VISCOSITYOF WATERAT HIGHPRESSURES 1187

Viscosity of Water at High Pressures and Moderate Temperatures

by E. M. Stanley and R. C. Batten'

Naval Ship Research and Development Laboratory, Annapolis, Maryland 2 1 40.2 (Received April 29, 1968)

The viscosity of water was measured at pressures from 0 to 1406.2 kg/cm2 gauge and over a temperature
range of 2-30' with a rolling ball viscometer. The data obtained show good agreement with most of the
previously reported results. Analysis of the viscosity data shows activation energies of viscous flow decreasing
in a uniform manner with increasing temperature and pressure and a possible transition phase at 36' and
atmospheric pressure, indicated by extrapolated data.

Introduction ture of the controlling liquid; electronic logic circuitry

In the last several years measurements of the
viscosity of water under pressure have become im-
portant because of the use of these data in studying the
structure of water, the processes of ionic conduction and
viscous flow, and certain anamolous behaviors of water.
Recently new data have appeared in the literature2-6
to supplement the older work of Bridgman,6 Cohen,'
Lederer," and Tammann and Rabe;g but correlation of
results among the different authors varies and in some
cases large discrepancies are noted. Horne and
Johnsons and Wonham2 have pointed out that more
information is needed to resolve the conflicting data
and to establish the correct values for use in other
studies.
Experimental Section
Apparatus. The instrument used in these experi-
ments is a rolling ball viscometer. Viscosity is de-
termined indirectly by measuring the time it takes
a ball to roll between two points in an inclined barrel
closed on both ends and filled with the fluid of interest.
Knowing the density of the fluid being investigated,
PF, the density of the rolling ball, PB, the measured time
of roll, t, between the two fixed points, and a previously
determined calibration constant, c, the viscosity, 17,
can be calculated from the equation
9 = Ct(PB - PF) (1)
The theory and limitations of this type of viscometer
for use at atmospheric pressure have been described by
Hubbard and Brownloand further considerations for its
use a t pressure have been discussed by Horne and
Johnson.*
The experimental instrumentation is a viscometer
which consists of a ball, a barrel, and sensing coils, in a
pressure vessel, and associated apparatus (Figure 1 ) .
The associated apparatus includes a pressurizing system
for generating and measuring pressure in the viscom-
eter; a temperature bath which circulates and regulates
the temperature-controlling liquid around the pressure
vessel; two quartz crystal thermometers, one a t each
end of the pressure vessel for measuring the tempera-
which measures the roll time interval of the ball be-
tween the two sensing coils; and electronic instrumenta-
tion for visually displaying the roll times and temper-
atures in digital form. Details of the viscometer and
its associated apparatus have been described previ-
ously."
Accuracy and Error. As previously indicated, veri-
fication is needed of existing experimental data for the
viscosity of water as a function of pressure. The
discrepancies in currently available data by different
investigators are probably caused by instrumentation
errors. A detailed study of rolling ball viscometers,
in general, and the laboratory experiments (with the
instrument used to obtain the results reported here)
indicated several possible sources of error in other data
taken with the same type of instrument. The primary
causes of error are pressure and temperature effects on
the ball and the barrel, turbulent flow, and nonlineapity
of the instrument throughout the viscosity and temper-
ature ranges used. The rolling ball viscometer em-
ployed was carefully designed and calibrated to avoid
or minimize these sources of error.
Changes in pressure and temperature affect the
clearance between the ball and barrel and were elimin-
ated by constructing the ball and barrel of materials
whose coefficient of thermal expansion and bulk moduli
were nearly the same. Furthermore, the measuring
barrel was placed inside the pressure vessel so that the
pressure was equal on all sides of the barrel. Changes
(1) The opinions expressed in this paper are those of the authors and
do not necessarily represent those of the U. 8. Navy or the Naval
establishment a t large.
(2) J. Wonham, Nature, 215, 1053, (1967).
(3) R. A. Horne and I).8. Johnson, J. Phys. Chem., 7 0 , 2182 (1965).
(4) K . E. Bett and 5. B. Cappi, Nature, 207, 620 (1965).
(5) W. Weber, Z . Angew. Phys., 15, 342 (1963).
(6) P. W. Bridgman, Pfoc. A m . Acad. Arts Sei., 61, 57 (1926).
(7) R. Cohen, Ann. Phys., 45, 666 (1892).
(8) E. L. Lederer, Kollotd-Beth., 34, 270 (1932).
(9) G . Tammann and H. Rabe, 2. Anorg. Allg. Chem., 1 6 8 , 7 3 (1927).
(IO) R. M. Hubbard and G . G. Brown, Ind. Eng. Chem., Anal. E d . ,
15, 212 (1943).
(11) E. M. Stanley and R. 0. Batten, Anal. Chem., 40, 1751 (1968).

