main 2012/4/9 12:45 page 19 #1

Volume 31, N. 1, pp. 1935, 2012

Copyright 2012 SBMAC
ISSN 0101-8205 / ISSN 1807-0302 (Online)
www.scielo.br/cam
A stabilized fnite element method to pseudoplastic
fow governed by the Sisko relation
MARCIO ANTNIO BORTOLOTI
1
and JOS KARAM FILHO
2
1
Departamento de Cincias Exatas, Universidade Estadual do Sudoeste da Bahia, UESB,
Estrada do Bem-Querer, Km 4, 45083-900 Vitria da Conquista, BA, Brazil
2
Coordenao de Mecnica Computacional, Laboratrio Nacional de Computao Cientfca,
LNCC, Av. Getlio Vargas, 333, 25651-070 Petrpolis, RJ, Brazil
E-mails: marciob@lncc.br / jkf@lncc.br
Abstract. In this work, a consistent stabilized mixed fnite element formulation for incom-
pressible pseudoplastic fuid fows governed by the Sisko constitutive equation is mathematically
analysed. This formulation is constructed by adding least-squares of the governing equations and
of the incompressibility constraint, with discontinuous pressure approximations, allowing the use
of same order interpolations for the velocity and the pressure. Numerical results are presented to
confrm the mathematical stability analysis.
Mathematical subject classifcation: Primary: 65M60; Secondary: 65N30.
Key words: fnite element method, numerical analysis, Sisko relation, non-Newtonian fow.
1 Introduction
In modeling some kinds of fuids, the Sisko constitutive relation comes out
from the Cross model [6], when the apparent viscosity lies in a range between
the pseudoplastic region and the lower Newtonian plateau. A good alterna-
tive constitutive equation of the Sisko type for blood fow has been proposed
by [13], for example.
#CAM-240/10. Received: 10/VIII/10. Accepted: 10/XI/11.
*Corresponding author
main 2012/4/9 12:45 page 20 #2
20 STABILIZED FEM TO PSEUDOPLASTIC FLOW
The nonlinearity of this relation together with the incompressibility constraint
may generate numerical instabilities when some classical numerical methods
are used. For classical methods, in case of velocity and pressure formulations,
it is well known that, even for the linear case, different interpolation orders for
these variables have to be used in order to satisfy the Babuka-Brezzi stability
condition [2, 4].
In this work, to avoid the use of penalization methods or reduced integration
and to recover the stability and accuracy of the solution of same interpolation
orders in primitive variables, a consistent mixed stabilized fnite element formu-
lation is presented. It is constructed by adding the least-squares of the governing
equations and the incompressibility constraint, with continuous velocity and
discontinuous pressure interpolations. The present formulation is here mathem-
atically analysed based on Scheurers theorem, [12]. Stability conditions and
error estimates are established when the Sisko relation is considered. Numerical
examples are presented to confrm the stability analysis.
2 Defnition of the problem
Let be a bounded domain in R
n
where the positive integer n, denotes the
space dimension. We consider the stationary incompressible creep fowproblem
governed by div = f in , where : R
n
R
n
denotes the Cauchy
stress tensor for the fuid and f denotes the body forces.
The governing equation, written above, is subjected to the incompressibility
constraint divu = 0 in , where u denotes the velocity feld.
The Sisko model is characterized by a linear supersposition between the
Newtonian and the pseudoplastic effects, presenting a dependence of the viscos-
ity with the shear-strain rate tensor, defning an apparent viscosity (|(u)|),
leading to the stress tensor of the form
= pI +(|(u)|)(u) with (|(u)|) =
1
+
2
(|(u)|)
and (s) = s
2
, where
1
,
2
are two positive constitutive constants, ]1, 2[
is the power index, p is the hydrostatic pressure, I R
n
R
n
is the identity
tensor, | | denotes the Euclidean tensor norm and
(u) =
1
2
_
u +
T
u
_
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 21 #3
MARCIO ANTNIO BORTOLOTI and JOS KARAM FILHO 21
is the symmetric part of the gradient of u.
With the above considerations, together with boundary condition of Dirichlet
type, the resulting problem is: fnd (u, p) C
2
() C
1
() such that
_

