You are on page 1of 12

Rheological properties of coarse food suspensions

in tube flow at high temperatures
Krittalak Chakrabandhu
1
, Rakesh K. Singh
*
Department of Food Science, Purdue University, W. Lafayette, IN 47907-1160, USA
Received 3 April 2001; received in revised form 27 January 2004; accepted 24 February 2004
Abstract
The effects of particle concentration and carrier fluid temperature on rheological behavior of model food suspensions consisting
of 1.5% CMC solution and green peas (15–30% v/v) were investigated using a tube viscometer. The flow behavior of the suspensions
was represented by the power law model. The suspension consistency coefficient (m
Ã
) increased with particle concentration and
decreased with temperature, whereas the opposite trends were observed for the suspension flow behavior index (n
Ã
). Among various
theoretical, semi-empirical, and empirical equations tested for suspension apparent viscosity (l
Ã
) estimation, the third order
expansion of Einstein equation, which was derived via the hydrodynamic approach, provided the best estimates for l
Ã
. Of equations
tested for m
Ã
estimation, those in which n
Ã
was included offered better estimates of experimental values, with an empirical equation
obtained based on the Einstein equation and the incorporation of n
Ã
term providing the best m
Ã
estimation. These findings suggest
that, for concentrated coarse suspensions subjected to conditions presented here, the dependence between m
Ã
and n
Ã
is of importance
and should be considered in order to achieve a better m
Ã
estimation. Besides, better representations for power law parameters of such
suspensions may be obtained based on a theoretical expression derived for l
Ã
via the hydrodynamic approach. The study presented
here provides a much-needed insight toward the flow behavior of concentrated coarse food suspensions at high temperature,
information of which is vital for various food processes.
Ó 2004 Elsevier Ltd. All rights reserved.
Keywords: Non-Newtonian fluids; Rheological properties; Coarse suspension; Particulate foods
1. Introduction
Designing an aseptic process requires information
about flow properties of the foods. Flow properties have
significant effects on the residence time distribution and
heat transfer of the particulate food system as reported
by several researchers (Awuah, Ramaswamy, Simpson,
& Smith, 1996; Ramaswamy, Awuah, & Simpson, 1996;
Sandeep & Zuritz, 1994). A number of studies have been
done on rheological behavior of fluid foods and many
are summarized by Holdsworth (1971). However, only a
few studies on rheological behavior of coarse food sus-
pensions can be found in the literature, especially with
the carrier fluids that exhibit non-Newtonian behavior
(Bhamidipati & Singh, 1990; Martinez-Padilla, Cornejo-
Romero, Cruz-Cruz, Jaquez-Huacuja, & Barbosa-Ca-
novas, 1999; Pordesimo, Zuritz, & Sharma, 1994).
Due to the larger particle size, only few types of
viscometers can be used in characterizing the flow
behavior of coarse suspensions. These include tube
viscometers (Bhamidipati & Singh, 1990), wide-gap
parallel plate viscometer (Pordesimo et al., 1994), wide-
gap rotational viscometer (Martinez-Padilla et al.,
1999). The above-mentioned equipments are although
functional, difficulties still exist especially in the studies
for dense suspensions. Some of the difficulties could be
caused by the interference to the equipment parts by the
large particles as reported by Martinez-Padilla et al.
(1999), centrifugal effects in the parallel plate viscometer
as pointed out by Pordesimo et al. (1994), or simply
large amount of materials and floor space required to
run the tests using the tube viscometer. This leads to the
scarcity of the knowledge on the flow behavior of coarse
food suspensions. More insight on the flow behavior of
Journal of Food Engineering 66 (2005) 117–128
www.elsevier.com/locate/jfoodeng
*
Corresponding author. Address: Department of Food Science and
Technology, University of Georgia, Food Science Building, Athens
30602, USA. Tel.: +1-706-542-2286; fax: +1-706-542-1050.
E-mail address: rsingh@arches.uga.edu (R.K. Singh).
1
Current address: Death Receptor Laboratory, Institute of Signal-
ing, Developmental Biology and Cancer Research, CNRS UMR 6543,
33 Ave. de Valombrose 06189, Nice, France.
0260-8774/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2004.02.039
coarse food suspensions particularly in the conditions
applicable to aseptic processing is clearly desirable. The
objectives of this study were (1) to characterize the flow
behavior of model food suspensions consisting of non-
Newtonian carrier fluid and high concentration of large
particulates at high temperatures, (2) to investigate the
effect of particle concentration and temperature on flow
behavior of non-Newtonian coarse suspensions, and (3)
to investigate the applicability of several mathematical
expressions in explaining the effect of particle concen-
tration on the flow behavior of coarse suspensions.
2. Materials and methods
2.1. Food system
Green peas were used as model particulate phase
because of their commercial availability and their
applications in real food in the market. Frozen green
peas were purchased from Purdue University food store.
Green peas were thawed and drained prior to each
experimental run.
An aqueous solution (1.5% w/v) of sodium carboxy-
methylcellulose (CMC) (TIC Gum, Inc., Belcamp, MD)
was selected as the carrier fluid. The CMC solution was
prepared by slowly adding the desired amount of CMC
powder into a mixing tank filled with water continu-
ously agitated by a mechanical mixer. The mixture was
agitated for approximately 45 min then left overnight
(12–15 h) for the CMC to dissolve. Prior to the experi-
mental run, the CMC solution was mixed with green
peas to attain the specified particle volume fractions
(15–30%v/v). Due to the delicate texture of green peas,
the CMC-pea suspension was made in several small
batches of 0.0379 m
3
to achieve the desired particle
concontration. The suspension was mixed manually to
ensure the uniform distribution of green peas in the
suspension and then delivered into the pump feeder
during the experimental run. The physical properties of
the food materials are summarized in Table 1.
2.2. Experimental setup
The rheological characterization of the fluid food,
with or without particulates in aseptic processing con-
ditions was performed using an in-line tube viscometer.
