Vous êtes sur la page 1sur 6

An Analysis of the Viscous Behaviour

of Polymeric Solutions
P . 1. CARREAU, D . DE KEE* and M . DAROUX**
Chemical Engineering Department, Ecole Polytechnique, Montrkal, Q d b e c , H3C 3A7

Viscosity data of an aluminum soap solution, a polyisobu- On a analysk, au nioyen de deux mod&les rhkologiques (le
tylene solution, a polyacrylamide solution, and ten polystyrene niodkle A de Carreau et celiii dEllis), les rksultats des viscositks
solutions have been analysed using two rheological models: d'nne solution de savon daluminum, une solution de poly-
the Carreau model A and the Ellis model. The model para- isobutylkne e t dix solutions de polystyrhe. On a obtenu les
meters were obtained by non-linear regression. The importance paramttres des niodhleles par regression non linkaire. On illustre
of the zero-shear viscosity for design purposes is illustrated l'importance pratique de la viscosit6 en l'absence de cisaille
by observing the appearance of a master curve in the repre- nient, en observant l'apparition d u n e courbe principale dans
sentation of the dimensionless viscosity versus the product la reprksentation de la viscositk a-diniensionnelle en fonction
of the zero-shear viscosity by the shear rate. A linear relation is du produit de la viscositk en l'absence de cisaillement par la
shown to exist between the logarithm of the zero-shear vis- vitesse de cisaillement. On niontre qu'il existe une relation
cosity and the polymer concentration. It is asserted that either linhire entre le logarithme de la viscositC en l'absence de
model should replace advantageously the commonly used power- cisaillement et la concentration des polymkes. On soutient que
law expression. les deux niodkles pourraient remplacer avantageusement l'ex-
pression de la loi de puissance eniployke communknient.

I t is well known that polymer melts or solutions and


highly concentrated solid suspensions do not obey
Newton's law of viscosity. Except for r a r e examples,
their form: each contains three parameters. In this
paper we focus attention on the zero-shear region
these materials are pseudoplastic, i.e. their viscosity of viscosity data obtained on the Rheogoniometer.
decreases as the rate of shear or the shear stress Our aim i s to show t h a t although the data appear
increases. Decreases in the viscosity by 2 o r 3 decades to be independent of the shear rate at low shear rate,
have been commonly observed for high molecular this is not exactly the case a s predicted by the two
weight polymer melts"' for shear rates ranging from models. The implication of this on falling sphere
10-2 to lo3 s-1. data is discussed. We f u r t h e r show t h a t the zero-
Recently most efforts in rheology of polymers have shear viscosity can be quite useful t o correlate non-
been aimed a t describing elastic effects. Quite a few Newtonian behaviour to the molecular weight and
phenomena, material functions or theories (see for concentration of a polymer solution.
examples references 2 and 3 ) have been reported o r
proposed to elucidate the so-called viscoelastic be- Rheological models
haviour of polymer solutions or melts. Although in The non-Newtonian viscosity is described respec-
some important industrial processes the time-effect tively for the two rheological models by
or elastic properties of these materials cannot be
ignored, the most important material function for Carreau Model A(1o) q / q o = 1/ [l + ( t ?)2]s.
~ .. . . . . . . . . .(1)
most engineering applications is the viscosity. I n Ellis model ( 1 1 ) : 11/11. = 1/[1 + I*-']. . . . . . . (2)
I712/71/2
fact the viscosity is sufficient to characterize rheo- where vo is the zero-shear viscosity
logically these materials in : tl is a timeconstant I
i ) steady simple-shear flow S, a are dimensionless parameters
T ~ is/ the
~ shear stress for which
ii) accelerated, decelerated or complex flows of 11 = 7 0 / 2 .
polymers of not too high molecular weight.
Both models a r e very similar i n form; for model A
Various models have been proposed to describe the the functional dependence is on the shear rate y where-
viscosity of polymer s01utions'~~~', melts"' suspensions as it is on the shear stress 7 1 2 f o r t h e Ellis model.
of solids'", biological fluids@', etc.. . Some of these
For characterization purposes, model A appears t o
equations have theoretical foundations'" but lack be more interesting since i t contains a time constant
flexibility t o f i t viscositv data f o r a large class of which is closely related to the elastic time constant
non-Newtonian fluids. Other models a r e empirical and f o r most polymeric materials"o'. However we note
their usefulness resides in their ability to correlate t h a t for the Ellis model t h e ratio of the two para-
adequately viscosity data of various types of fluids meters 7. and rh has the unit of time. The parameters
through a small number of meaningful parameters. S and a! a r e associated with the power-law behaviour,
The primary purpose of this work is t o evaluate and obviously for pseudoplastic materials S is greater
extensively two successful empirical models which than zero and a! greater than 1.
are known to f i t adequately viscosity data of various At very low s h e a r rates ( (tl 9) << 1and T~~t:- qo9 )
polymeric materials. These two equations, the Ellis Equation (1) and Equation (2) reduce respectively
model and one proposed previously by one of us (re- to :
ferred to, in this work, as model A) a r e similar in
7 / q o A 1 - S(tlY')2.. . .. . . . . . . . .. . . . . . . .. . . .. .. .. .(3)
'Present address : Department of Chemistry and Chemical Engineering, and
Royal Military College of Canada, Kingston. Ont.
""Present address: Labratoire des Sciences du GBnie Chimiaue. Ecole
Nationale Superieure des Industnea Chimipues. 71/70 = 1/11 +I 11"+ / 7 1 / 2 l n - ' 1 =
640 Nancy. cedex. France. 1 - ~ ~ o - y / T l ~ z ~ .* .......
- l . . . . ... . . . ... . . . . ...(4)