_
div((|(u)|)(u)) +p = f in
divu = 0 in
u = u on
(1)
where denotes the boundary of .
Physically, pseudoplastic fows are characterized by a viscosity decreasing
continuously and smoothly with increasing of shear rate, and this behaviour
occurs in a limited range of shear rate, generating the viscosity plateaus, where
we can see that the apparent viscosity (s) is a bounded continuous function
such that

(s)
0
(2)
for 0 < g
0
s g

with g
0
and g

being the shear rate fnite limits and

0
and

corresponding to the fnite limiting Newtonian plateaus for low and

high shear rate, respectively, [3].
It can be seen that from continuous classical Galerkin formulation associated
with problem (1), we can obtain
(u)
0

C

f
0
(3)
for all u W
1,2
0
(), where C is the constant in Poincars inequality.
3 Petrov-Galerkin-Like formulation
To generate the stabilized fnite element method proposed here, the following
defnitions will be used.
Let L
p
() = {u|u is measurable,
_

|u(x)|
p
d < } be the class of all
measurable functions u, such that, u is p-integrable in and let L
p
0
() = {u
L
p
(),
_

u d = 0} be the class of functions in L

p
() such that u has null
mean. Let W
m, p
() be the Sobolev space W
m, p
() = {u L
p
()|D

u
L
p
(), 0 || m} with
D

u =

||
u

1
x
1

m
x
m
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 22 #4
22 STABILIZED FEM TO PSEUDOPLASTIC FLOW
where
i
is a natural integer and || =
1
+ +
m
. The W
m, p
0
() is defned
as the space of functions u W
m, p
() such that D

u = 0 on , for all with

|| m 1, [1].
The norm in the space W
m, p
() is defned as
u
m, p
=
_
_

||m
D

u
p
_
_
1/p
, 1 p < ,
where u
p
is the L
p
() norm defned as
u
p
=
__

|u(x)|
p
dx
_
1/p
, x .
In this paper, we will denote u
1
= u
1,2
and u
0
= u
2
.
We assume for simplicity R
n
, a polygonal domain discretized by a
classical uniform mesh of fnite elements with N
e
elements, such that
=
N
e
_
e=1

e
,
e
i

e
j
= for all i = j
where
e
denotes the interior of the e
t h
element and
e
is its closure.
Let S
k
h
() be the fnite element space of the Lagrangean continuous poly-
nomials in of degree k and Q
l
h
() the fnite element space of the Lagrangean
discontinuous polynomials in of degree l. Thus we can defne the approxima-
tion spaces V
h
= (S
k
h
() W
1,2
0
())
n
and W
h
= Q
l
h
() L
2
() to velocity
and pressure respectively, that can be generated by triangles or quadrilaterals.
Remark 3.1. For the Galerkin method, k and l must be different orders even for
k 2 and one can follow [7, 8] and [9], for example, to see the limitations for
the combinations of k and l.
Remark 3.2. With the present consistent stabilized formulation, all the combi-
nations of different orders are possible, optimal and suboptimal, but it is possible
to use same orders for k and l, with complete polynomials, providing k 2, as
follows.
In this work, to obtain velocity and pressure approximations to problem (1),
we defne the following variational form, constructed by adding the least squares
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 23 #5
MARCIO ANTNIO BORTOLOTI and JOS KARAM FILHO 23
of the linear momentum and of the continuity equations to the Galekin formula-
tion, generating the following problem with homogeneous boundary condition
considered without lost of generalities:
ProblemPG
hd
. Given f W
1,2
(), the dual of W
1,2
(), fndU
h
V
h
W
h
,
such that
_
_
_
(A
h
(U
h
), V
h
) + B
h
( p
h
, v
h
) = F
h
(V
h
) V
h
V
h
W
h
B
h
(q
h
, u
h
) = 0 q
h
W
h
where
(A
h
(U
h
), V
h
) = ((|(u
h
)|)(u
h
), (v
h
)) +
2
(divu
h
, divv
h
)
+

