Energy & Fuels 2005, 19, 1935-1947

1935

Experimental Studies of Jet Fuel Viscosity at Low

Temperatures, Using a Rotational Viscometer and an
Optical Cell
Daniel L. Atkins and Jamie S. Ervin*
University of Dayton, Mechanical and Aerospace Engineering Department, 300 College Park,
Dayton, Ohio 45469-0210

Linda Shafer
University of Dayton Research Institute, 300 College Park, Dayton, Ohio 45469-0116
Received December 3, 2004. Revised Manuscript Received May 18, 2005

A temperature scanning rotational viscometer and visualization techniques are utilized to

evaluate the viscosity behavior of kerosene-based jet fuels from -20 C to sample cloud point
temperatures. The uncertainty of using the rotational viscometer technique is evaluated and
determined to be acceptable. The rotational viscometer is used to perform shear rate studies on
JP-8, Jet A, and JP-8 blended with a low-temperature additive designed to improve fluidity.
Clouding of the neat fuel samples during cooling, visualized in a unique optical cell using crosspolarized light and a He-Ne laser, was determined to not cause non-Newtonian viscosity behavior.
Polarized light is used to demonstrate that neat fuel clouding during cooling is likely due to free
droplets of a water/di-ethylene glycol monomethyl ether (water/DIEGE) mixture, while clouding
of the jet fuel/additive blend was caused by crystallization. Viscosity results are presented for all
jet fuel samples. Selected results are compared with viscosity measurements obtained from other
methods and are determined to be in good agreement. Complex non-Newtonian viscosity behavior,
including both pseudo-plasticity and dilatancy, was identified for the fuel/additive blend. Viscosity
results and the effects of shear rate changes matched observed solidification behavior.

Introduction
Operators and designers of aircraft, both manned and
unmanned, that fly at altitudes above 12 000 m for
extended periods require high-fidelity fuel freezing
prediction tools. An impediment to the understanding
and simulation of jet fuel freezing has been the absence
of published thermophysical properties of jet fuel at low
temperatures. Common jet fuels, such as Jet A, Jet A-1,
and JP-8, have little published property measurements
at temperatures near their freeze point temperatures.
The freeze point for a hydrocarbon liquid fuel (ASTM
D2386-03) represents the temperature at which visible
wax crystals melt upon warming. Therefore, jet fuel
consists mostly of liquid several degrees (3-4 C)
below its freeze point. The absence of published property
data is likely due to the variation of jet fuels, based on
the composition of the original crude oil and the refinery
methods used. Also, property measurements at temperatures approaching the onset of solidification are difficult to obtain, requiring specialized equipment and
techniques. Commercial and military jet fuel specifications are not sufficiently rigorous to ensure consistent
thermophysical properties. Moreover, the introduction
of non-petroleum-derived fuels, such as Fischer-Tropsch fuels, are on the horizon. Therefore, jet fuel properties must be assessed regularly and new techniques
* Author to whom correspondence should be addressed. Telephone:
(937) 252-8878, ext. 114. Fax: (937) 252-9917. E-mail address:
jamie.ervin@notes.udayton.edu.

developed to determine their low-temperature properties accurately.

The Coordinating Research Council (CRC) has provided a compilation of aviation fuel properties that are
widely used in the aerospace industry.1 Unfortunately,
these properties are incomplete for temperatures near
jet fuel freeze point temperatures. The low-temperature
properties of jet fuels must be known to model and
design aircraft fuel systems adequately. At present, the
specified freeze point temperature and viscosity at a
single temperature are used by aircraft designers and
operators to determine a fuels ability to be pumped at
lower temperatures. The usefulness of the specified
viscosity requirement and freeze point temperature is
limited because jet fuel is a liquid below its freeze point
temperature. In addition, fuel viscosity greatly impacts
the ability to re-light an engine at low temperature.
More-complete data are needed for improved combustor
design. The present specification method of determining
the viscosity of jet fuel is the capillary method (ASTM
D445). Continuous viscosity measurements of jet fuel,
as a function of temperature, would provide important
information on actual fuel flow characteristics at temperatures near the onset of crystallization.
Viscosity measurements at varying shear rates could
reveal possible non-Newtonian behavior caused by
forming wax or water crystals. Fuel clouding behavior
(1) Handbook of Aviation Fuel Properties, Coordinating Research
Council Technical Report 635, Atlanta, GA, 2004.

10.1021/ef049683k CCC: $30.25 2005 American Chemical Society

Published on Web 06/28/2005

1936

Energy & Fuels, Vol. 19, No. 5, 2005

upon cooling, which is caused by precipitating water or

other phenomena (e.g., large n-alkane gelation), may
introduce non-Newtonian behavior. Two-phase slurries
(wax particles suspended in liquid fuel), emulsions, or
dispersions may change the flow properties from Newtonian to non-Newtonian behavior.2 The effects of fuel
additives on flow characteristics could also be determined. The proposed method utilizes a temperature
scanning rotational viscometer (ASTM D5133-01), which
has been applied to low-temperature automotive lubricants, to investigate low-temperature jet fuel fluidity.3
The use of a rotational viscometer to determine the
low-temperature viscosity of jet fuels has been limited
but is emerging as a new technique.4 The viscosity of
Jet A has been measured with a rotational viscometer
at temperatures as low as -50 C; however, these
measurements did not include shear rate studies or the
effects of additives. Moreover, variations in shear rate
were required to maintain instrument scale, concealing
potential non-Newtonian behavior.5 Other researchers
investigated the viscosity of Russian kerosene-based jet
fuels.6-11 For example, Mitin et al.6 used a kerosenebased thermally stable fuel (T-6) that was designed for
high-performance aircraft in a rotational viscometer.
They concluded that the fuel is non-Newtonian below
0 C, because of the formation of rheological structures. T-6 has greater density than JP-8, because it
consists mainly of cycloparaffins, reducing the lowerdensity n-alkane content.12 Gorenkov et al.7 measured
the viscosity of Russian jet fuels using the capillary
method (GOST 33-82) and identified the lack of continuous viscosity measurements at low temperatures.
Likhterova et al.8 also identifies the lack of viscosity
measurements at low temperatures and suggests that
jet fuels may be non-Newtonian near the initial crystallization temperature. In this study, a rotational viscometer is used to study flow characteristics of Russian
kerosene-based jet fuels at low temperatures.8 Unfortunately, this research provides data on only Russian
fuels, lacks experimental detail, has translation issues
(2) Pedersen, K. S.; Ronningsen, H. P. Energy Fuels 2000, 14, 4351.
(3) Standard Test Method for Low Temperature, Low Shear Rate,
Viscosity/Temperature Dependence of Lubricating Oils Using a Temperature-Scanning Technique, ASTM D5133-01; American Society of
Testing Materials: West Conshohocken, PA. Available via the Internet
at URL: http://www.astm.org [cited June 2004].
(4) Selby, T. W. The Scanning Brookfield Technique of LowTemperature, Low-Shear RheologysIts Inception, Development, and
Applications, ASTM STP 1143; American Society for Testing and
Materials: Philadelphia, PA, 1992; pp 33-64.
(5) Stockemer, F. J. NASA Technical Report CR-159615, National
Aeronautics and Space Administration, Washington, DC, 1979.
(6) Mitin, M. B.; Lyashenko, B. N.; Karpov, V. A.; Goncharov, L. A.;
Kutuzova, O. V. Chem. Technol. Fuels Oils (English translation of
Khim. Tekhnol. Topl. Masel) 1989, 24, 219-221.
(7) Gorenkov, A. F.; Lifanova, T. A.; Klyuchko, I. G.; Saleev, V. A.
Chem. Technol. Fuels Oils (English translation of Khim. Tekhnol. Topl.
Masel) 1983, 19, 297-299.
(8) Likhterova, N. M.; Gorenkov, A. F.; Perepelkina, O. A. Chem.
Technol. Fuels Oils (English translation of Khim. Tekhnol. Topl. Masel)
1983, 19, 599-602.
(9) Belousov, A. I.; Bushueva, E. M.; Rudyakov, D. G. Chem. Technol.
Fuels Oils (English translation of Khim. Tekhnol. Topl. Masel) 1984,
20, 79-80.
(10) Dubovkin, N. F.; Malanicheva, V. G. Chem. Technol. Fuels Oils
(English translation of Khim. Tekhnol. Topl. Masel) 1980, 16, 543547.
(11) Gorenkov, A. F.; Stepanenko, V. S.; Saleev, V. A. Chem. Technol.
Fuels Oils (English translation of Khim. Tekhnol. Topl. Masel) 1982,
18, 94-98.
(12) Hanson, F. V. Air Force Wright Research and Development
Center Technical Report No. WRDC-TR-89-2097, 1989.

