ Lcs ½qdðy xÞ ¼
"
qkT ln q þ
#
3 2h ð1 hÞ
)
1 lq aq2 þ pbulk dðy xÞ 2
ð45Þ
Eq. (40) may also be used to describe other fluids given the form of the function v(q). For example, for a dense gas, v was given by Chapman and Cowling (1970)
5 8
v ¼ 1 þ ðbqÞ þ 0:2869ðbqÞ2 þ 0:1103ðbqÞ3 þ 0:0386ðbqÞ4 þ
ð46Þ
4.4. Local fluid with density gradient effects (smooth nonlocal fluid) To illustrate the results of Section 3, and similar to Abraham (1979), Nadiga and Zaleski (1996), Swift, Osborn, and Yeomans (1995), Anderson et al. (1998),ee and Lin (2003, 2005) and Chang and Alexander (2005), we assume Eq. (22) takes the form
L½qðyÞ; qðxÞ ¼
1 2
qðxÞf ½qðxÞ þ j½rx qðxÞ2 dðy xÞ
ð47Þ
in which j is a constant capillary coefficient. Substituting Eq. (47) into Eq. (26) gives RT
1 ¼ P 0 þ jðrqÞ2 þ jqr2 q I jrq rq 2
ð48Þ
In which P0 is given by
P0 ¼ q2 ðxÞ
@f ½qðxÞ @q
ð49Þ
130
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
and is the thermodynamic pressure. In practice this could take a variety of forms depending on the structure of the Helmholtz free energy f[q]. Eq. (48) is known as Kerteweg’s stress tensor. This may be rewritten in the form of a pure pressure part plus a pure shear stress part as RT
1 ¼ P0 jðrqÞ2 þ jqr2 q I þ j½ðrqÞ2 I rq rq 2
ð50Þ
The matrix form of the second part of the right hand side of Eq. (50) is then
0
@@xq
0
B
@q @q j½ðrqÞ2 I rq rq ¼ jB @ @y @x
@q @y
0
@@zq @@xq
@@zq @@yq
@@xq
@q 1 @z
@@yq
@q C @z A
C
ð51Þ
0
which represents a pure shear stress produced by the density gradient. For a plane interface, all variables depend only on z and not x or y. Under this particular condition Eq. (51) equals zero. However in the most general cases, the shear stress component is non-trivial. The stress tensor, Eq. (50), was introduced to the Lattice Boltzmann scheme by Swift et al. (1995) to model the hydrodynamics of phase separation in two phase flow. In their model, a nonlocal term was added to the equilibrium distribution, so that the Navier–Stokes–Kerteweg equation could be recovered from the Lattice Boltzmann equation directly by performing a Chapman–Enskog expansion. Recently Lee and Lin (2003, 2005) and Chang and Alexander (2005) provided a standard Lattice Boltzmann model to simulate the diffusion interface in two phase flow. It can be proved using the Chapman–Enskog scheme that the standard Lattice Boltzmann procedure recovers the ideal gas Navier–Stokes equation. Thus in their Lattice Bhatnagar–Gross–Krook equation in the Lattice Boltzmann model, the intermolecular interaction force
F ¼ rðqc2s Þ þ r ðR TÞ
ð52Þ
was introduced to modify the momentum equation from the ideal gas Navier–Stokes equation to the Navier–Stokes–Kerteweg equation, where qc2s is the pressure of an ideal gas. Substituting Eq. (50) into Eq. (52), the intermolecular interaction force may be rewritten as
F ¼ rðqc2s P 0 Þ þ jqrðr2 qÞ
ð53Þ
in which P0 is given by Eq. (49). 4.5. Nonlocal fluid without gradient effects A particular case is a nonlocal fluid in the absence of a density gradient. In this case, Eq. (6) is reduced to
W½q ¼
Z Z v
v
3
3
L½y; x; qðyÞ; qðxÞd yd x
ð54Þ
Eq. (9) then reads
Ls ¼ L½y; x; qðyÞ; qðxÞ þ L½x; y; qðxÞ; qðyÞ
ð55Þ
and Eq. (21) becomes RT
¼
Z Z 1 @Ls 3 3 Ls d y qðxÞ d y I 2 v v @ qðxÞ
ð56Þ
The reversible stress now reduces to a pure pressure part. Introducing the thermodynamic pressure
P ¼ qðxÞ
Z
Z @Ls 3 1 3 d y Ls d y 2 v v @ qðxÞ
ð57Þ
Eq. (56) may then be rewritten as RT
¼ PI
ð58Þ
4.6. Van-Kampen nonlocal fluid As a case of Section 4.5, following Li and Tafti (2007), Sullivan (1981) and van Kampen (1964), we assume the dual grand potential takes the special form
