Professional Documents
Culture Documents
Document PDF
Document PDF
HASSAN MOHAMMED
List of Figures iv
List of Abbreviations ix
Abstract x
Acknowledgement xi
1 Introduction 1
2 Literature Review 5
2.1 Indentation of Viscoelastic Materials . . . . . . . . . . . . . . . . 5
2.1.1 Indentation Creep/Relaxation with Constant Poisson’s Ra-
tio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.2 Lee and Radok’s Functional Integrals . . . . . . . . . . . 7
2.1.3 Other Methods . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Indentation of Bitumen and Bitumen-based Materials . . . . . . 10
2.2.1 Indentation Creep/Relaxation with Constant Poisson’s Ra-
tio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.2 Lee and Radok’s Functional Integrals . . . . . . . . . . . 11
2.2.3 Other Methods . . . . . . . . . . . . . . . . . . . . . . . . 12
3 Analytical Solutions 15
3.1 The Linear elastic solution . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Relaxation indentation test . . . . . . . . . . . . . . . . . . . . . 17
3.3 Ramp test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
ii
4 Validation of Analytical Solutions Using Simulations 19
4.1 Models Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1.1 Indentation Relaxation Test . . . . . . . . . . . . . . . . . 21
4.1.2 Indentation Ramped Load Test . . . . . . . . . . . . . . . 22
4.2 Back-Calculation of the Viscoelastic Functions from Model Data 22
4.2.1 Indentation Relaxation Test . . . . . . . . . . . . . . . . . 22
4.2.2 Indentation Ramped Load Test . . . . . . . . . . . . . . . 27
4.3 Effect of Finite Specimen Dimensions . . . . . . . . . . . . . . . . 32
4.4 Effect of Errors in Strain Measurement . . . . . . . . . . . . . . . 32
5 Experimental Tests 34
5.1 Tensile Test on POM Samples . . . . . . . . . . . . . . . . . . . . 34
5.1.1 The Experimental Setup . . . . . . . . . . . . . . . . . . . 35
5.1.2 Mathematical Formulation . . . . . . . . . . . . . . . . . 36
5.1.3 Experimental Results . . . . . . . . . . . . . . . . . . . . 37
5.2 Indentation Tests on POM Samples . . . . . . . . . . . . . . . . . 41
5.2.1 The Experimental Setup . . . . . . . . . . . . . . . . . . . 41
5.2.2 Machine Compliance . . . . . . . . . . . . . . . . . . . . . 43
5.2.3 Relaxation Indentation Tests Results . . . . . . . . . . . . 43
5.2.4 Ramped Indentation Tests Results . . . . . . . . . . . . . 49
6 Discussion 53
6.1 Analytical Solutions . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.2 The Tensile Test . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.3 The Indentation Relaxation Test . . . . . . . . . . . . . . . . . . 54
6.4 The Indentation Ramped Test . . . . . . . . . . . . . . . . . . . . 55
7 Conclusion 56
Appendices 61
B Experimental Results 64
B.1 Uniaxial Tensile Tests . . . . . . . . . . . . . . . . . . . . . . . . 64
B.2 Indentation Relaxation Tests . . . . . . . . . . . . . . . . . . . . 66
B.3 Indentation Ramped Tests . . . . . . . . . . . . . . . . . . . . . . 69
iii
List of Figures
iv
23 The time dependent Poisson’s ratio (ramp test) . . . . . . . . . . 30
24 The error in the viscoelastic functions (ramp test) . . . . . . . . 30
25 The error in the viscoelastic bulk function (ramp test) . . . . . . 31
26 The errors obtained in the calculation of the viscoelastic functions
vs. the quotient of the contact radius to the distance from the
center of contact to the sample edges . . . . . . . . . . . . . . . . 32
27 The errors in the obtained viscoelastic functions due to errors in
strain measurement . . . . . . . . . . . . . . . . . . . . . . . . . . 33
v
51 Difference error between shear relaxation modulus from indenta-
tion ramped test and the uniaxial creep test . . . . . . . . . . . . 52
vi
List of Tables
vii
List of Symbols
A Contact area of indentation
a Contact radius
A0 Constant contact area during relaxation indentation
Au Smallest area in the uniaxial test
D Indenter diameter
ei Prony series parameter for relaxation viscoelastic function
˜θ Laplace transform of circumferential strain–time curve
θ Circumferential strain
x Uniaxial strain
G Relaxation modulus or relaxation viscoelastic function
G0 Instantaneous shear modulus
G̃ Laplace transform of the shear viscoelastic function
gi Prony series parameter for shear relaxation function
h Indentation depth
h0 Constant indenation depth applied during indentation relaxation
hi Prescribed initial indentation depth
hr Prescribed final ramp depth
ji Prony series parameter for creep viscoelastic function
J Compliance or compliance viscoelasic function
J0 Instantaneous creep compliance
J˜ Laplace transform of the compliance viscoelastic function
K Bulk modulus or bulk viscoelastic function
K0 Instantaneous bulk modulus
K̃ Laplace transform of the bulk viscoelastic function
ki Prony series parameter for bulk viscoelastic function
ν Poisson’s ratio
P Indentation load
P0 Constant load applied during indentation creep
P̃ Laplace transform of load–time curve
r Strain gauge distance from the center of indentation
R Indenter radius
s Laplace variable
σx Uniaxial stress
t Time
τ and ξ Dummy variables for the time integral
τi Relaxation time of a Prony series branch
ūz Surface displacement
viii
List of Abbreviations
AFM Atomic Force Microscopy
BBR Bending Beam Rheometer
DSR Dynamic Shear Rheometer
DTT Direct Tension Tester
FEM Finite Element Method
LLDP Linear Low-Density Polyethylene
PC Polycarbonate
PDMS Polydimethylsiloxane
PMMA Poly(methyl methacrylate)
POM Polyoxymethylene
PVAc Polyvinyl Acetate
ix
Abstract
x
Acknowledgement
xi
Chapter 1
Introduction
1
Figure 1: ZHU zwickiLine a commercial indenter for universal hardness testing
2
or mastic sample and mounted on the BBR apparatus. The BBR apparatus is
used to measure the how much deflection or creeping the sample sustains at
a certain temperature. The beam theory is used to calculate the compliance
viscoelastic function, Wang (2010). An alternative to the BBR is the Direct
Tension Tester (DTT) which is basically a uniaxial tensile test for the sample.
