IJMMS 2004:47, 2525–2535

PII. S0161171204309105
http://ijmms.hindawi.com
© Hindawi Publishing Corp.

AN APPLICATION OF THE METHOD OF MATCHED ASYMPTOTIC

EXPANSIONS FOR LOW REYNOLDS NUMBER FLOW PAST
A CYLINDER OF ARBITRARY CROSS SECTION

MIRELA KOHR

Received 11 September 2003

We study the low Reynolds number flow of an incompressible Newtonian fluid of infinite
expanse past a cylinder of arbitrary cross section by using the method of matched asymp-
totic expansions. The analysis that will be made in this paper is equivalent to that devel-
oped by Power (1990) in order to solve the resulting inner (or Stokes) problems with the
completed double-layer boundary integral equation method (CDLBIEM) due to Power and
Miranda (1987). We will solve these problems by the boundary integral method developed
by Hsiao and Kress (1985).

2000 Mathematics Subject Classification: 76D03, 76M15, 76M45.

1. Introduction. The method of matched asymptotic expansions has been developed

by Kaplun [2] and Proudman and Pearson [11] for the problem of the low Reynolds num-
ber flow past a circular cylinder or a sphere. This method was used by many authors
to treat several low Reynolds number flow problems. For example, Umemura [13] ob-
tained a matched asymptotic analysis of low Reynolds number flow past two equal
cylinders. The same method was also applied by Shintani et al. [12] to study the low
Reynolds number flow due to a uniform stream at infinity past an elliptic cylinder. Lee
and Leal [6] treated the low Reynolds number flow past cylindrical bodies of arbitrary
cross section. In addition to the method of matched asymptotic expansions, they used
the boundary integral formulation of Youngren and Acrivos [14], in order to obtain the
corresponding solutions of the Stokes and Oseen approximations. Power [7] developed
a matched asymptotic analysis for low Reynolds number flow past a cylinder of arbi-
trary cross section by using the completed double layer boundary integral equation
method (CDLBIEM) to solve the resulting inner problems, and the singularity method
to treat the resulting outer problems (see also [9, Section 6.3]).
Note that Youngren and Acrivos [14] proposed a boundary integral method in order to
treat the unbounded Stokes flow due to the motion of a solid particle of arbitrary shape
in an incompressible Newtonian fluid of infinite expanse. This method uses the direct
boundary integral representation of an exterior Stokes flow, in which the variables are
the boundary velocity and traction. Also, the method leads to a set of Fredholm integral
equations of the first kind with unknown boundary traction. The method of Youngren
and Acrivos [14] has been applied by many authors to obtain the numerical solutions
of several problems dealing with solid particles and drops in Stokes flows, the motion
of a particle near a solid wall or a fluid interface, particle-particle interactions, Stokes
2526 MIRELA KOHR

flows in containers, and so forth (see, e.g., [4, 9, 10]). However, the direct boundary
integral representations, in particular, those encountered in the method of Youngren
and Acrivos [14], lead to a set of Fredholm integral equations of the first kind, which,
after the discretization of the involved surface integrals, is ill-conditioned at a large
number of boundary elements. On the other hand, it is known that the boundary inte-
gral methods which lead to Fredholm integral equations of the second kind are more
preferable than those which lead to Fredholm integral equations of the first kind, since
the Fredholm integral equations of the second kind give rise to numerical solutions that
are more stable than those due to Fredholm integral equations of the first kind. The
indirect boundary integral methods are designed so that they provide a set of Fredholm
integral equations of the second kind, and therefore they are always well-behaved nu-
merically. In particular, effective iterative solution procedures can be applied to solve
large scale problems with indirect formulations. An alternative indirect boundary in-
tegral formulation was proposed by Power and Miranda [8] for the three-dimensional
exterior Stokes flow around a solid particle (see also [9]). Power and Miranda’s method is
a completion plus a deflation procedure that leads to a bounded and invertible integral
operator (with a spectral radius strictly less than one), and therefore iterative solution
strategies are guaranteed to converge to a unique solution. Karrila and Kim [3] called
Power and Miranda’s method the completed double-layer boundary integral equation
method because of the involved completion procedure. This method applies to both
two- and three-dimensional Stokes flow problems. For the two-dimensional Stokes flow
problem due to the motion of a cylinder of arbitrary shape in an unbounded domain,
there are two equivalent integral formulations available: one was provided by Hsiao
and Kress [1] and uses a combination of double- and single-layer potentials. This for-
mulation leads to a system of Fredholm integral equations of the second kind that has
a unique continuous solution. The second formulation was developed by Power [7] and
is given in terms of a double-layer potential and two singularities located inside the
cylinder.
In this paper, we study the low Reynolds number flow of an incompressible New-
tonian fluid of infinite expanse past a cylinder of arbitrary cross section by using the
method of matched asymptotic expansions and the method of Hsiao and Kress [1] in
order to solve the resulting inner (or Stokes) problems.

