You are on page 1of 22

Internship

Polymer Engineering Group

2nd May 2022 - 15th July 2022

By: R.Frees
Supervisor: Luca Andena

Milano, July 15, 2022


Internship Robbe Frees

Abstract
During the Internship which took place at Politecnico Milano in the Polymer Engineering group from
the 2nd of May until 15 July a research on environmental stress crack (ESC) on high impact polystyrene
(HIPS) was executed with the use of 4-point bending. Due to some paperwork issues the start of this
period should be spend on different projects. Therefore the start of the Internship was dedicated to a
different subject: toughness of 3D printed VeroWhite and VeroGray. In this first part the toughness of
these materials was tested by performing scratch tests. With the use of these tests a visualization of the
toughness with respect to aging and printing direction could be given, this was executed with the use
of a MATLAB-script. After this visualization the effect of the applied load was researched, the same
MATLAB-script could be used to analyze this results. Besides that difference between the two sides of
the specimens were examined on toughness.
After the paperwork was finished the ESC on HIPS could be examined. For this experiment the effect of
oil on HIPS was tested by performing 4-point bending on notched samples with a constant displacement
rate. The first part of the project consist on evaluating the material parameters where-after the crack
initiation and propagation phase for both in air and active environment where analyzed. With the use of
load separation and the J-integral these phenomena could be researched numerically. In the end there was
found that the influence of the active environment is present under the assumption of a low displacement
rate ( 1 mm/min).

TU/e i
Internship Robbe Frees

Contents
1 DMA-testing 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Correcting for Inertia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Correction for strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Scratch tests 4
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5 Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3 ESC 4-point bending on HIPS 8


3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.3 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

A Appendix 19
A.1 Crack initiation and propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

TU/e ii
Internship Robbe Frees

1 DMA-testing
1.1 Introduction
During the first stage of my internship I helped Samuelle Zalaffi, a Research Assistant, with DMA-testing.
Since most of them were already performed there was not a lot of testing left. However, Samuelle found
out that the samples were not perfectly rectangular. This could lead to a different result of the modulus
and the strain due to the error in inertia. The goal of this part is therefore to research the effect of a
change in inertia on the results of the modulus and the strain and perform this correction on the data.

1.2 Correcting for Inertia


When conducting a 3-Point-Bending measurement the DMA machine assumes that the tested sample
has a rectangular cross-section. However, if the sample shape is significantly different, an inaccurate
computation of the moment of inertia will be the result. This inaccurate inertia will be used to compute
the modulus with the use of the following equation:

F L3
Emeasured = (1)
48δIrect
where F is the force applied, L is the length of the sample, δ is the measured displacement and Irect is
the moment of inertia of a rectangle. To correct for the inertia a factor depending on the actual inertia
of the sample should be used. Since Equation 1 gives an inverse linear relationship between the modulus
and the inertia, the factor is the ratio between the inertia of a rectangle and the actual inertia, as given
below:
F L3 Irect Irect
Eactual = ∗ = Emeasured ∗ (2)
48δIrect Iactual Iactual
where Iactual is the real inertia of the tested sample.

In the case studied in this internship the actual cross-section of the sample is a trapezoid with a top
width of a and a bottom width of b and a height of h, as is visualised in Figure 1.

Figure 1: Trapezoid with top width a, bottom width b and height h

The inertia of the trapezoid is therefore dependent on these measures a,b and h and can be written as
[5]:
h3 (a2 + 4ab + b2 )
Itrap = . (3)
36(a + b)
bh3
With the use of this equation and the knowledge that the inertia of a square is equal to 12 , the correction
factor can be rewritten as:
Irect a2 + 4ab + b2
= . (4)
Itrap 3b(a + b)
To check this equation a and b could be set equal and this would give a ratio of 1, which is the expected
result.

TU/e 1
Internship Robbe Frees

During the start of this project 4 measurements were performed on different trapezoidal specimen with
the same length but width different cross-sections. Therefore, different correction factors has to be taken
into account. The results can be seen below:

Table 1: Modulus correction factors

Specimen A(mm) B(mm) Factor


1 12.2033 12.6767 1.0191
2 12.1833 12.6833 1.0202
3 12.1533 12.6900 1.0217
4 12.1700 12.6767 1.0205

From this result can be found that the correction factor is quite low and will increase the modulus with
only 2%.