1 L½y; x; qðyÞ; qðxÞ ¼ L0 ½qðxÞdðy xÞ þ wðjy xjÞ½qðyÞ qðxÞ2 4
ð59Þ
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
131
Substituting the first part of the right hand side of Eq. (59) into Eq. (57) gives
P0 ¼ qðxÞ
@L0 ½qðxÞ L0 ½qðxÞ @q
ð60Þ
which is the same as Eq. (27) of Li and Tafti (2007). Noting that for the second part of Eq. (59) (denote as L0 )
L0s ¼ 2L0 ¼
1 wðjy xjÞ½qðyÞ qðxÞ2 2
ð61Þ
Substituting Eq. (61) into Eq. (57) we find 0 RT
Z Z 1 @L0s 3 3 ½ L0s d yI þ qðxÞ d y I 2 v v @ qðxÞ Z Z 1 3 3 ¼ wðjy xjÞ½qðyÞ qðxÞ2 d y I þ qðxÞ wðjy xjÞ½qðyÞ qðxÞd y I 4 v v ¼
ð62Þ
Eq. (60) together with Eq. (62) gives RT
Z @L0 ½qðxÞ 1 3 ¼ qðxÞ L0 ½qðxÞ I þ wðjy xjÞ½qðyÞ qðxÞ2 d y I @q 4 v Z 3 þ qðxÞ wðjy xjÞ½qðyÞ qðxÞd y I
ð63Þ
v
or
P ¼ qðxÞ
Z Z @L0 ½qðxÞ 1 3 3 L0 ½qðxÞ wðjy xjÞ½qðyÞ qðxÞ2 d y qðxÞ wðjy xjÞ½qðyÞ qðxÞd y @q 4 v v
ð64Þ
which is the same as that given by Li and Tafti (2007). If we assume L0 ½qðxÞ is expressed as
L0 ½qðxÞ ¼ qðxÞf ½qðxÞ
ð65Þ
Substituting Eq. (65) into Eq. (60) we have
P0 ¼ q2 ðxÞ
@f ½qðxÞ @q
ð66Þ
This is as same as Eq. (49). Li and Tafti (2007) built a Lattice Boltzmann model by generalizing the thermodynamic pressure method. In order to validate their pressure formula, they carried out the following Taylor expansion
1 2
qðyÞ qðxÞ ðy xÞ rqðxÞ þ ½ðy xÞ r2 qðxÞ
ð67Þ
Li and Tafti (2007) took w(jy xj) in Eq. (64) as the Lennard–Jones potential. Substituting Eq. (67) into Eq. (64) and ignoring the terms of order higher than jy xj2, they proved that the pressure given in Eq. (64) is consistent with the pressure component in Eq. (50), i.e. the first term in the right hand side of Eq. (50). It is worth emphasizing that the first term in the right hand side of Eq. (48) is not a pure pressure component of stress tensor, as the second term is not pure shear stress. Comparing Eq. (59) with Eq. (1) in Li and Tafti (2007), one may find that Li and Tafti chose the special case of Van-Kampen Nonlocal fluid. 4.7. Zhang and Kwok model A mean field free energy model had been proposed by Zhang et al. (2004) and Zhang and Kwok (2004b). In their model the dual grand potential took the form
1 L½y; x; qðyÞ; qðxÞ ¼ fw0 ½qðxÞ þ qðxÞVðxÞgdðy xÞ þ qðxÞ/ff ðjy xjÞ½qðyÞ qðxÞ 2
ð68Þ
Substituting the first part of the right hand side of Eq. (68) into Eq. (57) gives
P0 ¼ qðxÞ
@fw0 ½qðxÞ þ qðxÞVðxÞg fw0 ½qðxÞ þ qðxÞVðxÞg @q
ð69Þ
P0 ¼ qðxÞ
@w0 ½qðxÞ w0 ½qðxÞ @q
ð70Þ
or
Thus the external potential V(x) makes no contribution to the thermodynamic pressure. Substituting the second part of the right hand side of Eq. (68) (denoted as L0 ) into Eq. (55) results in
132
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
1 L0s ¼ /ff ðjy xjÞ½qðxÞ qðyÞ2 2
ð71Þ
and substituting Eq. (71) into Eq. (57) gives
P0 ¼ qðxÞ
Z v
3
/ff ðjy xjÞ½qðyÞ qðxÞd y þ
1 4
Z v
3
/ff ðjy xjÞ½qðyÞ qðxÞ2 d y
ð72Þ
Adding Eqs. (72) and (70) gives
P ¼ qðxÞ
Z Z @w0 ½qðxÞ 1 3 3 w0 ½qðxÞ þ qðxÞ /ff ðjy xjÞ½qðyÞ qðxÞd y þ /ff ðjy xjÞ½qðyÞ qðxÞ2 d y @q 4 v v
ð73Þ