Volume 79,Number 6 Ma# 1959

1188 AND R. C. BATTEN
E. M. STANLEY

Table I: Relative Viscosity of Water qp/ql for Various Temperatures and Pressures

Pressure, ?P/?1 ?Pi?l ?P I?I 7P/?l ?Pl?l le191

kg /cm2 a t 2.233 a t 5,998O a t 10.005 a t 15.020O a t 19.997 a t 20.001
176 0.9761 0.9822 0,9844 0.9905 0.9952 0,9990
0.9743 0.9821 0.9870 0.9904 0.9935 0.9977
0.9947

352 0,9596 0.9686 0.9755 0.9822 0.9917 0.9990

0.9575 0.9683 0.9779 0.9822 0.9912 0,9991

527 0.9475 0.9595 0,9701 0.9769 0 9898 I 1 .0000

0,9456 0.9601 0,9693 0.9777 0.9887 1.0004

703 0.9403 0.9532 0.9610 0.9750 0.9899 1 .0049

0.9411 0.9533 0.9633 0.9754 0.9886 1.0046
0.9873

879 0.9347 0.9500 0.9602 0.9752 0.9919

0.9357 0.9501 0.9622 0.9759 0.9907

1055 0.9310 0.9491 0.9607 0.9777 0.9939 1.0143

0.9322 0.9485 0.9614 0.9788 0.9941 1.0151

1230 0.9305 0.9510 0.9806 0,9999

0.9314 0.9495 0.9818 0 9984
I

1406 0.9322 0.9522 0.9672 0.9832 1 0050

I 1.0283
0.9339 0.9505 0.9726 0.9863 1.0059 1.0315
1.0042

1406.2 kg/cm* and 0.03% for a temperature of 30.

The barrel length changes less than 0.03% for the same
TIMER
temperature change. With these variations, estimated
PRESSURE changes in the roll time, by equations in Hubbard and
Brown,l0 are on the order of O.l%, maximum, which is
the magnitude of the repeatability of the roll times.
QUARTZ CRYSTAL
Standard calibration techniques1J 3 were employed

(12) The change in diameter, due to pressure, of a thick-walled

PRESSURE
ACCUMULATOR cylinder open on both ends where pressure is equal on all sides is
TEMPERATURE given by Timoshenko (S. Timoshenko, Strength of Materials,
D. Van Nostrand Co., Inc., New York, N. Y., 1948) as

WATER JACKET

VISCOSIMETER/

I
Figure 1, Block diagram of viscometer and associated apparatus.
P Ad
in the radius of ball and barrel, and therefore the
clearance between them, can be calculated,12 but
because the clearance between the ball and barrel is
small, 0.0203 cm, exact calculations of the change are
difficult.
Calculations using equations of ref 12 indicate that
the average clearance between the ball and barrel
changes less than 0.65% for a maximum pressure of
= -Pd(l - N ) / E
where d is the inside diameter of the cylinder (barrel) a t atmospheric
pressure (0.6502 cm), p is Poissons ratio, P is the gauge pressure,
E is the modulus of elasticity of the cylinder, and the minus sign
indicates a decrease in diameter. The change in diameter of a solid
ball due to pressure can be calculated from the rearranged com-
pressability equation
Ad = P d j B K
where d is the diameter of the ball (0.6299cm), P is the gauge pressure,
and K is the bulk modulus. The change in diameter of the ball and
barrel with temperature cain be expressed by the equation
Ad = d a A T
where d is the diameter of the ball or barrel, a is the coefflcient Of
linear expansion, and T is the temperature. The change in length
A2 of the barrel due to temperature expansion is shown by a similar
equation
A1 =ZaAT
(13) W.W.Langston and V. Orr, Calibration of the Ruska Rolling-
Ball Viscosimeter, United Gas Corp. Intraofflce Report No. 22-52,
Jan 8, 1952.

