• 検索結果がありません。

PERISTALTIC TRANSPORT OF AN OLDROYD-B FLUID IN A PLANAR CHANNEL

N/A
N/A
Protected

Academic year: 2022

シェア "PERISTALTIC TRANSPORT OF AN OLDROYD-B FLUID IN A PLANAR CHANNEL"

Copied!
30
0
0

読み込み中.... (全文を見る)

全文

(1)

PERISTALTIC TRANSPORT OF AN OLDROYD-B FLUID IN A PLANAR CHANNEL

T. HAYAT, Y. WANG, K. HUTTER, S. ASGHAR, AND A. M. SIDDIQUI Received 5 June 2003 and in revised form 20 May 2004

The effects of an Oldroyd-B fluid on the peristaltic mechanism are examined under the long wavelength assumption. Analytical expressions for the stream function, the axial velocity, and the pressure rise per wavelength are obtained up to the second order in the dimensionless wave number. The effects of the various parameters of interest on the flow are shown and discussed.

1. Introduction

The word peristalsis derives from the Greek wordπερισταλτικoswhich means clasping and compressing. It is used to describe a progressive wave of contraction along a channel or tube whose cross-sectional area consequently varies. Peristalsis is regarded as having considerable relevance in biomechanics and especially as a major mechanism for fluid transport in many biological systems (as it is in the human). It appears in the ureter, in the intestines, and in the oviducts, to name just a few instances.

Great strives have been undertaken, both experimentally and theoretically, to study the propagation of waves in peristaltic motion [3,12,14,18,26,52,53]. Arbitrary shapes of these waves [27,29,31,32,33] as well as sinusoidal waves [1,10,12,15,23,49] have been analyzed and measuring techniques [9,24] were designed to test and verify early hydrodynamic models [15,40,41].

The governing equations are nonlinear, so assumptions are made about the amplitude ratio, the wavenumber, and the Reynolds number. The amplitude ratio is the ratio of the amplitude of the wave to the half-width of the channel and is usually taken to be small.

The case of vanishingly small Reynolds number has also received considerable attention [3,12,41]. To include nonlinear effects due to nonvanishing Reynolds number, solutions are usually presented as expansions in terms of a small parameter. They are generally of two types:

(1) expansion parameter is the amplitude of the wave that disturbs the wall; such an expansion was pursued in [15] for the channel and in [13,55] for the pipe, up to second order. Only zero-mean flow was considered for the second-order terms,

Copyright©2004 Hindawi Publishing Corporation Mathematical Problems in Engineering 2004:4 (2004) 347–376 2000 Mathematics Subject Classification: 76A05

URL:http://dx.doi.org/10.1155/S1024123X04405043

(2)

a restriction that was removed in [35]. This approach, while valid for all Reynolds numbers and wavelengths of the disturbing wall, is restricted to small amplitudes and has been applied only to a sinusoidal wave;

(2) expansion in terms of the wavenumber and the Reynolds number for all wave amplitudes. This has been done only for sinusoidal waves, and up to first order in the square of the wavenumber (the natural parameter) and second order in the Reynolds number [22,28,56]. The nonsinusoidal wave was studied in [27] for small intestine and for zero Reynolds number only.

To obtain information about flows at moderate Reynolds number it has been neces- sary to use numerical methods. Several investigators [2,10,21,49] used numerical meth- ods for the solution of the Newtonian hydrodynamical equations. The results, in general, agree with the analytical perturbation solutions in their range of validity with the excep- tion of the calculations of the pressure field by Takabatake and Ayukawa [49]. It is noted that the higher-order terms of Reynolds and wavenumbers do not significantly extend the range of validity of the results.

The application of the theory of particle-fluid mixture is also very useful in under- standing a number of diverse physical problems concerning peristalsis. An interesting ex- ample is the particulate suspension theory of blood [5,20,34,36,46,47,50]. Peristaltic transport of solid particles with fluid has first been attempted in [21]. Various geometric and dynamic effects on the particle transport in a channel with flexible walls were exam- ined. The peristaltic motion for the case of two-phase flow was studied in [48] where a perturbation solution for a small amplitude ratio is given.

