Professional Documents
Culture Documents
Viscosity-Temperature Correlation For Liquids
Viscosity-Temperature Correlation For Liquids
Received 9 March 2006; accepted 19 April 2006; published online 7 June 2006
A new viscosity–temperature equation and corresponding chart have been developed to extend the range of the current ASTM
viscosity–temperature charts. This new chart and equation extends the temperature and viscosity range for hydrocarbons and, for
the first time, has the ability to extend to the low viscosity regime of halocarbons and low temperature fluids. The new equation and
chart can linearize liquid viscosity data from 0.04 cSt and covers the temperature range from )210 to 500 C for halocarbons and
hydrocarbons. With a modification to the temperature scaling, the new equation also has the ability to fit liquid metal viscosity
data. The new chart and equation cannot accurately linearize the viscosity with respect to temperature of fluids exhibiting strong
molecular bonding (water, ammonia), fluids whose molecular structure consists of long coils (some long chained silicones), or fluid
mixtures in which one fluid precipitates out of solution (wax precipitation).
KEY WORDS: viscosity, temperature, hydrocarbon, halocarbon, metal
1. Introduction 2. Background
The purpose of this work is to build upon the In an attempt to correct problems that have perpe-
knowledge of our predecessors and provide an improved trated through history, a critical literature review is
chart that will linearize the liquid viscosity of hydro- presented to correct the forms of the equations and to
carbons, halocarbons and liquid metals over an ex- correctly reference the authors of those works. Table 1
tended temperature and viscosity range. In 1886, contains the historical development of viscosity–tem-
Osborne Reynolds [1] wrote ‘‘On the Theory of Lubri- perature relations in their original nomenclature. In
cation and Its Application to Mr. Beauchamp TowerÕs 1886, Osborne Reynolds found it necessary to know the
Experiments, including an Experimental Determination viscosity of olive oil as a function of temperature and
of the Viscosity of Olive Oil’’ and started the process of used a simplistic exponential equation to fit the available
scaling the coordinate system to plot liquid viscosities as data over a small temperature range (16–49 C). In this
straight lines. The current ASTM standard viscosity paper, Reynolds formulated and integrated the hydro-
charts, ASTM 341-93 [2], are based on the work of dynamic equations that define hydrodynamic lubrica-
Wright [3], which improved the linearization of hydro- tion, and derived the Reynolds equation. In his lubricant
carbons down to 0.18 cSt on charts available from experiments, Reynolds noted that as the load on the
ASTM and to 0.21 cSt with the developed equation in bearing was increased so did the lubricant temperature.
the standard. The new hydrocarbon and halocarbon The temperature increase generated a new geometric
chart and equation described in this paper will now stability, which is based upon a decrease in the fluid
extend from the low temperature realm of liquid meth- viscosity with increasing temperature. It was therefore
ane to the high temperature, high viscosity realm of necessary for the interpretation of the results to know
petroleum oils. The new chart utilizes the historical how the viscosity of the lubricant varied with respect to
work at high viscosities and extends the linearized low the temperature. At this point in history fluid dynamics
viscosity range to 0.04 cSt and below. Using the his- were only starting to become understood; Reynolds
torical volume based mixing rules, the new chart also proposed that fluids have two viscosities, one viscosity
allows for calculations of the viscosity of mixtures. An for what we now know to be laminar flow and another
alternate scaling of the temperature axis linearizes the for turbulent flow to answer problems being probed by
viscosity of metals. Stokes, Poiseuille, and Darcy.
One of the most misreferenced viscosity equations is
that of Vogel [4], which represents the absolute viscosity
as:
tt1
*To whom correspondence should be addressed.
E-mail: seeton@uiuc.edu gt ¼ gtt
1
1
ð1Þ
Table 1.
Historical development of viscosity–temperature relations in original nomenclature.
t, temperature (C) T, temperature (Kelvin or Rankine).g, go & l, absolute viscosity (mPa-s, cP) m, kinematic viscosity (mm2/s, cSt).R, a, t1, t¥,
g¥, M, P, m, mo, To, a, b, b0, b1, b¢, c, K, h, b, c, A, B, So, Go are empirical constants.
where gt is the viscosity at a specified temperature t, g¥ Also in 1921, The Texas Company (Texaco) pub-
is the viscosity as t fi ¥, t1 is the temperature where lished a viscosity–temperature chart developed by
g = 1, and t¥ is the temperature where the g fi ¥. N. MacCoull in the June edition of Lubrication [15].