Zhang et al. (2004) and Zhang and Kwok (2004b) derived the pressure formula following Yang et al. (1976). Comparing Eq. (73) with Eq. (2) in Zhang et al. (2004) and Zhang and Kwok (2004b), it is found that the last term in the right hand side of Eq. (73) is missing in their Eq (2). Li and Tafti (2007) also found that by substituting the Taylor expansion into the Eq. (2) of Zhang et al. (2004) and Zhang and Kwok (2004b), they were unable to recover the classical square-gradient theory in Li and Tafti’s model. However, they did not point out that this was due to an incorrect formulation of the pressure. As pointed out by Li and Tafti (2007) an incorrect total nonlocal stress tensor formulation may cause the system to move away from the critical point. Following the same approach as Li and Tafti (2007) by substituting Eq. (67) into Eq. (73) and ignoring higher order terms, it is found that resulting pressure agrees with the pressure component in Eq. (50), which is the stress tensor for the classical square gradient theory. 4.8. Fluid with both density gradient and nonlocal effects In some complex fluids, such as polymers or copolymers, the dual grand potential may result from both density gradient effects and general nonlocal interactions. Following Maurits and Fraaije (Maurits & Fraaije, 1997) it is assumed that
L¼
1 2 1 1 1 qðxÞqðyÞ aq ðxÞ þ bq4 ðxÞ þ jjrx qðxÞj2 dðy xÞ c2 2 2 2 2 jy xj
ð74Þ
where a,b,j,c are material constants. According to Helfand’s penalty on high density fluctuation (Uneyama & Doi, 2005; van Vlimmeren, Maurits, Zvelindovsky, Sevink, & Fraaije, 1999), the first term is a short range interaction. The second term represents the long range cohesive interaction (see also Drossinos & Kevrekidis (2003), Drossinos, Kevrekidis, Lazaridis, & Georgopoulos (2000) and Roy et al. (1998)). Substituting Eq. (74) into Eq. (21) results in RT
Z 1 1 1 qðyÞ 3 ¼ q2 ðq 1Þða þ bq2 Þ þ jðrqÞ2 þ jqr2 q I jrq rq c2 qðxÞ d y I 2 2 2 v jy xj
ð75Þ
With Eq. (75), one may model the mesoscale phase separation kinetics of long range interaction copolymer melts in terms of an elegant nonlocal fluid dynamics framework. 5. Conclusion An explicit generalized formulation of the reversible part of the stress tensor for nonlocal fluids is obtained by means of the functional variational principle. It is well known that the equation of state of an ideal gas is required to close the set of equations for classical gas dynamics, similarly the generalized nonlocal stress tensor is an indispensable constitutive equation required to close the thermodynamic system of nonlocal fluid dynamics. Unlike previous pressure or stress tensor formulations, which are only suitable for a particular fluid, the stress tensor obtained in this paper is general, and is suitable for any nonlocal fluid, where the dual grand potential L½y; x; qðyÞ; qðxÞ; ry qðyÞ; rx qðxÞ could be an arbitrary function. We demonstrate that different forms of the non-local fluid stress tensors that can be found in the literature may be produced from the general formulation, and indeed that some of them can only be generated in this way. Using this approach we were also able to clarify some ambiguous formulations of pressure/stress tensors found in the literature using the general formulation. In a similar manner to the dynamic part of nonlocal stress tensor given by Eringen (1972, 2002), the form of the static part of the stress tensor obtained in this paper is also expressed in terms of nonlocal integration over the entire 3 dimension spatial domain occupied by the fluid,, which is essential for practical engineering application. The derived form of the nonlocal stress tensor is able to represent micro scale intermolecular interactions, and provides an efficient mesoscale numerical tool using Lattice Boltzmann simulation to carry out a multi scale analysis. Acknowledgment Authors would like thank Prof. David Bradley for his constructive discussing and correcting of the manuscript. JJZ would like thank RCUK for the support of fellowship.