Most studies on the peristaltic motion assume the physiological fluids to behave like Newtonian fluids with constant viscosity. However, this approach fails to give an ade- quate understanding of the peristaltic mechanism involved in small blood vessels, lym- phatic vessels, intestine, and ductus efferentes of the male reproductive tracts. In these body organs, the viscosity of the fluid varies across the thickness of the duct [11,16,19].

Also, the assumption that the chyme in small intestine is a Newtonian material of vari- able viscosity is not adequate in reality. Chyme is undoubtedly a non-Newtonian fluid.

Some authors (see, e.g., [51]) feel that the main factor responsible for moving the chyme along the intestine is a gradient in the frequency of segmentation (a process of oscillating contraction and relaxation of smooth muscles in the intestine wall) along the length of intestine. Moreover, peristaltic waves die out after travelling a very short distance; peri- staltic waves which travel the entire length of small intestine do not occur in humans ex- cept under abnormal conditions. Also, in transport of spermatozoa in the cervical canal, there are some other important factors, responsible for the transport of semen in ductus efferentes. One of the major factors is cilia, which keep semen moving towards the epi- didymis [7,17,25,30,54]. The phenomenon of peristalsis has also been proposed as a mechanism for the transport of spermatic fluid (semen) in vas deferens [39]. Movement through vas deferens is accomplished by means of peristaltic action of contractile cells in the duct wall [39,51]. However, there is no doubt that peristalsis aids in moving semen in ductus efferentes, the chyme in the intestine, and flow of semen in vas deferens.

The above review of physiological flows indicates that non-Newtonian viscoelastic rhe- ology is the correct way of properly describing the peristaltic flow through channels and

(3)

tubes. Only a few studies [4, 8,42,43,44,45,47] have considered this aspect of the problem. Although the second, and third-order models in [43,44,45] take into account normal stress differences and shear-thinning/thickening effects, they lack other features such as stress relaxation. The Oldroyd-B fluid, which includes elastic and memory ef- fects exhibited by dilute solutions, has been extensively used in many applications, and also results of simulations fit experimental data quite well [6,37,38]. However, so far, no attempt has been made to understand the peristaltic motion for an Oldroyd-B fluid.

We propose to study the effects of an Oldroyd-B fluid on the mechanism of peristaltic transport in a planar channel. Of course the natural coordinate system is axisymmet- ric; however, the planar case has been predominantly studied. Qualitatively the transport phenomenon of the fluid is similar for both configurations [23]. Also, experimental data are available for channel flows [26,53]. Therefore, the present mathematical model con- siders an Oldroyd-B fluid between parallel walls on which a sinusoidal travelling wave is imposed. The assumption for the present analysis is that the length of the peristaltic wave is large compared with the half-width of the channel. This assumption is similar to those used in [22,43] for the peristaltic motion of Newtonian and second-order flu- ids, respectively. A regular perturbation technique is adopted to solve the present prob- lem and solutions are expanded in a power series of the small dimensionless wavenum- ber. The Reynolds number and material time constants are left arbitrary. The analysis is completely analytical but lengthy, and closed-form solutions up to second order of the wavenumber are presented. The effects of the nonlinear terms of the governing equations on the fluid transport are constructed. Comparison is made between the results for the Newtonian and Oldroyd-B fluids. The explicit non-Newtonian terms are obtained and their effect on peristaltic motion is examined. The results for Maxwell and Newtonian fluids are obtained as special cases of the presented analysis.

2. Basic equations

Consider an incompressible fluid whose balance laws of mass and linear momentum are given by

div ¯V=0, (2.1)

ρV˙ =div ¯T+ρ¯f, (2.2)

whereρ, ¯V, ¯T, and ¯f are mass density, velocity, Cauchy stress tensor, and specific body force and the dot (·) denotes material time derivative. In the ensuing analysis, body forces will be ignored and isothermal conditions will be implied. The above system of equations will be closed by a constitutive equation for the stress tensor. The constitutive equation for the Cauchy stress ¯Tin an Oldroyd-B fluid is given by [37]

T¯ = −pI¯ + ¯S, (2.3)

where the extra stress tensor ¯Sis given by S¯+Λ1

dS¯

dt¯L¯S¯S¯L¯T

=µ

A¯12

dA¯1

d¯t L¯A¯1A¯1L¯T

, (2.4)

(4)

in whichpI¯ is the spherical part of the stress due to the constraint of incompressibility, d/dt¯denotes material time derivative,µis the viscosity, andΛ1andΛ2are material time constants referred to as relaxation and retardation time, respectively. It is assumed that Λ1Λ20. The tensors ¯Land ¯A1are defined as follows:

L¯=grad ¯V, A¯1=L¯+ ¯LT, (2.5)

where ¯V is the velocity vector. It should be noted that this model includes the classi- cal linear case forΛ1=Λ2=0, and when Λ2=0, the model reduces to the Maxwell model.