The Journal of Phyeical Chemiatry

VISCOSITY
OF WATERAT HIGHPRESSURES
SlANLIV L BAllEN 2.Z.C
BRIDGMAN O'C
6 HORNE 6 JOHNSON 2 S C
PRESSURE, k g h *
Figure 2. Comparison of experimental values of relative viscosity at 2.2,6, and 10"with literature values.
I
c
(with distilled water, 10 and 22% glycerol, and 92%
ethyl alcohol solutions) to ensure that the instrument
was linear in the viscosity range of interest.
A Stanton-type diagram10 was used to assure that all
data reported were in the laminar region of flow.
The excellent repeatability and overall accuracy of
the instrument is a result of careful design, meticulous
cleaning and filling techniques, and rigorous oper-
ational procedures, all of which have been previously
described."
Roll times for all experiments range from 19 to 35 sec.
Average relative standard deviations vary from 0.05 to
0.2% with a mean of 0.1%. Temperatures were
measured to f0.015' and pressures to f1.4 kg/cma.
The density of the fluid was calculated with an em-
1189
SlANLel h BATlEN
A HORNE h JOHNSON 6.2%
'J ZHUZE @I01 5.C
PRESSURE, kolcmr PRESSURE, k ~ b a
pirical formula derived by Wilson and Bradley" and
is accurate to fO.0001 g/cm8. The density of the ball
was calculated from a modification of an equation from
Horne and Johnsona and is accurate to fO.OO1 g/cm?
The absolute accuracy of the calculated viscosities is
better than k0.015 oP.
Results
The results of the present viscosity measurements are
presented in Table I in terms of relative viscosity
(14) W. Wilson and D. Bradley, "Speciflc Volume, Thermal Expan-
sion and Isothermal Compressibility of Sea Water." U. 8. Naval
Ordnance Laboratory Report NOLTR 66-103, June 2, 1966.
Volume Y 8 Number 6 Maa, 1969
1190
where the subscript p refers to the values at pressure
and the subscript 1 to the values at 1 atm. Each
data point shown is an average of a minimum of ten
separate time determinations for both v p and 71.
Collation of these data with those reported by other
investigators is given in Figures 2 and 3. The graphical
limits in these illustrations designate the 99% con-
fidence intervals calculated for the present experi-
mental work; that is, the probability is 0.99 that a
single value of relative viscosity at a given pressure
and temperature will lie between the limits shown.
Discussion
For 2.2" (Figure 2) the present results lie in the
middle of all previously reported data. The best
agreement is with the data of Horne and J o h n ~ o n , ~
Cohen7 a t lo, and Bett and Cappia4 The Bett and
Cappi values were read from a photographic enlarge-
ment of their published graph. Good agreement is
also shown with Tammann and Rabe,9 as reported by
Dorsey,I5 at 0" and below 600 kg/cm2. The poorest
correspondence at this temperature is with Ledere?
and Bridgman4 whose values are high and low, re-
spectively.
The 6" work (Figure 2) shows good agreement with
the 5" work of T. P. Zhuze, V. 1. Sergeevich, and
A. I. Chestnov as quoted by Richardson, el al.,lB but
no agreement with the 6" work of Horne and Johnson.*
At 10" (also Figure 2) the accordance with Bett and
C a ~ p i Bridgman,6
,~ and also Zhuze, Sergeevich, and
Chestnov, as was reported by Richardson, et al,,le
is good. Wonham's2 values, although at 12.65", in-
dicate that if 10" data were available, they would be in
good agreement. Lederer8 follows the present ex-
perimental data up to 500 kg/cm2, beyond which his
values are low. Horne and Johnson3 and Tammann
and Rabe,%as reported by Dorsey,16appear to be in some
agreement with each other, but their values are high as
compared to the present data and those of Bett and
CappL4
At 15" (Figure 3) excellent agreement is shown with
Cohen' while the values of Horne and Johnson8are high
as they were a t 10". Wonham's2 values, although not
exactly at 15", again indicate good agreement.
Cohen,' Weber,5 and Wonham2 agree very well with
the present values a t 20' (Figure 3 ) . Bett and
Cappi4 and Zhuze, Sergeevich, and Chestnov, as
quoted by Richardson, et U Z . , ~ ~only agree up to 700
kg/cm2 beyond which their values are low, while
Horne and Johnsona are again high.
Finally, at 30" (Figure 3) excellent accordance is
again reached with Bett and Cappi4 and Weber.6
Very poor agreement is shown with the work of Bridg-
mant6 Tammann and Rabe,g as reported by Dorsey,15
and Lederer,8 whose values are high and agree with
those of Zhuze, Sergeevich and Chestnov a t 40" as
quoted by Richardson, et ~ 1 . ' ~
The Journal of Physical Chemiatry
AND R. C. BATTEN
E. M. STANLEY
l 4 O 0 7
TEMPERATURE, 'C
Figure 4. Pressure of minimum relative viscosity as a
function of temperature.
In summary, the results of the present experiments
show good agreement with Bett and C a ~ p i Weber,6,~
Wonham,2 Cohen,' and Zhuze, Sergeevich, and
Chestnov as quoted by Richardson, et at.,16 for all
common temperatures and pressures. As for the re-
maining authors, Bridgman's and Horne and Johnson's
experimental procedures and instrumentation have
been examined for possible causes of their variations
from the results reported here. (Lederer's work was
not experimentally determined, and details of the
Tammann and Rabe experimental procedure and
instrumentation were not available.) Deviation of
Bridgman's results in the high-temperature region may
be due to roll timing difficulties and/or variations in the
clearance between the ball and barrel due to pressure.
Horne and Johnson's experimental data, which were
constantly high except for 2.3", may have errors intro-
duced by turbulence and/or leaving the ends of the
barrel open during the rolling of the ball, as opposed to
closing them, the normal procedure.
The minima in the curves of the relative viscosity
(Figures 2 and 3) are a function of two processes: (1)
the breakup of the hydrogen-bonded water, which causes
a decrease in viscosity, and (2) the compaction of the
associated and unassociated water molecules, which
results in an increase in viscosity. A t the pressure
where the relative viscosity is a minimum, the two
(15) N. E. Dorsey, "Properties of Ordinary Water-Substance,"
Reinhold Publishing Corp., New York, N. Y., 1940.
(16) J. L. Richardson, P. Bergsteinsson, R. J. Geta, D. L. Peters,
and R. W. Sprague, "Sea Water Mass Diffusion Coefficient Studies,"
Philco Corp. Publication No. U-3021 (Feb 26, 1961), Office of Naval
Research Contract No. Nom-4061 (00) I
VISCOSITYOF WATERAT HIGH PRESSURES 1191