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
133
References Aarts, D., Schmidt, M., & Lekkerkerker, H. N. W. (2004). Direct visual observation of thermal capillary waves. Science, 304, 847–850. Abraham, F. F. (1979). On the thermodynamics, structure and phase-stability of the nonuniform fluid state. Physics Reports-Review Section of Physics Letters, 53, 93–156. Anderson, D. M., Cermelli, P., Fried, E., Gurtin, M. E., & McFadden, G. B. (2007). General dynamical sharp-interface conditions for phase transformations in viscous heat-conducting fluids. Journal of Fluid Mechanics, 581, 323–370. Anderson, D. M., McFadden, G. B., & Wheeler, A. A. (1998). Diffuse-interface methods in fluid mechanics. Annual Review of Fluid Mechanics, 30, 139–165. Chakraborty, S. (2007). Order parameter modeling of fluid dynamics in narrow confinements subjected to hydrophobic interactions. Physical Review Letters, 99, 094504. Chang, Q. M., & Alexander, J. I. D. (2005). Application of the Lattice Boltzmann method to two-phase Rayleigh–Benard convection with a deformable interface. Journal of Computational Physics, 212, 473–489. Chapman, S., & Cowling, T. G. (1970). The mathematical theory of non-uniform gases. Cambridge University Press. Davies, A. R., Summers, J. L., & Wilson, M. C. T. (2006). On a dynamic wetting model for the finite-density multiphase lattice Boltzmann method. International Journal of Computational Fluid Dynamics, 20, 415–425. Ding, H., & Spelt, P. D. M. (2007). Wetting condition in diffuse interface simulations of contact line motion. Physical Review E, 75, 046708. Drossinos, Y., & Kevrekidis, P. G. (2003). Classical nucleation theory revisited. Physical Review E, 67, 026127. Drossinos, Y., Kevrekidis, P. G., Lazaridis, M., & Georgopoulos, P. G. (2000). Translational invariance in the theory of nucleation. Nucleation and Atmospheric Aerosols, 534, 217–220. Eringen, A. C. (1972). On nonlocal fluid mechanics. International Journal of Engineering Science, 10, 561–575. Eringen, A. C. (2002). Nonlocal continuum field theory. New York: Springer-Verlag. Fradin, C., Braslau, A., Luzet, D., Smilgies, D., Alba, M., Boudet, N., et al (2000). Reduction in the surface energy of liquid interfaces at short length scales. Nature, 403, 871–874. Frink, L. J. D., Salinger, A. G., Sears, M. P., Weinhold, J. D., & Frischknecht, A. L. (2002). Numerical challenges in the application of density functional theory to biology and nanotechnology. Journal of Physics-Condensed Matter, 14, 12167–12187. Galliero, G. (2010). Lennard–Jones fluid–fluid interfaces under shear. Physical Review E, 81, 056306. Galliero, G., Pineiro, M. M., Mendiboure, B., Miqueu, C., Lafitte, T., & Bessieres, D. (2009). Interfacial properties of the Mie n-6 fluid: Molecular simulations and gradient theory results. Journal of Chemical Physics, 130, 104704. Gonnella, G., Lamura, A., & Sofonea, V. (2007). Lattice Boltzmann simulation of thermal nonideal fluids. Physical Review E, 76, 036703. Iatsevitch, S., & Forstmann, F. (1997). Density profiles at liquid–vapor and liquid–liquid interfaces: An integral equation study. Journal of Chemical Physics, 107, 6925–6935. Iatsevitch, S., & Forstmann, F. (2000). Structure of fluid interfaces: An integral equation study. Molecular Physics, 98, 1309–1322. Khatavkar, V. V., Anderson, P. D., Duineveld, P. C., & Meijer, H. E. H. (2007a). Diffuse-interface modelling of droplet impact. Journal of Fluid Mechanics, 581, 97–127. Khatavkar, V. V., Anderson, P. D., & Meijer, H. E. H. (2007b). Capillary spreading of a droplet in the partially wetting regime using a diffuse-interface model. Journal of Fluid Mechanics, 572, 367–387. Lee, T., & Lin, C. L. (2003). Pressure evolution lattice-Boltzmann-equation method for two-phase flow with phase change. Physical Review E, 67, 056703. Lee, T., & Lin, C. L. (2005). A stable discretization of the lattice Boltzmann equation for simulation of incompressible two-phase flows at high density ratio. Journal of Computational Physics, 206, 16–47. Li, B. M., & Kwok, D. Y. (2003). Discrete Boltzmann equation for microfluidics. Physical Review Letters, 90, 124502. Li, S. M., & Tafti, D. K. (2007). A mean-field pressure