The schematic diagram of the experimental setup is
shown in Fig. 1. The system consisted of a 150 l feed
tank, a moving pocket type pump (Moyno
â
Product
Progressive cavity pump, Robbins and Myers, Inc.,
Springfield, OH), a helical double tube heat exchanger
(46.3 m long, 0.032 m I.D.) as a heater, a helical double
tube heat exchanger (33.5 m long, 0.032 m I.D.) as a
cooler (Stork, Amsterdam, Netherlands), an electro-
magnetic flow meter (Promag 33, Endress and Hauser
Inc. Greenwood, IN), an insulated straight stainless steel
tube (0.022 m I.D.), a differential pressure transducer
(Rosemount Model 3051 Smart Pressure Transmitter,
Rosemount Inc., Eden Prairie, MN), resistance tem-
perature devices (RTD’s), and a programmable logic
controller (PLC, 5-15, Allen-Bradley Co., Inc. Milwau-
kee, WI) for control and data acquisition. The tube
viscometer (2.89 m long) is located in the straight tube
Nomenclature
D tube diameter, m
L length, m
L
e
entrance length, m
m fluid consistency coefficient, Pa s
n
m
Ã
suspension consistency coefficient, Pa s
n
n fluid flow behavior index
n
Ã
suspension flow behavior index
P pressure, Pa
Q volumetric flow rate, m
3
/s
Q
ws
volumetric flow rate without slip, m
3
/s
Q
m
measured volumetric flow rate, m
3
/s
r tube radius, m
Re Reynolds number
T temperature, °C
Greek characters
b slip coefficient, mPa
À1
s
À1
/ particle volume fraction
/
m
maximum particle volume fraction
_ c shear rate, s
À1
_ c
w
wall shear rate, s
À1
l
Ã
suspension apparent viscosity, Pa s
l
f
fluid apparent viscosity, Pa s
l
r
relative viscosity (l
Ã
=l
f
)
s shear stress, Pa
s
w
wall shear stress, Pa
Table 1
Physical properties of model food system
Property Value
Green peas
Average diameter (m) 0.0084 ±0.0025
Density (kg/m
3
) 1026 ±9.770
CMC solution (1.5% w/v)
Density (kg/m
3
) 1022 ±5.000
118 K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128
section of the system. The entrance length (L
e
) of 3.44 m
(L
e
> 100 I.D.) was used to avoid the entrance effect.
The entrance length was verified against the equation
given by Kays (1966).
L
e
D
¼
Re
20
ð1Þ
2.3. Experimental run
Water was pumped through the system until the
system reached the desired temperature. Then the test
food suspension was delivered to the positive displace-
ment pump. Once the flow and temperature stabilized,
the data of flow rate and pressure drop were collected
with the data acquisition system. The suspension was
not reused after heating due to the change in rheological
properties of CMC solution being heated at high tem-
perature. The experimental parameters are summarized
in Table 2.
2.4. Wall shear stress and wall shear rate calculations for
tube viscometry
The relationship between shear rate and shear stress
was obtained from the measurement of pressure gradi-
ent and volumetric flow rate based on the following
assumptions: (a) steady flow, (b) time independent fluid
properties, (c) fluid velocity has no radial or tangential
components, (d) no slip condition at pipe wall, (e)
incompressible fluid, and (f) isothermal flow. Wall shear
stress was calculated from:
s
w
¼
rDP
2L
ð2Þ
Shear rate was calculated using Rabinowitsch–Mooney
equation
_ c
w
¼
3Q
pr
3
þs
w
dðQ=pr
3
Þ
ds
w
ð3Þ
The derivative term in Eq. (3) was obtained by taking
the derivative of a function from curve fitting the plot of
Q
pr
3
vs. s
w
.
When a tube viscometer is used to study the flow
behavior of dense coarse suspension, wall effects would
be present. The slip at wall occurs when fluid, having
lower viscosity than the bulk suspension, forms a thin
lubricating layer at the wall of the tube. Since the Ra-
binowitsch-Mooney equation was derived based on the
assumption of no slip at wall, the effect of the slippage
must be corrected in order to use the equation to obtain
the true shear rate. The classical method for slip cor-
rection can be found in the literature (Mooney, 1931;
Jastrzebski, 1967; Steffe, 1996). The correction is done
by introducing a term to account for the added flow due
to wall slippage to the overall flow rate. This results in
the following equation:
Q
m
s
w
pr
3
¼
Q
ws
s
w
pr
3
þ
b
r
ð4Þ
Generally, the effective slip coefficient, b, is evaluated
using tubes of different radii. However, to determine the
slip coefficient by varying the tube radius would require
large amount of food materials in this study. Recog-
nizing the similar effect of varying tube radius and
varying particle concentration on the wall shear stress,
an alternative procedure based on the utilization of
variable particle concentration was used to determine
the slip coefficient. The procedure has been described by
Chakrabandhu (2000). The slip coefficient was then used
for flow rate correction. The corrected volumetric flow
rate was subsequently used to calculate the true wall
shear rate.
Table 2
Experimental parameters for study of rheological properties of coarse
food suspensions
Parameter Value
Carrier fluid temperature (°C) 85, 110, 135
Particle concentration (% v/v) 0, 15, 20, 25, 30
Volumetric flow rate (m
3
/s) 1.26·10
À4
–3.15 · 10
À4
Fig. 1. Schematic diagram of the experimental setup for the study of coarse suspension flow behavior at high temperature.
K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128 119
2.5. Flow behavior characterization of coarse food
suspensions
Once the true shear rate was calculated, flow behavior
parameters of the suspension were computed. As several
studies have reported that the flow behavior of suspen-
sions follows power law pattern, the power law para-
meters of the coarse suspension were investigated. The
suspension consistency coefficient (m
Ã
) and flow behav-
ior index (n
Ã
) were estimated by fitting the experimental
data in the power law model (Eq. (5)) using linear
regression (SAS, 1989).
ln s ¼ ln m
Ã
þn
Ã
ln _ c ð5Þ
The suspension apparent viscosity (l
Ã
) was then calcu-
lated from
l
Ã
¼ m
Ã
_ c
n
Ã
À1
ð6Þ
An analysis of variance (ANOVA) was performed for
the effects of particle concentration and suspension
temperature on the consistency coefficient and flow
behavior index of suspensions.
