Академический Документы
Профессиональный Документы
Культура Документы
Contents
Contents ....................................................................................................................................... i
List of Figures ............................................................................................................................ iii
List of Tables ............................................................................................................................. iv
List of Diagrams ......................................................................................................................... v
List of Symbols and Abbreviations .......................................................................................... vii
1
Introduction ........................................................................................................................ 1
1.1
Motivation ................................................................................................................... 1
1.2
Problem Statement....................................................................................................... 2
Background ......................................................................................................................... 3
2.1
2.2
2.3
2.4
2.5
2.6
2.7
3.2
3.3
Geometry ................................................................................................................... 20
4.2
Boundaries ................................................................................................................. 24
4.3
Loads ......................................................................................................................... 26
4.4
4.5
List of Figures
5.2
Geometry ................................................................................................................... 35
5.3
Boundaries ................................................................................................................. 36
5.4
Loads ......................................................................................................................... 37
5.5
5.6
6.2
6.3
6.4
7.2
7.3
7.4
7.5
8.2
8.3
8.4
Appendix ..................................................................................................................................... I
Bibliography ............................................................................................................................. XI
Declaration of Authorship ..................................................................................................... XIII
List of Figures
ii
List of Figures
Figure 1: crack propagation (s. [2] p. 1) ..................................................................................... 4
Figure 2: K-dominant region (s. [3] Lecture 5 p. 101) ............................................................... 5
Figure 3: mode I-III (s. [4] p. 85) ............................................................................................... 6
Figure 4: stress field in vicinity of the crack tip (s. [1] p. 23) .................................................... 7
Figure 5: Paris' Law (s. [2] p. 3) ................................................................................................. 8
Figure 6: crack closure effect (s. [7] p. 123) ............................................................................ 10
Figure 7: crack closure effects .................................................................................................. 10
Figure 8: stress increase along a screwed joint (s. [9] p. 4)...................................................... 12
Figure 9: stress distribution within a screw thread (s. [9] p. 6) ................................................ 13
Figure 10: crack tip models with -point-elements: quadr. and triang. (s. [7] p. 194) ............ 15
Figure 11: crack tip meshing (s. [3] Lecture 4 p. 74) ............................................................... 16
Figure 12: hoop stress around crack tip .................................................................................... 17
Figure 13: schematic flow diagram in FRANC2D code .......................................................... 18
Figure 14: geometry of threaded M16 connection (s. [12]) ..................................................... 20
Figure 15: symbols of the geometric parameters of the bolt and the nut (s. [9] p. 13) ............ 21
Figure 16: meshed assembly of threaded connection ............................................................... 22
Figure 17: boundaries (s. [9] p. 36) .......................................................................................... 24
Figure 18: boundaries of the bolt thread and the nut thread ..................................................... 25
Figure 19: normal cut surface and stress area (s. [9] p. 12)...................................................... 26
Figure 20: assign line................................................................................................................ 27
Figure 21: q-vector ................................................................................................................... 27
Figure 22: meshed crack tip region .......................................................................................... 28
Figure 23: deformed crack tip region ....................................................................................... 28
Figure 24: numbering of the pitches ......................................................................................... 29
Figure 25: maximal principal stress above the first loaded threads flank ............................... 30
Figure 26: crack initiation in respect of the inclination ......................................................... 32
Figure 27: model of threaded connection in FRANC2D.......................................................... 35
List of Figures
iii
List of Tables
Table 1: tolerance zone for M16 for 6g and 6H [mm] out of DIN13 ....................................... 21
