Journalof

Non-Newtonian
Fluid
Mechanics
ELSEVIER J. Non-Newtonian Fluid Mech.. 57 (1995) 119 136

A generalized Giesekus constitutive model with

retardation time and its association to the spurt effect
D i m i t r i s V l a s s o p o u l o s ~"* S a v v a s G . H a t z i k i r i a k o s ~
~' Foundation./br Research & Technoh~gy Hellas, ln.~'titute O~ Electronic Structure & La.wr,
P.O. BOl\ 152Z 71110 Heraklion, ('re to, Greece
~ Department o! (Twmical Engineering, 2216 Main Mall, Univcr.~ity , 4 Britisll ('Muml~ia. l'amoulcr.
British Columbia V6T IZ4, ('umukl

Received [I August 1994

Abstract

This paper presents a simple generalization of the single-mode Giesekus constitutive model
with retardation time. It is based on the utilization of the Gordon Schowalter derivatives l\~r
the stress and rate-of-deformation tensors, which account for the non-afline motion of
polymer strands, and make the model more versatile in terms of its ability to predict the
complex flow behaviour. The model is consistent with molecular kinetic theory, and has been
tested in several well-characterized shear flows, whereas its ability to describe shear-fiec flows
has been proved as well. Finally, a comparison against steady shear data of nearl~
monodisperse polybutadienes demonstrates its superiority with respect to other continuum or
molecular models exhibiting non-monotonic constitutive behaxiour.

Kevwords: Giesekus model; Gordon Schowalter convected derivatives: Polvbutadienes:

Spurt effect

1. Introduction

It is well k n o w n that the usefulness o f a constitutive e q u a t i o n for describing

processing flows o f p o l y m e r solutions and melts rests on its ability to accurately
predict rheological d a t a , as well as on its numerical t r a c t a b i l i t y in several flow
geometries. Such an e q u a t i o n should successfully account for shear d e p e n d e n t
viscosity, n o r m a l stress effects in steady shear flows, d y n a m i c response in u n s t e a d y
shear flows, elastic effects in shear-free flows, a n d n o n - v i s c o m e t r i c flow p h e n o m e n a .

* Corresponding author.

0377-0257/95/$09.50 ~ 1995 - Elsevier Science B.V. All rights reserved

S S D 1 0377-02570(94)01 337-3
120 D. V/assopoulos, S.G. Hatzikiriakos / J. Non-Newtonian Fluid Mech. 57 (1995) 119 136

To date, no available model is capable of accurately describing all of the above

phenomena [1,2]. It is, therefore, a matter of best judgment to choose a rheological
model that is most suitable for a particular application; this is especially true for
non-viscometric flows. It is common practice that any model chosen for studying
such a complex flow should produce at least qualitatively correct response mainly
in viscometric flows, as well as some well-characterized complex ones. But as
Tiefenbruck [3] has pointed out, the only motivation for this practice is the fact that
the above flows are heavily documented; this however, may not always be jus-
tifiable. For example, whereas shear flows are weak [4,5], many non-viscometric
flows of practical interest are strong. In such a case, a shear flow does not serve as
a good test for a constitutive equation. Therefore, it is important to find out which
material functions are most closely related to the flow problem in question, and
check how well the model describes these properties. Nevertheless, even this
approach involves some degree of ambiguity, since this assessment is often based on
experience [6,7].
The Giesekus model [8 10] has been a popular choice, because of its relative
success in several flows, and also because it reduces to several well-known simpler
models, useful in a variety of flow situations [6]. It was originally developed for
polymer melts and concentrated solutions, but has also been applied to dilute
solutions; it is based on a kinetic theory analysis of closely packed polymer chains.
The key characteristic of this model is that it includes non-linear terms in stress. Its
form is the following:
21
Tp-f-2~ I To<~>+ ~ - - Tp" T p = 2qpD, (la)
qp
T = T ~ + T v and Ts=2qsD. (lb)
T is the extra-stress tensor; D is the rate-of-deformation tensor; q is the viscosity;
2t is the relaxation time of the polymer; the subscripts p and s refer to polymer and
solvent, respectively; c~ is the mobility factor, whose origin is associated with
anisotropic Brownian motion and/or anisotropic hydrodynamic drag of the poly-
mer molecules [8]. Finally, To<l> is the upper convected derivative of the stress
tensor defined as
DT
r,<~>---- Vv ~- Tp- Tp" Vv, (2)
Dt
where D T / D t is the substantive time derivative, and Vv is the velocity gradient
tensor.
From the above original Giesekus equation, one can obtain the extra stress for a
Giesekus model with a retardation time [6,11],

T + Z t T < I > + A 2-LT. T - 2 A 2 2 ( T . D + D . T)

qo
z5
=20o D + 2 2 D < j > - 2 A - r - D "D , (3a)
Al
 D. l'lassopouh>~, S.G. Hat=ikirMkos / J. Non-Nelctonian Huid Mech. 57 (1995) 1/9 136 121

where

',o = ;,~ + 'lp, ,i~ = ;t,(rl~ t A - Y" (~b)

" \qol' 1 - (2,,;il) "

