Вы находитесь на странице: 1из 7

Ind. Eng. Chem. Res.

1999, 38, 1729-1735

1729

RESEARCH NOTES
Does the Viscosity of Glycerin Fall at High Shear Rates?
Prasannarao Dontula, Christopher W. Macosko,* and L. E. Scriven*,
Coating Process Fundamentals Program, Center for Interfacial Engineering and Department of
Chemical Engineering & Materials Science, University of Minnesota, Minneapolis, Minnesota 55455

The sole report of glycerin-water viscosities at high shear rates, by Ram, has so far been
interpreted as evidence of shear thinning. Two alternative interpretations, both of which invoke
experimental artifacts, have now been explored. One of these, viscous dissipation, qualitatively
explains the progressive fall in the viscosity of glycerin-water as the shear rate is ramped-up
beyond 60 000 s-1 to 90 000 s-1. New measurements with parallel disks separated by a narrow
gap indicate that the viscosity of 60 wt % glycerin in water is constant up to 90 000 s-1.
In a 1967 communication to Eirichs Rheology, Ram1
stated that certain glycerin-water solutions are shear
thinning at shear rates of 60 000 s-1 and above. No data
or experimental details were provided, only seven
literature citations.2-8
Neale2 had reported shear thinning in castor oil in a
journal-bearing apparatus in which the current drawn
by the motor was measured as a function of continually
rising journal speed. On the basis of well-conducted
experiments, Weltmann3 had shown that poor temperature control and wide-gap fixtures can lead to apparent
shear thinning in rheological measurements with oils
and inks, though not glycerin. Hagerty4 had observed
that the viscosity of concentrated solutions of glycerin
in water (under unspecified experimental conditions) fell
with time over 5 min and had speculated it was caused
by striae visible in these liquids. Dumanskii and Khailenko5 had reported viscosities of glycerin-water solutions
that are 10-100 times larger than accepted values, an
aberration that was not resolved. Merrill6 and Ram and
Tamir7 had mentioned results from Rams Sc.D. thesis8
in passing.
The seventh citation was of Rams8 unpublished Sc.D.
thesis at Massachussetts Institute of Technology (M.I.T.),
which contains the sole set of experimental data. The
thesis can be examined only by application to M.I.T.,
which we have done; it seems not to have been scrutinized before on this account. Ram measured the torque
on the stationary outer cylinder of a narrow-gap (150
m) concentric cylinder fixture, as the angular velocity
of the inner cylinder was ramped-up linearly over
anywhere between 10 and 20 s, a standard procedure
with the Merrill-Brookfield viscometer9 (also Figure
1a). The maximum shear rate was about 96 000 s-1 at
the end of the ramp-up. The ends of the fixture were
not sealed; sample liquid leaked out slowly and was
continually replenished by a syringe. The temperature
of the brass fixtures was controlled to within (1 C by
water circulating at 2 L/min in each annular channel
about 1.5 mm beneath the cylindrical surfaces (Figure
E-mail: pjensen@cems.umn.edu. Phone: (612) 625-1058.
Fax: (612) 626-7246.

