CHAPTER 3.

NUMERICAL ANALYSIS
3.1 Scope of Numerical Work

This thesis work includes numerical analysis to simulate the crack growth retardation due
to periodic overloads induced residual stress field at the crack tip and to calculate the
stress intensity factor for the plane strain and plane stress cases at different applied
loading conditions for 2-D & 3-D modeling. The first part of the numerical analysis
provides a significant insight to understand the mechanism of overload induced residual
stress field in spite of its limitation. In further work, cyclic strain hardening, node release
method, fracture criteria and non-linear material behavior should be considered in the
future work.

In this work, the crack tip displacement is recorded as the crack growth length. Instead of
the real crack growth modeling, elastic strain and plastic strain are recorded at the crack
tip. In second part of the numerical work, modeling crack region and calculating the
fracture parameter (stress intensity factor: KI) are focused on. The most important region
in modeling the fracture region is the region around the crack. The singularity at the crack
tip should be picked up when modeling the crack. The PREP7 KSCON command, which
assigns element division sizes around a key-point, is used in a 2-D model to create 2-D
crack tip elements with nodal singularity. This command removes the nodal singularity at
the crack tip.

3.2 Crack Growth Retardation due to Overload Induced Residual Stress Field

In this study, a finite element analysis using ANSYS 5.7 was conducted to simulate the
crack-growth retardation due to the single-peak overload under cyclic loadings. The
objective of simulation is to predict the crack-growth retardation due to the influence of
overload in Aluminum Alloy with a center pre-crack specimen. The compressive residual
stress at crack-tip after the overload is the major factor causing retardation (Figure 3.1).

Residual stresses are produced when one region of a part experiences permanent plastic
deformation while other regions of the same part either remain elastic or are plastically
deformed to a different degree. The overload introduces a large plastic zone in which the
material experiences permanent deformation. Upon unloading, the surrounding elastic
material attempts to resume its original size (the plastic zone is permanently deformed)
and by doing so, exerts compressive stresses on the plastically deformed material at the
crack tip. When the crack has grown through the region (compressive plastic zone) of
residual stresses after a further period of cyclic loadings, the crack growth resumes at the
propagation rate expected under constant amplitude cyclic loadings. It means that if we
can introduce compressive residual stress by the controlled yielding (overload), the
fatigue life of a component will be significantly increased (Figure 3.2). This enhancement
of fatigue life depends on the peak and magnitude of the overload ratio.

36
 Y

Tensile Stress
A fte r O v e rlo a d
O v e rlo a d P la s tic
Zone

X
(+ )
0
(--)
D is ta n c e fro m c ra c k tip
Compressive Stress

Figure 3.1 Residual compressive stresses ahead of crack tip

Crack Length(a)

Constant amplitude
cycles Overload

Periodic
overload
retardation

Number of cycles(N)
The slower crack growth continues until the crack grows beyond the
overload plastic region(The Wheeler model of crack growth inside an
overload plastic zone). This beneficial residual stress effect of overload
is called crack growth retardation.

Figure 3.2 Fatigue life enhancement due to overloads

37
During the crack propagation, the crack growth is affected by environmental conditions,
material properties, and loading conditions including the magnitude and stress ratio of
overload. Careful attention must be given to a series of critical decisions about element
type, mesh method (Mapped mesh, Free mesh), element length size, the selection of
material behaviors, model design (geometry of the specimen, symmetric boundary
condition or loading condition) and crack tip modeling (singular point) if the analysis is
to be reliable. A model for the elastic-plastic finite element simulation in plane stress is
presented by running a nonlinear analysis with ANSYS 5.7. The bilinear inelastic
isotropic hardening (an elastic-plastic model) is considered as element material behavior.
Large deformation effects were also considered during the nonlinear analysis. The crack
growth simulation was based on the stress-strain curve of the node point near crack-tip.
The displacement near crack-tip, the stress-strain curve and stress redistribution along the
crack plane after overload were investigated during cyclic loadings. The specific results
that are being aimed for are the effects of overload induced residual compressive stresses
on the crack growth near the crack tip.

