Food Research International 26 (1993) 109-l 13

A capillary viscometer for characterization of

fluid foods

J. P. White, V. J. Davidson & L. Otten

School of Engineering, University of Guelph, Guelph. Ontario, Canada, NlG 2 WI

A multi-section capillary viscometer was designed and tested for measuring the
rheological properties of food materials (e.g. power-law parameters, thixo-
tropy). The apparatus included a data acquisition system to collect flow rate,
temperature and differential pressure measurements as inputs to a BASIC algo-
rithm which computed estimates of the power-law parameters in real time. The
apparatus was tested for steady-state and dynamic measurements, and an anal-
ysis of the effects of measurement errors on the parameter estimates was com-
pleted. Rheological properties of representative food materials (carboxymethyl
cellulose solutions, pregelatinized starch, locust bean gum) as well as systems
with enzymes hydrolysing complex polysaccharides were characterised.

Keywords: capillary viscometer, power-law parameters, non-Newtonian fluids

INTRODUCTION Newtonian and can be characterised by a power-

law model (Eqn (1))
Rheological properties are commonly measured in
rw = k k” (1)
food processing either as direct indications of
product quality or as indirect measures of other where TV= wall shear stress (Pa), yw = wall shear
properties (e.g. viscosity as an indication of total rate (s-l), k = consistency index (Pas”), and n =
solids, changes in molecular size). Since many power-law index then the pressure drop (AP)
food materials behave as non-Newtonian fluids, along the capillary tube is:

the rheological properties must be measured over
a wide range of shear rates. These data are also
required for design calculations in any process
that involves fluid flow (e.g. pump sizing, filtra-
tion, extrusion) and are critical for the analysis of
flow conditions in pasteurization and aseptic pro-
cessing to ensure adequate thermal treatment.
Capillary viscometers are commonly used in
laboratories and on-line process equipment to
measure rheological properties (Dealy, 1984). The
fundamental design is relatively simple and is
based on the theory of laminar flow in a straight
tube. The shear stress on the fluid at the wall is
related to the pressure drop along the length of
the tube and to the geometry. If the fluid is non-
Food Research International 0963-9969/93/$06.00
0 1993 Canadian Institute of Food Science and Technology
1 (2)
Ap =4kL 8Q(3n+I) n
D nnD3
where Q = volumetric flow rate (length3 ssl), D =
diameter of capillary section (length), L = length
between pressure taps (length).
Since the viscosity of many food and bio-
materials is dependent on the shear rate, several
researchers have used variations in the capillary
design to measure apparent viscosity or to esti-
mate the power-law parameters, n and k (Antonini
et al., 1989; Corrieu et al., 1982, 1983; Windhab &
Kuhlmann, 1989). In order to use the relationship
in Equation (2) to estimate the power-law parame-
ters, pressure drops must be measured at a mini-
mum of two different shear rates. Different shear
rates can be achieved by varying the flow rate for
a single capillary section or very simply by using