Atkins et al.
Table 1. Volume Percentages of Species Classes within
Four Fuel Samples, as Determined by ASTM 2425
Composition of Jet A Fuel (vol %)
species classes

F3686

F3688

F3773

F3804

paraffins
cycloparaffins
dicycloparaffins
tricycloparaffins
alkylbenzenes
indan and tetralins
indenes, CnH2n-10
naphthalene
naphthalenes
acenaphthenes
acenaphthylenes
tricyclic aromatics

43.6
20.0
11.7
0.9
12.7
6.6
0.3
0.4
3.5
0.3
<0.2
<0.2

60.5
15.4
4.3
<0.2
13.7
4.3
<0.2
<0.2
1.5
<0.2
<0.2
<0.2

57.2
17.4
6.1
0.6
13.5
3.4
<0.2
<0.2
1.7
<0.2
<0.2
<0.2

58.5
15.7
3.7
<0.2
16.9
3.2
<0.2
<0.2
2.0
<0.2
<0.2
<0.2

100.0

100.0

100.0

100.0

total

Table 2. Volume Percentages of Species Classes within

Four Fuel Samples, as Determined by ASTM 6379
Composition of Jet A Fuel (vol %)
species classes

F3686

F3688

F3773

F3804

aromatics
monoaromatics
diaromatics
total
total saturates

19.9
4.0
23.9
76.1

17.8
2.0
19.8
80.2

16.6
2.1
18.7
81.3

19.9
2.3
22.2
77.8

and conflicting non-Newtonian viscosity results, and

does not consider the effect of additives.
In this paper, the rotational viscometer method is
used with commonly employed commercial and military
jet fuels (Jet A and JP-8) at low temperatures approaching initial observable crystallization (the cloud
point temperature). The objective is to determine the
applicability of the technique to jet fuel viscosity measurement and shear rate studies. We investigate the
Newtonian and non-Newtonian viscosity behavior before
initial crystallization, because the previous research is
incomplete and inconsistent. The viscosity behavior of
fuels mixed with additives that improve low-temperature fluidity is also studied, and any resulting nonNewtonian behavior is characterized. Visualization
techniques are used to investigate clouding behavior and
to determine if the phenomena responsible for clouding
cause non-Newtonian behavior.
Experimental Section
JP-8 is the current primary jet fuel used by the United
States Air Force and is essentially Jet A-1 fuel that has been
combined with three additives. The first additive is a corrosion
inhibitor that also improves lubricity. The second is a fuel
system icing inhibitor, and the third is a static dissipater. Jet
A-1 is a commercial jet fuel that is used internationally,
whereas Jet A is the primary U.S. commercial jet fuel.
Therefore, JP-8 and Jet A are representative of extensively
used jet fuels and were selected for analysis in the present
work. Jet fuel samples for the present work were obtained from
the Fuels Branch of the Turbine Engine Division of the
Propulsion Directorate of the Air Force Research Laboratory
at Wright-Patterson Air Force Base. In addition, a proprietary
low-temperature additive was used in some experiments.
Samples are identified by an alphabetic character, F or A,
which designates fuel or additive, followed by four identification digits. The digits have no meaning other than to identify
the sample.
Table 1 lists the compositions (ASTM 2425) by species
classes of the four fuel samples studied in this work. In
addition, Table 2 describes the fuel composition (on a volumetric basis, ASTM 6379) in terms of the monoaromatics,

Studies of Jet Fuel Viscosity at Low Temperatures

Energy & Fuels, Vol. 19, No. 5, 2005 1937

Figure 1. Measured distribution of n-alkanes for JP-8 and Jet A samples.

Table 3. Selected Low-Temperature Properties of Jet Fuel Samples

fuel

freeze point
((0.5 C)

cloud point
((1.0 C)

pour point
((2.0 C)

JP-8 (F3804)
JP-8 (F3773)
Jet A (F3686)
Jet A (F3688)

-48.3
-49.4
-42.0
-41.6

-52.6
-53.7
-46.2
-46.5

-58.0
-58.0

Density, F
((0.001 g/mL)
@ -20 C
@ -40 C

Kinematic Viscosity,
((0.5 cSt)
@ -20 C
@ -40 C

Viscosity,
((0.5 cP)
@ -20 C
@ -40 C

0.822

0.834

3.9

8.4

3.2

7.0

0.824

0.833

4.9

11.9

4.0

9.3

diaromatics, total aromatics, and total saturates. In lowtemperature studies of jet fuel, the freeze point temperature
is a parameter commonly used to characterize the fluidity of
the fuel. Table 3 lists both the measured freeze and cloud point
temperatures for the fuel samples (using a Phase Technology
Series 70V petroleum analyzer; ASTM D5972-96). For the
temperature regime in which fuel crystallization occurs, the
incipient solid crystals consist of nonideal solid solutions of
the larger n-alkanes (such as C16-C19) within the fuel.13 These
large n-alkanes represent a small fraction of the total n-alkane
concentration. Thus, the precipitation of n-alkanes from the
fuel (cloud point), and, therefore, the freeze point, are dependent on the n-alkane distribution of these species. Figure 1
shows the n-alkane distribution for the fuel samples considered. In addition, Figure 1 shows that fuels F3688 and F3686
have n-alkane distributions that have a greater fraction of the
larger n-alkanes (C16-C19). Therefore, it may be anticipated
that they would have the highest freeze point temperatures,
as demonstrated by Table 3.
The n-alkane analyses of the fuels were performed using
gas chromatography (Agilent model 6890) combined with mass
spectrometry (GC/MS) and flame ionization detection (GC/
FID). The same column model was used in both detectors
(0.25 mm inside diameter (ID) 30 m DB-5MS with 0.25-m
film). The GC temperature program involved an initial temperature of 40 C (2-min hold) followed by ramping (5 C/min)
to 200 C and then an increase (30 C/min) to 280 C. A
constant column flow rate of 1 mL/min and a 40:1 split ratio
were used. The injector temperature was 250 C. The mass
spectrometer, transfer line, and FID detector were held at a
(13) Zabarnick, S.; Widmor, N. Energy Fuels 2001, 15, 1447-1453.