3.3 Finite Element Modeling

In this study, 8 node PLANE82 was used as an element type with the plane stress option.
Nonlinear analysis was performed with an elastic-plastic material model under the action
of cyclic loadings when a single high peak overload was introduced. A bilinear inelastic
isotropic hardening curve (an elastic-plastic model) was considered as material behaviors.
The tangent modulus is taken as 0 GPa. Therefore, material behaves like an elastic-
perfect plastic model. The mechanical properties of specimen (Aluminum) are shown in
Table 3.1. The fracture toughness of Al-2024-T3 is 34MN/m3/2.

Table 3.1 Mechanical properties of Aluminum Alloy

Elastic modulus Poisson Yield Temperature Tangent modulus

(E) Ratio (ν) Stress (T) (H)

325 20 0
70 [GPa] 0.33
[MPa] [°c] [GPa]

The region around the crack tip that is under investigation needs to be finely meshed. In
addition, the results are compared between the mapped and free mesh (with KSCON
command which assigns element division sizes around a singular key point (crack tip)). It
automatically generates singular elements around the specified key point.

To advance the crack, the crack tip advance scheme involving node release immediately
after maximum load on each cycle is the preferred technique. The crack advancement at
maximum loading was found to yield a closer representation of the actual crack
phenomenon. The node release technique can also be carried out by using a series of
nonlinear compression-only and tension-only springs along the crack plane or element

38
death option provided by ANSYS code. However, in this work, the node near the crack
tip was employed to determine the crack growth or displacement. The displacement and
stress-strain curve of the node near the crack tip are recorded.

The mesh refinement study was also conducted to determine the size of element length
(Le) required along the crack plane. The parameter selected for rating mesh size was the
ratio of the element length (Le) to the plastic zone size (rp). The equation of the plastic
zone radius given by Irwin’s expression assuming plane stress condition is given in
equation (2.3).

The value of KI (Stress Intensity Factor) was chosen as 35 [MN/m3/2]. To avoid the
discontinuity between the crack stress and strain, element size less than Le/rp =0.1 are
required to appropriately capture the crack plasticity effects. The element length size
could be obtained as 0.5mm. Symmetry boundary conditions were adopted (Figure 3.3).
The boundary conditions used for the specimen were as follows.

(1) All nodes along the crack plane were constrained from the crack tip to the right
bottom edge along the crack plane.
(2) All nodes were constrained along the left plane.

σ
Symmetric
Boundary
Condition
4
0

m
m
Element Size
around crack
tip : 0.5mm

Element type :
2
0

8 node PLANE 82 m
m

Crack tip 10mm

40mm

Figure 3.3 Geometry and dimension of finite element modeling

39
The external load within a load step was applied in increments over a certain number of
sub-steps. The size of the load increment is controlled by defining these subsequent sub-
steps. For the cycling loading steps, 5 sub-steps were specified, with a maximum of 25
and a minimum of 3 sub-steps. For the overload step, 10 sub-steps were specified, with a
maximum of 50, and a minimum of 5 sub-steps. Since, the specimen exhibits a severe
nonlinear behavior during the overload, the load increment was increased up to 1/10th of
the size of the overload. In order to investigate stress relaxation after overload, constant
cyclic loadings after the peak high overload have also been applied (Figure 3.4).

σ a pp 70M Pa

30M Pa

1 3 5 7 9 11 13
6
2 4 8 10 12 14
T IM E (t)

Figure 3.4 Applied cyclic loading conditions with a high peak overload.
(The number of load step is shown on the load line.)