2.6. Mathematical representation of apparent viscosity for
coarse suspensions
Selected theoretical expressions from the literature
developed from various approaches, i.e. hydrodynamic
approach (Einstein, 1905; Ting & Luebbers, 1957), en-
ergy approach (Frankel & Acrivos, 1967) as well as
semi-empirical (Mooney, 1951; Thomas, 1965) expres-
sions were considered for the prediction of the apparent
viscosity of suspensions that contain high concentra-
tions of coarse particulates. A statistical test (v
2
-test)
was performed to determine the goodness of fit of the
mathematical expressions to the experimental results.
The v
2
value is expressed as:
v
2
¼

n
i¼1
ðY
measured
ÀY
predicted
Þ
2
Y
predicted
_ _
ð7Þ
The null hypothesis of no significant differences between
the predicted and experimental results is rejected if the
calculated v
2
value is larger than the tabulated critical
value (c) depending on the confidence level (a) and de-
gree of freedom (df). The expressions tested are listed
below.
Einstein’s equation
l
r
¼
l
Ã
l
f
¼ 1 þ2:5/ ð8Þ
Second-order expanded Einstein’s equation
l
r
¼ 1 þ2:5/ þ10:05/
2
ð9Þ
The second-order coefficient of 10.05 arrived after con-
sidering the hydrodynamic interaction of spheres and
the effect of double formation due to collisions (Thomas,
1965).Third-order expanded Einstein’s equation
l
r
¼ 1 þ2:5/ þ10:05/
2
þa/
3
ð10Þ
Thomas’s equation
l
r
¼ 1 þ2:5/ þ10:05/
2
þae
b/
ð11Þ
Instead of expanding the power series to the third order,
Thomas proposed that the fourth term which represents
the particle interaction to be an exponential term as
suggested by Eyring, Henderson, Stover, and Eyring
(1964). The empirical constants, a and b, in Eq. (11)
were given by Thomas as 0.00273 and 16.6, respectively.
Mooney’s equation:
log l
r
¼
2:5/
1 Àk/
ð12Þ
where k is an empirically determined parameter sug-
gested by Mooney to be in the range of 1.35–1.91. Since
k is a parameter related to the packing geometry, several
numerical values that represent theoretical packing
geometries along with some that have been presented in
the literature were used to investigate the fit of the
Mooney equation to the experimental data from this
study. The values and their significance are summarized
in Table 3. The value of k was also empirically deter-
mined to obtain the best fit to the experimental data.
Frankel–Acrivos’s equation
l
r
¼ b
ð/=/
m
Þ
1=3
1 Àð/=/
m
Þ
1=3
_ _
ð13Þ
The Frankel–Acrivos equation was derived based on the
assumption that the particle shell is spherical and b takes
the value of 9/8. Instead, if a cubical particle shell was
assumed, b would take the value of 3p=16. In the veri-
fication of their equation, Frankel and Acrivos used the
/
m
value of 0.625 which was obtained by Thomas (1965)
from available data in the literature. In this study, the
values of b and /
m
were varied (Table 4). The /
m
value
was also empirically determined to obtain the best fit to
the experimental data.
Table 3
Values of k tested for best fit of Mooney’s equation to experimental
data
k Value Significance
1.350 Lower bound for k suggested by Mooney
1.910 Upper bound for k suggested by Mooney
0.340 Maximum volume fraction for completely tetrahe-
dral packing
0.524 Maximum volume fraction for completely cubical
packing
0.625 Maximum volume fraction used by Thomas (1965)
and Frankel and Acrivos (1967)
120 K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128
Ting–Luebber’s equation
l
r
¼
/
m
/
m
À/
ð14Þ
In the same manner as the tests for other expressions
mentioned above, the values of /
m
in the Ting–Lueb-
ber’s equation was also varied (0.34 and 0.524) and an
empirical value of /
m
was also determined to obtain the
best fit for the equation.
2.7. Mathematical representation of power law para-
meters for coarse suspensions
Compared to the expressions for apparent viscosity
of suspension, smaller number of mathematical expres-
sions relating the power law parameters to the solid
fraction exist. In this study, the following equations
proposed by Jarzebski (1981) were tested for the fit to
experimental results.
Modified Frankel–Acrivos’s equation
m
Ã
m
¼
9
8
ð/=/
m
Þ
1=3
1 Àð/=/
m
Þ
1=3
_ _
n
ð15Þ
m
Ã
m
¼
3p
16
ð/=/
m
Þ
1=3
1 Àð/=/
m
Þ
1=3
_ _
n
ð16Þ
Modified Mooney’s equation:
ln
m
Ã
m
¼
2:5/
1 Àk/
_ _
n
ð17Þ
Empirical equations of simple forms listed in Table 5
were also fitted to the experimental data to describe the
dependence of power law parameters on the particle
volume fraction in the conditions covered in this study.
Due to the fact that the consistency coefficient and the
flow behavior index together represent the flow curve of
suspensions, it was considered that a better prediction of
consistency coefficient could be obtained if the descrip-
tive equation also contained the flow behavior index
term or vice versa. Also, most of the previously de-
scribed mathematical expressions, which have been
shown in the literature to describe well the l
Ã
and m
Ã
values, contain the particle fraction terms in the form of
the volume fraction-to-maximum volume fraction ratio
(/=/
m
) or those similar to the what given in the Einstein
equation. Therefore the following equations were also
investigated:
m
Ã
m
¼ a
/
/
m
_ _
b
n
Ã
n
_ _
c
ð18Þ
m
Ã
m
¼ að1 þ2:5/Þ
b
n
Ã
n
_ _
c
ð19Þ
3. Results and discussion
3.1. Effects of particle concentration and temperature on
power law parameters of coarse suspensions
From the known value of slip velocity, the corrected
volumetric flow rate was computed and used in calcu-
lation of the true shear rate according to the Rabi-
nowitsch–Mooney’s equation. The flow curves were
plotted for suspensions at 85, 110, and 135 °C in Figs. 2–
4. The suspensions exhibited power law behavior. Flow
behavior parameters of the suspensions were determined
according to the power law model. An analysis of var-
iance (ANOVA) was performed for the effects of particle
concentration and temperature on the consistency
coefficient (m
Ã
) and flow behavior index (n
Ã
) of suspen-
sions. The power law parameters of the suspensions are
summarized in Table 6.