temperature of 280 C. The GC/MS and GC/FID systems were

calibrated with standards containing n-alkanes (C8-C16) and
an internal standard. Calibration curves were generated by
obtaining response factors between the area responses for the
n-alkanes and the area response of the internal standard for
each level of calibration. The area responses for the GC/MS
calibrations were extracted ion areas of the primary characteristic ions for each n-alkane. A minimum of four calibration
levels was used to calibrate each compound. An average
response factor and the relative standard deviation (expressed
as a percentage) were calculated for each of the n-alkanes. Fuel
samples were diluted so that the concentrations of the nalkanes were in the linear range of the calibrations. Each sample was diluted to at least two different concentration ranges.
The higher concentration n-alkanes were quantified by GC/
MS. The GC/MS quantitation involved the extracted ion areas
of the primary characteristic ions, which provided baseline
separation of the n-alkanes from other fuel components. The
lower concentration n-alkanes were already baseline separated
from other fuel compounds and, thus, could be quantified by
GC/FID. The estimated uncertainty in the n-alkane measurements is (10% for weight percentages of >1.0% (for example,
the uncertainty in a stated value of 23% is (2.3%), (15% for
weight percentages in the range of 0.11%-1.0%, and (25%
for weight percentages in the range of 0.010%-0.10%.
ASTM D5133 describes the test method for the temperature
scanning rotational viscometer of Figure 2. This test method
provides a means to determine the continuous rheological
properties of fluids over a desired temperature range. Its
application to petroleum products other than engine oils, such
as jet fuels, has not been established by ASTM. The temperature scanning rotational viscometer utilized in this study

1938

Energy & Fuels, Vol. 19, No. 5, 2005

Atkins et al.
Table 4. Properties of Calibration Fluid LNF-45, as
Provided by the Manufacturera
temp
(C)

viscosity,
(cP)

kinematic viscosity,
(cSt)

density
(g/mL)

-30
-40
-45

31.79 ( 0.12
66.04 ( 0.10
101.8 ( 0.8

36.14 ( 0.18
74.59 ( 0.06
114.5 ( 1.0

0.8791 ( 0.0005
0.8857 ( 0.0009
0.8904 ( 0.0004

Tannas Company provided the LNF-45 compound.

Table 5. Uncertainty for 95% Confidence Level and

Maximum Relative Torque for JP-8 (F3804)
rotor uncertainty
maximum
speed
((%)
relative torque
60
24
12
60
24
12

Figure 2. Rotational viscometer test cell.

(Tannas Company, model PlusTwo) is equipped with two
TAV-1 viscometer heads, each with an adapter that couples
the rotor and glass stator cell (Figure 2) to the viscometer head
with a calibrated spring. The TAV-1 head contains a spring
that has a full scale torque of 6.73 10-5 N m. A single
viscometer head was used to obtain all data to minimize
uncertainty. The unit is equipped with a programmable digital
controller and data acquisition system that controls and
records the sample temperature, as well as records the
percentage of full-scale torque that is required to turn the rotor
(hereafter referenced as relative torque). The desired sample
temperature is achieved by submersion of the rotor and stator
cell into a stirred methanol bath. The maximum bath cooling
rate is -15 C/h, and a sample cooling rate of -5 C/h was
selected to ensure that the sample was cooled at the same rate
as the bath.
The recorded relative torque is that required to maintain a
constant angular velocity of the rotor. The sample fluid exerts
a shear stress on the rotor, providing the opposing torque and
deflecting a torsional spring. The calibrated spring deflection
is measured with a rotary transducer and converted to relative
torque. For a Newtonian fluid, shear stress () is directly
proportional to the shear rate ( ). Viscosity () is defined as
the ratio of shear stress to shear rate:

(1)

For a rotational viscometer, eq 1 and the equipment dimensions can be used to derive a relationship between torque (M)
and the rotor angular velocity (). Equation 2, which is called
the Margules equation,14 shows the relationship between the
viscosity, the dimensions of the viscometer, and the torque in
a rotational viscometer. Equation 2 also shows that the torque
is directly proportional to the rotor angular velocity for a
Newtonian fluid:

M)

4h

1
1
R2R R2S

(2)

where h is the rotor height, RR the rotor radius, and RS the

stator radius (Figure 2).
The instrument is preprogrammed for multiple rotor speeds,
including 12, 24 and 60 rpm, which equate to shear rates of
1.26, 2.51, and 6.28 s-1. Shear rate studies are performed by
altering the rotor rotational speed to determine whether the
(14) Margules, M. Jahrb. Zentralanst. Meteor. Wien. 1903, 40, 1-26.

60
24
12

-20 to -25 C
4.0
9.6
10.0
3.9
18.0
2.2
-30 to -35 C
3.0
13.3
7.0
5.3
14.0
2.8
-40 to -45 C
2.0
20.0
4.0
7.7
12.0
4.1

uncertainty
maximum
((%)
relative torque
-25 to -30 C
3.0
11.2
10.0
4.5
18.0
2.5
-35 to -40 C
2.0
16.2
5.0
6.3
14.0
3.4
-45 to -50 C
2.0
25.0
3.0
9.7
8.0
5.1

jet fuel sample is a Newtonian fluid at low temperature.

Equation 2 is valid for laminar flow only, and, because the
rotational viscometer has been applied to more-viscous lowtemperature automotive lubricants, verification of laminar flow
is required for jet fuel. Moreover, turbulent flow could result
in erroneously high viscosity measurements and mask nonNewtonian viscosity behavior. A Reynolds number of Re <
1500 indicates laminar flow within the rotational viscometer.15
The maximum Re value calculated was <1 for all shear rates
and temperatures studied here, indicating that the flow within
the rotational viscometer is laminar.
The accuracy of the measured absolute viscosity of jet fuel
samples is determined using a Tannas Company certified,
Newtonian, hydrocarbon mixture calibration fluid (LNF-45)
that is similar in composition to a narrow cut jet fuel. Known
values of kinematic viscosity () and density (Table 4) are used
to fit absolute viscosity with temperature (absolute temperature, given in degrees Kelvin), (T), using the MacCoull,
Walther, and Wright equation:16-20

ln[ln( + 0.7)] ) m(ln T) + b

(3)