While Vogel does not note the units used, his choice of This chart linearizes different cuts of crude oil from their
fluids (water, mercury, and petroleum oils) indicates that cloud point to 40 Saybolt Universal (about 4 cSt).
temperature is in degrees Celsius and viscosity is in MacCoullÕs chart also appears in the 1927 International
centipoise (cP). This equation performs reasonably well Critical Tables [16] with dual axes for Saybolt Universal
over a moderate temperature range; however, its non- and centistokes. Neither the Lubrication publication nor
linear nature and three fluid specific coefficients make it the International Critical Tables gives the method or
difficult for generalization. Most referenced works list equation used to determine the scale of the axes.
VogelÕs equation as: In 1928, Walther [6] introduced an equation credited
to Le Chatelier.
l ¼ KeðtþhÞ
b
ð2Þ
logðlogðlÞÞ ¼ P M t ð4Þ
or
B where l is the absolute viscosity in centipoise (cP), t is
loge ðlÞ ¼ A þ ð3Þ the temperature in degrees Celsius (C). WaltherÕs key
T T0
point is that given two measured points and finding P
which found in [10] and has been carried forward and M, calculation of the viscosity allows for estimation
through the lubrication literature; however, these equa- at other temperatures. Unfortunately, in this equation,
tions are not equivalent to VogelÕs [4] equation. Fulcher viscosities of 1.0 or less cause an undefined double log-
[5] is the first to list equation (3) in the analysis of the arithm, and exposes the main drawback of all the double
viscosity of molten glasses. Scherer [14] reviews the his- logarithm viscosity–temperature relations that have
tory behind the ‘‘Fulcher equation’’ or more standard followed to this date. In 1931, Walther [7,8] published
designation as the VFT (Vogel–Fulcher–Tammann) two additional papers where he changed to correlating
equation. As noted by Scherer, the VFT equation works the kinematic viscosity (m) with temperature and cor-
well over wide viscosity range for glasses and many liq- rected for the undefined double logarithm by adding a
uids, however, like the Vogel equation, its three fluid constant of 0.95 to the viscosity allowing viscosities
specific coefficients make it difficult for generalization. down to 0.05 cSt to be correlated.
C.J. Seeton/Viscosity–temperature correlation for liquids 69
T where A = 1, a = 0.8, and b and c are constants for
logðlogðm þ 0:95ÞÞ ¼ m log þ logðlogðmo þ 0:95ÞÞ each oil. Here the additive constant, a, is left as it was
To
approved by Geniesse and Delbridge. Barr, realizing
ð5Þ that there is an inadequacy in the formula at high tem-
The simplification of the above equation has historically peratures and low viscosities, attempts a different
been named the ‘‘Walther formula’’ and is of the form1: correlation
Tribology, if D is equal to 10 or e and b is the second Manning [13] simplified WrightÕs complex form of
constant, an incorrect representation of the Walther f(m) by replacing it with a polynomial and extended the
formula is formulated. Crouch and Cameron compare chartable (not linear, due to its basis on WrightÕs work)
other viscosity–temperature relationships before range down to 0.12 cSt.
attempting to modify the ‘‘Vogel’’ equation, as pre-
sented by Barr, with a third constant in developing a log log m þ 0:7 þ exp 1:47 1:84m 0:51m2
ð16Þ
generalized equation. ¼ A B logðTÞ
Roelands [12] developed correlations for the viscos-
ity–temperature and viscosity–pressure relationships. While ManningÕs new polynomial provides a simplifi-
RoelandsÕ viscosity-temperature correlates the absolute cation of WrightÕs complex form, it also exhibits non-
viscosity in similar form to the Walther formula: linear nature below 0.18 cSt due to its derivation.
Manning also provides an approximation to find the
T viscosity knowing the A and B coefficients for the fluid.
log ðlog ðgo þ 4:2ÞÞ ¼ So log 1 þ þ log ðGo Þ
135
m ¼ðZ 0:7Þ exp ½0:7487 3:295ðZ 0:7Þ
ð12Þ i ð17Þ
þ 0:6119ðZ 0:7Þ2 0:3193ðZ 0:7Þ3
where So and Go are fluid specific constants. This for-
mulation performs over a narrow temperature range; where
however, viscosity correlation breaks down over an ex-
tended temperature range. Recently, the RoelandsÕ vis- logðlogðZÞÞ ¼ A B logðTÞ ð18Þ
cosity–pressure correlation is being scrutinized due to a
Figure 1 charts the viscosity of pure HFC134a, CO2,
truncation of data in its derivation of the high-pressure
propane, HCFC123 and mixture HFC410A. These
regime [21].