Results
Now that the factor has been computed the results of the DMA measurement can be corrected. The
result is shown in Figure 2. The figure shows a small difference between the corrected and uncorrected
data as was expected.

102
Corrected
Uncorrected
E [GPa]

101

100

0 50 100 150 200 250


T[°C]

Figure 2: Corrected modulus (red) and Uncorrected modulus (blue) versus temperature

1.3 Correction for strain


Besides the inertia the maximum strain on the sample is also dependent on the cross-section of the sample.
Also in this computation the machine assumes a perfect rectangular shape and uses the following equation:
σ0
ϵ0 = (5)
E

TU/e 2
Internship Robbe Frees

where the correction factor for the modulus E has already been derived. The stress on the sample is
however also dependent on the cross-section because the neutral line changes per shape. The standard
equation for the stress is:
My
σ= (6)
I
where M is the bending moment, which is equal to F4L for three point bending, y is the farthest distance
from the neutral axis, which is dependent on the cross-section and I is the moment of inertia. When
plugging in Equation 1 and Equation 6 into Equation 5 the effect of inertia cancels out and gives the
following equation:
12δy
ϵ0 = (7)
L2
where only the y changes for different cross-section shapes. The DMA-machine uses y = h2 since it
assumes a rectangular shape while this could differ in real life, therefore a correction factor dependent y
should be introduced:
12δyrec yreal yreal
ϵ0,real = 2
∗ = ϵ0,rec ∗ (8)
L yrec yrec
where yreal is the actual distance from neutral line to the farthest point.

In the case of a trapezoid the neutral line is located at:


h 2a + b
yneutral = ∗ . (9)
3 a+b
Assuming that a < b, the distance to the top part is h − yneutral and this leads to the following correction
factor:
2a+b
ytrap h(1 − 3(a+b) ) 2 a + 2b
= h
= ∗ (10)
yrec 2
3 a+b
When applying this equation to the 4 tested samples the following correction factors can be found:

Table 2: Strain correction factors

Specimen A(mm) B(mm) Factor


1 12.2033 12.6767 1.0063
2 12.1833 12.6833 1.0067
3 12.1533 12.6900 1.0072
4 12.1700 12.6767 1.0068

This factor indicates that for this case the correction factors are insignificant.

1.4 Conclusion
For the modulus the correction factor is dependent on the inertia:
Irect
Eactual = Emeasured ∗ . (11)
Iactual

For the case of a trapezoid this results in an increase of only 2%.

For the strain the correction factor is dependent on the distance between neutral line and the farthest
point:
yreal
ϵ0,real = ϵ0,rec ∗ (12)
yrec
For the case of a trapezoid this results in an increase of 0.60%.

TU/e 3
Internship Robbe Frees

2 Scratch tests
2.1 Introduction
During this part of my internship 3D printed samples will be tested on scratch and will be performed
together with Luca Andena. The goal of the tests is to visualize the toughness of the 3D samples for
different configurations.

2.2 Experimental
The scratch tests will be performed on 8 different samples with the V227 indenter with a tip radius of
200 microns, a constant sliding speed of 1 mm/min and a constant load of 10,5,2 or 1 N. The samples can
be distinguished by three parameters: the date of production, the material and the printing direction.
The first parameter, the date of production, could be an indication how the samples age over time. In
this experiment there are two batches, where the first batch is produced on 14 April 2022 and the second
batch is printed on 21 and 22 April. For the material there are two different options: VeroWhite (VW)
versus VeroGrey (VG). The last parameter is the direction of printing. This parameter is divided in three
directions x,y and z. The producer (PhD student Tim Volders from L’Université de Liège) defined the
directions with the use of the following figure:

Figure 3: Schematic figure of the three printing directions: x,y and z.

With the use of these three parameters the samples can be defined as: Batch 2y VW/VG, Batch 2 z
VW/VG, Batch 2 x VW/VG and Batch 1 x VW/VG. Besides that the load applied during the scratch
test will be mentioned in the name.