1b). The temperature of the fixtures themselves was not


measured; presumably it was inferred from the water
temperature in the cooling bath. The shear viscosity at
each moment of this short test was taken from the
instantaneous slope of the flow curve of shear stress
(from measured torque) versus shear rate (from imposed
angular velocity). The reproducibility was reported to
be within 2-5% for liquids of viscosity larger than 2.5
mPa s. The inferred shear viscosity of a 62.5 wt %
solution of glycerin in water at an unspecified temperature (29 C according to the now standard measurements of Sheely10) was 9 ( 0.089 mPa s up to 40 000
s-1, i.e., during the first 8.5 s of a 20 s ramp-up; it fell
gradually above 55 000 s-1 and then more rapidly to
about 8.44 mPa s at 90 000 s-1, i.e., about 6.2% smaller
(cf. Figure 51 in Rams thesis8). From these data, Ram
concluded that glycerin-water solutions above 50 wt %
concentration thin at high shear rates. Ram also reported shear thinning in ethylene glycol, aqueous sugar,
and solutions of paraffin in decalin (cf. Table XIII in
Rams thesis8).
Today this conclusion is surprising, because no supporting data have been reported, as far as we know, on
glycerin-water solutions or other low molecular weight
hydrogen-bonding liquids. The possibility that glycerinwater solutions are mildly shear thinning above 60 000
s-1 at room temperature is important, because they are
now often employed as Newtonian test liquids in various
viscous free-surface coating flows where strain rates
often exceed 10 000 s-1, and even 100 000 s-1. Hence,
we consider here possibilities that the apparent shearthinning was an experimental artifact, in particular due
to instrument eccentricity and shear heating. By this
note we also report our own measurements of viscosities
of aqueous glycerin up to 100 000 s-1.
Effect of Hydrodynamic Instabilities
Couette flow with a rotating inner cylinder is susceptible to Taylor instability; the toroidal vortices that form
raise the torque. Up to the highest speed possible in
Rams device, the critical Taylor number above which
the secondary flow appears can be exceeded only with
liquids whose viscosity is less than 4 mPa s.

10.1021/ie9805685 CCC: $18.00 1999 American Chemical Society


Published on Web 02/25/1999

1730 Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999

Figure 1. Schematic of the Merrill-Brookfield viscometer used by Ram.8 (a) Plan view: the rotor, stator and rotating frame form liquid
and air bearings that support the reaction force applied by the knife connected to the torque-measuring member. (b) Radial cross-section
A-A of the apparatus: the brass stator and rotor are both cooled by water circulating at 2 L/m in annuli 1.5 mm beneath the surface.

Effect of Eccentricity
Figure 1a also shows a plan view schematic of Rams
device. The outer cylinder (stator) was floated in the
tapered frame (clearance unspecified) by compressed air
supplied at about 0.17 M Pa from eight equispaced holes
around the periphery (size and location unspecified).
The stator was restrained by a knife resting against a
V-shaped bearing on its surface. The knife actuated a
torque transducer via a lever, and the reaction force by
the knife deflected the stator by 3 at maximum torque
(cf. Appendix D in Rams thesis8). From dimensions of
the apparatus and measured torque, this reaction force
can be estimated to be 1.614 N at 55 000 s-1 and 2.642
N at 90 000 s-1 with 60 wt % glycerin in water. This
load was borne by the difference in restoring forces set
up in the liquid and air bearings. The speeds were
such that the gas-bearing number was small and the
compressibility of the gas was negligible.11 A reasonable
value of the nominal radial clearance in the air bearing
is 25.4 m (0.001 in.); the eccentricity would then be
only 0.0747% of the liquid gap at 55 000 s-1 and
imperceptibly higher at 90 000 s-1.
The effect of eccentricity on the shear stress distribution, drag force, and torque on the walls of the liquidfilled gap is well-known. From Sommerfelds lubrication
theory12

2x1 - 2
M
)
M0
2 + 2
where M0 is the torque that would be measured if
cylinders were coaxial, and  is the eccentricity
pressed as a fraction of the nominal gap between
cylinders. The torque is virtually unaffected by
centricities below 1%, as in this case.

(1)
the
exthe
ec-

Effect of Shear Heating


The sample liquid slowly flows axially through the
gap between the cylinders, draining out the bottom
around the circumference and being replenished halfway up the outer cylinder wall at one circumferential
location. The maximum axial velocity of a 11 mPa s
liquid under gravity in the 150 m gap, about 3 10-3
m/s, is negligibly small compared to the maximum
imposed azimuthal velocity, 14.4 m/s. Gravity flow thus
contributes inconsequentially to shear rate.