3. Formulation of the problem and flow equations

Consider a two-dimensional flow of an Oldroyd-B fluid in an infinite channel having width 2a. Assume an infinite wave train travelling with velocitycalong the walls. Choose a rectangular coordinate system for the channel with ¯X along the central line in the di- rection of wave propagation, and ¯Y transverse to it. Let the geometry of the wall surface be defined as

h( ¯¯ X, ¯t)=a+bsin

λ ( ¯Xct)¯

, (3.1)

wherebis the wave amplitude andλthe wavelength. Assume, moreover, that there is no motion of the wall in the longitudinal direction (this assumption constrains the deforma- tion of the wall; it does not necessarily imply that the channel is rigid against longitudinal motions, but is a convenient simplification that can be justified by a more complete anal- ysis. The assumption implies that for the no-slip condition ¯U=0 at the wall).

For unsteady two-dimensional flows,

V¯ =U( ¯¯ X, ¯Y, ¯t), ¯V( ¯X, ¯Y, ¯t), 0, (3.2)

and we find that (2.1)–(2.5), in the absence of body forces, take the following form:

∂U¯

∂X¯ +∂V¯

∂Y¯ =0, ρ

∂t¯+ ¯U

∂X¯ + ¯V

∂Y¯

U¯ = −∂p( ¯¯ X, ¯Y, ¯t)

∂X¯ +∂S¯X¯X¯

∂X¯ +∂S¯X¯Y¯

∂Y¯ , ρ

∂t¯+ ¯U

∂X¯ + ¯V

∂Y¯

V¯ = −∂p( ¯¯ X, ¯Y, ¯t)

∂Y¯ +∂S¯X¯Y¯

∂X¯ +∂S¯Y¯Y¯

∂Y¯ ,

(5)

S¯X¯X¯1

¯t+ ¯U

∂X¯ + ¯V

∂Y¯

S¯X¯X¯2∂U¯

∂X¯S¯X¯X¯2∂U¯

∂Y¯S¯X¯Y¯

=∂U¯

∂X¯ + 2µΛ2

∂t¯+ ¯U

∂X¯ + ¯V

∂Y¯ ∂U¯

∂X¯ 2 ∂U¯

∂X¯ 2

∂U¯

∂Y¯ ∂U¯

∂Y¯ +∂V¯

∂X¯

, S¯X¯Y¯1

∂t¯+ ¯U

∂X¯ + ¯V

∂Y¯

S¯X¯Y¯∂U¯

∂Y¯S¯Y¯Y¯∂V¯

∂X¯S¯X¯X¯

=µ ∂U¯

∂Y¯ +∂V¯

∂X¯

+µΛ2

∂t¯+ ¯U

∂X¯ + ¯V

∂Y¯ ∂U¯

∂Y¯ +∂V¯

∂X¯

2 ∂U¯

∂X¯

∂V¯

∂X¯ +∂U¯

∂Y¯

∂V¯

∂Y¯

, S¯Y¯Y¯1

∂t¯+ ¯U

∂X¯ + ¯V

∂Y¯

S¯Y¯Y¯2∂V¯

∂X¯S¯X¯Y¯2∂V¯

∂Y¯S¯Y¯Y¯

=∂V¯

∂Y¯ + 2µΛ2

∂t¯+ ¯U

∂X¯ + ¯V

∂Y¯ ∂V¯

∂Y¯ 2 ∂V¯

∂Y¯ 2

∂V¯

∂X¯ ∂U¯

∂Y¯ +∂V¯

∂X¯

, (3.3) where ¯Uand ¯Vare the longitudinal and transverse velocity components.