where slope (m) and intercept (b) of the linear fit are
determined. Relative torque measurements of the calibration
fluid are then correlated to (T), allowing any measured
relative torque of a sample fluid to be correlated with an
absolute viscosity.
An uncertainty analysis was performed to determine the
precision of the rotational viscometer instrument. Relative
torque was continuously measured from -20 C to -45 C for
rotor speeds of 12, 24, and 60 rpm. The experiment was
repeated four times at each rotor speed for the JP-8 F3804
sample. Table 5 summarizes the results of the uncertainty
analysis and shows that the instrument is most precise at
relative torque measurements >6%. Table 6 summarizes the
propagated error in the relative torque ratio calculation, which
is used in shear rate studies to demonstrate the direct
proportionality requirement of eq 2.
(15) Wilcox, D. C. Basic Fluid Mechanics; DCW Industries: La
Canada, CA, 1997.
(16) MacCoull, N. Lubrication 1921, 7, 65.
(17) MacCoull, N. Lubrication 1950, 36, 63.
(18) Walther, C. Erdoel Teer 1928, 4, 510.
(19) Walther, C. Erdoel Teer 1931, 7, 382.
(20) Wright, W. A. J. Mater. 1969, 4, 19-27.

Studies of Jet Fuel Viscosity at Low Temperatures

Energy & Fuels, Vol. 19, No. 5, 2005 1939

Table 6. Torque Ratio Uncertainty for 95% Confidence

Level for JP-8 (F3804) Fuel
Uncertainty (%)
temp (C)

torque ratio )
60/24

torque ratio )
60/12

torque ratio )
24/12

-20 to -25
-25 to -30
-30 to -35
-35 to -40
-40 to -45
-45 to -50

(10.8
(10.4
(7.6
(5.4
(4.5
(3.6

(18.4
(18.2
(14.3
(14.1
(12.2
(8.2

(20.6
(20.6
(15.7
(14.9
(12.6
(8.5

Fuel clouding and freezing behavior visualization was

accomplished utilizing an optical cell,21-24 in conjunction with
a He-Ne gas laser. Viscosity behavior is analyzed and
interpreted utilizing the information and images obtained. The
optical cell enables visualization down to -75 C and determination of the temperatures at which solidification begins,
as well as macroscopic clouding behavior. The laser enables
changes in the relative intensity of the scattered light to be
observed, which are caused by the formation of fuel-clouding
particles.

Figure 3. Relative torque measurements at rotor speeds of

12, 24, and 60 rpm for JP-8 (F3804) cooled at a rate of
-5 C/h.

Results and Discussion

Shear Rate Studies Involving JP-8. The rotational
viscometer was used to obtain relative torque measurements for two JP-8 samples: F3804 and F3773. The
relative torque was measured continuously during cooling at rotor speeds of 12, 24, and 60 rpm. Figure 3 shows
the measured relative torque, relative to temperature,
as the jet fuel is cooled at a rate of -5.0 C/h from
-20 C down to the cloud point temperature. For
reference, the measured freeze point temperature of
F3804 is -48.3 C, and the cloud point temperature is
-52.6 C (Table 3). At each rotor speed, the relative
torque has an almost-exponential increase with decreasing temperature, which is similar to the relationship
between viscosity and temperature of eq 3. Moreover,
Figure 3 shows a sudden and rapid increase in relative
torque at -52.7 C, which occurs simultaneously with
observed crystallization in the optical cell and the
measured cloud point temperature, independent of rotor
speed. The relatively high rotor speeds (as high as
243 rpm) in previous studies were reported to have
broken up the crystalline structure of the frozen jet fuel,
which reduced the measured viscosity.8 The introduction
of two-phase non-Newtonian behavior with high shear
rates demonstrates the practical upper limit on the use
of a rotational viscometer for jet fuel. The maximum
rotor speed of 60 rpm (a shear rate of 6.28 s-1) was not
sufficient to shear or disrupt the forming crystalline
structures here. Therefore, the measured relative torque
obtained with the rotational viscometer at its maximum
rotor speed can be used in shear rate studies. The ability
to use the relative torque measured at the maximum
(as well as the minimum) rotor speed is essential to
perform shear rate studies in the present work.
(21) Atkins, D. L.; Ervin, J. S.; Vangsness, M.; Obringer, C. Flow
Visualization of the Freezing of Jet Fuel, U.S. Department of Energy.
In Proceedings of the 7th International Conference on Stability and
Handling of Liquid Fuels, Austria, 2000; pp 841-856.
(22) Atkins, D. L.; Ervin, J. S. Energy Fuels 2001, 15, 1233-1240.
(23) Atkins, D.; Ervin, J. S.; Obringer, C. Simulations of the Freezing
of Jet Fuel. In Proceedings of the NATO RTO AVT Symposium,
Norway, 2001, Paper No. RTO-MP-069(I).
(24) Atkins, D. L.; Ervin, J. S.; Saxena A. J. Propul. Power 2005,
21, 356-367.

Limitations on the minimum rotor speed (12 rpm)

may also exist and must be considered. The rotational
viscometer is typically applied to fluids that are more
viscous than jet fuel and the instrument used in the
present work is equipped with the most sensitive
torsional spring available from the manufacturer. Table
5 summarizes the uncertainty analysis performed on the
measured relative torque of the JP-8 sample obtained
from the instrument. The table demonstrates that nonNewtonian viscosity behavior may not be detectable,
because of the lack of precision for some conditions.
From Table 5, the uncertainty in the relative torque
measurements is relatively high (>5%) until the measured relative torque is >6% of full scale. Figure 3
shows that the measured relative torque is <6% below
-50 C at 12 rpm and below -35 C at 24 rpm. In
addition, the figure shows that relative torque is always
>6% at 60 rpm. Consequently, the figure shows that
the relative torque measurements at 12 rpm and 24 rpm
(above -35 C) have the greatest noise. Low relative
torque readings (<6%) result in higher uncertainty. For
this JP-8 sample (F3804), uncertainty in the measured
relative torque is lowest at temperatures below -35 C
at a shear rate of 24 rpm, and temperatures below
-20 C at a shear rate of 60 rpm. For low temperatures,
relative torque readings measured for a shear rate of
60 rpm are the most precise; however, the measured
relative torque at rotor speeds of 12 and 24 rpm are
useful and will be needed to evaluate the potential nonNewtonian viscosity behavior of jet fuel.
A method to analyze viscosity behavior requires at
least three rotor speeds. A linear relationship between
torque and rotor speed (recorded in terms of rpm) is
expected for a Newtonian fluid in a rotational viscometer (eq 2). The instrument is capable of three rotor
speeds and data from all are required to plot relative
torque with rotor speed in order to evaluate linearity.
The relative torque values were determined by averaging measured data 0.5 C above and below selected
discrete temperatures for each shear rate, to reduce
noise. Figure 4 shows this averaged relative torque
plotted for three shear rates for temperatures from

1940

Energy & Fuels, Vol. 19, No. 5, 2005

Figure 4. Relative torque versus rotor speed for JP-8 (F3804)

cooled at a rate of -5 C/h.