The developed test would be able to measure the bitumen and bitumen–
based mastics viscoelastic properties across different length scales and temper-
ature with minimal changes on the experimental setup. The measuring across
different size scale will allow the advancement of the multi-scale modeling of bi-
tumen, mastic and asphalt as illustrated by Jäger, Lackner, and Stangl (2007)
in figure 2. The developed test has the ability to measure local and in situ of bi-
tumen viscoelastic properties, an advantage that the conventional bitumen and
mastics tests lack. An important benefit of this test is that the bitumen and
mastics properties measured can be incorporated into the asphalt deformation
and fracture degradation models that have a direct effect on the cost and service
life of designed roads.
This thesis will start by exploring the state of the art published research,
in chapter 2, on the indentation of viscoelastic materials in general and then
narrows the focus on the bitumen and bitumen based materials indentation re-
search. In chapter 3 the formulas for two testing cases will be derived: controlled
depth test and ramp load test, that should be used in order to extract two in-
dependent viscoelastic functions form the test data. In chapter 4 numerical
simulations will be performed using finite element software, in order to mimic
the laboratory indentation test and re-calculate the independent viscoelastic
function using the solution methods derived in chapter 3. In addition, in chap-
ter 4 the effect of finite sample size and circumferential strain measurement
3
uncertainties will be explored.
In chapter 5 the results will be presented and described of the experimental
work performed on Polyoxymethylene (POM) samples. POM is used as a model
viscoelastic material to explore the possibility of applying the test on bitumen
and bitumen–based mastics samples. The reason for that is to bypass the re-
quirement of cooling and controlling the temperature of bitumen samples, that
would introduce additional uncertainties into the measurements. Performing the
experiments on POM would act as a verification of the developed method and
one step ahead in indentation testing of bitumen and bitumen–based mastics.
In chapter 6 the results obtained from the experiments will be thoroughly
discussed and their implications explored. The conclusions obtained form the
tests and their impact on bitumen and bitumen-based mastics are presented in
chapter 7.
4
Chapter 2
Literature Review
• Indentation creep: p
8h(t) A(t)
J(t) = √ (1)
3 π(1 − ν) × P0
5
• Indentation relaxation:
√
3 π(1 − ν)P (t)
G(t) = √ (2)
8h0 × A0
Larsson and Carlsson (1998) looked into the possibility of using the indentation
test to measure the viscoelastic functions of polymers. Analytical and experi-
mental analysis were performed for the flat-punch and spherical indenters on two
different kinds of polymers; an epoxy resin and PMMA. The authors indented
the test materials rapidly and kept the indentation depth constant to mimic a
relaxation test, while measuring the evolution of force in time and measuring
the residual imprint at the end of the test. They showed using FEM simulations
that the residual imprint can be measured well after the test before it recovers
fully. In addition the authors suggested using a strain gauge to measure the
evolution of the circumferential strain in order to measure two independent vis-
coelastic functions, they used this method on the flat punch indentation. The
authors used uniaxial relaxation tests to validate the indenation tests, the re-
ported errors for the flat punch indentation are shown in table 1. While the
error is 3.5% in the instantaneous shear modulus (G0 ) for ball indentation of
PMMA.
Table 1: Flat punch indentation validation error, Larsson and Carlsson (1998)
• The dynamic model used to determine the indentation area did not take
damping into consideration.
• Tip–sample adhesion.
6
Table 2: Maximum validation errors for different materials, VanLandingham
et al. (2005)
Huang and Lu (2006) used the functional integrals method in order to mea-
sure two independent viscoelastic functions; by using two different types of
indenters, one axis-symmetric while the other is not, to obtain two indepen-
dent equations solved simultaneously in a non-linear mean square minimization
7
for the all experimental points to obtain the independent viscoelastic functions.
The loading type used was ramp loading. The method of functional integrals
was used to obtain the analytical solutions for the viscoelastic indentation prob-
lem. For validation the authors used conventional tests results referenced from
other published literature, table 4 shows the average error difference between the
conventional tests and performed indentation tests for bulk and shear functions.
