Validering av materialmodell for

polypropylen (HDPE)

Tonje Hegni

Master i produktutvikling og produksjon

Innlevert: Juni 2012
Hovedveileder: Arild Holm Clausen, KT
Medveileder: Marius Andersen, KT

Norges teknisk-naturvitenskapelige universitet

Institutt for konstruksjonsteknikk
Department of Structural Engineering ACCESSIBILITY

Faculty of Engineering Science and Technology OPEN

NTNU- Norwegian University of Science and Technology

MASTER THESIS 2012

SUBJECT AREA: DATE: NO. OF PAGES:
POLYMER ENGINEEING 11 JUNE 2012 72 + 45

TITLE:

Validation of material model for polyethylene (HDPE)

Validering av materialmodell for polyetylen (HDPE)

BY:

Tonje Hegni

SUMMARY:
This thesis is concerned with validation of a hyperelastic – viscoplastic material model, which
previously has been implemented in the finite element code LS – Dyna. The material considered
in this thesis is a high – density polyethylene (HDPE) produced by a German company SIMONA.
The material parameters of the constitutive model are found from experimental testing and
calibration.

The model is validated by comparing results from two types of well – defined experimental tests
with numerical predictions. Firstly, tension tests on a plate with a hole have been performed. The
reason for using a plate with a hole is that the geometry is more complex due to the hole which
results in a more complex stress and strain state. It also has well – defined boundary conditions.
There was also performed experimental testing of an impact problem. A real component used in
the industry is stretched, compressed and subjected to different velocities, and an impact
problem provides a realistic representation of real life circumstances.

The experimental and the simulated tests are evaluated and compared. The model captures the
main features observed in the tests.

RESPONSIBLE TEACHER: Professor Arild H. Clausen

SUPERVISOR(S): Arild H. Haugen, Marius Andersen and Anne Serine Ognedal

CARRIED OUT AT: SIMLab, NTNU

MASTEROPPGAVE 2012

Tonje Hegni

Validation of material model for polyethylene (HDPE)

(Validering av materialmodell for polyetylen (HDPE))

As a part of the development of the finite element method, significant effort has been
devoted to propose new material models which are able to represent the material
behaviour at different conditions. Relevant parameters involve strain level, strain rate,
temperature, and the material at hand may also exhibit anisotropy, viscoelasticity etc.
Today, material models for metals are in general more accurate than models for e.g.
polymers. This is a general tendency for all non-linear finite element programs, including
Abaqus and LS-DYNA. SIMLab is involved in the development of improved models for
thermoplastics. A model applicable for ductile polymers was ready in 2010, and the
research work continues with other classes of models.

All new models need to be validated against experimental benchmark tests for different
types of polymers. Validation involves also calibration, requiring material tests on the
material at hand. This master thesis is concerned with a HDPE material produced by the
German company SIMONA. The selected benchmark tests are tension of a plate with a
hole and impact tests on small plates. Subsequently, the test results are to be compared
with numerical predictions applying the material model for ductile thermoplastics.

Some keywords for activities related to this master thesis project are:
 Literature: Polymers in general, relevant test results, material models.
 Experimental tests: Material and component tests. Presentation of test results.
 Calibration: Identify the coefficients of the material model.
 Numerical modelling: Simulation of experimental tests. Evaluation of the model.

The candidate may agree with the supervisors to pay particular attention to specific parts
of the investigation, or include other aspects than those already mentioned.

The thesis is to be organized as a research report, recognising the guidelines provided by

Department of Structural Engineering.

Supervisors: Arild H. Clausen, Marius Andersen and Anne Serine Ognedal

The report is to be handed in at Department of Structural Engineering not later than 11

June 2012.

NTNU, 18 January 2012

Arild H. Clausen

faglærer
Acknowledgement

I wish to express my gratitude to supervisor Arild H. Clausen for outstanding guidance and
support throughout the work. I would also like to thank co – supervisors PhD student Marius
E. H. Andersen for facilitation and guidance. Thanks to Anne – Serine Ognedal for help with
the software 7D and Matlab codes.

The experimental work was performed in cooperation with students Torgrim Østen, Kristin
Sælen and Kjetil Vange. I would like to give them credits for being great supporters and
sharing pictures and Matlab codes during the work. They have contributed both
academically and socially throughout the work.

I would also like to thank Trond Auestad for support during testing in the laboratory at
NTNU.

I
II
Sammendrag
Denne rapporten tar for seg validering av en hyperelastisk – viscoplastisk material modell som
tidligere har blitt implementert i elementmetode koden LS – Dyna. Materialet som er evaluert er
polyetylen (HDPE) som er levert av det tyske firmaet SIMONA. Material parameterne som er satt inn i
den konstitutive material modellen er funnet fra eksperimentell testing og kalibrering.

Modellen er validert ved å sammenligne resultater av to veldefinerte eksperimentelle tester ved

numeriske forutsigelser. Det har blitt utført tester på plate med hull. Platene har en mer kompleks
geometri grunnet hullet i platen, og vil derfor gi mer komplekse spenning og tøynings tilstander. I
tillegg har platene veldefinerte randbetingelser. Disse har en mer komplisert sammensetning av
trykk og strekk og vil derfor være mer realistiske, da en virkelig komponent brukt i industrien vil bli
utsatt for både trykk og strekk samt påført last av ulike hastigheter.

De eksperimentelle og simulerte testene er sammenlignet og evaluert i valideringskapittelet.

Modellen fanger opp de viktigste egenskapene observert i testene.

III
IV
Contents
Acknowledgement....................................................................................................................... I
Sammendrag ............................................................................................................................. III
1. Introduction ............................................................................................................................ 1
2. Theory..................................................................................................................................... 2
2.1 High density polyethylene (HDPE) .................................................................................... 2
2.2 Mechanical behavior ........................................................................................................ 2
2.3 Introduction to the constitutive model ............................................................................ 3
2.4 Intermolecular resistance, Part A ..................................................................................... 3
2.5 Intramolecular resistance, Part B ..................................................................................... 5
3. Material testing ...................................................................................................................... 7
3.1 Introduction .......................................................................................................................... 7
3.2 Experimental program for tensile testing ........................................................................ 7
3.3 Applying an optical technique to obtain stress and strain ............................................... 8
3.4 Results obtained from the tensile tests............................................................................ 9
3.4.1 Comments on force – displacement curves of tensile tests of thickness 5 mm, T1 –
T8......................................................................................................................................... 9
3.4.2 Comments on force – displacement curves of tensile tests of thickness 10 mm, T9
– T13 .................................................................................................................................. 11
3.4.3 Comments on the true stress – strain curve, and determination of yield stress .... 11
3.4.4 Longitudinal versus transverse strain ...................................................................... 14
3.5 Experimental program of compression tests, C1 – C5 ................................................... 15
3.6 Test results of compression tests ................................................................................... 16
3.6.1 Comments on force – displacement curve of the compressions tests.................... 16
3.6.2 Specimens with different extrusion directions ........................................................ 17
3.6.3 Comments on true stress – true strain curves of compression test C1 – C5........... 18
3.6.4 Poisson’s ratio from compression tests ................................................................... 19
3.7 Discussion ....................................................................................................................... 20
4. Calibration ............................................................................................................................ 21
4.1 Introduction .................................................................................................................... 21
4.2 Calibration of parameters............................................................................................... 21
4.2.2 Determination of parameters based on the stress – strain curve........................... 22

V
 4.2.3 Calibration of Part B ................................................................................................. 23
5. Validation ............................................................................................................................. 26
5.1 LS-Dyna simulations of the tensile tests ......................................................................... 26
5.1.1 Modeling the tensile test in LS - Dyna ..................................................................... 26
The Bridgman factor ......................................................................................................... 28
5.1.2 Results from the simulation of tensile tests ............................................................ 29
5.1.3 Simulations of different elasticity modulus’s .......................................................... 32
5.2 Simulation of the compression test................................................................................ 33
5.2.2 Results obtained from the simulation of the compression test.................................. 34
5.3 Plates with holes of diameter 30 mm and 20 mm. ........................................................ 35
5.3.1 Introduction ............................................................................................................. 35
5.3.2 Procedure ................................................................................................................. 35
5.3.3 Applying the DIC method ......................................................................................... 36
5.3.4 LS-Dyna simulations of the plates ............................................................................ 37
5.4 Results obtained from the simulations of plate with a hole ...................................... 38
5.4.2 Results obtained from the plate with a hole of test 02 through test 04 ................. 42
Force – displacement curves of all four tests ................................................................... 49
5.4.4 Young’s modulus ...................................................................................................... 49
5.5 Drop Tower testing ......................................................................................................... 52
5.5.1 Introduction ............................................................................................................. 52
5.5.2 Experimental program of drop tower testing .......................................................... 52
5.5.3 Simulation of drop tower ......................................................................................... 53
5.5.4 Results obtained from the drop tower tests ........................................................... 54
5.5.5 Stresses at the tip of the plate ................................................................................. 58
5.6 Discussion and source of error ................................................................................... 59
6. Conclusion ............................................................................................................................ 61
7. Reference ............................................................................................................................. 62
Appendix A ............................................................................................................................... 63
Appendix B ............................................................................................................................. 101

VI
1. Introduction
This thesis is written in cooperation with Structural Impact Laboratory (SIMLab) which is
located at the Department for Structural Engineering at NTNU. One of SIMLab’s research
areas is dedicated to polymers. They are cooperating with several industries to improve the
existing models of polymers. The most important features that the material model for
polymers should capture is strain rate, temperature change, pressure dependency and
change in volume.

Polymers are becoming more of interest in the industries as there is a higher focus on more
slender and lighter structures. This has led to a need for understanding how polymers work
when exposed to different kinds of loading. The ductility and the low weight make
thermoplastics ideal for energy absorption. This makes polymers a desirable material in car
industries and as protection in impact incidents in the pipeline industry.

SIMLab has been developing a constitutive material model for thermoplastics which is built
up of two parts; Part A is the hyperelastic – plastic model and Part B the entropic
hyperelastic model. The model is built up of thirteen parameters which can be determined
from uniaxial tensile and compression tests. The material that is considered is high density
polyethylene (HDPE).

This thesis first presents some theory about the material and the material model that is
implemented in LS – Dyna. Then, the experimental tests are described by explaining the
procedure and illustrating the results obtained. The calibration method is presented with
necessary equations and explanations. The final section contains the validation of plate with
a hole and the impact tests performed in the drop tower with test results and discussion of
source of error. The thesis is rounded off with a conclusion and a proposition for further
work.

1
2. Theory
2.1 High density polyethylene (HDPE)
High density polyethylene is the polymerization of ethylene, and is built up of long molecule
chains. The material is delivered by a German company SIMONA under the name PE-HWU,
and was delivered as extruded sheets of 2000 x 1000 mm.

HDPE has a high chemical and corrosion resistance, good sliding properties, high UV-
protection and has very good electrical isolation features. This means that the material is
stabilized for outdoor use. The manufacturer SIMONA has not reported any use of fiber or
particle reinforcement in the material. The mechanical properties of HDPE according to the
manufacturer are given in Table 2.1 [1].

Table 2.1: Material properties of HDPE according to the manufacturer SIMONA

Density 0.950 g/cm3

Yield stress 22.0 MPa
Elongation at yield stress 9%
Elongation at tear 300 %
Tensile, E - modulus 800 MPa

This material is used in the building industry, machines, storage technology like boxes and
pallets, as well as in the vehicle sector and where physiological safety is required. HDPE is
though at room temperatures, but not in elevated temperatures at approximately 100˚C.
The reason is that HDPE has a glass transition temperature Tg of approximately 120˚C [1].

2.2 Mechanical behavior

The HDPE monomer has two carbon atoms and four hydrogen atoms linked together by
covalent bonds. The monomers are linked together forming a polymer chain with a carbon
backbone as shown in Figure 2.1. The chains are linked together by weak intermolecular
bonds called van der Waals bonds. This gives the material a more flexible structure
facilitating large plastic deformations caused by relative sliding between the chains. The
weak van der Waals bonds may be overcome by thermal activation even at room
temperature. This is because polymers are in the high temperature regime even at room
temperature. The deformation is therefore time-dependent, and it is not always easy to
distinguish between elastic and plastic deformations. These weak intermolecular bonds are
what affect the elastic properties of the material. The bonds that link the monomers
together are a lot stronger, for these bindings to break there has to be plastic deformation
present [2-3].

2
Figure 2.1: Molecule structure of HDPE

2.3 Introduction to the constitutive model

The material model has two parts. Part A is a hyperelastic-viscoplastic model representing
intermolecular resistance. There are strong bonds in the longitudinal direction and weak
bonds in the transversal direction. Part B is an entropic hyperelastic model representing re-
orientation of molecular chains, which means that the molecules optimize the amount of
entropy. This is intramolecular resistance. Both parts are kinematically described by the
same deformation gradient, and the two parts will therefore have the same change of
volume, expressed with the Jacobian determinant, J = det F =JA = JB. By summing the
contribution from both Part A and Part B the Cauchy stress tensor, σ = σA + σB, is obtained.
Figure 2.2 illustrates the material behavior [4-5].

Figure 2.2: Hyperelastic-viscoplastic model with intermolecular (A) and network contributions (B),
[6].

2.4 Intermolecular resistance, Part A

The Jacobian, which represents the volume change of Part A is decomposed as JA = JeA ∙ JpA.
Similarly, the deformation gradient FA is decomposed into elastic and plastic parts, FA = FeA ∙
FpA, where the plastic part defines an intermediate configuration which is invariant to rigid

3
body rotations of the current configuration. For allowing large elastic deformations a Neo –
Hookean material is chosen for the spring of Part A. The Kirchhoff stress tensor, τ A = JeA σA, is
the elastic stress tensor of Part A, and is written

(2.1)

where BeA is the elastic Cauchy – Green deformation tensor, and is the second order unit
tensor. The coefficients λ0 and μ0 are the Lamé constants from linearized theory, and may be
expressed as functions of Young’s modulus, E0, and Poisson’s ratio, ν0, [4].

For Part A to be isotropic-elastic the Mandel stress tensor is implied to be symmetric. This
stress tensor applies to the viscoplastic contribution of Part A, which is represented in Figure
2.2 with the dashpot and friction element. The Mandel and the Kirchhoff stress tensors are
related through and [4].

The pressure – sensitive behavior that is observed in polymeric materials is accounted for by
the equivalent stress, , and is defined as specified by Raghava,

(2.2)

Invariants of the Mandel stress tensor is represented in the Raghava equation as I1A = tr
and , and the material parameter α = σC / σT 1 represents the
pressure sensitivity, σC is the compressive yield stress. When α = 1, the equivalent stress σA is
equal to the von Mises equivalent stress, [4].

The yield criterion is expressed as

(2.3)

where σT is the yield stress in uniaxial tension and R is an isotropic variable defining
hardening or softening of the material. The R is an expression for the accumulated plastic
strain and is written as

(2.4)

4
σs represents the saturated stress level of Part A. By selecting σs > σT hardening is obtained,
and when σs < σT softening is obtained, i. e. when is negative [4-5].