However, thermal effects can be significant. Shearing


of Newtonian liquid between infinite parallel plates
separated by 2h, a good approximation to the actual gap,
generates heat. When the speed of the inner cylinder is
raised linearly, the rate of heat generation per unit
volume, 2, rises quadratically with time if the shear
viscosity is independent of shear rate . If the liquid
also has constant thermal properties and the channel
walls are maintained at constant temperature (see
Figure 2a), the temperature at position x (-h e x e h)
within the liquid is given by13

Tl(x,t) )

4R2
Fc

(-1)n

n)02n

+1

cos

](

(2n + 1)x t2
2h

2t
2

2(1 - e-t)
3

(2)

where R is the angular acceleration and (2n +


1)2l2/4h2. Here l is the thermal diffusivity kl/Flcl, about
10-7 m2/s in liquids. All thermal properties of 62.5 wt
% glycerin in water are not available; hence, subsequent
calculations are carried out for 60 wt % glycerin in water
at 20 C. The time constant in eq 2 for 60 wt % glycerin
in water in a 150 m gap is large (density Fl of 1156
kg/m3, specific heat cl of 2410 J/(kg K), and thermal
conductivity kl of 0.381 J/(s m K) at 20 C14). The liquid
temperature is nearly constant at short times (less than
0.05 C rise in 5 s) and rises slowly and then rapidly
with the square of time. The temperature at the
midplane of the 11 mPa s 60 wt % glycerin-water
solution would rise by 0.24 C in the first 11.5 s of a 20
s ramp-up, i.e., at 55 000 s-1, and about 0.65 C when
the shear rate was 90 000 s-1. The estimated temperature rise in shorter ramps is smaller, but the estimated
difference at 90 000 s-1 between 10 and 20 s ramps is
less than 0.01 C.
However, the plate surfaces did not, in fact, remain
at constant temperature, for two reasons. One is the
thermal inertia of the brass between them and the
coolant passages; the other is the thermal resistance of
the same material. If the brass has constant thermal
properties and the brass-cooling water interfaces are
maintained at constant temperature, the method of
Laplace transforms can be used to calculate the temperature Tl in the liquid (0 e x e h) and Tb in the brass
(-l e x e 0; see Figure 2b):

Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 1731

Figure 2. Schematic of the cases studied: (a) liquid layer only with isothermal walls and (b and c) heat transfer through the liquid and
brass wall composite. In part b, the outermost surfaces of the brass walls remain at constant temperature, whereas in part c, they are
allowed to vary. One wall is ramped-up linearly in speed.

Tl(x,t) )

2R2
Fc

by integration across the gap,

cos nl cos qn(h - x)

(l + q3h) sin nl cos qnh + (qh + q2l) cos nl sin qnh

t2

bn2

Tb(x,t) )

2R2q2
Fc

dx
1 0
) - V dv )
02h(x,t)

n)1 n

2t

2(1 - e-bn t)
2

(bn2)2

(bn2)3

(3)

n)1 n

sin qna sin n(l + x)

(l + q3h) sin nl cos qnh + (qh + q2l) cos nl sin qnh

t2

bn2

2t
(bn2)2

2(1 - e-bn t)
2

(bn2)3

(4)

