Modulus = Stress(t) / Strain

Compliance = Strain(t) / Stress

Spring

Dashpot

Figure 1. The two basic building blocks of viscoelasticity. The spring is elastic and the dashpot is viscous.

The next step is to consider how each of these elements behaves when subjected to an
oscillating strain. First we'll consider the case of the spring which is portrayed in Figure 2.

').70

+
~.~.

90

Figure 2. Simplified diagram of an oscillating spring.

The strain can be written as:

(3)

'Y

'Yosin wt

The term 'Yo is the amplitude or maximum strain, and w represents the angular velocity.
Substituting equation (3) into equation (1) leads us to the expression:
(4)

TE '"

G'Yo sin wt

It is clear that the strain and the resulting shear stress acting on the spring are in-phase with each
other (i.e., they are both sine waves).
The next consideration is that of an oscillating dashpot which can be visualized in Figure 3.

90

Figure 3. Simplified diagram of an oscillating dashpot.

An inspection of equation (2) leads us to first taking the derivative of the assigned strain defined
in equation (3):
(5)

,y

= d)' / dt =

w'Yocos wt

If this expression is substituted into equation (2), the resulting shear stress acting on the dashpot
is:
(6)

TV '"

1jw'Yocos wt

Contrary to the spring response, it is apparent that the applied strain and the resulting shear stress
on the dashpot are 90 out-of-phase (i.e., the strain is a sine wave and the stress is a cosine
wave). We can map these results in the following manner:

time

Ley de Hooke de la Elasticidad

1:El'------~------~time

Ley de Newton de la Viscosidad

time

Figure 4. The assigued strain and stress responses as a function of time for a spring and a dashpot.

At this point the elastic and viscous limits are apparent and a measure of the phase-shift
indicates how viscoelastic the substance is under given test conditions. If the phase-shift is 0,
then the substance can be described as purely elastic. If the phase-shift is 90, then the substance
can be classified as purely viscous. A phase-shift of 45 would indicate for example that the
material is 50% elastic and 50% viscous.

2.2 Kelvin-Voigt Model

In order to further picture exactly what viscoelasticity is, mechanical models have been
introduced consisting of springs and dashpots. The Kelvin-Voigt model corresponds to a spring
and dashpot in parallel and is portrayed in Figure 5.

Figure 5. Diagram of the Kelvin-Voigt mechanical model.

The spring has a constant modulus of G and the dashpot has a constant viscosity of 1/. Any
strain 'Y applied to this system will result in the exact same strain being applied to the dashpot
Yv and spring 'YE' The total stress necessary to deform the system is equal to the spring stress
plus the dashpot stress.
(7)

'Y = 'Yv = 'YE

(8)

TE

TV

If we substitute equations (I) and (2) into this expression, then the constitutive equation
describing the Kelvin-Voigt mechanical model can be written as:
(9)

= G'Y + 1/d'Y/dt

If a sinusoidal strain 'Y

(10)

Kelvin-Voigt Equation of State

'Yosin wt is applied, then the resulting solution is:

= G'Yosin wt + 1/w'Yocos wt

It is clear that the shear stress consists of two parts: a) The in-phase or sine-wave contribution

and b) The out-of-phase or cosine-wave contribution. As reasoned earlier, the in-phase part is
the elastic component and the out-of-phase part is the viscous component.

2.3 Maxwell Model

The Maxwell model corresponds to a mechanical system consisting of a spring and dashpot in
series. This is diagramed in Figure 6.

Figure 6. Diagram of the Maxwell mechanical model.

The spring has a constant modulus of G and the dashpot has a constant viscosity of "I. Because
these elements are in series, the stress on each element will be equal to the total stress. The total
strain exerted on the system is equal to the dashpot strain plus the spring strain.

(11)

'T

(12)

"I -

'Tv

'TE

v + "IE

If the derivative of equation (12) is taken with respect to time, then the strain rate can be
expressed as:

(13)

d'Y/dt

d'Yv/dt + d'YE/dt

Taking the derivative of equation (1) and substituting it and equation (2) into equation (13), the
following constitutive equation is found:
(14) lIG(d'T/dt) + 'T/"I

d'Y/dt

Maxwell Equation of State

If a sinusoidal strain "I = 'Yosin wt is applied, then the following differential equation is the
result:

(15) lIG(d'T/dt) + 'T/"I = w'Yocos wt

The steady-state solution of this first-order differential equation is:

(16)

= {G(Aw)2'Yo/[1 + (AW)2]) sin wt

+ {GAw'Yo/[1 + (AW)2]}COS wt

The term A is the relaxation time and defined as A = r/G. Although the final solution is more
complicated than that of the Kelvin-Voigt model, it is nevertheless apparent that the solution can
be broken into two parts: a) The in-phase or sine-wave contribution and b) The out-of-phase or
cosine-wave contribution. The in-phase part is the elastic component and the out-of-phase part is
the viscous component.