TU/e 4
Internship Robbe Frees

2.3 Results
First all the samples were scratched 3 times with a constant load of 10 N and with a 1mm distance
between scratches. This leads to the following measurements:

Figure 4: Penetration and residual depth for all the measurements

One can observe that the results of ’Batch 1 X VG 10 N’ are not very consistent. This was probably
because the sample was not clamped properly, therefore more measurements where performed of this
sample to be able to delete the false measurements. In this way every sample has at least 3 proper
results, when only using these results this gives the following figures:

Figure 5: Scratch data used for averaging the results

With the use of these measurements the average penetration and residual depths of every measurement

TU/e 5
Internship Robbe Frees

can be computed so the measurements can be compared. This gives the following results:

Figure 6: Bar plots of penetration and residual depth (P d and Rd) for the white and grey samples at 10
N

From this figure can be concluded that there is no difference between the aged samples of Batch 1 and
the new samples of Batch 2 with respect to there material properties. However there is a clear distinc-
tion between the depths when comparing the direction of printing. Also a small difference in toughness
between VeroWhite and VeroGrey can be observed.

After analysing this data the same procedure is repeated for 1,2 and 5 Newton. With the use of this data
the following results are obtained:

Batch 2 X Batch 2 Y Batch 2 VW Batch 2 VG


150 150 150 150
VG Side 1 VG Side 1 X Side 1 X Side 1
VG Side 2 VG Side 2 X Side 2 X Side 2
VW Side 1 VW Side 1 Y Side 1 Y Side 1
VW Side 2 VW Side 2 Y Side 2 Y Side 2

100 100 100 100


Depth [ m]

Depth [ m]

Depth [ m]

Depth [ m]

50 50 50 50

0 0 0 0
0 5 10 0 5 10 0 5 10 0 5 10
Load [N] Load [N] Load [N] Load [N]

(a) Depth versus load divided in X and Y direction for (b) Depth versus load divided in VeroWhite and Vero-
both sides of the samples. Grey for both sides of the samples.

Figure 7

In this figure a distinction between the two sides of the sample is made. From these results can be
concluded that the side of the sample only has influence when printing the sample with VG. Furthermore

TU/e 6
Internship Robbe Frees

the relationship between load and penetration depth is not dependable on the printing direction or
material of the sample.

2.4 Conclusion
In this part an overview of the most important conclusions and insights are listed:
• Aging has a minor influence regarding the penetration depth of the samples.
• Printing the sample in the z-direction gives the highest toughness, while printing in x or y-direction
only has a small difference.
• VeroWhite has a slightly higher toughness then VeroGrey
• When analyzing the sides of the samples there is only a difference in toughness in VeroGrey and
not in VeroWhite
• The relatonship between load and penetration depth does not depend on the material or the x or
y-direction.

2.5 Recommendation
In this part a few recommendation of future researches are discussed. First, in this set of tests the load
of the x and y-direction are varied however when printing the sample in the z-direction the toughness of
the sample can vary in a whole new way. With this test also the influence of the side of the toughness of
both sides of the material should be tested. Secondly, in this experiment the sliding speed has been held
constant but it is not known how this parameter influences the toughness of the material. Thirdly, the
effect of aging is tested by only using two different batches with only a difference in time of two weeks.
To get a better insight in the effect of aging on this material more tests has to be executed over a longer
period of time to be able to make some proper conclusion and to find a numerical relation. Lastly, to test
the fracture toughness of this 3D-printed material different kind of test should be performed to be able
to analyze the mechanical properties of this material, think about tensile tests and 3/4-point-bending.

TU/e 7
Internship Robbe Frees

3 ESC 4-point bending on HIPS


3.1 Introduction
In this part of the Internship the influence of Environmental Stress Cracking (ESC) on high-density
polystyrene (HIPS) will be tested. To test HIPS mechanically a 4-point bending will be introduced in an
environment of oil and air to compare the results and investigate the influence of the oil on the mechanical
properties of the material. The theory of load-separation will be used to have a numerical values which
can be compared.