1
h
2

u
h
+p
h
,

v
h
+q
h
)
h
,
(4)
B
h
( p
h
, v
h
) = ( p
h
, divv
h
), (5)
F
h
(V
h
) = f(v
h
) +

1
h
2

(f,

v
h
+q
h
)
h
, (6)
with h denoting the mesh parameter, (|(u
h
)|) the apparent viscosity,
(u, v) =
_

u v dx, (u, v)
h
=
N
e

e=1
_

e
u v dx,

1
and
2
being positive constants denoted as stability parameters,

u
h
=
div((|(u
h
)|)(u
h
)), U
h
= {u
h
, p
h
}, V
h
= {v
h
, q
h
} and being a dimensional
parameter. We can note that when
1
=
2
= 0, Problem PG
hd
reduces to
the Galerkin formulation which, for interpolations of same order, is unstable
exhibiting spurious pressure modes or presenting the locking of the velocity
feld, [9]. The nonlinear Problem PG
hd
preserves the good properties of the
linear analogous of [11] in the sense that it accommodates, for k 2 equal-
order interpolations for velocity and pressure as will be shown in the following
analysis and confrmed later by the obtained numerical results.
This formulation is consistent, being easy to verify that the exact solution of
problem defned in (1), U = {u, p}, satisfes the PG
hd
problem.
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 24 #6
24 STABILIZED FEM TO PSEUDOPLASTIC FLOW
4 Finite element analysis
The fnite element analysis is developed here by considering solutions in Hilbert
Spaces, as in [10]. We start the analysis by rewriting the discontinuous pres-
sure approximation p
h
as
p
h
= p

h
+ p
h
, (7)
with p

h
W

h
and p
h
W
h
where W

h
= { p

h
W
h
L
2
0
(
e
), p
e
h
=
p

h
} is the subspace of the pressure with zero mean at the element level and
W
h
= { p
h
W
h
; p
e
h
= 0, p
e
h
=
_

e
p
e
h
d
e
/
_

e
d
e
} is the subspace of
the piecewise constant pressure, where p
e
h
represents the constraint of p
h
in
element
e
.
The discontinuous pressure allows the satisfaction of the incompressibility
constraint at element level in contrast to the continuous approximations, which
satisfes the constraint only in global sense. Considering this segregation, the
Problem PG
hd
can be rewritten as the following variational form, which consid-
ers the pressure variable p
h
written as functions of p

h
W

h
() and p
h
W
h
,
as was described above:
Problem PG

hd
. Given f W
1,2
(), fnd {u
h
, p

h
, p
h
} V
h
W

h
W
h
,
such that
_
_
_
(A

h
(U

h
), V

h
) + B
h
( p
h
, v
h
) = F

h
(V

h
) V

h
V
h
W

h
B
h
(q
h
, u
h
) = 0 q
h
W
h
where
(A

h
(U

h
), V

h
) = ((|(u
h
)|)(u
h
), (v
h
)) + B
h
( p

h
, v
h
) + B
h
(q

h
, u
h
)
+
2
(divu
h
, divv
h
) +

1
h
2

u
h
+p

h
,

v
h
+q

h
)
h
,
(8)
B
h
( p

h
, v
h
) = ( p

h
, divv
h
), (9)
B
h
( p
h
, v
h
) = ( p
h
, divv
h
), (10)
F

h
(V

h
) = f(v
h
) +

1
h
2

(f,

v
h
+q

h
)
h
, (11)
U

h
= {u
h
, p

h
} and V

h
= {v
h
, q

h
}.
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 25 #7
MARCIO ANTNIO BORTOLOTI and JOS KARAM FILHO 25
Lemma 4.1. For (s) bounded, continuous and smooth real function such that
|d(s)/ds| M, there exists a positive constant C, independent of h, such that
h

u
h

0h
C(u
h
)
0
where u
2
0h
= (u, u)
h
.
Proof. From the inverse estimate
hdiv(u
h
)
0h
C
h
(u
h
)
0
, (12)
typical of fnite element methods, [5], 0 <