Table 5
List of empirical equations of simple forms used to describe the
dependence of power law parameters on the particle volume fraction
Power law
parameter
described
Basis of empirical equation
Volume fraction to
maximum volume
fraction ratio
Einstein equation
Consistency
coefficient
m
Ã
m
¼ a
/
/
m
_ _
b
m
Ã
m
¼ að1 þ2:5/Þ
b
m
Ã
m
¼ a
/
/
m
_ _
b
n
c m
Ã
m
¼ að1 þ2:5/Þ
b
n
c
Flow Behavior
Index
n
Ã
n
¼ a
/
/
m
_ _
b
n
Ã
n
¼ að1 þ2:5/Þ
b
a, b and c are empirical constants.
Fig. 2. Shear stress–shear rate relationships for suspensions at 85 °C
and various particle concentrations.
Table 4
Constants tested for best fit of Frankel–Acrivos equation to experi-
mental data
Parameter Values
B 9/8, 3p=16
/
m
0.340, 0.524, 0.625
K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128 121
Particle concentration and temperature showed sig-
nificant effects on the consistency coefficient and the flow
behavior index (P < 0:0001). In general, the consistency
coefficient increased with particle concentration and
decreased with temperature as shown in Fig. 5. For all
experimental temperatures, as the particle concentration
was increased from 0% to 15%, the consistency coeffi-
cient increased significantly (a ¼ 0:05). However, the
suspension at 85 °C, which possessed highest carrier
fluid viscosity, exhibited a much higher rate of increase
in consistency coefficient than what was observed for the
suspensions at higher temperatures. As the particle
concentration level increased from 15% to 20% and
25%, although slight increase in consistency coefficient
was observed for suspensions at 110 and 135 °C, no
significant change in the consistency coefficient was
observed at 95% confidence level for all experimental
temperatures. As the particle concentration level in-
creased from 25% to 30%, the consistency coefficient
increased significantly (a ¼ 0:05) for suspensions at
85 °C and 110 °C whereas no significant change in the
consistency coefficient was observed at 135 °C. For all
particle concentrations, suspensions at 85 °C showed
significantly higher consistency coefficient than those at
higher temperatures. When comparing the suspensions
at 110 °C and 135 °C at the same level of particle con-
centration, however, the difference between the suspen-
Fig. 5. Effects of particle concentration and temperature on the con-
sistency coefficient of coarse suspension.
Table 6
Power law parameters for CMC solution with and without particulates
Temperature (°C) Particle concentration (%v/v) Consistency coefficient, m (Pa s
n
) Flow behavior index, n Shear rate, _ c (1/s)
85 0 5.1 ±0.47

0.510 ±017

159–374
15 26 ±4.1
b
0.30 ±0.016
b
44.6–125
20 23 ±4.0
b
0.35 ±0.038
b
35.7–115
25 26 ±1.7
b
0.32 ±0.023
c
69.1–248
30 39 ±8.7
c
0.25 ±0.050
d
33.0–118
110 0 1.6 ±0.21
d
0.58 ±0.020
e
115–372
15 2.1 ±0.62
e
0.58 ±0.082
e
73.4–205
20 1.8 ±0.09
e
0.62 ±0.008
e
114–190
25 3.0 ±0.35
e
0.56 ±0.019
f
71.8–126
30 7.9 ±0.88
f
0.36 ±0.017
g
151–247
135 0 0.35 ±0.085
g
0.75 ±0.047
h
125–209
15 1.4 ±0.11
e;h
0.55 ±0.017
e;i
108–266
20 2.0 ±0.71
e;h
0.51 ±0.081
e;i
99.1–263
25 1.5 ±0.29
h
0.59 ±0.056
f;i
82.8–158
30 1.5 ±0.10
h
0.56 ±0.014
i
155–229
*
Values having same superscripts (a–i) are not significantly different at a ¼ 0:05.
Fig. 4. Shear stress–shear rate relationships for suspensions at 135 °C
and various particle concentrations.
Fig. 3. Shear stress–shear rate relationships for suspensions at 110 °C
and various particle concentrations.
122 K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128
sions at these temperatures was not as discernable. Al-
though the suspensions at 110 °C generally had higher
consistency coefficient than those at 135 °C, as can be
seen at 25% and 30% particle concentration levels
(P < 0:01), the difference was insignificant at the particle
concentration levels of 15% and 20%.
The flow behavior indices indicate shear-thinning
behavior for all suspensions (n < 1). The effects of par-
ticle concentration and temperature on flow behavior
index of coarse suspensions were in opposite trends of
the effects on the consistency coefficients. That is, in
general, the flow behavior index decreased with an in-
crease in particle concentration and increased with
temperature as shown in Fig. 6. As for the effect of
temperature, even though the difference between the
flow behavior indices of the suspensions at 110 and 135
°C was not evident at the particle concentration levels of
20–25%, the difference at 30% particle concentration
level was very noticeable. As the particle concentration
was increased from 0% to 15% the flow behavior index
decreased significantly (a ¼ 0:05) for suspensions at 85
°C and 135 °C. At 110 °C, however, the flow behavior
index did not exhibit a significant difference as the par-
ticle concentration increased from 0% to 15%. When the
particle concentration was increased from 15% to 20%,
no significant changes in the flow behavior index was
observed at all temperature tested. As the particle con-
centration was increased from 20% to 25% and 30%, an
evidently decreasing trend (a ¼ 0:05) could be observed
for suspensions at 85 °C and 110 °C. The suspensions at
135 °C, however, did not exhibit any significant changes
as the particle concentration increased.