3.2 Experimental
To test the HIPS on fracture with 4-point bending Single Edged Nodge Bending (SENB) geometry will
be used. With the use of the molded sheets ESC-resistance samples were manufactured. The dimensions
are shown in Figure 8 and are in accordance with ISO 13856. To create consistent samples they are
thermally treated before notching them. The heat treatment consist of one hour of 90 degrees hereafter
one cools the samples of slowly to room temperature. A majority of the samples are produced with
a sharp notch and the others are prepared with a blunt notch. This will show the difference in crack
initiation and propagation. The size of the notches are 0.5 a/W for the majority of the samples but to
analyze the impact of crack length a few blunt notches have a crack size ranging from 0.3 to 0.7 a/W .
The production of the samples started by cutting square bars from the plates by using a band saw. The
desired dimensions were realised by using a milling machine with a helical blade with a cutting depth of
0.5 mm. The sharp notch was introduced via automated ”chisel-wise” cutting. This realises a notch root
radius lower then 10 µm. The blunt notches were realised by using a diamond-studded circular blade
on the milling machine, this introduced a radius of 1 mm. This same blade has been used to introduce
grooves in all the samples.

Figure 8: SENB specimen (measures in mm)

For the analysis of the mechanical properties of the HIPS material there are two types of tests: shape
factor tests and crack initiation and propagation. For the shape factor test 5 blunt specimen with a notch
size of 0.3, 0.4, 0.5, 0.6 and 0.7 a/W are tested in air and with a constant displacement of 1 mm/min.
The test to analyze the crack initiation and propagation a constant notch size of 0.5 a/W will be used.
The displacement rate will be set on 1, 0.1 and 0.001 mm/min and for every speed 4 sharp (2 in air and
2 in oil) and 1 blunt (in air) notched samples are necessary to create valid results.

TU/e 8
Internship Robbe Frees

3.3 Data analysis


In this part of the report the data analysis and its underlying theory will be discussed. The 4-point
bending test will result in load versus displacement data which will be used to analyse this material and
testing method. This analysis will be divided in two parts: the shape factor and the crack initiation
and propagation. In the first part the elastic and plastic shape factor will be computed with the use of
load separation. These parameters are then necessary to evaluate the crack initiation and propagation in
phase two with the use of the J-integral approach [7]. In phase two the start of the crack propagation will
be evaluated by computing the energy added to system and the energy needed to propagate the crack.
Due to the fact that extensive plastic deformation will take place in the 4-point bending of the HIPS
specimen there is chosen to make use of the J-Integral approach to evaluate the fracture data. This
approach makes use of the following equation:
Uel Upl
J = Jel + Jpl = ηel + ηpl (13)
B(W − a) B(W − a)
where Jel is the elastic contribution and Jpl the plastic contribution to the J-integral. These two contri-
butions depend on the thickness of the specimen (B), the ligament length (W − a), the shape factors (ηel
or ηpl ) and the elastic and plastic components of the mechanical input energy (Uel and Upl ). In the first
part the two shape factors of the material will be computed and analyzed while the J will be computed
and analyzed in the second part.

Shape factor
The first step is to compute the compliance of every specimen (C0 ), which is defined as the inverse value
of slope of the first linear part of the load vs displacement curve, see Figure 9. After the compliance is
calculated the raw curve at the beginning will be replaced for the linear line with the slope of the inverse
of the compliance. In the end the curve will be shifted to assure the load vs displacement curve starts at
(0,0) as shown in Figure 9.

Figure 9: Determination of the compliance (C0 ) and shift of the loading curve to zero.[2]

With the use of the compliance, the displacement (u) can be divided in elastic and plastic displacement
(uel and upl ) by the use of the following equations:
uel = P ∗ C0 (14)
upl = u − uel = u − P ∗ C0 (15)
where P is the load in Newton. Since the J-integral and the energy release rate are equivalent for a
linear elastic material [3] the elastic shape factor can be computed with the use of the energy release rate
a
calibration (ϕ( W )) [1]:
a
Uel (W − a) Uel 1− W Uel Uel
Jel = G = = ∗ = ∗ = ηel ∗ (16)
BW ϕ Wϕ B(W − a) ϕ B(W − a) B(W − a)
TU/e 9
Internship Robbe Frees

in the case of 4-point bending the energy release rate calibration is presented by an analytical function
in Contino’s dissertation [4]. With the use of this expression the elastic shape factor will be computed,
with the following equation:
a
1− W
ηel = (17)
ϕ