2.4 General Approach to Forced-Oscillation

When applying a sinusoidal strain to a substance, the angular velocity w and strain amplitude
'Yo are assigned parameters. The resulting shear stress is measured as a function of time and
varies sinusoidally with a measurable phase-shift and amplitude. Experimentally, we can thus
write the expressions:
(17)

'Y = 'Yosin wt (Assigned)

(18)

T =

Tosin(wt + o) (Measured)

where
is the phase-shift and TO is the shear stress amplitude. Equation (18) can be expanded
trigonometrically as:
(19)

T = TO[COSO sin wt

sino cos wt]

As in the cases of the Kelvin-Voigt and Maxwell models, there are two contributions
corresponding to the elastic and viscous compQnents.
The magnitude of the complex modulus G is defined as:

and can be readily calculated. The complex modulus is a measure of the substance's total
resistance to strain. Upon consideration of the in-phase sin wt term, the elastic component can be
defined as the storage modulus G':
(21) G' = G*coso = (TO/'YO)coso

The viscous component can be likewise defined from the out-of-phase cos wt term as the loss
modulus G":
(22) G" = G*sino = (To/'Yo)sino

If the test substance is purely elastic (i.e., a spring), then the phase angle o = 0 0 and G* = G',
G" = O. If the test substance is purely viscous (i.e., a dashpot), then the phase-angle is 90 0 and
G" = G*, G' = O.
As an alternative to the complex modulus, the complex viscosity r * is defined as:
(23)

r*= G*/w = TO/'YOW

and is a measure of the magnitude of the total resistance to a dynamic shear (i.e., the maximum
shear rate is 'Yow and the maximum shear stress is TO). It can likewise be broken into two
components, the storage viscosity r" (the elastic component) and the dynamic viscosity r' (the
viscous component):
(24)

r' = G"/w = (To/'Yow)sino

(25)

r" = G' /w = (TO/'YOW)coso

The stress response of equation (19) can be thus written in terms of the moduli or viscosities:
(26)

= G''Yosin wt + G"'Yocos wt

(27)

= 1J"'Yowsin wt

1J' 'Yowcos wt

The magnitudes of these parameters can be readily calculated by applying a known strain
amplitude and frequency to the test substance and measuring the subsequent stress amplitude and
phase-shift. The next step in this anal ysis is to match the predicted model terms with the
measured terms.

2.5 Matching Tenns and Making Sense Out oC the Models

The Kelvin-Voigt model of Figure 5 will be considered first. The stress response of this
mechanica1 model is expressed in equation (10). If the terms of equation (10) are matched with
those of the general equation (26), the magnitudes of the loss modulus and the storage modulus
for the Kelvin-Voigt mode1 can be written as:
T = (Mod)Yosin(wt) + (Visc)wYocos(wt)
(28) G' = G
G" = 1JW
The elastic response is thus entirely due to the spring and the viscous response is entirely due to
the dashpot. If one measures G' and G" experimentally as a function of frequency and G'
remains constant while G" increases proportionally with the frequency, then the substance
behaves in accordance with the Kelvin-Voigt model. Such a material can be described as being a
viscoelastic salid since it always retums to its initial equilibrium position (Le., if one imposes a
stress on the model and then re1eases the stress, the model will eventually retum to its prestressed position.) A viscoelastic solid is said to possess memory. The time-scale necessary for
the response is the retardation time A = 1J/G. If the dashpot viscosity is great, then the
dampening of the response to an imposed stress is great. Most rubbers and gels are viscoelastic
solids.
The Maxwell model of Figure 6 will now be considered. The stress response of this
mechanical model is expressed in equation (16). If the terms of equation (16) are matched with
those of the general equation (26), the magnitudes of the 10ss modulus and storage modulus for
the Maxwell model can be written as:
(29) G' = G(Aw)2/[1 + (AW)2]

G" = GAw/[1 + (AW)2]

In this case the characteristic time-scale is referred to as the relaxation time and is again defined
as A = 1J/G. The elastic and viscous responses are dependent on the magnitude of the
dimensionless term AW. There are two distinct limits when G' and G" are measured as a
function of frequency:
Case (1): (AW)2

