On Flow of Asphalt

F. MOAVENZADEH and R. R. STANDER, JR.

Respectively, Assistant Professor and Research Assistant, Department of Civil
Engineering, Massachusetts Institute of Technolog·y, Cambridge

This paper reviews some concepts suggested to analyze the flow

characteristics of rheological materials with emphasis on those
which are Jromising in the analysis of flow of asphalts. The sug-
gested concepts are categorized as experimental, mathematical,
structural, and physical-chemical. It is shown that for the two
asphalts used in this study, it is possible to construct flow dia-
grams over a wide range of shear stress, rate of shear, and tem-
perature by using three different viscometers and the principle
of reduced variables. The application of Eyring's rate process
theory to the analysis of asphalt flow is examined and the varia-
tion of viscosity, determined at constant shear stress or constant
shear rate, with temperature is discussed.

•KNOWLEDGE OF the flow properties of asphaltic materials is of special interest in

the design of asphaltic mixtures. Such knowledge should lead to a better understanding
of the properties of the asphalt-aggregate mixture which in turn should result in the de -
velopment of a more rational method of mixture design. An understanding of the effect
of loading and climatic variables on the flow properties of the binder could be of great
assistance in the selection of the most suitable asphalt for a specified job.
The complex chemical structure and numerous varieties of asphalts have, thus far,
prevented development of a specific mechanism for describing its flow behavior over
the useful range of loading and climatic variables. There are, however, a great num -
be I" of en1piTical n1odels, relatively few se111i-en1pirical iuodels. and few theoretical
models proposed to describe the flow behavior of asphaltic materials over a limited
range of sot1e specific variables. The extent of work in this area was well reviewed
by Schweyer in 1958 (1), who said, "Acceptance of the statement that 'knowledge in a
field is measured by ffie brevity with which the concepts can be presented' makes the
asphalt technologist pause when the voluminous literature on asphalt is considered. "
This, in other words, indicates that the "art of asphalt" is a very real and necessary
part of asphalt technology, and in order to achieve a scientific base for this art, funda-
mental concepts involving the flow properties of asphalt must be developed. Thus, it is

briefly reviewing the fundamentals involved in analyzing the flow behavior of materials
and discussing the difficulties encountered in their application to asphalt. The response
of certain asphalts under loading as measured by different methods is presented and the
applicability of some rheologic concepts to the analysis of the results is discussed.

CURRENT TRENDS
The objective of rheology is to yield a distinct fundamental, or rational, description
of the deformation and flow of matter. To achieve this end the physical chemist has
approached the problem by considering the molecular characteristic of materials. The
quantum mechanics mathematician approaches the problem by formulating constitutive

Paper sponsored by Committee on Characteristics of Bituminous Materials and presented at the 45th
Annual Meeting.
8
 9

equations and/ or conservation relations for particular materials. Although advances

in both of these fields are quite evident, the practical requirements of industry have out-
run basic research. This situation has created analytical and empirical solutions based
on simplifying assumptions as applied to observed flow conditions. More and more, the
applied approach has become that of using as much as possible of the physical-chemical
and continuum mechanic works. This is quite evident from the recent works of Herrin
and Jones (2) and Majidzadeh and Schweyer(3). The former have used the advances in
the application of rate process theory (4) to the rheological studies of asphalt, and the
latter have used a kinetic approach (~) to analyze the flow behavior of the asphalt.

FLOW REPRESENTATION
In general, there are two basic methods to represent the flow of a fluid, by using a
shear stress -rate of shear diagram or by using a viscosity-shear rate diagram. It
might also be argued that the single point viscosity (coefficient of viscosity) is sufficient
to represent the data; however, for research purposes , the use of this parameter is
limited to Newtonian materials which, unfortunately, do not include a great number of
paving asphalts.
The simplest plot of flow data is that of shear stress vs rate of shear on arithmetic
coordinates. A Newtonian material will plot as a straight line passing through the ori-
gin. If the material is non-Newtonian, the data will, in general, pass through the ori-
gin but will not be linear. The disadvantage of this plot is that the degree of deviation
from Newtonian characteristics cannot be determined. This can be accomplished in
most cases by using a log-log plot. Generally, the data will describe a straight line
which may be represented by a power formula

T = A ('51)n (1)

where n is a measure of the deviation from Newtonian behavior. For a Newtonian mate -
rial, r, is equal to unity.
Consistency cannot be represented by A because of the dimensional difficulties; how-
ever, the viscosity, defined as T/ -9 for a given ,j3 (sometimes referred to as apparent
viscosity), may be used. Since the f used is arbitrary, and not standardized, con-
fusion may result. An alternate method is proposed by Traxler (6) in which viscosities
are compared at a particular power input per volume of sample. -Although the power
input is arbitrary (r x -j)/unit volume = 1,000 given as convenient), the method has the
advantage that very little extrapolation of data is necessary to determine values over a
wide range of materials and test temperatures. The degree of data treatment required
to arrive at the basic shear stress-rate of shear diagram, generally depends on the
geometry of the test apparatus used.
An alternate method oI flow representation is in terms of the viscosity-shear rate
diagram. For Newtonian materials, viscosity is defined as the ratio of shear stress
to shear rate which at constant temperature is independent of the shear stress level.
For non-Newtonian materials, the ratio is defined as apparent viscosity, and is shear-
dependent. The slope of the shear diagram, also used to define viscosity, is called the
differential viscosity, 77D = aT/af. For a power law material the plots of both apparent
viscosity and differential viscosity vs shear rate should be straight and parallel lines.
A third term, referred to as plastic viscosity, is defined as the slope of the straight
line portion of 1 vs T on arithmetic scales. This viscosity is independent of shear rate
and is generally used to determine the yield stress of pseudoplastic materials.
Materials exhibiting rheologic be havior are usually divided into different groups ac-
cording to their flow characteristics. The general division is that of Newtonian and
non-Newtonian flow behavior. Although Reiner (7) objects to this division and believes
that all materials showing viscosity should be called Newtonian fluids, and those with
variable viscosity, generalized Newtonian fluids, in most rheologic work and for the
purpose of this discussion, the Newtonian and non-Newtonian division is used.
Non-Newtonian flow is generally either shear thinning or shear thickening. The first
group would include the ideal Bingham plastic and the pseudoplastics, as they both exhibit
10

decreasing apparent viscosity with increasing shear rate The term thixotropy, as
sometimes applied to this whole group, is used here to describe a special case of shear
thinning in that it is time-dependent. The structure, which is broken down by shear, is
recoverable with time, or, in other words, the process is reversible. Dilatancy is
sometimes used to describe shear thickening material; however, this is a misnomer in
that many materials which show increasing viscosity with increasing shear do not dilate
at all. Thus, dilatancy is a special case of shear thickening. Another unfortunate mis-
use of terms results by describing time-dependent, shear thickening behavior as rheo-
pec tic. There may exist a material which displays shear thickening with time, but r heo-
pexy is the process by which certain thixotropic materials regain their structure faster
with the application of a gentle shear. This is analogous to the flocculation process
which is speeded by a gentle stirring action. In both cases, increasing the shear beyond
a certain limit will start to destroy the structure.

DATA INTERPRETATION
The basic objective of data interpretation is the prediction of flow properties, given
limited information.

Experimental Representation
Until the development of a fundamental concept describing the flow behavior of mate-
rials in general, use of the existing models is generally limited to the range of variables
incorporated in the development of such models. Thus application of these models, while
showing good results when properly employed, is limited to the exact test conditions and
materials as those used in the development of the particular relationship. In other
words, these equations are usually developed to cover a very specific range of data.
This can, perhaps, best be illustrated by allusion to the developments for a pseudo-
plastic material as defined by the flow diagram below.

Shear Stress

In this figure, the total flow diagram is the line ABCDEF; however, for many cases,
description of only part of this line is necessary. For the straight line portion AB, the
material is Newtonian and can be described by

T = 110 'i) (2)

 11

This is again true for EF with 71 0 replaced by 71 where Tim < 71 0 .

00
In the portion CD, the
data are described by the Bingham plastic relationship

(3)

The power function, first proposed by Ostwald and described by Reiner (2_) is a good
approximation of the portion AC, where

T = A (y)n (4)

and n for a pseudoplastic is less than unity. To extend the description to point D, the
relationship

y = A sinh BT (5)

gives a very good approximation. Finally, to describe the entire curve, the equation

(6)

has been suggested (8).