In order to investigate the effect of residual stresses in fatigue crack growth, it is

necessary to know (1) the magnitude and shape of the residual stress field, and (2) the
combined effect of the residual stress and applied stresses on the fatigue crack growth
rate. Therefore, the displacement of the node near crack tip was monitored over the entire
solution. Furthermore, the reaction forces at the same node were monitored. The crack tip
stress-strain curve at the same node was also plotted. The strain was obtained by adding
elastic and plastic strain. Combined residual stresses and applied stresses along the crack
plane were shown by using the path operation command. The results showed the
influence of overload induced compressive residual stresses on the crack plane. These
compressive residual stresses will decrease the crack growth rate. Overloads ratio, strain
hardening, node release technique, creep and various loading conditions should be
conducted in future work.

40
3.4 Results and Discussions

Strain range
after overload
D
B

A
Strain range before
overload

Figure 3.5 Stress-Strain Curve after one overload (80MPa)

In Figure 3.5, strain range before overload between A and B moves back to negative
strain rage C-D after overload 80 MPa. Cyclic strain range is considerably reduced after
one overload of 80 MPa. Therefore, the crack growth rate will decrease accordingly.
Strain range becomes negative after overloading due to the residual compressive stress.
This result might be considered crack growth retardation. This decreased strain will affect
a change of displacement near crack tip. This graph shows the effect of overloading near
crack tip displacement. The combined effect of the residual stress and applied stresses on
the crack plane before overload and after overload was shown in order to know the effect
of residual stresses in fatigue crack growth. The initial objective of this study is to
investigate the influence of overload induced residual compressive stress field on fatigue
crack growth. Accordingly, residual compressive stress is induced due to the overload
when the model is stressed over yield point 200 MPa, but the stress-strain curve turns out
to be unrealistic. The reason was a bilinear inelastic isotropic hardening (BISO) curve (an
elastic-perfect-plastic model) was considered as a material behavior. Actually, the BISO

41
(material model behavior) cannot be used in order to simulate accurate rate-dependent
plasticity as well as cyclic hardening or softening. Therefore, careful attention should be
given to selections like element type, material model behavior. Further work needs to be
carried out in the future. In order to simulate the actual crack growth, mesh at the crack-
tip need to be re-meshed whenever the crack advances along the crack plane.

79MPa

100MPa

200MPa

Figure 3.6 von-Mises Stress during overloading (after load step 5)

In Figure 3.6, the yield has occurred at the crack tip due to the stress concentration when
overload of 70 MPa was applied. Therefore, the plastic deformation occurred at the crack
tip. This plastic deformation at the crack tip was squeezed after overload and then,
compressive residual stress field has been induced at the crack tip.

42
 σ app

80MPa

5 15
30MPa 16
6

1 3 7 9 11 13 17 19 21 23
2 4 8 10 12 14 18 20 22 24
TIME(t)

Figure 3.7 Applied cyclic loading conditions with two overloads condition
(80MPa-80MPa)

To induce proper residual stresses, the peak and magnitude of the overload ratio are
important, and therefore various loading conditions need to be conducted. Two overloads
condition are carried out as shown in Figure 3.7.

The plot of total strain curve (Figure 3.8) didn’t show any significant decreased in strain
reading after second overloading for 80MPa-60MPa. After the second overload, the
second reduced strain was recorded only in 80MPa-100MPa. In case of 80MPa-80MPa
(Figure 3.7), the strain range after the second overload was plotted at the same strain
range level following the first overload. This is because the material used in this study did
not remember the load history. The effect of first overload was completely eliminated
during the second overload.

Plasticity is a non-conservative, path-dependent, time-dependent and temperature-

dependent phenomenon. Two overloads effects on the crack plane and crack growth rate
were not well documented in this study, because the material did not remember the load
history. Von-Mises stress redistribution along the crack plane after first overload (80MPa)
was reduced as shown in Figure 3.9.