It should be noted that for mixtures containing high
concentration of coarse particulates (i.e., 15–30%), as
the viscosity of the carrier fluid increased, the influence
of particle concentration was more pronounced. Also, as
the particle concentration increased, the influence of the
carrier fluid viscosity was more pronounced. This indi-
cates a significant interaction (P < 0:005) between par-
ticle concentration and temperature (which reflects the
effect of carrier fluid viscosity) on the power law
parameters of coarse suspensions. Also, the effect of the
interaction between these variables was more observable
on the consistency coefficient than on the flow behavior
index.
The effect of temperature on the consistency coeffi-
cient of CMC solution containing no particulates (m)
can described satisfactorily by an Arrhenius-type equa-
tion
m ¼ Ae
Ea=RT
R
2
¼ 0:986 ð20Þ
where the pre-exponential constant (A) and the activa-
tion energy are 0.0000014 mPa s
n
and 66 kJ/(g mol),
respectively.
For the flow behavior index (n) of the CMC solution,
the effect of temperature, can be described well by an
exponential equation,
n ¼ 0:2138 expð0:0092TÞ R
2
¼ 0:996 ð21Þ
Exponential function have been used successfully to
express the effect of temperature on CMC solution in the
temperature range below 70 °C (Bhamidipati & Singh,
1990).
3.2. Mathematical expressions for apparent viscosity of
coarse suspensions
Variations of Einstein, Mooney, Frankel–Acrivos,
and Ting–Luebber expressions, as well as the expression
presented by Thomas (1965) were tested for the good-
ness of fit to the experimental data. The results are
summarized in Table 7. It was found that most expres-
sions tested adequately represented the apparent vis-
cosity for the entire range of particle fraction covered in
this study. The original Einstein equation (Eq. (8)) ex-
plained the suspension’s apparent viscosity quite well for
the entire range of /. As a matter of fact, the Einstein’s
equation was the best in explaining the suspension
apparent viscosity at the higher temperature range (110
and 135 °C), i.e., for the lower levels of fluid apparent
viscosity. Expanding the Einstein’s equation to second-
order (Eq. (9)) improved the prediction of suspension
apparent viscosity at the lower temperature level (85
°C), i.e., higher fluid apparent viscosity. However, it did
not appear to improve the prediction for the entire range
of particle fraction. When the Einstein’s equation was
expanded to the third-order, with the second order
coefficient of 10.05 and the best-fit third-order coefficient
(a ¼ 20:84), i.e.,
l
r
¼ 1 þ2:5/ þ10:05/
2
þ20:84/
3
ð22Þ
the prediction for the entire range of l
r
improved
markedly when compared to the prediction by the ori-
ginal Einstein’s equation. Thomas (1965) noted that
the fourth term in the expanded Einstein’s equation
Fig. 6. Effects of particle concentration and temperature on the flow
behavior index of coarse suspension.
K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128 123
Table 7
Summary of goodness of fit test values (v
2
) for apparent viscosity expressions of coarse suspensions
Equation Variation v
2
value
0:15 < / < 0:30 (df ¼37, c ¼ 24.08) 0:15 < / < 0:20 (df ¼17, c ¼8.68) 0:25 < / < 0:30 (df ¼19, c ¼10.12)
T (°C) T (°C) T (°C)
85 110 135 85 110 135 85 110 135
Einstein – 14.82 0.55
a
1.16
a
6.46 0.26
a
0.38 8.37 0.36
a
0.78
a
Expanded
(2nd order)
2.44 8.20 6.15 1.79 2.25 0.62 0.65 6.50 5.52
Expanded
(3rd order),
a ¼ 20:84
b
6.21 3.05 1.83 3.08 1.30 0.26 3.14 2.05 1.58
Mooney k ¼ 1:35 3.94 16.51 14.20 1.48 2.64 0.81 2.45 14.55 13.39
k ¼ 1:91 19.20 42.72 40.17 0.67 4.44 1.98 18.53 39.44 38.19
k ¼ 0:340 5.26 4.00 2.91 3.54 1.08 0.19 1.72 3.19 2.72
k ¼ 0:524 4.27 5.13 3.80 3.16 1.27 0.22 1.11 4.18 3.58
k ¼ 0:625 3.79 5.89 4.43 2.95 1.39 0.26 0.85 4.85 4.18
k ¼ 0:078
b
6.87 2.81 2.04 4.08 0.86 0.17 2.78 2.17 1.87
Thomas – 2.34
a
12.23 9.86 1.37 2.73 0.87 0.97 10.16 8.99
Frankel and
Acrivos
b ¼ 9=8,
/
m
¼ 0:340
288.15 346.16 352.86 25.96 46.08 41.46 262.19 310.53 311.41
b ¼ 9=8,
/
m
¼ 0:524
23.59 52.70 49.48 1.80 11.93 8.18 21.78 43.54 41.30
b ¼ 9=8,
/
m
¼ 0:625
7.73 28.07 24.84 0.40
a
6.33 3.41 7.33 23.28 21.43
b ¼ 3p=16,
/
m
¼ 0:524
9.31 7.46 7.44 8.85 0.53 0.76 0.46
a
7.04 6.68
b ¼ 3p=16,
/
m
¼ 0:625
19.08 2.61 5.27 14.87 1.08 3.18 4.21 1.53 2.09
b ¼ 9=8,
/
m
¼ 1:01
b
8.82 2.89 2.76 6.72 0.39 0.33 2.11 2.63 2.44
b ¼ 3p=16,
/
m
¼ 0:55
b
11.54 5.38 6.00 10.79 0.52 1.22 0.75 4.91 4.78
Ting and
Luebber
/
m
¼ 0:340 42.66 73.32 70.86 0.47 6.00 3.16 42.19 68.85 67.69
/
m
¼ 0:524 6.67 3.53 3.04 4.95 0.60 0.17
a
1.72 3.10 2.87
/
m
¼ 0:539
b
7.61 2.88 2.60 5.35 0.50 0.20 2.27 2.53 2.40
a
Smallest v
2
in the range of / and temperature.
b
Best fit value.