The flow rule is taken as non-associative to avoid potential unrealistic volumetric strains
utilizing a plastic potential function similar to Raghava, the equation reads

(2.5)

where the volumetric plastic strain is controlled by the material parameter β, and isochoric
plastic strain is obtained when β = 1. By differentiating the plastic potential function, ,
with respect to the total stress invariant, , and the deviatoric stress invariant, , the
functions and are obtained. They are written as

(2.6)

(2.7)

The gradient of the plastic potential is obtained by differentiating the plastic potential
function with respect to Mandel stress tensor, given as

(2.8)

The plastic velocity can then be written as

(2.9)

The is the viscoplastic multiplier defined as

(2.10)

where C and are constants [4-5].

2.5 Intramolecular resistance, Part B

In Part B the theory is built on the fact that the stretch resistance of the polymer network is
of a hyper – elastic nature [4].

The Cauchy stress, σB, is given as

5
 (2.11)

Where is the inverse function of the Langevin function defined (x) = coth x-1/x.

is the distortional effective stretch , where the distortional left Cauchy-

Green deformation tensor is B*B = F*B ∙ (F*B)T, and F*B = JB-1/3∙FB denotes the distortional part
of FB. The two main parameters for the model of intramolecular resistance are the initial
elastic modulus, CR, and the locking stretch, . When all polymer chains are directed in the
stretching direction, the stretching is equal to the locking stretch. In addition the stretching
is done on the chains themselves. When approaches the stresses proceed towards
infinity [4, 6].

6
3. Material testing

3.1 Introduction
The quasi – static material tests were performed under displacement control in a servo-
hydraulic Dartec machine. The capacity of the load cell was 20 kN. The HDPE material was
delivered by SIMONA as extruded sheets of size 2000 x 1000 mm, and with two different
thicknesses, 5 mm and 10 mm. All specimens for tensile testing and compression testing
were machined in the laboratory at NTNU. The layout of the extrusion direction of the
specimen is illustrated in Figure 3.1.

Figure 3.1: Layout of the specimen extrusion direction

3.2 Experimental program for tensile testing

The tensile specimens were made in the xy – plane of the extruded sheet. Altogether, 13
tensile tests were performed. The first eight tests had a thickness of 5 mm, Figure 3.2 a)
illustrates the exact geometry of the specimen. A specimen of thickness 10 mm was applied
in the last five tensile tests, the exact geometry is illustrated in Figure 3.2 b).

Figure 3.2: Nominal Geometry of tensile specimen; Specimen a) was used in test 1 - 8, specimen b)
was applied in test 9 – 13 [7].

The main purpose of performing tensile tests was to determine true strain and true stress
curves at different strain rates. Only two identical tests were performed at each strain rate
to determine contingencies within the tests. If the force – displacement curves at each strain
rate had a significant abnormal variation a third test was performed. Each specimen where
placed into the servo - hydraulic Dartec machine by a wedge-mechanism in mechanical grips.
All specimens had a gauge length of 33 mm. Each test was monitored with a Canon camera

7
placed close to the specimen, but in range to get the full displacement length in the xy –
plane. The amount of pictures taken during the test was dependent on the strain rate
subjected to each specimen. This was to ensure at least 100 photos of each test performed.
There was only one camera taking photos from only one direction, this means that the
displacement over the thickness was not measured during the test. However, each specimen
was measured using a digital sliding caliper before and right after the test, this was to see
the change in thickness in both x and z direction, calculating the ratio of these two values
shows whether the material is isotropic or anisotropic. An overview of all tensile tests can be
seen in Table 3.1 below.

Table 3.1: Overview of all tensile tests, T denotes tension specimen.

Test Strain rate Thickness, t0 Width, w0 Max load Yield stress

# [s-1] [mm] [mm] [N] [MPa]
T1 10-3 4.91 12.00 1218 22.7
T2 10-3 4.90 12.01 1218 22.9
T3 10-2 4.91 11.94 1475 27.8
T4 10-2 4.90 11.96 1425 26.6
T5 10-2 4.90 11.96 1409 26.6
T6 10-1 4.89 11.95 1619 30.2
T7 10-1 4.91 11.95 1621 29.9
T8 10-1 4.90 11.95 1677 30.9
T9 10-2.5 9.76 11.97 2750 26.2
T10 10-2.5 9.76 11.95 2767 26.5
T11 10-2 9.76 11.96 2985 28.0
T12 10-1 9.76 11.96 3307 30.5
T13 10-3 9.76 11.96 2552 24.2

3.3 Applying an optical technique to obtain stress and strain

Thermoplastics may experience volume change during plastic deformation, therefore the
well known formula for true stress, σ = S(1+e), relating nominal and true stress values cannot
be used. This equation assumes that the material is incompressible, i.e. A∙L = A0 ∙ L0, which
necessarily does not apply to plastics as HDPE. However, it is reasonable to assume that
HDPE is an incompressible material, but due to the localization of stresses implying a non –
homogenous strain field in the sample, a measurement of the full deformation field is
needed. The formula for true strain, ε = ln(1+e), on the other hand is applicable for HDPE [8].

Another aspect of thermoplastics is that necking occurs rather early in tension testing, this
result in a non-homogeneous deformation field. A good way to avoid these problems is to
measure the strains by using an optical technique [8].

Each specimen was first sprayed with white paint, and then a quick spray with black painting
on top, see Figure 3.3. This made each specimen contain a black and white spotted
appearance that would be visible on the images. The pattern deforms with the specimen,

8
and the deformation is captured by a camera with high resolution during the test. This
optical technique for measuring displacement is called digital image correlation (DIC). A
program called 7D processed the pictures by applying a mesh to the specimen, and
calculating the displacements as it recognizes the positions of the mesh nodes when
minimizing a DIC. Each picture was compared with the first one. From the displacements of
each node the in-plane Green-Lagrange strain tensor, E, was determined [8].

Figure 3.3: Black and white pattern of the tensile test.

The longitudinal strains and the transverse strains were obtained from the software 7D
through a Matlab script. By plotting an image of the specimen in Matlab it was easy to see
where the necking had initiated and the area where the highest strains were. A Matlab script
was written to find the exact section with the largest strains. These were then used to
calculate the cross section area of the specimen in each picture by using the formula
, where εw is the transversal strains, and then calculating the true stress, σ =
F/A. The transverse strains are assumed to be the same both in the thickness direction, z,
and in the width direction, y. This has not been tested in this thesis, but in a previous master
thesis [6].

3.4 Results obtained from the tensile tests

3.4.1 Comments on force – displacement curves of tensile tests of thickness 5 mm,

T1 – T8
Figure 3.4 shows the force – displacement curves from all tensile tests with thickness 5 mm.
By comparing the two first tests with strain rate 10-3 s-1 it can be seen that they are close to
identical, thus a third test was not needed in this case. The strain rate is increased to 10-2 s-1
and it is seen that the force peak is elevated with approximately 200 N. Tensile test T3 and
T4 turned out to have some variations; the two tests have different maximum force, and the
falling part of the force – displacement curve is not in conformity like in test T1 and T2. A
third test, T5, was necessary to reveal which one of them gave an abnormal force –
displacement curve. By comparing all three of these curves it came clear that test T3 was the
abnormal curve, it had a higher max force and the yield stress differs with 1 MPa from both
test T4 and T5, this can be seen in Figure 3.4 (b). The strain rate was increased to 10-1 s-1,
this is quite fast, and it is clear that it affects the material behavior. The force peak increases
with yet another 200 N, and also in this case a third test seemed reasonable. By comparing

9
test T6 and T7 it is seen that T6 has a steeper declining than T7, and that the maximum force
differ some. However, it appears that test T8 came in between test T6 and T7 in the falling
part of the force – displacement curve, and in addition the maximum force increased with 50
N. This rate-dependent behavior is called viscoplasticity and this theory is applicable in
structural impact problems, forming operations and creeps and stress relaxation. Quasi –
static strain rate at yield stress is lower than the yield stress at elevated strain rate, this can
be seen in Figure 3.7 [9]. Temperature may also be an effect to this phenomena, the melting
temperature for HDPE is between 126 – 130 ˚C, this means that an elevation of 30 – 40 ˚C is
quite significant for this material. An increased velocity leads to an increase in temperature
in the material which changes the material behavior [10].

1.5 1.5
T1 T3
T2 T4
T5
1 1

Force [kN]
Force [kN]

0.5 0.5

0 0
0 10 20 30 40 50 60 -10 0 10 20 30 40 50 60
Displacement [mm] Displacement [mm]

(a) Force – displacement at (b) Force – displacement at

2 2
T6 T1 - 0.001
T7 T5 - 0.01
1.5 T8 1.5 T7 - 0.1
Force [kN]

Force [kN]

1 1

0.5 0.5

0 0
0 20 40 60 80 0 20 40 60 80
Displacement [mm] Displacement [mm]

(c) Force – displacement curve at = 10-1 s-1 (d) Comparison of the force – displacement
curves, the strain rate is indicated in the
figure.

Figure 3.4: Force - displacement of all tensile tests with strain rates ranging from 10-3 s-1 to 10-1 s-1

10
3.4.2 Comments on force – displacement curves of tensile tests of thickness 10
mm, T9 – T13
It was performed five tensile tests of thickness 10 mm. Tensile tests T9 and T10 were both
subjected to a strain rate of 10-2.5 s-1, and by illustrating Figure 3.7 the two curves are seen to
be close to identical. Further on, test T11 was subjected to a strain rate of 10-2 s-1, test T12 a
strain rate of 10-1 s-1 and tensile test T13 was subjected to a strain rate of 10-3 s-1. There was
only performed one test at each strain rate. By comparing the five curves illustrated in Figure
3.5 it shows that HDPE is a strain rate dependent as seen for the tensile tests of thickness 5
mm. The maximal force is higher for these tests, but this is expected since the thickness of
the specimen is twice as thick. The yield stress is not particularly influenced by the change in
thickness.

HDPE is a ductile material, and it does not experience fracture. The tensile tests were
therefore stopped when the force stabilized itself. This applies for all tensile tests, T1 – T13.

3.5
T9
3 T10
T11
T12
2.5 T13

2
Force [kN]

1.5

0.5

-0.5
0 10 20 30 40 50 60 70 80
Displacement [mm]

Figure 3.5: Force - displacement of tensile tests made from sheets of thickness 10 mm

3.4.3 Comments on the true stress – strain curve, and determination of yield stress
Figure 3.6 show that the strain rate has a significant influence on the stress – strain curve as
well as for the force – displacement curve. The reason why the stress – strain curve keeps
rising is due to hardening in the material.

11
After the yield stress has been reached, the specimen starts to neck in some region due to
local softening. If the deformation continues, more and more chain molecules are drawn and
straightened in parallel. The more noticeable the drawings of the chains are the more loaded
will the covalent bonds be, and this causes a local hardening. This will overcompensate for
the reduction in cross section and prevents further necking in this area. The necking will
continue to grow until the whole specimen consists of drawn molecule chains. The strains
can be as high as 300 % [2].

50 40

40

True stress [MPa]

True stress [MPa]

30
30
20
20
T1
T3
T2 10
10 T4
T5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1
True strain True strain

(a) True stress – strain at s-1 (b) True stress – strain at s-1

50 50
T6
40 T7 40
True stress [MPa]

True stress [MPa]

T8
30 30
T9
20 20 T10
T11
10 10 T12
T13
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain True strain

(c) True stress – strain at s-1 (d) True stress – strain at

12
 50

40
True stress [MPa]

30

20
T1 - 0.001
10 T5 - 0.01
T7 - 0.1
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain

(e) Comparisson of true stress – strain curves

at strain rates indicated in the figure
Figure 3.6: True stress - strain curves a – c tensile specimens of thickness 5 mm, d tensile specimens
of thickness 10 mm.

The true stress – true strain curves in Figure 3.6 show that the curves do not have a distinct
stress peak. The reason for this is that the material is experiencing hardening rather than
softening. This means that the stress curve continues to rise, and it therefore makes it hard
to see where the yield stress is. If the material were experiencing softening, there would
have been a distinct stress peak, and calculations would not have been necessary. Anyway,
the yield stress is calculated by plotting the true stress and the differentiated stress against
the true strain. By plotting both of these curves in the same plot a point of intersection is
obtained, this point is the yield stress for the test [10]. The equation for the differentiated
stress for discrete data points reads

(3.1)

An overview of the yield stresses for all tensile tests are given in Figure 3.7.

13
 31

30

29
Yield stress [MPa]

28

27

26

25

24

23 T1 - T8, t = 5 mm
T9 - T13, t = 10 mm
22 -3 -2 -1
10 10 10
Logarithmic strain rate

Figure 3.7: Yield stress for all tensile tests, T1 - T13

3.4.4 Longitudinal versus transverse strain

Figure 3.8 shows the longitudinal and absolute value of transverse strain as a function of
time for test T1. An important observation of this is that the transverse strains are close to
half the longitudinal strains, this result in a Poisson’s ratio, ν, equal to 0.47. The strains
through the thickness and over the width are considered equal, as mentioned in Section 2.2,
and both of them are negative and multiplied with the Poisson’s ratio, while the longitudinal
strains are not multiplied with the Poisson’s ratio. By adding together the strains, εx, εy and
εz, it can easily be seen that HDPE has a volume change of 6 %. This is not a significant
amount for the material, and HDPE is considered an isochoric material.

14
 1.4

1.2

1
True strain

0.8

0.6

0.4

0.2
Tranversal strain
Longitudinal strain
0
0 200 400 600 800 1000 1200
Time [s]

Figure 3.8: True stress - time curve of tensile test 1

3.5 Experimental program of compression tests, C1 – C5

Five compression tests were performed, four of them with height 8 mm and the last had a
height of 5 mm. The compression specimens of height 5 mm was made in the xy – plane
from the plate of thickness 5 mm, while the specimen of height 8 mm was made in the yz –
plane from the plate of thickness 10 mm. The geometry illustrated in Figure 3.9 a) and b)
shows the exact geometry of the test specimens.

Figure 3.9: Exact geometry of compression specimen.

The axisymmetric compression specimen was inserted into the servo – hydraulic Dartec
Machine, and three different strain rates were used on the five tests. The first two tests, C1
and C2, was subjected to a strain rate of 10-3 s-1, the third, C3, and the fifth,C5, tests were
subjected to a strain rate of 10-2 s-1, these two tests is made out of two different extrusion
sheets with two different thicknesses of respectively 5 mm and 10 mm. Only compression
test four, C4, was subjected to a strain rate of 10-1 s-1. The main purpose of testing the
specimens at different strain rates is to see how the material behaves, if the yield stress will
differ from one test to another by only changing the velocity of compressing.

