J. Non-Newtonian Fluid Mech.

129 (2005) 177–181

Short communication

Non-Newtonian fluids with a yield stress

H. Zhu a , Y.D. Kim b , D. De Kee a,∗
a Department of Chemical and Biomolecular Engineering and Tulane Institute for Macromolecular Engineering and Science (TIMES),
Tulane University, New Orleans, LA 70118, USA
b Faculty of Applied Chemistry, Chonnam National University, Kwangju 500-757, Korea

Received 11 April 2005; received in revised form 24 May 2005; accepted 10 June 2005

Abstract

A modified Herschel–Bulkley model [E. Mitsoulis, S.S. Abdali, Flow simulation of Herschel–Bulkley fluids through extrusion dies, Can. J.
Chem. Eng. 71 (1993) 147–160] predicts an infinite apparent viscosity at vanishing shear rate. Furthermore, the dimensions of one parameter
depend on another parameter. In this contribution, we propose a generalized model based on earlier work by De Kee and Turcotte [D. De
Kee, G. Turcotte, Viscosity of biomaterials. Chem. Eng. Commun. 6 (1980) 273–282] and on the work of Papanastasiou [T.C. Papanastasiou,
Flows of materials with yield, J. Rheol. 31 (1987) 385–404] to solve the problems associated with the modified Herschel–Bulkley model.
Compared to the responses of the Papanastasiou model and the modified Herschel–Bulkley model, the proposed generalized model provides
the expected improvements and is capable of predicting successfully the rheological behavior (viscosity and yield stress) of Carbopol 980
dispersions.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Herschel–Bulkley model; Papanastasiou model; Shear thinning/thickening; Yield stress

1. Introduction yield surfaces and applying two different constitutive equa-

tions (Eqs. (1) and (2)) across them. In addition, at vanishing
Many materials such as paint, slurries, pastes and some shear rates, the apparent viscosity in Eq. (2) becomes infinite,
food products have been described by the Bingham model, i.e. which leads to a discontinuity and numerical difficulties.
a yield stress followed by Newtonian behavior. The Bingham To overcome these issues, Papanastasiou [3] proposed
model [4] is given by: a modified Bingham model to approximate the rheological
behavior of Bingham type materials, i.e.
γ̇ = 0 if τ ≤ τ0 , (1)



 1 − exp (−mγ̇)
τ0 τ = η + τ0 γ̇, (3)
τ = η+ γ̇ if τ > τ0 , (2) γ̇
γ̇
where parameter m is a large number to guarantee large appar-
where τ is the stress tensor, γ̇ the rate of deformation tensor, τ
ent viscosity at vanishing rates of strain. Eq. (3) indicates
the second invariant of the stress tensor, η the viscosity, τ 0 the that as the shear rate approaches zero, the apparent viscos-
yield stress and γ̇ is the second invariant of the rate of defor- ity is given by η + mτ 0 , which is a large but finite value. By
mation tensor. However, the Bingham model is not amenable using the Papanastasiou model, tracking the material yield
to numerical analysis because in some complex applications, surfaces is completely unnecessary. The validation and effec-
parts of material flow while the rest behaves as a solid. This tiveness of the Papanastasiou model has been studied by
causes difficulties in tracking the shape and the location of the several researchers including Elwood et al. [5], Blackery and
Mitsoulis [6] and Smyrnaios and Tsamopoulos [7]. How-
∗ Corresponding author. ever, applications of the Papanastasiou model are limited to
E-mail address: ddekee@tulane.edu (D. De Kee). Bingham materials, which show a yield stress followed by

0377-0257/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnnfm.2005.06.001
178 H. Zhu et al. / J. Non-Newtonian Fluid Mech. 129 (2005) 177–181