formulation for liquid–vapor flows. Journal of Fluids Engineering-Transactions of the ASME, 129, 894–901. Li, S. M., & Tafti, D. K. (2009). Near-critical CO2 liquid-vapor flow in a sub-microchannel. Part I: Mean-field free-energy D2Q9 lattice Boltzmann method. International Journal of Multiphase Flow, 35, 725–737. Li, Z., & Wu, J. (2006). Density functional theory for polyelectrolytes near oppositely charged surfaces. Physics Review Letters, 96, 048302. Llovell, F., Galindo, A., Blas, F. J., & Jackson, G. (2010). Classical density functional theory for the prediction of the surface tension and interfacial properties of fluids mixtures of chain molecules based on the statistical associating fluid theory for potentials of variable range. Journal of Chemical Physics, 133, 024704. Lovett, R., & Baus, M. (1999). van der Waals theory for the spatial distribution of the tension in an interface. I. Density functional theory. Journal of Chemical Physics, 111, 5544–5554. Lovett, R., Mou, C. Y., & Buff, F. P. (1976). Structure of liquid–vapor interface. Journal of Chemical Physics, 65, 570–572. Maurits, N. M., & Fraaije, J. (1997). Application of free energy expansions to mesoscopic dynamics of copolymer melts using a Gaussian chain molecular model. Journal of Chemical Physics, 106, 6730–6743. Maurits, N. M., Zvelindovsky, A. V., & Fraaije, J. (1998). Equation of state and stress tensor in inhomogeneous compressible copolymer melts: Dynamic mean-field density functional approach. Journal of Chemical Physics, 108, 2638–2650. Nadiga, B. T., & Zaleski, S. (1996). Investigations of a two-phase fluid model. European Journal of Mechanics B-Fluids, 15, 885–896. Noether, E. (1971). Invariant variation problems. Transport Theory and Statistical Physics, 1, 186–207. Nourgaliev, R. R., Dinh, T. N., Theofanous, T. G., & Joseph, D. (2003). The lattice Boltzmann equation method: Theoretical interpretation, numerics and implications. International Journal of Multiphase Flow, 29, 117–169. Onuki, A. (2007). Dynamic Van der Waals theory. Physical Review E, 75, 036304. Patra, C. N. (2007). Effect of attractions on the structure of polymer solutions confined between surfaces: A density functional approach. J Chem Phys, 126, 074905. Percus, J. K. (1996). The stress tensor for nonlocal field equations. Journal of Mathematical Physics, 37, 1259–1267. Qian, T. Z., Wu, C. M., Lei, S. L., Wang, X. P., & Sheng, P. (2009). Modeling and simulations for molecular scale hydrodynamics of the moving contact line in immiscible two-phase flows. Journal of Physics-Condensed Matter, 21, 464119. Reguera, D., & Reiss, H. (2004). The role of fluctuations in both density functional and field theory of nanosystems. Journal of Chemical Physics, 120, 2558–2564. Romero-Rochin, V., & Percus, J. K. (1996). Stress tensor of liquid-vapor states of inhomogeneous fluids. Physical Review E, 53, 5130–5136. Roy, A., Rickman, J. M., Gunton, J. D., & Elder, K. R. (1998). Simulation study of nucleation in a phase-field model with nonlocal interactions. Physical Review E, 57, 2610–2617. Slattery, J. C., Oh, E.-S., & Fu, K. (2004). Extension of continuum mechanics to the nanoscale. Chemical Engineering Science, 59, 4621–4635. Slattery, J. C., Sagis, L., & Oh, E.-S. (2007). Interfacial transport phenomena. Springer. Sullivan, D. E. (1981). Surface-tension and contact-angle of a liquid–Solid Interface. Journal of Chemical Physics, 74, 2604–2615. Swift, M. R., Osborn, W. R., & Yeomans, J. M. (1995). Lattice Boltzmann simulation of nonideal fluids. Physical Review Letters, 75, 830–833. Tarazona, P., Checa, R., & Chacon, E. (2007). Critical analysis of the density functional theory prediction of enhanced capillary waves. Physical Review Letters, 99, 196101. Tarazona, P., Marconi, U. M. B., & Evans, R. (1987). Phase-equilibria of fluid interfaces and confined fluids - nonlocal versus local density functionals. Molecular Physics, 60, 573–595. Tavel, M. A. (1971). Milestones in mathematical physics - Noethers theorem. Transport Theory and Statistical Physics, 1, 183–185. Todd, B. D., Evans, D. J., & Daivis, P. J. (1995). Pressure tensor for inhomogeneous fluids. Physical Review E, 52, 1627–1638.