15
Grease was applied to each specimen to avoid as much friction as possible. This facilitates
free movement of the bottom and the top surface in radial direction, which contributes to
keep the edges straight. The formula for true stress, σ = F/A, and true strain, ε = ln(H/H0), is
not applicable when the specimen has a barrel shape because the stresses varies over the
height of the specimen as the area is changing. It was therefore important to avoid friction
to retain straight edges. Figure 3.10 illustrates the set up of a compression test.

Figure 3.10: Setup of a compression test

Although the camera was used on the compression tests as well as the tension tests, digital
image correlation was not performed on the compression tests. The displacements were
obtained by using Matlab since the black and white pattern could not be applied to the small
compression specimen. A Matlab script was then written to obtain the change in diameter
and height, since Matlab does not operate with units all lengths were given in pixels. True
strain and true stress was calculated from the lengths obtained. The diameter was measured
at the middle of the sample, where the strains evolve most in the beginning of the test. The
cross section area of the specimen was calculated by using A = (π/4)d2. The diameter was
possible to calculate since d0 was measured before placing the specimen into the servo –
hydraulic machine, and the Matlab script gave the diameter in pixels at every photo.

3.6 Test results of compression tests

3.6.1 Comments on force – displacement curve of the compressions tests

The force – displacement curve for all five of the compression tests are shown in Figure 3.11.
The two first tests, C1 and C2, were subjected to a strain rate of 10-3 s-1. These two tests are
as similar as one can hope for. Further on test 3, C3, is subjected to an increased strain rate
of 10-2 s-1. Test C5 was also subjected to the same strain rate, but as the height was only 5
mm, the force – displacement curve is quite different from the other tests. The deviation is
due to the height which is 3 mm shorter than the other test specimens. Finally, test C4 was
subjected to a strain rate of 10-1 s-1. Similar to the tension tests, the compression tests also
experience an elevation in forces when the strain rate is increased. Clearly, the material
behaves quite similar both in tension and compression. An overview of the five compression
tests are given in Table 3.2.

16
 7000

6000

5000
Force [N]

4000

3000
C1 - 0.001
2000 C2 - 0.001
C3 - 0.01
1000 C4 - 0.1
C5 - 0.01
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Displacement [mm]

Figure 3.11: Force - displacement curves of compression specimens, the strain rate is indicated in
the figure.

Table 3.2: Overview of compression tests, C denotes compression specimen.

Test # Strain rate [s-1] Height, h0 [mm] Yield stress [MPa]

C1 10-3 8.02 25.1
-3
C2 10 8.01 24.3
-2
C3 10 8.02 28.0
-1
C4 10 8.02 31.2
C5 10-2 4.94 27.0

3.6.2 Specimens with different extrusion directions

The longitudinal axis of the C3 compression specimen was parallel with the extrusion
direction of the plate, while specimen C5 was directed in the thickness direction of the
extruded sheet. In addition the C5 specimen has a height of 5 mm instead of 8 mm. Both
test specimen C3 and C5 was subjected to a strain rate of 10-2 s-1, and therefore it is
reasonable to compare these two tests. Figure 3.12 illustrates the longitudinal strains in x
and z direction of respectively compression test 3 and 5. The two curves are close to parallel.
This means that there are no significant differences in the longitudinal strains when the
specimen is machined from a different direction of the extruded sheet. The true stress – true
strain curve, see Figure 3.13, illustrate no significant difference between compression test 3
and 5.

17
 1.4

1.2

1
True Strain

0.8

0.6

0.4

0.2
C3 - 0.01
C5 - 0.01
0
0 10 20 30 40 50 60 70
Time [s]

Figure 3.12: Comparing longitudinal strains in x and z direction

3.6.3 Comments on true stress – true strain curves of compression test C1 – C5

Figure 3.13 illustrates the true stress – strain curves of all compression tests performed.
These curves show that hardening is present in compression similar as for tension. Although,
the stress hardening is larger in compression than in tension, and therefore makes the two
stress – strain curves different in the two loading modes. This means that the shape of the
curve is pressure dependent.

40

35

30
True stress [MPa]

25

20

15

10

C1 - 0.001
5
C2 - 0.001
C3 - 0.01
0 C4 - 0.1
C5 - 0.01
-5
-0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain

Figure 3.13: True stress - strain of all compression tests, the strain rate is indicated in the figure.

18
The yield stress must be calculated by plotting the stress and the differentiated stress as a
function of the true strains. The point of intersection that is obtained is the yield stress for
the test. This is the same procedure as used in tension, Section 3.4.3. Figure 3.14 illustrates
the yield stress for the five different compression tests performed. As seen in tension, the
yield stress increases when the strain rate is elevated due to hardening and elevated
temperature during the compression test.

32

31

30
Yield stress [MPa]

29

28

27

26
C1 - C4, h = 8 mm
25 C5, h = 5 mm

24 -3 -2 -1
10 10 10
Logarithmic strain rate

Figure 3.14: Yield stress of all five compression tests, the height is indicated as h in the figure.

3.6.4 Poisson’s ratio from compression tests

Figure 3.15 illustrates the transverse strains and the longitudinal strains. The transversal
strains are half the longitudinal strains, resulting in no change of volume when the material
is subjected to compression. This was seen during tension testing as well, since the material
has no significant change in volume neither in compression nor in tension, the material is
isochoric.

19
 0.6

0.5
Tranverse strain

0.4

0.3

0.2 C1
C2
C3
0.1 C4
C5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

Figure 3.15: Transversal strains versus longitudinal strains of all compression tests

3.7 Discussion
The temperature within the material increases due to increased strain rates. As seen in
Figure 3.6 (e) the tensile test of strain rate 10-3 s-1 is intersecting with the two curves of a
higher strain rate due to elevated temperature within the material. The temperature can rise
as much as 20 – 40 ˚C this is a lot for a material with a transition temperature of
approximately 120 ˚C. [10]

The yield stress also increases due to increased strain rate, and one of the reasons is
viscosity due to plastic flow. The material model is compared to a spring with a damping
effect. The forces working on a stretched spring is the displacement multiplied with the
spring stiffness, F = k∙x. The force working on a damper is calculated by multiplying the
viscosity factor, c, with the velocity, F = c∙ [10].

Though, the velocity is constant throughout the test the strain rate however will change as
the length of the sample extends. Another aspect of this matter is the localization of the
strains, as the diffuse necking forms the strains are accelerating in the cross section. As the
localized neck approaches the cross section the strain rate will decrease again. The cross
section is depicted from the DIC measurements.

20
4. Calibration
4.1 Introduction
The material testing has now been done, and it is time to evaluate the results obtained from
the experiments. The yield stress and Poisson’s ratio have already been determined in
Section 3.4.3 and 3.6.3 respectively. When performing the calibration the test results from a
tensile test of strain rate 10-3 s-1 is used as a baseline case. The results from the first tensile
test, T1, were selected. There are also needed results from a tensile test of a different strain
rate as well as results from a compression test. The second compression test C2 was chosen.
This test also had a strain rate of 10-3 s-1.

4.2 Calibration of parameters

First the strain rate dependency has to be determined because of the accurate calibration
based on the baseline tension test. The strain rate that is used corresponds to the yield
stress found in tensile test T1. The parameters ν0 and β are related to the strain rate, and the
slope of the curve in the elastic domain gives the Poisson’s ratio, by plotting the contraction
ratio, ρ = - εt /εl, and the true longitudinal strains the curve will converge toward a value.
This contraction ratio applies to the plastic area.

The volumetric change is controlled by the material constant β. The coefficient can be
calculated from the following equation

(4.1)

The value of ρ is close to 0.5, and β = 1, this means that the isochoric plastic behavior is
obtained [4,6].

In the strain rate sensitivity part of the model the following three coefficients are involved
σT, and C. The equation for the yield stress, , reads

(4.2)

It is obvious that the expressions in the parentheses are equal to one, this means that
equation 4.2 will predict a yield stress, , higher than σyield. To avoid this problem the yield
stress σT is replaced by the static yield stress σT0 by extrapolating back to a fictitious strain
rate . Further on, the reference strain rate, , is fetched from the section where the
specimen experienced the onset of necking, and thereby the initial yielding. According to the
results given from 7D the strain rate at yielding is in the tension tests. The baseline
case is tensile test T1 with the lowest strain rate, this means that the nominal strain rate =
10-3 s-1. This gives a reference strain of = 0.0007 s-1. The logarithmic strain rate, , are
determined from the DIC measurements [4, 10].

21
The parameters C and σS can now be determined as the yield stress and the corresponding
strain rate is found. By plotting the yield stresses from all tension tests of thickness 5 mm
against the logarithmic strain rate and making a logarithmic curve fit, the C is obtained.

(4.3)

where a = C, b is the yield stress σT, y is the yield stress at the reference strain rate and x =
. The abscissa value is observed to be ln(2) = 0.693 for the slowest tests, the
reason for this is that the strain rate is selected as . By the linear curve fit the C =
1.873/21.94 = 0.0854 [4, 6].

4.2.2 Determination of parameters based on the stress – strain curve

The parameters still remaining from Part A are E0, σT, σS and H. These parameters have to be
determined from a stress – strain curve. Again the test with strain rate 10-3 s-1 was selected
as the baseline case.

The response of the material is strongly non-linear in the elastic domain, and it is therefore
complicated to calculate the Young’s modulus. In addition the DIC measurements are
inaccurate for small strains. Therefore the Young’s modulus is taken from the manufacturer
SIMONA, were E0 = 800 MPa. The linear Young’s modulus was plotted in the stress - strain
curve to see if this seemed like a suitable value. Figure 4.1 illustrates that this value is
reasonable.

As previously mentioned the material experience hardening which means that σs > σT.

(4.4)

The best curve fit is to choose a rather low yield stress for σT, and let equation 4.4 above
represent the gradual reduction of stiffness towards the saturated stress, σ S. By plotting the
Young’s modulus and the true stress – true strain curve, the σT is the interception point
between the straight line, E0, and the test data. The σS is equal to the yield stress of strain
rate = 0, this means that σS = 21.9 MPa. According to equation 4.4 the hardening is strain
rate dependent. This means that a similar adjustment have to be done for σ T. The difference
between σS and σT has to be the same for the fully static conditions, thus (σS – σT) = 12.2
MPa. A curve fit between these two test data finds the rate – independent hardening
parameter H = 55.3. Figure 4.1 shows Part A from the observed test data, the analytical
method and the Young’s modulus [4]

22
 30

25
True stress [MPa]

20

15

10
Part A
5 Experimental Part A
E modulus
0
0 0.05 0.1 0.15 0.2 0.25
True strain

Figure 4.1: Comparisson of analytical stress and experimental data of Part A, including the Young’s
modulus E0 = 800 MPa.

4.2.3 Calibration of Part B

The remaining parameters are CR and related to Part B. This identification part has to be
performed with some care due to the factor (B*B - ). The transverse deformation is
assumed isotropic, ε2 = ε3, and the stretches are defined as λ1 = exp(ε1) and λ2 = exp(ε2),
where the logarithmic strains where obtained from the DIC measurements [4, 6]. The
deformation gradient of a uniaxial tension test reads

(4.5)

The stress of Part A is assumed to saturate at σS because Part B is most important at large
deformations [4].

The parameter α = 1.07, and is close to obeying the Mises’ yield criterion. This means that
Part A in the plastic domain, σeq, is corresponding to the uniaxial Part A, . Both Part A
and Part B observed from the uniaxial tension test is allocated as

(4.6)

The longitudinal stress of Part B from the test data can be isolated as follows

(4.7)

The stress component σB1 has to be fitted to the stress of target Part B, . The
Langevin function has no inverse function which is complicating the regression fit of
. By introducing a Padé approximation to proposed by Cohen. [4, 6]

23
 (4.8)

The following expression for the stress component σB1 is given as

(4.9)

The numerical values CR = 1.932 MPa and = 6.095 were obtained by a regression fit of
Equation 4.9 to the target stress of equation 4.7. The total uniaxial stress was found from
equation 4.6. The first part is sensitive to the strain rate, while is independent of
the strain rate [4].

In addition the bulk modulus and the shear modulus had to be calculated, and inserted in
the material card of LS –Dyna. The following equations are used

(4.10)

(4.11)

45

40

35
True stress [MPa]

30

25

20

15

10
Experimental
5 Analytical
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain

Figure 4.2: True stress – strain of experimental data and analytical test.

Figure 4.2 illustrates both Part A and Part B. It can be seen that the curves are almost
identical. An overview of all parameters that are calculated is given in Table 4.1 below.

24
Table 4.1: An overview over the parameters calculated in the calibration

Parameter Value Unit

E0 800 MPa
ρ0 0.47
ν0 0.47
σs 21.9 MPa
σT 10.5 MPa
(σS – σT) 12.2 MPa
α 1.07
H 55.3
β 1.0
0.0007 s-1
C 0.0854
CR 1.93 MPa
6.095

25
5. Validation
In this section the experimental tests are modeled and simulated in the non – linear finite
element program LS – Dyna. The models are evaluated by using an explicit dynamic
calculation method. A material model for ductile plastic materials has been developed, the
material model for elastic – viscoplastic materials is implemented. The main purpose of
modeling and simulating the experimental tests is to see how this implemented material
model works for HDPE. All necessary material properties were calculated in Section 4, and
are being implemented in the material model in Ls – Dyna. The simulations are supposed to
be as similar to the experimental tests as possible. All tests are not simulated, but a variety
of tests subjected to different strain rates are chosen.

5.1 LS-Dyna simulations of the tensile tests

There were simulated three tensile tests in LS – Dyna at nominal different strain rates of
respectively 10-3 s-1, 10-2 s-1 and 10-1 s-1. There was not simulated any tensile tests of
thickness 10 mm since there was no significant differences.

5.1.1 Modeling the tensile test in LS - Dyna

The whole tensile specimen was modeled in LS – Dyna by using solid elements. There was
not used any symmetry planes on this model, although it would have been possible to have
two symmetry planes; one in the thickness direction and one in the width direction. During
the test the specimen was clamped at each end and pulled upwards by one of the
mechanical grips. The model is therefore fixed at one end, and subjected to a displacement
controlled loading in the positive x – direction. It was necessary to restrict the rotation
around the loading direction because the end of the plate started to rotate around the x –
axis. This would never have happened in the experimental test since the mechanical grips
would not have allowed this. Figure 5.1 illustrates the model.

Figure 5.1: Mesh of tensile test

The tensile test of strain rate 10-3 s-1 was the first test that was modeled in LS – Dyna. All
necessary parameters have been determined. The material model explained in Section 2.3

26
was implemented in LS – Dyna as a user-defined material model working for shell and brick
elements. A fracture criterion is not implemented in the model [4].