Table 2: averaged normal pressure magnitudes of all five thread flanks ................................. 39
Table 3: crack length of change between the FE programs ...................................................... 44
Table 4: crack length of change between the FE programs ...................................................... 67
Table 5: inclination of initial cracks ......................................................................................... 67
Table 6: comparison of the crack growths (s. chapter 7.5)....................................................... 74
List of Tables
iv
List of Diagrams
Diagram 1: stress distribution along the screw axis ................................................................. 30
Diagram 2: principal stress along the thread root ..................................................................... 31
Diagram 3: J-Integral depending on the angle of inclination ................................................ 32
Diagram 4: linearization of normal pressure for thread flank 1 ............................................... 38
Diagram 5: linearization of tangential pressure for thread flank 2 ........................................... 38
Diagram 6: normal contact pressure of all five loaded thread flanks ....................................... 39
Diagram 7: comparison between the interaction theories ........................................................ 43
Diagram 8: crack path predicted by Franc2D........................................................................... 45
Diagram 9: normed normal force on thread flanks (Case Study 1) .......................................... 46
Diagram 10: stress redistribution in thread roots (Case Study 1) ............................................. 47
Diagram 11: accuracy of the SIFs between ABAQUS and FRANC2D (Case Study 1) .......... 48
Diagram 12: deviation of the SIFs before and after software change ...................................... 49
Diagram 13: K-a-curve (Case Study 1) .................................................................................. 52
Diagram 14: a-N-curve (Study Case 1) .................................................................................... 54
Diagram 15: Crack 2 path......................................................................................................... 60
Diagram 16: comparison of Crack 1 path between Case Study 1 and 2 .................................. 60
Diagram 17: normed normal force on thread flanks (Case Study 2) ........................................ 61
Diagram 18: stress redistribution in thread roots (Case Study 2) ............................................. 61
Diagram 19: accuracy of the SIFs Kv between ABAQUS and FRANC2D (Case Study 2)..... 62
Diagram 20: deviation of the SIF results before and after software change ............................ 62
Diagram 21:comparison of Kv between Case Study 1 and Case Study 2 ................................ 63
Diagram 22: K-a-curve of Crack 1 (Case Study 2)................................................................ 64
Diagram 23: SIF range curve of Crack 2 (Case Study 2) ........................................................ 65
Diagram 24: a-N-curve (Case Study 2) .................................................................................... 66
Diagram 25: comparison of Crack 1 path of Case Study 1 and 3 ............................................ 68
Diagram 26: crack paths of Crack 2-5 ...................................................................................... 69
Diagram 27:normed normal force on thread flanks (Case Study 3) ......................................... 70
List of Diagrams
Diagram 28: deviation of the SIFs Kv between ABAQUS and FRANC2D (Case Study 3) .... 71
Diagram 29: jumps of the SIFs due to software change ........................................................... 71
Diagram 30: comparison of Kv between Case Study 1 and Case Study 3 ............................... 72
Diagram 31: K-a-curve od Crack 1 (Case Study 3) ............................................................... 73
Diagram 32: SIF range curve of Crack 2-5 .............................................................................. 74
Diagram 33: a-N-curve (Case Study 3) .................................................................................... 75
List of Diagrams
vi
da/dN
C*
Di
di
FA
amplitude [kN]
FM
mid-load [kN]
g(a,w)
Ki
Kc
MCTS
NI
NP
R1
SIF
vii
Introduction
1.1 Motivation
In mechanical engineering, most components are subjected to variable load cycles. In
conjunction with geometry and material discontinuities, these mechanical components in
general fail because of fatigue.
By improvements of the methods determining the mechanical material behavior of
engineering components subjected to high stress, an increasing material-saving design was
made possible. Caused by this, material fatigue came to the fore. In the past high efforts were
expended to understand the phenomenon of fatigue and to develop avoidance strategies. In
many cases, the avoidance is nevertheless not possible and the development of techniques for
predicting the life-time of components was unalterable. For this purpose, it was important to
acquire knowledge about the initiation and the propagation behavior of cracks.
The linear elastic fracture mechanics is one of the fracture mechanical concepts that are
applicable to linear elastic materials. In this concept, the existing crack inside a structure is
researched. The stress intensity factors Ki characterize the nearby stress field around the crack
tip. The trajectory of the fatigue crack and the life-time cycle is mainly controlled by these
factors. In the past a lot of investigations were concerned with the development of analytical,
experimental and later numerical methods for determining the stress intensity factors. In
recent years, the numerical methods like the finite element methods gained more and more
importance caused by the increase of computer performance.