43
0.015

0.01

0.005

-0.00 5

-0 .01

-0.01 5
0 5 10 15 20 25
Sub step tim
e
Real material doesn’t behave like the elastic perfect-plastic model. Furthermore, in this
study, the strain hardening was not considered. For more accurate modeling, appropriate
material behavior should be considered, such as, cyclic strain hardening or softening and
path or time dependent plasticity, which might remember the load history. Large
deflection and large strain are associated with nonlinear analysis. Various loading
conditions like overload ratio, stress ratio, and stress range should be conducted in the
future. To advance the crack, the crack tip advance scheme (involving node release
immediately after maximum load on each cycle) needs to be carried out. Further work is
recommended.

The results of these studies vary depending on the following

the type of nonlinear material behavior being defined

the selection of appropriate element types
the type of mesh (Mapped or Free)

3.5 Calculation of Stress-Intensity Factor in Center-Crack Specimen

Cracks and flaws occur in many structures such as bridges, aircrafts, and offshore
structures, sometimes leading to disastrous failures. The engineering field of fracture
mechanics was established to develop a basic understanding of such crack problems and
the state of stress around the crack region. Fracture mechanics deals with the study of
how a crack initiates and propagates through the material under applied loads. It involves
comparing analytical predictions and simulations with experimental results. Typically, the
crack length increases with application of cyclic load.

In order to estimate the exact crack growth rate analytically, fracture parameters such as
stress intensity factor in the crack region should be known exactly. Crack growth rate
(da/dN vs. ∆K) is based on the concept of the stress intensity factor that defines the near
crack-tip driving forces for crack growth and thus, able to characterize crack growth for
different geometries and loads.

Calculation of fracture parameters such as the stress intensity factor is important to

fundamental studies of fracture mechanics, crack growth rate and fatigue mechanism.
The objective of this study is to calculate the stress intensity factor (KI) at the crack tip in
the center-crack plate (2-D and 3-D) by Ansys 5.7, in the center crack plate specimen of
Aluminum Alloy. This fracture parameter will be used as basic information to obtain the
crack growth rate in the structure.

Some typical fracture parameters are:

Stress intensity factors (KI, KΠ, Kш) associated with the three basic modes of
fracture.
J-integral, which may be defined as a path-independent line integral that
measures the strength of the singular stresses and strains near a crack tip.

45
 Energy release rate (G), which represents the amount of work associated with a
crack opening or closure.

3.6 Research Methodology

Modeling the crack region provides a sound basis to predict some typical parameters.
Solving a fracture mechanics problem in ANSYS involves performing a linear elastic or
elastic-plastic static analysis and then using specialized post-processing commands or
macros to calculate desired fracture parameters. In this study, two main aspects have been
concentrated on.

1. Modeling the crack region

2.Calculation of stress intensity factor KI

The most important region in a fracture model is the region around the edge of the crack.
To pick up the singularity around the crack, the crack faces should be coincident, and the
elements around the crack tip (or crack front) should be quadratic, with mid-side nodes
placed at the quarter points. Such elements are called singular elements. The
recommended element type for a two-dimensional fracture model is 2-D plane strain
elements (PLANE82), the 6-node triangular solid, and for three-dimensional model is
SOLID95, the 20-node brick element. A one-quarter model is used because of symmetry.
A long plate with a center crack is subjected to an end tensile stress σ as shown in Figure
3.11. The objective of this study is to determine the fracture mechanics stress intensity
factor KI in 2-D and 3-D model. Geometric properties are shown in Table 3.2 with
loading conditions. The dimension is also shown in Figure 3.10.