1
2
4
K
.
C
h
a
k
r
a
b
a
n
d
h
u
,
R
.
K
.
S
i
n
g
h
/
J
o
u
r
n
a
l
o
f
F
o
o
d
E
n
g
i
n
e
e
r
i
n
g
6
6
(
2
0
0
5
)
1
1
7

1
2
8
represents the existence of particle interaction. The
improvement in prediction brought by including the
third order term to the expanded Einstein’s equation
confirms the significance of particle interaction. Essen-
tially, this equation provided the closest estimate con-
sidering the entire range of experimental variables in this
study (v
2
¼ 11:10 for all conditions combined).
The Thomas’s equation (a ¼ 0:00273, b ¼ 16:6), an-
other modification of the Einstein’s equation, suffi-
ciently represented the suspension’s apparent viscosity
in the lower range of particle fraction (0:15 < / < 0:20).
However, it did not represent the apparent viscosity at
all temperature levels in the higher particle fraction
range (0:25 < / < 0:30).
It has been reported in the literature (Dabak & Yucel,
1987) that prediction for suspension’s viscosity im-
proved when the maximum volume fraction was incor-
porated in the mathematical expressions. As for the
Mooney’s equation, it was found that using maximum
volume fraction as the k value, indeed, provided better
predictions than assuming k in the range proposed by
Mooney. The prediction improved as the k value de-
creased from 0.625 to 0.340. The best-fit k value
(k ¼ 0:078) for the experimental data, however, was
lower than maximum volume fraction reported in the
literature. This best-fit Mooney equation provided the
second closest estimation considering the entire range of
experimental variables in this study (v
2
¼ 11:71 for all
conditions combined).
It should also be noted that when the k values pro-
posed by Mooney were used, the equation adequately
represented the suspension’s viscosity in the particle
fraction range of 0.15–0.20 but failed to represent the
suspension’s viscosity for the higher particle fraction
range (0:25 < / < 0:30). This predictive behavior is
similar to that of the Thomas’s equation. The observa-
tion that the Thomas’s and Mooney’s equations exhibit
similar predictive behavior has been reported previously
by Metzner (1985).
The Frankel–Acrivos’s equation is one of the
expressions containing the maximum volume fraction
that has been reported to provide a good estimation for
apparent viscosity for rigid solid in polymer melts at
high solids concentration levels (Metzner, 1985). Using
maximum volume fraction values frequently cited in the
literature, it was found that the equation adequately
represented the entire range of experimental results only
when the particle shell was assumed cubical (b ¼ 3p=16)
instead of spherical (b ¼ 9=8) often assumed in the lit-
erature. When the particle shell was assumed cubical, a
slightly better prediction was obtained when the /
m
for
cubic packing was used, compared to the /
m
value
proposed by Thomas (1965). The best-fit /
m
for the
equation took the value of 0.549 which is very close to
the maximum volume fraction for the cubical packing
pattern (/
m
¼ 0:524).
When spherical particle shell was assumed, using the
maximum volume fraction presented by Thomas
(/
m
¼ 0:625), the Frankel–Acrivos’s equation provided
good prediction in the lower range of volume fraction
(0:15 < / < 0:20) but over predicted the relative
apparent viscosity in the higher volume fraction range
(0:25 < / < 0:30). When considered the best-fit /
m
va-
lue in the Frankel–Acrivos’s equation, with spherical
particle shell assumption, the /
m
took the value of 1.
Although it is physically impossible for the maximum
particle packing volume fraction to be unity in the
experiment, this finding indicated that the experimental
results for this study could also be expressed by a re-
duced form of the theoretical Frankel–Acrivos’s equa-
tion
l
r
¼
9
8
/
1=3
1 À/
1=3
_ _
ð23Þ
The Ting–Luebber’s equation (Eq. (14)) is also one of
the equations that includes the maximum volume frac-
tion. It was found that assuming tetrahedral particle
packing pattern (/
m
¼ 0:340), the equation provided
satisfactory estimations for the experimental results in
the particle fraction range of 0.15–0.20. However, sim-
ilar to the Thomas’s equation and the Mooney’s equa-
tion (1:35 < k < 1:91), the Ting and Luebber’s equation
failed to represent the suspension viscosity for the higher
particle fraction range (0:25 < / < 0:30). On the other
hand, when cubical packing was assumed (/
m
¼ 0:524),
the equation satisfactorily depicted the experimental
results in the entire range of experimental variables.
Similar to the findings of the Frankel-Acrivos’s equa-
tion, the best-fit maximum volume fraction value for the
Ting–Luebber’s equation (/
m
¼ 0:539) was also close to
that of the cubic packing. This strengthens the impli-
cation that the particle packing pattern in this study may
be approximated as a cubic packing. Fig. 7 illustrates
the fit of some mathematical expressions to the experi-
mental apparent viscosity data.
3.3. Mathematical expressions for power law parameters
of coarse suspensions
In the equations being statistically tested, the /
m
of
0.524 was used following the results from the apparent
viscosity expression section that cubic packing pattern
was a close approximation for the packing pattern in
this study. Goodness of fit test (v
2
) was performed for
the Jarzebski’s modification to the Frankel–Acrivos’s
and Mooney’s equations (Jarzebski, 1981) for suspen-
sion’s power law parameters (Eqs. (15)–(17)). It was
found that the equations did not satisfactorily describe
the experimental data (v
2
> 4:575, 95% confidence).
Linear regression procedure was used to fit the experi-
mental data to empirical equations listed in Table 5. The
K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128 125
regression coefficients indicated that the equations that
contain only the volume fraction terms could not satis-
factorily predict the consistency coefficient or the flow
behavior index of the suspensions (0:22 < R
2
< 0:47).
Including the flow behavior index of the carrier fluid in
the equations for consistency coefficient improved the
prediction only slightly (0:30 < R
2
< 0:48).