To evaluate the plastic shape factor experimental data is necessary to perform load separation proposed
in [7]. This theory mentions that for a given material, geometry and loading condition the load can
be split if the load can be formulated as the product of a crack geometry function F and of a material
deformation function H, as shown below:
a upl
P = F( ) ∗ H( ). (18)
W W
This relation can be used to compare two samples, from the same material with the same geometry and
under the same loading conditions, only the initial crack length can differ, with the following separation
parameter (Si,j ):
ai u ai
P (ai ) F(W ) ∗ H( Wpl ) F(W )
Si,j = = a u = a (19)
P (aj )
upl F ( j ) ∗ H( pl )
W W
F( j )
W
upl upl

With the use of multiple blunt notched specimen the separation values can be computed for one specific
upl . Following this theory the Si,j vs WW−a relation should result in a power law where the slope is equal
to the plastic shape factor ηpl [3]. Therefore, multiple blunt notched specimen with different crack size
will be tested on 4-point bending to be able to create a proper fit.

Crack initiation and propagation


With the use of the computed shape factors in the previous section J-integral can be computed, to be
more precise the Jlim . This is the moment of the crack initiation and can be compared to the video
recording of the tests. To compute this value the separation parameter of a blunt and sharp notched
sample (Ssb ) with equal crack size should be evaluated.
The separation parameter versus plastic displacement consist of three different zones: the first region
shows an unstable Si,j at the lower plastic displacements, which is called the ”unseparable” region, in
the second region the separation parameter shows a nearly constant value therefore called the ”plateau”
region, and the last region shows a decrease in the separation parameter, this is were the region of fracture
propagation.
To find the moment of crack initiation the value of the ”plateau” region (Ssb,pl ) is important to give
the plastic displacement limit (upl,lim ). There are three different kind of separation parameter curves
possible which are shown in Figure 10. These different options ask for different ways to analyze these
points, to do so the Protocol of S.Agnelli and F.Baldi [2] has been used. This protocol uses the following
definitions:
• Type 1: The “plateau” region, where Ssb holds an almost constant value, is easily recognizable.
– Ssb,pl : identify a plastic displacement interval where Ssb maintains a reasonably constant value
(“plateau” region); Ssb,pl is defined as the arithmetic mean value of Ssb data over such interval.
– upl,lim : is the plastic displacement beyond the “plateau” region, at which the separation
parameter reaches the value of Ssb,lim = 0.9975 ∗ Ssb,pl .
• Type 2: The Ssb curve shows a broad peak (at Ssb,max ), after the “unseparable” region.
– upl,lim : is the plastic displacement at which the curve, after the maximum point, reaches a
level of Ssb,lim = 0.9975 ∗ Ssb,max .
– Ssb,pl : is the average between Ssb,max and Ssb,lim , i.e. Ssb,pl = 0.9975 ∗ Ssb,max .
• Type 3: The curve does not show any evident change of the slope after the “unseparable” region:
a plateau region cannot be properly identified.

TU/e 10
Internship Robbe Frees

Figure 10: Types of separation parameter curves. Plateau levels (broken lines) and limit points (circles)
are indicated. [2]

With the use of this protocol the plastic deformation limit can be defined for type 1 and 2. This value can
be used to compute the displacement limits for the elastic and total displacement by using Equation 14
and 15.
When all the displacements of crack initiation are known the value of the J-integral at this point can
be computed, by Equation 13. In this equation Uel and Upl are the area under the curves until uel,lim
and upl,lim respectively. Lastly, the time of crack initiation (tlim ) can be obtained by using the total
displacement at crack initiation. These values can be compared to the recordings obtained during testing
to validate the protocol.
Now that the moment of crack initiation is evaluated the crack propagation will be analyzed. For this
part the third region of the separation parameter will be of importance. To be able to compare the
crack propagation the separation parameter should be normalized as: Ri = Si,j /Si,j,pl . With the use of
these normalized curves the opposite of the slope is said to be Ms . This value provides a classification of
fracture propagation processes in function of the amount of crack growth that takes place in the plastic
region. This can be computed by applying a linear fit at the end of each curve.

TU/e 11
Internship Robbe Frees

3.4 Results
In this section the theory discussed in subsection 3.3 will be executed on the measurements of the HIPS
samples under 4-point bending with a constant displacement. This chapter is therefore divided in the
same two parts: the shape factor and the crack initiation and propagation.