Each equation is valid only within the range of the variables considered in its devel-
opment and subsequent substantiation. Each has served and will continue to serve its
purpose, but their limitations and conditions must be thoroughly known before they can
be applied with meaning.

Mathematical Representation
To examine the flow behavior from the standpoint of continuum mechanics, certain
relationships regarding the state of the material must be derived. These equations,
commonly called the equations of change, are the relationships for conservation of
mass, momentum, and mechanical and thermal energy, each of which is covered for
both isothermal and non-isothermal systems.
The general procedure for developing any one of the equations of change is to assume
a volume element of dimensions t::..x, t::..y, and 6.z. The flow of the quantity of interest is
then formulated for each face of the element, and the dimensions t::..x, 6.y, t::..z are al-
lowed to approach zero. This result, with some simple manipulation, is the final form
of interest.
With this general procedure, each case can be taken in turn. For conservation of
mass, the mass balance equations result in

-(v • pv) (7)

where 'J = (a: cl; cl~).

+ + An important form of this equation is for a fluid of con-
stant density, or

(v - v) = O (8)

Considering conservation of momentum, the problem of momentum balance across

any face becomes more involved, since momentum terms must be accounted for in all
three coordinate directions for each face. The momentum balance equation not only
12

includes momentum in and out, but also the sum of the forces acting on the face such
that

\ pv = - [v . pvv] - [ v • T] - vp + pg (9)
0

where [ v · pvv-] and [ v · T J are vectors because of the tensoral nature of pvv and T.
To use Eq. 9 for flow analysis, the expressions for the components of T must be known.
For a Newtonian fluid, they are of the form

r xx = -2µ -avx +
ax
(2-3 µ - K ) ("v • v) (10)

and
a vx a vy)
T xy = ryx = -µ ( - (11)
aY + - o_-X

where µ and Kare the coefficient of Newtonian viscosity and the coefficient of bulk vis-
cosity, respectively. Some authors (9) suggest that the K term can be dropped as neg-
ligible, while others (7) are of the opinion that K is of the same order of magnitude asµ.
A useful form of Eq. 9 is for flow between parallel plates

= T zz -= T yz = T (12)
xz =
TXX ;
Tyy 0

and

Txy -µ (d;;) (13)

which is Newton's law of vis cosity.

The rate of change of kinetic ene rgy per unit mass is found by considering the scalar
pr oduct of loc,tl velocity wit h th e e rprntion fnr r.ons Prv a tion of momPntum

'\
0
(½ pv2) = -(v-½ pv2v) - (v• pv) - p(-v- v) -

\7• (T• v) - (-r: vv) + p(v• g) (14)

So far, this analysis is based on an isothermal system; however, the term (-r: 17v) de-
scribes the irreversible conver::;iun uf inlernal energy. Slnce some mechanical energy
is thus obviously degraded to thermal, the system is not strictly isothermal unless an
isothermal system is defined as one in which generated heat does not cause appreciable
temperature change. This is true in all but high-speed flow systems with large velocity
gradients. After examining the equation for mechanical energy, the subject can now be
expanded to thermal energy which also allows the examination of non-isothermal sys-
tems. Again assuming the general procedure as first set forth, the energy balance

+
equation, including the effects of kinetic and heat energy and the rate of work done on
the system, may be written as

pv ( U + ½v' )) - (,q) +

p (v• g) - (v · pv) - (17· [T· v ] ) (15)

 13

where U is the internal energy per unit mass of the fluid within the original volume
element.
By using the five equations shown here, alone or in combination, flow problems may
be set up. With the help of numerous assumptions, the resulting equations may be re-
duced to a solution. The difficulty in applying the equations of change to practical flow
problems is that in all but a few select and simple cases, the simplifying assumptions
that are necessary damage the validity of the analysis so much that the results are of
questionable value. It is the current trend of research in this area to apply as much as
possible the mathematical approach to empirically developed models.

Structural Representation
In recent years, the most promising approach for developing a general theory of flow
has been through a structural explanation. Three basic procedures have been suggested
depending on whether colloidal theory, rate theory, or kinetic theory is taken as the
basis of the argument. The colloidal theory is the oldest and simplest of the three.
The approach is based on low concentrations of particles in a solvent. Einstein first
deduced the equation

rJ = rJo (1 + 2. 5 0) (16)

for solutions below about two-tenths percent of spherical particles. Here, Tio is the
viscosity of the pure solvent and r/J is the volume of spherical particles per unit volume
of suspension. The only assumption is that there is no interaction between the individ-
ual spheres. Guth, Gold, and Sim ha (.!.Q_) later amended the relationship as

(17)

which has been experimentally validated for up to 6 percent solutions. Using the basic
idea that relative viscosity-the solution viscosity divided by the solvent viscosity-is
a function of the energy dissipation caused by the presence of the suspended particles,
relationships have been developed for particles of varied shapes at a range of concen-
trations. The effects of particle interaction, absorption of the solvent, particle de-
formability, system thermodynamic conditions, and system electrical conditions are
also investigated in the light of colloidal theory (10).
The second approach is that of Eyring and his co-workers. As stated by Brodkey,
"their work is based on the application of the theory of rate processes to the relaxation
processes that are believed to play an important part in determining the nature of the
flow. . . Very briefly, the theory postulates an activated complex as an intermediate
unstable state, which would be formed from the reactants and decompose into the prod-
ucts. The assumption is made that the decomposition of the complex is the rate con-
trolling step, and that there is an activation energy associated with this (11)." To ex-
pand this argument, the following is presented as digested from Glasstoneet al. (4).
Liquids and gases may be considered as opposites, one consisting of matter moving
around holes (gases), and the other, holes moving around matter. The energy neces-
sary to provide molecular movement is made up of two parts, the energy to create a
hole, and the energy to move a mole cule into it. Taking E as the energy required to
vaporize the molecule and leave behind a hole, then ½ E is required if the hole is closed,
and E - ½ E or ½ E is required merely to make the hole; ½ E is then the energy of
vaporization of the molecule.
In order to extend this development to viscosity, reference is made to the figure on
the following page. For a molecule to move the distance 6, it must be transported
across an energy barrier, E 0 , the energy of activation. The net rate of flow in terms
of shear rate is

(18)
14

where k is the initial rate of motion and Fis the applied shearing force. Using statis-
tical mechanics, k may be shown to be

k = hKT exp (-Eo)

RT (19)

where K, h, and Rare respectively the Boltzmann, Planck, and universal gas con-
stants. Then

A sinh BT (20)

where

2KT
A -h- (21)

and

B (22)

The third concept of structural flow is that based on a homogeneous reaction kinetic
approach as developed by Denny and Brodkey (5, 11, 12). The basic hypothesis is that
non-Newtonian behavior is caused by a structuralbreakdown of the forces between par-
ticles. At very low shear rates, the viscosity is a constant, r, 0 , and at high shear
rates, the viscosity of the material again approaches a constant f/co- These two repre-
sent the limiting cases of no breakdown and complete breakdown. At any point between
these extremes, the viscosity r, is some function of the portion of unbroken forces within
the material. By applying the inverse lever principle, the portion of broken forces may
be taken as

f (r, 0 ) - f (r,)
F (23)
f (11) - I (tJo,)
 15

and the unbroken portion is 1 - F. The function f (rJ) is a relationship for the concen-
tration of changing structure to viscosity and for polymer melts and polymers in solution
the proper form would be (.!..!,)

f (rJ) = A tJl/3. 5 (24)

It is noted that all viscosities are the true apparent viscosities, or the arctangent of a
straight line drawn from the origin to the point on the flow diagram in consideration.
In order to establish the kinetic reaction rate, consider the reaction

k2
unbroken forces ':; broken forces (25)
k1

which has the kinetric reaction equation

d (1 - F)
(26)
ci t

where k ~ must include the effect of shear stress. Therefore

Brodkey (11) has suggested the use of shear stress rather than shear rate so as to sep-
arate the effect of temperature on viscosity from the effect of shear rate on the struc-
tural reaction. The reasoning is that if shear rate is used, then the reaction rate would
be affected by both changes in the F and k terms, whereas, if shear stress is used, the
F term is constant with constant temperature, and the rate is changed only by changes
in the k's.
Following this development, then

d (1 - F)
(28)
dt

At the limiting viscosity

- cl (1 - F) = O (29)
clt

and

(30)
(1 - F)n rP

where K is of the form of an equilibrium constant. The integer constants, m and n,

may be taken from initial rates, or in the case of time-dependent materials, by inte-
gration with assumed values for m and n along a constant shear line.
The attractions of this theory are that the exact nature of the fundamental mechanism
is not assumed, and that the constants are easy to determine and interpret. It is sug-
gested that ultimately all of the parameters should eventually be related to the specific
breakdown mechanism involved for each material or type of material, but it is empha-
sized that this relation need not be known to apply the method.
16