2. Inner and outer expansions. We consider the problem of determining the low
Reynolds number flow of an incompressible Newtonian fluid of infinite expanse past a
stationary cylinder of arbitrary cross section. At infinity the flow is a uniform stream
with velocity U∞ in the direction of the x1 -axis.
The flow is governed by the continuity and steady Navier-Stokes equations, which in
nondimensional form are given by

∇·v = 0 in D,
2
(2.1)
−∇p + ∇ v − Re(v · ∇)v = 0 in D,

where D is the two-dimensional unbounded domain exterior to the cross section of

the cylinder in the x1 x2 -plane. Let Γ denote the boundary of this domain, assumed to
 AN APPLICATION OF THE METHOD OF MATCHED ASYMPTOTIC. . . 2527

be a simple closed Lyapunov curve (i.e., Γ has a continuously varying normal vector;
more exactly, there exists α ∈ (0, 1] such that Γ is of class C 1,α ). Also, let D0 denote
the bounded domain inside Γ . Equations (2.1) are nondimensionalized with respect
to the characteristic variables Uc = U∞ , lc = a (a characteristic cylinder radius), and
pc = µU∞ /a (the characteristic pressure). Also, the Reynolds number is defined by
Re = ρaU∞ /µ, where ρ and µ are the density and dynamic viscosity of the fluid.
We have to require the following boundary and asymptotic conditions:

v(x) = 0 for x ∈ Γ , (2.2)

v(x)
→ i1 , p(x)
→ 0 as |x|
→ ∞, (2.3)

where x = (x1 , x2 ) and i1 denotes the unit vector along the x1 -axis of a frame of Carte-
sian coordinates whose origin is inside Γ .
According to the method of Kaplun [2] and Proudman and Pearson [11], the region
around the cylinder is divided into two separate but overlapping regions, called the
inner and outer regions. In the inner region, where Re  1 (and hence the inertial term
is small), we consider the following expansions (see also [7] and [9, Section 6.3.4]):

v = f0 (Re)v0 + f1 (Re)v1 + · · · ,
(2.4)
p = f0 (Re)p 0 + f1 (Re)p 1 + · · · ,

such that

fn+1 (Re)

→ 0 as Re
→ 0. (2.5)
fn (Re)

The leading-order terms v0 and p 0 of these expansions satisfy the Stokes system of
equations

∇ · v0 = 0, −∇p 0 + ∇2 v0 = 0, (2.6)

whereas the first-order terms v1 and p 1 satisfy the following equations:

2
f0 (Re)
0 
∇ · v1 = 0, −∇p 1 + ∇2 v1 − Re v · ∇ v0 = 0. (2.7)
f1 (Re)

On the other hand, in the outer region, where |x| ≥ ᏻ(Re−1 ), the inertia term is not
negligible, and hence it must be taken into consideration. Therefore, in this region
the expansions (2.4) are not valid. For this reason, we introduce the new characteristic
c = µUc /lc = Re pc (the characteristic pressure), and vc = U∞
variables lc = lc / Re, p
(the characteristic velocity). Also, we denote by x , v  the position vector of an
 , and p
arbitrary point and, respectively, the velocity and pressure fields corresponding to the
outer region. Then the governing equations take the form

i
∂v ∂p j
∂2v j
∂v
= 0, − + i
−v = 0, (2.8)
i
∂x j ∂x
∂x i∂xi i
∂x
2528 MIRELA KOHR

where the summation convention rule after the repeated indices is used. The solution
(  is expressed in the form (see also [7])
v, p)

v v0 + f1 (Re)
 = f0 (Re) v1 + · · · ,
(2.9)
p  0 + f1 (Re)p
 = f0 (Re)p 1 + · · · ,

such that

fn+1 (Re)