Table 4: Shear and bulk viscoelastic functions validation errors, Huang and Lu
(2006)
Oyen (2006) on the other hand derived an analytical solution for a com-
bination of ramp-creep test by controlling the load for both spherical and
Berkovich pyramidal indenters. The author used three different polymer ma-
terials: Polyurethane, PL-1 and PMMA. The validation utilized the material
properties of one of the loading conditions using spherical indenter to predict the
load–depth data of another loading condition of the Berokovich indenter, and
the error is the difference between the analytically predicted load–depth data
and the obtained data experimentally. Figure 3 shows the difference between the
predicted data, and the data obtained from the experiment for Polyurethane and
PL-1. The material parameters used in the predicted data are obtained using
spherical indentation, and the experimental data are obtained using Berkovich
indentation. The author attributes the large difference in the figure due to the
effects of plasticity introduced by the sharp tip of Berkovich indenter, which
should be explicitly included in the derivation of the analytical formulation of
the viscoelastic indentation problem. Another potential cause for the deviation
is the large strains induced by the sharp indenter.
8
Figure 3: Predicted and experimental results for Polyurethane (left) and PL-1
(right), Oyen (2006)
9
through nano–indentation and the conventional Dynamic Mechanical Analysis
(DMA). Herbert, Oliver, and Pharr (2008) on the other hand, used a flat punch
nano indenter and reported a 15% error between the nano–indentation and the
conventional DMA test.
10
2.2.2 Lee and Radok’s Functional Integrals
Jäger, Lackner, and Eberhardsteiner (2007), Jäger, Lackner, and Stangl (2007),
and Veytskin, Bobko, and Castorena (2014) have utilized the functional integrals
in their bitumen and bitumen based mastics evaluations.
Jäger, Lackner, and Eberhardsteiner (2007) used the method of functional
integrals to transform the elastic solution into the viscoelastic one while taking
the real deforming geometry of the indenter into account. The authors used
different spring-damper models to represent the elastic and viscous behaviours
of the viscoelastic materials and by relating these parameters (shear modulus
and viscosity) to the force and indentation depth an analytical formula can be
obtained. By fitting the model parameters to the experimental data and per-
forming least square minimization the viscoelastic parameters can be obtained
numerically. The authors used bitumen type (B50/70) in Berkovich indenta-
tion test. Grid nano–indentation technique was used to get the morphology of
the bitumen in the micro–meter range as well as the ability to use statistical
methods to interpret the nano–indentation results. In addition, the authors also
explored the effect of loading rate and maximum load on the obtained param-
eters; concluding that the loading rate has no significant effect on the results
in contrast to the maximum load, where the authors have noted that the ob-
tained model parameters keep decreasing until they reach a limiting value. The
authors have not used an independent viscoelastic test results obtained using a
different methods to validate the obtained material properties. In their next pa-
per Jäger, Lackner, and Stangl (2007) further extended the research by utilizing
the grid indentation technique to identify the properties of the different phases
in the bitumen. The nano-indentation tests were performed at different loading
rates, maximum loads and temperature in order to investigate their effects. The
non-linear dash-pot model is used to describe the viscoelastic functions. Fig-
ure 4 and equations 3 and 4 show the model used by the authors; where F (t)
is a function that depends on the geometric properties, Ȳ is the viscoelastic
function, Ja and k are the model parameters.
11
Z t
dP (τ )
F (t) = Ȳ (t − τ ) (3)
0 dτ
k !
1 t
Ȳ (t) = Ja (4)
4 τ̄
Depending on the temperature range, the authors used DSR and BBR to
validate the results. At approximately 0◦ C the error difference in Ja is about
40% as shown in figure 5.
12
Figure 6: Validation of nano-indentation results, Veytskin, Bobko, and
Castorena (2014)
tion that the stress and strain fields under the indenter are self-similar. Different
types of loading were used. The comparison between the analytical solution and
the experiments showed close agreement at the beginning of the test but they
started to diverge in the later parts of the time domain, the maximum error
between the measured and predicted force is 12.5% to 13.5% in the considered
time frame. No independent validation test was used. Ossa and Collop (2007)
further expanded the application of the formulas to asphalt mixtures to good
agreement between the experiments and the predicted analytical solution. The
downside to this analysis is the requirement to assume the bitumen as a power
law creeping solid with no flexibility to test other models.
Hamzaoui, Guessasma, and Bennabi (2013) did not use analytical solutions
to calculate the viscoelastic functions from indentation readings, but instead
the authors have used FEM simulations with varying model parameters in or-
der to fit the results with the experiments. Prony series parameters with one
summation term were used as a model for the functions with Poisson’s ratio
assumed as a constant. the authors used flat punch indentation technique on
asphalt binder specimens. The authors observed non-linear effects for the higher
loads and higher temperatures, while low loads have also resulted in incorrect
measured quantities; the authors attributed that due to the contact conditions,
friction forces or low resolution measurement error. There were no independent
validation tests performed on the bitumen. A severe limitation of this method
is the computational cost required to fit the parameters especially if more Prony
series summation parameters were used.
Allen et al. (2013) used the atomic force microscopy (AFM) technique to
13
obtain semi-quantitative values of the asphalt binder properties. The authors
identified three phases within the bitumen, for each one the phases the relaxation
modulus was calculated. The authors used the composite theory to calculate
the lower and upper bounds of the bulk binder material properties.
14
Chapter 3
Analytical Solutions
From reviewing the current literature, the bitumen and bitumen-based inden-
tation testing have two proposed loading schemes:
The analytical manipulation required for each type of loading varies sig-
nificantly; for the relaxation indentation test the viscoelastic functions can be
obtained directly from the correspondence principle, while the ramp loading
scheme requires the use of the more convoluted method of functional integrals
developed by Lee and Radok (1960).