A is known as the modified mobility parameter, and 2, is the retardation time of the
polymer and T~> and D<~> are the upper convected derivatives of the stress and
rate-of-deformation tensors defined previously. This equation is like the Oldroyd B
model, except for the modified mobility terms. The usefulness of the Giesekus
model with 2~ can be seen by the fact that it reduces to some well-known models
for particular values of the three parameters "~1" "~2' and :~ [11 13]: Newtonian,
Second Order Fluid, Upper Convected Maxwell, Oldroyd B, Leonov-like, Leono,'-
like with retardation time, and Corotational Maxwell-like. The material functions
can be found in Ref. [6]; they are in general more realistic then the above mentioned
special cases or the Oldroyd 8-constant model, due to the inclusion of ll;c
non-linear stress terms and the retardation terms. The model can account for a
large decrease in viscosity and normal stress coefficients with shear rate, non-zero
second normal stress coefficient, a bounded elongational viscosity which reaches a
constant value at large strain rates, and a fairly diverse behaviour in transient flow,s.
including viscosity and normal stress overshoot [9]. Some limitations of this model
(sensitivity to flow strength, not to alignment strength) are discussed by Larson [1 ].
The purpose of the present contribution is to generalize the Giesekus model in
order to expand the range of its applicability, presenting therefore an improved and
more versatile constitutive formulation, while at the same time making it physically
more robust. In the next sections, the generalized Giesekus model is presented.
Emphasis is placed on its performance in shear flows, which is tested and ewduated.
The shear-free flow behaviour is briefly discussed. Finally. in order to demonstratc
its versatility, this general model is used to describe the steady shear beha~iour c4"
monodisperse polymer melts, and compared against the experiments of Lira and
Schowalter [14] with polybutadienes extruded through slit capillary dies.

2. Generalized Giesekus model with retardation time (GGM)

The Giesekus model with retardation time has been generalized by altering the
convected derivatives for the stress and rate-of-deformation tensors. Instead ol'
using the upper convected derivative, we have implemented the more general
G o r d o n - S c h o w a l t e r derivative (GS), defined for a second order tensor T as [15]

,~ T
T,,~s, = ~ - ~ - ~(D • T + T . D), 14)

where ~/~t is the Jaumann derivative defined as

~T DT
= (o ~ • T - T.~o, (5)
~t Dt
122 D. Vlassopoulos. S.G. HatzikiHakos / J. Non-Newtonian Fluid Mech. 57(1995) 119 136

where ~o is h a l f o f the vorticity tensor, a n d 4 is an empirical p a r a m e t e r v a r y i n g

from 0 to 1 (actually { lies between - 1 a n d 1, but cases ~ < 0 are o f no interest
here). F o r ~ = 0, the J a u m a n n derivative is recovered, while for ~ = 1, T~c;s~ reduces
to the u p p e r c o n v e c t e d derivative o f T.
T h e r a t i o n a l e for using the G S derivative is that it represents a non-attine m o t i o n
t h r o u g h the n o t i o n o f " s l i p " , which characterizes the relative m o t i o n o f a s t r a n d
e n d - t o - e n d vector (in p o l y m e r systems) with respect to the m o t i o n ( m a c r o s c o p i c
d e f o r m a t i o n ) o f the c o n t i n u u m ; this leads the s t r a n d to t r a n s m i t only a fraction o f
its tension to the c o n t i n u u m . Therefore, with the G S non-aftine f o r m u l a t i o n , the
e q u a t i o n for the stress t e n s o r is reduced from that for an affine m o t i o n by a factor
o f ~ (for ~ = 1, the m o t i o n is affine). T h e G o r d o n - S c h o w a l t e r derivative is actually
o b t a i n e d if the basis vectors o f a c o n t i n u o u s l y d e f o r m i n g c o o r d i n a t e system slip
with respect to the d e f o r m i n g m a t e r i a l lines. It has been a p p l i e d to dilute p o l y m e r
solutions [15], as well as p o l y m e r melts [I 6]. In a d d i t i o n to the physical significance,
there are also some p r a c t i c a l reasons, which underline the usefulness o f such a
g e n e r a l i z a t i o n o f the G i e s e k u s model. T h e r e is a p l e t h o r a o f w e l l - k n o w n m o d e l s (as
seen in T a b l e 1 below), which are special cases o f the G G M , a n d this m a k e s this
m o d e l m o r e flexible a n d attractive. M o r e o v e r , the a d d i t i o n a l p a r a m e t e r , ~, m a y
help in o b t a i n i n g a better q u a n t i t a t i v e a c c o u n t o f the e x p e r i m e n t a l d a t a in various
a p p l i c a t i o n s o f engineering interest, and thus the m o d e l m a y p r o v e m o r e successful
in that respect t h a n the original G i e s e k u s one.
T h e constitutive e q u a t i o n for the extra stress is given by

T+21T(~s)+A 2-2 T . T - 2A).2( T . D + D • T)

rio

=2qo D+z2D~(}s)-2A--:D'D . (6)

2L

Table 1
Some special cases of the generalized Giesekus model with retardation time

21 ),~ a Model

O,I 0,1 0 Newtonian

>0 0 0 UCM
>0 >0 0 OB
0 <0 Second Order Fluid
>0 0 1/2 Leonov-like
>0 >0 > 1/2 Leonov-like with retardation
>0 0 I Corotational Maxwell-like
>0 >0 >0 Giesekus with retardation
>0 0 >0 Original Giesekus
>0 -- 0 Oldroyd 4 constants
0 >0 0 0 Maxwell Jaumann derivatives
0 >0 >0 0 Oldroyd-Jaumann derivatives
>0 >0 0 Johnson-Segalman
 D. I7axsopouh~s, S.G. Hatzikiriako~ J. Non-Newmnian Fluid Mech. 57 (19951 119 136 123

' ' ' ' ' ' ' H ' ' ' . . . . . I ' ' ' . . . . . I ' ' ' ' ' ' '

°0"1 ~

0.01
ex=O.1
X@/XI=O.O01
0.001 ........ , ........ i . . . . . . . . I . . . . . . .

0.1 1 10 100 1000

5'x~
Fig. 1. Steady shear viscosity f o r the G G M (z- 0,1, 2~ 2- - 0.001). F r o m 1 to 5, ~ = 0, 0.75, 0.gfl. 0.95,
and 1.