The application of these theories, especially rate process theory and kinetic theory,
to any specific group of materials such as paving asphalts requires some further as-
sumptions and mathematical treatments. The following is a brief review of such ap-
plications to the analysis of flow behavior of asphalts.
Herrin and Jones (2) applied the absolute rate theory, or rate process theory, to
asphalts. Using Eq. 18, they assumed that the values of o and 01 may be considered
of the same order of magnitude, and 26/ 61 may be taken "" 1. Further, the term 66 2 63
was taken as equal to Vf, the effective volume of the flow unit. Using these assumptions
and considering the equation

t.F = t.H - Tt.S (31)

where F is the free energy, H is the heat of activation, T is the temperature in absolute
degrees, and S is the system entropy, they found

T
J = y0 sinh (32 )
To

where T 0 = Vf/2KT and Yo= (KT/h) exp (-t.H/RT) exp (t.S/R). Rewriting the expres-
sion for -,! 0 in logarithmic form

log ~ = log
K
h
t.S
+ 2. 303R
t.H
2. 303R (½) (33)

Then if y0 is constant at any given T, the plot of log 1) 0 /T vs 1/T should be a straight
line. The work of Herrin and Jones showed this relationship to be valid for their
material.
It can also be shown that if T O is taken as larg then the quantity ')I = yO (T/ T 0 ) may
be calculated at any test temperature from the s emilog plot, yo/T vs 1/T for any se-
lected y0 . Using this T O in the expression 1/T O = Vf/2KT, the size of the flow unit at
any temperature may be found. It can also be shown that t.S is not a function of tem-
perature. and thus the heat of activation LlH can be found as the slope of the plot. Since
the intercept of the line is t.S, the free energy of activation AF can then be found and is
a linear function of temperature.
This analysis applied to one asphalt showed that the flow behavior of asphalt can be
predicted using absolute rate theory. It also showed that the flow unit size is much
larger than the individual molecules, and that they decrease in size as the temperature
increases.
Majidzadeh and Schweyer (3), working with thirteen different asphalts, applied ki-
netic reaction theory to their data to determine the equilibrium constants, K, as afunc-
tion of shear rate. For each asphalt, a hyperbolic sine relationship was assumed in
the form

')I = A sinh BT, (34)

and the constants A and B were determined. The limiting viscosities were found and
the equilibrium viscosity for any constant rate of shear was then

1
11 (35)
AB cash BT

which is the inverse of the first derivative of y = A sinh BT for any stress level. Eq.
30 may be written in log form as
(?lo - 77)m
log p log f + log K (17 0 - 11 "')m - n (36)
(77 - 77..,)11
 17
Thus, the slope of the straight line plot of log (r, 0 - r,)m/ (r, - r, )n vs log y will be p and
00

the intercept at y = 1 will make possible the calculation of K. Of course, in order to

obtain a straight line, the proper choic e of m and n must be made . In this case, as with
the case of many polymer melts (11), m = 2 and n = 1.
Thus , if the equilibrium constant of a material and its limiting viscosities are known,
the viscosity for any shear rate within the range of the study may be calculated knowing
one additional point. The only requirements on the material are that it must be repre-
sented by a hyperbolic sine relationship, and that the proper choice of the integers m
and n can be made such that the log-log plot of (r, 0 - r,)m/ (T) - T)ro)n vs y is a straight line .

Physical-Chemical Repr esentation

Early att empts to arrive at a rational explanation for the flow behavior of bitumens
assumed a colloidal system. This theory, postulated by Nellensteyn (13), was based
on observations of solutions of bitumens in benzene. Mack (6) questions the validity
of extrapolation of Nellensteyn findings to a solvent-free material because evidence
indicates that asphaltenes are not completely dissolc.,ed in benzene but exist in the solu-
tion as partially saturated associated particles. This is not compatible with the as-
sumption that the asphaltic bitumens are solutions of asphaltenes in petrolenes which,
at low temperatures (less than 120 C), form molecular complexes. Mack shows quali-
tatively that non-Newtonian flow increases as the aromaticity of the oil constituents de-
creases. Thus, it would seem that non-Newtonian flow is some function of the concen-
tration of the asphaltenes and aromaticity or the ability of the petrolenes to dissolve
the asphaltenes.
This work is substantiated by several studies on asphalt using the electron micro-
scope. Katz and Betu (14) found that photographs of films of bitumens formed from ben-
zene solutions showed associated particles of asphaltenes. Swanson (15) found that to
form a homogeneous bitumen film by th e same proc ess, the ratio of resins to asphal-
tenes had to be at least three to one. Finally, if the film is not formed by evaporation
of benzene from a bitumen-benzene solution, examination indicates no particles of typi-
cal colloidal dimensions.
Although most of the molecular forces in bitumens are dispersion forces of attraction
produced by carbon and hydrogen, electron microscope studies seem to indicate that
strong polar bonds are also operative. The structural buildup is then somewhat like
crystallization in that molecular orientation takes place. On the other hand, unlike
crystallization, the operative forces are unable to attract like neighbors because of the
relatively large distances over which attractive forces would have to act. Thus, bitu-
men structure is rather random with unlike molecules attracting each other only if their
forces and shapes are such that they can adapt to each other. Molecules with aromatic
rings, because of their side groups, lend well to cavity formation in which other mole-
cules, if they are of the right shape, may be trapped. Therefore, again the structure
of bitumen is dependent on asphaltene concentration and the aromatic properties of the
other constituents.
To relate structure to flow properties, it is necessary to consider the shape of the
associated complexes. Considering the internal thermodynamics of the system, change
will take place spontaneously only if the free energy is diminished. This decrease is
associated with a decrease in surface area, and thus it could be expected that the com-
plexes are spherical in shape. On application of a shear stress the particles elongate
into ellipsoids and flow through the oily medium. In the case of Newtonian materials,
the association bonds within the particles are too strong to be broken by the applied
stress. On application of higher stresses, some point is reached where these bonds
begin to break, and continue to break until breakup is in equilibrium with the applied
stress. The material then shows Newtonian behavior in two ranges, one of no break-
down (low shear stress) and one of breakdown in equilibrium with stresses (high shear
stress). Thus, chemical structure can be related to physical behavior.

THERMAL EFFECTS
Asphaltic materials are thermoplastic and, therefore, will show variation in con-
sistency with change in test temperature . Since climatic variations and construction
18

conditions represent wide variations in temperature, the prediction of flowbehaviorwith

temperature is of great importance. The general method used to represent the viscosity-
temperature relationship is some form of graphical plot that produces a straight-line
relationship for the particular data. Neppe (16) has listed six different plots along with
a description of the equations and useful temperature ranges.
To date, the most commonly used graphical representation of viscosity-temperature
data is the Walther plot of log-log viscosity vs log absolute temperature (16). Follow-
ing the general objective of finding a straight line relationship, this method has been by
far the most successful. Another plot, log viscosity vs reciprocal absolute test tem-
perature, was used by the authors in a previous study with good results over a tempera-
ture range of 10 C to 60 C (17).
The viscosity-temperature relationship may be explained structurally by considering
conditions within the material at low temperatures. In this state the asphaltenes are
precipitated from solution and exist as relatively large associated complexes. As the
temperature increases, thermal activity forces the individual molecules farther and
farther apart. Since the attractive forces diminish rapidly with distance, the complexes
subjected to shear stresses begin to break down and thus the viscosity is reduced. Another
contributing factor is that at low temperatures varying amounts of oily constituents are
held within the asphaltene complexes by association bonds. As the thermal energy in-
creases the strength of the bonds decrease, and the oils are freed to give added lubri-
cation to the system.
Perhaps the most significant advance in the study of temperature effects is the de-
velopment of the time-temperature superposition principle. The principle was origi-
nally developed in the field of polymeric sciences and was successfully used by Ferry
(18) to obtain shear diagrams over a wide range of shear rates. The technique is such
that the shear diagrams of a temperature-susceptible material like asphalt, determined
over a wide range of temperatures, are reduced to a common arbitrary temperature.
The result is a shear stress-rate of shear diagram of the material over a wide range
of shear rate at that particular temperature.
The procedure involves determining a shift factor, aT, either analytically or graph-
ically such that when the shear diagram curves for different temperatures are multi-
plied by their respective aT's, the curves superimpose on each other in one continuous
curve at an arbitrary base temperature. The applicability of the principle to asphaltic
materials has been well s111).5t::i.11tic1.ted. Sis!<"o (19) obtained a master curve of viscosity
vs shear rate for a wide range of shear rates. P hillippoff et al. (20) developed a master
curve for the dynamic properties of asphalt. The authors (17) haveshown a reduced
shear diagram over seven decades of shear rate. -