(s)
0
and |d(s)/ds|
M, we have the inverse estimate proposed, with C =
0
C
h
+ M.
Lemma 4.2. Assuming the same considerations of Lemma 4.1, there is a
positive constant C
l
, independent of h, such that, h

u
h
+

v
h

0h

C
l
(u
h
) (v
h
)
0
for all u
h
, v
h
V
h
with h > 0.
Proof. From the triangular inequality, the Lemma 4.1 and (12) we have
h
2

u
h
+

v
h

2
0h
4[(
2
0
C
2
h
+ M
2
)(u
h
v
h
)
2
0
+(1 +C
2
h
)(v
h
)
2
0
(u
h
) (v
h
)
2
0h
].
The mean value theorem yields
h
2

u
h
+

v
h

2
0h
4
_
(
2
0
C
2
h
+ M
2
)(u
h
) (v
h
)
2
0
+(1 +C
2
h
)(v
h
)
2
0
sup
01
(|((1 )u
h
+v
h
)|)
2
0h
u
h
v
h

2
0h
_
.
Since |d(s)/ds| M for all u
h
V
h
and from Korns inequality, we can
conclude the result, with the constant C
l
given by
C
l
= 4
_
(
2
0
C
2
h
+ M
2
) +(1 +C
2
h
)M
2
C
2
K
sup
v
h
V
h
(v
h
)
2
0
_
where C
K
is the constant of the Korns inequality. The lemma is obtained as a
consequence of
(u
h
)
0
C(
1
,
2
,

, C,
1
, )f
0
for all u
h
V
h
, (13)
that is, naturally, obtained from Problem PG
hd
and consequently yields C
l
as a
fnite constant, where C is the constant of the Poincar inequality.
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 26 #8
26 STABILIZED FEM TO PSEUDOPLASTIC FLOW
Defnition 4.3. Let |U
h
| = U
h
+h(div(u
h
)
0h
+(u
h
)
0h
+p
h

0h
)
be a mesh-dependent norm on the product space H
1
0
() L
2
(), where h
denotes the mesh parameter and U
h

2
= u
h

2
1
+p
h

2
0
is the norm defned in
V
h
W
h
.
Lemma 4.4 (Equivalence of the norms). There exists a positive constant
such that U
h
|U
h
| U
h
for all U
h
V
h
.
Proof. The inequality U
h
|U
h
| is immediate. In other hand, from the
defnition of |U
h
|, fromthe inverse estimate (12) and fromthe inverse estimate
for the pressure, see [5],
hp

0h
C
p
p

0
,
we have |U
h
| U
h
+ (C
h
+ 1)(u
h
)
0
+ C
p
p
h

0
. Using the classical
inequality,
1

n
divu
0
(u)
0
u
1
, (14)
we complete the proof of Lemma 4.4, with = 1 +max{C
h
+1, C
p
}.
With the above results we can establish the following result that will be needed
later to generate the estimates in Theorem 4.9.
Theorem 4.5. There exists a positive constant
c
such that
|(A

h
(U

) A

h
(V
h
), U

h
V

h
)|
c
|U

h
| U

h
V

for all U

W
1,2
0
() L
2
() and U

h
, V

h
V
h
W

h
.
Proof. By the consistency of the problem PG

hd
and from (4) we have
|(A

h
(U

) A

(V

h
), U

h
V

h
)|
1
|((u v
h
), (u
h
v
h
))|
+
2
|((|(u)|)(u) (|(v
h
)|)(v
h
), (u
h
v
h
))| +|B
h
( p

h
q

h
, u v
h
)|
+
2
|(divu divv
h
, divu
h
divv
h
)| +|B
h
( p

h
, u
h
v
h
)|
+

1
h
2

u +

v
h
+( p

h
),

u
h
+

v
h
+( p

h
q

h
)
_
h

Comp. Appl. Math., Vol. 31, N. 1, 2012

main 2012/4/9 12:45 page 27 #9
MARCIO ANTNIO BORTOLOTI and JOS KARAM FILHO 27
From the continuity of the two frst terms in the right hand side above, [12], and
using the inequality (14) we have
|(A