The Canadian Journal of Chemical Engineering, Vol. 57J A # d J 1979 135


I

0= I 5 % SEPARAN

@ : 2 X PIE
_ .. MODEL A

: E L L I S MODEL

I I I 1 1 1 , 1 1 , 1 I !,,Ud I ,,1,,,,1 ,
10- 10-2 10-1 100 10' 102
y, SEC.1

Figure 1 -Viscosity of the aluminum soap (E.L.), Separau, Figure 2 - Viscosity of the polystyrene solutions.
and polyisobutyleue (P.I.B.) solutions.

mers in Aroclor 1248 solvent (chlorinated diphenyls


Except for the case of a = 3, the functional de- with a viscosity of 0.3 Pa*s a t 25°C). The poly-
pendence in the low shear rate region is different. styrene solutions investigated are listed in Table 1.
For model A, the viscosity is a linear function of The P.I.B. solution was a 2% solution (by weight)
the second invariant 05 the rate-of-strain tensor, in Primol 355, a pharmaceutical grade white oil with
I1 = 2 y 2 The differences in behaviour are brought a viscosity of about 0.15 Pa's a t 25°C. The P.I.B. was
out by differentiating the respective expressions for a well fractionated cut having a molecular weight
7) with respect to 11: of 1.5 x 10' g/mol.
The polyacrylamide solution consisted of 1.5%
Model A aq / ,311 = - Sqe 1: / 2 . . . . . . ... . . . . . . . . . . . .(5) (by weight) of Dow Chemical Separan AP-30 in a
50/50 mixture by weight of water and glycerine. The
Ellis model aluminum soap solution was made of 70 g of alu-
minum laurate in 1000 ml of decalin and 20 ml of
a -3 m-cresol. A description of the preparation of this
aqiazr = - (a - 1) ( ~ ~ [ 2~ ~1~ 2 7 ~ / ~ 1111
1a--t (6)
solution is given by Huppler et al"*'.
For model A, the derivative is a negative constant
whereas for the Ellis model the derivative is a Eualuation of the models
decreasing or increasing function of the second in- Although the Ellis and the Carreau models are
variant depending on whether a is greater than or strongly non-linear with respect to the parameters,
smaller than 3. In no case is a plateau predicted, i.e. the optimum parameters can still be obtained from a
the viscosity is a decreasing function of the second least-squares method. However, the set of equations
invariant. Although the dependence of q on I1 is resulting from setting the partial derivatives with
weak in the very low shear-rate region, Newtonian respect to each of the parameters equal to zero is
behaviour is not predicted and q,,has to be interpreted difficult to solve. Three methods can be used to
as a limit a t which the viscosity approaches as I1 solve this problem : the Gauss-Newton, the steepest
tends to zero. descent, and the Marquardt methods. The Gauss-
Newton method converges only if the initial guesses
At very high shear rates, the viscosity a s predicted of the parameters are close to the final values"";
by Equations (1) or (2) decreases to zero. This is the steepest descent method converges for any initial
unreasonable, but the models can be easily corrected estimate, but may require a prohibitive computer
by inclusion of a high-shear viscosity term, qm.For time. We chose the Marquardt which is
example, the left member of Equations (1) or (2) a combination of the first two.
may be written as (7 -+>/ (yo- qm>.Equation (2)
so written is referred to as the Meter 3-
has a physical meaning, that is the value of the vis- TABLE1
cosity as the shear rate becomes very large. It is INVESTIGATED SOLUTIONS OF POLYSTYRENE
reasonable to assume that any pseudoplastic material
would behave as a Newtonian fluid a t high enough
shear rates. For polymer solutions, the limiting vis- Molecular
cosity is that of the solvent, but there are indications weight
dmol &?w/Mn wt. %
that ymis larger than q s (solvent viscosity). Never-
theless, for most engineering applications qmis very 411,000 1.06 2.5
small and negligible compared to q. 5.0
7.5
Fluids 860,000 1.15 1.0
The rheological equations are evaluated using vis- 2.5
cosity data published by Ashare''') and Huppler e t 5.0
a1<14)
. Quite a range of rheological behaviour is 7.5
covered by this choice of polystyrene solutions, a
polyisobutylene (P.I.B.) solution, a 1.6% polyacryl-
amide solution, and an aluminum soap solution. The
1,800,000 1.20
I 1.o
2.5
5.0
polystyrene solutions were nearly monodisperse poly-

136 The Canadian Journal of Chemical Engineering, VoZ. 57, A f i l , 1979


100

FLUID REFERENCE @
8

r
E 30 -
"
0
YI
-

I
00 005 0 10 0 15 0 20
2 7 2 :t

Figure 3 - Low-shear-rate viscosity


------ predictions of model A
- - -- _ - - predictions of the Ellis model.

The criterion for parameter value selection was


that the following quantity be a minimum:
N
n = .z wi (712iE - 71ZiC)2.. ...................(7)
2= 1
where N is the number of data, wi the weight associated
with each datum, and ~ ~ and 2 8T ~ respectively
~ ~ C the .OO DI .02 .03 .04 .OS .06 .07 .O 8 .09 .I0
experimental and calculated values of the shear stress (the
Polymer weight fraction, w
dependent variable as experimentally measured on a cone-
and-plate viscometer). To give equal importance to the Figure 4-Correlations of the zero-shear viscosity for the
deviations in the low shear stress region as well as in the polystyrene solutions.
high shear stress region, the weight was taken to be:
%
w i = (1/71ziE)'. ................................. .(8)
The principle of the non-linear least-squares method
is to expand the model into Taylor series, about the

TABLE
2
RHEOLOGICAL
PARAMETERS

I I Model A Parameters
Ellis Model
Parameters I Variance
T o
Pa s
7i/z
Pa I Model A Ellis
7% Alum. Soap
1.5% Separan
2% P.I.B.
171
1
211.7
I 3 1 876.0 1
91.8
1 169
1.061
45.2 ~