GENERAL FLOW MEASURING DEVICES

The measurement of viscosity is a field more widely investigated than the theory of
viscosity itself, a fact to which present literature will attest. Viscometers themselves
present quite a large variety ranging from small simple rising-bubble types to quite
large supmsut:aLeu rneumeLers. The basit: problem wiLh musL insLrumenLs is that they
are designed to do a specific job, and in most studies, unless the more sophisticated
instruments are available, more than one type of instrument is necessary.
Viscometers are divided into three basic types: rotational, capillary, and miscel-
laneous. Rotational viscometers may be of the coaxial cylinder type or the cone and
plate type. The first is characterized by two concentric cylinders with a small gap
between them in which the sample is placed. A constant rotation, or a constant torque
is applied to one or both of the cylinders, and the viscosity coefficient is calculatedfrom
the torque required to maintain a constant shear, or to keep the other cylinder station-
ary. These instruments are simple of design and easy to use; however, the data rela-
tionships are derived for Newtonian fluids.
For non-Newtonian materials, Brodkey (21) has suggested the following procedure.
The correction factor n' may be found by plotting T vs

(3 7)
 19

on a log-log scale where T/ is the apparent viscosity, r1 and r 2 are the inner and outer
cylinder radii, respectively, n is the angular velocity at the face of the inner cylinder,
and h is the inner cylinder height. Then for any value of T, n' is the slope of the line,
and

- (rz/r1l2 ] [ 2r;flh ] (38)

f = [n' [:
(r/r 1)2/ n'J (r ~ - r!)

If n' is unity, then

2r~nh T
f = = (39)
7)
(r! - r!)

which is the Newtonian case. The measured T and calculated f are then used to con-
struct the shear stress-rate of shear diagram.
The second basic group, capillary viscometers, consists of a fluid reservoir, a cap-
illary tube, a pressure control device, a rate of flow measuring device, and a tempera-
ture control device. Rheometers are part of this general group and are distinguished
by a piston used to drive the fluid through the capillary tube. Their particular advan-
tage is that their high driving pressures allow viscosity measurements at high shear
rates. Orifice viscometers are the simplest and most widely used of the capillary
group. Their simplicity makes them highly adaptable to industrial uses, but for re-
search, the difficulties of analyzing the flow mechanism of such a short capillary ex-
clude their use.
Glass capillary tubes are of two types, pressure flow type and gravity flow type. The
gravity flow types are generally limited to materials of low viscosity, but are conven-
ient in that the driving force is usually the hydrostatic head of the test liquid itself. The
kinematic viscosity may thus be measured directly.
Again, for the capillary viscometer, the relationships are derived for Newtonian
materials such that

Ywall = (- ~~ )wall = ~ (40)

where A is taken as constant and t is fill time, and

T wall = f [K2 (H - h) t] (41)

where K2 is a calibrated instrument constant and (H - h) is a vacuum head term. For

non-Newtonian materials, A is not a constant. Brodkey (21) has suggested the follow-
ing analysis -

fwall = [(3n' + 1)/4n'J (4v/r 0 ) (42)

where n' should satisfy the relationship

(43)

T hus, n' can be obtai ned from the slope of a log- log plot of (-r 0 ~ p/2L) vs (4v/r 0 ) . The
corr ected shear t·at e at the wall may then be ca lcula ted using n' and its corr es ponding
value of (4v/r 0 ). It should be noted that n' is not necessarily a constant and may have
to be found for each value of shear stress at the wall. Shear stress may then be easily
20

calculated point by point from the relationship

T = y T/ (44)

and the shear stress-rate of shear diagram constructed.

Of all the miscellaneous group, the most useful and widely used is the sliding plate
microviscometer. The procedure for this instrument is relatively simple as the geom-
etry is that used to define viscosity. The data, therefore, may be used as taken for
both Newtonian and non-Newtonian materials. The instrument produces most favorable
results when applied to studies on asphaltic cement. It has the advantage of requiring
a small sample, and the viscosity is measured in very thin films of the same order of
magnitude as those maintained around aggregates in bituminous mixtures.

MATERIALS AND PROCEDURE

The two asphalt cements used in this study are a 60/70 penetration asphalt cement
from a Venezuelan crude and an AC-20 grade asphalt cement used in the "Asphalt In-
stitute Bureau of Public Roads Cooperative Study of Viscosity-Graded Asphalts" coded
as B-3056. The results of conventional tests on these asphalts are given in Table 1.
The 60-70 penetration asphalt (No. 1) is nearly Newtonian in behavior and the second
asphalt (No. 2) is non-Newtonian, as indicated from their viscosity results (Table 1).
Three different instruments were used to obtain viscosity data ov r a range of tem-
4
peratur e of 10 C to 160 C a nd a range of shear rates of 10- 4 to 10 reciprocal seconds.
A sliding plate mi croviscom -ter , made by Hallikainen Instruments, and a Va rian
Model G-14 graphic recorder were used to obtain data in a shear rate range of 10--i to
10- 1 reciprocal seconds and a temperature range of 10 C to 45 C. Specimen prepara-
tion and testing were done following the procedure described by ASTM (22) with one ex-
ception-each specimen was loaded only once. This exception was usedbecause in the
upper range of shear rates, and especially with more complex asphalt, a considerable
deformation was necessary to establish a measured constant slope. In many cases this
deformation was 200 microns, which represents a one-percent change in area of the
plates and was thus used as the maximum deformation permitted on any one specimen.
In order to provide data for low shear rates at higher temperatures, a Haake Roto-
visco with coaxial cylinders was used. This instrument and its test procedure is well
,.:i,...,,-,r,,,..;1-.,.,...,,-1 hTT ,:r,... ...... "tl7n1"'7r-.VI ...... + ,...,, lo\ n ..... ;,....-flTT -I-hr. Dr.+r.,,;c,nr,. nr,,nc,ic,+c, r.-f ~ nr,,rd---,,,,-..1 ,,,.,;.,_
UC,OV.1..1.U'C,U UJ YU,.lJ. ,tu..LIC,.1. C,l.. '-L.l. 0 \U/o .LJ.I.J..'C.L.J.J, l..J..Lc; .l.l,Vl..VV.1.UVV VVJ.J.U.I.Ul,..::J V.L L-4. '-'V.Ll,l,.LV.1. \ ..U.J..Ll..

and a measuring head. The control unit houses electrical circuitry for converting
torque to potential difference, the drive motor, and a ten-speed reduction transmission.
The measuring head is connected to the control unit with a flexible shaft which acts as
both an electrical and mechanical connection.