h
(U

) A

h
(V

h
), U

h
V

h
)| [(
1
+
2
+n
2
)u v
h

1
+

np

0
]u
h
v
h

1
+

1
h

_
(|(u)|)div(u) (|(v
h
)|)div(v
h
)
0h
+p

0h
+(u)(u) (v
h
)(v
h
)
0h
_

_
u
h
v
h

1
+p

h
q

0
_
.
By using Lemma 4.2 and identifying the | | norm, we can conclude
|(A

h
(U

) A

h
(V

h
), U

h
V

h
)|
c
|U

h
| U

h
V

h
,
where
c
= max{
1
+
2
+n
2
,

n,

1

}.
Theorem 4.6. Let K
h
= {v
h
V
h
, B
h
(q
h
, v
h
) = 0, for all q
h
W
h
}. Then,
there exists a positive constant
e
such that (A

h
(U

h
) A

h
(V

h
), U

h
V

h
)

e
U

h
V

2
for all U

h
, V

h
K
h
W

h
.
Proof. From (8) we obtain
(A

h
(U

h
) A

h
(V

h
), U

h
V

h
)
1
(u
h
v
h
)
2
0
+
2
divu
h
divv
h

2
0h
+2B
h
( p

h
q

h
, u
h
v
h
)
+

1
h
2

u
h
+

v
h
+( p

h
q

h
)
2
0h
+
2
_
(|(u
h
)|)(u
h
) (|(v
h
)|)(v
h
), (u
h
v
h
)
_
.
From the ellipticity of the last term in the right hand side above, presented in
[12], and using the Young inequality with =
1

we have
(A

h
(U

h
) A

h
(V

h
), U

h
V

h
) (
2
+
1
)(u
h
) (v
h
)
2
0
+

1
h
2

u
h
+

v
h
+( p

h
q

h
)
2
0h

1

h
q

2
0
.
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 28 #10
28 STABILIZED FEM TO PSEUDOPLASTIC FLOW
Applying again the Young inequality it yields
(A

h
(U

h
) A

h
(V

h
), U

h
V

h
) (
2
+
1
)(u
h
) (v
h
)
2
0

h
q

2
0
+

1
h
2

(1
1

u
h
+

v
h

2
0h
+

1
h
2

(1 )p

h
q

2
0h
,
with > 0. From Lemma 4.2 and considering r
1
and r
2
as two positive con-
stants, such that
1
r
1
+
1
r
2
= 1, we can rewrite the previous inequality as
(A

h
(U

h
) A

h
(V

h
), U

h
V

h
)

2
+
1
r
1
(u
h
) (v
h
)
2
0
+h
2
_

2
+
1
r
2
C
l
+

1

_
1
1

_
_

u
h
+

v
h

2
0h
+

1
h
2

(1 )p

h
q

2
0h

1

h
q

2
0
.
Choosing =

1
r
2
C
l
(
2
+
1
)+
1
r
2
C
l
)
, we obtain
(A

h
(U

h
) A

h
(V

h
), U

h
V

h
)

2
+
1
r
1
(u
h
) (v
h
)
2
0

h
q

2
0
+

1
(
2
+
1
)h
2
(
2
+
1
) +
1
r
2
C
l
)
( p

h
q

h
)
2
0h
.
By using the inequality
h
2
q

2
0h
q

2
0
, (15)
as in [10], and the Korns inequality, we have
(A

h
(U

h
) A

h
(V

h
), U

h
V

h
)
e
(u
h
v
h

2
1
+p

h
q

2
0
)
with

e
= min
_
(
2
+
1
)C
r
1
,

1
(
2
+
1
)
(
2
+
1
) +
1
r
2
C
l

1

_
,
since

1
(
2
+
1
)
(
2
+
1
) +
1
r
2
C
l

1

> 0. (16)