As seen in Figure 5.1 the elements are longer in the width direction than in the longitudinal
direction. This is to avoid that the elements get too drawn out in the necking zone. When
this happens the accuracy is poor, and the force – displacement curve gets rough. There are
four elements trough the thickness of the tensile specimen. The element formulation was
set to -1 for fully integrated solid elements. This formulation is efficient and intended for
elements with poor aspect ratio [11]. In addition, the hourglass effect is avoided when
applying full integration. The model is quite big and contains many elements, to reduce the
simulation time the model has been mass scaled. This means that the density is increased to
make the time step larger and less iterations are performed [12]. The density of the material
was reduced from 0.950 ∙ 10-9 ton/mm3 to 0.950 ton/mm3. The reduction of the density is
large, and it may cause error. Therefore it has been simulated tests specimens with a lower
density, but this does not have any particular influence on the simulations, except longer
simulation time. This applies for the tests of a nominal strain rate of 10 -3 s-1. When the
nominal strain rates have been increased to 10-2 s-1 and 10-1 s-1 the density have been
reduced to 0.0950 ton/mm3 and 0.00950 ton/mm3 respectively. The equation for the critical
time step reads

(5.1)

During tensile testing there has to be accounted for two types of necking; diffuse and
localized necking. When the material reaches its maximum force the test specimen starts to
neck, this can barely or not at all be seen on the test specimen. As the neck keeps growing
the thickness and width of the specimen will be reduced and a clear notch will be visible as
illustrated in Figure 5.2. This necking condition is called localized necking. [13, 14]

Figure 5.2: Localized necking in tensile specimen T1.

As the neck keeps growing the necking area gets harder due to part B. The deformations will
then come to an halt, and the neighboring areas of the neck will start to neck as well, this is
called propagating or drawing, this is illustrated in Figure 5.3 [13, 14].

27
Figure 5.3: Propagated necking in tensile specimen T1, comparing experimental and simulated
neck.

By comparing the simulated and experimental necking it is clear that they are similar. The
neck in the simulation is more distinct. Figure 5.2 is taken from the middle of the
experimental test and simulation, and Figure 5.3 is taken from the last picture in the
experimental test, and in the end of the simulation.

The Bridgman factor

The stress state in a test specimen will change from uniaxial to triaxial after the onset of
diffuse necking. To account for this triaxial behavior Bridgman developed a correction to the
average stress based on the geometry of the neck. Bridgman examined the triaxial stress
state for circular cylindrical test specimens with parameters a and R, where a is the radius of
the test specimen during necking and R is the curvature of the neck. The parameters are
defined in Figure 5.4 [6, 13]

Figure 5.4: Defining the constants R and a in the Bridgeman correction.

The Bridgman corrector factor is expressed as

(5.2)

where is the average longitudinal stress and the is the equivalent uniaxial stress
[Hovden + Diffuse necking]. The Bridgeman corrector was accounted for in the master thesis
of Martin T. Hovden [6], but did not have any significant outcome on the stress – strain curve
and is not accounted for in this thesis.

28
5.1.2 Results from the simulation of tensile tests
The force – displacement of the experimental and simulated results of T1 is compared in
Figure 5.5. There are three main characteristics of importance; the initiation of the curve up
to the force peak, the force peak itself and the stabilized force at the end of the simulation.
The force – displacement curves illustrated in Figure 5.5 shows that the initiation of the
experimental curve is too stiff compared to the simulation done in LS – Dyna. The force peak
of the simulated curve is slightly higher than the experimental force peak. For the
experimental test the force stabilizes at 800 N. The simulated test reaches approximately
900 N, however, the force does not stabilizes but continues to rise slowly due to a secondary
hardening. Altogether the overall shape of the curves is similar.

1400
T1 Experimental
1200 T1 Simulation

1000
Force [N]

800

600

400

200

0
0 10 20 30 40 50 60
Displacement [mm]

Figure 5.5: Force – displacement curve of tensile specimen T1

The true stress – strain curve illustrated in Figure 5.6 shows the similarities between the
simulated stress – strain curve and the experimental. The simulated curve is slightly higher
from about yield stress, but other than that they are approximately similar. It can be seen
from the curves that the experimental curve does not go all the way up due to large
deformations. The black and white pattern could not be recognized by the 7D program any
longer.

29
 70

60

50
True stress [MPa]

40

30

20
T1 Experimental
10 T1 Simulation

0
0 0.5 1 1.5
True strain

Figure 5.6: True stress – true strain curve of tensile specimen T1

Tensile test T5 is subjected to a nominal strain rate of 10-2 s-1. By comparing the two curves
in Figure 5.6 it can be seen that the overall shape of the two curves are approximately
similar. The force peak is shown to be slightly higher for the simulated curve. As seen in
tensile test T1 the simulated curve is stiffer for the simulated curve. The overall shape of the
simulated curve is a good approximation. The force stabilizes at approximately 800 N. Again,
the analysis is seen to experience a second hardening towards the end of the analysis.

1500
T5 Experimental
T5 Simulation

1000
Force [N]

500

0
0 10 20 30 40 50 60
Displacement [mm]

Figure 5.7: Force – displacement curve of tensile specimen T5

Comparison between the true stress – strain curve in Figure 5.8 shows that the experimental
results and the simulation is close to identical.

30
 70
T5 Experimental
60 T5 Simulation

50
True stress [MPa]

40

30

20

10

0
0 0.5 1 1.5
True strain

Figure 5.8: True stress – strain curve of tensile specimen T5

There were three tensile tests subjected to a nominal strain rate of 10-1 s-1, and test T7 was
selected to be simulated. The force – displacement curve, shown in Figure 5.9 (a), of tensile
specimen T7 is similar from the start and up to peak force, the stabilized force is a little
higher as for all other tensile tests. There is a difference in the descending part of the two
curves.

The true stress – strain curve illustrated in Figure 5.9 (b) shows that the experimental and
the simulated curves are close to identical up to approximately 40 MPa, after that the
experimental values of the stress are constant. When the neck starts to propagate the stress
and strains may be approximately constant.

2000 100
T7 Experimental
T7 Simulation
80
True stress [MPa]

1500
Force [kN]

60
1000
40
500
20 T7 Experimental
T7 Simulation
0
0 20 40 60 80 100 0
Forskyvning [mm] 0 0.5 1 1.5 2
True strain

(a)
(b)
Figure 5.9: (a) Force – displacement curve and (b) True stress – true strain curve of tensile test T7

Figure 5.10 shows the true stress – true strain curve of the three tensile tests that was
simulated in LS – Dyna. The nominal strain rate subjected to each of the three specimens is
indicated in the figure. By comparing the three curves it is seen that tensile specimen T7
subjected to a nominal strain rate of 10-1 s-1 is experiencing higher stresses than the other
tests. The experimental true stress – strain curves in Section 3.4.3 illustrated in Figure 3.6 (e)
shows that the curves of a nominal strain rate of 10-3 s-1 crosses the curves subjected to a

31
nominal strain rate of 10-1 s-1. The reason for this is that a specimen subjected to a nominal
strain rate of 10-1 s-1 is experiencing an increased temperature which makes the material
softer. The true stress – strain curve will then flatten. LS – Dyna does not take the
temperature changes into account, and the tensile specimen subjected to a nominal strain
rate of 10-1 s-1 will give a higher true stress – strain curve for the simulated results [10]. This
means that the simulated tensile tests subjected to a nominal strain rate of 10 -3 s-1 will not
cross a simulated tensile test subjected to a nominal strain rate of 10 -1 s-1 as seen for the
experimental results. This presupposes that the tensile test subjected to a nominal strain
rate of 10-3 s-1 is used in the calibration.

90

80

70
True stress [MPa]

60

50

40

30
T1 - 0.001
20 T5 - 0.01
T7 - 0.1
10

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
True strain
Figure 5.10: Comparison of stress – strain curve of tensile specimen T1, T5 and T7.

5.1.3 Simulations of different elasticity modulus’s

In some of the force – displacement curves the elastic part of the curve is seen to be slightly
stiffer. It has therefore been performed a variation of simulation with elasticity modulus
ranging from 400 MPa to 800 MPa, where 800 MPa is the value found in the calibration.
Figure 5.11 illustrates test T1 applied to the various elasticity modulus’s mentioned.

1400
T1 Experimental
1200 T1 Simulation, E = 800 MPa
T1 Simulation, E = 600 MPa
1000 T1 Simulation, E = 400 MPa
Force [N]

800

600

400

200

0
0 10 20 30 40 50 60
Displacement [mm]

Figure 5.11: Comparison of Young’s modulus of tensile test T1.

32
By looking at the force – displacement curves it is clear that the elasticity modulus of 400
MPa is the best fit for the uniaxial quasi – static tensile test T1. Both tensile test T5 and T7
are simulated, and an overview of which elasticity modulus that makes the best fit is
illustrated in Table 5.1.

Table 5.1: An overview of elasticity modulus for each tensile test

Test # Strain rate [s-1] E modulus [MPa]

T1 10-3 400
-2
T5 10 600
-1
T7 10 800

5.2 Simulation of the compression test

The compression test subjected to a strain rate of 10-3 s-1 was modeled in LS – Dyna. As this
was the only test that was used in the calibration of the material model this was the only
compression test that was modeled. The compression test was modeled after the
dimensions that were measured with the sliding caliper before inserting the test specimen
into the Dartec machine. There were used eight node solid elements in the model, and ten
elements through the height of the specimen. The element formulation that was selected
was fully integrated solid elements that are intended for elements with poor aspect ratio
[11]. The same method as used in the tensile simulations. This method is an efficient
formulation.

The only constraint applied is for displacement in the z – direction of the model to prevent it
from moving downwards when the specimen is compressed. In addition there is applied a
prescribed motion in the negative z – direction, and a smooth curve is defined to prevent the
velocity to initiate abruptly which can lead to incorrect simulation results. In the laboratory
each compression specimen was applied a layer of grease in between the specimen and the
metal plates to avoid friction. The friction coefficient was therefore neglected in the
simulations.

The mesh has a distance of 0.5 mm, and is illustrated in Figure 5.12 below.

33
Figure 5.12: The simulation model of the compression specimen C1.

5.2.2 Results obtained from the simulation of the compression test

The force – displacement curve from the experimental and the simulated tests are
approximately very similar. As mentioned in Section 3.6.2 the DIC method was not used in
the compression test. All results from the compression test are therefore determined from a
Matlab script. By comparing the experimental results with the simulated results it can be
seen from Figure 5.13 that the force is slightly higher in the beginning of the ascending part
of the curve. The overall shape is similar.
4500

4000

3500

3000
Force [N]

2500

2000

1500
C1 Experimental
1000
C1 Simulation
500

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Displacement [mm]

Figure 5.13: Force – displacement curve of compression test C1.

Figure 5.14 illustrates the stress – strain curves obtained from the experimental results and
the simulation. The stress and strains from the simulation are taken from selected elements
in the centre of the compression specimen where the strains are supposed to be at its
largest. The average value of the strains and the stresses are plotted into Matlab. The two
curves have a similar overall shape, and the test is in consistency with the simulation. The
force peak seems to be slightly higher for the simulated curve, and the hardening is not as
distinct in the simulation as it is for the experimental curve.

34
 30

25
True stress [MPa]

20

15

10

C1 - Experimental
5 C1 - Simulation

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
True strain

Figure 5.14: True stress – true strain of compression specimen C1.

5.3 Plates with holes of diameter 30 mm and 20 mm.

5.3.1 Introduction
The main purpose of performing the tests on plates with a hole is that the geometry is more
complex than a uniaxial tensile test due to the hole. This results in a more complex stress
and strain state. In addition the boundary conditions and loading conditions are well defined
which is important to make it a good validation object for the material model. The test can
be evaluated both globally and locally. The force – displacement comparison as well as
geometrical considerations as diameter and lateral faces are the global aspects of the
validation. The comparison between the strains close to the hole obtained from the pictures
and the simulated results obtained through the analysis is the local responses of the plate
with a hole tests [10].

5.3.2 Procedure
There were performed four tests on plates with a hole in the centre. The holes had two
different diameters of 30 mm and 20 mm respectively. All plates were dimensioned as 160 x
60 x 5 mm. Each specimen was measured to make sure that the geometries were exact. It
was, however, noted that the holes were not placed perfectly in the centre of the specimen.
A line was drawn 50 mm from centre across the test specimen. This was to place the
specimen correctly into the Dartec machine with the same distance from the hole in both
ends. After the specimen had been measured the specimen was inserted into the Dartec
machine using a load cell with a maximum capacity of 20 kN.

The DIC method was applied on these tests as well as the tensile tests, and the specimens
were sprayed with white and black painting to obtain a black and white spotted pattern. The
specimen was painted on the top side of the specimen covering both lateral faces as well as
the parts over and below the hole. During these tests there were placed to cameras in two
different directions to obtain the change in thickness. As the DIC method was not applied to
the thickness of the specimen the lateral side turned towards the camera was marked with

35
two points 35 mm from centre on each side of the hole. These marks were visible on both
cameras. Figure 5.15 illustrates the two lateral faces w1 and w2.

Figure 5.15: Geometry of the plate with indications of lateral faces and diameter

Two different nominal strain rates were applied to four test specimens. They were estimated
so that the strain rate in the necking area would correspond to 10-3 s-1 – 10-2 s-1. Since the
distance between the mechanical grips and the centre of the hole, where the necking area is
assumed to develop, is approximately 50 mm the velocity will be 0.05 mm/s and 0.5 mm/s.
The first test, 01, and the second test, 02, are subjected to a strain rate of approximately 10-3
s-1, where the diameter is 30 mm and 20 mm respectively. The two last tests, 03 and 04, are
subjected to a nominal strain rate of about 10-2 s-1. An overview of all plates with a hole tests
performed is illustrated in Table 5.2 with their respective measurements.

Table 5.2: An overview of all plates with a hole tests performed.

Test Estimated strain Velocity Width [mm] Diameter Thickness

# rate [s-1] [mm/s] [mm] [mm]
Lateral Lateral
face, w1 face, w2
01 10-3 0.05 15.69 15.29 30.09 4.89
02 10-3 0.05 20.21 20.16 20.06 4.92
03 10-2 0.5 15.12 15.56 29.87 4.89
04 10-2 0.5 20.33 20.11 20.02 4.92

5.3.3 Applying the DIC method

The displacement were obtained by using the DIC method, but to be able to compare the
test results and the simulations the displacement would have to come from the exact same
cross section of the plate. It was chosen to take out the displacements 40 mm from the
centre of the hole. It was necessary to choose a cross section placed high enough so that the

36
displacements had stabilized. In the program 7D it was important that the first row in the
mesh was 40 mm from the centre, if not it would be impossible to find the right row in the
matrix. A Matlab script was then written to obtain only the first row of displacements in
every picture taken. The test setup is illustrated in Figure 5.16, the two marks placed 35 mm
from the centre is clearly seen on the test specimen.

Figure 5.16: Test setup of plate with a hole

The change in diameter and the change of the lateral faces were determined by a Matlab
script. The contrast between the white painted plate and the black background makes it
possible for Matlab to distinguish the width of the lateral faces. Black color is defined as zero
and white as one, by summing the value over the face the width is obtained. Unfortunately
there was not used a perfectly black background this made it difficult for Matlab to
distinguish the background from the plate. The grey color scale from the background have
affected the results, this can be seen from the results obtained as the width suddenly
increases during the experiment. This was not a problem for the compression test as the
background was perfectly white.