TABLE 1
STANDARD SPECIFICATION TESTS ON ASPHALTS

Asphalt Asphalt
Test
No. 1 No. 2

Specific gravity 77/77 F 1. 010 1. 020

Softening point, ring and ball 123 F
Ductility 77 F 150 + cm 250 + cm
Penetration
100 gm, 5 sec, 77 F 63 30
200 gm, 60 sec, 39. 4 F 23.5
Flash point, Cleveland open cup 455 F 545 F
 21

The sample was introduced into a circular gap between two coaxial concentric cyl-
inders at the base of the measuring head. The cylinders were surrounded by an oil
jacket heated to within ±0. 1 C of test temperature by a Haake Model F oil bath. The
outer cylinder was held stationary while the inner cylinder was caused to rotate at a
constant rate by the drive motor through the flexible shaft. Located in the measuring
head, between the flexible shaft and the rotating cylinder, is an electrical torsion dyna-
mometer consisting of two coaxial shafts mechanically coupled by a creep-resistant
torsion spring. The angular displacement of the spring caused by the torque created on
the cylinder immersed in the test material is transmitted electronically through the
flexible shaft to the control unit. This torque is then a measurement of viscosity.
The Rotovisco may be equipped with several sizes of inner and outer cylinders of
which two, denoted MV and SV, were used depending on the expected viscosity of the
material at the particular test temperature. Any one of three measuring heads , 50,
500, or 5000, was used depending on the expected level of stress (the numbers indicate
the approximate torque measured in gm -cm). The ten basic rotation speeds may be re-
duced by 10, 100 or 1000 times using a 10-to-1 and/or 100-to-1 gear reducer in the
m echanical transmission line. This provides a minimum shear rate of 3 x 10- 3 recip-
rocal seconds. The minimum test temperature used with this instrument and coaxial
cylinders was 45 C. Below this, stress levels exceed the instrument limits.
To obtain the viscosity at high shear rates Cannon-Manning vacuum viscometers
were used in conjuction with the Cannon vacuum regulator, and a model H-1 high tem-
perature oil bath. Accuracy is maintained to ±0. 5 mm Hg for a range 5-50 cm Hg with
the regulator, and to ±0. 01 C for a range of 68 to 400 F in the bath. The geometry of
the viscometers is described in detail elsewhere (8).
At least three tests were run at each test temperature to describe the range of shear
rates available. Test temperatures varying from 45 C to 150 C were used.

RESULTS AND DISCUSSION OF RESULTS

The values of viscosity obtained for the two asphalts used in this study are tabulated
in Table 2. As mentioned before, these values were obtained using a capillary, a slid-
ing plate, and a rotational viscometer. Due to the instrument limitations, each type of
viscometer could be used only for a certain range of temperature, shear stress, and
rate of shear. However, as shown in Figure 1, it was possible to obtain values of vis -
cosity for the asphalts at a test temperature of 45 C with all three viscometers. Fig-
ure 1, which shows the variation in the viscosity of asphalt No. 1 with rate of shear,
indicates that the three viscometers used gave overlapping data, and there exists a
continuity in their results. The slight deviation observed at low rates of shear is be-
lieved to be due to an error in the viscosity measurement of the sliding plate viscom-
eter. At the test temperature of 45 C the magnitude of load required to cause flow of
the asphalt between the two parallel plates was of the order of a few grams, as shown
in Table 2. For such small shear stresses, any small error in the balancing system

& M1crov1scometer
0 Rotov1sco
8Cap1ll,ny
4
-2 -1 0

Figure l. Viscosil·y vs shear rate for asphalt No. l at 45 C with all viscometers.
22

TABLE 2
VISCOMETRY RESULTS OF ASPHALTS

Description
Shear Stress T Shear R ate f Viscosity r,
(dynes/ cm 2 ) (sec- 1 ) Poises

Asphalt No. 1
T=5C 3. 30 X 10 5 6. 70 X 10- 4 4. 93 X 10 8
5. 00 X 10 5 8. 55 X 10- 4 5 . 85 X 108
5 3 8
7. 00 X 10 1. 11 X 10- 6. 31 X 10
6 3 8
l, 00 X 10 1. 58 X 10- 6. 33 X 10
3
1. 65 X 10 8 3. 08 X 10- 5, 36 X 10
8
5
T = 10 C 1. 61 X 10 l, 42 X 10- 3 1. 13 X 10
8

3. 50 X 10 5 2.67 X 10-
3
1. 31 X 10
8
5 3 8
6. 00 X 10 4. 68 X 10- 1. 28 X 10
5 2 8
1. 20 X 10 1. 12 X 10- l, 07 X 10
5
1. 65 X 10 l_ 73 X 10-
2
9. 54 X 10 7
4 3
T = 20 C 1. 60 X 10 1. 31 X 10- 1. 22 X 10 7
4 3
3. 00 X 10 2. 56 X 10- L17 X l0 7
5. 50 X 104 4. 90 X 10- 3 1. 12 X 10 7
5 2 7
l_ 30 X 10 1. 23 X 10- 1. 06 X 10
5
3. 55 X 10 3 . 62 X 10-
2
9 .81 X 10 6
3 6
T = 25 C 2. 50 X 10 8. 75 X 10- 4 2 , 86 X 10
3 3
6. 00 X 10 2. 29 X 10- 2. 62 X 10 6
4
1. 20 X 10 4. 59 X lQ- 3 2. 62 X 10 8
4
l_ 43 X 10- 2 6
3, 50 X 10 2 .45 X 10
5 2 6
8. 10 X 10 3, 43 X 10- 2.36 X 10
2 3 5
T = 35 C 8 . 10 X 10 2. 37 X 10- 3 . 42 X 10
3 3 5
1. 50 X 10 4 . 70 X 10- 3 . 19 X 10
3 2 5
4. 00 X 10 1. 39 X 10- 2 . 88 X 10
3 2 5
9. 00 X 10 3. 40 X 10- 2 . 65 X 10
4 2 5
l_ 63 X 10 6, 55 X 10- 2. 49 X 10
2 3 4
T = 45 C 2. 80 X 10 6 . 48 X 10- 4. 32 X 10
n nn. .. 1 n2 n n -t - , 1 n -2 A nr-, v -, n4
~- VU ~ .lV G , C..l A .lV '-t: . VI A .lV
3 2
2.80 X 10 7, 45 X 10- 3, 76 X 104
9. 50 X 103 2. 67 X 10- 1 3. 56 X 104
3. 00 X 104 9, 10 X 10- 1 3 . 30 X 104
T = 80 C 2.38 x l02 7, 40 X 10-l 3.22 X 102
7. 00 X 102 2. 17 X 10° 3 . 23 X 102
1. 80 X 103 5, 55 X 10° 3 . 24 X 102
5. 50 X 103 l.68 x l0 1 3 . 27 X 102
1. 78 X 104 5. 43 X 10 1 3.28xl02
1
,,, -
.J. -
1 I")(),
..l.~V
r,
V
A
-::z:,
1r\ v
.LU A
1(\2
.LV 2.68x10 1 l, 53 X 10
9.00 X 102 5. 88 X 10 1 1. 53 X 10 1
1. 80 X 103 l, 19 X 10
2
1. 51 X 101
4. 00 X 103 2 . 60 X 102 1. 53 X 101
7. 70 X 103 5. 00 X 102 l, 54 X 10
1

T = 160 C 8.15 X l0 2 5. 22 X 102 1. 56 X 10°

1. 40 X 103 8. 95 X 102 1,56 X 10°
2.80 X 103 l, 79 X 10
3
1. 56 X 10°
4. 00 X 103 2. 56 X 10
3
1, 56 X 10°
5.00 X 103 3 , 20 X 103 1, 56 X 10°
 23