-20 C to -50 C, at 5 C increments, for the JP-8

(F3804) sample. Figure 4 also shows the relative torque
at each shear rate just before solidification begins
(-52.0 C). The curve fits of relative torque as a function
of shear rate (rotor speed) in Figure 4 have a high level
of linearity at all the temperatures, including -52 C.
(The coefficient of determination or correlation coefficient, R2, is >0.9995 for all fits.) The fits have different
slopes at each temperature, which is expected as the
viscosity varies with temperature. In addition, these fit
lines show a nonzero relative torque at 0 rpm. This
offset varies with temperature and is likely due to
mechanical friction and spring-constant variations with
temperature. The lines in Figure 4 show that the JP-8
sample behaves as a Newtonian fluid at the low temperatures investigated, until the cloud point temperature is reached. The relatively high level of uncertainty
in the measured relative torque and the averaging
technique applied at discrete temperatures require
further consideration to verify Newtonian behavior.
Another method to analyze viscosity behavior requires
data from only two rotor speeds and makes use of all
the data without averaging. Equation 2 indicates that
measured torque is directly proportional to rotor speed
and that this relationship is independent of temperature
for a Newtonian fluid in a rotational viscometer. Only
two rotor speeds are required to evaluate the direct
proportionality of eq 2. Moreover, continuous measured
relative torque, as opposed to average values at discrete
temperatures, may be utilized, enabling the use of all
measurements. From eq 2, relative torque measurements obtained at 60 rpm divided by relative torque
obtained at 12 rpm should yield a torque ratio of 5.0 at
all temperatures for Newtonian behavior. Similarly,
torque ratios of 2.5 (60/24) and 2.0 (24/12) would be
expected for the other rotor speed combinations. Figure
5 shows plots of the torque ratios. Table 6 shows that
the propagated uncertainty in the calculated torque
ratios is lowest for the 60/24 ratio. Consequently, Figure
5 shows that the 60/24 torque ratios have the least
noise, especially below -35 C. This is expected, because
the measured relative torque exceeded 6% for both shear
rates below -35 C (Figure 3). The calculated torque

Atkins et al.

Figure 5. JP-8 (F3804) torque ratios versus temperature for

varying shear rates at a cooling rate of -5 C/h. Open circles
represent corrected values, and solid data points represent
measured values.

ratio for the 60/24 rotor speeds for a Newtonian fluid

should be 2.5. From Figure 5, 60/24 torque ratios varied
from 2.34 to 2.51 below -35 C. A statistical analysis
of the 60/24 torque ratios yields a mean torque ratio of
2.45 with a standard deviation of 0.029, which indicates
Newtonian behavior. Figure 5 also shows torque ratios
at selected temperatures (Figure 4) that have been
corrected for the offset derived from Figure 4. The
nonzero relative torque value at a rotor speed of 0 rpm
(from Figure 4) has been subtracted from all relative
torque measurements. The corrected torque ratios have
a mean value of 2.50 and a standard deviation of 0.018.
Similarly, the corrected 24/12 and 60/12 torque ratios
of Figure 5 have mean torque ratios of 2.0 (standard
deviation of 0.032) and 5.0 (standard deviation of 0.046).
Analysis of the torque ratios provides further evidence
that the JP-8 sample is Newtonian and that the
rotational viscometer can be used for shear rate analysis
of jet fuel.
The compounds within a jet fuel each contribute to
the overall shear behavior and effective viscosity of the
fuel mixture, and the contribution of each species also
is dependent on its concentration within the fuel.25
Because jet fuel samples that originate from different
refinery sources likely may have different flow behaviors, a similar analysis was performed on the measurements obtained for a different JP-8 sample (F3773),
which has a cloud point (-53.7 C) that is lower than
that of F3804 (-52.6 C). Figures 3 and 6 together show
that the measured relative torque curves for F3804 have
a tendency to be higher than those for fuel sample
F3773. The cloud point temperature of F3773 is slightly
lower than that for F3804; therefore, it is reasonable
that the sudden increase in viscosity for F3773 (Figure
6) occurs at a somewhat lower temperature than that
for F3804 (Figure 3). Figure 7 shows the shear rate
analysis of the F3773 JP-8 sample. The linearity of the
relative torque versus shear rate fit (R2 > 0.9995) at
all temperatures, indicates that this JP-8 sample
(25) Boned, C.; Zeberg-Mikkelsen, C. K.; Baylaucq, A.; Dauge, P.
Fluid Phase Equilib. 2003, 212, 143-164.

Studies of Jet Fuel Viscosity at Low Temperatures

Figure 6. Relative torque measurements at rotor speeds of

12, 24, and 60 rpm for JP-8 (F3773) cooled at a rate of
-5 C/h.

Figure 7. Relative torque versus rotor speed for JP-8 (F3773)

cooled at a rate of -5 C/h.

exhibits Newtonian behavior from -20 C to the cloud

point temperature.
Shear Rate Studies Involving Jet A. A shear rate
analysis was performed on the two Jet A samples F3686
and F3688. The analysis was similar to that performed
on the neat JP-8 samples. The Jet A samples chosen
have similar measured freeze point temperatures
(Table 3). As in the case of the JP-8 fuels and as
anticipated, the Jet A samples were determined to
behave as Newtonian fluids from -20 C to their
respective cloud point temperatures. Fuel clouding
during cooling does not cause non-Newtonian viscosity
behavior in the JP-8 and Jet A fuel samples.
Effects of Low-Temperature Additive. The effect
on viscosity of a proprietary additive (A4647) designed
to improve the fluidity of kerosene-based jet fuel at low
temperatures was analyzed. Relative torque measurements were obtained for JP-8 (F3804) that contained
the additive (Figure 8) for each shear rate over the
temperature range of -20 C to -70 C. For reference,
Figure 8 shows relative torque measurements obtained