5.3.4 LS-Dyna simulations of the plates

The plates were simulated by using a symmetry plane across the length direction. Since it
turned out that the plates was not symmetrically placed over the width a symmetry plane
could not be used in the x - direction. This means that half the model was sketched up in LS –
Dyna, and simulated. By using a symmetry plane in the y - direction the simulation time will
be reduced due to fewer nodes and therefore less iterations. Mass scaling is applied in this
model, and the equation for the critical time step is given in Section 5.1.1 equation (5.1). The
density is set from 0.95 ∙ 10-9 ton/mm3 to 0.095 ton/mm3, though this is a lot there is no
particular difference when applying a lower density except increased running time of the

37
simulation. The element formulation is set to -1 for fully integrated solid elements. This
formulation is efficient and intended for elements with poor aspect ratio [11].

In the symmetry plane the edges are detained in the y direction and all rotations are fixed. In
addition the middle row of nodes on the edges in the symmetry plane is detained in the z –
direction. This way the thickness can vary through the simulation but not the whole
specimen. The top of the model is prescribed a motion in the positive y – direction to stretch
the plate as done in the testing. A smooth curve is defined to prevent abrupt initiation of the
velocity which can lead to incorrect results. The velocity will then reach constant speed after
approximately 10 % of full running time.

Solid elements with eight nodes are chosen for the model. The lower part of the model has
elements that are wider in the x – direction than in the y – direction. The reason for this is
that the material experience large strains due to tensile loading which makes the simulation
inaccurate if not the elements are rectangular before the test is initiated. The mesh is
illustrated in Figure 5.17.

Figure 5.17: The mesh of the plate with a hole.

5.4 Results obtained from the simulations of plate with a hole

All plates with hole have been simulated in LS – Dyna with the correct geometry for each and
every plate since the thickness, diameter and the lateral faces differ some for each test. The
four different plates are applied to two different velocities and two diameters of respectively
20 mm and 30 mm. All four tests are presented as either the velocity or the diameter is
different for all four tests.

5.4.1 Plate with hole of diameter 30 mm, test 01.

The first test performed in the laboratory had a diameter of 30 mm and was subjected to a
velocity of 0.05 mm/s. A comparison of the force – displacement curves of the experimental

38
and simulated data is illustrated in Figure 5.18. There are three main characteristics that are
taken into consideration; the peak force, the stabilized force and the overall shape of the
curve. By looking at Figure 5.18 it is seen that the force peak is close to identical, but that the
simulated curve is slightly stiffer. In addition the simulated force – displacement curve is
experiencing secondary hardening. The overall shape of the simulated force – displacement
curve is satisfying.

3500
01 - Experimental
3000 01 - Simulation

2500
Force [N]

2000

1500

1000

500

0
0 5 10 15 20 25 30
Displacement [mm]

Figure 5.18: Force - displacement curves of plate with hole, d = 30 mm

The change in lateral faces, diameter in both directions and thickness

The lateral faces and the diameter of the plate have been determined by a Matlab script,
where the contrast between the black test specimen and the white background gives the
width of the lateral faces and the dimensions of the diameter in both x – and y – directions.
The main reason for measuring this is to see how the material deforms during tensile loading
of a plate with a hole. By considering the displacement – time curve of lateral face w1 in
Figure 5.19 it is seen that the two curves have the same initial value and ending value. The
overall shape is similar, and the experimental and simulated curves give a good
approximation for the validation. The lateral face w1 stabilizes itself after approximately 300
s. Lateral face w2, on the other hand, is not quite as similar. There was a mark at the outer
side of the lateral face which made the width incorrect. This curve is therefore not illustrated
as there is no reason to compare with the simulation.

39
 16
01 - Experimental, w1
01 - Simulation, w1
Displacement [mm]

14

12

10

8
0 100 200 300 400 500 600 700
Time [s]

Figure 5.19: Width of lateral face w1.

The same procedure was used to determine the thickness of the specimen. Figure 5.20
illustrates the change in thickness during both the simulation and the experimental testing.
It is seen that the shape and the ending value are very similar to the experimental curve. By
comparing the curve shape to the lateral faces it can be seen that neither the thickness nor
the lateral faces are decreasing from approximately 400 s and throughout the test. This is
because the specimen has propagated in the necking area which is illustrated in the figure.
The reduction of the thickness is happening further away from the cross section. The ratio of
the change in thickness is 0.38 while the ratio for the width is 0.56. It is seen that the plate
deforms more in the width direction than in the thickness direction. The thickness of the test
specimen was measured right after testing on the inner and outer sides of each lateral face.
This showed that the inner thickness was thinner than the outer. This was not possible to
measure during the test as only the widest part of the lateral face will be visible on the
pictures.

5
01 - Experimental
4.5 01 - Simulation
Displacement [mm]

3.5

2.5

2
0 100 200 300 400 500 600 700
Time [s]
Figure 5.20: Thickness of specimen plate with hole 01.

By studying Figure 5.21 it is seen that the diameter in the x – direction first contracts slightly
due to the tensile loading. As the mechanical grips is clamped around the specimen the
width of the upper specimen will not change this makes the specimen wider in the upper
part which pulls the lateral faces out to each side, and the diameter expands. This happens
after approximately 150 s. [10] The two curves are similar in the beginning but differ more
during the analysis. The final diameter is quite alike.

40
The change in the diameter in the y direction is illustrated in Figure 5.22. The experimental
test and the simulated test are both parallel through the test. They start off at the same
value, but differ slightly as the test continues.

36 60

55
Displacement [mm]

Displacement [mm]
34
50

32 45

40
30
01 - Experimental, dx 01 - Experimental, dy
35
01 - Simulation, dx 01 - Simulation, dy
28 30
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Time [s] Time [s]

Figure 5.21: Displacement – time curve of Figure 5.22: Displacement – time curve of the
diameter in the x – direction. diameter in y – direction.

Strain rate
The strains are plotted against time during the simulation as shown in Figure 5.23. As
necking initiates the strains will increase rapidly in the necking area until the specimen
experiences propagation. The strains will be largest where the necking initiates which is
approximately at the middle of the lateral faces. The strains are therefore taken from a cross
section in that area and plotted against simulated time. By differentiating with respect to
time the strain rate is obtained. This is illustrated in Figure 5.24. Although the strain rate was
estimated to be quasi – static the strain rate will vary locally. The strain rate increases as the
necking progresses. When the specimen reaches localized necking the strain rate in the
necking area is close to zero, this is logical as the change in strains is very small. The
experimental values of the strains obtained are very poor because the 7D program is not
able to recognize the black and white pattern due to large strains in the area. There is not
much to compare other than the initiation of the strains in the experimental test and
simulated model is alike. The experimental test values for the other three tests are not
plotted and compared to the simulation as the strains cannot be evaluated when the
necking develops.
-3
1.5 x 10
10
01 - Experimental
8 01 - Simulation
1
True strain

6
Strain rate

0.5 2
01 - Experimental
0
01 - Simulation
0 -2
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Time [s] Time [s]

Figure 5.24: True strain – time curve of test Figure 5.25: Strain rate – time curve of
specimen 01. specimen 01.

41
5.4.2 Results obtained from the plate with a hole of test 02 through test 04
The test results from tests 02, 03 and 04 are represented under the same section as the
procedure is identical to the first test described above.

Test 02, plate with a hole of 20 mm

The plate with hole of diameter 20 mm has been simulated in LS – Dyna to see if there are
any significant differences from the plate with hole of diameter equal to 30 mm. The force –
displacement curves from the experimental results and the simulated results are illustrated
in Figure 5.25, and shows that the simulated curve is stiffer than the experimental curve. The
displacements for both the curves are taken from the same node and at the same distance
from the centre of the hole which is 40 mm. The force peak is slightly higher for the
simulated curve, and as seen earlier it is experiencing a second hardening. This is seen from
the elevation in force at the end of the simulation. The experimental curve stabilizes at a
force of approximately 3000 N. The overall shape of the two curves is satisfactory equal. The
experimental curve is a little shorter than the simulated curve due to invalid pictures in that
area. As it is clear at which value the force stabilizes it was chosen not to add experimental
values measured by the machine.

5000
02 - Experimental
02 - Simulation
4000

3000
Force [N]

2000

1000

0
0 5 10 15 20 25 30 35
Displacement [mm]

Figure 5.25: Force – displacement curve of plate with a hole test 02.

Dimensions of the lateral faces and the diameters of test 02

The lateral faces and the diameter of the hole in both directions are found from the
experimental results and from the simulation in LS – Dyna. First, by comparing the changes
in lateral faces during testing and simulation it is seen that the initial values and the end
values of both curves are close to identical; this can be seen in Figure 5.26 (a) and (b).
Though there are differences during the analysis. The ending value is larger for the simulated
curves in both x and y directions due to error in the Matlab calculations. This can be said as
the test specimen was measured right after the test was performed, and the diameter was
measured to be d = 27.76 mm which is similar to the analysis. Though, these results cannot
be fully trusted as the plate contracts slightly when the test specimen is removed from the

42
test machine, but it does however give an indication of the value. The diameter in the y –
direction was neither measured before nor after testing, and it is therefore difficult to say
which one of the values that is most correct. The diameters are illustrated in Figure 5.26 (c)
and (d). Figure 5.26 (e) shows the experimental and simulated curves of the thickness of the
specimen. The two curves start off at the same point and follow the same shape during the
first 300 s. In the end the displacements differ with about 1 mm which is quite a lot when
the thickness was 5 mm in the first place. Nevertheless, the overall shape of all five of the
curves is similar and comparable.

22 22
Lateral face, w1, Experimental Lateral face, w2, Experimental
20 Lateral face, w1, Simulation 20 Lateral face, w2, Simulation

Displacement [mm]
Displacement [mm]

18 18

16 16

14 14

12 12

10 10
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Time [s] Time [s]

(a) Displacement – time of lateral face w1 (b) Displacement – time of lateral face w2

28 50

26
Displacement [mm]
Displacement [mm]

40
24

22 30

20
20 Diameter Y, experimental
18 Diameter X, experimental Diameter Y, simulation
Diameter X, simulation
16 10
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Time [s] Time [s]

(c) Displacement – time curve of the (d) Displacement – time curve of the
diameter in the x - direction diameter in the y - direction

6
02 - Experimental
02 - Simulation
Displacement [mm]

1
0 100 200 300 400 500 600 700
Time [s]

(e) Displacement – time curve of the thickness

43
Figure 5.26: Comparison between the experimental and the simulated curves of changes in lateral
faces and diameter

Strain rate
Test 02 was applied the same velocity as test 01, but the diameter of the hole is 10 mm
wider. The strains and the strain rate are plotted against time and compared to test 01 to
see if there will be changes as the lateral faces are wider for test 02. By considering Figure
5.27 and Figure 5.28 it is seen that the strains in test 01 is slightly higher which leads to a
difference in strain rate. The reason for this difference is due to more localized necking in
test 01. As the plate from test 02 has wider lateral faces the necking will be less localized
which leads to less strains in the necking area i.e. the strain rate will be lower for test 02
[10].

1.5 0.01
Test 01
0.008 Test 02
1
True strain

True strain 0.006

0.5 0.004

Test 01
0.002
Test 02
0
0 100 200 300 400 500 600 700 0
Time [s] 0 100 200 300 400 500 600 700
Time [s]
Figure 5.27: True strain – time curve of tests
01 and 02. Figure 5.28: Strain rate – time curve of tests
01 and 02.

Plate with hole of 30 mm, test 03

The third test, 03, was applied a velocity of 0.5 mm/s, and had the same diameter as test 01
that is 30 mm. As seen in Figure 5.29 the force – displacement curve of the two curves are
close to identical. The force peak hits exactly the same point as well as the declining part of
the curves. The model is experiencing secondary hardening and does not follow the
stabilizing force, at approximately 2400 N. During the test the camera was placed too close
to the specimen which means that the cross section 40 mm from the centre is not in the
picture. Therefore the test data obtained from the machine is inserted from the dot marked
in Figure 5.29.

44
 4000
03 - Experimental
3500 03 - Simulation

3000

2500
Force [N]

2000

1500

1000

500

0
0 10 20 30 40 50
Displacement [mm]

Figure 1.29: Force – displacement curve of plate with a hole of diameter 30 mm, test 03.

Change in lateral faces, thickness and diameter, test 03

The change in the width of the lateral faces and the diameter is shown in Figure 5.30 (a), (b)
and (c). The overall shape between the experimental and simulated curves is more similar
for this test as the pictures obtained during testing had fewer disturbances. By comparing
the displacements in the thickness direction and in the transverse direction it is seen that
there is a slightly difference, taking the ratio for the transverse displacement gives 0.58 and
the ratio for the thickness direction gives 0.5. This means that the material HDPE is not
perfectly isotropic but close to. This ratio is not calculated for the other tests as there was
too much insecurity in the experimental tests due to background noise in the pictures taken.

16 16
03 - Experimental, w1 03 - Experimental, w2
03 - Simulation, w1 03 - Simulation, w2
Displacement [mm]

Displacement [mm]

14 14

12 12

10 10

8 8
0 20 40 60 80 100 0 20 40 60 80 100
Time [s] Time [s]

(a) (b)

45
 5
03 - Experimental
03 - Simulation
Displacement [mm]

1
0 20 40 60 80 100
Time [s]

(c)

Figure 5.30: (a) Displacement – time curve of lateral face w1, (b) displacement – time curve of
lateral face w2 and (c) displacement – time curve of the thickness.

The diameter in the x – direction is illustrated in Figure 5.31 (a) and shows how it first is
pulled together and then pulled out again due to the wider upper part of the test specimen.
It can also be seen that the displacements are approximately constant during the localized
necking condition at the end of the test. Figure 5.31 (b) illustrated the displacement of the
diameter in the y – direction follows the displacement controlled tensile loading that the
specimen was subjected to.

36 80

70
Displacement [mm]
Displacement [mm]

34
60

32 50

40
30 03 - Experimental, dx 03 - Experimental, dy
30
03 - Simulation, dx 03 - Simulation, dy
20
28 0 20 40 60 80 100
0 20 40 60 80 100 Time [s]
Time [s]

(a) (b)

Figure 5.31: (a) Displacement – time curve of the diameter in the x – direction and (b) displacement
– time curve of the diameter in the y – direction.

Strain rate
As the third test, 03, was applied a velocity of 0.5 mm/s the strain rate was determined to
see the changes of the strains through the test. The diameter of the hole is 30 mm as for test
01. The strains and the strain rate are both plotted against time, and illustrated in Figure
5.32 (a) and (b) where the simulated and the experimental curves are both represented. It is
seen from the curve that the strain rate is 10-2 s-1 in the beginning of the test, but as the
necking progresses the strain rate increases up to 10-1 s-1.

46
 2 0.1
Test - 03 Test - 03

1.5 0.08

Strain rate [1/s]

True strain

0.06
1
0.04
0.5
0.02

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Time [s] Time [s]

(a) (b)

Figure 5.32: (a) True strain – time curve of test 03 and (b) strain rate – time curve of test 03.