The findings from the goodness of fit test on the
equations proposed by Jarzebski, and the regression
tests on the empirical equations clearly indicate that the
assumption made by Jarzebski that the addition of
particles to the carrier fluid affects only the consistency
coefficient of the suspension does not apply to the
experimental data in this study due to the high particle
concentration. Considering the experimental data, it can
be seen that the addition of particles to the carrier fluid
significantly modified the consistency coefficient as well
as the flow behavior index. Also, the consistency coef-
ficient and the flow behavior index are related due to the
fact that they are obtained from the same regression
equation. They together are representatives of the flow
curves. Thus attempts to fit each parameter separately
without considering the close relation between the two
parameters may be the reason why all the correlations
did not depict the experimental results. In fact, a ques-
tion has been raised in the literature about analyzing
each of the power law parameters separately (Scott
Blair, 1971). According to this observation, linear
regression procedure was used to fit the experimental
data to empirical equations (Eqs. (18) and (19)). Each
equation contains both the consistency coefficient and
the flow behavior index of the suspension. The following
equations were obtained.
m
Ã
m
¼ 2:20
/
/
m
_ _
0:29
n
Ã
n
_ _
À2:17
R
2
¼ 0:87 ð24Þ
m
Ã
m
¼ 1:05ð1 þ2:5/Þ
1:10
n
Ã
n
_ _
À2:14
R
2
¼ 0:90 ð25Þ
The prediction of the consistency coefficient was mark-
edly improved when the flow behavior index of the
suspension was included in the correlation. This suggests
that for concentrated coarse suspensions subjected to
conditions presented in this study, the relationship be-
tween m
Ã
and n
Ã
is of importance and should be con-
sidered in order to achieve a better m
Ã
estimation. The fit
of Eqs. (24) and (25) to the experimental data is shown
in Fig. 8. Even though both Eqs. (24) and (25) estimated
the experimental results well, Eq. (25) provided a slightly
better prediction according to the regression coefficient.
Considering the constants in both equations, it can be
Fig. 8. Plot of experimental and predicted relative consistency coeffi-
cient (m
Ã
=m
f
) for coarse suspension [(}) Eq. (24); (·) Eq. (25); (––)
Experimental].
Fig. 7. Plots of experimental and predicted values of relative apparent
viscosity for suspension at different temperatures. [(}) Eq. (8); (M)
Eq. (10); (Ã) Eq. (12) (k ¼ 0:078), (·) Eq. (13) (b ¼ 3p=16, /
m
¼ 0:549),
(
Ã
) Eq. (13) (b ¼ 9=8, /
m
¼ 1:01), (–) Eq. (13) (b ¼ 9=8, /
m
¼ 0:625);
(s) Eq. (14); (–+–) Experimental].
126 K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128
seen that the exponentials of the relative flow behavior
index in both Eqs. (24) and (25) are very close to each
other in magnitude. Another interesting observation is
that in Eq. (25), which contains the familiar terms from
the Einstein’s equation, the right hand side of the
equation is approximately the right hand side of the
Einstein’s equation multiplied by the square of the in-
verse of the relative flow behavior index of the suspen-
sion, i.e.,
m
Ã
m
¼ ð1 þ2:5/Þ
n
Ã
n
_ _
À2
ð26Þ
This striking similarity implicate that while the Einstein
equation, which was originally derived via the hydro-
dynamic approach to represent the apparent viscosity of
suspension with Newtonian carrier fluid, it may be slight
modified with the information of the flow behavior
index of carrier fluid and suspension, such that it pro-
vides a satisfactory representation for the consistency
coefficient of suspensions whose carrier fluids are non-
Newtonian.
While Eqs. (24)–(26) provide an important implica-
tion that for concentrated coarse food suspension sub-
jected to such conditions used in this study, the
consistency coefficient and the flow behavior index are
strongly related, in general practice, they alone do not
offer convenience for prediction of the parameters since
both the suspension consistency coefficient and the flow
behavior index are unknown and need to be predicted in
most cases. It appears to be more practical if the power
law parameters were computed using these equations
along with the information on the suspension apparent
viscosity of which several expressions, particularly the
third-order Einstein equation, were found to provide
satisfactory estimation.
4. Conclusions
Knowledge of flow behavior of food suspensions is
critical for many food processes. For those involving
tube flow at high temperatures such as aseptic process-
ing of particulate foods, such essential information is
scarce, particularly for concentrated coarse suspensions
subjected to aseptic processing conditions, which include
ultra-high temperature and high pressure. The shortage
in the much-needed information is due partly to the lack
of viscometers applicable to such material and process-
ing conditions. With the goal to alleviate the situation,
the study described herein was carried out. A tube vis-
cometer, designed with care to meet all theoretical
requirements, was used to characterize the flow behavior
of a model food suspension consisting of a non-New-
tonian carrier fluid and high concentration of large
particles at high temperature. An approach based on the
Rabinowitch–Mooney equations (Chakrabandhu, 2000)
was used to correct slippage at tube wall, allowing the
subsequent computation of wall shear rate and power
law parameters of the suspensions. It was found that the
consistency coefficient of the suspension (m
Ã
) increased
with particle concentration and decreased with temper-
ature. The suspension flow behavior index (n
Ã
) indicated
shear-thinning behavior for the entire range of study. In
general, the n
Ã
decreased with particle concentration and
increased with temperature. Significant effect of inter-
action between particle concentration and temperature
(and thus carrier fluid viscosity) on power law parame-
ters of coarse suspension was observed, with the effect
more evident on m
Ã
than n
Ã
.
To obtain mathematical representations and more
insight to the flow behavior of such concentrated coarse
food suspensions, apparent viscosity and power law
parameters estimated using theoretical, semi-empirical,
and empirical expressions were compared to the exper-
imental values. In general, v
2
test showed that most of
the classical theoretical, semi-empirical, and empirical
equations tested offered good overall estimations of
suspension apparent viscosity. Of these equations, those
derived based on the hydrodynamic approach provided
more satisfactory estimations, with the third order
expansion of Einstein equation offering the best esti-
mates when compared to the experimental values for the
entire range of particle fraction and temperature used in
this study.