1
0.39
0.33
0.32
89.8
253.2
846.0
63.8
4.33
4.16
5.08
3.04
2.60
3.3 x 10-3
3.7 x 10-3
1.3 x
1.12
4.43
2.2
x
x
x
10-3
10-3
lo+
I I 1 I 1.32 X 10-4
P.S. 5.0 - 411
P.S. 1.0 - 1800
P.S. 2.5 - 411
1713.31
4

I 6 I 3.49 I
24.6
1 0.06
0.20
0.01
1I 0.17
0.11
0.15
24.6
3.84
3.53 1I
1230.1
109.0
1159.4 11
2.61
1.52
2.25 1I
1.7 X
1.9
2.7 X
x
lo-'
10-4
1I 2.1
9.6 X
x 10-3

P.S. 7.5 - 411 1 7 I 118.0 I 0.34 I 0.13 118.7 1654.0 2.36 6.3 x 5.2 x 10-4
1 I I I 4.7
P.S. 5.0 - 1800
P.S. 2.5 - 1800
P.S. 1.0 - 860
17132.4
8

I 10 1 2.05 1
746.0
I 6.95
0.81
0.05
1I 0.30
0.21
0.10
746.0
34.4
2.11 1
158.7
102.0
579.0 1
2.85
2.14
1.71 1
3.3
2.6
4.7
x
x
x
10-4
10-4 I 1.7
4.8
x
x
x
10-3
10-4
P.S. 2.5 - 860 I 11 1 10.2 I 0.07 I 0.20 11.2 1 401.9 1 2.13 I 2.5 X low3 I 3.1 X lo-'

1 1 1 1 1 1
~

P.S. 5.0 - 860 12 98.2 0.05 0.26 98.7 444.0 2.86 4.2 x 10-4 3.1 x 10-4
P.S. 7.5 - 860 ~ 1
7
1 617.3 1.30 0.33 594.4 652.0 3.75 2.2 x 10-4 7.5 x 10-4