→ 0 as Re
→ 0. (2.10)
fn (Re)

Clearly, the first term in each of the above asymptotic expansions corresponds to the
uniform flow. Hence, we consider

0 0

f0 (Re) = 1,  = i1 , 0 .
 ,p
v (2.11)

v1 , p 1 ) are
Therefore, the governing equations for (

j1
∂v 1
∂p j1
∂2v j1
∂v
= 0, − + − = 0, (2.12)
j
∂x j ∂x
∂x k∂xk ∂x1

that is, the continuity and Oseen’s equations.

We require that the boundary condition (2.2) be satisfied by the first of the expan-
sions (2.4), and that the uniform stream conditions at infinity (2.3) be satisfied by the
asymptotic expansions (2.9). Additionally, we have to apply the matching principle in
the overlapping domain between the inner and outer regions, from which we obtain
other asymptotic conditions for each expansion and the possibility to compute succes-
sive terms of these expansions.

3. The solution of the leading-order problem in the inner region. We next deter-
mine the solution (v0 , p 0 ) of the leading-order problem in the inner region by using the
boundary integral method of Hsiao and Kress [1]. Therefore, we consider the following
boundary integral representation of the flow field v0 :
     
F 1 1
v0 (x) = V x, − + η0 V x, Φ− Φ dl
4π |Γ | 4π |Γ | Γ
   (3.1)
1
+ W x, Φ − η1 Φ(y)dl(y),
4π Γ

where V(·, Ψ ) is the single-layer potential with continuous density Ψ , given by


Vj (x, Ψ ) = Ᏻji (x − y)ψi (y)dl(y), (3.2)
Γ

W(·, h) is the double-layer potential with continuous density h, given by

 
S S
Wj (x, h) = Kij (y, x)hi (y)dl(y) = Sijk (y − x)nk (y)hi (y)dl(y), (3.3)
Γ Γ
 AN APPLICATION OF THE METHOD OF MATCHED ASYMPTOTIC. . . 2529
S
Ᏻji are the components of the two-dimensional Stokeslet Ᏻ, and Sijk are the components
S
of the stress tensor S associated with the two-dimensional Stokeslet. These compo-
nents are given by (see, e.g., [5, 10])

xj x i S xi x j x k
Ᏻji (x) = −δji ln |x| + , Sijk (x) = −4 . (3.4)
|x|2 |x|4

In addition, |F| is the total force exerted by the flow (v0 , p 0 ) on Γ , n is the outward
unit normal vector to Γ , η0 and η1 are two real constants such that η0 > 0 and η1 = 0,

|Γ | = Γ dl is the length of the curve Γ , and Φ is an unknown continuous vector density
on Γ . The total force F will be determined from the matching principle.
The corresponding boundary integral representation of the pressure field p 0 is given
by
       
F 1 1 1
p 0 (x) = P s x, − + η0 P s x, Φ− Φ dl + P d x, Φ , (3.5)
4π |Γ | 4π |Γ | Γ 4π

where P s (·, Ψ ) is the pressure field associated with the single-layer potential V(·, Ψ ),
that is,

P s (x, Ψ ) = ΠiS (x − y)ψi (y)dl(y), (3.6)
Γ

and P d (·, h) is the pressure field associated with the double-layer potential W(·, h),
that is,

P d (x, h) = ΛSik (x − y)nk (y)hi (y)dl(y), (3.7)
Γ

ΠiS and ΛSik being the components of the pressure vector ΠS and of the pressure tensor
ΛS , respectively, associated with the two-dimensional Stokeslet. These components are
given by the formulas (see, e.g., [10])

xi δik xi x k
ΠiS (x) = 2 , ΛSik (x) = −4 +8 . (3.8)
|x|2 |x|2 |x|4

Note that the single-layer potential V(·, Ψ ) is continuous across the Lyapunov contour
Γ , but the double-layer potential W(·, h) has a jump provided by the following limiting
values on both sides of Γ :
 PV