Energy & Fuels, Vol. 19, No. 5, 2005 1941

for the neat fuel. Several differences in the behavior of

the neat fuel and the additive-containing fuel are
obvious. The figure shows that the fuel with the additive
exerts more torque on the rotor, indicating that it is
more viscous. The 60-rpm shear rate curves show this
greater viscosity beginning near -20 C. Unlike the neat
fuel, the figure shows that the relative torque measurements of the fuel containing this low-temperature
additive do not have a sudden increase in value at
-52.7 C. Relative torque measurements for all three
shear rates show an abrupt increase near the temperature at which adhered solids form on the optical cell
surfaces (-55 C). To understand the differences in the
measured relative torque, the additive-containing fuel
was cooled in an optical cell and significant clouding of
the fuel was observed, beginning near -40 C.
Flow Visualization in an Optical Cell without
the Low-Temperature Additive. Jet fuel clouding
has been observed previously during cooling in fuels that
contain additives designed to improve low-temperature
fluidity, as well as neat fuels.21,26 Figure 9 shows images
of samples being cooled in a 114.3-mm-deep optical cell.
Figure 9a shows a Jet A sample (F3219), obtained from
a vented underground storage tank, cooled to -48.5 C.
Adhered crystals are visible on the surface of the cell
and significant clouding is observed. Figure 9b shows
laboratory-grade (bottled) kerosene cooled to -48.5 C,
with adhered crystals, but no clouding behavior. The
cooling-induced clouding in jet fuels has not been
extensively reported; however, it is believed to be caused
by dissolved water precipitating from solution as the
water solubility decreases with decreasing temperature.27 GC was used to determine the dissolved water
content of the samples.28 The Jet A (F3219) and JP-8
(F3804) samples were found to contain as much as 100%
more dissolved water than the kerosene. The handling
of the kerosene sample limits its exposure to water.
Moreover, kerosene likely contains fewer polar compounds, resulting in lower water solubility. Consequently, the cooling of weathered jet fuel, such as the
JP-8 and Jet A samples, results in a jet fuel/water
emulsion and/or ice crystal dispersion that causes
clouding as the fuel is cooled.
To better understand the clouding phenomena, the
JP-8 sample (F3804) was cooled from room temperature
(18 C) in a 44.5-mm-deep optical cell.22 Unlike the
114.3-mm-deep optical cell (Figure 9), visible clouding
was difficult to observe in the 44.5-mm-deep cell. Cross
polarization has been shown to increase contrast between wax crystals and liquid jet fuel in an optical cell
greatly, because of light depolarization by the crystals.22
Figure 10 shows the results of using cross polarization
to improve the contrast of particles responsible for fuel
clouding during cooling. Figure 10a shows the JP-8
sample in the optical cell with cross polarization and
with the surfaces cooled to -53.0 C. Clouding was not
observed during cooling; however, wax crystals were
observed forming on the cell surface near the cloud
(26) Leo, A.; von Meerwall, E.; Bruno, R.; Schruben, D.; Joshua, J.
J. Aircr. 1989, 26, 465-469.
(27) Aviation Fuels Technical Review (FTR-3), Chevron Products
Company. Available via the Internet at URL: http://www.chevron.com/
prodserv/fuels/bulletin/aviationfuel [cited March 2004].
(28) Anderson, S. D.; Rubey, W. A.; Striebich, R. Prepr. Pap.sAm.
Chem. Soc., Div. Pet. Chem. 1998, 43, 471-475.

1942

Energy & Fuels, Vol. 19, No. 5, 2005

Atkins et al.

Figure 8. Relative torque measurements at rotor speeds of 12, 24, and 60 rpm for JP-8 (F3804) with 5000 mg/L of A4647 cooled
at a rate of -5 C/h.

Figure 9. Observed clouding behavior in optical cell of (a)

Jet A (F3219) and (b) kerosene at estimated wall surface
temperatures. (From ref 21.)

point temperature. Figure 10b-d shows that continued

cooling of the cell surfaces to the pour point temperature
(-58 C) produced surface-adhered wax structures and
visible free-floating wax particles that have been described previously.22,24 The figure shows how wax
crystals are visible using cross polarization, but the
particles responsible for fuel clouding are not, indicating

that the particles responsible for clouding are not

crystalline.
Because the cross-polarization filtering did not show
the expected clouding of a weathered fuel, a He-Ne gas
laser was directed through the optical cell that contained the JP-8 sample (F3804). Figure 11 shows that
the beam intensity was observed to increase as the fuel
is cooled. Beam intensity continued to increase gradually as the sample was cooled to its cloud point temperature. Figure 11a shows no detectable scattering at
20 C, whereas Figures 11b-d shows increased scattering as the temperature is reduced. Light scattering
is a function of wavelength, as well as the number and
size of particles. The wavelength of the He-Ne laser is
0.638 m. Increased scattering indicates that the number and/or size of particles is increasing and is likely
the result of reduced water solubility as the fuel is
cooled. It is believed that if the particles were wax or
free water crystals, they would have depolarized the
light and become visible, using cross-polarization filtering. If the particles are free water, they may not crystallize, because of their size and the presence of a fuel
system icing inhibitor additive. Studies have shown that
significant supercooling is required to freeze emulsified
water droplets, depending on their size.29 Moreover, the
presence of DIEGME, which is intended to lower the
freeze point of free water in jet fuel and is present in
the fuel samples analyzed, inhibits crystallization. The
resulting water/DIEGME mixture freeze point can approach the cloud point temperature of the fuel, inhibiting free water crystallization.30 Therefore, the particles
are likely free droplets of a water/DIEGME mixture.
Shear rate studies described above for JP-8 show that
(29) Langer, G. Air Force Wright Air Development Division Technical Report No. 60-826, 1960.
(30) Grosjean, D. F. Efforts to Advance Fuel-System Ice-Inhibitor
(FSII) Technology; Prepared under contract for the Air Force Research
Laboratory, Propulsion Directorate, Wright-Patterson Air Force Base,
OH, September 18, 1998.

Studies of Jet Fuel Viscosity at Low Temperatures

Energy & Fuels, Vol. 19, No. 5, 2005 1943

Figure 10. JP-8 (F3804) in an optical cell with surfaces cooled to (a) the cloud point temperature (-53.0 C). With further
cooling, the surfaces reach -58.0 C and are held at that temperature for (b) 37.5 min, (c) 42.5 min, and (d) 47.5 min.

Figure 11. Laser scattering in JP-8 in an optical cell at (a)

20 C, (b) -10 C, (c) -30 C, and (d) -40 C.

the observed cloudiness of the fuel/water emulsion does

not influence viscosity behavior and does not cause nonNewtonian viscosity behavior.
Flow Visualization Using the Low-Temperature
Additive. Visualization is useful when analyzing fuels
that have been blended with additives. Visualization
from the optical cell shows that the JP-8 (F3804) fuel
blended with 5000 mg/L of A4647 additive begins to
cloud near -40 C, which is 8 C above the neat fuel
freezing point (Figure 12). The A4647 low-temperature
additive is believed to encourage nucleation sites to form
in the cooled liquid fuel at temperatures well above the
freezing point of the neat fuel. The forming particles
depolarize the light and clouding becomes visible,
indicating that these particles are crystals. Figure 12a
shows that the cloudiness increases in the coldest region

of liquid fuelsat the bottom of the cellsat -45 C.

Figure 12b and c shows that this clouding behavior
expands throughout the cell until solids begin to form
on the sides of the optical cell at -55 C and extend
into the liquid, as shown in Figure 12d. Figure 12d
shows that the clouding is more evenly distributed in
the cell. The relatively large, free-floating crystals that
formed in the JP-8 sample do not form in the fuel with
the additive. It was anticipated that the two-phase
dispersion resulting from the use of the additive would
cause non-Newtonian viscosity behavior.2
Rotational viscometer shear rate studies are performed to evaluate the viscosity behavior of the fuel/
additive mixture. The increased relative torque measurements shown in Figure 8 for the additive-containing
fuel is likely the result of the presence of a two-phase
dispersion, which causes fuel clouding. Moreover, near
-55 C, the measured relative torque briefly increases
rapidly but then returns to the previous rate of increase.
It is believed that the additive alters the crystal
morphology by inhibiting the formation of large flat
plate crystals that entrap liquid and cause an indefinite
increase in relative torque measurements (Figure 3).
The complex relative torque measurements of Figure 8
suggest non-Newtonian behavior. Figure 13 shows the
smoothed relative torque plotted at specific temperatures, as a function of shear rate for the JP-8 with
additive. The curve fits have a high level of linearity
from -20 C to -45 C with coefficients of determination
of >0.9995. However, at -50 C, the linearity drops
to 0.998, and at -55 C, drops to 0.986, indicating a
degree of non-Newtonian behavior relative to the neat
fuel (Figures 4 and 7). To further evaluate the nonNewtonian behavior, torque ratios are plotted with
temperature. Figure 14 shows that the torque ratios are
stronger functions of temperature, when in comparison
to the neat fuel (Figure 5), further indicating nonNewtonian behavior as the mixture is cooled. Corrections were obtained, as done previously, from the
relative torque plot versus rotor speed (Figure 13) for
validation of non-Newtonian behavior. These corrections
were determined to be invalid due to the nonlinear
behavior and are not shown. The nonlinear plots of
relative torque versus rotor speed (Figure 13) and the

1944

Energy & Fuels, Vol. 19, No. 5, 2005

Atkins et al.