<<< 1

Case (2): (AW)2 > > > 1

G' = G(AW)2
G' = G

G" = 1JW
G" = G2/1JW

In case (1) where the frequency is low, the storage modulus scales as the frequency squared
whereas the 10ss modulus is proportional to the frequency. In this regime, the viscous component
is greater than the elastic component since the model has enough time to respond to a given
strain. In case (2) the frequency is high and the dashpot does not have enough time to respond to
the given strain. In this regime, the elastic component dominates and the substance behaves like
a spring. If a material behaves according to these two limits, then it can be said to be a
Maxwellian viscoelastic liquido Such a material has no singular equilibrium position and thus no
memory since it does not retum to its original pre-stressed position. Any applied deformation is
permanent. It is readily apparent that if the dashpot viscosity is great, the model will act as a
spring and if the spring modulus is great, the mode1 will act as a dashpot. Polymer solutions are
viscoelastic liquids.

The results for both of these models are diagramed in Figure 7 in order to graphically
demonstrate how the moduli behave in each case.

G"= TjCJ)
log G

-----7"'-----G'= G

log ACJ)

-------G'

log G

G"

log ACl)
Figure 7. The behavior ofG' and G" as a function offrequency for the Kelvin-Voigt model (above) and for the
Maxwell model (below).

It is not practical to expect that a given substance should behave as either of these models. In
fact most materials behave like various models in conjunction with a spectrum of relaxation
times. The models portrayed in this report only provide a basic physical foundation for
explaining viscoelastic effects. Nevertheless, it has been demonstrated that viscoelastic materials
can behave as either solids or liquids.

5.

EVALUATION

With the above mentioned type of instrument, it is possible to meke e

defined strain and measure a defined stress.

Wa know that these two

signals can be in-phase or out-of-phase depending on the Viscoelastic

behavior of the measured material.

= ((;, sin [w t l

r.- to-'
(
L = 1. 0 sin (w t + cJ )

0'0=

max. amplitude of strain

rr-

l-o = max. amp l i tuda of stress

Now we have to find a way to describe these Viscoelestic properties.

Tha relationship between stress and strain is defined as:

= G*'

using the complex modulus G*.

The complex modulus G* includes the complete informetion of the
Viscoelsstic properties; the elastic component, the viscous component
as well as the phase shift between stress end strein.

G*

= G'

+ iG"

elastic
viscous
where G' is the storaga mmmm(elastic)modulus and Gil is the loss modulus.

The storege modulus G' is e meesurement for the energy stored end
recovered in the materiel, while the loss modulus is e measurement for
the energy lost es heet in the material.
This relationship cen be visualized in the following greph:

Gil

G'

The absolute magnitude of the complex modulus G* can now be calculated

as the peak stress dividad by the peak strain:

IG* I =
Out of this value, the storaga and the loss modulus can be calculated
using trigonometric identities:
G' = IG*/cos eS
G" = IG*/ sin eS
A useful parameter which is a measurement of the ratio of energy lost
to energy sto red is the loss tangent:
ten b = G" / G'

As an alternative to G*, the phasa relationship can also be dascribed

by e complex viscosity:
~ .. = ~' -

wi th

~' =

G" / w and

i~"

~" = G' / w

When both tha amplituda of strain and the frBquancy bacomes very small,
a ViscoBlastic fluid will bahave more lika e newtonian fluid.
case, tha dynamic viscosity

approaches the steady-shBar viscosity

lim~'=~

w-o

In this

3. Experimental
The dynamic properties of viseoelastie substanees ean be readily measured with a rheometer
that ean be made to oseillate at an assigned frequeney and strain amplitude. The major
experimental requirement is the ability to measure the shear stress as a function of time sinee the
strain varies with time. This information can then be used to calculate the various viscoelastic
parameters as outlined in Section 2.4 of this reporto It should be noted that a controlled-stress
rheometer oseillates with an assigned frequency and stress amplitude, and the strain is measured
as a funetion of time.
Various fxture geometries are possible with a rheometer, but the most common are
concentric-cylinder, cone-and-plate, and parallel-plate. In order to characterize a given
substance, various tests are possible. The material can be subjected to a:
(1) Strain sweep in order to locate the sample's linear viscoelastic region.
(2) Frequeney sweep in order to charaeterize the sample's degree of viseoelastieity at various
time scales.
(3) Time sweep in order to traek sueh kinetie phenomena as euring and gel-formation.
(4) Temperature sweep to study how the viseoelastieity alters as a funetion of temperature.
This is also useful for monitoring temperature-initiated reaetions.
The following sections offer brief deseriptions of relating strain and shear stress to the
measuring geometry and of interpreting strain and frequeney sweeps.