processes have equal and opposite effects. Figure 4 is 5.0 , I , 1 1 1 1 1 1 1 1 I

a plot of the pressure of minimum viscosity as a func-
tion of temperature. It shows that the pressure of 4.8- -
minimum relative viscosity decreases with an increase
in temperature. If the curve is extrapolated (dashed
line), it intersects the zero pressure (atmospheric) axis
4.6 - -
at about 36". The temperature of 36" agrees well with
the temperature of 35" quoted by Franks and Good"
4.4 - -
where they postulated that at 1 atm water undergoes
a change from a quasi-crystalline structure to a sus-
f5 4.2- -
pension of clusters.
The present data are plotted with isobars of relative 2 4.0,
viscosity as a function of temperature (Figure 5).
The dashed curve represents the data at 1500 kg/cm2
from a similar graph of Horne and Johnson.a The
3.8 -
lack of sharp changes in slope in this graph near 4",
such as illustrated by the dashed curve of Horne and
3.6 -.
Johnson,a indicate that the breakup of the structured
regions of water due to temperature is a gradual process 3A I l ~ ~ i c ~ ~ r l l l l ! l
0 2 4 6 8 10 12 14 I6 18 20 22 24 26 28 $0
and supports the calculations of Nemethy and
TEMPERATURE,'%
Scheraga.l*
Figure 6 is a graph of the isobars of the calculated Figure 6. Isobars of activation energy of viscous flow of water
activation energies of viscous flow, E,, as a function of (kcal/mol) as function Of temperature+

temperature. This plot shows a marked difference in

features a t low temperatures when compared to the
results of Horne and J ~ h n s o n ,represented
~ by the
dashed line.lg At low temperatures no sharp increase
in activation energies is observed, but a small relative
minimum is present in the data taken at low pressure.
Some small fluctuations are also observed between 16
and 20" a t high pressure.
Although the curves of E , at pressure are not as
smooth as E , for atmospheric pressure, data values of
the activation energies of viscous flow decrease with
increasing pressure and temperature in a uniform
manner and tend to follow the shape of the atmospheric
pressure curve.
At present no explanation can be given for the
relative minima of the activation energies at low
temperature. However, it should be noted that (1)
the data used and calculations of E , are not precise
enough (standard deviations of f 0 . 1 kcal/mol) to
establish the absolute existence of the minima at the
low temperature and (2) the relative minima do not
correlate with the decrease in the temperature of
maximum density at pressure as quoted by Dorsey.16
Therefore, a more detailed and accurate investigation
should be undertaken in this low-temperature region a t
pressure.

(17) F. Franks and W. Good, Nature, 210, 85 (1966).

(18) G. Nemethy and H. A. Scheraga, J. Chem. Phys., 36, 3382
5 10 15 20 25 30 (1962).
TEMPERATURE, "C (19) The sharp changes in slope of the curye of Horne and Johnson
at low temperature for Figures 5 and 6 are caused by the high wp/71
Figure 5. Isobars of relative viscosity as a function of pressure. values at 4' and above (see Figures 2 and 3).

Volume 78,Number 6 May lQ80