It should be emphasized that this constitutive equation is consistent with the

general molecular kinetic theory of polymeric liquids with anisotropic effects (as is
the Giesekus model) as presented by Bird et al. [17]. One may start with their Eq.
(13.7-4) which describes the change for (Q_Q), and then use the anisolropic tensors.
their Eq. (13.7-6) and (13.7-7), to describe non-Maxwellian velocity distributions
and anisotropic frictional effects. Finally, defining an "effective" velocity gradient
as

,( = ,, - ½(1 - ( ) D , (7)
where x = Vv, and including a retardation time. one can arrive at the above
generalized Giesekus equation with redefinitions of terms.
Special cases of the G G M are presented in Table I. Of particular interest is the
Oldroyd 4-constant model, which has been used to study the motion of particles in
viscoelastic liquids [3]; it was found that the best agreement between theory and
experiments was obtained for a wdue of ( between 0 and 1. Similar conclusions
were drawn from other studies involving polymer solutions, and suspensions as well
[18 201.

3. Results

3. 1. Application to shear flows

Particular attention has been paid to the performance of the G G M in a variety

of shear deformation histories. Figs. 1 and 2 depict the model's response in plane
124 D. Vlassopoulos, S.G. Hatzikiriakos / J. Non-Newtonian Fluid Mech. 57 (1995) 119 136

steady Couette flow (simple shear). Each one of these figures represents five cases,
depending on the value of 2; from 1 to 5, they are, respectively: ~ = 0 (the
corotational limit), ~ = 0.75, ~ = 0.90, 4 =0.95 and ~ = 1 (the codeformational
limit, which is the original Giesekus model). The parameters involved are dimen-
sionless, scaled as follows: the shear viscosity with qo, the shear rate with 1/),1, the
time with 21 , the first normal stress coefficient with 2 % ( 2 1 - ,i2), and the second
normal stress coefficient with a q o ( 2 1 - 2~). The mobility factor was ~ = 0.1, while
the ratio 22/2 ~ was fixed at 0.001. It is clear that the limiting behaviour of the
original Giesekus model, with a plateau viscosity at high shear rates, is reached for
= 1 (Fig. 1). The effect of the GS formulation can also be appreciated, where a
clear deviation from the codeformational limit occurs at quite high values of ~,
accompanied by a corresponding change of slope in the power-law shear thinning
region. Moreover, for values of ~ < 1, a high-shear rate plateau viscosity limit (~/~j)
is reached. Consequently, the versatility of the G G M , and thus its ability to
accurately describe a big variety of shear viscosity data, is demonstrated. Similar
remarks hold for the first and second normal stress coefficients. In the former case,
which is depicted in Fig. 2, the five curves are distinctly different at high shear rates,
and collapse at low shear rates, in agreement with the shear viscosity behaviour; in
the latter case however, the opposite behaviour is observed (the curves are different
at low shear rates and nearly collapse at high shear rates).
Representative stress growth results are depicted in Fig. 3 (shear viscosity growth
function, q+) for a = 0.3, and 22/21 = 0.001, ;:'2, = 3, and three limiting values of ~,
namely ~ = 0, 0.75, and 1. They suggest that as we deviate from codeformational
behaviour, an undershoot develops at times t/21 > 1.5; in the limiting corotational
case, a second oscillation is observed, with a clear maximum occurring at t/21 = 2.5.

0 0 . . . . . . . . I . . . . . . . . I . . . . . . . . I . . . . . .
1

10 -1

,<I 1 0-2
,<~ 0 -3 , 5
"%1
0-4 1

10 . 5
i0_ 6 Xe/X1 =0"001 "~
, , , , , , , , I . . . . . . . ,I L = i ,,,,,I , , , , , , , "

O. I I I0 I O0 1000
ix 1
Fig. 2. First normal stress coefficient in steady shear for the G G M (:~=0.1, 2~/22-0.001 ). From 1 to
5, ~ = 0, 0.75, 0.90, 0.95, and I.
 D. Vlassopoulos. S.G. Hatzikiriakos ; J. Non-Newtonialt Fluid Mech. 57 (1995) II9 136 I25

0.5
.... ' .... . =. . . 0 . 3 ' X "6 . boi

0.4

o 0.3 ~=1

g" 0 . 2 ~ ~--=n.75
% ~

0.0 ,, i . . . . . . f,,
0 1 2 3 4 5
t/x
Fig. 3. Stress growth in steady shear for the G G M ( ~ - 0.3, /i ;;= 3, / ~ 2 ~ = 0.001 t. Results ~\~r shear
viscosity.

Moreover, the overshoot of q+ becomes less significant as ~ increases. Similar

conclusions are drawn from the calculation of the first and second normal stress
coefficient growth functions, hv(, and Wj-, respectively. However, it should be
pointed out that the presence of more than one oscillation in the material functions
has not been observed experimentally; instead, an undershoot following the initial
maximum is usually observed. Therefore, the cases ~ = 0 and ~ = 1 are considered
not realistic. This of course justifies the use of the G o r d o n Schowalter derivative in
the Giesekus model, and proves the effectiveness of the G G M . A further remark
deals with the observed maxima. Results for q~(, and ~u;+ indicate that they occur
at times twice the time of the q+ maximum (t/;.l is about 0.6). This is in agreemenl
with experimental observations reported in the literature.
Fig. 4 represents typical relaxation predictions in steady shear for the same values
of" parameters as in Fig. 3. It is clear that for the shear viscosity relaxation function,
q the same general observations made before, hold. Moreover, results for the first
( ~ l ) and second (u/~) normal stress relaxation functions follow the same trends as
well. It is noted however, that the latter two functions are monotonic, in contrast
to their stress growth analogues. A final note concerning the (I~1) results is m
order: at short times t/21 < 2, the predictions for ( = 0.75 are close to the corota-
tional ones, while at longer times t/2~ > 2 they become closer to the codeforma-
tional ones.
It is evident from the results discussed here that the model is able to predict all
of the known phenomena observed in steady and transient shear flows of viscoelas-
tic fluids. More importantly though, by varying the parameter ~ it is possible lo
predict behaviours ranging from convected derivative to corotational derivative.
126 D. Vlassopoulos, S.G. Hatzikiriakos / J. Non-Newtonian Fluid Mech. 57 (1995) 119 136