refrigerants demonstrate nearly linear nature above
The most significant modification to the Walther
0.21 cSt, and the linear correlation breaks down into
formula and MacCoull charts came in 1969. Wright [3]
nonlinear behavior below 0.21 cSt, demonstrative of the
extended the ability to linearize the charting viscosity of
Wright and Manning modifications.
hydrocarbons over the temperature range )73–371 C
In order to understand the viscosity–temperature
()100–700 F) and viscosities down to 0.21 cSt (0.18 cSt
relationships of Vogel, VFT, Walther, and Wright, it is
in charts available from ASTM). In order to linearize
useful to illustrate their boundaries by selecting the
hydrocarbon data over the temperature range, Wright
saturated liquid viscosity of propane ranging from the
developed a form of the Walther formula where a
triple point to the critical point, [22]. The Vogel equa-
function of viscosity f(v) was added to the standard
tion, figure 2(a), is able to represent the liquid viscosity
historical constant, c = 0.7.
over a limited temperature range, however, as the
log10 ðlog10 ðm þ c þ fðmÞÞÞ ¼ A B log10 ðTÞ ð13Þ temperature approaches the critical point, the formu-
lation is not able to accurately represent the data. The
where VFT equation, figure 2(b), is also able to represent
c þ fðmÞ ¼ 0:7 þ C D þ E F þ G H ð14Þ the viscosity over a limited temperature range; however,
the representation falls off near the triple and critical
and points. While there is not a limit in the temperature
field, the Walther formula, figure 2(c), only approxi-
C ¼ exp ð1:14883 2:65868mÞ; mates the viscosity of propane due to its limitation of
the viscosity approaches 0.2 cSt as T޴. The Wright
D ¼ exp ð0:00381308 12:5645mÞ
modification to the Walther formula, figure 2(d), does
E ¼ exp ð5:46491 37:6289mÞ; not exhibit an asymptotic behavior as the temperature
ð15Þ
F ¼ exp ð13:0458 74:6851mÞ increases and is able to represent the saturated liquid
G ¼ exp ð37:4619 192:643mÞ; viscosity above 0.21 cSt; however, it can not be used
below a viscosity of 0.21 cSt. The purpose of this work
H ¼ exp ð80:4945 400:468mÞ
is to find a generalized equation that can be utilized
As with all the previous double log forms, once the term over the entire temperature and viscosity range,
in the first log term becomes zero, the second log is figure 2(e).
undefined; this occurs at a viscosity of 0.21 cSt with the
Wright formulation. While this equation will not allow
the correlation of viscosities below 0.21 cSt, it performs
3. Development of a new improved generalized
well with hydrocarbons above that threshold. The cur-
formulation
rent state of the art is still based entirely on WrightÕs
modification to the Walther formula and its charts re- It is desired to construct a generalized chart and
flect the improvement. mathematical formulation that is able to:
C.J. Seeton/Viscosity–temperature correlation for liquids 71
3
2
1
0.9
0.8
0.7
0.6
0.5
0.4
Viscosity (cSt)
0.3
0.28
0.26
0.24
0.22
0.2
0.18
0.16
0.14
R123
R32 Propane
CO 2
R134a
0.12
–60
–50
–40
–30
–20
–10
10
20
30
40
50
60
70
80
90
100
110
120
130
Temperature(oC)
Figure 1. ASTM standard D 341-93 (m ‡ 0.21); Manning equation (m < 0.21).
precedence. The new modification must also linearize This scaling condition on viscosity allows k and w to be
hydrocarbon fluids in the range of 0.04 cSt to their glass dependent and the value of k was optimized using his-
transition temperature, and incorporate those hydrocar- torical lubricant and current fluid viscosity data sets.