In the laboratory frame ( ¯X, ¯Y), the flow in the channel is unsteady, but if we choose moving coordinates ( ¯x, ¯y) which travel in the positive ¯X-direction with the same speed as the wave, then the flow can be treated as steady. This coordinate system is known as the wave frame. The coordinate frames are related through

¯

x=X¯ct,¯ y¯=Y¯, (3.4)

and the velocity components in the laboratory and wave frames are related by

¯

u=U¯c, v¯=V¯, (3.5)

where ¯uand ¯vare dimensional velocity components in the directions of ¯xand ¯y, respec- tively. Employing these transformations in (3), we obtain

∂u¯

∂x¯+∂v¯

∂y¯ =0, ρ

¯ u

∂x¯+ ¯v

∂y¯

¯ u= −∂p¯

∂x¯ +∂S¯x¯x¯

∂x¯ +∂S¯x¯y¯

∂y¯ , ρ

¯ u

∂x¯+ ¯v

∂y¯

¯ v= −∂p¯

∂y¯ +∂S¯x¯y¯

∂x¯ +∂S¯y¯y¯

∂y¯ ,

(6)

S¯¯x1

u¯

∂x¯+ ¯v

∂y¯

S¯¯x2∂u¯

∂x¯S¯¯x2∂u¯

∂y¯S¯¯y

=∂U¯

∂x¯ + 2µΛ2 u¯

∂x¯+ ¯v

∂y¯ ∂u¯

∂x¯ 2 ∂u¯

∂x¯ 2

∂u¯

∂y¯ ∂u¯

∂y¯+∂v¯

∂x¯

, S¯x¯y¯1

¯ u

∂x¯+ ¯v

∂y¯

S¯¯y∂u¯

∂y¯S¯¯y∂v¯

∂x¯S¯¯x

=µ ∂u¯

∂y¯+∂v¯

∂x¯

+µΛ2

¯ u

∂x¯+ ¯v

∂y¯ ∂u¯

∂y¯ +∂v¯

∂x¯

2 ∂u¯

∂x¯

∂v¯

∂x¯+∂u¯

∂y¯

∂¯v

∂y¯

, S¯¯y1

¯ u

∂x¯+ ¯v

∂y¯

S¯¯y2∂v¯

∂x¯S¯x¯y¯2∂¯v

∂y¯S¯¯y

=∂v¯

∂y¯+ 2µΛ2 u¯

∂x¯+ ¯v

∂y¯ ∂v¯

∂y¯ 2 ∂¯v

∂y¯ 2

∂v¯

∂x¯ ∂u¯

∂y¯ +∂v¯

∂x¯

.

(3.6) The formulation of the boundary conditions is postponed untilSection 5.

4. Dimensionless formulation

To set the important parameters of the outlined problem in evidence, a scale analysis is performed and the equations are nondimensionalized. Using the dimensionless variables

¯ x= λx

2π, y¯=ay, u¯=cu, v¯=cv, S¯=µc

aS, p¯= λµc

2πa2p, h¯=ah

(4.1)

in (3), we arrive at

δ∂u

∂x+∂v

∂y=0, (4.2)

᏾eδu

∂x+v

∂y

u

= −∂p

∂x+δ∂Sxx

∂x +∂Sxy

∂y , (4.3)

δ᏾e

δu

∂x+v

∂y

v

= −∂p

∂y+δ2∂Sxy

∂x +δ∂Sy y

∂y , (4.4)

Sxx+λ1

δu∂

∂x+v

∂y

Sxx∂u

∂xSxx2∂u

∂ySxy

=∂u

∂x+ 2λ2 δ

δu∂

∂x+v

∂y ∂u

∂x2 ∂u

∂x 2

∂u

∂y ∂u

∂y+δ∂v

∂x

,

(4.5)

(7)

Sxy+λ1

δu

∂x+v

∂y

Sxyδ∂v

∂xSxx∂u

∂ySy y

= ∂u

∂y+δ∂v

∂x

+λ2

δu

∂x+v

∂y ∂u

∂y+δ∂v

∂x

2

δ2∂u

∂x

∂v

∂x+∂u

∂y

∂v

∂y

, (4.6)