The Canadiun Journal of Chemical Engineering, Vol. 57, A @ d , 1979 137


F
, FLUID R E F E R E N C E
F k
NUMBER FLU 3 REFERENCE
1011
UUMBER
b
I 10

v 11
0 12
A 1 3
L L L - L L A _iL

I0 ' 00 10' 102 10' 101 100 101 10 ' 10' 10'
y 71, pa
I
Y 7). pa

Figure 5 - Master curve for the polystyrene solutions with Figure 6 -Master curve for the polystyrene solutions with
M, = 411000 g/mol. M, = 860000 g/mol.
initial value of the parameters, retaining only the f i r s t
order terms : Low-shear-rate region
P Using principles of continuum mechanics, Dealy"*'
~ 1 2 , ' = 7 1 z i C (Ti, b l o ) 2 + j= 1
has shown that the concept of a perfectly Newtonian
fluid is inconsistent with that of a "simple fluid",
( d r i 2 , l d b j ) Abj, i = I, 2 . . . .N.. ...... . . . ... . . . .(9) i.e. if the viscosity of any pseudoplastic fluid can
where P is the number of parameters ( 3 in our case). be described by a simple polynomial expansion of
The correction vector A b appears linearly in t h i s terms ( t j j ) " this
, material cannot exhibit a New-
equation; hence it can be solved by a linear least- tonian behaviour in the low-shear-rate region.
squares method. The convergence criterion proposed Although the low-shear-rate viscosity data obtained
by Marquardt"", is defined as on the Weissenberg Rheogoniometer a r e not very
accurate (+ 10% f o r some data according to Huppler
IAb,fI / ( r + lb,"] < E , j = 1, 2, 3 .. . . . . . . . . ... ..(lo)
e t al(I4), we have analysed the data of every fluid
For this work, E was set equal to lo5 and r equal which appear to be, on a log-log basis of Figures 1
to low3,which assures convergence even when a para- and 2, in a plateau region. Polynomial regressions
meter bj approaches zero. were tested N i t h each set of data and a best f i t
I n all cases only a few seconds of computer time was obtained, in all cases, with a linear expression
on a n IBM 360/75 were necessary to reach con- of the viscosity versus the second invariant of the
vergence, even for inaccurate initial estimates of t h e rate-of-strain tensor. This is in agreement with the
parameters. For the Ellis model, which is implicit predictions of model A a s given by Equation ( 3 ) .
in the shear stress, slightly more computer time was The behaviour in the low-shear-rate region is shown
required. for three different fluids i n Figure 3. The solid
The fits obtained with the two models a r e shown lines a r e the predictions of model A whereas the
in Figures 1 and 2. The solid and dashed lines a r e dashed lines a r e those of the Ellis model. Both models
respectively the predictions of model A and of t h e predict an increasing viscosity as the second invariant
Ellis model. The optimum parameters and the vari- of the rate-of-strain tensor (I1 =.i/') decreases to
ances are presented in Table 2. A s indicated by t h e zero. Although the scattering in the data is too large
values of the parameters, the various fluids, labeled to be conclusive, model A appears to be more satis-