Wj± x0 , h = ±2π hj x0 + S
Sijk y − x0 nk (y)hi (y)dl(y), x0 ∈ Γ , (3.9)
Γ

where the plus sign applies for the external side of Γ (in the direction of the unit normal
vector) and the minus sign applies for the internal side of Γ . Also, the symbol PV means
the principal value of the double-layer potential at an arbitrary point x0 ∈ Γ (note that
S
the kernel Kij of Wj is weakly singular, but the corresponding double-layer integral has
a well-defined value at any point of Γ . For more details, see, for example, [9, Chapter 5]).
2530 MIRELA KOHR

Now, applying the boundary condition (2.2) to the flow field defined by the boundary
integral representation (3.1), and using the above-mentioned properties of single- and
double-layer potentials, we obtain the following Fredholm integral equation of the sec-
ond kind with unknown continuous vector density Φ = (φ1 , φ2 ):
   
1 F
I + K d + η0 K s M − η1 |Γ |(I − M) Φ = V ·, on Γ , (3.10)
2 4π |Γ |

where I : C 0 (Γ ) → C 0 (Γ ) is the identity operator, M : C 0 (Γ ) → C 0 (Γ ) is the operator

given by

1
Mh = h − h dl, h ∈ C 0 (Γ ), (3.11)
|Γ | Γ

and K s : C 0 (Γ ) → C 0 (Γ ) and K d : C 0 (Γ ) → C 0 (Γ ) are the single- and double-layer integral

operators given by the relations
   PV

s  1
d  1
K h (x) = V x, h , K h j (x) = Sijk (y − x)nk (y)hi (y)dl(y) (3.12)
4π 4π Γ

for h ∈ C 0 (Γ ) and x ∈ Γ . Note that both single- and double-layer integral operators are
compact on C 0 (Γ ).
By using the notation Φ = |F|Ψ , the above equation becomes
   
1 i1
I + K d + η0 K s M − η1 |Γ |(I − M) Ψ = V ·, on Γ . (3.13)
2 4π |Γ |

We mention that Hsiao and Kress [1] proved that (3.13) possesses a unique contin-
uous solution Ψ . Therefore, this density provides the unique continuous solution Φ of
(3.10).

4. The solution of the first-order problem in the outer region. As in [7], we take into
account the fact that far from the cylindrical body the role of the cylinder is similar to
that of a point force. Consequently, if we allow some point force located at the origin to
act on the fluid, then the first-order problem (i.e., the Oseen problem) could be satisfied
in the outer region. Therefore, it is sufficient to consider the two-dimensional Oseenlet,
that is, the fundamental solution of Oseen’s equation, which in outer variables is given
by (see [6, 7] and [9, page 238])
 
O 1 x 1 /2 |
x|
Ᏻij x) = −
( e K0 δij
4π 2
    (4.1)
1 |
x| 2 i
x x  1 δi2
 2 δi1 − x
− ex 1 /2 K1 − δij + δ2j ,
4π 2 |
x| |x| |
x|

where K0 and K1 are the modified Bessel functions of the second kind and of orders
0 and 1, respectively. Note that the above fundamental solution can be derived if we
include the term −ij δ(x) on the left-hand side of the second equation of (2.12), and
then use the method of Fourier transform. Here, δ denotes the Dirac distribution or
the delta function in R2 .
 AN APPLICATION OF THE METHOD OF MATCHED ASYMPTOTIC. . . 2531

5. The matching principle for the inner and outer expansions. We have seen that
the zeroth-order solution for the outer region is a uniform flow described by f0 (Re) =
 0 ) = (i1 , 0). Therefore, the matching principle requires that the zeroth-order
v0 , p
1, (
solution for the inner region should become



lim f0 (Re) v0 , p 0 = i1 , 0 . (5.1)
Re→0
|x|1

On the other hand, the singular behavior of the flow field v0 at large distances is
provided by the two-dimensional Stokeslet, and hence it is of logarithmic type (see
(3.1)). Therefore, for large |x| = ᏻ(Re−1 ), where Re → 0, we have the relation

f0 (Re)
f0 (Re)vi0 (x) ∼ (ln Re)Fi , (5.2)
4π

which shows that the matching condition (5.1) is satisfied to leading order if

4π
i1 = f0 (Re)F. (5.3)
ln Re

Consequently, the complete velocity field up to the leading-order solution of the Stokes
problem is given by

f0 (Re)v0 (x)
   