Shape factor
For this part five blunt specimen with a different crack size (a/W ): 0.3, 0.4, 0.5, 0.6 and 0.7 were tested
with a constant displacement rate of 1 mm/min. This results in a load versus displacement graph shown
in Figure 11a.
With this raw data the compliance of the measurement can be computed. Since the measurement is
not exactly linear at the start of the measurement the compliance is computed at the first linear part
and not at the start. The compliance is computed by creating a MATLAB-function which searches for
the highest slope by interpolating between a manually chosen point and the other points. The results
of C0 are presented in Table 3 and are used to create a linear start of the plot. Then, as explained
in subsection 3.3, the graph should be shifted to zero and the data after the maximum load will be
discarded, since the part after fracture is not interesting for this Internship. After these alterations the
load displacement curve results in Figure 11b.

400 400
0.3 0.3
0.4 0.4
350 350
0.5 0.5
0.6 0.6
300 0.7 300 0.7

250 250
P (N)

P (N)

200 200

150 150

100 100

50 50

0 0
0 0.5 1 1.5 2 2.5 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
u (mm) u (mm)

(a) Raw data of load versus displacement for vary- (b) Corrected data of load versus displacement for
a a
ing W . varying W .

Figure 11

Table 3: Compliance

(a/W)
0.3 0.4 0.5 0.6 0.7
[-]
C 0
0.00149 0.00202 0.00273 0.00463 0.00768
[mm/N]

The next step is to compute ηel with the use of Equation 17 and the definition of ϕ found by M.Contino
[4]. This leads to the following figure:

TU/e 12
Internship Robbe Frees

0.7 1.8

el 1.6
0.6

1.4
0.5
1.2

0.4 1

el
0.3 0.8

0.6
0.2
0.4

0.1
0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
a/W

Figure 12: Energy calibration factor ϕ and elastic shape factor ηel for HIPS specimens.

Now that the elastic shape factor is defined, the plastic shape factor can be computed. To do so the
plastic deforation has to be computed with the use of the compliance and Equation 15. In this way
Figure 11b can be changed to a load versus plastic displacement to give Figure 13.

400
0.3
350 0.4
0.5
300 0.6
0.7
250

200
P(N)

150

100

50

-50

-100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
u pl

Figure 13: Load versus plastic displacement

Next, the separation parameter can be computed where every sample will be used as reference line to
give a clear overview of the result. These results are visualized in the following figure:

TU/e 13
Internship Robbe Frees

Ref = 0.3 Ref = 0.4 Ref = 0.5


2 2.5
1
2
0.8 1.5
1.5
S ij

S ij

S ij
0.6 1
1
0.4
0.5
0.5
0.2
0 0
0.2 0.4 0 0.5 0 0.5
upl upl upl
Ref = 0.6 Ref = 0.7
4 8 0.3
0.4
3 6 0.5
0.6
0.7
S ij

S ij

2 4

1 2

0 0
0 0.5 0 0.5 1
upl upl

Figure 14: Separation parameter versus plastic displacement for different references.

When plotting the values of Si,j at a plastic displacement of 0.1 mm versus the crack size with respect
−ai
to width of the specimen ( WW ) on a log-log scale the ηpl can be computed performing a linear fit. The
slope of this linear fit is equal to the plastic shape factor. The result at a plastic deformation of 0.1 mm
has been visualized below:

pl
= 2.2479
0.8

0.6

0.4

0.2
log(Sij)

-0.2

-0.4

-0.6

-0.8

-1
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
log((W-a)/W)

W −ai
Figure 15: Sij versus W on a log-log scale for upl = 0.1, with linear fit.

There can be observed that changing upl will change the values of Sij and would result in a different fit
with a different slope. Therefore the plastic shape factor is plotted versus the plastic displacement in
TU/e 14
Internship Robbe Frees

Figure 16. From this figure can be concluded that a low plastic displacement would lead to a drastically
decrease of ηpl and the same happens when increasing upl higher then 0.25 mm, however still not bigger
then 0.03 lower. Therefore the average ηpl between upl = 0.1 and upl = 0.25 will be used as the final
plastic shape factor: ηpl = 2.2410 ± 1.1993e − 05.

2.255

2.25

2.245

2.24

pl
2.235

2.23

2.225

2.22

2.215
0.1 0.15 0.2 0.25 0.3
u pl

Figure 16: Plastic shape factor versus the plastic displacement.