The experiments will use a spherical indenter to alleviate the difficulties en-
countered on other indenter types; sharp indenters will always introduce plas-
ticity and flat indenters introduce singular stress field leading to plastic defor-
mations at higher loads.
15
Figure 7: A schematic of the indentation showing relevant parameters
8 Ga3
P = (5)
3 (1 − ν) D
1 (2 ν − 1) P
θ (r) = (6)
4 πr2 G
Where: r
D
a= h (7)
2
and:
3K − 2G
ν= (8)
6K + 2G
By substituting equations 7 and 8 into 5 and 6, and utilizing D = 2R the
following final formulae are obtained:
8 √ 3KG + G2
P = R × h3/2 (9)
3 (3K + 4G)
3 P
θ (r) = − (10)
4 π r2 (3 K + G)
16
3.2 Relaxation indentation test
The relaxation indentation test is obtained by applying the correspondence prin-
ciple to 9 and 10, the Laplace transformed equations are:
8√ sG̃ 3K̃ + G̃ h
3/2
P̃ = R× × o (11)
3 3K̃ + 4G̃ s
3 P̃
˜θ (r) = − (12)
4 π r2 3 K̃ + G̃ × s
h3/2
While L(h3/2 ) = o
s has been utilized since h = ho is constant during the
test and the tilde (˜) symbol above the variables denotes the Laplace transform
of these parameters.
The steps required to obtain the relaxation functions are as follows:
• The recorded P and θ data are fitted into a four terms Prony series
• The Laplace domain relaxation functions, G̃ and K̃, are obtained from
solving 11 and 12 simultaneously.
17
h i R
t
L−1 B s × Ã = 0 B(t − τ ) dA
dτ dτ
• The recorded P , h3/2 and θ data are fitted into a polynomial function.
If P , h3/2 and θ cannot be fitted by a suitable polynomial the following
alternative way should be performed:
n
!
X
−t/τi
K(t) = K0 × 1− ki × 1 − e (18)
i=1
• The two independent equations are solved for all data points in a non-
linear square fit to obtain the Prony series parameters that represent the
material.
18
Chapter 4
Validation of Analytical
Solutions Using Simulations
19
The model used indenter radius R of 12.7 mm, the half space size from the
center of the indentation to the edge is equal to 200 × R to avoid any edge
effects.
20
Figure 9: The zoomed-in view of the mesh and the different zones used
Figure 10: A plot showing the indented geometry with 26 elements in contact
21
in the first stage the half space was modelled as an elastic material and the
displacement is increased linearly from zero to the prescribed value, in the sec-
ond stage the displacement is kept at the prescribed value after changing the
material of the half space into a viscoelastic material with the desired properties.
The controlled displacement was kept for 50 seconds with 5000 data points to
cover the time period. At each data point the summation of the reaction forces
was recorded as the value of the indentation load as well as the displacement
and the circumferential strain at a surface point chosen at a large distance from
the indentation area; in order to be far enough from the local effects of the
indenter as well as far away from the boundaries.
Where:
G0 = 82 M P a K0 = 792.6667
g1 = 0.447 g2 = 0.288 g3 = 0.165 g4 = 0.073
22
k1 = 0.150 k2 = 0.120 k3 = 0.100 k4 = 0.098
τ1 = 0.017 τ2 = 0.165 τ3 = 1.310 τ4 = 11.400
The load and circumferential strain are shown in figure 11 and 12, for both
the obtained data and the fitted data.
The viscoelastic shear relaxation function G(t), bulk relaxation function
K(t) and time-dependent Poisson’s ration ν(t) are shown in 13, 14 and 15
respectively, and the percentage error between the back-calculated and input
parameters is shown in figures 16 and 17.
50
Data Points
45 Fit Function
40
35
30
P (N )
25
20
15
10
0
0 5 10 15 20 25 30 35 40 45 50
time (s)
23
0
-0.5
-1
ǫθ (t) × 106
-1.5
-2
-2.5
-3
Data Points
Fit Function
-3.5
0 5 10 15 20 25 30 35 40 45 50
time (s)
70
60
G (M P a)
50
40
30
20
10
0
0 5 10 15 20 25 30 35 40 45 50
time (s)
24
Bulk Relaxation M odulus
2200
Obtained
2000 Input
1800
1600
K (M P a)
1400
1200
1000
800
600
0 5 10 15 20 25 30 35 40 45 50
time (s)
ν(t)
0.5
0.495
0.49
ν
0.485
0.48
obtained
input
0.475
0 5 10 15 20 25 30 35 40 45 50
time(s)
25
V iscoelastic F unctions Error
0.7
G-error
ν-error
0.6
0.5
P ercentage error (%)
0.4
0.3
0.2
0.1
0
0 5 10 15 20 25 30 35 40 45 50
time (s)
25
P ercentage error (%)
20
15
10
0
0 5 10 15 20 25 30 35 40 45 50
time (s)
Figure 17: The error in the viscoelastic bulk function (relaxation test)
26
4.2.2 Indentation Ramped Load Test
Similarly to the previous section, The load, slope of square–root of cubed depth
and slope of circumferential strain are shown in figures 18, 19 and 20. While
the viscoelastic functions are shown in figures 21, 22 and 23. The percentage
error between the back-calculated and input parameters is shown in figures 24
and 25. It is worth noting that the error in the viscoelastic bulk relaxation
function is larger that the other functions due to the fact the the material is
almost incompressible.