4π i1 1 1 1
= V x, − + η0 V x, Ψ− Ψ dl + W x, Ψ − η1 Ψ dl ,
ln Re 4π |Γ | 4π |Γ | Γ 4π Γ
(5.4)

where Ψ is the unique continuous solution of (3.13). Moreover, we have

f0 (Re)p 0 (x)
   (5.5)
4π i1 1 1 1
= P s x, − + η0 P s x, Ψ− Ψ dl + P d x, Ψ .
ln Re 4π |Γ | 4π |Γ | Γ 4π

Furthermore, taking into account the asymptotic expansions (2.9) and the expres-
sions (5.4) and (5.5), we find that

4π
f1 (Re) = (5.6)
ln Re

and that (  1 ) is the flow due to an Oseenlet located at the origin and oriented in the
v1 , p
x1 -direction. Therefore, we have

f1 (Re)v
i1 (
x)
 
4π 1 x 1 /2 |
x| 1 |
x| 2 x i (5.7)
 1 /2
x
= − e K0 δi1 − e K1 − δi1 .
ln Re 4π 2 4π 2 |
x| |
x|
2532 MIRELA KOHR

In order to study the asymptotic behavior of the inner flow velocity field far from
the origin, we expand the two-dimensional Stokeslet Ᏻ(x − y) in a Taylor series with
respect to y about the origin. Then we obtain the following asymptotic expansion of
the inner flow velocity field:
  
4π δi1 1
f0 (Re)vj0 (x) = − Ᏻji (x) + · · · + Wj x, Ψ − η1 ψj dl
ln Re 4π 4π Γ
   (5.8)
4π 1 1
= − Ᏻj1 (x) + · · · + Wj x, Ψ − η1 ψj dl .
ln Re 4π 4π Γ

The above expansion shows that the asymptotic form of the inner flow velocity field
far from the origin is the velocity field due to a Stokeslet located at the origin plus
a constant vector. Moreover, we use the fact that the outer flow velocity field up to
the first-order approximation at large distances from the origin is the velocity field of
a uniform flow in the x1 -direction plus the velocity field due to an Oseenlet located
at the origin and oriented in the x1 -direction. Therefore, we find that the mismatch
between f0 (Re)v0 , from the inner region, and the sum f0 (Re)
v0 + f1 (Re)
v1 , from the
outer region, has the same terms as the mismatch provided by Proudman and Pearson
[11], up to the same order of approximation for the singular perturbation solution of
a uniform flow past a circular cylinder at small Reynolds number (see also [7] and [9,
page 240]). This mismatch denoted by

v0 + f1 (Re)
∆ ≡ f0 (Re) v1 − f0 (Re)v0 (5.9)

is hence given by

1

∆∼ γ0 − ln 4 i1 + 4π η1 Ψ dl , (5.10)
ln Re Γ

where γ0 = 0.5772 . . . is the Euler constant and Ψ is the unique continuous solution of
(3.13). Further, according to the fact that this mismatched uniform flow is ᏻ((ln Re)−1 ),
we deduce that (v1 , p 1 ) will be a solution of the Stokes equation, since the term (v0 ·
∇)v0 in the second equation of (2.7) is asymptotically negligible with respect to any
inverse power of ln Re. Consequently, the first-order inner velocity and pressure fields
are also given by (3.1) and (3.5) with

4π

f1 (Re) = , F = γ 0 − ln 4 i 1 + 4π η 1 Ψ dl. (5.11)
(ln Re)2 Γ

Hence the vector density of the corresponding double-layer potential for the first-order
approximation is given by the unique continuous solution of (3.10) with the constant
vector F given by the second equation of (5.11). Accordingly, the hydrodynamic force
FT acting on the cylinder is provided by the single-layer potentials V(·, −(4π )−1 |Γ |−1 i1 )
and V(·, −(4π )−1 |Γ |−1 F), where F has been mentioned previously. The components FT ,j
of this force, up to ᏻ((ln Re)−2 ), are given by

4π 4π


FT ,j = δ1j + γ0 − ln 4 δ1j + 4π η1 ψj dl + ᏻ (ln Re)−3 . (5.12)
ln Re (ln Re)2 Γ
 AN APPLICATION OF THE METHOD OF MATCHED ASYMPTOTIC. . . 2533