TABLE 2 (Cont'd)
VISCOMETRY RESULTS OF ASPHALTS

Shear Stress T Shear Rate 'ft Viscosity ri

Description
(dynes/ cm 2) (sec- 1 ) Poises

Asphalt No. 2
T = 10 C 4. 00 X 10 5 2. 70 X 10- 4 1. 48 X 10 9
5
6. 50 X 10 5. 15 X 10- 4 1. 26 X 10 9
9. 50 X 105 l, 02 X 10- 3 9. 31 X 10 8
1. 30 X 105 1. 77 X 10- 3 7. 34 X 10 8
1. 64 X 10 6 2. 68 X 10-3 6. 12 X 10 8
T = 20 C 1. 15 X 105 1. 84 X 10-3 6. 25 X 107
2. 20 X 105 4. 40 X 10-3 5.00 X 10 7
3. 80 X 105 9. 3 5 X 10- 3 4. 06 X 107
7. 00 X 105 2. 07 X 10- 2 3.38 X 10 7
1. 16 X 10 6 4. 08 X 10- 2 2.84 X 107
4
T = 25 C 4. 80 X 10 2. 08 X 10- 3 2.3lx l0 7
4
8. 00 X 10 3.81 X 10- 3 2. 10 X 107
1. 50 X 10 5 8. 15 X 10-3 1. 84 X 10 7
2.70 X 105 1. 62 X 10- 2 1. 66 X 107
4. 90 X 105 3. 35 X 10- 2 1. 46 X 10 7
6
T = 35 C 3.28xl03 2. 40 X o-3 l, 37 X 10
6. 00 X 103 4. 77 X 10- 3 1. 26 X 10 6
1.00 X 104 8. 53 X J0- 3 1. 17 X 10 6
1. 80 X 104 l, 66 X l0- 2 1.08 X 10 6
3.20 X 10
4
3.15 X 10- 2 1. 02 X 10 6
2 l_ 83 X 10- 2
4
T = 45 C 7. lQ X 10 3. 88 X 10
2. 00 X 103 4.01 X lQ- 2 4 , 99 X 104
5. 00 X 103 1. 39 X 10- 1 3 .60 X 104
4 4
l, 50 X 10 4. 38 X 10- 1 3. 42 X 10
4
2. 90 X l 0 8. 65 X 10- 1 3 . 35 X 104
T = 80 C 2. 68 X 10 2 8. 80 X 10-l 3 . 05 X 102
7. 00 X 102 2. 3 8 X 10° 2.94 X 102
3
1. 80 X 10 5, 90 X l 0° 3 .05xl02
3. 20 X 10
3
1. 03 X 10° 3. 11 X 102
7. 50 X 10
3
2. 44 X 10
1
3.07 X 102
T = 120 C 2. 28 X 102 1. 87 X 101 1. 22 X 101
2 4. 10 X J0 1 1. 22 X 101
5. 00 X 10
3
l, 20 X 10 9.95'110 1 1. 21 X 101
2.50 x 103 2 1. 23 X 10 1
2. 04 X 10
3 5, 35 X 02 1. 21 X 101
6.50 X 10
2. 16 X 103
2
T = 140 C 6.65xl0 3 . 25 X 10°
3. 40 X 103 1. 03 X 10 3
3 .30 X 10°
4 . 50 X 10 3 1. 36 X 03 3 . 31 X 10°
T = 160 C 1. 18 X 103 9.25xl02 1. 28 X 10°
2. 00 X 103 1. 60 X 103 1. 25 X 10°
3
2. 93 x lO'l 2. 32 X 10 l, 26 X 10°
24

5 6 6
5X10 1 X10 1.5X,06 2X1Q
004
I I I I i
fosl IemQec • t11re...Q 0

. I l I,...
002
2
2 3 4 5 7
u
~
,n
Log- Cnr,·cctinn F:1c101· (i\T) 0
crt 4 4
2x10•
" 01 0 4X10 6X10 8x10•
Figure 2. Correction factor vs shear stress for the '"""
T...e..s.L.Iempec..a.wr_ !~ .~
rl
rotovisco on asphalt No. 1 at 45 C. ~ 075 =

o5
I I

of the viscometer would result in signifi-

cant variation in the calculated values of 0 ~ "-'--l--..L,_-1----L-....,..._.___.J
viscosity. 4

It was mentioned previously that for 800o,-.~o7_5_x_10~•__1_05_x_10_'_•_ 2_0_2_5~X'I0

_ '_ 3_x ~
10

capillary and rotational viscometers the I I I I I

GOO _Ies_Lil.;illpecutwr. = J?0° C
calculation of viscosity is based on New-
tonian flow. In order to determine the
I 1..,...0'
viscosity of non-Newtonian materials, 400 --r·--t--t--J---=-~l~
.,............
- + -I
some corrections are necessary. Follow- 2,vv1----1---1-........1.e:::~--l---+--II----I
ing the correction procedure as suggested
by Brodkey, Figure 2 was constructed. r. . ,._,
This figure shows the iog of shear stress 0 UAl\,J

vs log of correction factor M for asphalt Shea r Strnss , (T) d.\ ncs/c m 2

No. 1 as obtained by the rotational vis-

Figure 3. Shear rate vs shear stress on arithmetic
cometer where
scales for asphalt No. 1 at various
test temperatures.
(45)

This figure shows not only that the results generate a straight line, but the slope of
such a line is unity, which indicates that the Newtonian analysis is sufficient for deter-
mination of viscosity using the rotational viscometer. Similar analysis was used to
determine the correction factors for the capillary viscom eters, and it was found that
for the test temperatures used in this study, the results of capillary viscometers could
also be treated as Newtonian .
Figures 3 and 4 show the flow diagrams at different temperatures for the two asphalts
used in this study. These figures, which are the plots of rate of shear vs shear stress
on arithmetic scales, show that asphalt No. 1 behaves as a Newtonian material at tem-
peratures above 25 C, while asphalt No. 2 exhibits some non-Newtonian behavior up to
45 C. These results when plotted on logarithmic scales generally give straight lines
as shown in Figures 5 and 6. This suggests that, over the range of shear stress and
for the temperatures used in this study, the shear stress-rate of shear relationship can
be approximated by a powe r formula as

T A (-,9)11 (46)
 25

where A and n are constants whose values

Test Turnpe.c.ot uce= ·,a·c- vary with temperature. For both asphalts,
1/ 1 at low temperatures the slope of the line
0 001 1
f -- l - --+----1---'-.,.""-i---+--+--- I is not equal to unity, which indicates that
the asphalts exhibit non-Newtonian behavior
in that temperature range.
0
Figures 7 and 8 show master flow curves
0 5x15 1x106
5
1. 5x10 2x10
6 6 obtained by reducing the curves of Figures
004 - - - ~ - - - - - - - -- - - ~ 5 and 6 with a horizontal shift factor (aT)
d,......- determined as

002 I -- + - - + - - - + - -
(47)

0 where T 0 = 318 K is the arbitrary base

0 1.2 5x10
5 5
2o5x10 3o75x10
5
temperature, and r, and r, 0 are the viscos-
5x10
5

1.2 ities of the asphalts at temperatures T and

T 0, respectively. The master curves
o9 show that when the log -j) vs log T curves
of Figures 5 and 6 were multiplied by the
06 appropriate aT's, they superimposed in a
continuous straight line. These results
o3
substantiate the usefulness of the super-
0 position principle in the prediction of the
4 4
0 1x10 2x10° 3x10 response of materials over wide ranges
400 of shear rate from the results of relatively
300 few tests.
Figure 9 is a plot of log aT vs 1/T -
200 1/T0 where T 0 = 318 K. The curve of each
material is made up of two straight-line
portions that intersect at a temperature
slightly above the softening point of the
2X1Q
3
<1x10" 6x10
3
8x10
3 materials. This inflection is also found
Shear Slrcss, (T) Dyncs/cm 2
in Figure 10, which is a log viscosity vs
reciprocal absolute temperature plot for
Figure 4. Shear rate vs shear stress on arithmetic
1both asphalts. Traxler and Scbweyer (24),
scales for asphalt No. 2 at various
noticing simila · behavior, altribuled this
test temperatures.
change in slope near the softening point to
the colloidal nature of the asphalt. They
suggested that the colloidal structure of
asphalt changes more rapidly with temperature in the region of the softening point than
in other temperature regions, which should result in a difference in temperature sus-
ceptibility of the material in that region.

TEMPERATURE EFFECT
For Newtonian materials, it is shown that viscosity is not dependent on shear stress
and/ or rate of shear, and is a constant at any one temperature. For such a material,
the relationship showing the temperature dependency of viscosity is independent of shear
rate or shear stress, and as shown before (17) this relationship can be expressed satis-
factorily by the Arrhenius equation -

'Y/ A exp (-~E/RT) (48)

where R is the gas constant, and A and ~E can be considered as constants over limited
temperature ranges.
For non-Newtonian materials, however, the viscosity at fixed temperatures is de-
pendent on shear stress or rate of shear. Therefore, in order to develop an expression
 I\:>
0)

4
/
/ /
/ 4
3

2
2

0
'C / '0
0
I-' C
' /
~ ~
0 0 0
d
~ 0 // "'" 0
]" -" "
Ul

is" ~o J
0
cl .-l
.1
/ ,
,/ -2
I
-2 , -3

-4
3
0 1 2 3 4 5 6 7
- M r- 1 1 - Lo~ Shear Stress (T), dyncs/cm 2

-4
2 3 4 5 6 7 Figure 6. Shear rate vs shear stress on log-log scales for
Log Shear ~,tress (T), dyncs/cm 2
asphalt No. 2 at all test temperatures.