from 1 to 13, cover a wide range of viscous prop- factory. The differences in behaviour predicted by
erties. Overall, both models a r e quite successful i n the Ellis model f o r various values of ~y is striking.
describing the data. We should point out that t h e For a! much greater than 3 (fluid reference number 1
large deviations observed at high shear rates f o r on Figure 3 with a! = 5.08), t h e viscosity is almost
the 2% P.I.B. (fluid number 3) could be considerably constant in the low-shear-rate region. For a! smaller
reduced by including in the models a fourth para- than 3, a s f o r fluid reference number 9, the Ellis
meter, qm (an infinite shear viscosity), as men- model predicts a rather rapid decrease of viscosity
tioned previously. in the vicinity of zero shear. This is unlikely, but
In all cases, the variances of t h e parameters f o r more accurate data are required to prove this point.
both models were found to be very small. For ex- Our findings confirm the observations of various
ample, the intervals of the parameters of Model A, at researchers with t h e experiment of falling spheres.
95% confidence, for t h e 1.6% Separan data are: I n order to obtain the low or zero-shear viscosity
from falling-sphere data, one has to assume t h a t the
211.1 < qo < 212.4 Pa s fluid obeys Stokes' law and the zero-shear viscosity is
45.0 < t , < 45.4 s determined by extrapolating the apparent viscosity so
0.3352 < S C 0.3353 obtained to the zero apparent shear rate or apparent
I n Model A, there is a strong correlation between shear stress. I n all cases reported, the data f o r poly-
the parameters 7. and tl. For the Ellis model, a meric fluids, even a f t e r corrections f o r wall effects
similar correlation is observed between t h e para- (see reference"') f o r example), show a clear depend-
meter 7. and m2. This is not surprising, since t h e ence on either t h e shear rate or t h e shear stress.
power-law regions (the straight lines on the log-log The assumption of Newtonian behaviour i n t h e low
plots of Figs. 1 and 2) can be adequately represented shear rate region is not correct, and since t h e t r u e
by a two-parameter model and most of the data shear stress o r shear rate cannot be determined f o r
are in those regions. the analysis of the flow about a sphere, t h e zero-

138 The Canadian Journal of Chemical Engineering, Vol. 57, April, 1979
101 , , , ,,, / , , , , , I I , , ., I 1 I
' " i ' " "
TABLE3
PARAMETERS
FOR THE SOLUTIONS.
POLYSTYRENE

t I -
M, 1
I

Fluid Reference
I

g/mol Number I s

F t
411 000
860000
1800000 1 ::z:
10, 11, 12, 13
1 405.0
1 0.206
0.279
0.315

.-I
FLUID REFERENCE
NUMBER
I I I
* 5
m a The values of the parameters of Table 3 differ from those pre-
A 9 sented in Table 2 ( T ~ v,/tl) since 7 1 and S are not really inde-
pendent of concentration.
2 ,-&d u 1 1 1 ' l i 1 L ' d l L U I LIillLLi
lo1 100 10' 102 10' I 0'
Y 77.. PO