Table 3.2 Geometric properties of specimen

Geometric Properties
Material Properties Loading
(Dimension as shown in Fig. 3.10)

E = 70 GPa a = 5 mm σ = 100 MPa

ν = 0.3 b = 25 mm

KI = 35 MN/m3/2 h = 25 mm

Thickness (t) = 5 mm (3-D)

(Fig. 3.12)

46
The simple 2-D model using PLANE82 is created by automatic (free) mesh generation.
The PREP7 KSCON command, which assigns element division sizes around a key-point,
is particularly useful in a 2-D model to create 2-D crack tip elements with nodal
singularity. This command removes the nodal singularity at the crack tip. POST1 is used
to get fracture mechanics stress intensity factor (KI) by displacement extrapolation
(KCALC command). The dimension of 2-D modeling is shown below with nodes. In
fracture mechanics, the KI value represents the resistance to failure. This stress intensity
factor that is given in equation (1.1), is associated with the geometry of modeling, the
crack size and the applied stress.

In middle-center crack specimen, the geometry factor “Y” is

Y = sec( π a / 2 b ) = 1 . 118 ASTM Standard E 647 (ASTM, 95) (3.1)

The applied stress (σ) is 100 MPa, the crack size (a) is 5 mm and the half width (b) is 25
mm. Hence, the calculated stress intensity factor at a center crack specimen (finite width
plate) is 14 MN/m3/2.

h =
25 m m

C rac k tip
a = 5 mm
b = W /2 = 2 5 m m

Figure 3.10 Dimension of finite element modeling

47
 σ

Symmetric boundary

Symmetric boundary
Representative finite element model with boundary
conditions (2-D model with Plane82)

Figure 3.11 Element modeling with boundary conditions

Thickness (t)

Representative finite element modeling

(3-D model with Solid95)

Figure 3.12 Three-dimensional element modeling with applied stress

48
 The element modeling is shown with symmetric boundary conditions in Figure 3.11. The
stress is applied at the top of the specimen. Generating a 3-D fracture model is
considerably more involved than a 2-D model. The KSCON command cannot be used.
The 3-D solid (SOLID95 elements) is used to simulate the 3-D fracture model. In the 3-D
analysis, the plane strain condition is achieved by constraining UZ degrees of freedom of
all the nodes (displacements in the Z-direction). The 3-D element modeling is shown in
Figure 3.12.

3.7 Results

(1) 2-D modeling

v o n -M is e s s tre s s d is tr ib u tio n n e a r c ra c k tip

Figure 3.13 von-Mises stress distribution near the crack tip in 2-D modeling and
KI = 11.96 MN/m3/2

Actually, the crack tip elements to designate singularity should be created by macro files
to remove the singularity at the crack tip in 3-D modeling. But, in this study, singularity
point at the crack tip was not removed. At the crack tip, the elements are disjointed. The
reason is that the compatibility couldn’t be maintained at the crack tip. Because the crack
tip elements are not assigned to remove the singularity at the crack tip in 3-D modeling.
This result shows the importance of the singularity at the crack tip element.

49
(2) 3-D modeling

v o n -M ise s s tre ss d istrib u tio n n e a r c ra c k tip in 3 -D m o d e l

Figure 3.14 von-Mises stress distribution at the crack tip in 3-D modeling and
KI = 12.93 MN/m3/2

(3) The results of KI for the plane strain and plane stress cases at different applied loading
conditions for 2-D & 3-D modeling are shown in Table 3.3 ~ Table 3.8. The column
chart in Figure 3.15 shows all results. These data provides significant insights when
calculating the fatigue life of the test material.

Table 3.3 Stress intensity factor (K) for the plane strain case, (σ = 100MPa)

% variation
KI (2-D model) KI (3-D model)
from 2-D to
MN/m3/2 MN/m3/2
3-D
By Displacement Extrapolation
11.96 12.93 8.11%
(KCALC command)

By the calculation of fracture

14 14
mechanics

50
 Table 3.4 Stress intensity factor (K) for the plane strain case, (σ = 200MPa)

% variation
KI (2-D model) KI (3-D model)
from 2-D to
MN/m3/2 MN/m3/2
3-D
By Displacement Extrapolation
23.92 25.86 8.11%
(KCALC command)

By the calculation of fracture

28 28
mechanics

Table 3.5 Stress intensity factor (K) for the plane strain case, (σ = 250MPa)