For the mathematical representations of power law
parameters of the coarse suspensions, it was found that
predictive expressions describing m
Ã
only as a function
of particle fraction and flow behavior index of carrier
fluid (n) did not satisfactorily describe the experimental
data. Empirical expressions showed that a successful
estimation of the consistency coefficient of coarse sus-
pension required the knowledge of the flow behavior
index of suspension. These findings imply that, for
concentrated coarse suspensions in tube flow at high
temperature as presented in this study, the relationship
between m
Ã
and n
Ã
is of importance and should be
considered in order to achieve a better m
Ã
estimation.
Among the equations obtained from literature as well as
those developed in this study for estimation of m
Ã
an
empirical equation describing m
Ã
as a function of par-
ticle concentration and relative suspension flow behav-
ior index (n
Ã
=n), which was developed in this study,
provided the best estimates for m
Ã
This equation con-
tains a striking similarity to the Einstein equation and
suggests that slight modification of the Einstein equa-
tion, which was originally derived via the hydrodynamic
approach to represent the apparent viscosity of sus-
pensions with Newtonian carrier fluids, could provide a
satisfactory estimation for the consistency coefficient of
suspensions whose carrier fluids are non-Newtonian,
at least for the conditions presented in this study. The
K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128 127
results presented here, including mathematical repre-
sentations of suspension flow behavior parameters,
provide a much-needed insight toward flow behavior of
concentrated coarse food suspensions in tube flow at
high temperature, information of which is vital for
various food processes.
Acknowledgements
The authors acknowledge the generosity of TIC
Gums Inc. (Belcamp, MD) for supplying carboxy-
methylcellulose used in the research.
References
Awuah, G. B., Ramaswamy, H. S., Simpson, B. K., & Smith, J. P.
(1996). Fluid-to-particle convective heat transfer coefficient as
evaluated in an aseptic processing holding tube simulator. Journal
of Food Process Engineering, 19, 241–267.
Bhamidipati, S., & Singh, R. K. (1990). Flow behavior of tomato sauce
with or without particulates in tube flow. Journal of Food Process
Engineering, 12, 275–293.
Chakrabandhu, K. (2000). Aseptic processing of particulate foods:
computational models and experimental studies. Phd thesis.
Purdue University, W. Lafayette, IN, USA.
Dabak, T., Yucel, O. (1987). Prediction of viscosity and characteriza-
tion of particle size distribution of highly concentrated slurries. In
Proceedings of the 11th international conference on slurry tech-
nology, March 16–18, 1986, Hilton Head, South Carolina.
Einstein, A. (1905). Eine neue bestimmun der molekul-dimnesionen.
Ann. Physik 19, 289–306, 34, 591–597. [English translation in
Einstein, A. (1956). Investigations on the theory of Brownian
movements. Trans. A.D Cowper. (Ed). with Notes R. Furth. Dover
Publications].
Eyring, H., Henderson, D., Stover, B. J., & Eyring, E. M. (1964).
Statistical Mechanics and Dynamics. New York: Wiley and Sons.
Frankel, N. A., & Acrivos, A. (1967). On the viscosity of a
concentrated suspension of solid spheres. Chemical Engineering
Science, 22, 847–853.
Holdsworth, S. D. (1971). Applicability of rheological models to the
interpretation of flow and processing behaviour of fluid food
products. Journal of Texture Studies, 2, 393–418.
Jarzebski, G. J. (1981). On the effective viscosity of pseudoplastic
suspensions. Rheologica Acta, 20, 281–287.
Jastrzebski, Z. D. (1967). Entrance effects and wall effects in an
extrusion rheometer during the flow of concentrated suspensions.
Industrial and Engineering Chemistry Fundamentals, 6, 445–454.
Kays, W. M. (1966). Convective heat and mass transfer. New York:
McGraw-Hill.
Martinez-Padilla, L. P., Cornejo-Romero, L., Cruz-Cruz, C. M.,
Jaquez-Huacuja, C. C., & Barbosa-Canovas, G. V. (1999). Rhe-
ological characterization of a model food suspension containing
discs using three different geometries. Journal of Food Process
Engineering, 22, 55–79.
Metzner, A. B. (1985). Rheology of suspensions in polymeric liquids.
Journal of Rheology, 29, 739–775.
Mooney, M. (1931). Explicit formulas for slip and fluidity. Journal of
Rheology, 2(2), 210–222.
Mooney, M. (1951). The viscosity of a concentrated suspension of
spherical particles. Journal of Colloid Science, 6, 162–170.
Pordesimo, L. O., Zuritz, C. A., & Sharma, M. G. (1994). Flow
behavior of coarse solid–liquid food mixtures. Journal of Food
Engineering, 21, 495–511.
Ramaswamy, H. S., Awuah, G. B., & Simpson, B. K. (1996). Influence
of particle characteristics on fluid-to-particle heat transfer coeffi-
cient in a pilot scale holding tube simulator. Food Research
International, 29, 291–300.
Sandeep, K. P., & Zuritz, C. A. (1994). Residence time distribution of
multiple particles in non-Newtonian holding tube flow: statistical
analysis. Journal of Food Science, 59, 1314–1317.
SAS (1989). SAS/STAT User’s Guide (6.03). New York: SAS Institute,
Inc. Cary.
Scott Blair, G. W. (1971). On power law equations relating stress to
shear rate in rheology. Journal of Texture Studies, 3, 431–
436.
Steffe, J. F. (1996). Rheological methods in food process engineering
(2nd Ed.). MI: Freeman Press.
Thomas, D. G. (1965). Transport characteristics of suspensions. Part
VIII. A note on the viscosity of Newtonian suspensions of uniform
spherical particles. Journal of Colloid Science, 20, 267–277.
Ting, A. P., & Luebbers, R. H. (1957). Viscosity of suspensions of
spherical and other isodimensional particles in liquids. AIChE
Journal, 3, 111–116.
128 K. Chakrabandhu, R.K. Singh / Journal of Food Engineering 66 (2005) 117–128