Figure 7 -Master curve for the polystyrene solutions with The viscosity master curves can be described by
M, = 1800000 g/mol. model A rewritten in t h e form
7 /?. = 1/ [l + 70/71)*Is.. .. . . . . . . . . . ... . .. . .(13)
shear viscosity obtained by extrapolation is probably where r1 has the units of stress and is independent
erroneous. Evidence of this was shown by Huppler of the polymer Concentration. The predictions of
e t a1114)who reported large discrepancies between Equation (13) a r e presented by the solid lines on
values of zero-shear viscosity determined from falling- Figures 5 to 7. The parameters r1 and S for t h e
sphere data and those measured on the Weissenberg three molecular weights a r e given in Table 3.
Rheogoniometer. Although these discrepancies can be Similar results have been observed recently by
explained by the non-existence of a viscosity plateau Parrini et al(*') for molten aliphatic polyamides.
at low shear rate, the elastic nature of the fluids
may also interefere in the flow of falling spheres Conclusions
since the strain history is not uniform across the Two rheological equations have been evaluated with
fluid for this flow. We suggest t h a t a complete viscosity data of 13 different polymer solutions.
ana1,ysis of falling-sphere data with model A be made These pseudoplastic fluids have a wide range of
to clarify the importance of the slightly non-New- viscous properties. Both the Carreau model A and
tonian behaviour in the determination of t h e zero- the Ellis model a r e quite successful in describing t h e
shear viscosity. viscosity data in terms of three parameters. There
Correlations with the zero-shear viscosity is some indication t h a t in the very low shear-rate
region the viscosity is not constant, but increases
We have investigated various possible correlations slowly a s t h e second invariant of t h e rate-of-strain
for Model A with the parameters reported i n Table 2. tensor decreases to zero.
The following two observations are of particular in- For the polystyrene solutions, the zero-shear vis-
terest. In Figure 4, we show t h a t there exists a cosity was correlated with the polymer concentration
linear relationship between log (zero-shear vis- and master curves of the dimensionless viscosity (q/
cosity) and w,the weight fraction of the polymer q o ) versus the product of the zero-shear viscosity by
in solution. This set of data however does not verify the shear rate were obtained.
the well-known correlation of Fox and F10ry"~': There exist in the literature many other rheological
IKM, ,Fw< M, equations. However, the two equations evaluated here
a r e believed to be the most flexible and capable of
characterizing any non-Newtonian fluids. These two
models should replace advantageously the commonly
where M , is a critical molecular weight. This cor- used power-law expression, since they contain mean-
relation has been verified mostly for melts and its ingful parameters that can be easily determined. It
validity for polymer solutions is questionable. is true t h a t the use of the simple power-law model
As suggested by Vinogradov et al. (in ref. 20) for leads to analytical solutions for a number of flow
polyethylene melts, the viscosity data of the various problems of practical importance. These solutions
polystyrene solutions for each molecular weight can may become quite complex when a three parameter
be correlated approximately with the apparent shear model is used. However, a s computers become more
stress defined by accessible, analytical solutions will be of decreasing
7]Z0 = -y Qa * . . , . , . . , . . . . . . . , . . . . . . . . . . , , . , . . . .(12) importance in comparison with more realistic solu-
tions obtained with models describing adequately t h e
The master curves of 11/71. vs. y 11. so obtained fluid behaviour.
ar e presented in Figures 5 to 7 for the three mole-
cular weights. Although there a r e important devia- Acknowledgments
tions a t large values of 9 yo, the influence of the The authors wish to acknowledge financial assistance from the
polymer concentration is almost eliminated, i.e. the Canadian National Research Council ( g r a n t A-6817).
effect of the concentration is accounted for by the
apparent shear stress. This is useful for design pur- Nomenclature
poses; from one set of data for a given polymer = any of the parameters of both models
weight one can predict approximately, with the help = proportionality constant in Equation (13)
of correlations such as those presented in Figures 4 = critical molecular weight
to 7, the viscous behaviour of polymer solutions of = number averaged molecular weight
various concentrations. = weight averaged molecular weight