% variation
KI (2-D model) KI (3-D model)
from 2-D to
MN/m3/2 MN/m3/2
3-D
By Displacement Extrapolation
29.9 32.32 8.09%
(KCALC command)

By the calculation of fracture

35 35
mechanics

Table 3.6 Stress intensity factor (K) for the plane stress case, (σ = 100MPa)

% variation
KI (2-D model) KI (3-D model)
from 2-D to
MN/m3/2 MN/m3/2
3-D
By Displacement Extrapolation
12.84 13.36 4.05%
(KCALC command)

By the calculation of fracture

14 14
mechanics

Table 3.7 Stress intensity factor (K) for the plane stress case, (σ = 200MPa)

% variation
KI (2-D model) KI (3-D model)
from 2-D to
MN/m3/2 MN/m3/2
3-D
By Displacement Extrapolation
25.67 26.71 4.05%
(KCALC command)

By the calculation of fracture

28 28
mechanics

51
 Table 3.8 Stress intensity factor (K) for the plane stress case, (σ = 250MPa)

% variation
KI (2-D model) KI (3-D model)
from 2-D to
MN/m3/2 MN/m3/2
3-D
By Displacement Extrapolation
32.1 33.4 4.04%
(KCALC command)

By the calculation of fracture

35 35
mechanics

35
30
25
2-D KI (Plane stress)
20 3-D KI (Plane stress)
15 Exact KI
2-D KI (Plane strain)
10 3-D KI (Plane strain)

5
0
100MPa 200MPa 250MPa

Figure 3.15 Stress intensity factor KI in 2-D and 3-D for plane stress and strain
conditions

3.8 Verifying Corrosion Fatigue Testing

The test for task #1 was conducted by using 1 specimen for each condition. There is a
large range of region test results in this thesis work due to experimental constraints. To
find out the fatigue life of test specimens theoretically, Walker (Walker, 1970) or Paris
(Paris, 1964) equations are used in this thesis work. Finally, the theoretical results were
compared with the experimental results of task #1. From the dimension of test specimen
geometry, half width is b=24.5mm, the thickness t is 4.1 mm and an initial crack length ai
is 5 mm.

P 25000 N at R = 0.2 (3.2)

S = = = 124 MPa
2 bt 2 × 0 . 0245 × 0 . 0041

52
The geometry factor Y is varying during the crack growth depending on the crack length.
This value of Y in equation (3.1) can be obtained in ASTM Standard E 647. This
approximation is accurate within 2% for (α = a / b) < 0.9. To find out the fatigue crack
growth life by Paris or Walker equation, both the initial crack length ai and final crack
length af should be known. The final crack length af can be obtained from the trial and
error method. Since, Y varies as crack grows, a iterative solution is needed to obtain af.
The fracture toughness K goes close to KIC = 34 MN/m3/2 if the trial crack length is final
crack length af. The procedure is shown in Table 3.9 for the case of R (stress ratio = 0.2).
The fracture toughness K is close to KIC = 34 MN/m3/2 when the trial crack length (a) is
14.4 mm. Therefore, the final crack length af is assumed as 14.4 mm for R = 0.2 from the
above analysis. Also, the real final crack length from experiments is 15 mm. For the case
of R = 0.5, the same procedure can be used to find out the final crack length. In this case,
the only difference is the gross stress S.

P 40000 N at R = 0.5
S = = = 200 MPa
2 bt 2 × 0 . 0245 × 0 . 0041

Table 3.9 Final crack length

Trial crack length a πα

“a” [mm]
α= Y = sec K = YS π a
b 2
5 0.20408 1.02648 15.9526
10 0.40816 1.11704 24.5508
11 0.44897 1.14598 26.4161
12 0.48979 1.17986 28.40649
13 0.53061 1.21959 30.562
14 0.57142 1.2664 32.933
14.4 0.58775 1.28752 33.95
14.5 0.59183 1.2930 34.219

In the case of stress ratio 0.5, at trial crack length a = 8 mm, the fracture toughness K =
33.967 is approximately close to KIC = 34 MN/m3/2. Therefore, we can assume that the
final crack length af is 8 mm for R = 0.5. But, the real final crack length from this thesis
work is about 10-11 mm. The result from this method is not the same in case of the stress
ratio 0.5.