bons typically seen in natural gas production. The zero The optimized constant was determined to be the
order modified Bessel function of the second kind, K0, was historical constant, k = 0.7. Which defines w as:
chosen to be the key component of the additive function
in the log–log form. The zero order modified Bessel k ¼ 0:7 ! w ¼ 1:244066584703
function of the second kind has the form:
The new proposed function in its generalized form be-
Z 1
cos ðxtÞdt come:
K0 ðxÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffi ð19Þ
0 t2 þ 1 loge ðloge ðm þ 0:7 þ em K0 ðm þ 1:244067ÞÞÞ
ð24Þ
The new function for the viscosity takes the same form ¼ A B loge ðTÞ
as WrightÕs function:
0TB
loge ðloge ðm þ k þ fðmÞÞÞ ¼ A B loge ðTÞ ð20Þ m þ 0:7 þ em K0 ðm þ 1:244067Þ ¼ eA ð25Þ
However, the function of viscosity is now changed where
in order to extend the viscosity range to ultra-low
viscosities: A ¼ loge ðA0 Þ ð26Þ
72 C.J. Seeton/Viscosity–temperature correlation for liquids
100 100
t +140.94
1296.53
Log e ( m ) = –5.714 +
ht = 0.021 t + 263.14
T + 363.7
1 1
t=t t=t1
0.1 0.1
h=h •
0.01 0.01
–273.15 –223.15 –173.15 –123.15 –73.15 –23.15 26.85 76.85 126.85 –273.15 –223.15 –173.15 –123.15 –73.15 –23.15 26.85 76.85 126.85
100 100
Log10 ( Log10 (n+ 0.8 ) ) = 7.312 3.663* Log10 (T ) (
Log10 Log10 (n + l + f (n ))) = 6.104 – 3.099 * Log10 (T )
Kinematic Viscosity (cSt)
1 1
0.1 0.1
0.01 0.01
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Temperature (K) Temperature (K)
c.Walther Formula d. Wright Equation
100
(
Log e Log e (n + l + f (n))) = 13.545 2.821* Log e (T )
Kinematic Viscosity (cSt)
l+ f (n) = 0.7 + e– nK 0 (n + y)
10
0.1
0.01
0 50 100 150 200 250 300 350 400
Temperature (K)
e. Proposed Fit
Figure 2. Viscosity of propane from triple point to critical point, data from Vogel et al. [22].
In comparison, the historical additive constants and be linearized over their entire temperature range, as
additive functions, as well as, the proposed additive shown in figure 7.
function are shown in figure 3. This proposed viscosity–
temperature relationship is compared to the Wright
equation in table 2 and figure 4. It is shown that the new 3.1. Mixing rule for liquids
and old equations perform equally well with fluids
above 0.21 cSt. Utilizing liquid viscosity data found at Traditionally, the mixing rule for the Walther for-
mula and MacCoull charts has been a volume fraction
the NIST Webbook [23] and extending the range of
based rule.
viscosities to the low-viscosity region yields a chart of
X
refrigerants, table 3 and figure 5; extending to low X¼ Vi ðAi Bi loge ðTÞÞ ð27Þ
temperature fluids yields figure 6. Due to their strong
hydrogen bonding, water and ammonia are not able to where
C.J. Seeton/Viscosity–temperature correlation for liquids 73
0.95
D
0.9
0.85 C
0.8 A
g + f( n )
0.75 B
0.7
0.65
0.6
0.55
0.5
0.01 0.1 1 10
Viscosity (cSt)
Figure 3. Additive function to log–log viscosity–temperature correlations (a). Geniesse and Delbridge, (b) ASTM D341–43, (c) Wright, (d)
proposed formulation.
X ¼ loge ðloge ðmmix þ k þ emmix K0 ðmmix þ wÞÞÞ ð28Þ where wi is the mass fraction of each fluid, and qi is
the density of each fluid varying with temperature.
Vi is the volume fraction of each fluid, and Ai and Bi are To simplify the above equation, if the densities of
the intercept and slope of the fluid on the new chartÕs the fluids and their mixture are assumed to be
transformed coordinate system, and k and w are con- nearly equal, the density ratio becomes unity and
stants defined above. This becomes convenient as the only the mass fraction is used. In another simplifi-
mixture viscosity can be estimated when two data points cation, if the densities of the individual fluids are
for each fluid in the mixture are known. different and ideal mixing assumed, then one can
The mixture viscosity, as a function using the mass write:
fraction, becomes:
X X 1
qmix ðTÞ wi
X¼ wi ðAi Bi loge ðTÞÞ ð29Þ qmix ðTÞ ¼ ð30Þ
qi ðTÞ qi ðTÞ
Table 2.
Comparison of Wright Equation to the proposed equation.