Sy y+λ1

δu

∂x+v

∂y

Sy y∂v

∂xSxy2∂v

∂ySy y

=2∂v

∂y+ 2λ2

δu

∂x+v

∂y ∂v

∂y2 ∂v

∂y 2

δ∂v

∂x ∂u

∂y+δ∂v

∂x

,

(4.7)

where the dimensionless wavenumberδ, the Reynolds number᏾e, and the Weissenberg numbersλ1andλ2are defined, respectively, as

δ=2πa

λ , ᏾e= ca

µ/ρ, λ1=Λ1c

a , λ2=Λ2c

a . (4.8)

These have easy physical interpretations:δ is a measure of how large the semidepth of the peristaltic motion is, as compared to its wavelength. It is an aspect ratio and thus an expression of shallowness. The Reynolds number᏾eis formed with the wave speed, the amplitude, and the kinematic viscosity of the Newtonian part of the constitutive behav- ior;λ1andλ2measure the elastic contributions of the stress behavior.

The continuity equation (4.2), after defining the dimensionless stream functionΨ(x, y) by the relations

u=∂Ψ

∂y, v= −δ∂Ψ

∂x, (4.9)

is identically satisfied and, from (4.3)–(4.7) we deduce

δ᏾e ∂Ψ

∂y

∂x

∂Ψ

∂x

∂y ∂Ψ

∂y

= −∂p

∂x+δ∂Sxx

∂x +∂Sxy

∂y , (4.10)

δ3᏾e∂Ψ

∂y

∂x

∂Ψ

∂x

∂y ∂Ψ

∂x

= −∂p

∂y +δ2∂Sxy

∂x +δ∂Sy y

∂y , (4.11)

δ᏾eΨ

∂y

∂x

Ψ

∂x

∂y 2Ψ

∂y2 +δ22Ψ

∂x2

=δ∂2SxxSy y

∂x∂y + 2

∂y2δ2 2

∂x2

Sxy,

(4.12)

(8)

Sxx+λ1

δ

Ψ

∂y

∂x

Ψ

∂x

∂y

Sxx2Ψ

∂x∂ySxx22Ψ

∂y2Sxy

=2Ψ

∂x∂y+ 2λ2

δ2

∂Ψ

∂y

∂x

∂Ψ

∂x

∂y 2Ψ

∂x∂y 2 2Ψ

∂x∂y 2

2Ψ

∂y2 2Ψ

∂y2 δ22Ψ

∂x2

,

(4.13)

Sxy+λ1

δ

∂Ψ

∂y

∂x

∂Ψ

∂x

∂y

Sxy+δ22Ψ

∂x2Sxx2Ψ

∂y2Sy y

= 2Ψ

∂y2 δ22Ψ

∂x2

+λ2

δ

Ψ

∂y

∂x

Ψ

∂x

∂y 2Ψ

∂y2 δ22Ψ

∂x2

+ 2δ 2Ψ

∂x∂y 2Ψ

∂y2 +δ22Ψ

∂x2

,

(4.14)

Sy y+λ1

δ

Ψ

∂y

∂x

Ψ

∂x

∂y

Sy y+ 2δ 2Ψ

∂x∂ySy y+ 2δ22Ψ

∂x2Sxy

= −2Ψ

∂x∂y+ 2λ2 δ2 ∂Ψ

∂y

∂x

∂Ψ

∂x

∂y

2Ψ

∂x∂y

2 2Ψ

∂x∂y 2

+δ22Ψ

∂x2 2Ψ

∂y2 δ22Ψ

∂x2

,

(4.15)

where the compatibility equation (4.12) is obtained by eliminatingpbetween (4.10) and (4.11); it represents the vorticity transport equation. Notice that (4.10) and (4.11) are formally decoupled from (4.12)–(4.15). So, the latter are used to determineΨandSxx, Sxy,Sy y, while the former is then employed to determine the pressure field.