Consider the situation where the crack growth rate is given by the Paris’s equation and
where F = F (a / b) is constant, or can be approximated as constant, over the range of
crack length ai to af. The value of C used can include the effect of the Stress ratio R, as
from the Walker approach (Walker, 1970) that is given in equation (2.6). The γ, C1, m1
are constants for the Al-2024-T3. These constants are determined from Dowling, 1993.
The values are given below, γ = 0.68, C1 = 1.42*10-8, m1 = m = 3.59.

The m1 in Walker’s equation (Walker, 1970) is equal to m in Paris’ equation (Paris, 1964).
Assume that Smax and Smin are constant, so that ∆S and R are also both constant.

53
 Therefore, C, Y, ∆S, R, and m are all constant. The only variable is “a”, and integration
can be expressed by the equation (2.7).

Equation (2.7) is for the crack propagation life that doesn’t include the crack initiation
life. Actually, the fatigue life results from this thesis work include both the crack
initiation life and crack growth life until failure. Therefore, the results from experimental
work need to be divided into two parts (fatigue crack initiation life and fatigue crack
growth life). But, the transition from crack initiation to crack propagation is not clear. In
this thesis work, the total fatigue life is used instead of fatigue crack growth life. As crack
accelerates during their growth, most cycles are exhausted while the crack growth is non-
exist, and few cycles are spent while the crack is growing. Experimentally, the fatigue
crack growth part is observed as around 30% of the total fatigue life, and the remaining
70% is the fatigue crack initiation part of the total fatigue life. The result which is
calculated from the closed form of equation is Nf = 3000 cycles. This fatigue crack
growth result Nf = 3000 cycles is around 23% of the total fatigue life cycles 13466 cycles
for the test case R=0.2, f=0.5.

In the case of R=0.5, the only difference is Smax = 240 MPa. The final crack length is
taken as 8 mm, while the geometry factor Y is taken as 1.3. Thus, Nf from the same
procedure is 2474 cycles. The fatigue crack growth result Nf =2474 cycles is 29 % of the
total fatigue life cycle, 8550 cycles for R= 0.5, f=0.5.

3.9 Discussions of Numerical Work

The value of the stress intensity factor (KI) from this study shows the valid results, if
these results are compared with the exact value of KI from fracture mechanics. The KI in
case of plane stress is close to the exact value of KI from the fracture mechanics (%
variation between result by KCALC command and result by fracture mechanics ranges
from 4.5% to 8.3% depending on 2-D or 3-D). The difference in value for KI between
plane stress and plane strain results from the state of modeling. In plane stress, ductility
dominates the state of modeling, while, in plane strain, brittleness dominates the state of
modeling. Generally, KI in ductile material (plane stress) is higher than that in brittle
material (plane strain).

The difference between 2-D and 3–D results from the singularity at the crack tip was
noted. In this study, the KI in 3-D modeling has a higher value. The reason is that the
singularity was not designated in this 3-D modeling. But, KI in 2-D modeling shows valid
values with the exact value from the fracture mechanics. The results show that the %
variation of KI (from 2-D to 3-D) depends on the plane stress or plane strain, not on the
various stress level. The % variation for plane strain is 8%, and the % variation for plane
stress is 4%. In this study, the importance of the singularity at the crack tip should be
taken into consideration in fracture modeling. The obtained value of stress intensity
factor (KI) by using displacement extrapolation (KCALC command) in ANSYS shows
the valid results in 2-D modeling. This data provides useful information in calculating
fatigue life.

54