Comp. Appl. Math., Vol. 31, N. 1, 2012

main 2012/4/9 12:45 page 29 #11
MARCIO ANTNIO BORTOLOTI and JOS KARAM FILHO 29
The inequality (16) gives a suffcient condition, providing a useful relation to
be used to choose the stabilizing parameters.
Theorem 4.7. There exists a positive constant
B
such that
B
h
( p p
h
, u
h
v
h
)
B
p p
h

0
U

h
V

for all p L
2
(), U

h
, V

h
V
h
W

h
.
Proof. This result comes from the application of the Hlder-Schwarz inequal-
ity and by the use of (14).
Theorem 4.8. For k 2 and since V
h
V
h
W
h
, there exists a positive
constant
h
, independent of h, such that
sup
v
h
V
h
|B
h
(q
h
, v
h
)|
v
h

1

h
q
h

0
for all q
h
W
h
.
Proof. This result may be seen in [7].
With the above results, we can establish the following error approximation
estimates.
Theorem 4.9. There exists a positive constant
h
, independent of h, such that
the following estimate holds |U U
h
|
h
|U V
h
|.
Proof. From the defnition of (A

h
(), ) and the consistency of the formula-
tion we can write
_
A

h
(U

h
) A

h
(V

h
), U

h
V

h
_
=
_
A

h
(U

) A

h
(V

h
), U

h
V

h
_
+
_
A

h
(U

h
) A

h
(U

), U

h
V

h
_
(17)
with
_
A

h
(U

h
) A

h
(U

), U

h
V

h
_
= B
h
_
p q
h
, u
h
v
h
_
. (18)
Replacing (18) in (17) we have
_
A

h
(U

h
) A

h
(V

h
), U

h
V

h
_
=
_
A

h
(U

) A

h
(V

h
), U

h
V

h
_
+ B
h
_
p q
h
, U

h
V

h
_
.
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 30 #12
30 STABILIZED FEM TO PSEUDOPLASTIC FLOW
From Theorem 4.5, Theorem 4.6, Theorem 4.7 and considering the norm equiv-
alence between | | and established in Lemma 4.4, we have
|U

h
|
_
1 +

c

e
_
|U

h
| +

B

e
p q
h

0
. (19)
In order to obtain an estimate to p p
h

0
, we note that from PG
hd
problem,
we have
B
h
_
p
h
, u
h
v
h
_
=
_
A

h
(U

) A

h
(U

h
), U

h
V

h
_
+ B
h
_
p, u
h
v
h
_
.
Since v
h
K
h
, then
B
h
_
p
h
q
h
, u
h
v
h
_
=
_
A

h
(U

) A

h
(U

h
), U

h
V

h
_
+ B
h
_
p q
h
, u
h
v
h
_
.
Using Theorem 4.5 and Theorem 4.7 we have
sup
w
h
V

h
B
h
( p
h
q
h
, w
h
)
w
h

1
+p

h
q

0
|U

h
| +p q
h

0
.
By the use of Theorem 4.8, we can see that
p p
h

0

1

h
U

h
+
_
1 +
1

h
_
p q
h

0
. (20)
Combining (19) and (20) we have
|U

h
|
h
|U

h
|,
where

h
= 1 +
1

h
+
1

e
max
_

c
,
B
_
,
since p
h

2
0
= p

2
0
+p
h

2
0
.
From Theorem 4.9, applying inverse estimates and the very classical interpo-
lation results presented in, for example, Chapter 3 of [5], we obtain the following
error estimate
U U
h

h
(2 +C
h
)c
1
h
k
|u|
k+1
+
h
(1 +C
p
)c
2
h
l+1
| p|
l+1
(21)
with c
1
, c
2
R, u
h
S
k
h
() and p
h
Q
l
h
() and | |
m+1
the semi-norm of
the W
m,2
().
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 31 #13
MARCIO ANTNIO BORTOLOTI and JOS KARAM FILHO 31
5 Numerical results
In order to obtain numerical results for the fnite element method presented here
to nonlinear problem, the following numerical algorithm will be used. There are
many methods to solve nonlinear equations. In this case, we lag nonlinear terms
in the systemof equations and start with an initial guess generating a sequence of
functions that is expected to converge for the solution. In this sense, our scheme
is constructed by: given u
0
h
V
h
, lets fnd (u
n
h
, p
n
h
) V
h
W
h
, n = 1, 2, 3, . . .
such that
((|(u
n
h
)|)(u
n+1
h
), (v
h
)) + B
h
( p
n+1
h
, v
h
) + B
h
(q
h
, u
n+1
h
)
+