5. Rate of volume flow and boundary conditions

The dimensional rate of fluid flow in the laboratory frame is given by Q=

h¯

0

U( ¯¯ X, ¯Y, ¯t)dY¯, (5.1) where ¯h, the position of the channel wall, is a function of ¯Xand ¯t. The rate of fluid flow in the wave frame is given by

q= h¯

0u( ¯¯ x, ¯y)dy,¯ (5.2)

where ¯his now a function of ¯xalone. With the help of (3.4) and (3.5), one can show that these two rates are related through

Q=q+ch.¯ (5.3)

(9)

The time-averaged flow over a periodTat a fixed position ¯Xis given by Q¯ = 1

T T

0 Q dt. (5.4)

On using (5.3) in (5.4), we find that

Q¯ =q+ac. (5.5)

If we define the dimensionless mean flows Θ, in the laboratory frame, and F, in the wave frame, according to

Θ= Q¯

ac, F= q

ac, (5.6)

one finds that (5.5) reduces to

Θ=F+ 1, (5.7)

where, according to the first equation of (4.9), F=

h

0

Ψ

∂yd y=Ψ(h)Ψ(0). (5.8) If we choose the zero value of the streamline along the central line (y=0)

Ψ(0)=0, (5.9)

then the shape of the wave at the wall boundary is the streamline with value

Ψ(h)=F. (5.10)

The boundary conditions for the dimensionless stream function in the wave frame are therefore

Ψ=0 (by convention)

2Ψ

∂y2 =0 (by symmetry) on the central liney=0, (5.11)

Ψ

∂y = −1 (no-slip condition) Ψ=F

at the wally=h. (5.12) We also note thathrepresents the dimensionless form of the surface of the peristaltic wall which will be chosen as a sinusoidal function, namely,

h(x)=1 +Φsinx, (5.13) where

Φ=b

a (5.14)

is the amplitude ratio or the occlusion and 0<Φ<1.

(10)

6. Perturbation solution

We note that (4.10)–(4.15) are higher-order nonlinear partial differential equations.

Therefore, it seems to be impossible to find the solution in closed form for arbitrary values of all parameters. Even for Newtonian fluids [22,41], all solutions obtained so far are based on the assumption that one or several parameters are zero or small. Following Jaffrin [22], we expand the flow quantities in a power series of the small parameterδas follows:

Ψ=Ψ0+δΨ1+δ2Ψ2+···, p=p0+δ p1+δ2p2+···, S=S0+δS1+δ2S2+···, F=F0+δF1+δ2F2+···.

(6.1)

On substituting (6.1) into (4.10)–(4.15) and then collecting terms of equal powers ofδ, one obtains the following sets of perturbed equations.

(i) Zeroth-order equations

2S0xy

∂y2 =0, (6.2)

∂p0

∂x +∂S0xy

∂y =0, (6.3)

∂p0

∂y =0, (6.4)

S0xx1

2Ψ0

∂y2 S0xy= −2

2Ψ0

∂y2 2

, (6.5)

S0xyλ12Ψ0

∂y2 S0y y=2Ψ0

∂y2 , (6.6)

S0y y=0. (6.7)

(ii) First-order equations

᏾e ∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ0

∂y2

=2S0xxS0y y

∂x∂y + 2

∂y2S1xy,

᏾e ∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y ∂Ψ0

∂y

= −∂p1

∂x +∂S0xx

∂x +∂S1xy

∂y ,

∂p1

∂y +∂S0y y

∂y =0,

(11)

S1xx+λ1

Ψ0

∂y

∂x

Ψ0

∂x

∂y

S0xx22Ψ0

∂x∂yS0xx22Ψ0

∂y2 S1xy22Ψ1

∂y2 S0xy

=22Ψ0

∂x∂y22Ψ0

∂y2

2Ψ1

∂y2 , S1xy+λ1

Ψ0

∂y

∂x

Ψ0

∂x

∂y

S0xy2Ψ0

∂y2 S1y y2Ψ1

∂y2 S0y y

=2Ψ1

∂y2 +λ2

Ψ0

∂y

∂x

Ψ0

∂x

∂y 2Ψ0

∂y2 + 22Ψ0

∂x∂y

2Ψ0

∂y2

, S1y y+λ1

Ψ0

∂y

∂x

Ψ0

∂x

∂y

S0y y+ 22Ψ0

∂x∂yS0y y

= −22Ψ0

∂x∂y.