Plate with hole of 20 mm, test 04

The last test had a diameter of 20 mm, and was applied a velocity of 0.5 mm/s. The force –
displacement curve illustrated in Figure 5.33 shows that the ascending part of the curve, the
force peak itself and the declining part is very similar. The difference is the stabilizing force
which starts off a lot earlier and experiences a second hardening in the simulated curve. Also
in this case the cross section taken 40 mm from the centre is not captured in the picture
after approximately 60 s of the running time. Therefore experimental data obtained from
the machine is inserted from the dot marked in the figure.

6000
04 - Experimental
04 - Simulation
5000

4000
Force [N]

3000

2000

1000

0
0 5 10 15 20 25 30 35 40 45
Displacement [mm]

Change in lateral faces, thickness and diameter

Further on, the lateral faces and diameters are following the overall shape nicely as seen in
Figure 5.34 (a) through (d).

47
 22 25
04 - Experimental, w1 04 - Experimental, w2
20
Lateral face w2 [mm]
04 - Simulation w1 04 - Simulation, w2

Displacement [mm]
20
18

16 15

14
10
12

10 5
0 20 40 60 80 100 0 20 40 60 80 100
Time [s] Time [s]

(a) Lateral face w1 (b) Lateral face w2

28 70

60

Displacement [mm]
Displacement [mm]

26
50
24
40
22
30
20 04 - Experimental, dx 04 - Experimental, dy
20
04 - Simulation, dx 04 - Simulation, dy
18 10
0 20 40 60 80 100 0 20 40 60 80 100
Time [s] Time [s]

(c) Diameter in x - direction (d) Diameter in y - direction

5
04 - Experimental
4.5 04 - Simulation
Displacement [mm]

3.5

2.5

2
0 20 40 60 80 100
Time [s]
(e) Thickness

Figure 5.34: Displacement – time curves of (a) lateral faces w1, (b) lateral face w2, (c) diameter in x
– direction, (d) diameter in y - direction and (e) thickness.

Strain rate
The true strain – time curve is illustrated in Figure 5.35 and maximum strains are 1.5 while
for test 03 the maximum strains are 1.53. The difference due to a smaller diameter in test
04, although the difference is very small compared to each other. The strain rate – time
curve is illustrated in Figure 5.36.

48
 2 0.1
Test - 04 Test - 04
0.08
1.5

Strain rate [1/s]

True strain

0.06
1
0.04

0.5 0.02

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Time [s] Time [s]

Figure 5.35 True strain – time curve Figure 5.36 Strain rate – time curve

Force – displacement curves of all four tests

Figure 5.37 illustrates the force – displacement curves of all plate with a hole tests
performed, and it is seen that the force – displacement curves varies depending on both the
diameter of the specimen and the velocity applied. A higher velocity and a wider diameter
results in a higher force peak.

5000
01 - Eksperimental
02 - Eksperimental
4000 03 - Eksperimental
04 - Eksperimental
Force [N]

3000

2000

1000

0
0 5 10 15 20 25 30 35 40
Displacement [mm]

Figure 5.37: Force – displacement curves are illustrated for all plate with a hole tests performed.

5.4.4 Young’s modulus

It is seen in the force – displacement curves that the elasticity modulus is too stiff in some
occasions. Therefore a variety of simulations were performed to see which Young’s modulus
that would fit each test the best. The reason for wanting to try this out is that polymers as
HDPE is experiencing viscoelasticity, and the material model in LS – Dyna does not account
for this, the material model implemented in LS – Dyna is elastic – viscoplastic.

Figure 5.38 below shows the first test of plate with a hole applied to different elasticity
modulus’. By studying the curves it is seen that a Young’s modulus of E = 600 MPa is a much
better fit. The force peak is closer as well as the initiation of the curve up to force peak.

49
 3500

3000

2500
Force [N]

2000

1500

01 - Experimental
1000
01 - Simulation, E = 800 MPa
01 - Simulation, E = 600 MPa
500
01 - Simulation, E = 400 MPa

0
0 5 10 15 20 25 30
Displacement [mm]

Figure 5.38: Force – displacement curves of plate with a hole test 01 applied different Young’ s
modulus’s.

All the plates with hole have been applied different Young’s modulus’ to see which one of
them would make the best fit. The Young’s modulus have been seen to differ quite a lot
depending on if the test have been applied a high strain rate or if the test performed is
quasi-static. The quasi-static tests shows that the Young’s modulus is lower for these tests,
as the strain rate increases the Young’s modulus increases as well. In addition the diameter
of the plate is significant. A wider diameter leads to more localization of strains. This leads to
a higher local strain rate and the Young’s modulus will therefore be higher [10]. The Young’s
modulus that results in the best fit for each test is illustrated in Figure 5.39 below.

3500 5000
3000
4000
2500
Force [N]

Force [N]

2000 3000

1500
2000
1000
Experimental 1000 Experimental
500
Simulation, E = 600 MPa Simulation, E = 400 MPa
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30 35
Displacement [mm] Displacement [mm]

(a) Force – displacement curve of plate 01 of (b) Force – displacement of plate 02 of

diameter 30mm and applied a velocity of diameter 20 mm and applied a velocity of
0.05 mm/s. 0.05 mm/s.

50
 4000 5000

3000 4000
Force [N]

Force [N]
3000
2000
2000
1000 Experimental
Simulation, E = 800 MPa 1000 Experimental
Simulation, E = 600 MPa
0
0 10 20 30 40 50 0
Displacement [mm]
0 10 20 30 40 50
Displacement [mm]

(c) Force – displacement of plate 03 of (d) Force – displacement of plate 04 of

diameter 30 mm and applied a velocity of 0.5 diameter 20 mm and applied a velocity of 0.5
mm/s. mm/s.
Figure 5.39: Force – displacement curves of all tests of plate with hole (a through d) applied the
Young’s modulus which makes the best fit for each test.

51
5.5 Drop Tower testing

5.5.1 Introduction
To evaluate the material model it is important that the validation model is subjected to more
complex situations according to stresses and strains. A real component will be stretched,
compressed and undergo different velocities when applied in the industry. In addition it is
important that the boundary conditions and loading condition are well defined to minimize
the uncertainty in the validation process. The friction between the sphere and the plate will
make some uncertainties in the model and will be a source of error in this testing [10].

5.5.2 Experimental program of drop tower testing

There were performed six drop tower tests all of different magnitudes. A plate of about 100
mm in both length and width was placed into the Instron CEAST 9350 machine. The
specimen was lying on top of a hollow cylinder with a 15 mm wide edge. A long shaft made
of metal was lifted up, and let fall. The tip of the shaft was shaped like half a sphere of radius
10 mm. There was applied different amount of weights, velocities and fall height dependent
on how much the material would endure. Two tests led to fracture and that was test DT – 4
and DT – 5. An overview of all tests performed in the drop tower is given in Table 5.3.

Table 5.3: Drop tower testing

Test # Weight Velocity Fall Energy Maximum Maximum

[kg] [m/s] height [J] force [N] displacement [mm]
[m]
DT - 1 5.045 4.4 1 50 5499 16.92
DT - 2 5.045 6.263 2 100 6303 23.89
DT - 3 10.045 4.4 1 100 6276 15.95
DT - 4 10.045 6.263 2 200 6469 24.17
DT - 5 10.045 5.4 1.5 150 6336 20.06
DT - 6 10.045 4.85 1.2 120 6114 18.17

The shaft was dropped, and the fall height was adjusted by a spring that is attached to the
shaft. The fall height differed from between one and two meter. There were added weight
on top of the plate attached to the shaft, and the two first tests had an additional weight of
5.045 kg, while the four last ones had 10.045 additional kilos. Figure 5.40 illustrates the
setup of the plate right after the specimen has been hit by the sphere.

52
Figure 5.40: The plate and the shaft right after impact.

5.5.3 Simulation of drop tower

The drop tower tests were simulated by using axisymmetry where only one thin section was
sketched and simulated both for the plate and the sphere. The right end is fixed, see Figure
5.41 for illustration. Since this end was clamped between two metal grips the model could
have been modeled with the clamps by adding contact definitions. However, this method
was not chosen. Further on, the left side is detained in the x – direction so as to prevent any
displacements other than in the y direction. The sphere is only constrained in the x –
direction on the left side. Both the sphere and the plate was modeled by shell elements, and
is applied a shell thickness of 1/2π. There was applied a mass of respectively 5.045 kg and
10.045 kg that was implemented by calculating the density of half a sphere including a
cylinder of height and radius equal to 10 mm. The density was then implemented in the LS –
Dyna material card. The material used for the sphere was chosen as 001 – Elastic and the
properties for steel were inserted. Figure 5.41 below illustrates the simulated model of the
drop tower test.

The sphere was dropped in the experimental tests, and this was modeled in LS – Dyna by
adding an initial velocity to the sphere.

53
Figure 5.41: The mesh of the simulated model of the drop tower test.

The mesh shown in Figure 5.41 is 0.6 mm in both x and y directions making the elements
quadratic. This applies to both the plate and the sphere. A four node axisymmetric solid
element formulation is chosen for the model. The elements have a symmetry around the y –
axis and are volume weighted [11].

In these models there have not been necessary to use mass scaling as the running time is set
to approximately 16 milliseconds which was the running time of the experiments performed.
The critical time step is then kept small which is important to avoid information to propagate
more than the distance between adjacent nodes during a single time step [12].

The contact definitions are important in this analysis, and the most common type for crash
analyses is 2D automatic single surface. The slave surface is typically defined as a list of part
ID s, and the master surface is not defined. This contact definition includes self – contact in
addition the contact is considered between all the parts in the slave list [15].

Friction is another important parameter during this test, and will also be an important
source of error. It has therefore been performed several simulations with different friction
coefficients ranging from 0.1 - 0.3. The first simulation was run with no friction at all [16].

5.5.4 Results obtained from the drop tower tests

The first test, DT – 1, was applied a kinetic energy of 50 J and a mass of 5.045 kg which
makes the velocity equal to 4400 mm/s. A drop tower test is a lot more complicated, and
there are more factors involved. Figure 5.42 illustrates the force – displacement where the
experimental and the simulated curves are compared. It is seen that the force peak is close
to identical but the simulated curve is stiffer than the experimental curve. The initial part of
the curve is a good approximation, and fits nicely. As viscoelastic parameters are not

54
implemented in LS – Dyna the unloading of the simulated test will not match the
experimental test. However, the first part of the declining is quite parallel to the
experimental curve.

6000
DT - 1, Experimental
5000 DT - 1, Simulation

4000
Force [N]

3000

2000

1000

-1000
0 2 4 6 8 10 12 14 16 18
Displacement [mm]

Figure 5.42: Force – displacement curves of drop tower test 1

When the test specimen is hit by the falling sphere the specimen is pushed downwards into
the cylinder making a plug where the sphere hit. As the specimen is fixed, due to the
mechanical grips clamping around the plate, the plate has no ability to contract and the
plate is forced to stretch. This means that the plate gets slightly longer, and a membrane
effect is obtained. The specimen will in particular experience a lot of strains near the plug.

Drop tower test 2 is applied the same mass, but the energy is increased to 100 J since the
first plate did not experience much deformation. The force – displacement curve of test 2 is
illustrated in Figure 5.43 and shows that the force peak is not very similar in this test. The
beginning of the ascending part of the curve is however similar. The force – displacement
curve does not match quite as good in this test as the first test, DT – 1.

55
 7000
DT-2, Experimental
6000 DT-2, Simulation

5000

4000
Force [N]

3000

2000

1000

-1000
0 5 10 15 20 25 30
Displacement [mm]

Figure 5.43: Force – displacement curves of drop tower test 2.

The tests DT – 4 and DT – 5 both experienced fracture, the specimens where added a mass
of 10.045 kg and kinetic energy of 200 J and 150 J respectively. In test DT – 4 the cap of the
plug was flung open, it was clear that there had been large forces working on the specimen.
It was desired to see how much displacement the material could hold before it fractured,
and therefore another test was performed of lower energy to hope for it to hold. It did not.
The cap of the plug on test DT – 5 was barely torn open which means that the plate could
almost hold the impact. Therefore a sixth test, DT – 6, was performed hoping that the sphere
would not penetrate the plate, but make a big plug. The specimen withheld the impact
forming a huge plug in the test specimen. The color of the material is black, but the plug had
changed to a grayish color. The grayish layer can be removed by scratching on it with a sharp
instance.

In the drop tower test the friction will be of major influence, and it has therefore been
performed several analysis of different coefficients of friction ranging from 0.1 to 0.3. Figure
5.44 illustrates that a friction coefficient of 0.1 fits best to the experimental curve.

56
 7000
Experimental
6000 Simulation
Simulation µ = 0.1
5000 Simulation µ = 0.2
Simulation µ = 0.25
4000 Simulation µ = 0.3
Force [N]

3000

2000

1000

-1000
0 2 4 6 8 10 12 14 16 18
Displacement [mm]

Figure 5.44: Force – displacement curves of drop tower test 1 applied different coefficients of
friction.

When considering the force – displacement curve it is clear that the simulated curves are
stiffer than the experimental ones. Various simulations with different modulus’ of elasticity
were run to see if there would be any significant outcomes, both near the force peak and the
unloading. Figure 5.45 illustrates the force – displacement curves of test DT – 1 applied
different Young’s modulus’. By applying a Young’s modulus of 400 MPa it is seen that the
initiation of the curve makes a poorer fit, but the unloading part is more parallel. The curves
shows that there are two different elasticity modulus’ for the drop tower test; one for the
loading and another for the unloading. The reason for this occurrence is that the specimen is
applied a high velocity leading to large strains and this makes the elasticity modulus higher.
When the projectile stops and reverses the velocity will be lower than the incoming velocity.
This makes the strain rates lower during unloading which leads to a lower elasticity modulus
in the descending part of the curve.

6000
Experimental
5000 Simulation, E = 800 MPa
Simulation, E = 600 MPa
4000 Simulation, E = 400 MPa
Force [N]

3000

2000

1000

-1000
0 2 4 6 8 10 12 14 16 18
Displacement [mm]

Figure 5.46: Force – displacement curves of drop tower tests applied various Young’s modulus’

57
The force – displacement curves of four of the tests performed is illustrated in Figure 5.46 to
compare the shapes of the tests. It is seen that the initiating part of the force – displacement
curve is similar for all the tests. The difference lies in the force – peak and the unloading. The
two first tests, DT – 1 and DT – 2, is applied the same amount of weights, but the velocity is
increased in the second test. Neither of these tests provided more than a dent in the plate.
The fourth and the sixth tests are both applied weights of 10.045 kg, and a velocity of 6263
mm/s and 4850 mm/ s respectively. Test DT – 4 went right through, and the cap of the plug
was flung open. The sixth and last test did not fracture, but it was close. The displacements
were not measured manually after the testing as it was very difficult to get the height of the
cap. For this to be done it would have been necessary to cut the specimen in two pieces. This
way both the displacements and the thickness of the specimen could have been measured.