Here, the n, n ) 1, 2, 3, ..., are the positive roots of

cos nl cos qnh - q2 sin nl sin qnh ) 0


and q xb/l, qkb/k, and kb and b are respectively
the thermal conductivity and thermal diffusivity of
brass. The properties of yellow brass are as follows:
density Fb, 8470 kg/m3; specific heat cb, 385 J/(kg K);
thermal conductivity kb, 116 J/(s m K).15 With Mathematica,16 the first seven terms of the series in eqs 3
and 4 were used to calculate temperature in the liquid
and brass. The temperature at the midplane of a 60 wt
% glycerin-water solution was thereby estimated to rise
by 0.3 C after 11.5 s of a 20 s ramp-up, i.e., at 55 000
s-1, and by about 0.75 C after 18.75 s, i.e., when the
shear rate reached 90 000 s-1. The estimated temperature rise in shorter ramps is smaller, but the estimated
difference at 90 000 s-1 between 10 and 20 s ramps is
less than 0.05 C. Figure 3a shows how the temperature
distribution in the glycerin-water and the confining
brass walls evolves during a 10 s ramp-up.
The viscosity of glycerin-water solutions falls with
temperature, the more so the higher the glycerin
concentration. The viscosity of 60 wt % glycerin in water
falls probably exponentially from 11 mPa s at 20 C to
about 7 mPa s at 30 C;10 the temperature coefficient of
viscosity is 0.0306 K-1 at 20 C. The shear stress )
- v/x is uniform across the gap of width 2h. Hence,

V(t)

(5)

where V is the velocity of the wall at x ) 0. Taking the


temperature, and hence viscosity distributions, to be
symmetric about the midplane (eq 3) yields the shear
stress. The shear stress , and hence the apparent
viscosity, would be 0.6% smaller than the true viscosity
at 55 000 s-1 and 1.6% smaller at 90 000 s-1. Thus,
shear heating explains only a fourth of the decrease
reported by Ram (Figure 4). Three terms instead of
seven in eq 3 lowered the estimate of the apparent
viscosity by less than 0.01%.
However, the two brass-cooling water interfaces did
not, in fact, remain at constant temperature. The
average temperature rise of about 600 cm3 of cooling
water used during this short test in Rams device was
estimated by integrating the flux at the brass-water
interface and neglecting the nonuniform distribution of
cooling water in the annuli; the estimate was about 15
C, i.e., 20 times the maximum temperature rise in
glycerin-water estimated with eq 3. The temperature
of water would, in reality, rise little at first and then
sharply, qualitatively mirroring the temperature rise in
the sheared liquid; this probable time course differs from
the ideal isothermal condition used in the analysis so
far (Figure 2b). A more accurate boundary condition
(Figure 2c) would be kb dT/dx ) H(Tb - Tw), where Tw
is the water temperature and H the heat-transfer
coefficient in forced convection.
If the brass has constant thermal properties and the
cooling water is at constant temperature, heat transfer
by forced convection would alter the temperature Tl in
the liquid (a e x e b; see Figure 2c for notation) and Tb
in the brass (0 e x e a) to (cf. O
zis ik,17 pp 273-298)

Tl(x,t) )

Xln(x)

n)1

Tb(x,t) )

Xbn(x)

n)1

(
(

g/nR2

g/nR2

t2

2t

)
)

2(1 - e-n t)

n2

n4

n6

t2

2t

2(1 - e-n t)

n2

n4

n6

(6)

(7)

1732 Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999

Figure 3. Calculated temperature distribution with position in the 60 wt % glycerin-water and confining brass walls as the shear rate
is ramped-up from 0 to 90 000 s-1 in 10 s. (a) The brass-cooling water interface is at constant temperature (Figure 2b). (b) The brasscooling water interface temperature is allowed to vary with time (Figure 2c).

where the n, n ) 1, 2, 3, ..., are the positive roots of

sin

n
H xb
cos
a)
kb n
xb

a-

xb

n
n
n
b
H xb
tan (b - a) cos
a+
sin
a
l
l
xb kb n
xb

kl
kb
and

n
H xb
Xbn(x) ) cos
x+
sin
x
xb kb n
xb
n

Xln(x) ) Cln cos


n

cos
Cln )
cos

xl

xl

x + tan

xl

cos

(b - a)
g/n )

N)

b sin

xl

xb

xb

abXln2(x) dx

kb
b

0aXbn2(x) dx + llabXln2(x) dx

The cooling water in Rams experiments flowed through


flat ducts (25.4 mm wide, 4.5 mm thick, and about 175
mm long; the thickness was read from Rams Figure
578). The flow was just within the transition zone from
laminar to turbulent flow (Reynolds number Re based
on the hydraulic radius is 2229). For lack of better
correlations, the cooling-water flow was considered to
be laminar (Re < 2100). The heat-transfer coefficient
H can be estimated from Graetzs solution (Jakob,18 pp
462-4)