1. Introduction
The fatigue-analysis is conducted with the help of the finite element methods. The threaded
connection is reduced to a two-dimensional axially symmetric model of the screw thread and
the nut thread with linear elastic material. The thread flanks of both mechanical components
are assembled with interface elements that control the contact and the frictional behavior
among each other. For this purpose, two finite element programs are used: ABAQUS and
FRANC2D. ABAQUS is used for the simulation of the interface between the thread flanks.
FRANC2D is used for the prediction of the crack path trajectory and the determination of the
stress intensity factors. In FRANC2D, just the screw thread is modeled and the normal and
tangential interface loads between the thread flanks are replaced by linearized equivalent
loads. The stress intensity factors are periodically verified with ABAQUS. The process of the
crack initiation and the micro-crack development is not considered in this work. This
procedure is replaced by existing initial cracks of 200m length at the decisive locations.
In the first case study (Case Study 1), the presence of one crack is investigated. The expected
results of this investigation are fracture mechanical curves that illustrate the relationship
between the number of load cycles and the crack length. For this purpose, the Paris Law is
used. Additionally, the stress intensity factor range is related to the crack length.
In a further case study (Case Study 2), the crack trajectories of co-existing cracks and the
possible influence on the life endurance are investigated. Especially the trajectory of the crack
that finally causes ultimate failure and its fatigue-life is regarded and compared to Case Study
1.
The final investigation (Case Study 3) is poorly academic. All the thread roots contain initial
cracks. The aim is investigating the influence of the co-existing cracks on the crack that
causes ultimate fracture.
1. Introduction
Background
Geometry of the structure: The geometry of a structure and the applied loads create a
stress field that favors the development of cracks at locations of high stress
concentration, e. g. sharp corners. After the micromechanical development of a crack,
the stress field notably determines its continuing path.
Geometry of the crack: The crack depth, the location and direction within the structure
and the notch base radius r mainly affect the stress concentration in vicinity of the
crack tip. By increasing the crack depth, the ratio of the hot spot stress and the
nominal stress is getting higher, therefore the crack is growing faster. The hot spot
stress is in inversion proportion to the notch base radius.
Material: the material type and the deformation behavior strongly influence the
deformation in the crack tip and affect the speed of the crack growth. Of course, it is
of importance in context of predicting the crack path direction, if the material is
isotropic or anisotropic.
Stress amplitude: the influence of the stress amplitude is visible in Whler curves (S/N
curves). The increase of the stress amplitude results in a decrease of the life
expectation, because the respective cycles have a more damaging influence.
Mean stress: the minimum-to-maximum stress ratio R, respectively the mean stress,
additionally affects the speed of the crack growth. It additionally influences the
parameters of the Paris law (s. chapter 2.5).
There are a lot of other factors like temperature and corrosive environment that influence the
proceeding of the damage of the material. Nevertheless, those factors are not taken into
consideration in this work.
2. Background
2. Background
In contrast, ductile fracture problems are handled with the methods of elastoplastic fracture
mechanics (EPFM). The methods that are mainly used are J-integral (a method based on
energy methods) and CTOD (method crack tip opening displacement that is based on the
displacement of the crack flanks).
The job definition pretends the analysis and evaluation with parameters of linear elastic
fracture mechanics. Therefore, the methods of LEFM is used below.
2. Background
Mode I causes the crack opening and results from the tensile stress orthogonal to the crack
flanks inside the plane. This mode has the decisive influence on the life expectancy and
mainly controls the crack growth speed. Mode II results from the shear stress tangential to the
crack flanks inside the plane and controls the crack trajectory. The ratio between KI (Mode I)
and KII (Mode II) is affected by the location and the inclination of the crack in the structure.
2. Background
In literature, a great many of calculation formulas for the determination of SIFs is listed for
special cases. For more accurate determinations, the shape of the structure must be considered
more precisely.