7000

6000 DT - 1
DT - 2
5000 DT - 4
DT - 6
4000
Force [N]

3000

2000

1000

-1000
-30 -20 -10 0 10 20 30
Displacement [mm]

Figure 5.46: Force – displacement curves of drop tower tests 1, 2, 4 and 6.

5.5.5 Stresses at the tip of the plate

The plate experiences a lot of stresses during the impact between the plate and the sphere.
There are both tensile and compression stresses present during the test. First the plate will
endure compression stresses as the sphere hits the plate, and they will be as large as
approximately 50 MPa. As the plate is bent further down the upper side of the plate
experience tensile stresses up to about 90 MPa. This is quite a lot, but the deformations
close to the projectile are large. Figure 5.47 illustrates the stress – time curve from the
simulation. As there was not used a camera during the drop tower test it is not possible to
compare the stresses and strains to the experimental results.

58
 100
DT - 06
80

60
True stress [MPa]

40

20

-20

-40

-60
0 0.005 0.01 0.015 0.02
Time [s]

Figure 5.47: True stress – time curve of drop tower test 6.

5.6 Discussion and source of error

The experimental tensile tests are experiencing an increase in temperature due to elevated
strain rates in the true stress – strain curves this is illustrated in Figure 3.6 (e). This makes the
material behavior more soft which leads to less steeper curve. LS – Dyna does not consider
temperature changes during an analysis. The material model is calibrated from the tensile
test of low strain rate which has a steeper curve. When the model is applied a higher velocity
the curve is just elevated without any softening effect present.

There are performed a variety of simulations of different elasticity modulus of both tensile
tests and plate with a hole. The reason for performing these analyses is to determine how
important the viscoelastic effect is, since this is not accounted for in the material model in LS
– Dyna. The simulations shows that the viscoelasticity has an influence on how the force –
displacement responds, and that there are some misleading results as this is not considered.
On the other hand, the viscoelasticity effect is not suppose to be that significant for large
deformations, and all tests performed in this thesis are applied to large deformations and
stress – strain relations. Another aspect of not implementing the viscoelastic effect into the
material card in LS – Dyna is that the simulations would have been a lot more complicated.

The plate with a hole has a slightly more complex geometry than the uniaxial tensile test,
and that was also the purpose of validating these specimens. To be able to make an
appropriate validation of the material model it was necessary to use a more complex stress –
strain response than the uniaxial tensile test. The inside of the hole is experiencing large
strains; it was not even possible to get out any strains after 200 s as the mesh was
completely ruined in 7D. This makes it very difficult to compare local variations in the
specimens. Therefore the experimental strains have not been taken into much
consideration.

59
As not all pictures have had a perfect solid – colored background this have affected the
results obtained from Matlab. When the thickness, diameters and the lateral faces was
measured there will be some errors due to background noise. The grayscale was adjusted to
the utmost ability, but there are some inaccuracies present. When the compression tests
was performed a solid – colored background was used which makes these results more
reliable. There is less inaccuracy during these tests.

When performing experiments and simulations of impact problems as the drop tower
testing there is many aspects to take in consideration such as complex stress – strain
responses. During these tests friction was an important factor to considerate, and it is room
for error when simulating these tests. It was seen by Figure 5.x that by applying different
coefficients of friction there was variations in forces. The friction is considered a “dark
horse” in the impact simulations [10].

The coefficient C is determined in the quasi – static tensile test at a strain rate of 10-3 s-1, in
the impact simulation the strain rate is experienced to be roughly 100 s-1 to 1000 s-1, the
yield stress is therefore extrapolated to a satisfactory value as this is unknown. This will be a
source of error during the simulation of the drop tower.

The flow criteria in the model determine how the plastic strains develop in the model. β
affects the change in thickness during the analysis, and was determined in the calibration to
be equal to 1. If this constant is incorrect this may lead to error in the plastic strains as well
as for the necking [10].

60
6. Conclusion
The experimental tests were subjected to nominal strain rates ranging from 10-3 s-1 to 10-1 s-
1
. It was seen that the stress – strain curves subjected to a nominal strain rate of 10-3 s-1
crossed the curves subjected to a nominal strain rate of 10-1 s-1. The reason is that when the
material is subjected to high strain rates the temperature increases which makes the
material softer. This shows that HDPE is strain rate dependent due to elevated temperatures
that leads to changes in the material behavior.

The tensile tests, plate with a hole tests and the drop tower tests were all simulated with
different Young’s modulus’. As the results of the force – displacement curves was more
similar to the experimental results in some of the occasions, it showed that the there are
viscoelasticity that is not taken in consideration during the simulations, and that this affects
the results.

During the material testing of the compression tests the longitudinal and the transverse
strains were plotted in the same figure. This showed that the transverse strains are close to
half the longitudinal strains. This means that HDPE does not experience any significant
change in volume and the material is considered an isochoric material.

A further study of this thesis would be to implement a viscoelastic effect into the material
model in LS – Dyna. In addition there could have been performed a thorough calibration
instead of the partial recalibration performed in this thesis.

61
7. Reference

[1] SIMONA - Product Information PE-HWU/HWST, 1995

[2] M. Polanco-Loria, A.H. Clausen, T. Berstad and O.S. Hopperstad – A constitutive model for
thermoplastics intended for structural applications, 2009

[3] J. Rosler, H. Harders and M. Baker – Mechanical Behaviour of Engineering Materials, 2006

[4] A.H. Clausen, M Polanco-Loria, MT Hovden, M Haugen, OS Hopperstad - “Experimental

and numerical study of HDPE and PVC at different loading conditions", In preparation for
possible journal publication (2012).

[5] A.H. Clausen, M. Polanco-Loria, T. Berstad and O.S. Hopperstad – A constitutive model for
thermoplastics with some applications, 2011

[6] Hovden M.T. – Tests and numerical simulations of polymer components, Master Thesis
2010.

[7] A. H. Clausen – Geometry of tensile specimen.

[8] ] R.F. Moura, A.H. Clausen, E. Fagerholt, M. Alves, M. Langseth – Impact on HDPE and PVC
– Experimental tests and numerical simulations, 2010

[9] O. S. Hopperstad and T. Børvik – Lecture Notes TKT4135 Mechanics of materials, 2012

[10] A. H. Clausen – Personal communication

[11] Livermore Software Technology Corporation (LSTC) – LS – Dyna Keyword manual

Volume1. Version 971, 2007.

[12] K. M. Mathisen – Lecture Notes, Non-Linear Finite element analysis. Lecture 9 – Solution
of the Nonlinear Dynamic Equilibrium Equations.

[13] Y. Ling – Uniaxial True Stress – Strain after Necking, 1996

[14] M. Andersen – Personal communication

[15] LS – Dyna support: http://www.dynasupport.com/tutorial/ls-dyna-users-guide/contact-

modeling-in-ls-dyna.

[16] matbase.com – friction coefficient of HDPE.

62
Appendix A

63
64
Tensile testing
The tensile tests were performed in the laboratory at NTNU in February 2012. The material
Polyethylene (HDPE) was delivered by a German company called SIMONA.

The uniaxial tensile tests are presented with key data and pictures. The test results of the
force – displacement curve, stress – strain curve and longitudinal versus transverse strains
are illustrated for each tensile test performed. The geometri of the tensile specimen is
illustrated below.

The most essential key data is given as:

w0 – initial width, [mm]

t0 – initial thickness, [mm]

- nominal strain rate, [s-1]

v – velocity, [mm/s]

Fmax – Maximum force, [N]

σy – yield stress, [MPa]

Δmax – Maximum displacement, [mm]

ν – contraction ratio

65
Tensile test T1

1500 50
T1 Experimental

True stress [MPa] 40

1000
Force [N]

30

500 20

10
T1 Experimental
0
0 10 20 30 40 50 60 0
Displacement [mm] 0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain

0.8
T1 Experimental
Transverse strain

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

w0 [mm] t0 [mm] [s-1] v [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

12.00 4.91 10-3 0.033 1218 22.7 51.85 0.49

66
Tensile test T2

1500 40
T2 Experimental

True stress [MPa]

30
1000
Force [N]

20
500
10
T2 Experimental
0
0 10 20 30 40 50 60 0
Displacement [mm] 0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain

0.5

0.4
Transverse strain

0.3

0.2

0.1
T2 Experimental

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

12.0 4.90 10-3 0.033 1218 22.9 50.14 0.49

67
Tensile test T3

1500 40
T3 Experimental T3 Experimental

True stress [MPa]

30
1000
Force [N]

20

500
10

0 0
0 10 20 30 40 50 60 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Displacement [mm] True strain

0.4
T3 Experimental
Transverse strain

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.94 4.91 10-2 0.33 1475 27.8 53.85 0.45

Comments: The contraction ratio is seen to be slightly lower for this test.

68
Tensile test T4

1500 40
T4 Experimental
True stress [MPa]
30
1000
Force [N]

20
500
10
T4 Experimental
0 0
0 10 20 30 40 50 60 0 0.2 0.4 0.6 0.8 1
Displacement [mm] True strain

0.5
T4 Experimental
0.4
Transverse strain

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.96 4.90 10-2 0.33 1425 26.6 57.13 0.48

69
Tensile test T5

1500 40
T5 Experimental

True stress [MPa] 30

1000
Force [N]

20

500
10
T5 Experimental

0 0
0 10 20 30 40 50 60 0 0.2 0.4 0.6 0.8
Displacement [mm] True strain

0.4
Transverse strain

0.3

0.2

0.1
T5 Experimental
0
0 0.2 0.4 0.6 0.8
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.96 4.90 10-2 0.33 1409 26.6 57.42 0.48

70
Tensile test T6

50
1500 T6 Experimental
40
True stress [MPa]
Force [N]

1000 30

20
500
10
T6 Experimental
0
0 20 40 60 80 0
Displacement [mm] 0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain

0.5

0.4
Transverse strain

0.3

0.2

0.1 T6 Experimental

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.95 4.89 10-1 3.3 1619 30.2 71.76 0.47

71
Tensile test T7

40
1500 T7 Experimental

True stress [MPa]

30
Force [N]

1000
20

500
10 T7 Experimental

0 0
0 20 40 60 80 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Displacement [mm] True strain

0.4
Transverse strain

0.3

0.2

0.1
T7 Experimental
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.95 4.91 10-1 3.3 1621 29.9 71.77 0.49

72
Tensile test T8

40
T8 Experimental
1500
True stress [MPa] 30
Force [N]

1000
20

500
10
T8 Experimental
0 0
0 20 40 60 80 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Displacement [mm] True strain

0.25

0.2
Transverse strain

0.15

0.1

0.05 T8 Experimental

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.95 4.90 10-1 3.3 1677 30.9 71.52 0.42

73
Tensile test T9

3000 50
T9 Experimental
2500
40
True stress [MPa]

2000
Force [N]

30
1500

1000 20

500 10
T9 Experimental
0
0 10 20 30 40 50 60 70 0
Displacement [mm] 0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain

0.8
T9 Experimental
Transverse strain

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.97 9.76 10-2.5 0.1443 2750 26.2 62.05 0.48

74
Tensile test T10

3000 50
T10 Experimental
2500
40
True stress [MPa]

2000
Force [N]

30
1500
20
1000

500 10 T10 Experimental

0 0
0 10 20 30 40 50 60 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Displacement [mm] True strain

0.8
T10 Experimental
Transverse strain

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

b0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.95 9.76 10-2.5 0.1443 2767 26.5 55.22 0.50

75
Tensile test T11

3000 50

2500 40

True stress [MPa]

2000
Force [N]

30
1500
20
1000
T11 Experimental
500 10
T11 Experimental
0
0 20 40 60 80 0
Displacement [mm] 0 0.2 0.4 0.6 0.8 1 1.2 1.4
True strain

0.8
T11 Experimental
Transverse strain

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

w0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.96 9.76 10-2 0.33 2985 28.0 70.13 0.48

76
Tensile test T12

4000 50
T12 Experimental
40
True stress [MPa]
3000
Force [N]

30
2000
20
1000
10 T12 Experimental

0 0
0 20 40 60 80 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Displacement [mm] True strain

0.5

0.4
Transverse strain

0.3

0.2

0.1 T12 Experimental

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

w0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.97 9.76 10-1 3.3 3307 30.5 71.49 0.47

77
Tensile test T13

40
2500 T13 Experimental

True stress [MPa]

2000 30
Force [N]

1500
20
1000

500 10
T13 Experimental
0
0 10 20 30 40 50 0
Displacement [mm] 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
True strain

0.4
T13 Experimental
Transverse strain

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Longitudinal strain

w0 [mm] t0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

11.96 9.76 10-3 0.033 2552 24.2 43.7 0.49

78
Compression tests
The compression tests were performed in the laboratory at NTNU in February. The
compression specimen was extruded in two different directions. The four first was extruded
in the yz - direction and had a height of 8mm. The fifth test was extruded in the xy - direction
and had a height of 5 mm. The geometry of the test specimens are illustrated below.

The results represented herein are the force – displacement curves, true stress – strain
curves and the transversal strains versus the longitudinal strains. In addition the most
important key data is given as

h0 – initial height, [mm]

d0 – initial diameter, [mm]

- nominal strain rate, [s-1]

v – velocity, [mm/s]

Fmax – maximum force, [N]

σy – yield stress, [MPa]

Δmax – maximum displacement, [mm]

ν – contraction ratio

79
Compression test C1

5000 30

4000 25
True stress [MPa]

20
Force [N]

3000
15
2000
10
1000 C1 Experimental C1 - Experimental
5
0
0 1 2 3 4 0
Displacement [mm] 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
True strain

0.4
C1 Experimental

0.3
Tranverse strain

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Longitudinal strain

h0 [mm] d0 [mm] [s-1] v [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

8.02 9.56 10-3 0.008 4048 25.1 3.93 0.49

80
Compression test C2

4000 30

25
3000
Stress MPa

20
Force [N]

2000 15

10
1000
C2 Experimental
C2 Experimental 5
0 0
0 1 2 3 4
0 0.2 0.4 0.6 0.8 1
Displacement [mm]
Strain

0.4
C2 Experimental

0.3
Tranverse strain

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
Longitudinal strain

h0 [mm] d0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

8.01 9.55 10-3 0.008 3925 24.3 3.84 0.43

81
Compression test C3

5000 40

4000
True stress [MPa]

30
Force [N]

3000
20
2000

1000 C3 Experimental 10
C3 Experimental
0
0 1 2 3 4 5 0
Displacement [mm] 0 0.5 1 1.5
True strain

0.4

0.3
Tranverse strain

0.2

0.1
C3 Experimental

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

h0 [mm] d0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

8.02 9.55 10-2 0.08 4978 28.0 4.39 0.42

82
Compression test C4

6000 40

5000
True stress [MPa] 30
4000
Force [N]

3000 20

2000
C4 Experimental 10 C4 Experimental
1000

0 0
0 1 2 3 4 5 0 0.2 0.4 0.6 0.8 1
Displacement [mm] True strain

0.5
C4 Experimental
0.4
Tranverse strain

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
Longitudinal strain

h0 [mm] d0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

8.02 9.55 10-1 0.8 5167 31.2 4.42 0.48

83
Compression test C5

8000 40

True stress [MPa]

6000 30
Force [N]

4000 20

2000 C5 Experimental 10
C5 Experimental

0 0
0 0.5 1 1.5 2 2.5 3 0 0.2 0.4 0.6 0.8 1 1.2
Displacement [mm] True strain

0.5
C5 Experimental
0.4
Tranverse strain

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Longitudinal strain

h0 [mm] d0 [mm] [s-1] V [mm/s] Fmax [N] σY [MPa] Δmax [mm] ν

4.94 9.99 10-2 0.05 6157 27.0 2.89 0.43

84
Plate with a hole testing
The plate with a hole tests were performed in the beginning of march in the laboratory at
NTNU. The plates were made from the 2000 x 1000 mm sheets that were delivered by the
German company SIMONA.