3.2. Behaviour in shear-free f l o w s

Since the main focus of the paper is on shear flows, detailed results for the
shear-free material functions will not be presented here. However, it is still
important as a matter of completeness to show that the G G M provides an
additional flexibility for predicting reasonable elongational properties over the
original Giesekus model with retardation time, due to the inclusion of the additional
parameter ~. Provided that the parameter c~ is not zero, the elongational viscosity
is bounded and reaches a constant value at large strain rates. Bird et al. [6] present
several shear-free material functions for the Giesekus model for selected values of
the parameter ~, and thus it is anticipated that the G G M will provide an even better
representation of such properties. Here, we only present one characteristic example
of the G G M performance (with ~ = 0 . 1 and 22/2j=0.001) in steady planar
elongational flows. Fig. 5 depicts the first viscosity function vs. the elongation rate
& for values of 4 = 0.2 (curve 1), 0.3, 0.5, 0.75, and 1 (curve 5). For the limiting
codeformational case, the original Giesekus results are reproduced [6]. It is noted
that as ¢ decreases, for the same ~ the "first viscosity function gap" disappears,
while the first viscosity values approach its minimum value at ¢ = 1. The depen-
dence of the second viscosity function on the elongation rate, for the same values
of 2, has also been investigated. The effect of ~ is similar to the first viscosity
function case, i.e. it tends to flatten the "viscosity gap"; here, with increasing 4 the
second viscosity function decreases for ~2~, while it increases for ~).] > 0. Finally,
for ~ = 1, the original Giesekus results are recovered. It should be mentioned that
the original multimode Giesekus model has been found capable of accurately fitting
shear-free material properties for a linear low-density polyethylene [6].

' 1 I I ' I I ' I ' 1 I t I ' ~ ' ' I ' ~ ' '

:o.oo

i ~=0.75
o.o,

0.001 i i ~ i

0 1 2 3 4 5
t/X I
Fig. 4. Stress relaxation in simple shear lbr the GGM (~ - 0.3, -~1;; = 3, 22,'2~- 0.001). Results for shear
viscosity.
 D. Vlussopoulo.s, S.G. Hat--ikiriakos .I. Non-,Vcwmniau FluiJ lh,ch. 57 (10951 II ~) I ;0 t27

100 ......... , ......... , ......... , .........

~x=0.1
ke/Xl=0.001

5
f D -- -

Ig" ,
-__..q/ - . . . . . .

-10 -5 0 5 10
~x
1

Fig. 5. First viscosity function ill steady exlmlsiona] flo,.,, (phmar dongation, /~-- I) Ior the G G M
(~'-0.1). From I to 5 j = O . 0.75, 0.9, 0.95, a n d I.

3.3. Comparison with experhJwnts on monodLv,erse too/tell polymers

It has been known for some time now that linear polymers exhibit cyclic or
oscillating melt fracture during flow through conduits, such as capillary and slit
dies. In other words, for wall shear stress greater than a critical value and apparent
shear rate within a certain range, the flow becomes unstable. In a cons:ant piston
speed rheometer, this instability manifests itself as oscillations in both the pressure
drop and mass flow rate [21 23]. The flow curves (wall shear stress vs. apparem
shear rate) for polymers associated with such phenomena exhibit a maximum and
a minimum shear stress, that is the shear stress is a multivalued function of the
shear rate.
This phenomenon (oscillating melt fracture) was recently studied m detail [br "l
series of high density polyethylenes (HDPE) by Hatzikiriakos and D e a d [22] using
a capillary rheometer. These authors demonstrated that the oscillating l]ow (spurt
phenomenon) is due to a combination of wall slip and compressibility of the molten
polymer. It should be noted that by employing a multivalued (non-monotonic)
constitutive equation exhibiting a maximum and a mminlum in the wall shear
stress shear rate curve and without using explicitly a slip boundary condition it is
not possible to describe the data reported by these authors. For example, the flow
curves obtained from a capillary rheometer are diameter-dependent in the pro-oscil-
lating rio,a, region (sharkskin region), an observation consistent with the assumption
of wall slip.
Lim and Schowalter [14] studied the shear behaviour of a series of nearly
monodisperse polybutadienes by using slip capillary dies. They found that the
obtained flow curves in the pre-oscillating flow region were independent of slit
128 D. Vlassopoulos, S.G. Hatzikiriakos / J. Non-Newtonian Fluid Mech. 57 (1995) 119 136

height. In addition, no change of slope was observed, which is another indication