4cpFQ
HDe
) 1.85
k
kL

()

HDe w
k b

0.14

[(

1.75

n
H xb
a
sin
kb n
xb

a+

water. The quantity within brackets is a modified


Graetz number for flow through ducts. The heattransfer coefficient H in eq 8 is based on the arithmetic
mean temperature of the cooling water and approaches
that based on the logarithmic mean temperature difference for Graetz numbers larger than 2000.18 Equation 8 is similar to the empirical relationship

1/3

(8)

where De ) 4Rh, Rh is the hydraulic radius, Q is the


cooling-water flow rate, L is the length of the annular
channel, and cp, k, and F are respectively the specific
heat, thermal conductivity, and density of the cooling

4cpFQ
kL

+ 0.04

) ]

L
Gr Pr
De

0.75 1/3

(9)

used to correlate a large number of heat-transfer


measurements. McAdams19 reports eq 9 fits the data
with a maximum deviation of 60% (pp 235-7). Here Gr
is the Grashof number and Pr is the Prandtl number of
the flow, and the subscripts b and w denote values in
the bulk and those at the wall, respectively. If the
Grashof and Prandtl numbers are small and the wall
temperature does not significantly differ from the bulk
temperature, then free convection effects are negligible
and eq 9 reduces to eq 8 with a 6% difference in the
numerical constant. In Rams8 device at 20 C, Pr ) 7,
Gr 20, and the Graetz number was 1860. Hence,
either eq 8 or eq 9 may be used to estimate the heattransfer coefficient. In subsequent calculations, a heattransfer coefficient H of 1500 J/(s m2 K) estimated with
eq 8 was used. With Mathematica,16 the first seven
terms of the series in eqs 6 and 7 were used to calculate
temperature in the liquid and brass. The temperature
at the midplane of a 60 wt % glycerin-water solution
would then rise by 1.45 C in the first 11.5 s of a 20 s
ramp-up, i.e., at 55 000 s-1, and about 4.5 C when the
shear rate was 90 000 s-1. The temperature would rise
by 1.13 and 3.65 C, respectively, at the same shear
rates during a 10 s ramp-up, while it would rise by 0.85
and 2.8 C during a 5 s ramp-up. The shear stress ,
and hence the apparent viscosity, during a 5 s ramp-up
would be 2% smaller than the true viscosity at 55 000
s-1 and 7.03% smaller at 90 000 s-1 and is enough to
account for the deviation in Rams experiments.8 The
longer the ramp, the greater the apparent shear thinning predicted (Table 1 and Figure 4b). Three terms
instead of seven in eq 6 lowered the estimate of the
apparent viscosity by less than 0.05%. Doubling the
heat-transfer coefficient H raised the estimated apparent viscosity at 90 000 s-1 in Table 1 by 1% during a 5

Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 1733

Figure 4. Comparison of Rams8 data on 62.5 wt % glycerin-water and calculations on 60 wt % glycerin-water. (a) Shear stress versus
shear rate: the dotted line corresponds to Newtonian behavior without shear heating. (b) Ratio of the apparent viscosity and viscosity at
low shear rate versus shear rate. The circles are representative data points from part a. The dashed lines are theoretical estimates for
60 wt % glycerin-water with an infinite heat-transfer coefficient (Figure 2b) and finite heat-transfer coefficient by forced convection
(Figure 2c).
Table 1. Comparison of Experiment with Theory:
Concentric Cylinders
% reduction
in apparent
viscosity at

ramp
heat-transfer
duration, coefficient,
)
)
s
J/(s m2 K)
55 000 s-1 90 000 s-1
experiment (Ram8)
theory