134
J. Zhu et al. / International Journal of Engineering Science 70 (2013) 124–134
Todd, B. D., & Hansen, J. S. (2008). Nonlocal viscous transport and the effect on fluid stress. Physical Review E, 78, 051202. Uneyama, T., & Doi, M. (2005). Density functional theory for block copolymer melts and blends. Macromolecules, 38, 196–205. van der Graaf, S., Nisisako, T., Schroen, C., van der Sman, R. G. M., & Boom, R. M. (2006). Lattice Boltzmann simulations of droplet formation in a T-shaped microchannel. Langmuir, 22, 4144–4152. van Giessen, A. E., Bukman, D. J., & Widom, B. (1997). Contact angles of liquid drops on low-energy solid surfaces. Journal of Colloid and Interface Science, 192, 257–265. van Kampen, N. G. (1964). Condensation of classical gas with long-range attraction. Physical Review A-General Physics, 135, A362–A369. van Vlimmeren, B. A. C., Maurits, N. M., Zvelindovsky, A. V., Sevink, G. J. A., & Fraaije, J. (1999). Simulation of 3D mesoscale structure formation in concentrated aqueous solution of the triblock polymer surfactants (ethylene oxide)(13)(propylene oxide)(30)(ethylene oxide)(13) and (propylene oxide)(19)(ethylene oxide)(33)(propylene oxide)(19). Application of dynamic mean-field density functional theory. Macromolecules, 32, 646–656. Varea, C., & Robledo, A. (1996). Stress tensor of inhomogeneous fluids. Physica A, 233, 132–144. Wertheim, M. S. (1976). Correlations in liquid–vapor interface. Journal of Chemical Physics, 65, 2377–2381. Wu, J. Z., & Li, Z. D. (2007). Density-functional theory for complex fluids. Annual Review of Physical Chemistry, 58, 85–112. Yang, A. J. M., Fleming, P. D., & Gibbs, J. H. (1976). Molecular theory of surface-tension. Journal of Chemical Physics, 64, 3732–3747. Zhang, J. (2010). Lattice Boltzmann method for microfluidics: Models and applications. Microfluidics and Nanofluidics, 10. http://dx.doi.org/10.1007/s1040410010-10624-10401. Zhang, J. F., & Kwok, D. Y. (2004a). Apparent slip over a solid–liquid interface with a no-slip boundary condition. Physical Review E, 70, 056701. Zhang, J. F., & Kwok, D. Y. (2004b). Lattice Boltzmann study on the contact angle and contact line dynamics of liquid–vapor interfaces. Langmuir, 20, 8137–8141. Zhang, J. F., & Kwok, D. Y. (2005). A 2D lattice Boltzmann study on electrohydrodynamic drop deformation with the leaky dielectric theory. Journal of Computational Physics, 206, 150–161. Zhang, J. F., & Kwok, D. Y. (2006). Contact line and contact angle dynamics in superhydrophobic channels. Langmuir, 22, 4998–5004. Zhang, J. F., Li, B. M., & Kwok, D. Y. (2004). Mean-field free-energy approach to the lattice Boltzmann method for liquid–vapor and solid–fluid interfaces. Physical Review E, 69, 032602. Zhou, S. Q., & Solana, J. R. (2009). Low temperature behavior of thermodynamic perturbation theory. Physical Chemistry Chemical Physics, 11, 11528–11537.