A. Cannon instrument Co. Calibration fluid K3A 233–423 10 9.920 0.99987 23.509 0.99971
4.076 4.047
B. Cannon instrument Co. Calibration fluid K200A 293–423 8 8.885 0.999992 21.294 0.999992
3.419 3.419
C. Koehler instrument Co. Calibration fluid S6S 293–423 8 9.528 0.999995 22.676 0.999997
3.855 3.838
D. Polyethylene glycol 200 [15] 233–588 15 10.277 0.9997 24.310 0.99985
4.063 4.030
E. Petroleum oil [15] Navy symbol 2135 255–644 14 10.631 0.99986 25.038 0.99996
4.158 4.111
F. Petroleum oil [15] Grade 1120 273–644 13 8.874 0.99967 21.208 0.99977
3.394 3.384
G. Halocarbon flurolube [15] Light grease 372–588 9 12.909 0.99886 30.409 0.99911
4.911 4.886
a
To compare the intercept of the new viscosity-temperature relationship to that of the Wright modification, divide by Loge(10) 2.303
74 C.J. Seeton/Viscosity–temperature correlation for liquids
10000
4000
2000
1000
500
300
200
100
60
40
30
20
10
8
7
6
5
4 B
Viscosity (cSt)
1 F
0.9 C
0.8 G
0.7
0.6 A
0.5 D
0.4 E
0.3
–40
–30
–20
–10
0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
170
190
210
230
250
270
290
310
330
350
370
390
Temperature(oC)
Figure 4. Data of different oils on proposed viscosity–temperature chart, table 2.
Table 3.
Liquid viscosity of halocarbons and low temperature fluids. Fluid data from webbook.nist.gov
Fluid Temp. range (K) Number of data points Proposed Eq. A B Proposed Eq. r2
1
0.9
0.8
0.7
0.6
0.5
Viscosity (cSt)
0.4
0.3
0.28
0.26
0.24
0.22
0.2
0.18
0.16
0.14
0.12
R123
0.1
Propane
0.08
CO2 R32 R134a
0.06
0.04
–60
–50
–40
–30
–20
–10
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
Temperature(oC)
Figure 5. Proposed viscosity chart with selected hydrocarbons and halocarbons. Fluid data from webbook.nist.gov.
10
8
6 Propane
5
4
3
2
Ethane
1
0.9
0.8
0.7
Viscosity (cSt)
0.6
0.5
0.4
Nitrogen Methane
0.3
0.28
0.26
0.24
0.22 Argon
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
–
– 210
– 200
– 190
– 180
– 170
– 160
– 150
– 140
– 130
– 120
– 110
– 100
– 90
– 80
– 70
– 60
– 50
– 40
– 30
– 20
220
Temperature ( oC)
Figure 6. Proposed viscosity chart extended to low temperatures. Fluid data from webbook.nist.gov.
76 C.J. Seeton/Viscosity–temperature correlation for liquids
1
0.9
0.8
0.7
0.6
0.5
0.4
Viscosity (cSt)
0.3
0.28
0.26
0.24
0.22
0.2
0.18
0.16
0.14
0.12
NH3 H2O
0.1
0.08
0.06
0.04
–90
–80
–70
–60
–50
–40
–30
–20
–10
0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
170
190
210
230
250
270
290
310
330
350
370
390
Temperature (oC)
Figure 7. Viscosity of ammonia and water from the triple point to the critical point. Fluid data from webbook.nist.gov.
0.6
0.5
0.4
0.3
0.28
Viscosity (cSt)
0.26
0.24
0.22
0.2
0.18
0.16
0.14
0.12
0.1 R134a
0.08
0.06
–60
–50
–40
–30
–20
–10
10
20
30
40
50
60
70
80
90
100
Temperature (o C)
Figure 8. Refrigerant Mixture HFC407C. Dashed line represents actual HFC407C viscosity. Individual fluid data from webbook.nist.gov.
R407C mixture data from NIST REFPROP v.7.
C.J. Seeton/Viscosity–temperature correlation for liquids 77
Table 4. Table 5.
Components of HFC407C, figure 8. Viscosity of liquid metals, figure 9.
1
0.9
0.8 Sodium
0.7 Potassium
0.6
0.5
Viscosity (cSt)
0.4
Gallium
0.3
0.28
0.26
0.24
0.22 Lea d
0.2
0.18 Me rcury Tin
0.16
0.14
0.12
0.1
0.08
0.06
0
–50
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
800
900
1000
1100
1200
1300
Temperature (o C)
Figure 9. Viscosity of liquid metals, table 5.
78 C.J. Seeton/Viscosity–temperature correlation for liquids