10-20
10-20
3
5
10
20

infinite
1500
1500
1500
1500

1.96a
0.6
1.73
2.15
2.94
3.89

6.2a
1.6
5.60
7.03
9.38
11.63

Rams experiments on 62.5 wt % glycerin-water.

s ramp-up, by about 2% during a 10 s ramp-up, and by


about 4% during a 20 s ramp-up and is still sufficient
to explain the decrease reported by Ram. Figure 3
compares the effect of the thermal boundary condition
on the temperature distribution in the glycerin-water
and confining brass walls during a 10 s ramp-up.
Measurements with Parallel Disks
We measured shear viscosity in a room at 22.5 C with
an ARES controlled-strain rheometer (Rheometrics
Scientific, Piscataway, NJ). The rheometer was equipped
with a force-rebalance transducer that measures torques
between 0.02 g cm (1.96 N m) and 2000 g cm (0.196 N
m) in two ranges to within 1% accuracy. The liquid was
sheared between two titanium alloy disks, each 50 mm
in diameter and 1.5 mm thick. Because of the unusually
small gaps used (down to 30 m), the alignment of the
actuator (motor) and transducer axes, their concentricity, and gap uniformity were crucial. The total indicated
runout of the actuator was measured by a Mitutoyo dial
gauge (0.0001 in. or 2.54 m, least count) to be less than
(0.0002 in. The transducer slide axis was parallel to
the actuator axis to within 0.0002 in. over 1.5 in. of its
travel. The face of one of the disks was perpendicular
to the actuator axis to within (0.0004 in. and the other
to within (0.0001 in. Actuator and transducer were
concentric to within (0.0003 in.
The shear viscosity of a silicone oil (nominal viscosity
0.6 Pa s) and 60 wt % glycerin in water was measured
at different disk separations between 0.02 and 0.75 mm
to quantify the effect of alignment errors. Figure 5b
shows the average of the apparent viscosity of these

liquids between 100 and 1000 s-1. The error bars bound
the maximum and minimum values; no thinning was
observed over this range of shear rates. The apparent
viscosity falls as the gap is narrowed and more sharply
so at smaller gaps. Two reasons for the drop in viscosity
can be attributed to canted fixtures (Figure 5a). First,
the measured gap between the disks corresponds to the
distance between the points of closest approach on the
disks (high points) and is consequently incorrect. Second, the gap is nonuniform; the azimuthal flow is hence
a combination of drag flow due to the moving plate and
azimuthal pressure gradient-driven flow. The torque
reduction owing to the nonuniform gap can be estimated
by integrating Sommerfelds solution for a flooded
journal bearing (eq 1) across the radius. The dimensionless eccentricity, defined by

(r) )

h
r
Rhnom

rises with radius r. Here h is the cantedness at the


largest radius R (obtained from perpendicularity measurements), and hnom is the nominal separation between
the plates (Figure 5a). Radial flow can be neglected at
the first approximation. Figure 5b shows that both of
these factors are significant and explain the drop in
apparent viscosity at small separations. The theoretical
estimates are based on the perpendicularity measurements, which were accurate only up to 2.54 m. Other
factors such as run-out of the actuator axis and departure of the transducer and actuator axes from coaxiality
may also affect the flow field and alter the measured
torque and thus the inferred viscosity.
High shear rates were achieved by ramping the
actuator to its maximum speed and down as described
by Connelly and Greener.20 The shear rate was raised
in two steps: from 0 to 1000 s-1 and from 1000 to 90 000
s-1. Each step was 3 s long. The shear rate was lowered
similarly. In order to collect torque data more rapidly
(every 10 ms) than normally possible with the ARES
rheometer, a software low-pass filter was turned off.21
The motor is rated to attain two-thirds of its programmed speed in less than 3 ms. Figure 6 shows the
apparent viscosity versus shear rate during the experiment. The apparent viscosity is constant over the range
of shear rates attained and varied by less than (1%
with three different sample loadings. Ramps longer than