(6.8)

(iii) Second-order equations

᏾e∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ1

∂y2 + ∂Ψ1

∂y

∂x

∂Ψ1

∂x

∂y 2Ψ0

∂y2

=2S1xxS1y y

∂x∂y +2S2xy

∂y2

2S0xy

∂x2 ,

(6.9)

᏾e∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y ∂Ψ1

∂y + ∂Ψ1

∂y

∂x

∂Ψ1

∂x

∂y ∂Ψ0

∂y

= −∂p2

∂x +∂S1xx

∂x +∂S2xy

∂y ,

(6.10)

∂p2

∂y +∂S0xy

∂x +∂S1y y

∂y =0, (6.11)

S2xx+λ1

∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y

S1xx+ ∂Ψ1

∂y

∂x

∂Ψ1

∂x

∂y

S0xx

22Ψ0

∂x∂yS1xx22Ψ1

∂x∂yS0xx22Ψ0

∂y2 S2xy22Ψ1

∂y2 S1xy22Ψ2

∂y2 S0xy

=22Ψ1

∂x∂y+ 2λ2

∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ0

∂x∂y2 2Ψ0

∂x∂y 2

22Ψ0

∂y2

2Ψ2

∂y2 +2Ψ0

∂x2

2Ψ0

∂y2 2Ψ1

∂y2 2

,

(6.12)

(12)

S2xy+λ1

Ψ1

∂y

∂x

Ψ1

∂x

∂y

S0xy+ Ψ0

∂y

∂x

Ψ0

∂x

∂y

S1xy

2Ψ0

∂y2 S2y y2Ψ1

∂y2 S1y y2Ψ2

∂y2 S0y y+2Ψ0

∂x2 S0xx

=2Ψ2

∂y2

2Ψ0

∂x2 +λ2

Ψ0

∂y

∂x

Ψ0

∂x

∂y 2Ψ1

∂y2 + Ψ1

∂y

∂x

Ψ1

∂x

∂y 2Ψ0

∂y2 + 22Ψ0

∂x∂y

2Ψ1

∂y2 +22Ψ1

∂x∂y

2Ψ0

∂y2

,

(6.13) S2y y+λ1

Ψ1

∂y

∂x

Ψ1

∂x

∂y

S0y y+ Ψ0

∂y

∂x

Ψ0

∂x

∂y

S1y y

+ 22Ψ0

∂x2 S0xy+ 22Ψ0

∂x∂yS1y y+ 22Ψ1

∂x∂yS0y y

= −22Ψ1

∂x∂y2

Ψ0

∂y

∂x

Ψ0

∂x

∂y 2Ψ0

∂x∂y+ 2 2Ψ0

∂x∂y 2

2Ψ0

∂x2

2Ψ0

∂y2

. (6.14)

Because the boundary conditions (5.11), (5.12) are linear, identical conditions fall on every order system of equations; this is why they are not repeated above.

After lengthy calculations (the interested reader may consult the principal author) with (6.2)–(6.14) and the boundary conditions (5.11), (5.12) at each order, the following boundary value problems for the stream function, pressure and stress components are deduced.

(i) Zeroth-order system

4Ψ0

∂y4 =0, (6.15)

∂p0

∂x =

3Ψ0

∂y3 , (6.16)

∂p0

∂y =0, (6.17)

S0xx=2λ1λ2

2Ψ0

∂y2 2

, (6.18)

S0xy=2Ψ0

∂y2 , (6.19)

S0y y=0 (6.20)

with boundary conditions

Ψ0=0, 2Ψ0

∂y2 =0 aty=0, Ψ0=F0, ∂Ψ0

∂y = −1 at y=h.

(6.21)

(13)

(ii) First-order system

᏾e∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ0

∂y2

2λ1λ2

2

∂x∂y

2Ψ0

∂y2 2

=4Ψ1

∂y4 λ1λ2

2

∂y2 Ψ0

∂y

∂x

Ψ0

∂x

∂y 2Ψ0

∂y2

2λ1λ2

2

∂y2 2Ψ0

∂y2

2Ψ0

∂x∂y

,

(6.22)

∂p1

∂x = −᏾e∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y ∂Ψ0

∂y

+ 2λ1λ2

∂x

2Ψ0

∂y2 2

+3Ψ1

∂y3 λ1λ2

∂y Ψ0

∂y

∂x

Ψ0

∂x

∂y 2Ψ0

∂y2

2λ1λ2

∂y 2Ψ0

∂y2

2Ψ0

∂x∂y

,

(6.23)

∂p1

∂y =0, (6.24)