of the absence of slip. Therefore, their experiments implied the absence of slip for
polybutadienes in the pre-oscillating flow regime. Although the authors offered as
possible explanation their inability to obtain sufficiently small increments in flow
rate, the matter is still not fully conclusive; further, for such cases an alternative
analysis other than the one presented by Hatzikiriakos and Dealy [22] may be
possible.
Predicted velocity profiles in capillary flow under constant stress mode (the other
case is constant flow rate mode) by constitutive equations exhibiting such a
"double-valued" non-monotonic behaviour in the high flow rate branch of the flow
curve indicate an apparent slip velocity [24-26] over an extremely thin layer of the
fluid close to the wall. This apparent slip may practically be considered as wall slip.
Thus, a possible physical interpretation of the spurt phenomenon may be provided
through "constitutive instabilities", i.e. by using a constitutive equation exhibiting
maxima and minima without invoking explicitly a slip boundary condition. How-
ever, this analysis may only be used for molten polymers exhibiting no-slip in the
pre-oscillating flow regime. This is indeed the case of the nearly monodisperse
polybutadienes, studied by Lim and Schowalter [14]. It is clear however, that a
good agreement between theory (constitutive behaviour) and experiments cannot
provide a full explanation of this spurt effect in the case of monodisperse polybuta-
dienes, without further experiments and extensive analysis. Moreover, it should be
emphasized that for molten (polydisperse) polymers which exhibit slip in their low
flow rate branch of the flow curve (sharkskin region), the analysis presented in this
section is inadequate, and an explicit slip boundary condition should be imposed as
an additional requirement.
The predictions of the G G M in simple shear were found to be in close agreement
with the experimental observations of Lira and Schowalter [14]. Indeed, Figs. 6-8,
represent dimensionless (the shear stress is scaled with qo/2~), typically non-
monotonic shear stress-shear rate flow curves predicted by this model for various
values of the parameters ~, 22/21 and ~; they show the relative importance of each
of those parameters on the shape of the flow curve. It can be observed that a wide
range of responses is obtained which range from monotonic flow curves (absence of
cyclic melt fracture), to flow curves exhibiting two extreme shear stress values. In
particular, it can be seen from Fig. 6 that for ~ = 0.1 and .~2/,~1 = 0.001, a clear
non-monotonic flow behaviour is observed, with a maximum (o-..... ) and then a
minimum (am~,,) of shear stress, when ~ = 0. However, with increasing the value of
(intermediate values from 0.25 to 0.99) the difference between a ..... and O-m~r,
decreases, while the shear rate range over which the spurt effect is obtained
increases. At .~ = 1 the shear stress curve is monotonic, with no a ...... and am,l; a
"nearly perfect" plateau can be reached for a value of f = 0.999. The effect of ~ in
obtaining a non-monotonic flow curve is thus significant. In Fig. 7, for :~ = 0.1 and
d = 0.85, it can be observed that an increase of the value of 2~/,i~ (from 0.001 to
0.3) decreases the difference between a .... and a,T,~n, but also decreases the shear
rate range over which the spurt effect is observed. For ,~2/,~1 ---0.3, the flow curve
becomes monotonic, but only a short plateau region can be reached by varying this
 D. Vlassopou/os, S.G. Hat_-ikiriako.s' J, Non-Newmnian Fluid Mech. 57 (1995) 119 13C~ 1",!9

1 0 -, ....... , . . . . . , ........ , ........ , .......

o(=O.1
10
x /x,=o.ool

2 1

tb
0.1

0.01 ........ ' ....... ' ........ [ ........ ~ . . . . .

0.0 0.1 1 10 100 1000

o

7X]

Fig. 6. Shear stress shear rate flow curve as predicted b} the G G M , for :~ = 0 . 1 . and /,l ,~2= 0 . 0 0 l .
F r o m curve 1 to c a r v e 10. the values of ,;- are: 0, 0.25. 0.5, 0.75, 0.85, 0.90. 0.95, 0,98, 0.99. and I.

parameter. Similar, but less pronounced, is the effect o f 7. on the shape o f the flou.
curve. This is exhibited in Fig. 8, for 5 = 0.1 and ,'{2/2~ = 0.001. As ~ varies from 0.9
to 0.1, the difference ~r..... - a m , , , decreases and the relevant "effective" shear rate
range increases, but only slightly. In all cases, however, shear stress behaviour
exhibiting m a x i m u m and minimum is clearly observed.

10 ........ , , ,~, ........ , ,/,,,/ ~

o I

b
0.1

0 00 1 ........ l .... ~ ......... l ......... [ ........

0.01 0.1 I 10 100 1000

Fig. 7. Shear stress -shear rate flow curve as predicted by the G G M , for ~ = 0.1. and ~ = 0,85. From
curve l to curve 8, the wdues of )q :,i 2 are: 0.001, 0.003, 0.005. 0.01. 0.03. 0.05, 0.1, and ( .3.
130 D. Vlassopouh)s, S.G. Hatzikiriakos / J. Non-Newtonian Fluid Mech. 57 (1995) 119 136

] 0 ........ I ........ I ........ I ........ I ..... ~'q

~--o.85
Xa/X1=O.O0 1

6
° I

,,..-i

0.1

0.01 ........ ~ ........ ' ........ ' ........ ' .......

0.01 0.1 1 10 1 O0 1000

F i g . 8. S h e a r s t r e s s shear rate flow curve as predicted by the GGM, f o r ~ = 0.85. a n d 22,'2 t -0.001.
From c u r v e 1 to c u r v e 6. t h e v a l u e s o f ~ a r e : 0.9, 0.8, 0.7, 0.5, 0,3, a n d 0.1.