3.1 Relating Strain and Stress to the Fixture Geometry

Depending upon the sample' s physical charaeteristies, different geometries are available in
order to optimize dynamie measurements. Since the assigned variable is in actuality the angle of
deformation '" and the measured quantity is the torque M, equations must be utilized in order to
calculate the shear stress T and the strain 'Y. The three most common geometries are pictured in
Figure 8 along with derivations of the strain as a function of the assigned deformation angle for
each confguration.

10

Cone-and-plate

-vr lfJ
~ z: r-t(lco

-=:.-'~

~e)
R

-~-~------

Figure 8. Three common geometries and how to calculate their respective strains as a function of the assigned

deformation angle.

The shear stress as a function of the measured torque M for each geometry can be calculated
from the following expressions.
(30)

= 3MI2'll"R3

(Pa)

Cone-and-plate

(31)

= 2MhrR3

(pa)

Parallel-plates

(32)

= M/2'll"Ri2L (Pa)

Concentric-cylinders

The expression for the parallel-plate fixture is valid when the applied strain lies within the
sample's linear-viscoelastic range (i.e., the moduli are independent of strain). Note that the
stress changes as a function of the radius, and that the maximum stress and strain values exist at
the edge of the plate.
Concentric-cylinders are beneficial when the test sample has a low viscosity (i.e., dilute and
semi-dilute polymer solutions).
The major advantages of a cone-and-plate are that the strain and stress are constant throughout
the gap (when 0 :s; 4) and that it is easy to load and clean. A cone-and-plate fixture is
commonly used for medium and high viscosity materials with extended linear-viscoelastic regions
(i.e., the moduli are independent of strain).
A parallel-plate geometry is recommended when a test sample contains particles or has a
limited linear-viscoelastic region. The strain can be readily adjusted by altering the gap h or the
plate radius R. This is the main advantage parallel-plates have relative to a cone-and-plate (see
the strain equations of Figure 8). The strain can be minimized by increasing the plate gap and
decreasing the plate radius. Of course there is a price to pay and one should be aware that these
actions consequently decrease the torque signal - thus increasing measurement uncertainty.
Optimization of the parallel-plate system is an operator intensive trial-and-error procedure. As
already pointed out, parallel-plates are employed when testing particulate samples. The general
rule when setting the gap is that it should be at least three times greater than the particle diameter.
11

3.2 Strain Sweep

A strain sweep is a dynamic test where the moduli are measured as a function of strain at a
constant frequency. The usual aim of such a test is to measure the point where the "stiffness" or
"strength" of the material is initially effected by the amount of strain. If the moduli and phaseshift remain constant as a function of strain, then the material is classified as linear-viscoelastic
and the test can be considered to be non-destructive (i.e., the internal structure of the sample has
not been disrupted). This is similar to Hooke's Law in which the force required to deform an
object is directly proportional to the strain. When the strain is too great, the proportionality
factor decreases with increasing strain and the object behaves non-linearly.
A simple example to illustrate the above is to consider the stretching of a rubber band. If the
applied strain is small, then the force necessary to stretch it will remain proportional to the
deformation. If the rubber band is stretched too far, it will lose its strength due to the destruction
of crosslinks until it finally breaks.
The strain region where the moduli are dependent on the strain is defined as the non-linear
viscoelastic region and the test is destructive. By performing a strain sweep, one can thus
measure the critical strain at which the substance becomes non-linear viscoelastic. This
information can be used towards predicting how stable or strain-resistant a particular product is.
As the intermolecular structure is increased (i.e., permanent instead of attractive crosslinks), the
linear viscoelastic region will be extended. A practical example is presented in Figure 9.

Delta (degrees)

Complex ModulU8

1000 r---"-'---'-=-=--=---------=--=-==.:..:..::..;

80

80
100

40

+Do'l&

20

10 L..---..J'---'-'-'-LLlJ..L_---'--'--'-l-l-LLLL---..J'----'--'-'-J...UuJ o
0.01
0.1
1
10

Strain

Figure 9. Strain sweep of pudding in order to determine its linear viscoelastic region.