The Canadian Journal of Chemical Engineering, Vol. 57, AM.1, 1979 139
N = number of data ( 6 ) Ree, F. H., Ree, T. and Eyring, H., Ind. Eng. Chem.
P = number of parameters 50, 1036 (1958).
S = dimensionless parameter in Equation (1) (6) Chu, K. K. and Rudin, A., Trans. SOC. Rheol. 18:1,
t1 = time constant in Equation (l),s 103 (1974).
wi = weight associated with each data, defined by Equation
(8)
(7) Dawson, R., Tans. SOC. Rheol. 19:2, 229 (1975).
zu =polymer weight fraction. (8) Whitmore, R. L., Biorheology I, 201 (1967).
a! = dimensionless parameter in Equation (2) (9) Graessley, W. W., J. Chem. Phys. 54, 5143 (1971).
= shear rate, s-l (10) Carreau, P. J., Trans. SOC. Rheol. 16:1, 99 (1972).
A = squares of deviations, defined by Equation (7) (11) Bird, R. B., Armstrong, R. C. and Hassager, O.,
?) = viscosity, Pa . s “Dynamics of Polymeric Liquids: Volume I, Fluid
T~ = zero-shear viscosity, Pa . s
qrn = infinite-shear viscosity, Pa s
7
-
= convergence parameter defined by Equation (10)
Mechanics”. Wiley, New York (1977).
(12) Meter, D. M. and Bird, R. B., AIChE J. 10, 878
~~2 = shear stress, Pa
(1964).
~ ~ =
/ 2characteristic shear stress in Equation (2), Pa (13) Ashare, E., Trans. SOC. Rheol. 12:4, 535 (1968).
~ ~ 2= , apparent shear stress, defined by Equation (12), Pa (14) Huppler, J. D., Ashare E. and Holmes, L. A., Trans.
T~ = characteristic shear stress in Equation (13), Pa SOC. Rheol. 11:2, (1967).
I1 = second invariant of the rate-of-strain tensor, s - ~ (15) Marquardt, D. W., J. SOC. Indust. Appl. Math. 11,
431 (1963).
Subscrifit (16) Marquardt, D. W., “Least Squares Estimation of
Nonlinear Parameters, A Computer Program in
i = refers to a data point Fortran IV Language,” IBM SHARE Library, Dis-
3 = refers to a parameter”
0 = refers1to the initial value of a parameter. tribution Number 309401, August (1966).
(17) Cygan, D. A. Caswell, B., Trans. SOC.Rheol. 15:4,
Subsmifit 663 (1971).
(18) Dealy, J. M., Trans. SOC. Rheol. 14:4, 461 (1970).
= refers to a calculated value (19) Fox, T. G. and Flory, P. J., J Phys. Chem. 21, 581
E = refers to an experimental data (1958).
= refers to the final value of a parameter.
(20) Semjonow, V., Adv. Polym. Sci., 5, 387 (1968).
(21) Parrini, P., Romanini, D. and Righi, G. P., Polymer
References 17, 377 (1976).
(1) Han, C. D., Yu, T. C. and Kim, K. U., J. Appl.
Polym. Sci. 15, 1149 (1971).
(2) Lodge, A. S., “Elastic Liquids”, Academic Press, Manuscript received January 14, 1978; accepted for
New York (1964).
(3) Han, C. D. “Rheology in Polymer Processing”, publication November 17, 1978.
Academic Press, New York (1976).
(4) Spriggs, T. W. and Bird, R. B., Ind. Eng. Chem.
Fundam. 4, 182 (1964). * * *

140 The Canadian Journal of Chemical Engineering, Vol. 57, A@il, 1979

Vous aimerez peut-être aussi