In summary, a typical non-monotonic shear stress flow curve can be divided into
three regions (this is also discussed by McLeish and Ball [27] and Cares et al. [28],
with a nice link to the wall slip): region A, which corresponds to low shear rates,
up to the characteristic shear rate that corresponds to a ...... ; region B covers the
range of shear rates from a ...... up to a,,,, ; and region C, at high shear rates. Region
A is controlled by qo and 21, region B is governed by ~ and (, and region C is
mainly affected by 2~. It should be pointed out that the non-monotonic behaviour
exhibited in the G G M rheograms presented in Figs. 6-8, is in qualitative agreement
with the molecular approaches of McLeish and Ball [27] and Cares et al. [28] who
have extended the D o i - E d w a r d s theory, and the continuum approaches of Kolkka
et al. [25], who used a Johnson-Segahnan fluid, Kolkka and lerley [29], who used
a Giesekus fluid, and Deiber and Schowalter [26] who used a linear superposition
of corotational and codeformational Maxwell models. It is also worth mentioning
that Kolkka and Ierley [29] have presented a phase space analysis of the spurt
phenomenon for a Giesekus fluid. Moreover, the works of McLeish and Ball [27],
and Kolkka et al. [25] were found to be consistent with experimental data reported
by Vinogradov et al. [30] on the spurt behaviour of polybutadienes (used in Ref.
[27]) and polyisoprenes (used in Ref. [25]), when examined in a constant stress
mode, while that of Deiber and Schowalter [26] compared nicely against the data of
Lim and Schowalter [14] on monodisperse polybutadienes.
Since only the single-mode G G M is presented here, for a fair quantitative
comparison with experimental results, steady shear flow data for nearly
monodisperse polymers are needed. Such data are available in the literature. They
were obtained by Lim and Schowalter [14], who reported wall slip experiments in
 D. Vlassopoulo.s, S.G. Hatzikiriakos ,I. Non-NewtoniaH Fluid Mech. 57 (19951 119 130 I ~1

slit capillary dies with nearly monodisperse polybutadienes (M,,/M,, ~ 1.05t. Tile
data by Vinogradov et al. [30] on nearly monodisperse polyisoprenes could also be
used; however an excellent fitting was obtained by Kolkka et al. [25] using the
Johnson Segalman model which is a partial case of the G G M . Thus the G G M w'as
only compared against data of Lim and Schowalter [14] for various naolecular
weights, as discussed below. Four different molecular ueights were tested, namely
M,, = 78 000. 99 000, 150 000, and 240 000. The high molecular weight data show a
maximum shear stress at a critical value of the shear rate (spurt effect). For the low
molecular weights (M,, = 78 000 and 99 000) however, the experimental data do not
show such a stick slip flow transition effect: this is attributed to the fact that the
relatively high shear rates necessary for this effect were not reached experimentally
[14]. For the application of the G G M , the values of % and 2~ obtained by Lira and
Schowalter [14] were used; the 2, value was assigned to the longest relaxation time
(of the order of the reptation time). The retardation time ./, here corresponds to the
Rouse rehixation time of a chain of random conformations; it is expressed b,~
i~-~ 2~ M~.3M,, [31] where M~.=6025 is the entanglement molecular weight for
polybutadiene [32]. Therefore. for the four polybutadienes tested. M,, = 78 00(1.
99 000. 150 000, and 240 000, the ratio 22)., was taken as 0.026. 0.020. 0.013. and
0.008. respectively. The best fit G G M curves were obtained with :~ = 0.80 and
.,- = 0.85. it is apparent that this set of parameters produces a very good litting of
the experimental results (Figs. 9 12).
Because of the molecular nature of the parameters inxolved, we believe that our
approach produces a physically meaningful and fairer fitting than a trial and error
approach (which would have undoubtedly produced a better fitting). It is important
to emphasize that the data of Lira and Schowalter appearing on the plateau regions

o.1 ,A//

b
0.01

~/ Mw=78000

/
0.001 ........ I ........ I
0.1 1 10 O0

"7' ( - 1 )
Fig. 9. Comparison of G G M predictions with the polybutadiene wall slip experimental rcsuhs of Lira
and Schowalter [14], for ?v/~ = 78 000. in slit extrusion dies (triangles, 0.21 cm slit: circles, 0.39 cm slit)
132 D. Vlassopoulos, S.G. Hatzikiriakos /J. Non-Newtonian Fluid Mech. 57 (1995) 119-136

I ' ' . . . . ''I ' ' ' . . . . . I ' ' ' ' ' ' ' '

..~ 0.1 t I

b
0.01

IV[W = 9 9 0 0 0

0.001 ........ L ........ l .......

0.1 I 10 100
(s-')
Fig. I0. Comparison of G G M predictions with the polybutadiene wall slip experimental results of Lirn
and Schowa]ter [14], for M~,.= 99 000; same symbols as in Fig. 9.

in these figures represent some average values o f the pressure drop, which oscillates
between two extreme values during the flow o f the polybutadiene melts through the
capillary dies; these extremes are nicely represented by the two extreme shear stress
values o f the G G M flow curve. A t high molecular weights (Figs. ] I and 12) the

1 , . . . . . . . i . . . . . . . . i . . . . . . . . i . . . . . .

,-. 0.1

b
0.01

0.001 ........ i ........ , ........ , .......

0.01 0.1 I 10 O0

Fig. 1 l. Comparison of G G M predictions with the polybutadiene wall slip experimental results of Lira
and Schowalter [14] for M w = 150000; same symbols as in Fig. 9.
 D. lTassopoulos, S.G. Hatzikiriakos J. Non-Nen'mnian Fluid ,~[('(ih. 57 (1995~ 119 /7¢'~ 133

1 . . . . . . i i] 1 - i i ] i , [ , I i i T . . . . . ] . . . . . .

0.01 / • •
/
M =240000

0.001 . . . . . . . I . . . . . . . . . I ........ I .......