1
h
2

(div((u
n
h
)(u
n+1
h
)) +p
n+1
h
,

v
h
+q
h
)
h
+
2
(divu
n+1
h
, divv
h
) = F

h
(V

h
), V
h
V
h
W
h
.
0
0.2
0.4
0.6
0.8
1
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
y
Horizontal Velocity

1
=0.1,
2
= 1.0, = 1.3

1
=1.0,
2
= 0.1, = 1.3

1
=1.0,
2
= 1.0, = 1.3
Newtonian Flow
Figure 1 Horizontal velocity at x = 0.5
The algorithm above, was applied to obtain numerical results for the clas-
sical driven cavity fow problem with bounbary conditions: u(x) = (1, 0) on
x [0, 1] {1} and u(x) = (0, 0) on the other boundaries. A fnite element
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 32 #14
32 STABILIZED FEM TO PSEUDOPLASTIC FLOW
(a) Velocity feld (b) Pressure feld
Figure 2 Results for
1
= 1.0,
2
= 0.1 and = 1.3 using
1
= 1.0 and
2
= 10.0.
(a) Velocity feld (b) Pressure feld
Figure 3 Results for
1
= 1.0,
2
= 1.0 and = 1.3 using
1
= 1.0 and
2
= 10.0.
(a) Velocity feld (b) Pressure feld
Figure 4 Results for
1
= 0.1,
2
= 1.0 and = 1.3 using
1
= 1.0 and
2
= 10.0.
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 33 #15
MARCIO ANTNIO BORTOLOTI and JOS KARAM FILHO 33
0
5
10
15
20
25
30
1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
I
t
e
r
a
t
i
o
n
s