Figure 12. Clouding behavior of JP-8 (F3804) with A4647 additive at surface temperatures of (a) -45 C, (b) -50 C, (c) -55 C,
and (d) -60 C.

Figure 13. Relative torque versus rotor speed for JP-8

(F3804) with 5000 mg/L of A4647 cooled at a rate of -5 C/h
at specific temperatures.

variation of torque ratio with temperature (Figure 14)

demonstrate non-Newtonian behavior of the JP-8 and
low-temperature additive mixture. Optical cell imagery
(Figure 12) shows that a sufficient number of evenly
distributed particles have formed that influence viscosity behavior.
Viscosity Determination and Analysis. From eq
3 and Table 4, the correlation between measured
relative torque and viscosity for the Newtonian calibration fluid allows the absolute viscosity of jet fuel to be
determined. The relative torque measurements obtained
at 60 rpm, which are least uncertain, are used to
determine the absolute viscosity of the jet fuel samples.
Figure 15 shows the viscosity of the JP-8, JP-8 with
additive, and Jet A samples. For validation, the viscosity
was determined by another method at -20 C and
-40C and compared to the results in Figure 15 for the
F3804 JP-8 and the F3688 Jet A samples. Kinematic
viscosity () was measured using the capillary method

(ASTM D445), and the density (F) was measured by

pycnometer to obtain absolute viscosity (, Table 3),
where is equal to F. Figure 15 shows that the viscosity
obtained from both methods agree reasonably well.
Figure 15 shows the viscosity of the neat fuel,
compared to that of the fuel containing the low-temperature additive, derived from relative torque measurements at 60 rpm. The additive-containing fuel shows a
growing increase in viscosity, when compared to the
neat fuel, as the temperature is reduced below -20 C.
This difference is likely due to the presence of a twophase dispersion caused by the action of the additive
and observed as clouding in Figure 12. The precipitated
wax/additive particles increase the viscosity, just as
precipitated wax particles increase the viscosity of oil
in pipelines.2 Moreover, crude oils with suspended wax
particles exhibit non-Newtonian viscosity behavior.2 A
plot of viscosity as a function of shear rate, as shown in
Figure 16, helps characterize the non-Newtonian behavior caused by the action of the additive. Figure 16
shows viscosity plotted for selected temperatures at each
of the shear rates. For Newtonian behavior, eq 1 dictates
no change in viscosity with shear rate. The curve fit of
viscosity at -30 C in Figure 16 shows no change in
viscosity with shear rate. Visualization using the optical
cell showed no clouding at -30 C, validating the
Newtonian viscosity behavior. The curve fits at -35 C
to -45 C show viscosity as an increasing function of
shear rate, decreasing with shear rate. Non-Newtonian
pseudo-plastic liquids exhibit decreasing viscosity with
increasing shear rate (shear thinning). Dispersions, such
as wax crystals dispersed in liquid hydrocarbons, have
been shown to exhibit pseudo-plastic viscosity behavior.2
In the present work, it is believed that the turning rotor
of the rotational viscometer has a tendency to align the
additive/wax particles parallel to the flow, reducing drag
and decreasing the shear stress exerted on the rotor.
Increasing the rotor speed (greater shear stress) would
further align the particles and lower the viscosity.
Visualization using the optical cell shows that the
particles become visible near -40 C surface temperatures in the coldest regions of the liquid (bottom of cell).
Figure 12 shows that the clouding behavior increases

Studies of Jet Fuel Viscosity at Low Temperatures

Energy & Fuels, Vol. 19, No. 5, 2005 1945

Figure 14. Torque ratios versus temperature for varying shear rates at a cooling rate of -5 C/h. JP-8 (F3804) with 5000 mg/L
of A4647 additive was used.

Figure 15. Absolute viscosity of JP-8, Jet A, and JP-8 with additive samples obtained using the rotational viscometer and capillary
tube methods. The two symbols represent two capillary tube measurements, and the curves represent the viscometer measurements.

as the fuel is cooled, validating the increasing pseudoplastic non-Newtonian behavior.

With decreasing temperatures, below -45 C, the
viscosity behavior is more complex. Figure 16 shows
that, at -50 C, the viscosity remains constant at shear
rates of 1.26 s-1 and 2.51 s-1 (12 and 24 rpm) but
decreases at a shear rate of 6.28 s-1 (60 rpm), indicating
viscosity behavior transition. The figure shows viscosity

increasing significantly from -50 C to -55 C. At

-55 C, viscosity increases from 1.26 s-1 to 2.51 s-1,
but decreases significantly at 6.28 s-1, more so than at
all higher temperatures. Increasing viscosity with an
increase in shear rate (dilatancy or shear thickening)
is observed in fluids that contain high levels of deflocculated, or finely divided, solid particles colliding with
each other. From the visualization of the clouding of the

1946

Energy & Fuels, Vol. 19, No. 5, 2005

Figure 16. Viscosity of JP-8 (F3804) with 5000 mg/L of A4647

additive, as a function of shear rate.

Figure 17. Magnified image of JP-8 (F3804) in an optical cell.

fuel with additive (Figure 12), it is believed that as the

temperature is reduced below -50 C, the particles have
increased sufficiently in number to cause shear thickening for the shear rates of 12 and 24 rpm. Near -55 C,
Figure 12c shows adhered solids forming on the optical
cell surfaces, revealing solidification behavior and partially explaining the significant increase in viscosity at
-55 C. This is in contrast to the solidification behavior
of the neat JP-8 (Figure 10). Adhered solids and
relatively large free-floating crystals begin to form in
the optical cell near -52 C for the neat JP-8. Figure
15 shows that the neat JP-8 viscosity increases suddenly
and rapidly at this temperature, corresponding to the
solidification behavior observed in the optical cell
(Figure 10).
Visualization of the solidifying JP-8 samples assists
in interrupting viscosity behavior. The magnified image
of freezing JP-8 in Figure 17 shows the adhered crystalline structure, consisting of relatively large flat interlacing platelike crystals, that traps liquid fuel and withstands the shear rate applied at rotor speeds of 12, 24,
and 60 rpm. After wax crystals begin to form, whether
suspended or adhered, the crystalline structure is

Atkins et al.