From (5.12) it follows that the contribution of ᏻ((ln Re)−1 ) to the hydrodynamic force
does not depend on the cylinder geometry. However, the contribution of ᏻ((ln Re)−2 )

to the drag force depends on the cylinder geometry, since it contains the term Γ ψ1 dl.
Also, the lift force of ᏻ((ln Re)−2 ) depends on the cylinder geometry by the fact that it

is expressed in terms of Γ ψ2 dl, where Ψ = (ψ1 , ψ2 ) is the unique continuous solution
of (3.13). Consequently, the dependency of the hydrodynamic force on the cylinder
geometry appears in the second-order approximation and is provided by the matching
asymptotic procedure.
In the case η1 = 1/(4π ), the asymptotic formula (5.12) reduces to that obtained by
Power [7]; see also [9, page 240].

6. Inertial effects. The inner and outer asymptotic expansions (2.4) and (2.9), for
which it can be proved that fn (Re) = (ln Re)−(n+1) and fn (Re) = (ln Re)−n , do not in-
clude the inertial effects of order ᏻ(Re) that are asymptotically smaller than the terms
of order ᏻ((ln Re)−n ) for each n as Re → 0. Proudman and Pearson [11] proposed a pro-
cedure for removing this inconvenience, which consists in the addition of the following
series to the expansions (2.4) and (2.9) (see also [7] and [9, page 242]):
∞
 ∞
 ∞
 ∞

Rem (ln Re)−n vm,n , Rem (ln Re)−n p m,n , (6.1)
m=0 n=0 m=0 n=0

and, respectively,
∞
 ∞
 ∞
 ∞

Rem (ln Re)−n v
 m,n , Rem (ln Re)−n p
 m,n . (6.2)
m=0 n=0 m=0 n=0

We have to require that these expansions satisfy the same boundary and matching
conditions as in the previous analysis. The terms vm,n and p m,n can be determined
by using similar arguments as before. These terms satisfy either the homogeneous
Stokes equation or nonhomogeneous versions of this equation. For example, the terms
v0,0 and p 0,0 lead to a nonhomogeneous equation for v1,2 and p 1,2 , whose particular
1,2 1,2
solution (vp , pp ) can be expressed in terms of the two-dimensional Stokeslet and its
associated pressure vector as follows:



v1,2
p (x) = Ᏻ(x − y) · v0,0 (y) · ∇v0,0 (y) dy,
D
 (6.3)


pp1,2 (x) = ΠS (x − y) · v0,0 (y) · ∇v0,0 (y) dy,
D

1,2 1,2
where D is the flow domain. Note that the particular solution (vp , pp ) has to be added
to the general solution given by (3.1) and (3.5) in order to complete the inner solution
(v1,2 , p 1,2 ).
If we use arguments similar to those for the terms (vm,n , p m,n ), all the terms
(
v m,n
 m,n ) can be computed too. They satisfy either the homogeneous Oseen equa-
,p
tion or nonhomogeneous versions of this equation. These outer higher-order approxi-
mations were obtained by Proudman and Pearson [11]. Both inner and outer solutions
can be completed by applying the above matching asymptotic method.
2534 MIRELA KOHR

7. Conclusions. In this paper, we have applied the method of matched asymptotic

expansions to the low Reynolds number flow of an incompressible Newtonian fluid
past a cylinder of arbitrary cross section. The hydrodynamic force on the cylinder is
expressed in terms of the unique continuous solution of the Fredholm integral equation
of the second kind (3.13). We note that this equation is uniquely solvable when the
cylinder cross-sectional boundary is an arbitrary simple closed Lyapunov curve and
the parameters η0 and η1 satisfy the conditions η0 > 0 and η1 = 0. For η1 = 1/(4π ),
the hydrodynamic force FT given by formula (5.12) is identical to the corresponding
result due to Power [7]. Note that the matched asymptotic analysis developed by Power
is based on the CDLBIEM, in order to solve the resulting inner problems, and on the
singularity method applied to solve the resulting outer problems (see also [9, Section
6.3.4]).
The matched asymptotic analysis developed in this paper differs from that of Power
[7], since we have used the compound double-layer method due to Hsiao and Kress [1]
instead of the CDLBIEM, which is the basis of Power’s approach.
Finally, we note that another matched asymptotic analysis of low Reynolds number
flow past a cylinder of arbitrary cross section was obtained by Lee and Leal [6]. This
analysis uses the boundary integral method due to Youngren and Acrivos [14], which
reduces the resulting inner problems to integral equations of the first kind.