Newtonian behavior. It cannot predict the rheological behav- reported that Carbopol 980 dispersions showed an apparent
ior of shear thinning fluids with a yield stress, which is typical yield stress [12]. Carbopol dispersions (0.09 and 0.08 wt%)
for complex fluids. were prepared according to Noveon TDS-103 [13] and neu-
Mitsoulis and Abdali [1] modified the Herschel–Bulkley tralized to pH 7 according to Noveon TDS-237 [14]. Rhe-
model [8] to predict shear thinning behavior by specifying ological measurements were performed at 25 ◦ C using a
the viscosity term in Eq. (3) as: TA Instruments AR-2000 rheometer. Geometries such as a
40 mm cone and plate (cone angle 1◦ , gap 0.028 mm) and a
η = kγ̇ n−1 . (4)
standard vane attachment (vane diameter 14 mm, vane length
Then the modified model becomes: 42 mm, cup diameter 15 mm) were used. A solvent trap was

 used to prevent drying. Rate-controlled measurements were
1 − exp(−mγ̇)
τ = kγ̇ n−1 + τ0 γ̇. (5) performed to observe the viscosity behavior, and the stress-
γ̇ controlled measurements were performed to observe the yield
Alexandrou et al. [9] investigated the flow instabilities of stress.
Herschel–Bulkley fluids by employing the modified model
(Eq. (5)). In their paper, they stated that for γ̇ ≈ 0, the appar-
ent viscosity is a finite value. However, according to Eq. (5), 3. Proposed generalized model
we can see that when the parameter n is less than one, which is
typical for shear-thinning fluids, the apparent viscosity tends Earlier, De Kee and Turcotte [2] proposed the following
to infinity when the shear rate approaches zero. The modified model:
Herschel–Bulkley model (Eq. (5)) does not follow Papanas-
tasiou’s [3] idea that the apparent viscosity should be a finite η = τ0 γ̇ −1 + η1 exp(−t1 γ̇). (6)
value at vanishing shear rates. Furthermore, in Eq. (5) the
dimension of parameter k depends on parameter n. This may Note that the second term in Eq. (6) is a finite number even
cause a problem when constructing a parameter with the unit at vanishing shear rate. In the current study, we generalized
of time based on the parameters k and n [10]. Therefore, it De Kee and Turcotte’s model [2] by incorporating Papanasta-
is desirable to have a generalized model that can solve these siou’s idea, i.e. we make the apparent viscosity a finite value
problems. at vanishing shear rates:
Carbopol dispersions show shear thinning behavior as well

1 − exp(−mγ̇)
as an apparent yield stress [11,12], i.e. they are ideal materials τ = η1 exp(−t1 γ̇) + τ0 γ̇. (7)
to test the generalized model proposed in this study. Carbopol γ̇
is a microgel, which is a collection of highly cross-linked Unlike the modified Herschel–Bulkley model (Eq. (5)),
polymer particles, which function as efficient thickeners and where the dimensions of parameter k depend on parameter n,
stabilizers for suspensions or emulsions. Carbopol cross- the dimensions of the parameters in (Eq. (7)) are well defined
linked polymer particles are individually gelled but together and independent of each other. In addition, the apparent vis-
act effectively as a concentrated dispersion even though the cosity is bounded at vanishing shear rate.
actual concentration is low. Uhlherr et al. [11] reported that
Carbopol dispersions showed a “ductile-type” yielding pro-
cess, due to the deformability of particles and entanglements 4. Comparison of the numerical response for
between particles. Much of the exceptional utility of Car- different models
bopol results from the hydrophilic nature of the polymer. The
viscosity of a Carbopol dispersion is sensitive to pH, with a Fig. 1 shows the apparent viscosity as a function of shear
maximum plateau in the range pH 5–9, and with a consider- rate predicted by three different models: the Papanastasiou
able decrease in viscosity above and below this range. model, the modified Herschel–Bulkley model and the pro-
Here, we introduce a generalization of a model intro- posed model. From this plot, we can see that both the Papanas-
duced earlier by De Kee and Turcotte [2]. The model is tasiou model and the proposed model predict a finite (large
well-posed and offers some advantages over the modified value) viscosity at vanishing shear rate, since a τγ̇0 contribu-
Herschel–Bulkey model for non-Newtonian fluids with a γ̇)
yield stress. The proposed generalized model successfully tion has been replaced by a term of the form τ0 1−exp(−m
γ̇ .
predicts the rheological behaviors of Carbopol 980 disper- However, for the modified Herschel–Bulkley model, the vis-
sions. cosity approaches infinity when the shear rate approaches
zero, even though the τγ̇0 contribution has been replaced.
When the shear rate is within the range of 10−3 to 103 s−1 ,
2. Experimental the apparent viscosity predictions from all three models are
comparable. The inset illustrates the difference in the appar-
Carbopol 980 (Noveon), which is a non-toxic version of ent viscosity predictions from the three models in the shear
Carbopol 940, was used in this experiment because it was rate range of 102 to 103 s−1 .
 H. Zhu et al. / J. Non-Newtonian Fluid Mech. 129 (2005) 177–181 179