Stress field in vicinity of the crack tip shown in Figure 4 (s. [5] p. 66):
Figure 4: stress field in vicinity of the crack tip (s. [1] p. 23)
Mode I:
=
!=
"
/
#
Mode II:
=
!=
"
/
/
Plane strain: $ =
)*
= *,
=/
Plane stress: $ =
2. Background
' +
'
'$ + +
'$ +
*.
/
/
/
/
&
/
(
&
(
/
/
+*
= 1 7 1, 9
2. Background
Region I:
Region 1 is the near threshold region. With exceeding the threshold Kth, the
crack propagation begins. The crack propagation stops below this threshold
value. The threshold value is influenced by the mean stress. A higher stress
ration R results in a lower critical SIF (s. [1] p. 87).
Region II:
Between the threshold Kth and the fracture toughness Kc, the stable
macroscopic crack propagation continues. In the double logarithmic diagram,
this section of the curve shows the linear relationship between the crack growth
rate da/dN and the stress intensity factor range K with gradient m. This
gradient is merely dependent on the material and additionally includes the
mean stress dependency. A higher stress ratio R results in a lower inclination
(s. [6] table 5-4).
Region III:
By approaching the fracture toughness Kc, the stable crack growth turns into
an instable crack growth and ends in ultimate fracture of the mechanical
component. The fracture toughness is dependent on the thickness of the
component, the quality of the material, the mean stress and the temperature. A
higher stress ration R results in a lower critical SIF (s. [1] p. 87). The critical
SIFs are in general listed for plane strain, because there is an influence of
thickness.
There are some other relationships between the fatigue crack growth rate and the stress
intensity factor range that respect the in reality rounded out intersection near the two threshold
values Kth and Kc (s. Figure 5), but the graphic rendition of the Paris Law has enforced
itself in science. It is quite easy to handle and merely requires the two material parameters m
and C.
2. Background
;<<
61
=>
The reasons are illustrated in Figure 7: plastic deformations (a) and phase transformations (b)
near the crack tip and the corrosion (c) and roughness (d) of the crack flanks. Additionally,
effects like viscous fluids between the crack flanks (e) can play a role.
2. Background
10
It is in the context of LEFM not possible to consider the decisive crack closure effects like
plastic deformation. The LEFM acts on the assumption that after an unloading of the
structure, the deformation state is the same like at the beginning of loading. The single
possibilities to consider this effect are solutions out of handbooks with the aim to directly
increase the stress ratio R and thereby the SIF range K.
,=>
,61
= AB C D, E/ + E D + E D + E D
E/ = /, G H /, ) I + /, /H I
E = /, ) H /, /P I
E =
E =
./ >
E/ E E
61
61
L
MN
<= D F /
OI
E/ + E
; >R1S; TU 1VS
=>
61
2. Background
= /. HH + /.
D + /.
<=
11
The screw thread consists of a slew of different sharp changes of cross section along the axis.
Figure 8 imagines a threaded connection with the stress concentration factors at several
locations along the bolt. There are three locations, where the stress concentration is quite high
and fatigue typically plays an important role:
1.) head-shank transition (location 1)
2.) runout of thread (location 4)
3.) first loaded thread flank at nut (location 6)
This work is solely focused on location 6, the first engagement of the thread at the beginning
of the nut. At this location, the root of thread has the highest stress concentration (s. Figure 8).
Therefore, the probability of the crack development is the highest. The other loaded thread
roots also have high stress concentration and are supervised while crack propagation.
2. Background
12
The infinitesimal element at the surface of the thread root is affected by all the stress
components (s. Figure 9). The maximum tensile stress component ax is tangential to the
surface. In this direction, the material has to bear the highest loads. The stress component tan
is a compressive stress component. The radial stress component perpendicular to the surface
rad is nearly zero.
2. Background
13
In addition to experimental and analytical results, the finite element method is the most
powerful tool for the numerical treatment of fracture mechanical problems. In the case of
linear-elastic fracture mechanics and the use of the Paris Law, the material parameters C and
m and the stress intensity factor range K are needed for an assessment about the amount of
load cycles N. In general, the material parameters can be found in literature. K results from
the finite element analyses.