The material used a one term Prony series for the indented half space:
G(t) = G0 × 1 − g1 × 1 − e−t/τ1
K(t) = K0 × 1 − k1 × 1 − e−t/τ1
Where:
10
6
P (N )
0
0 5 10 15 20 25 30 35 40 45 50
time (s)
27
×10-4
-1
Data Points
Fit Function
-2
-3
-4
dh3/2 (t)
dt
-5
-6
-7
-8
0 5 10 15 20 25 30 35 40 45 50
time (s)
×10-8
-3
Data Points
Fit Function
-4
-5
-6
dǫθ (t)
dt
-7
-8
-9
-10
0 5 10 15 20 25 30 35 40 45 50
time (s)
28
Shear Relaxation M odulus
90
Obtained
80 Input
70
60
G (M P a)
50
40
30
20
10
0 5 10 15 20 25 30 35 40 45 50
time (s)
700
K (M P a)
600
500
400
300
0 5 10 15 20 25 30 35 40 45 50
time (s)
29
ν(t)
0.475
obtained
input
0.47
0.465
ν
0.46
0.455
0.45
0 5 10 15 20 25 30 35 40 45 50
time(s)
0.4
0.35
P ercentage error (%)
0.3
0.25
0.2
0.15
0.1
0.05
0
0 5 10 15 20 25 30 35 40 45 50
time (s)
30
V iscoelastic Bulk F unction Error
4.5
3.5
P ercentage error (%)
2.5
1.5
0.5
0
0 5 10 15 20 25 30 35 40 45 50
time (s)
Figure 25: The error in the viscoelastic bulk function (ramp test)
31
4.3 Effect of Finite Specimen Dimensions
The solutions obtained for the viscoelastic contact problem apply to infinitely
large samples. Applying the same equations to a real sample with a finite size
would introduce errors. These errors have been estimated by obtaining the shear
relaxation function and the bulk viscoelastic function for models with a different
size. The results are represented in figure 26.
45
1
40
35
P ercentage error (%)
0.6 25
20
0.4
15
10
0.2
5
0 0
0 50 100 150 200 0 50 100 150 200
Indentation Radius/Distance to Edge Indentation Radius/Distance to Edge
Figure 26: The errors obtained in the calculation of the viscoelastic functions
vs. the quotient of the contact radius to the distance from the center of
contact to the sample edges
32
Average V iscoelastic F unctions Errors
30
G(t)
K(t)
ν(t)
25
P ercentage error (%)
20
15
10
0
0 2 4 6 8 10 12 14 16 18 20
ǫθ error (%)
Figure 27: The errors in the obtained viscoelastic functions due to errors in
strain measurement
33
Chapter 5
Experimental Tests
To validate the method used, a tensile test will be firstly performed on the POM
samples and extract two independent viscoelastic functions. The results from
the tensile tests will be compared to the indentation of the POM samples using
the method developed in this thesis.
34
Table 5: Smallest diameter du for each uniaxial test sample
35
command on the load frame’s software, after that one specimen at a time was
mounted on the load frame using uniaxial test fixtures.
For measuring the viscoelastic relaxation function, the extensometer was
mounted on the specimen as in figure 29, and the load frame was controlled
manually until at least a force of one newton was recorded by the load cell.
The Specimens were tested in uniaxial creep by assigning the constant loads of
400, 300 and 200 Newtons that approximately correspond to the pre-determined
stress levels, at each load level the specimen creep (i.e. the strain) was measured
in the time domain. Due to the load frame limitations the load was applied from
zero to the specified level in approximately 0.86 seconds to mimic a Heaviside
step function. Each specimen was tested at least two times at each load level
for 15 minutes and allowed to recover the strain in 25 minutes. The recovery
period has been determined after recording the amount of time required for the
extensometer to record approximately zero strains.
As for measuring the Poisson’s ratio, the extensometer was changed with a
5 mm strain gauge glued around the specimen circumference as in figure 30.
The Specimens were tested in uniaxial creep by assigning the constant loads
of 400 and 200 Newtons. The procedure then followed the same steps used for
measuring the viscoelastic relaxation function but the circumferential strain was
recorded instead of axial strain.
P 4P
σx = = (19)
Au πd2u
For uniaxial creep the constant stress is related to the strain by the creep
compliance function:
x (t)
J(t) = (20)
σx
The relaxation function can be obtained from the creep compliance function
by using the correspondence principle as in equation 21.