109
110 J. P. White, V. J, Davidson. L. Otten
several capillary sections of different diameters in
series. As long as the appropriate lengths are used
at transitions in diameter, it can be assumed that
the flow is fully developed between the pressure
taps. The series design allows the power-law
parameters to be measured in a single pass of fluid.
Davidson et al. (1990) fabricated a multi-section
capillary viscometer and tested it using
glycerol/water solutions (Newtonian behaviour)
and a 1% carboxymethyl cellulose solution
(pseudo-plastic behaviour) for steady-state flow
conditions. They found that the rheological data
obtained with the capillary viscometer were in
close agreement with the results obtained with a
Haake RV3 rheometer. The objective of this paper
is to present the results obtained when the vis-
cometer was used to measure real-time values of
power-law parameters for two bio-polymer solu-
tions, carboxymethyl cellulose and pregelatinized
starch. Both steady-state and transient flow condi-
tions were considered, and the effects of measure-
ment errors on the parameter estimates were
analysed. Hydrolytic enzymes were added to the
carboxymethyl cellulose and starch solutions in
order to determine the feasibility of using the vis-
cometer to monitor rheological changes in react-
ing systems.
MATERIALS AND METHODS
Apparatus
A schematic of the experimental apparatus with
three capillary sections of different diameters is
shown in Figure 1. To extend the range of mea-
surement, the capillary sections were interchange-
able. In these experiments, the capillary diameters
I Pmcaumsumra
Fig. 1. Schematic of multi-section capillary viscometer with
piping for recycle or single-pass flow.
varied from 2.00 to 6.35 mm. Differential pressure
transducers (MPXlODP, Motorola Inc., Phoenix,
AZ) were used to measure the pressure drop over
a known length (150-200 mm) between two taps
in each capillary section. Each differential pressure
transducer has two ports which were connected to
the taps by short lengths of flexible tubing. The
output of the transducers was linear over the nor-
mal operating range (100-10000 Pa) with a sensi-
tivity of 33 Pa mV’. The transducers were
calibrated before each experiment by measuring
the output voltage at known values of static head
with one side of the transducer open to atmo-
spheric pressure.
Flow was maintained in the apparatus by
means of a small gear pump (MICROPUMP@’
Model 1800-00, Cole Parmer Instrument Com-
pany, Chicago, IL). Although a constant input
voltage was maintained to power the pump during
each experiment, the flow rate changed as the vis-
cosity of the working fluid changed. Therefore, a
circuit was built to measure the number of pump
revolutions (directly proportional to flow rate) by
counting the voltage spikes on the pump input
voltage and converting this signal to a DC voltage
which was used as a measure of flow rate (White,
1991). A flow rate calibration was performed be-
fore each experiment and periodically the calibra-
tion was checked at the end of the experiment.
Using regression analysis, it was estimated that
the maximum error in flow rate measurement in
these experiments was 15%.
The apparatus also included an automatic data
acquisition system to monitor differential pres-
sures, flow rate and temperature in real-time. The
data acquisition system consisted of a multi-
plexer/screw terminal board (EXP-16 Metrabyte,
Taunton, MA) which provided the signal condi-
tioning for the analog voltage inputs. This was
I
connected to an analog-to-digital board (DAS-8
Metrabyte) in an Industry Standard Architecture
(ISA) computer (Intel 80386 microprocessor, 20
MHz). All of the software for data collection, cali-
bration and power-law parameter calculation was
written in Microsoft QuickBasic 4.5. Newton’s
method for simultaneous non-linear equations
(Hildebrand, 1976) was used to estimate the
power-law parameters, n and k, from the differen-
tial pressure and flow rate measurements.
A vibrational viscometer (Viscoliner Model 1810,
Nametre Co., Metuchen, NJ) was available for
some of the experimental studies. This viscometer
was designed for on-line measurements and had a
 A capillary viscometer for characterization of fluid foods 111

control console equipped with a temperature For example, the effects of relative errors in Q and
probe. Data were collected directly from the con- AP on the estimate of k were calculated as:
trol console through an RS 232 port.
Ak = [(g AQ~ + (3)
Methods
where:
In the present study, two fluid systems were used:

1
aqueous solutions of carboxymethyl cellulose Jk -nDAP 8(3n + 1) -n -in + I) = 2
(CMC) (Food Specialties, Halton Hills, ON) and
-=
aQ 4L ?7?lD3
Q (4)

pregelatinized starch (Nacan Products Ltd., ak k

Brampton, ON). The CMC solutions were pre-
pared by adding carboxymethyl cellulose powder
to water and heating to 80°C with stirring until
the powder was completely dissolved. The solu-
tions were cooled to room temperature and al-
lowed to sit for 12 h to allow entrained air
bubbles to escape. The pregelatinized starch was
dissolved into water and also allowed to stand
undisturbed for 12 h. The hydrolytic enzymes
used in the studies of reacting systems were cellu-
lase (CMC solutions) and P-amylase (pregela-
tinized starch solutions) (Sigma Chemical Inc., St.
Louis, MO). The dynamics of the response were
tested by making step changes in the concentra-
-=-
(5)
JAP AP
The relative error in k was evaluated over a range
of typical operating conditions and for estimates
of the maximum measurement errors in flow rate
(&15%) and pressure difference (+5%). The maxi-
mum relative error in k was a function of n and
for Newtonian fluids (n = 1) was estimated to be
*15.8%. For pseudoplastic fluids, the relative error
in k was lower (e.g. at n = 0.5, Ak/k = 9.0%).
A similar analysis was done to estimate the rela-
tive error in n resulting from measurement errors
in Q and AP. The partial derivatives of n with re-
spect to Q and AP are:

(6)
tion of CMC solutions being pumped through the
_=_ m’
viscometer. These step changes were achieved
using two fluid reservoirs connected to the system
via a two-way valve, so that the reservoirs could
-1
be interchanged very quickly. The extent of hy-
drolysis was measured using the reducing-sugar (7)
method detailed by Schellhorn & Forsberg (1984).
The estimate of n was less sensitive to measure-
ment errors in Q and AP. For typical operating
conditions, the maximum relative error in n was
RESULTS AND DISCUSSION calculated to be less than 4%.