Figure 5. Shear rate vs shear stress on log-log scales for

asphalt No. lat all test temperatures.
 4
-
4
I kl I
LE~ENl:l LEGEND
I--
Q 160'C
/
3 3 n Q 160"C
0 12C'C
~ D 120° C
6 a o·c
1-- / 6 80°C
2 /:.i 45'c
-l5QC
[:) 35'C

'w ~
0 2s'c j V ' ~ 8 35°C
T,
" @ 20' c ~ 1 '86 0 2:-..c
- ~ ,ere V "E· ~ 20°C

- & s'c ~ ~ 0 m"c

0
~ 0
~ V
in <r.=
.3 -1
/ 5
-1

~
w
-2
-2
.§ef
-~
-3
p -3

-4
-4
2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9
')
LOI-{ Rl'duecd $he.ii· SI t'l'.'-S (T), d_, nes/cm ~
Log lh•tluc(• d S hc:tt St l'\ •,-;:,,; (,) . rl,\ 11\' :-- / t 111:.!

Figure 8. Master flow diagram for asphalt No. 2 .

Figure 7. Master flow diagram for asphalt No. 1.

1:\:1
-'l
 NI
ex:,

5
9
4

8
3
7
2
m
0
m 6
0
"-

s
i 5
g, 0 ~
"""' >bO
0
..:i 4
-1

3 I I I/ 0 Asphalt #-1
-2 8 Asphalt#2
I I ______I__
2

-3

-4
0 =·- - + - - + - - --!--- --+-- --I- -+---'
230 250 270 290 310 3.30 350 370
-5 ~ t• OK X 10-5
-90 -70 -50 -30 -10 10 30 50
T Tn oi,;: :,.: 10-.J
Figure 10. Viscosity vs reciprocal absolute test temperature for
both asphalts at a constant power input.
Figure 9. Shift factor vs 1/r - 1/r0 for both asphalts at T,, = 318 K.
 29

like Eq. 12 to represent the temperature dependency of the viscosity of a non-Newtonian

material, either A or ~E or both must be considered as functions of shear stress or rate
of shear. This, in most cases, is avoided by using viscosity at a constant shear rate,
a constant shear stress, or a constant power input.
At a fixed shear rate, the viscosity can be expressed as

T (-j!, T)
r, (-j!, T) 0
(49)
'Y

and at a fixed shear stress as

T
r, (T, T) (50)
-,1 (T, T)"

Formal differentiation of these two equations with respect to temperature at fixed

shear stress and fixed shear rate, respectively, results in (23)

_ori/ aT){l =
(01'1 TT
(dll\1'))T
;i {n~
+ l (51)

and

(?l7¥ r'J T )T = l _ 1) (cl1')) (52)

(011h1Tl-9 oT T

For materials such as asphalt which show a decrease in viscosity with an increase
in shear rate and/or increase in shear stress, (2Jr,/2JT)T and (illn71/2Jlny)T are negative.

10

_J~ I
8 ----+-~420k -- --- ----
?c'N'

_____ ,________,__ - - - - - ____ ,_ - - ---<

1---- - -
2
.1?0°C~

0
-1 ------1----=--1-----1--- ---==--!--
-4 .3 -2 -1 0
-_- _--+-- ----
2
'~T·c. .
3 4

Figure 11. Viscosity vs shear rate for asphalt No. 1 at all test temperatures.
30

10
I
- c ,"c
I •Cf!.C
I
2 0° C
25°C
I
I'
I

6
35°cJ
>
..,g'
I 45°C
I
4
I
rn°c
2
20°c
I
150°c
0
3 4 5 6 7 8
Lo~- Shear Stress (T), dyncs/cm 2

Figure 12. Viscosity vs shear stress for asphalt No. I at all test temperatures.

10

~ ,d·c
8
==--- r- 2o~c O
35•c 25 C
- - - - - -,l- 45°C
I
~6 - -- - - - - ; - - - - - ; - - - - - - - - + - - - --I

2
_J --r-----i---- 80°C

_ _ _ __ _1, 0°c
140°C
160"C
0
-1
_4 -3 2 -1 0 2 3 4

Figure 13. Viscosity vs shear rate for asphalt No. 2 at all test temperatures.

Therefore, the (:;)T/(:;)'Y is larger than or equal to one, or (~;)-l}/(:; )r is smaller

than one, which indicates that the te mperatu:re variation of viscos ity al fix d s hear stress
is greater than this variation at fixed rate of shear.
The above-mentioned relation can readily be obse rved from Figures 5, 11, and 12 for
asphalt No. 1, and Figures 6, 13, and 14 for asphalt No. 2. Figures 11 and 13 are plots
of log viscosity vs log shear rate and Figures 12 and 14 are log viscosity vs log shear
 31
10

- _10°c

-
-~'
~
0
u
~
8

6
-- 35°c
~ -0 20°C
I

4 5° :::
2~c

>gi
..., 4
t . I
I
I
2
80°C
I I I

120 °c
_ r~~~c
0 - ~ 100 C I I
2 3 4 5 6 7 8

J,o~· Shear SLrc s.:, (T), d,\ncs/c 111 2

Figure 14. Viscosity vs shear stress for asphalt No. 2 at al I test temperatures.

stress for asphalts No. 1 and No. 2, respectively. Using Figures 5 and 6, it can be
seen that the variation at fixed shear stress is obtained by a vertical cross plot, whereas
the variation at fixed rate of shear is obtained by a horizontal cross plot. It is clear
that the distances between the different temperatures are greater along the vertical
cross plots than along the horizontal cross plots. This relation results because the
slopes of the lines in Figures 5 and 6 are greater than or equal to one, and the plots
for different temperatures tend to lie roughly parallel to each other:

EXAMINATION OF HYPERBOLIC SINE FLOW EQUATION

It is believed that the study of the effect of variation of temperature on the viscous
behavior of materials may lead to some important information about the molecular
structure and mechanism involved in the flow behavior of asphalts . One approach to
such a study is the examination of the suggested structural models for the flow of mate-
rials through the effect of temperature on their parameters. One of the structural models
suggested for the flow behavior of viscous material is the hyperbolic sine equation of
Eyring, which has also been applied to flow of asphalt.
Eyring's rate process theory develops a flow relation as given by Eq. 10. This
equation can also be written as

9 = A sinh BT (53)

where, at a fixed temperature, the values of A and B are constant. It is shown by

Herrin and Jones (2) that when the flow behavior of asphalt can be represented by this
equation, the size of the flow units and their dependency on temperature can be deter-
mined using

(54)
32

where Vf is the size of flow unit and T is

the absolute test temperature. Further-
C A!,pfluit. more, it is shown that considering a hy-
El Asphalt 2 pe rbolic sine relation for flow leads to the
calculation of the heat of activation ~Hand
the free energy of activation.
In order to examine the validity of such
an analysis for the asphalts used in this
study, a curve-fitting procedure was used
to calculate the values of A and B at each
test temperature for the two asphalts. For
asphalt No. 1 and No. 2 a plot of log A/T
vs 1/ T , where T is the absolute tempera-
ture, is shown in Figure 15. Herrin and
Jones have shown that this plot should be
a straight line

log ( $) p - ~H
2.303R (-Tl) (55)

where p is a constant over a normal range

of temperature, R is the universal gas
- 6'----L---..J..--.....1,---..__ __, constant, and ~H is the heat of activation.
200 240 280 320 360 400 This indicates that the slope of the curve
1/ T, :) K x 10-5
representing log A/T vs 1/T is a measure
Figure 15. A/T vs reciprocal absolute test temp- of the heat of activation. Using the cor-
erature for both asphalts. responding values of B at each temperature
for each asphalt, the size of flow units and
its variation with temperature were cal-
culated as shown in Figure 16.

a.-----.-----.---------
? i--- - -- t - - - - - 1 - - -- -- + - - - ------1

,~<
~ 6 r - - - ~ ~- - ~ ~ - -- + - - - ~
s
o Asphalt
0 Asphalt 2

4 _______________________,

250 300 400 450

Figure 16. Flow unit volume vs absolute test temperature for both asphalts .
 33

This figure shows that the size of the flow units of both the asphalts increased with
temperature to a leveling-off point at about 45 C. According to presently accepted ideas,
the flow units should break down and become smaller with increasing temperature. Be-
cause of the apparent contradiction shown in Figure 16, a further examination of the
validity of using the hyperbolic sine to represent the flow behavior of asphalts was made.
Eyring's flow relation, given by Eq. 18, can be written as

(56)

where ~F = AH - T~S and r, = T/,j1.