1
E(t) = L−1 (21)
˜
s2 × J(s)
The Poisson’s ratio is calculated from the axial strain and the circumferential
strain as follows:
36
˜θ
s × ν̃ = − (22)
˜x
−1 ˜θ
ν(t) = L − (23)
s × ˜x
Axial Stress
16
14
12
10
σx (M P a)
0
0 100 200 300 400 500 600 700 800 900 1000
time (s)
Using equation 21 on the data collected from the creep test yielded the
following relaxation function:
h
E(t) = E0 × 1 − e1 × 1 − e−t/τ1 − e2 × 1 − e−t/τ2
i (24)
− e3 × 1 − e−t/τ3 − e4 × 1 − e−t/τ4
Where:
37
Axial Strain
0.6
0.5
0.4
ǫx (%)
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700 800 900 1000
time (s)
2.9
2.8
ǫθ (%)
2.7
2.6
2.5
2.4
0 100 200 300 400 500 600 700 800 900 1000
time (s)
38
Creep Compliance F unction
0.037
0.036
0.035
0.034
J ( MP a)
1
0.033
0.032
0.031
0.03
0.029
0 100 200 300 400 500 600 700 800 900
time (s)
E0 = 3366 M P a
e1 = 0.077 e2 = 0.053 e3 = 0.042 e4 = 0.038
τ1 = 609.4 τ2 = 80.3 τ3 = 15.6 τ4 = 2.5
Using equation 23 on the data collected from the creep test yielded the
following Poisson’s ratio:
h
ν(t) = ν0 × 1 − ν1 × 1 − e−t/τ1 − ν2 × 1 − e−t/τ2
i (25)
− ν3 × 1 − e−t/τ3 − ν4 × 1 − e−t/τ4
Where:
ν0 = 0.3973 M P a
ν1 = 0.0274 ν2 = 0 ν3 = 0.0047 ν4 = 0.0217
τ1 = 609.4 τ2 = 80.3 τ3 = 15.6 τ4 = 2.5
Figures 35 and 36 show the average relaxation function and Poisson’s ratio
for POM graphically with a 95% confidence interval.
Results from individual tests are presented in appendix B.
39
E(t)
3500
3400
3300
3200
E(M P a)
3100
3000
2900
2800
2700
2600
-200 0 200 400 600 800 1000
time(s)
P oisson′ s ratio
0.42
0.41
0.4
ν (−)
0.39
0.38
0.37
0.36
0.35
0 100 200 300 400 500 600 700 800 900
time (s)
40
5.2 Indentation Tests on POM Samples
A cubic POM sample was prepared from the same batch as the uniaxial test
sample, with dimensions of 7 x 7 x 7 cm. The surface was initially ground using
fine sandpaper and then using a super fine ISO P1200 sandpaper. Two kinds
of tests were performed, relaxation indentation test and ramped displacement
test.
41
Figure 37: The indentation tests setup
hi (µm) hr (µm)
20 50
30 60
42
5.2.2 Machine Compliance
To measure the compliance of the testing setup and approximate method was
used. An indentation was performed on a high strength steel specimen with the
fully assembled indenter setup, with the assumption that the high strength steel
specimen deformation is negligible then all the recorded displacement could be
attributed to the testing machine. Only the unloading curve was considered to
mitigate any effects of plastic deformation. The unloading load-depth curve for
the machine is shown in 39, fitted with a linear equation.
All the obtained indentation depth results are then corrected by subtracting
the amount of machine displacement that corresponds to the applied load.
10
6
h (µm)
2
Data
Fit
0
-2
0 50 100 150 200 250 300 350 400
Load (N )
43
Indentation Load
75
70
65
Load (N )
60
55
50
45
0 100 200 300 400 500 600
time (s)
Figure 40: Load change in time during the relaxation indentation test
Indentation Depth
43
42
41
h (µm)
40
39
38
37
0 100 200 300 400 500 600
time (s)
Figure 41: Indentation depth change in time during the relaxation indentation
test
have been performed with strain measurements. The distance from one strain
gauge to the indent has been measured using a microscope to be 6.7873 mm.
44
Axial Strain
-19
-20
-21
ǫx (microstrain)
-22
-23
-24
-25
-26
-27
0 100 200 300 400 500 600
time (s)
Figure 42: Strain change in time during the relaxation indentation test
For each indentation depth 5 tests were performed and averaged, figure 43
shows the averaged results of the shear relaxation modulus for each indentation
depth together with the result from the uniaxial tensile test.
Figure 44 shows the averaged shear relaxation modulus between all tests
except the test at 70 µm; due to its clear deviation from the linear viscoelasticity
regime. The error bars represent a 95% confidence interval. A comparison
between the shear relaxation modulus obtained from the indentation relaxation
test and the uniaxial test is shown in figure 45 and the calculated difference
error is shown in figure 46.
The mean shear relaxation modulus obtained follows the following function:
h i
G(t) = G0 × 1 − g1 × 1 − e−t/τ1 − g2 × 1 − e−t/τ2 (26)
Where:
G0 = 1297 M P a
g1 = 0.1260 g2 = 0.2152 τ1 = 27.35 τ2 = 273.50
45
Indentation Relaxation T est
1700
40 µm
1600 50 µm
60 µm
1500 70 µm
Uniaxial
1400
1300
G (M P a)
1200
1100
1000
900
800
700
0 100 200 300 400 500 600
time (s)
Figure 43: Shear relaxation modulus obtained at different depths for the
indentation relaxation test
1400
1300
1200
G (t)
1100
1000
900
800
700
-100 0 100 200 300 400 500 600 700
time (s)
Figure 44: Averaged shear relaxation modulus for the indentation relaxation
test with 95% confidence interval
46
M ean Shear Relaxation M odulus
1350
Indentation Relaxation Test
1300 Uniaxial Creep Test
1250
1200
1150
G (t)
1100
1050
1000
950
900
850
0 100 200 300 400 500 600
time (s)
47
P ercentage Dif f erence Between the U niaxial and Indentation Relaxation T ests
12
10
8
P ercentageerror(%)
0
0 100 200 300 400 500 600
time(s)
Figure 46: Difference error between shear relaxation modulus from indentation
relaxation and the uniaxial creep test
48
5.2.4 Ramped Indentation Tests Results
The direct recording from the ramped indentation tests are the load and in-
dentation depth. Due to the inaccuracy of the used strain gauges the strain
measurement was not used. This allowed the freedom to indent on any surface
of the specimen. The average bulk modulus function value obtained from the
uniaxial tests was used and considered a constant value. Equation 15 is used
to obtain the the viscoelastic parameters of the shear relaxation modulus. Two
Prony series parameters were used to describe the shear relaxation function;
and thus the non-linear regression has to obtain five parameters that would
minimize the equation. Examples of the obtained load and indentation depth
are shown in figures 47 and 48.