Fig. 1. Comparison of viscosity prediction for three different models: (—) proposed model (Eq. (7)): (t1 = 4.52 × 10−4 s, η1 = 0.0115 Pa s, τ 0 = 40 Pa, m = 50 s);
(- - -) Papanastasiou model (η = 0.0115 Pa s, τ 0 = 40 Pa, m = 50 s); (– · – · –) modified Herschel–Bulkley model (k = 0.187 Pa sn , n = 0.3, τ 0 = 40 Pa, m = 50 s).

The relationships between shear stress and shear rate cal- predict the shear thinning behavior. Both the proposed model
culated from the three different models are shown Fig. 2. It (Eq. (7)) and the modified Herschel–Bulkley model show a
can be seen from this figure that all the three models predict a shear-thinning response after reaching the yield stress. The
yield stress. As expected, the Papanastasiou model does not inset shows that after the yield stress is reached, the modified

Fig. 2. Comparison of shear stress prediction from three different models: (—) proposed model (t1 = 4.52 × 10−4 s, η1 = 0.0115 Pa s, τ 0 = 40 Pa, m = 50 s); (- - -)
Papanastasiou model (η = 0.0115 Pa s, τ 0 = 40 Pa, m = 50 s); (– · – · –) Modified Herschel–Bulkley model (k = 0.187 Pa sn , n = 0.3, τ 0 = 40 Pa, m = 50 s).
180 H. Zhu et al. / J. Non-Newtonian Fluid Mech. 129 (2005) 177–181

the lines are the predictions of the proposed model (Eq. (7)).
The experimental data were obtained using the standard vane
geometry by the stress ramp method, which is one of the pos-
sible direct yield stress measurements [15]. The yield stress
values for 0.08 and 0.09 wt% Carbopol solutions are 2.15 and
5.84 Pa, respectively. As pointed out by Kraynik [16], there
are three yield stress values characterizing the rheological
behavior of a material: (1) an elastic-limit yield stress, (2) a
static yield stress and (3) a dynamic yield stress. The elastic-
limit yield stress is not the limit of linear behavior, but rather
the limit of reversibility for the material. The absence of linear
behavior generally occurs before the elastic limit. The static
yield stress is defined as the minimum stress level leading to
unbounded strain or deformation of the material. The plateau
Fig. 3. Viscosity and shear stress as a function of shear rate for Cabopol stress for large strains is termed the dynamic yield stress. It
980 dispersions. (
), () viscosity data; (), (䊉) shear stress data. (—)
Prediction of Eq. (7) (0.09 wt% Carbopol dispersion with τ 0 = 9.17 Pa,
is the extrapolated stress in the limit of zero-shear rate on a
η1 = 0.148 Pa s, m = 2000 s and t1 = 0.00294 s; 0.08 wt% Carbopol dispersion shear stress-shear rate plot.
with τ 0 = 3.0 Pa, η1 = 0.15 Pa s, m = 2000 s and t1 = 0.0089 s). The results in Fig. 4 show that the proposed model can
successfully predict the stress controlled rheological behav-
ior. Note that the proposed model can reproduce the dramatic
Herschel–Bulkley model predicts lager shear stress values
change in the viscosity/shear-stress behavior and therefore
than those predicted by the Papanastasiou model at relatively
predict the sample yield stress.
small shear rates, due to the unbounded characteristic of the
modified Hershel–Bulkley model in terms of the viscosity.
The proposed model predicts shear stress values smaller than
those obtained via the Papanastasiou model. 5. Conclusions
Viscosity and shear stress as a function of shear rate for
Cabopol 980 dispersions are presented in Fig. 3. The filled We proposed a generalized model and compared its
symbols represent shear stresses and the open symbols are response with the Papanastasiou model and the modified
viscosities. As shown in Fig. 3, both 0.08 and 0.09 wt% Herschel–Bulkley model. We find that the proposed model
Carbopol dispersions show yield stresses and shear thinning offers advantages over the modified Herschel–Bulkley model
behavior. Quantitative agreement with Eq. (7) is shown. The in that it predicts a finite value of the apparent viscosity when
model parameters were determined from the stress-shear rate the shear rate approaches zero. Also, the dimensions of the
data. The proposed (generalized De Kee–Turcotte) model parameters are independent on each other. The Papanasta-
with the parameters obtained from the shear stress-shear rate siou model does not predict the shear thinning behavior after
data was used to predict the stress-controlled experimental the yield stress is obtained. We have evaluated the predic-
data to determine whether the model can predict the yield tive capabilities of the proposed model with the rheological
behavior. Viscosity as a function of shear stress is presented properties of Carbopol 980 dispersions. The results show
in Fig. 4. The symbols represent the experimental data and quantitative agreement between the data and model predic-
tions.