S1xx= −1

λ1λ2

Ψ0

∂y

∂x

Ψ0

∂x

∂y

2Ψ0

∂y2 2

+ 4λ1λ2

2Ψ0

∂y2

2Ψ1

∂y2

1

λ1λ2

2Ψ0

∂y2 ∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ0

∂y2

+ 22Ψ0

∂x∂y,

(6.25)

S1xy=2Ψ1

∂y2

λ1λ2

∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ0

∂y2

2λ1λ2

2Ψ0

∂y2

2Ψ0

∂x∂y, (6.26) S1y y= −22Ψ0

∂x∂y (6.27)

with boundary conditions

Ψ1=0, 2Ψ1

∂y2 =0 aty=0, Ψ1=F1, ∂Ψ1

∂y =0 aty=h.

(6.28)

(14)

(iii) Second-order system

᏾e∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ1

∂y2 + ∂Ψ1

∂y

∂x

∂Ψ1

∂x

∂y 2Ψ0

∂y2

+ 2λ1

λ1λ2

2

∂x∂y

∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y

2Ψ0

∂y2 2

3 2

∂x∂y 2Ψ0

∂x∂y

+ 2λ1

λ1λ2

2

∂x∂y 2Ψ0

∂y2 ∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ0

∂y2

4λ1λ2

2

∂x∂y 2Ψ0

∂y2

2Ψ1

∂y2

=2S2xy

∂y2 ,

(6.29)

∂p2

∂x = −᏾eΨ0

∂y

∂x

Ψ0

∂x

∂y Ψ1

∂y + Ψ1

∂y

∂x

Ψ1

∂x

∂y Ψ0

∂y

1

λ1λ2

∂x

Ψ0

∂y

∂x

Ψ0

∂x

∂y

2Ψ0

∂y2 2

+ 2

∂x 2Ψ0

∂x∂y

1

λ1λ2

∂x 2Ψ0

∂y2 Ψ0

∂y

∂x

Ψ0

∂x

∂y 2Ψ0

∂y2

+ 4λ1λ2

∂x 2Ψ0

∂y2

2Ψ1

∂y2

+∂S2xy

∂y ,

(6.30)

∂p2

∂y = −

∂y 2Ψ0

∂x∂y

, (6.31)

S2xx+λ1 ∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y

S1xx+ 2λ1λ2

∂Ψ1

∂y

∂x

∂Ψ1

∂x

∂y

2Ψ0

∂y2 2

22Ψ0

∂x∂yS1xx4λ1λ2

2Ψ1

∂x∂y 2Ψ0

∂y2 2

22Ψ0

∂y2 S2xy

22Ψ1

∂y2 S1xy22Ψ0

∂y2

2Ψ2

∂y2

=22Ψ1

∂x∂y+ 2λ2 ∂Ψ0

∂y

∂x

∂Ψ0

∂x

∂y 2Ψ0

∂x∂y2 2Ψ0

∂x∂y 2

22Ψ0

∂y2

2Ψ2

∂y2 +2Ψ0

∂x2

2Ψ0

∂y2 2Ψ1

∂y2 2

,

(6.32)

参照

関連したドキュメント

The main technique in the proof of their theorem is the computation of the fixed point index of all iterates of an orientation preserving homeomorphism in a neighborhood of a

pole placement, condition number, perturbation theory, Jordan form, explicit formulas, Cauchy matrix, Vandermonde matrix, stabilization, feedback gain, distance to

In this case, the extension from a local solution u to a solution in an arbitrary interval [0, T ] is carried out by keeping control of the norm ku(T )k sN with the use of

Proof.. One can choose Z such that is has contractible connected components. This simply follows from the general fact that under the assumption that the functor i : Gr // T is

Then it follows immediately from a suitable version of “Hensel’s Lemma” [cf., e.g., the argument of [4], Lemma 2.1] that S may be obtained, as the notation suggests, as the m A

Applications of msets in Logic Programming languages is found to over- come “computational inefficiency” inherent in otherwise situation, especially in solving a sweep of

Shi, “The essential norm of a composition operator on the Bloch space in polydiscs,” Chinese Journal of Contemporary Mathematics, vol. Chen, “Weighted composition operators from Fp,

[2])) and will not be repeated here. As had been mentioned there, the only feasible way in which the problem of a system of charged particles and, in particular, of ionic solutions