Indentation Load
80
70
60
Load (N )
50
40
30
20
10
0 100 200 300 400 500 600
time (s)
Figure 47: Load change in time during the ramped indentation test
49
Indentation Depth
55
50
45
40
h (µm)
35
30
25
20
15
0 100 200 300 400 500 600
time (s)
Figure 48: Indentation depth change in time during the ramped indentation
test
1800
1600
G (t)
1400
1200
1000
800
0 100 200 300 400 500 600
time (s)
Figure 49: Averaged shear relaxation modulus from indentation ramped test
compared to the indentation relaxation and uniaxial creep tests
50
G0 = 2025 M P a
g1 = 0.3475 g2 = 0.2754 τ1 = 24.28 τ2 = 305.90
2500
2000
G (t)
1500
1000
500
-100 0 100 200 300 400 500 600 700
time (s)
Figure 50: Averaged shear relaxation modulus for the indentation ramped test
with 95% confidence interval
51
P ercentage Dif f erence Between the U niaxial T est and Indentation Ramped T ests
90
80
70
60
P ercentageerror(%)
50
40
30
20
10
0
0 100 200 300 400 500 600
time(s)
Figure 51: Difference error between shear relaxation modulus from indentation
ramped test and the uniaxial creep test
52
Chapter 6
Discussion
53
6.2 The Tensile Test
Creep tests were performed due to the observation that the load frame could
control the load better than the axial extension or piston position. As mentioned
in section 5.1.1, two different tests were performed at the same load level to
record the axial extension/strain and the circumferential strain; this has been
done due to the limitations of the output channels that could be connected to
the testing machine simultaneously.
From figures 35 and 36 is is apparent that the POM material shows a large
scatter which is expected from polymer materials. The scatter depends on the
specimen, test number and the load level. The scatter due to the different
specimens is expected, though the scatter due to the test number is probably
due to the relaxation behaviour of the material, i.e. the material needs a very
large amount of time to return to its original state. The scatter due to the load
level however shows that the material has a slight load dependency although
the efforts made to stay in the linear viscoelastic regime. The tests data with
the mentioned scatter can be seen in figures 52, 53 and 54 in appendix B. It is
worth mentioning that the relaxation behaviour or the E(t) curve retains the
same shape for the different tests.
• The strain gauge backing should have a lower stiffness than the tested
material to avoid specimen reinforcement.
54
figure 46 can be considered small. Further examining of individual tests as in
figures 55, 56, 57 and 58 shows that there is little stress dependency except for
the case of indentation at the depth of 70 µm. This is speculated to be due
to the deviation of the material from the linear viscoelastic behaviour at high
stress/strain levels.
55
Chapter 7
Conclusion
• The machine acting as a spring exerting force on the sample for low stiff-
ness high relaxation materials.
From chapter 4 it is determined that the solutions obtained using the functional
integrals method are valid and can accurately measure the shear relaxation
modulus and the bulk viscoelastic function/Poisson’s ratio.
In experiments on POM however, measurement uncertainties introduced er-
rors into the extracted viscoelastic functions, especially in the beginning of the
testing period, these errors could be summarized into three sources: depth mea-
surement, load measurement and strain measurement.
56
• Load Measurement Errors: This is the major source of error in this
report as argued in the discussion chapter in section 6.4. A small load
cell should be used. This is especially essential for a soft material like
bitumen.
57
Bibliography
58
of nanoindentation”. In: International Journal of Materials Research 98.5,
pp. 404–413.
Jelagin, D. and P.-L. Larsson (2013). “Measurement of the Viscoelastic Proper-
ties of Bitumen Using Instrumented Spherical Indentation”. In: Experimental
Mechanics 53.7, pp. 1233–1244.
Johnson, K. L. (1985). Contact Mechanics. Cambridge Books Online. Cambridge
University Press. isbn: 9781139171731.
Larsson, P.-L. and S. Carlsson (1998). “On microindentation of viscoelastic poly-
mers”. In: Polymer Testing 17.1, pp. 49 –75.
Lee, E. H. and J. R. M. Radok (1960). “The Contact Problem for Viscoelastic
Bodies”. In: Journal of Applied Mechanics 27.3, pp. 438–444.
Lu, H. et al. (2003). “Measurement of Creep Compliance of Solid Polymers by
Nanoindentation”. In: Mechanics of Time-Dependent Materials 7.3, pp. 189–
207.
Odegard, GM, TS Gates, and HM Herring (2005). “Characterization of vis-
coelastic properties of polymeric materials through nanoindentation”. In:
Experimental Mechanics 45.2, pp. 130–136.