The results of the change in diameter, thickness and the width of the lateral faces are
illustrated as well as the force – displacement curves for each test performed. In addition
some key data is provided.

Key data:

w1,0 – initial width of lateral face 1 dy,0 – initial diameter in the y – direction

w2,0 – initial width of lateral face 2 v – velocity, [mm/s]

t0 – initial thickness Fmax – maximum load

dx,0 – initial diameter in the x – direction Δmax – maximum displacement

85
Plate with a hole, test 01

w1,0 [mm] w2,0 [mm] t0 [mm] dx,0 [mm] dy,0 [mm] v [mm/s] Fmax [N] Δmax [mm]
15.69 15.29 4.89 30.09 30 0.05 3361 29.06

4000 5
01 Experimental 01 Experimental
4.5
3000
Thickness [mm]
Force [N]

4
2000
3.5

1000 3

0 2.5
0 5 10 15 20 25 30 0 100 200 300 400 500 600
Displacement [mm] Time [s]

36 60
Diameter x - direction [mm]

Diameter y - direction [mm]

01 Experimental
55
34
50

32 45

40
30
35
01 Experimental
28 30
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Time [s] Time [s]

86
 36 14
Lateral face w1 [mm] 01 Experimental 01 Experimental

Lateral face w2 [mm]

12
34
10
32
8
30
6

28 4
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Time [s] Time [s]

Comments:

 Lateral face w2 is incorrect due to poor quality of the pictures taken.

 The diameter in the y – direction was not measured before testing, and therefore
value of the diameter is given as approximately 30 mm in the table.

87
Plate with a hole, test 02

w1,0 [mm] w2,0 [mm] t0 [mm] dx,0 [mm] dy,0 [mm] v [mm/s] Fmax [N] Δmax [mm]
20.21 20.16 4.92 20.06 20 0.05 4321 29.79

5000 6
02 Experimental 02 Experimental
4000 5
Thickness [mm]
Force [N]

3000 4

2000 3

1000 2

0
0 5 10 15 20 25 30 1
Displacement [mm] 0 100 200 300 400 500 600 700
Time [s]

26 50
Diameter x - direction [mm]

Diameter y - direction [mm]

24
40
22
30
20

20
18 02 Experimental
02 Experimental
16 10
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Time [s] Time [s]

88
 22 22
Lateral face w1 [mm] 02 Experimental 02 Experimental

Lateral face w2 [mm]

20 20

18 18

16
16
14
14
12
0 100 200 300 400 500 600 700 12
Time [s] 0 100 200 300 400 500 600 700
Time [s]

Comments: The diameter in the y – direction was not measured before testing and is
therefore given as a an approximately value of 20 mm.

89
Plate with a hole, test 03

w1,0 [mm] w2,0 [mm] t0 [mm] dx,0 [mm] dy,0 [mm] v [mm/s] Fmax [N] Δmax [mm]
15.56 15.12 4.89 29.87 30 mm 0.5 3806 36.42

4000 5
03 Experimental 03 Experimental

4
Thickness [mm]

3000
Force [N]

2000 3

1000 2

0 1
0 10 20 30 40 0 20 40 60 80 100
Time [s]
Displacement [mm]

36 60
Diameter y - direction [mm]
Diameter, x direction [mm]

55
34
50

32 45

40
30
35 01 Experimental
03 Experimental
28 30
0 20 40 60 80 100 0 100 200 300 400 500 600
Time [s] Time [s]

90
 16 16
03 Experimental 03 Experimental

Lateral face w2 [mm]

Lateral face w1 [mm]

14 14

12 12

10 10

8 8
0 20 40 60 80 100 0 20 40 60 80 100
Time [s] Time [s]

Comments: The diameter in the y – direction was not measured before testing and is given as an
approximately value of 30 mm.

91
Plate with a hole, test 04

w1,0 [mm] w2,0 [mm] t0 [mm] dx,0 [mm] dy,0 [mm] v [mm/s] Fmax [N] Δmax [mm]
20.33 20.11 4.92 20.02 20 0.05 4999 36.44

5000 5
04 Experimental 04 Experimental
4000 4.5
Thickness [mm]

4
Force [N]

3000
3.5
2000
3
1000
2.5
0
0 10 20 30 40 2
Displacement [mm] 0 20 40 60 80 100
Time [s]

28 60
Diameter, x direction [mm]

Diameter, y direction [mm]

26 50

24 40

22
30
20
04 Experimental 20
04 Experimental
18
0 20 40 60 80 100 10
Time [s] 0 20 40 60 80 100
Time [s]

92
 22 25
04 Experimental 04 Experimental
20
Lateral face w1 [mm]

Lateral face w2 [mm]

20
18

16 15
14
10
12

10 5
0 20 40 60 80 100 0 20 40 60 80 100
Time [s] Time [s]

Comments: The diameter in the y – direction was not measured before the test and is
therefore given an approximate value of 20 mm.

93
Drop tower testing
The drop tower machine Instron CEAST 9350 had never been used before, and was first tried
out in the end of April when the impact tests were performed. The test specimen was a 100
mm x 100 mm plate of thickness 5 mm made from the extruded sheets delivered by the
manufacturer SIMONA.

The test results that are given are the force – displacement curve of all six of the impacts
tests performed. In addition some essential key data is given as

v – velocity, [mm/s]

W – weight applied during testing, [kg]

t0 – initial thickness

H – fall height of the shaft, [m]

Ep – potential energy, [J]

Fmax – maximum force, [N]

Δmax – maximum displacement, [mm]

94
Impact test, DT -1

v [mm/s] W [kg] t0 [mm] H [m] Ep [J] Fmax [N] Δmax

4400 5.045 4.91 1 50 5499 16.92

6000
DT - 1 Experimental
5000

4000
Force [N]

3000

2000

1000

-1000
0 2 4 6 8 10 12 14 16 18
Displacement [mm]

95
Impact test, DT – 2

v [mm/s] Mass [kg] t0 [mm] Fall Height [m] Ep [J] Fmax [N] Δmax
6263 5.045 4.90 2 100 6303 23.89

7000
DT-02 Experimental
6000

5000

4000
Force [N]

3000

2000

1000

-1000
0 5 10 15 20 25
Displacement [mm]

96
Impact test, DT – 3

v [mm/s] Mass [kg] t0 [mm] Fall Height [m] Ep [J] Fmax [N] Δmax
4400 10.045 4.90 1 100 6276 15.95

7000

6000 DT - 3 Experimental
5000

4000
Force [N]

3000

2000

1000

-1000

-2000
-50 -40 -30 -20 -10 0 10 20
Displacement [mm]

97
Drop tower, test 4

v [mm/s] Mass [kg] t0 [mm] Fall Height [m] Ep [J] Fmax [N] Δmax
6263 10.045 4.90 2 200 6469 24.17

7000
DT - 4 Experimental
6000

5000

4000
Force [N]

3000

2000

1000

-1000
0 5 10 15 20 25
Displacement [mm]

98
Drop tower test 5

v [mm/s] Mass [kg] t0 [mm] Fall Height [m] Ep [J] Fmax [N] Δmax
5400 10.045 4.90 1.5 150 6336 20.06

7000

6000 DT - 5 Experimental

5000

4000
Force [N]

3000

2000

1000

-1000
-20 -15 -10 -5 0 5 10 15 20 25
Displacement [mm]

99
Drop tower test 6

v [mm/s] Mass [kg] t0 [mm] Fall Height [m] Ep [J] Fmax [N] Δmax
4850 10.045 4.90 1.2 120 6114 18.17

7000
DT - 6 Experimental
6000

5000
Force [N]

4000

3000

2000

1000

0
-30 -20 -10 0 10 20
Displacement [mm]

100
Appendix B

101
102
Matlab script

This script finds the average strains in the necking area. The image is
used to see where the largest strains can be obtained. The script is used
for the tensile tests.

pic=elnmaxi085;
x=11;
y=60;
elnmaxi(1,1)=0;

for i = y-20:y+20
n=1;
for j = x-10:x+10
elnmaxi_teller(n)=(pic(i,j));
n=n+1;
end
ave_elnmaxi(i)=mean(elnmaxi_teller);
end

k = max(ave_elnmaxi)

This script is used to evaluate the strains, and gives the average strains
in each picture taken during the testing as long as the mesh in 7D holds.
The strains are given in a vector, and are applied to the tensile tests.

ant=112; % Number of pictures

y=57;
x=11;

elnmaxi(1)=0;
elnmini(1)=0;

for i=1:ant-1
teller=1;
for k=1:10
(sprintf('elnmaxi%.3d(y,%d)',i,k))
elnmaxi_skritt(teller)=(eval([sprintf('elnmaxi%.3d(y,%d)',i,(x-
10/2)+k)]));
elnmini_skritt(teller)=(eval([sprintf('elnmini%.3d(y,%d)',i,(x-
10/2)+k)]));

teller=teller+1;
end
elnmaxi(i+1)=mean(elnmaxi_skritt);
elnmini(i+1)=mean(elnmini_skritt);
end

103
This script gives the height and the diameter of the compression specimens.
The pictures are inverted to black and white pictures, white = 1, and black
= 0. By summing up the values the diameter and the height is obtained.

% compression test
% HDPE-C1

ant_images = 50;
files = dir('HDPE-C1*');

for i = 1:1

image = imread(files(i).name);
BW = im2bw(image,0.4); % invertes the pictures in a
grayscale of 0.4
V = 1-BW;
imshow(BW);

row_first_nonzero = find(BW(:,1),1,'first'); % Gives the location of where

% the first value is.

row_last_nonzero = find(BW(:,1),1,'last'); % Gives the location of the

% where the last value is.

row_diameter(i) = round(row_first_nonzero + (row_last_nonzero -

row_first_nonzero)/2); % Finner raden midt i teststykket

col_first_nonzero = find(V(row_diameter(i),:),1,'first');

col_last_nonzero = find(V(row_diameter(i),:),1,'last');

height_pixels(i) = row_last_nonzero - row_first_nonzero;

diameter_pixels(i) = col_last_nonzero - col_first_nonzero;

end

height_pixels = transpose(height_pixels);
diameter_pixels = transpose(diameter_pixels);

The diameter and the height are obtained in pixels.

104
This script gives the displacement of the first row in each matrix, and put
them into a new matrix containing the first row of displacements for each
picture taken during the plate with a hole test.

ant = 93;
dy(1)=0;
y = 1;
x = 1;

for i = 1:ant-1
p = 1;
for k = 1:113
row(i,p) = eval(sprintf('dy%.3d(y,%d)',i,k)); % i = ant bilder,
k = x verdi,
p = p+1;
end
matrix(i,:) = row(i,:);
end

%All_first_rows = matrix();

This script gives the average strains in lateral face w2. Though the
deformations were too large, and the strains were not possible to contract,
this script was only used for test 1.

%% Plate with diameter 30 mm, test 01.

pic = dy112;

x = 26;
y = 50;
elnmaxi(1,1)=0;

for i = x-4:x+3
n = 1;
for j = y-6:y+5
elnmaxi_teller(n)=pic(j,i);
n = n+1;
end
ave_elnmaxi(i)= mean(elnmaxi_teller);
end
k = max(ave_elnmaxi)

105
This script is used to get the strains from the plate with a hole. Though
this script was only used for test 1 as the strains were not possible to
extract due to large deformations.

ant = 212;
y = 49;
x = 28;

elnmaxi(1) = 0;
elnmini(1) = 0;

for i = 1:ant
n = 1;
for k = 1:10
elnmaxi_skritt(n) = eval(sprintf('elnmaxi%.3d((y-10/2)+k,%d)',i,k));
elnmini_skritt(n) = eval(sprintf('elnmini%.3d((y-10/2)+k,%d)',i,x));
n = n+1;
end
elnmaxi(i+1) = mean(elnmaxi_skritt);
elnmini(i+1) = mean(elnmini_skritt);
end

longitudinal_strain = elnmaxi';
transverse_strain = elnmini';
This script gives diameter in x and y directions and both the lateral
faces.

% HDPE-01. Test av plate med stort hull.

clear all
clc

ant_images = 300;
files = dir('HDPE-01*');

for i = 1:ant_images

image = imread(files(i).name);
BW = im2bw(image,0.2);
V = 1-BW;
% imshow(V)

col_first_nonzero(i) = find(BW(463,:),1,'first');
col_last_nonzero(i) = find(BW(463,:),1,'last');
Plate_center = round(col_first_nonzero + (col_last_nonzero -
col_first_nonzero)/2);

length_diameter_Y(i) = sum(V(463:end,1248));

row_last_nonzero(i) = find(V(:,1248),1,'last');
row_first_nonzero(i) = row_last_nonzero(i) - length_diameter_Y(i);
mid_point(i) = round(row_first_nonzero(i) + (row_last_nonzero(i) -
row_first_nonzero(i))/2);
length_diameter_X(i) = sum(V(mid_point(i),830:1650));

length_side1(i) = sum(V(mid_point(i),1:830));
length_side2(i) = sum(V(mid_point(i),1650:2448));

106
 Half_width_image(i) = length(V)/2;
Length_b2(i) = Half_width_image(i) - length_side1(i) -
(length_diameter_X(i)/2);
Length_b1(i) = Half_width_image(i) - length_side2(i) -
(length_diameter_X(i)/2);
end

Dx = transpose(length_diameter_X);
Dy = transpose(length_diameter_Y);
b1 = transpose(Length_b1);
b2 = transpose(Length_b2);

This script gives the thickness of the plate with a hole tests. The images
is converted into black and white pictures, and the contrast gives the
thickness of the specimens.

clear all
clc

ant_images = 300;
files = dir('HDPE-01*')

for i = 1:ant_images
image = imread(files(i).name);
BW = im2bw(image,0.2);
V = 1-BW;

col_first_nonzero(i) = find(V(1025,:),1,'first');
col_last_nonzero(i) = find(V(1025,:),1,'last');

thickness(i) = sum(V(1025,1134:1315));

end

t = transpose(thickness);

107