References

[1] G. C. Hsiao and R. Kress, On an integral equation for the two-dimensional exterior Stokes
problem, Appl. Numer. Math. 1 (1985), no. 1, 77–93.
[2] S. Kaplun, Low Reynolds number flow past a circular cylinder, J. Math. Mech. 6 (1957),
595–603.
[3] S. J. Karrila and S. Kim, Integral equations of the second kind for Stokes flow: direct solution
for physical variables removal of inherent accuracy limitations, Chem. Eng. Commun.
82 (1989), 123–161.
[4] S. Kim and S. J. Karrila, Microhydrodynamics: Principles and Selected Applications,
Butterworth-Heinemann, London, 1991.
[5] O. A. Ladyzhenskaya, The Mathematical Theory of Viscous Incompressible Flow, Gordon and
Breach Science Publishers, New York, 1963, translated from the Russian by Richard
A. Silverman.
[6] S. H. Lee and L. G. Leal, Low-Reynolds-number flow past cylindrical bodies of arbitrary
cross-sectional shape, J. Fluid Mech. 164 (1986), 401–427.
[7] H. Power, Matched-asymptotic analysis of low-Reynolds number flow past a cylinder of ar-
bitrary cross-sectional shape, Mat. Apl. Comput. 9 (1990), no. 2, 111–122.
[8] H. Power and G. Miranda, Second kind integral equation formulation of Stokes’ flows past
a particle of arbitrary shape, SIAM J. Appl. Math. 47 (1987), no. 4, 689–698.
[9] H. Power and L. C. Wrobel, Boundary Integral Methods in Fluid Mechanics, Computational
Mechanics Publications, Southampton, 1995.
[10] C. Pozrikidis, Boundary Integral and Singularity Methods for Linearized Viscous Flow,
Cambridge Texts in Applied Mathematics, Cambridge University Press, Cambridge,
1992.
[11] I. Proudman and J. R. A. Pearson, Expansions at small Reynolds numbers for the flow past
a sphere and a circular cylinder, J. Fluid Mech. 2 (1957), 237–262.
[12] K. Shintani, A. Umemura, and A. Takano, Low-Reynolds-number flow past an elliptic cylin-
der, J. Fluid Mech. 136 (1983), 277–289.
 AN APPLICATION OF THE METHOD OF MATCHED ASYMPTOTIC. . . 2535

[13] A. Umemura, Matched-asymptotic analysis of low-Reynolds-number flow past two equal

circular cylinders, J. Fluid Mech. 121 (1982), 345–363.
[14] G. K. Youngren and A. Acrivos, Stokes flow past a particle of arbitrary shape: a numerical
method of solution, J. Fluid Mech. 69 (1975), 377–403.

Mirela Kohr: Department of Applied Mathematics, Faculty of Mathematics and Computer Sci-
ence, Babeş-Bolyai University, 1 M. Kogălniceanu Street, 3400 Cluj-Napoca, Romania
E-mail address: mkohr@math.ubbcluj.ro
Advances in Advances in Journal of Journal of
Operations Research
Hindawi Publishing Corporation
Decision Sciences
Hindawi Publishing Corporation
Applied Mathematics
Hindawi Publishing Corporation
Algebra
Hindawi Publishing Corporation
Probability and Statistics
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

The Scientific International Journal of

World Journal
Hindawi Publishing Corporation
Differential Equations
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Submit your manuscripts at

http://www.hindawi.com

International Journal of Advances in

Combinatorics
Hindawi Publishing Corporation
Mathematical Physics
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of Journal of Mathematical Problems Abstract and Discrete Dynamics in

Complex Analysis
Hindawi Publishing Corporation
Mathematics
Hindawi Publishing Corporation
in Engineering
Hindawi Publishing Corporation
Applied Analysis
Hindawi Publishing Corporation
Nature and Society
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

International
Journal of Journal of
Mathematics and
Mathematical
Discrete Mathematics
Sciences

Journal of International Journal of Journal of

Hindawi Publishing Corporation Hindawi Publishing Corporation Volume 2014

Function Spaces
Hindawi Publishing Corporation
Stochastic Analysis
Hindawi Publishing Corporation
Optimization
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014