Ossa, E. and A. Collop (2007). “Spherical Indentation Behavior of Asphalt
Mixtures”. In: Journal of Materials in Civil Engineering 19.9, pp. 753–761.
Ossa, E.A., V.S. Deshpande, and D. Cebon (2005). “Spherical indentation be-
haviour of bitumen”. In: Acta Materialia 53.11, pp. 3103 –3113.
Oyen, M. L. (2006). “Analytical techniques for indentation of viscoelastic ma-
terials”. In: Philosophical Magazine 86.33-35, pp. 5625–5641.
Qiang, Bo et al. (2011). “Estimating material elasticity by spherical indentation
load-relaxation tests on viscoelastic samples of finite thickness”. In: Ultra-
sonics, Ferroelectrics, and Frequency Control, IEEE Transactions on 58.7,
pp. 1418–1429.
Starkova, O. and A. Aniskevich (2007). “Limits of linear viscoelastic behavior of
polymers”. In: Mechanics of Time-Dependent Materials 11.2, pp. 111–126.
Ullner, Christian et al. (2010). “Effect and measurement of the machine compli-
ance in the macro range of instrumented indentation test”. In: Measurement
43.2, pp. 216–222.
VanLandingham, M. R. et al. (2005). “Viscoelastic characterization of poly-
mers using instrumented indentation. I. Quasi-static testing*”. In: Journal
of Polymer Science Part B: Polymer Physics 43.14, pp. 1794–1811.
Veytskin, Yuriy, Christopher Bobko, and Cassie Castorena (2014). “Nanoinden-
tation investigation of asphalt binder and mastic viscoelasticity”. In: Inter-
national Journal of Pavement Engineering 17.4, pp. 363–376.
Wang, L. (2010). Mechanics of Asphalt: Microstructure and Micromechanics:
Microstructure and Micromechanics. McGraw-Hill Education. isbn: 9780071640978.
59
Zofka, A. and D. Nener-Plante (2011). “Determination of Asphalt Binder Creep
Compliance Using Depth-Sensing Indentation”. In: Experimental Mechanics
51.8, pp. 1365–1377.
60
Appendices
61
Appendix A
This appendix pertains to the derivation of the linear elastic equations in section
3.1 that are used to derive the viscoelastic solution.
p(r) = po a2 − r2 /a
(27)
Which is used to obtain the relation between the pressure and the total load
can be obtained as:
3 P
po = (28)
2 πa2
The indentation depth is given by:
1 − ν2 πpo a
h = ūz (r = 0) = × (29)
E 2
62
A.2 Circumferential Strain Relation
Using the relation relating the circumferential strain, outside the contact area,
to the contact pressure and the contact area from Johnson (1985):
1 (1 − 2 ν) (1 + ν) pa2
θ (r) = − (30)
2 Er2
The relation between the pressure and the applied load outside the contact
area can be given as in equation 31:
P
p= (31)
π a2
63
Appendix B
Experimental Results
3000
2900
2800
2700
0 100 200 300 400 500 600 700 800 900
time (s)
64
U niaxial T ensile T est E(t) [Specimen 2]
3400
400 N, Test 1
400 N, Test 2
3200 300 N, Test 1
300 N, Test 2
200 N, Test 1
200 N, Test 2
3000
All tests average
E (M P a)
2800
2600
2400
2200
0 100 200 300 400 500 600 700 800 900
time (s)
3100
3000
2900
2800
2700
2600
0 100 200 300 400 500 600 700 800 900
time (s)
65
B.2 Indentation Relaxation Tests
The results from the relaxation indentation tests are presented in the figures
below for the different indentation depths. The uniaxial tensile test averaged
value for G(t) is also presented in the figure for comparison.
Relaxation T est 40 µm
1600
test1
1500 test2
test3
1400 test4
test5
Uniaxial
1300
G (M P a)
1200
1100
1000
900
800
700
0 100 200 300 400 500 600
time (s)
66
Relaxation T est 50 µm
1600
test1
test2
1400 test3
test4
test5
Uniaxial
1200
G (M P a)
1000
800
600
400
0 100 200 300 400 500 600
time (s)
Relaxation T est 60 µm
1800
test1
test2
1600
test3
test4
test5
1400
Uniaxial
G (M P a)
1200
1000
800
600
400
0 100 200 300 400 500 600
time (s)
67
Relaxation T est 70 µm
2000
test1
test2
1800 test3
test4
test5
Uniaxial
1600
G (M P a)
1400
1200
1000
800
0 100 200 300 400 500 600
time (s)
68
B.3 Indentation Ramped Tests
The results from the ramped indentation tests are presented in the figures below
for the different indentation depths. The uniaxial tensile test averaged value for
G(t) is also presented in the figure for comparison.
20 µm (initial) 50 µm (f inal)
4500
test1
4000 test2
test3
test4
3500 test5
Uniaxial
3000
G (M P a)
2500
2000
1500
1000
500
0 100 200 300 400 500 600
time (s)
Figure 59: G(t) for the ramped indentation test at depths 20 µm initial and 50
µm final
69
30 µm (initial) 60 µm (f inal)
2600
test1
2400 test2
test3
2200 test4
test5
2000 Uniaxial
1800
G (M P a)
1600
1400
1200
1000
800
600
0 100 200 300 400 500 600
time (s)
Figure 60: G(t) for the ramped indentation test at depths 30 µm initial and 60
µm final
70
www.kth.se