References

[1] E. Mitsoulis, S.S. Abdali, Flow simulation of Herschel–Bulkley flu-

ids through extrusion dies, Can. J. Chem. Eng. 71 (1993) 147–
160.
[2] D. De Kee, G. Turcotte, Viscosity of biomaterials, Chem. Eng. Com-
mun. 6 (1980) 273–282.
[3] T.C. Papanastasiou, Flows of materials with yield, J. Rheol. 31
(1987) 385–404.
[4] E.C. Bingham, Fluidity and Plasticity, McGraw-Hill, New York,
1922, pp. 215–218.
[5] K.R.J. Elwood, G.C. Georgiou, T.C. Papanastasiou, J.O. Wilkes,
Laminar jets of Bingham-plastic liquids, J. Rheol. 34 (1990)
787–811.
Fig. 4. Viscosity as a function of shear stress for Carbopol 980 dispersions. [6] J. Blackery, E. Mitsoulis, Creeping motion of a sphere in tubes filled
(
), () Experimental data; (—) prediction of Eq. (7). The model parame- with a Bingham plastic material, J. Non-Newtonian Fluid Mech. 70
ters are as in Fig. 3. (1997) 59.
 H. Zhu et al. / J. Non-Newtonian Fluid Mech. 129 (2005) 177–181 181

[7] D.N. Smyrnaios, J.A. Tsamopoulos, Squeeze flow of Bingham plas- [12] G.P. Roberts, H.A. Barnes, New measurements of the flow-curves for
tics, J. Non-Newtonian Fluid Mech. 100 (2001) 165. Carbopol dispersions without slip artefacts, Rheol. Acta 40 (2001)
[8] W.H. Herschel, R. Bulkley, Konsistenzmessungen von Gummi- 499–503.
Benzol-Losungen, Kolloid-Z. 39 (1926) 291–300. [13] Noveon, TDS-103, 1993, http://www.personalcare.noveon.com/
[9] A.N. Alexandrou, T.M. McGilvreay, G. Burgos, Steady Herschel– techdata/Carbopol980.asp.
Bulkley fluid flow in three-dimensional expansions, J. Non- [14] Noveon, TDS-237, 1998, http://www.personalcare.noveon.com/
Newtonian Fluid Mech. 100 (2001) 77. techdata/Carbopol980.asp.
[10] P.J. Carreau, D. De Kee, R.P. Chhabra, Rheology of polymeric sys- [15] L. Zhu, N. Sun, K. Papadopoulos, D. De Kee, A slotted plate
tems: principal and applications, Hanser/Gardner Publications Inc., device for measuring static yield stress, J. Rheol. 45 (2001) 1105–
1997. 1122.
[11] P.H.T. Uhlherr, J. Guo, C. Tiu, X-M. Zhang, J.Z.-Q. Zhou, T.-N. [16] A.M. Kraynik, ER fluid standards: comments on ER fluid rheol-
Fang, The shear-induced solid–liquid transition in yield stress mate- ogy, in: J.D. Carlson, A.F. Sprecher, H. Conrad (Eds.), Proceedings
rials with chemically different structures, J. Non-Newtonian Fluid of the Second International Conference on ER Fluids, Technomic,
Mech. 125 (2005) 101–119. Lancaster, PA, 1990.