1734 Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999

Figure 5. Apparent viscosity measured with parallel disks. (a) Schematic of the parallel disks, as used. When the disks are canted, the
nominal gap is incorrectly measured and the gap is nonuniform. (b) Apparent viscosity of silicone oil (nominal viscosity 0.61 Pa s, circles)
and 60 wt % glycerin-water (triangles) at different apparent gaps between the parallel disks. Shear rates varied from 100 to 1000 s-1;
error bars bound the maximum and minimum values. The apparent viscosity measured with the largest gap was used to normalize the
experimental results with each liquid. The dotted line shows the change in viscosity just due to h, the nominal error in reading the gap.
The dashed line shows the estimated total effect of canted disks. The apparent viscosity at a gap of 0.5 mm was used to normalize the
theoretical estimates.

Figure 6. Measured shear stress and apparent viscosity of 60


wt % glycerin in water as it is sheared in a 40 m apparent gap
between two 50 mm diameter parallel disks. The shear rate is
raised from 1000 to 90 000 s-1. Torque at low rates is smaller than
the minimum that can be measured, and hence the spread in
viscosity is larger.

5 s showed hysteresis in the measured torque when the


shear rate was ramped-up and then ramped-down (also
see Connelly and Greener20). After the correction for
small gaps interpolated from experimental results
(Figure 5b) is applied, the viscosity is 9.66 ( 0.14 mPa
s, 1.73% lower than Sheelys10 reported value at 22.5 C.
Conclusions
Rams8 measurements of glycerin-water viscosities
at high shear rates have been interpreted by others as
evidence of shear thinning. Those measurements were
of shear torque versus rotation rate in a Couette-type
instrument, the Merrill-Brookfield viscometer. Two
alternative interpretations have now been explored.
Both invoke artifacts that can account for the torque
rising progressively less than in proportion to the
rotation rate as the mean shear rate in the gap passes
55 000 s-1 and ramps-up to 90 000 s-1.
The liquid gap itself and the instruments largerradius air-bearing gap (Figure 1a) could not have been
appreciably uncentered by the thrust on the latter by
the reaction force of the torque-measuring member and
hence cannot account for the deviation reported.
Heating by viscous dissipation can cause apparent
shear thinning as shown by Sukanek and Laurence.22
The liquid in the gap in Rams device was certainly

heated by viscous dissipation. Our estimate of the


efficacy of the instruments provision for heat removal
and of the consequent viscosity variation in the gap
suggests that the entire reported 6.2% deviation at
90 000 s-1 stems from viscous heating. A 5 s ramp with
heat removal by forced convection of cooling water
predicts a 7.03% deviation, close to Rams8 observations
(Table 1). The slower the ramp, the greater the shear
thinning predicted. Other liquids in the same gap must
have been viscously heated; Ram reported shear thinning in 14 mPa s ethylene glycol at 60 000 s-1. Equation
3 predicts a 3% drop in viscosity of ethylene glycol at
the same shear rate at 17.5 C (c of 2398 J/(kg K), F of
1114.9 kg/m3, of 23 mPa s, temperature coefficient of
viscosity of 0.041 K-1, and k of 0.241 at 25 C23,24). More
realistic thermal boundary conditions, i.e., forced convection incorporated in eqs 6 and 7, predict higher
deviations that also depend on ramp duration.
Our measurements with parallel disks separated by
narrow gaps indicate that the viscosity of 60 wt %
glycerin in water is constant up to 90 000 s-1.
Acknowledgment
This work was supported by industrial and National
Science Foundation funds through the Center for Interfacial Engineering at the University of Minnesota.
Literature Cited
(1) Ram, A. High-shear Viscometry. In Rheology: Theory and
Applications; Eirich, F. R., Ed.; Academic Press: New York, 1967;
Vol. 4, p 251.
(2) Neale, S. M. The Viscosity of Oils at High Rates of Shear.
Chem. Ind. (London) 1937, 37, 140.
(3) Weltmann, R. N. Consistency and Temperature of Oils and
Printing Inks at High Shearing Stresses. Ind. Eng. Chem. 1948,
40, 272.
(4) Hagerty, W. W. Use of an Optical Property of GlycerineWater Solutions to Study Viscous Fluid-flow Problems. J. Appl.
Mech. 1950, 17, 54.
(5) Dumanskii, I. A.; Khailenko, L. V. Rheological Properties
of Glycerine Solutions. Kollidn. Zh. 1960, 22, 277.
(6) Merrill, E. W. Non-Newtonianism in Thin liquids: Molecular and Physical Aspects. In Modern Chemical Engineering
Series; Acrivos, A., Ed.; Reinhold: New York, 1964; Vol. 1, p 141.
(7) Ram, A.; Tamir, A. A Capillary Viscometer for NonNewtonian Liquids. Ind. Eng. Chem. 1964, 56, 47.

Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 1735


(8) Ram, A. High-Shear Viscometry of Polymer Solutions. Sc.D.
Thesis, Massachusetts Institute of Technology, Boston, MA, 1961.
(9) Merrill, E. W. A Coaxial Cylinder Viscometer for the Study
of Fluids under High Velocity Gradients. J. Colloid Sci. 1954, 9,
7.
(10) Sheely, M. L. Glycerol Viscosity Tables. Ind. Eng. Chem.
1932, 24, 1060.
(11) Gross, W. A. Fluid Film Lubrication; Wiley-Interscience:
New York, 1980. This is an extension and revision of the 1962
publication Gas Film Lubrication by W. A. Gross.
(12) Tipei, N. Theory of Lubrication; Stanford University
Press: Stanford, CA, 1962.
(13) Carslaw, H. S.; Jaeger, J. C. Conduction of Heat in Solids,
2nd ed.; Oxford University Press: Oxford, U.K., 1959; p 131.
(14) Perry, R. H.; Green, D. W.; Maloney, J. O. Perrys Chemical
Engineers Handbook, 6th ed.; McGraw-Hill: New York, 1984.
(15) Baumeister, T.; Avallone, E. A.; Baumeister, T., III. Marks
Standard Handbook for Mechanical Engineers, 8th ed.; McGrawHill: New York, 1978.
(16) Wolfram, S. Mathematica, 2nd ed.; Addison-Wesley: Reading, MA, 1991.
(17) O
zis ik, M. N. Boundary Value Problems in Heat Conduction; Dover: New York, 1989. This is a republication of the 1968
International Textbook Company edition.

(18) Jakob, M. Heat Transfer Vol. 1; Wiley: New York, 1949.


(19) McAdams, W. H. Heat Transmission, 3rd ed.; McGrawHill: New York, 1954.
(20) Connelly, R. W.; Greener, J. High-shear Viscometry with
a Rotational Parallel-disk Device. J. Rheol. 1985, 29, 209.
(21) Mackay, M. E.; Liang, C.-H.; Halley, P. J. Instrument
Effects on Stress Jump Measurements. Rheol. Acta 1992, 31, 481.
(22) Sukanek, P. C.; Laurence, R. L. An Experimental Investigation of Viscous Heating in Some Simple Shear Flows. AIChE
J. 1974, 20, 474.
(23) Marie, C.; Bodenstein, M.; Bruni, G.; Cohen, E.; Wilsmore,
N.-T.-M. Annual Tables of Constants and Numerical Data; GauthierVillars: Paris, France, 1925/26; Vol. 7, p 89.
(24) Marie, C.; Bodenstein, M.; Bruni, G.; Cohen, E.; Wilsmore,
N.-T.-M. Annual Tables of Constants and Numerical Data; GauthierVillars: Paris, France, 1927/28b; Vol. 8, pp 31 and 59.

Received for review September 2, 1998


Revised manuscript received January 12, 1998
Accepted January 13, 1998
IE9805685

Вам также может понравиться