Figure 18. Magnified image of JP-8 (F3804) with A4647

additive in an optical cell.

sufficiently rigid to stop flow and behave as a solid.

Modeling the solidification behavior of jet fuel, using a
momentum exchange source term in the governing
equations, yielded similar conclusions.24 This solidification behavior explains why viscosity increases rapidly
near -52 C for neat JP-8.
Figure 15 shows that the viscosity of JP-8 fuel with
additive experiences a rapid increase initially at
-55 C, corresponding to solidification behavior observed in the optical cell (Figure 12), but then returns
to lesser rate of increase. Figure 18 shows markedly
different solidification behavior than the neat JP-8 fuel.
Large, flat interlacing crystals do not form and the
structure may not be sufficiently rigid to withstand
higher shear rates, such as 60 rpm. A shearing of the
structures would result in a significant reduction in
exerted rotor torque measurements and the resulting
viscosity. Solidification and resulting shearing likely
explain the complex viscosity behavior at -55 C in
Figures 15 and 16. The phenomenon causing shear
thickening, apparent at 12-24 rpm, is significantly less
than the effects on viscosity of shearing the structures,
hence, the large decrease in viscosity at 60 rpm.
Equation 1 has been modified to characterize nonNewtonian behavior. The effective viscosity of a pseudoplastic fluid, such as the JP-8 fuel with additive above
-50 C, could be represented by a simple two-parameter
power function (Ostwald-de Waele power law):31,32

(uy)

eff ) K

n-1

(4)

The coefficients K and n are fit over a range of shear

rates. Representing the non-Newtonian viscosity behavior below -50 C for the JP-8 fuel blended with the
low-temperature additive is more complex, because of
particle-induced shear thickening and structure-shearing behavior. Viscosity models that incorporate the
presence of particles and shear rates have been proposed. Krieger and Dougherty33 proposed eq 5, which
(31) Ostwald, W. Kolloid. Z. 1925, 36, 99.
(32) De Waele, A. Oil Color Chem. Assoc. J. 1923, 6, 33.
(33) Krieger, I. M.; Dougherty, T. J. Trans. Soc. Rheol. 1959, 3, 137152.

Studies of Jet Fuel Viscosity at Low Temperatures

Energy & Fuels, Vol. 19, No. 5, 2005 1947

shows an increase in liquid viscosity when particles are

added and the increase is dependent on particle concentration, shape, and size.

) liq 1 -

-B
A

(5)

Here, is the particle volume concentration, whereas

A and B are constants that vary with particle shape and
size. Equation 5 does not include a shear rate term and,
thus, could not reproduce the behavior observed in
Figure 16 below -30 C. Pedersen and Ronningsen2
proposed a more complex empirical representation for
the effects of dispersed wax particles in crude oil (eq 6),
which includes the effects of particle volume concentration and shear rate:

) liq exp(Dwax) +

Ewax

xdu/dy

F4wax

xdu/dy

(6)

where D, E, and F are constants that are fit, and wax

is the volume concentration of wax particles. Equations
5 and 6 require knowledge of the particle volume
concentration. Unfortunately, the particles observed in
the optical cell for the JP-8 blended with the lowtemperature additive were too small to determine the
particle volume concentration. Estimates of volume
concentration from phase-change energy obtained using
a differential scanning calorimeter proved inconclusive.
Moreover, eq 6 does not include a shear thickening term.
Based on the complex shear-thinning and shearthickening behavior, as well as the adhered structure
shearing, multiple viscosity models would be required
to represent the viscosity of the fuel/additive mixture
at temperatures below -30 C.
Conclusions
The present work contains the first set of JP-8 jet fuel
viscosity measurements obtained using a rotational
viscometer. The rotational viscometer can be used to
measure the viscosity of kerosene-based jet fuels, such
as Jet A and JP-8, down to the cloud point temperature.
Viscosity measurements from the rotational viscometer
were continuous from -20 C to the cloud point temperature, making the results useful for computational
fluid dynamics modeling. The viscosity of the neat fuel
was observed to increase exponentially with decreasing
temperature to the point of solidification, where viscosity values increased suddenly and rapidly. The rotational viscometer technique was determined to be
applicable to jet fuel; however, relative torque measurements must be >6% of full scale for this instrument to
minimize uncertainty. The viscosity measurements
agree with results obtained from other techniques and
previously published values for jet fuel. Moreover,
visualization techniques were useful in interpreting the
effects of clouding and solidification behavior on viscosity.

The rotational viscometer may be used to perform

shear-rate studies on jet fuel to evaluate Newtonian
behavior, even with relatively high levels of uncertainty.
JP-8, Jet A-1 (by inference), and Jet A fuels exhibit
Newtonian viscosity behavior at low temperatures,
down to temperatures approaching the onset of solidification. The linearity of the fits of relative torque versus
shear rate and the relationship between torque and
shear rate, both as predicted by eq 2, demonstrate that
JP-8 and Jet A are Newtonian fluids near their cloud
point temperature. Clouding of jet fuel during cooling
was determined to be likely caused by free water/
di-ethyl glycol monomethyl ether (water/DIEGME) droplets that do not cause non-Newtonian behavior. The
adhered crystalline structure of the JP-8 fuel that was
forming at its cloud point temperature was sufficiently
rigid to withstand shear rates up to 6.28 s-1.
The rotational viscometer technique can also be used
to evaluate the viscosity of fuels containing additives
that increase fluidity at low temperature. A class of lowtemperature additives thought to encourage nucleation
above the fuel freeze point was observed to exhibit nonNewtonian viscosity behavior. This behavior was complex and seemed to be a function of solidification
behavior and shear rates. Visualization techniques were
useful in determining the relationship between clouding,
solidification, and viscosity behavior. At temperatures
near the freeze point, a dispersion of particles caused
shear thinning non-Newtonian behavior. At temperatures near the cloud point, these particles are believed
to be of sufficient size, number, and distribution to cause
shear thickening. At higher shear rates (6.28 s-1), the
forming adhered crystalline structures of the fuel with
additive were unable to withstand the shear rate.
Nomenclature
DIEGME ) di-ethylene glycol monomethyl ether
h ) rotor height (m)
M ) spring torque (N m)
RR ) rotor radius (m)
RS ) stator (or cup) radius (m)
T ) temperature (C)
u/y ) velocity profile or shear rate (s-1)
) rotor rotational speed (rad/s)
) absolute or dynamic viscosity (Pa s)
) kinematic viscosity (cSt)
) shear stress (N/m2)
) shear rate (s-1)

Acknowledgment. This material is based on research sponsored by the Air Force Research Laboratory,
under Agreement No. F33615-03-2-2347. The U.S.
Government is authorized to reproduce and distribute
reprints for Governmental purposes notwithstanding
any copyright notation thereon. The views and conclusions contained herein are those of the authors and
should not be interpreted as necessarily representing
the official policies or endorsements, either expressed
or implied, of the Air Force Research Laboratory or the
United States Government.
EF049683K