In this case, a strain sweep was performed on pudding with a parallel-plate fixture in order to
determine its linear viscoelastic region and thus the strain range where the pudding retained its
elasticity. Common sense leads us to the assumption that pudding is elastic under low strain.
This can be demonstrated by placing a spoon in pudding and then nudging it a bit. The result the spoon will spring back after the strain is released! In fact, children are delighted to show that
pudding can be made to swing. If the spoon is pushed too far, the strain will be too great and the
spoon will not spring back. We can conclude in this case that the pudding changes from an
elastic to a viscous substance after a critical strain is surpassed. This result makes sense when
12

one recalls that pudding can be made to flow in the mouth. Figure 9 shows that the linear
viscoelastic region extends to almost 0.2, at which point the storage modulus G or pudding
"rigidity" falls off sharply with increasing strain. This is also reflected in the change of the phase
shift 0 from the elastic region (0 < 45) to the viscous region (0 > 45) with increasing strain.
These results are desirable since: I) The pudding should remain relatively stiff at zero or low
strain; and 2) The pudding should flow readily when subjected to high strain.
After all, the pudding should feel like a thick liquid as opposed to bubble gum while being
swallowed.

3.3 Frequency Sweep

A frequency sweep is a dynamic test in which the response of a material is measured as a
function of frequency at a constant strain amplitude. This test is normally performed such that
the strain amplitude lies within the linear viscoelastic region so that the measured properties are
strain independent. A frequency sweep provides a fingerprint of a viscoelastic substance under
non-destructive conditions. As discussed earlier in this report, one obtains direct information
regarding the viscous and elastic behavior of the test material.
Typical results of a frequency sweep performed on a viscoelastic solid are shown in Figure
10. The test substance was a hair gel [2].

.M:.::o..::d..::ul.:..i-'.(.:..Pa=')e.-

loo0~

......::C..::o:.::m:sp..::18..::X_V..::I..::SC:.;O:..:S.:..lt..<.Y-'(.:..P..::as.:..;)

1000

100
100
10

j-e-G'

-b-e"

-+-Eta-I

10 L....--'--'-'-..J...L.LI..l.L....---'--'-'-..J...L.LI..l.'---'---'.....J....w..LUJ 1
0.1
1
10
100

w (1/s)

Figure 10. Frequency sweep of a hair gel.

As the word 'gel' implies, hair gel should be highly elastic so that after styling the hair it stays in
place. In order to judge how elastic the hair gel is, a frequency sweep was performed at low
strain. It is clear from the results in Figure 10 that the elastic component G' (storage modulus) is
far greater than the viscous component G" (loss modulus). Based on this result, one can describe
the hair gel as exhibiting extreme elasticity at low strain. This result agrees with common sense
since the gel is intended to hold the hair in place.
Typical results of a frequency sweep performed on a viscoelastic liquid appear in Figure II.
The test substance was a bath soap containing collagen which can be considered as a polymer
solution (i.e., collagen is the polymer and bath soap is the solvent) [2].

13

ccM..:O:...:d_U_ _II

..:C..:o:...:m2P:...:le:::x-'-V_18_C_O_8_I--',ty

1000F

10

100

10

I-a--

Q'

-A-G"

-+-Ete:'

10

W (118)

Figure 11. Frequency sweep of a bath soap containing collagen.

Collagen is a natural polymer that exhibits elastic effects due to its long-chain nature. It serves
no purpose as a cleanser but acts instead as a temporary therapeutic means for making the skin
feel "more elastic" and thus the user "younger and refreshed". Any measurable elasticity in the
bath soap is thus due to collagen. The frequency sweep portrayed in Figure 11 reveals that the
loss modulus G" is greater at low frequency but that the elastic component G' increases rapidly
with increasing frequency. This behavior is typical of a viscoelastic liquid. In a separate study,
the storage modulus G' of various concentrated collagen solutions was directly correlated to
molecular-weight [3].

4. Summary
The intention of this paper was to introduce the theory and practice of forced-oscillation.
Background information was provided about the terms viscous, elastic and viscoelastic. Simple
mechanical models were then constructed in order to illustrate how viscoelasticity manifests itself.
The Maxwell model was presented as a viscoelastic liquid and the Kelvin-Voigt as a viscoelastic
solid. A general approach to forced-oscillation was outlined and terms were matched with the
mechanical models. This exercise was intended to give the various moduli a more immediate and
practical meaning. This exercise also illustrated the tremendous difference between the Maxwell
and Kelvin-Voigt models in the behavior of the moduli G' and G".
An experimental section described various fixture geometries and their advantages. Equations
necessary for calculating the strain and shear stress as a function of the assigned deformation
angle and the measured torque were also derived for three fixture geometries. In order to round
out the paper with a practical ring, real strain and frequency sweeps were presented along with
interpretations of the results.
5. References
[1] H.A. Barnes, J.F. Hutton and K. Walters, "An Introduction to Rheology". Elsevier
Science, Amsterdam, 1989.
[2] D.A. Holland, Rheology 91,2, June 1991, 108-1l2.
[3] Haake Information Report V91-40E, "Elasticity and Its Effect upon Mold-Filling" 1991.
14