0.01 0.1 1 10 100

9 (~-~)
Fig. 12. Comparison of G G M predictions w'ilh tile polybutadiene wall slip experimental results ot Linl
and Schowaher [14] for M,, = 240 000: same symbols as in Fig. 9.

transition between the two governing relaxation times (reptation and Rouse) is
clearly observed with data departing from the corotational limit and proceeding
towards the codeformational one. This transition was not observed with the low
molecular weight samples (Figs. 9 and 10), probably because the shear rates
reached were not high enough [14]. More importantly however, this fitting seems
superior to the previously reported comparison of the same data with a linear
superposition of corotational and codeformational Maxwell models (as it has been
already demonstrated in the previous sections); the key message, nevertheles,;
remains unchanged, i.e. neither of these two limiting cases is realistic when
associating constitutive behaviour with the spurt effect. It is clear that the incorpo-
ration of the two distinct relaxation mechanisms represents an important ingredienl
for the correct physical interpretation of the experimental data [27]. A special notc
with respect to the "best-fitting" choice of :~ and ( is in order. It is clear that these
parameters are molecular in origin, and thus should be scaled with molecular
characteristics of the four polybutadienes used. However, we think it is better to
report the fitting results here with their values constant, since otherwise our fitting
problem would reduce to one of fitting four sets of data with four adjustable
parameters, which is rather ambiguous. Therefore, since we do not know the
dependence of :~ and ~ on molecular weight, we used fixed values for both. in
addition, it should be mentioned that calculations suggest that ,;.2/.~ is a more
sensitive parameter than the others, ~ being the least sensitive; thus, the choice not
to vary :~ and ~ which are the least understood at the molecular level, is justified.
Finally, the issue of the plateau exhibited by the shear stress above a critical
shear rate will be briefly addressed. The experimental data of Lim and Schowalter
[14] represent average values for the stress and apparent shear rate in the cyclic melt
134 D. Vlassopoulos, S.G. Hatzikiriakos / J. Non-Newtonian Fluid Mech. 57 (1995) 119-136

fracture regime. In fact, their dynamic data show that the pressure (and therefore
the shear stress) oscillates between two extreme values. However, the flow rate
also oscillates between two extreme values under such conditions. Thus, under
these conditions one obtains a hysteresis in the shear stress-apparent shear rate
curve, as demonstrated by Bagley et ai. [21] and Hatzikiriakos and Dealy [22].
The G G M predictions also exhibit these characteristics, as seen in Figs. 9-12.
However, it is stressed that the association between constitutive behaviour and
experiments constitutes a systematic observation rather than an explanation. Di-
rect numerical shear flow computations have to be performed to demonstrate
that the G G M will exhibit pressure gradient oscillations beyond the shear stress
maximum in the constant flow rate mode. It is noted, however, that sustained
pressure gradient oscillations were obtained in the constant flow rate mode by
Malkus et al. [33] who used a Johnson Segalman fluid which is a partial case of
the G G M model.
A final remark on the model is related to the values that the mobility parame-
ter, A, may take, in order to obtain admissible and stable solutions. Schleiniger
and Weinacht [34] performed such an analysis for the Giesekus model and found
that for :~ < 1/2 solution always exist, however for 1/2 < ~ < 1 there is a range of
shear rates (usually large) in which a solution either does not exist or is not
realizable (admissible and stable). One could perform a linear stability analysis to
determine the restrictions applied to the values of A for the G G M [34] although
in our calculations such a problem was not noticed at least for ~ = 0.85 and
dimensionless rates up to 10 3 (Fig. 8).

4. Conclusions

A simple generalization of the Giesekus model with retardation time has been
presented in this paper. The modified model is consistent with molecular kinetic
theory. It exhibits distinct differences from the original Giesekus model in the
prediction of shear and shear-free flows, and is much more versatile. Further-
more, it is quite successful in describing the steady-state shear behaviour of
monodisperse molten polymers, as confirmed from the comparison against rele-
vant experiments with polybutadienes. The molecular parameters 2j, 22, c~ and
allow prediction of the shear stress plateau for four molecular weights. It is
worth noting that this model does a superior job in that respect, compared to
other continuum and molecular models proposed in the literature to fit the same
sets of experimental data. More importantly though, it represents further evi-
dence of the physical importance of incorporating two distinct relaxation mecha-
nisms in order to associate constitutive instabilities with the spurt effect.
Although it is proposed that this model may successfully describe more complex
non-viscometric behaviour, it is premature to anticipate its success at this stage.
Further analysis and extended experiments, both in well-defined shear-free and
non-viscometric flows, are needed.
 D. I/lassopoulos, S.G. Hatzikiriakos ,,J. Non-Newtonian FluM Mech. 57(1995) 119 136 15;5

Acknowledgements

This work was partially supported by a NATO collaborative Grant No. CRG
940064. The constructive criticisms of the manuscript by all reviewers is also
appreciated.