Crack initiation and propagation


Now that the shape factors of the HIPS material are determined the crack propagation and initiation can
be computed. This will be executed for 3 different displacement rates of: 1, 0,1 and 0,01 mm/min and
in air and an active environment of oil. To start the same procedure as explained in section 3.4 will be
used to get the separation parameters between the sharp and blunt specimen. Before these results are
shown a very important result should be explained first.
When testing the specimen with slow displacement rates (0.1 and 0.01) there is a very clear instability in
the raw data, as shown in Figure 17a. In this figure can be seen that there are frequent drops in the force
for all the specimen tested in air at at displacement rate of 0.1 mm/min. The cause of these drops are
probably due to an unstable crack growth or a different material characteristic like ”stick-slip” [6]. Since
the amount of time for this internship is limited a deeper research in this phenomenon was not possible.
These drops however make it impossible to apply the same steps to compute the compliance and more
important to apply load separation. To resolve for this problem a polynomial fit of the 7th power will
be used to smooth out the curves. This however could however result in loss of important information
and therefore loss of important conclusions. The result of the fitted and directly corrected data is shown
in Figure 21b. This same procedure has been repeated for 0.01 mm/min but this is not necessary for 1
mm/min, these results are shown in subsection A.1.

180 160
Sharp Air
160 Sharp Air
140
Sharp Air
Sharp Air
140
Sharp Oil 120
Sharp Oil
120 Air Blunt
100
100
P (N)
F (N)

80
80
60
60 Sharp Air
Sharp Air
40 Sharp Air
40
Sharp Air
20 Sharp Oil
20 Sharp Oil
Air Blunt
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 0.2 0.4 0.6 0.8 1 1.2
u (mm) u (mm)

(a) Raw data of load versus displacement at 0.1 (b) Fitted and corrected data of load versus dis-
mm/min. placement at 0.1 mm/min.

Figure 17

TU/e 15
Internship Robbe Frees

With the use of these results the normalized separation parameters of the sharp specimen with respect
to the blunt specimen (Rsb ) can be computed. The result of this is visualized below:

0.01 mm/min 0.1 mm/min


1.05 1.05
Sharp Air Sharp Air
Sharp Air Sharp Air
1 Sharp Oil Sharp Air
1
Sharp Oil Sharp Air
Sharp Oil
0.95 Sharp Oil
0.95

0.9
R sb (-)

R sb (-)
0.9
0.85

0.85
0.8

0.8
0.75

0.7 0.75
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3
u pl (mm) u pl (mm)

(a) Separation parameter versus plastic displace- (b) Separation parameter versus plastic displace-
ment at 0.01 mm/min. ment at 0.1 mm/min.
1 mm/min
1.02
Sharp Air
1 Sharp Air
Sharp Oil
0.98 Sharp Oil

0.96

0.94
R sb (-)

0.92

0.9

0.88

0.86

0.84

0.82
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
u pl (mm)

(c) Separation parameter versus plastic displace-


ment at 1 mm/min.

Figure 18

These results show that the majority of curves shows a type 2 kind of curve. There is only one exception
for one of the Sharp Oil sample at 0.01 mm/min, therefore this result will not be taken into account in
the further research.
With these results the Jlim at the time of crack initiation (tlim ) can be computed by using Equation 13
and the results of the shape factors in section 3.4, which is 1.5302 for ηel at a crack size of 0.5 a/W
and 2.2410 for ηpl , this results in Figure 19a for all the data. The second parameter is the Ms which is
computed by fitting a linear line through the last part of the curves and obtaining the slope of this fit.
This results in Figure 19b.

TU/e 16
Internship Robbe Frees

0.3 0.3
Air Air
0.2 Oil 0.2 Oil

log(Jlim [J/m2 ]) 0.1 0.1

0 0

log(Ms [J/m2 ])
-0.1 -0.1

-0.2 -0.2

-0.3 -0.3

-0.4 -0.4

-0.5 -0.5
1.5 2 2.5 3 3.5 1.5 2 2.5 3 3.5
log(tlim [s]) log(tlim [s])

(a) Log-log plot of Jlim versus tlim for air (blue) (b) Log-log plot of Ms versus tlim for air (blue) and
and oil (red) with a linear fit (black). oil (red) with a linear fit (black).