1
= 1 and
2
= 1

1
= 1 and
2
= 10

1
= 1 and
2
= 100

1
= 1 and
2
= 1000
Figure 5 Number of iterations with Tol 10
6
for 17 17 nodes with biquadratic
elements in the case of various combinations between
1
and
2
.
mesh of 17 17 nodes and 8 8 biquadratic quadrilateral elements has been
used. The numerical results were performed using the following stabilizing
parameters:
1
= 1.0 and
2
= 10.0. For the convergence of the algorithm,
we imposed a tolerence of 10
6
. Numerical results are shown for some com-
binations of the constitutive parameters
1
and
2
. In Figure 1 we can note
the characteristic of the pseudoplastic behavior comparing the velocity profles
on x = 0.5 for the
i
combinations presented. Velocity and pressure felds
are shown in Figures 2-4 for the same
i
combinations of those in Figure 1.
We can see, from these results, how
1
and
2
control the Newtonian and the
pseudoplastic contributions respectively. We note magnitude decreasing in the
velocity and in the pressure felds and also the fatteness of the pressure next to
the two corners (0, 1) and (1, 1) due to the pseudoplastic effect, as expected.
Formulations that use the continuous pressure interpolations may present lack
of accuracy in those regions, where critical boundary conditions exist. Unlike,
discontinuous pressure interpolations, as is the case here, recover the accuracy at
those regions, due to the satisfaction of the incompressibility constraint locally.
The convergence behaviour of the algorithm used, together with the Petrov-
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 34 #16
34 STABILIZED FEM TO PSEUDOPLASTIC FLOW
Galerkin-like formulation PG
hd
, is shown in Figure 5, as a function of the
power index for four Sisko fuids. It can be seen that the greater is the non-
Newtonian effect, the larger is the number of iteractions required to achieve
convergence, as expected for a fxed mesh. Note that even for a higher nonlin-
earity (higher power index ) convergence and stability are achieved.
6 Summary and conclusions
In this work, a consistent stabilized mixed Petrov-Galerkin-like fnite element
formulation in primitive variables, with continuous velocity and discontinuous
pressure interpolations, has been mathematically analyzed for fows governed
by the nonlinear Sisko relation. Stability, convergence and error estimates have
been proven for same order interpolations of the primitive variables for any
combinations when k 2.
To generate the mathematical stability conditions, it was possible to split the
discontinuous pressure. Only the constant by part pressure resulted as respon-
sible to fulfll the LBB condition. The other part, the null mean pressure part,
contributed to achieve the required ellipticity in the Scheurers theorem sense
together with the stabilizing terms. For this formulation ellipticity was the key
for the stability, since the constant part of the pressure fulflls in standard ways
the LBB. It was possible from the ellipticity to provide a suffcient condition to
choose the stabilizing parameters not only as a function of the quasi-newtonian
constitutive parameter but considering both constitutive constants coming from
the Sisko relation.
Numerical results have been presented for the benchmark driven cavity fow
problem to confrm the mathematical analysis.
Fromthe results, stability and convergence have been reached for several com-
binations of the constitutive parameters of the Sisko relation, ranging fromlower
to highly pseudoplastic (low and/or high
2
) effects, although with more in-
teractions in the last case.
The use of discontinuous pressure interpolations ensured accuracy for the
pressure feld even in the regions where discontinuous boundary conditions
are present, since, now, the weakened internal constraint is satisfed locally, in
contrast with continuous approximations.
Comp. Appl. Math., Vol. 31, N. 1, 2012
main 2012/4/9 12:45 page 35 #17
MARCIO ANTNIO BORTOLOTI and JOS KARAM FILHO 35
REFERENCES
[1] R.A. Adams, Sobolev Spaces. Academic Press (1975).
[2] I. Babuka, Error bounds for fnite element method. Numerische Mathematik,
16 (1971), 322333.
[3] R.B. Bird, R.C. Armstrong and O. Hassager, Dynamics of Polimeric Liquids.
Wiley, Vol 1 (1987).
[4] F. Brezzi, On the existence, uniqueness and approximation of saddle-point prob-
lems arising fromLagrange multipliers. Revue Franaise dAutomatique Informa-
tique et Recherche Oprationnelle, Ser. Rouge Anal. Numr., 8 (1974), 129151.
[5] P.G. Ciarlet, The fnite element method for elliptic problems. North-Holland
(1991).
[6] M.M. Cross, Rheology of non-Newtonian fuids: a new fow equation for pseu-
doplastic systems. Journal of Colloid Science, 20 (1965), 417437.
[7] M. Fortin, Old and new fnite elements for incompressible fows. International
Journal for Numerical Methods in Fluids, 1 (1981), 347431.
[8] V. Girault and P.A. Raviart, Finite Element Methods for Navier-Stokes Equations.
Theory and Applications. Springer-Verlag, Berlin (1986).
[9] T.J.R. Hughes, The Finite Element Method: Linear Static and Dynamic Finite
Element Analysis. Dover Publications (2000).
[10] A.F.D. Loula and J.N.C. Guerreiro, Finite element analysis of nonlinear creeping
fow. Computer Methods in Applied Mechanics and Engeneering, 79 (1990),
87109.
[11] J. Karam Filho and A.F.D. Loula, A Non-standard application of Babuka-Brezzi
theory to fnite element analysis of Stokes problem. Computational and Applied
Mathematics, 10 (1991), 243262.
[12] B. Scheurer, Existence et approximation de points selles pour certains prob-
lems non Lineaires. Revue Franaise dAutomatique Informatique et Recherche
Oprationnelle, Ser. Rouge Anal. Numr., 11 (1971), 369400.
[13] I. Wang and J.F. Stoltz, Infuence of non-Newtoninan properties of blood on the
global transport of red blood cells. Clinical Hemorheology, 14 (1994), 789796.
Comp. Appl. Math., Vol. 31, N. 1, 2012