References

[1] R.G. Larson, Constitutive Equations for Polymer Melts and Solutions, Butterworths, Boston, 198.~.
[2] D.L. Davidson, W.W. Graessley and W.R. Schowalter, Velocit3 and stress profiles of polymeric
liquids flowing in a periodically constricted channel. Part 2. Observations of non-Newtonian
behavior, J. Non-Newtonian Fluid Mech., 49 (1993) 345 375.
[3] G.F. Tiefenbruck, The translational motion of particles in a viscoelastic liquid, Ph.D. Dissertation,
California Institute of Technology, Pasadena, CA, 1980.
[4] R.I. Tanner, A test particle approach to flow classification for viscoelastic fluids, AIChE J, 22
(1976) 910 924.
[5] L.G. Leal, Studies of flow-induced conformation changes in dilute polymer solutions, in Y. Rabm
(Ed.) Polymer Flow Interaction, AIP Conference Proceedings No. 137, AlP, New York, NY, 1985.
[6] R.B. Bird, R.C. Armstrong and O. Hassager, Dynamics of Polymeric Liquids, Vol. I, Fluid
Mechanics, Wiley-lnterscience, New York, 1987.
[7] W.R. Schowalter, Mechanics of Non-Newtonian Fluids, Pergamon Press, Oxford. 1978.
[8] H. Giesekus, A simple constitutive equation for polymer fluids based on the concept of deforma-
tion-dependent tensorial mobility, J. Non-Newtonian Fluid Mech., I I (1982) 69 109.
[9] H. Giesekus, Stressing behaviour in simple shear flow as predicted by a new constitutive model fi,r
polymer fluids, J. Non-Newtonian Fluid Mech., 12 (1983) 367 374.
[10] lq. Giesekus, A comparison of molecular and network-constitutive theories ~br polymer fluids, in
A.S. Lodge, M. Renardy and J.A. Nohel (Eds.), Viscoelasticity and Rheology, Academic Press,
New York, 1985.
[1 I] R.C. Armstrong, R.A. Brown, A.N. Beris, J.V. Lawler and S.J. Muller, Evaluation of constitutive
equations: material functions and complex flows of viscoelastic fluids, in A.S. Lodge, M. Renardy
and J.A. Nohel (Eds.), Viscoelasticity and Rheology, Academic Press, New York, 1985.
[12] A.N. Beris, Calculations of viscoelastic flow in a journal bearing, Ph.D. Dissertation, M assachuset:s
Institute of Technology, Cambridge, MA, 1985.
[13] K.P. Jackson, K. Walters and R.K. Williams, A rheometrical study of Boger fluids, J. Non-Newto-
nian Fluid Mech., 14 (1984) 173 188.
[14] F.J. Lim and W.R. Schowalter, Wall slip of narrow molecular weight distribution polybutadienes,
J. Rheol., 33 (1989) 1359 1382.
[15] R.J. Gordon and W.R. Schowalter, Anisotropic fluid theory: a different approach to the dumbbell
theory of dilute polymer solutions, Trans. Soc. Rheol., 16 (1972~ 79 97.
[16] M.W. Johnson and D. Segalman, A model for viscoelastic fluid behavior which allows non-affine
deformation, J. Non-Newtonian Fluid Mech., 2 (1977) 255 276.
[17] R.B. Bird, C.F. Curtiss, R.C. Armstrong and O. Hassager, Dynamics of Polymeric Liquids. Vol. :2,
Kinetic Theory, Wiley-interscience, New York, 1987.
[18] J.L. Lumley, Applicability of the oldroyd constitutive equation to flow of dilute polymer solutions,
Phys. Fluids, 14 (1971) 2282-2284.
[19] E.J. Hinch and L.G. Leal, The effect of Brownian motion on the rheological properties of a
suspension of non-spherical particles, J. Fluid Mech., 52 (19721 683 712.
[20] N. Phan-Thien and R.I. Tanner, A new constitutive equation derived from network theory, /.
Non-Newtonian Fluid Mech., 2 (1977) 353 365.
[21] E.B. Bagley, I.M. Cabott and D.C. West, Discontinuity in the flow curve of polyethylene, J. Appl,
Phys., 29(1958) 109 110.
136 D. Vlassopoulos, S.G. Hatzikiriakos / J. Non-Newtonian Fluid Mech. 57 (1995) 119 136

[22] S.G. Hatzikiriakos and J.M. Dealy, Role of slip and fracture in the oscillating flow of HDPE in a
capillary die, J. Rheol., 36 (1992) 845-884.
[23] W.R. Schowalter, The behavior of complex fluids at solid boundaries, J. Non-Newtonian Fluid
Mech., 29 (1988) 25 36.
[24] J.K. Hunter and M. Slemrod, Viscoelastic fluid flow exhibiting hysteretic phase changes, Phys.
Fluids, 26 (1983) 2345 2351.
[25] R.W. Kolkka, D.S. Malkus, M.G. Hansen, G.R. lerley and R.A. Worthing, Spurt phenomena of
the Johnson Segalman fluid and related models, J. Non-Newtonian Fluid Mech., 29 (1988)
303 335.
[26] J.A. Deiber and W.R. Schowalter, On the comparison of simple non-monotonic constitutive
equations with data showing slip of well-characterized polybutadienes, J. Non-Newtonian Fluid
Mech., 40 (1991) 141 150.
[27] T.C.B. McLeish and R.C. Ball, A molecular approach to the spurt effect in polymer melt flow, J.
Polym. Sci., Part B: Polym. Phys., 24 (1986) 1735-1745.
[28] M.E. Cates, T.C.B. McLeish and G. Marrucci, The theology of entangled polymers at very high
shear rates, Europhys. Lett., 21 (1993) 451 456.
[29] R.W. Kolkka and G.R. lerley, Phase-space analysis of the spurt phenomenon for the Giesekus
viscoelastic fluid model, J. Non-Newtonian Fluid Mech., 33 (1989) 305 323.
[30] G.V. Vinogradov, A. Ya. Malkin, Yu. G. Yanovskii, E.K. Borisenkova, B.V. Yarlykov and G.V.
Berezhnaya, Viscoelastic properties and flow of narrow distribution polybutadienes and polyisopre-
nes, J. Polym. Sci. Part A-2: Polym. Phys., 10 (1972) 1061 1075.
[31] M. Doi and S.F. Edwards, The Theory of Polymer Dynamics, Clarendon Press, Oxford, 1986.
[32] D.W. Van Krevelen, Properties of Polymers, Elsevier, Amsterdam, 1990.
[33] D.S. Malkus, J.A. Nobel and B.J. Plohr, Approximating piston-driven flow of a non-Newtonian
Fluid, CMS Technical Summary Report #93-9, University of Wisconsin, Center for the Mathemat-
ical Sciences, 1993.
[34] G. Schleiniger and R.J. Weinacht, A remark on the Giesekus viscoelastic fluid, J. Rheol., 35 (1991)
1157 1170.