Figure 19

These results give a result as would be expected. For the crack initiation (Jlim ) there is a clear difference
in the slope between the samples in air and in oil, especially for higher tlim . For the crack propagation a
similar kind of result can be observed but in this plot the slope for oil is higher then the slope for air.
With this result can be concluded that the effect of oil is only present when applying a low displacement
rate (< 1). The amount of energy to initiate crack is lower in this case while the crack propagation
increases when the samples are in an active environment.

3.5 Conclusion
In this part the conclusion of the ESC 4-point bending on HIPS will be worked out. With the use of load
separation and het J-integral the shape factor of HIPS under 4-point bending are gathered. To come up
with these values the compliance has to be computed for every sample so a MATLAB-script has been
designed to be able to easily compute this value in the same way for every sample. This results in an
elastic shape factor for a crack length of 0.5 a/W equal to 1.5302 and a plastic shape factor equal to
2.2410. Furthermore there is concluded that the plastic shape factor is independent of the crack size but
does depend on the plastic displacement where the shape parameter is compared. A visualization of this
result has been given and from this plot a proper mean value can be given.
With this results the J-integral for all samples can be obtained to be able to analyze the crack initiation
and the crack propagation. This results are used to compare the results between the samples tested in
an active environment (oil) and air. In the end there can be concluded that the influence of environment
is the most active when applying low displacement rates. In this case the applied energy to initiate crack
is lower and the crack propagation is easier in oil with respect to air.
A very important observation that has been made while testing the samples under low displacement
rates is the fact that an instability in the force versus displacement curve is present. The cause of this
instability is more then probably a material characteristic but since this Internship was getting to an end
the exact cause of this problem could not be analyzed. To solve this problem for this research a 7th-order
polynomial fit has been applied to smooth the curves.

TU/e 17
Internship Robbe Frees

References
[1] ISO 13586: “Plastic - Determination of the fracture toughness (GIC and KIC) - Linear elastic fracture
mechanics (LEFM) approach”. In: (2003).
[2] S. Agnelli and F. Baldi. “A testing protocol for the construction of the “load separation parameter
curve” for plastics”. In: (2015).
[3] Marco Contino. “Engineering Fracture Mechanics”. In: (2020). doi: https://doi.org/10.1016/j.
engfracmech.2020.107422.
[4] Marco Contino. “Long term fracture behaviour of HDPE for household detergent containers: a study
on environmental stress cracking”. In: (2020), pp. 181–183.
[5] eFunda. Centroid,Area, Moments of Inertia, Polar Moments of Inertia, Radius of Gyration of a
Isosceles Trapezoid. url: https://www.efunda.com/math/areas/IsosTrapezoid.cfm. (accessed:
11.05.2022).
[6] Robert J. Young Salim Yamini. “Crack propagation in and fractography of epoxy resins ”. In: Journal
of Material Science 14 (1979), pp. 1609–1618.
[7] Landes JD. Sharobeam MH. “The load separation criterion and methodology in ductile fracture
mechanics”. In: Int J Fract 47 (1991), pp. 81–104. issn: 02712091. doi: https://doi.org/10.
1007/BF00032571.

TU/e 18
Internship Robbe Frees

A Appendix
A.1 Crack initiation and propagation
In this part the extensive results will be shown.

250 250
Sharp Air Sharp Air
Sharp Air Sharp Air
Blunt Air Blunt Air
200 Sharp Oil 200 Sharp Oil
Sharp Oil Sharp Oil

150 150
P (N)

P (N)
100 100

50 50

0 0
0 0.5 1 1.5 2 2.5 3 0 0.2 0.4 0.6 0.8 1 1.2
u (mm) u (mm)

(a) Raw data of load versus displacement at 1 (b) Fitted and corrected data of load versus dis-
mm/min. placement at 1 mm/min.

Figure 20

200 200
Sharp Air Sharp Air
180 Sharp Air 180 Sharp Air
Sharp Oil Sharp Oil
160 Sharp Oil 160 Sharp Oil
Blunt Air Blunt Air
140 140

120 120
P (N)

P (N)

100 100

80 80

60 60

40 40

20 20

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
u (mm) u (mm)

(a) Raw data of load versus displacement at 0.01 (b) Fitted and corrected data of load versus dis-
mm/min. placement at 0.01 mm/min.

Figure 21

TU/e 19

You might also like