Error analysis Steady-state and transient measurements

The experimental errors in the flow rate (Q) and
pressure differences (AP) measurements intro-
duced uncertainty in the estimates of the param-
eters y1and k. The magnitude of the uncertainty
was estimated by evaluating the partial derivatives
with respect to flow rate and pressure difference.
Estimates of n and k obtained from steady-state ex-
periments are summarized in Table 1. These pa-
rameters were estimated over a range of shear rates
from 100 s’ to 5000 s’. The dynamic response of
the capillary viscometer was also tested using step
changes in the CMC solutions (Figs 2 and 3).

Table 1. Estimates of Power-law parameters obtained for solutions of carboxymethyl cellulose (CMC), pregelatinized starch and
locust bean gum

Solution Temperature Flow behaviour index, n Consistency index, k

(“c> (Pas”)

0.75% CMC 24 0.75 0.38

1.O% CMC 24 0.75 0.58
4.0% Pregelatinized starch 23 0.57 0.80
0.25% Locust bean gum 75 0.8 0.023
 J. P. White, V. J. Davidson, L. Otten

apparent viscosity measurements. Figure 4 shows

the changes in rheological characteristics of a
1.O% CMC solution as it was hydrolysed by cellu-
lase. These estimates of n and k were obtained

r r
800 1200
600 1000
Th*(==m
Fig. 2. Response in estimate of n to a step change in car-
boxymethyl cellulose concentration.
When the CMC concentration was increased from
0.75% to 1.0% (indicated on figures as initial
source change), the system response was in the
order of 1 min (i.e. time to reach steady-state esti-
mates). However, the response to a step decrease
in concentration was somewhat slower, largely
due to a transient in the flow rate. The 1.0% solu-
tion was extremely viscous (apparent viscosity =
71 mPas at 1000 s-l) and the pump could not
maintain a constant flow rate.
Reacting systems
To investigate the potential of the capillary viscom-
eter to characterise changes in reacting systems, sev-
eral experiments were conducted using hydrolytic
enzymes. The purpose of this work was not to con-
sider a specific food process, but to test the system
over a range of rheological characteristics and to
compare changes in the power law parameters to
0.9 I
From 0.75 to 1 .O wt96 CMC
using the 4.70 and 6.35 mm capillary sections.
However, these values were in close agreement with
the calculations of n and k based on the pressure
differences from the 6.35 and 9.53 mm sections
(steady-state estimates within 4% for n and within
10% for k). This agreement was also an indication
that wall slip was not a factor in these experiments.
i
The analysis of the number of reducing end
1400
groups indicated that the rate of hydrolysis was
constant during the experiment. However, not sur-
prisingly, this did not result in simple changes in
the rheological characteristics. Cellulase is a
generic term for a group of endo- and exo-glu-
conases as well as P-glucosidase, and each of these
enzymes attacks the molecule in a different man-
ner. In the first 10 min, the action of the endo-
gluconase had a substantial effect on solution
viscosity and the effect was clearly evident in the
change in the parameter k. In the second half-
hour of the reaction, the parameter k was not a
useful measure of the extent of reaction because
the uncertainty in the estimate was +15%. During
most of this time, the parameter n was a sensitive
indicator of the extent of hydrolysis. The Nametre
viscometer was not sensitive to the changes in
molecular structure until 10 min into the reaction
because the effective shear rate for this instrument
was high relative to the wall shear rates for any of
the capillary sections. In Figure 5, the apparent
viscosity measured by the Viscoliner was com-
pared to the apparent viscosity calculated at the
wall shear rate in the smallest capillary section (?,
= 500-600 s-l). The high effective shear rate at the
Viscoliner surface reduced the apparent viscosity
of the solution and there had to be more substan-

-I
0 400 600 1200
200 600 1000 1400
~mmmw

Fig. 3. Response in estimate of k to a step change in car- Fig. 4. Changes in n and k estimates during hydrolysis of
boxymethyl cellulose concentration (CMC). 1.0 wt% carboxymethyl cellulose.