Differentiating this expression with respect to 1/T once at constant shear stress
and once at constant rate of shear gives

and

~Ff
~ = [ olm7/d(l/T)] 'l1 = -T + (~H/RT + 1) (2kT 2/T662fo) tanh (T6fo<'l3/2kT) (58)

These two equations indicate that if Eyring's flow equation is valid for a material,
then the change in logarithm of viscosity, at constant shear stress and/ or constant
shear rate, with the change in the reciprocal of absolute temperature is not only tem-
perature-dependent but also depends on the shear stress or rate of shear used for the
viscosity calculation. Furthermore, these equations show that there exists a certain
shear stress at which olnr,/o (1/T) will go through zero and change sign. For asphaltic
materials it is shown that the viscosity always decreases with an increase in shear rate,
while the above equations predict a shear stress beyond which the viscosity would start
increasing with an increase in shear rate. It is obvious that at shear stresses around
this value, the hyperbolic sine function cannot represent the flow behavior of asphalt.
In order to obtain an approximate numerical value for this critical shear stress, the
following calculated values of the flow unit size and the heat of activation were used at
a temperature of 300 K as taken from Figures 15 and 16:

Temperature °K H cal/mole °K Vf (A°)3

Asphalt No. 1 300 21,400 3. 70 X 10 5

Asphalt No. 2 300 32,000 3. 55 X 10 5

Substitution of these values in th quatio11s and setting the right side equal to ze r o will
r esult in olnrv' ~ (1/ Th and oln71/c1 (1/T){I going through zero at T = 7. 4 x 10 5 dynes/cm 2
for the first as phalt and 1 = 2 . 4 x 10 6 dynes/ cm 2 for the second asphalt.
Figure 17 shows the variation of olnri/ci(l/Th with shear stress for the two asphalts
used as determined by Eq. 13. This figure indicates that up to a s hear stress of about
104 dynes/ cm 2 for the asphalts, the behavior of the materials can probably be approxi-
mated by a hyperbolic sine relation. This is probably the reason that values of B in
jl = A sinh BT were contradictory . This point is fur ther substantiated by considering
that shear stresses below 10~ dynes/ cm 2 could be used only at high tempe ratures where
the behavior of the materials is Newtonian, and the hyperbolic sine relation degenerates
into T = rif.
The results discussed above indicate that Eyring's hyperbolic sine relation may pro-
vide an excellent means to correlate experimental data to a structural analysis of flow
34

20

15

--- I'\
5 I--
0 Asphalt 1
'I\ '\
\
\
o Asphalt 2

0
l
2 3 4 5 6 7
Shear Stress (T), dynes/ cm 2

Figure 17. Temperature variation of viscosity at fixed shear stresses according to hyperbolic sine flow
equation.

of asphalts in some cases, but not in others. As shown, for both of the asphalts used,
such a hypothesis fails to establish conceptually consistant values of the constants nec-
essary to define the expression.

SUMMARY AND CONCLUSIONS

This paper reviews some concepts suggested to analyze the flow characteristics of
rheological materials with emphasis on those which are promising in the analysis of
flow of asphalts. The suggested concepts are categorized as experimental, mathemat-
ical, structural, and physical-chemical. It is shown that for the two asphalts used in
this study, it is possible to construct flow diagrams over a wide range of shear stress,
rate of she2r, and temperature by using three different viscometers and the principle
of reduced variables. The application of Eyring's rate process theory to the analysis
of asphalt flow is examined and the variation of viscosity, determined at constant shear
stress or constant shear rate, with temperature is discussed. From this study, the
following conclusions are drawn:

1. The flow data, when obtained by different viscometers, are consistent. There-
fore, different viscometers can be used to obtain shear data over a wide range of shear
rate or shear stress. The principle of reduced variables will further extend these
ranges by reducing the data obtained at different temperatures to an arbitrary base
temperature.
2. The temperature dependency of viscosity requires that viscosity variation with
temperature at fixed shear stress be larger than that at fixed rate.
3. An examination of the applicability of Eyring's hyperbolic sine relation to anal-
ysis of the flow of the two asphalts used reveals that (a) for these two materials there
exist critical shear stresses, beyond which the hyperbolic sine relation fails to repre-
sent the flow behavior of each material, and (b) when these critical shear stresses are
within the experimental range, the flow results cannot be represented by such a
relationship.
 35

ACKNOWLEDGMENTS
This research was performed as a part of a project entitled "Durability Character-
istics of Asphaltic Materials" sponsored by The Ohio Department of Highways in coop-
eration with the U. S. Bureau of Public Roads, at the Department of Civil Engineering,
The Ohio State University. The authors are grateful to the above agencies for their
financial support of this study.

REFERENCES
1. Schweyer, H. E. Asphalt Composition and Properties. HRB Bull. 192,
p. 33, 1958.
2. Herrin, M., and Jones, G. E. The Behavior of Bituminous Materials from the
Viewpoint of the Absolute Rate Theory. Proc. AAPT, Vol. 32, 1963.
3. Majidzadeh, K., and Schweyer, H. E. Non-Newtonian Behavior of Asphalt
Cements. Paper presented at the annual meeting of the AAPT, Feb. 1965.
4. Glasstone, S., Laidler, K., and Eyring, H. The Theory of Rate Processes.
McGraw-Hill, New York, 1941.
5. Denny, D. A., and Brodkey, R. S. Kinetic Interpretation of Non-Newtonian Flow.
Jour. Appl. Phys., Vol. 33, No. 7, p. 2269, 1962.
6. Hoiberg, Arnold J. (ed.). Bituminous Materials: Asphalts, Tars, and Pitches,
Vol. 1. Interscience Pub., New York, 1964.
7. Reiner, Markus. Deformation, Strain, and Flow. lnterscience Pub., New York,
1960.
8. Van Wazer, J. R., Lyons, J. W., Kim, K. Y., and Colwell, R. E. Viscosity
and Flow Measurement. Inter science Pub., New York, 1963.
9. Bird, R. B., Stewart, Warren E., and Lightfoot, Edwin N. Transport Phenomena.
John Wiley and Sons, New York, 1960.
10. Andrade, E. N. da C. Viscosity and Plasticity. Chemical Pub. Co., New York,
1951.
11. Brodkey, Robert S. Unpublished notes.
12. Denny, D. A., Kim, H. T., and Brodkey, R. S. Kinetic Interpretation of Non-
Newtonian Flow. Paper presented at the annual meeting of the Society of
Rheology, Oct. 1964.
13. Nellensteyn, F. J. Jour. Inst. of Petroleum Technology, 1924.
14. Katz, D. L., and Betu, K. E. Ind. Eng. Chem., 1945,
15. Swanson, J. M. Jour. Phys. Chem., 1942
16. Neppe, S. L. Classification of Bitumens in Asphalt Technology by Certain Rhe-
ological Properties. Jour. Inst. of Petroleum, Part 1, p. 38, 1952.
17. Moavenzadeh, F. , and Stander, R. R. , Jr. Effect of Aging of Asphalt on Its
Rheological Properties. Paper presented at the annual meeting of ASTM,
June 1965.
18. Ferry, John D. Viscoelastic Properties of Polymers. John Wiley and Sons,
New York, 1961.
19. Sisko, A. W. Determination and Treatment of Asphalt Viscosity Data. Highway
Research Record 67, p. 27, 1965.
20. Gaskins, F. H., Brodnyan, J. G., Phillippoff, W., and Thelen, E. The Rheology
of Asphalt, II. Flow Characteristics of Asphalt. Trans. Society of Rheology,
Vol. 4, p. 265, 1960.
21. Brodkey, R. S. Translating Terms for Non-Newtonian Flow. Ind. Eng. Chem.,
Vol. 54, p. 44, Sept. 1962.
22. ASTM Standards. Proposed Method of Test for Aging Index of Bituminous
Materials. 1964.
23. Bestul, A. B., and Belcher, H. V. Temperature Coefficients of Non-Newtonian
Viscosity at Fixed Rate of Shear. Jour. Appl. Phys., Vol. 24, No. 6,
June 1953.
24. Traxler, R. N., and Schweyer, H. E. Physics, Vol. 7, p. 67, 1936.