80
70 -
--v*oo*ry
60
5 50-
14,_ vlaoalhr~lul
5 30
20 -
10
04
0 500 1000 1500 2000
Time
Fig. 5. Changes in the calculated apparent viscosity at the
wall shear stress in 4.70 mm diameter section and the apparent
viscosity measured by the Viscoliner during the hydrolysis of
1.0 wt% carboxymethyl cellulose.
tial changes in molecular structure before the sig-
nal from the Nametre probe changed substantially.
A pregelatinized starch solution (4 wt.%) was
hydrolysed at ambient temperature (23 + 1“C) with
P-amylase, and the change in rheological proper-
ties was followed in the capillary viscometer as
well as in the Nametre Viscoliner (White, 1991).
As was observed for the CMC hydrolysis, the con-
sistency index k was the most sensitive indicator
of enzyme action in the initial stages of the hy-
drolysis. The power-law index n changed linearly
from an initial value of 0.57 to 0.95 as the hydrol-
ysis proceeded.
Some preliminary studies were also conducted
to investigate the hydration kinetics of polysac-
charides that are commonly used as stabilizers in
dairy fluids (e.g. locust bean gum, carboxymethyl
cellulose). The capillary apparatus was useful for
determining the effects of temperature and shear
rate on the solution viscosity. The results for a
0.25% solution of locust bean gum, a typical level
for an ice cream mix, are also included in Table 1.
The solution showed non-Newtonian behaviour
and, based on the experimental values for the
power law parameters, the estimated apparent vis-
cosity at 500 ss’ and 75°C was 7 cP. This charac-
terization is being used to evaluate the potential
for laminar flow for viscous dairy products in the
holding tubes of HTST (high temperature, short
time) pasteurizers (Goff & Davidson, 1992).
CONCLUSIONS
The multi-section capillary viscometer was proven
as a versatile laboratory instrument for rheological
measurements. It was used to determine the power-
law characteristics of non-Newtonian food solu-
tions over a range of steady-state shear rates. The
A capillary viscometer for characterization of fluid foods 113
I
data acquisition system made it possible to calcu-
late changes in the power-law parameters in real-
time and to follow reacting food systems. Since the
capillary sections are interchangeable, it would
also be useful for studying other rheological prop
erties such as thixotropy. The capillary sections
could be sized so that two, equal diameter sections
1 were separated by a section of a different diameter.
2500 3000 3500 4000 4500 5000
(seconds) This apparatus has substantial potential for the
design of control strategies in a range of food pro-
cesses. Since the changes in properties can be mea-
sured during physical (heating, cooling, mixing,
etc) and chemical processes, preliminary testing
with the capillary viscometer would be useful for
the design engineer to determine the rheological
parameters that can be used as control variables.
In cases where a single measurement of apparent
viscosity is ultimately used on-line in a process,
the capillary apparatus could be used to determine
the appropriate design parameters for the on-line
instrument.
REFERENCES
Antonini, G., Francois, 0. & Duverd, G. (1988). La carac-
terisation en continu de la rheologie d’ecoulement des
fluides complexes (non-Newtoniens): le rheometre capillaire
en ligne. In Automatic Control and Optimisation of Food
Processes, ed. M. Renard & J. J. Bimbenet, Elsevier Ap-
plied Science, London, pp. 3147.
Corrieu, G., Lalande, M. & Fissette, A. (1982). Correlation be-
tween the dry matter content of fat-free cottage cheese and its
on-line measured apparent viscosity. Sci. Aliments, 2, 41-54.
Corrieu, G., Maingonnant, J. F. & Ferret, R. (1983). Un vi-
cosimbtre industriel pour la caracterisation en continu des
produits alimentaires de viscosite Clevee. Znd. Aliment.
Agric., 100, 23540.
Davidson, V. J., White, J. P. & Hayward, G. L. (1990). On-
line viscosity measurement in food systems. In Engineering
and Food, Vol. 1, ed. W. E. L. Spiess & H. Schubert. El-
sevier Applied Science, London, pp. 752-60.
Dealy, J. M. (1984). Viscometers for on-line measurement
and control. Chem. Eng., 91, 62-70.
Goff, H. D. & Davidson, V. J. (1992). Flow characteristics
and holding time calculations of ice cream mixes in HTST
Holding Tubes. J. Food Prot., 55, 347.
Hildebrand, F. B. (1976). Advanced Calculus for Applications.
2nd ed. Prentice-Hall Inc., Englewood Cliffs, New Jersey.
Schellhorn, H. H. & Forsberg, C. W. (1984). Multiplicity of
Extracellular p-( 1,4)-endogluconases of Bacteroides Suc-
cinogenes S85, Can. J. Micro., 30, 930-7.
White, J. P. (1991). The on-line characterization of non-New-
tonian reacting fluid food systems. M.Sc. thesis, University
of Guelph, Guelph, ON, Canada.
Windhab, E. & Kuhlmann, D. (1989). In-line control of elas-
tic and viscous fluid properties. Poster presented at 5th In-
ternational Congress on Engineering and Food, Cologne,
Germany, 28 May-3 June.
(Received 8 June 1992; accepted 11 August 1992)