The determination of the stress and strain distribution around the crack tip is the main
challenge for the engineer obtaining the needed stress intensity factors. Several methods were
developed to handle this problem. Three common methods are described below.
3.1.1
At the beginning of using the finite element methods, the problem of achieving the stress
singularity near the crack tip was solved by increasing the mesh refinement with
isoparametric regular standard elements in the vicinity of the crack tip. A higher element
density in general trends to result in an improved accuracy of the finite element result. The
results for the stress and strain distribution are acceptable. Despite the finer discretization with
regular standard elements that accompanied with the increasing computational power, the
result had still some unsatisfying error [10]. Nowadays, this method is rather used for rough
calculation.
In [7] p. 189, the following mesh density is recommended:
1. element size near crack tip: L = a/100 a/20
2. number of circumferential elements: >12
14
3.1.2
Crack tip elements are special elements that are developed for a more precise determination of
the singularity of the stress and strain field near the crack tip.
The order of quarter point elements around the crack tip is often used because of its very
simple application and its excellent results. The implementation in finite element software is
quite easy, thus the use of those elements is the most common method nowadays (Kuna,
2008). Those special elements are both implemented in Franc2D and Abaqus, therefore they
are used in the context of this work.
The characteristic of quarter point elements is the modification of an isoparametric element
with quadratic shape. The mid nodes of two neighbored element edges are moved to the
quarter position towards the crack tip. This simple modification is possible for both
quadrilateral and triangular isoparametric elements with quadratic trial function (s. Figure 10).
The gain of this method is a shape of the obtained stress function in the vicinity of the crack
tip that is similar to the Y Z[,\ -singularity. The assessment of the stress intensity factors is
Figure 10: crack tip models with -point-elements: quadr. and triang. (s. [7] p. 194)
15
Figure 11 illustrates the mesh of a crack tip with triangular quarter-point elements.
The following equations for the SIF extraction are from [3] Lecture 4 p. 74:
=
')
')
`(
`(
3.1.3
Hybrid elements are an alternative to general elements with pure displacement functions. The
underlying idea of hybrid elements are stress functions in addition to displacement functions.
The results are slightly better than the ones of the quarter tip elements (s. [7] p. 217).
However, these elements are quite difficult to implement in the finite element software and
are in commercial software like ABAQUS not implemented by default.
16
, O -a
b + cO
, O - O
: `
: ` Z
dB
e f
)
TV7S
gK
M + Ghi
17
18
With the objective of a realistic stress analysis of the threaded connection, the ABAQUS
version 6.12-3 (2012) is used below. Furthermore, this software is able to handle sharp cracks
and to determine the fracture mechanical parameters. It both can use the contour integral in
combination with quarter-tip elements and since version 6.12-3 XFEM for axially symmetric
problem types. The quality of XFEM results is still discussed in circles of experts, thus the
contour integral is used for an independent verification of the SIF results of Franc2D. The
crack tip can be meshed with a rosette of collapsed quarter tip elements and the fatigue
problem is solved by contour integral.
Abaqus has a whole slew of opportunities establishing the frictional contact between the bolt
thread and the nut thread. The modeling requires a high effort, but it guarantees a more
realistic result compared to equivalent loads out of special literature.
19
4.1 Geometry
Material:
Youngs modulus:
E = 204900 MPa
Poissons ratio:
= 0.30
Paris constants:
C = 7.58E-08
m = 2.63
Loads:
Mid-load:
FM = 76.5 kN
Amplitude:
FA = 8.5 kN
20
In the following table (s. Table 1), the dimension of the geometry (s. Figure 14) is controlled
relating to the tolerance of the ISO standard DIN 13. The required tolerance zone of the nut
thread is 6H, and the one of the bolt thread is 6g. Figure 15 describes the relevant
geometric parameters. The geometry complies with the tolerance zone of DIN 13
specification.
Figure 15: symbols of the geometric parameters of the bolt and the nut (s. [9] p. 13)
d2
d3
R1
parameter
0.269
2.000
upper limit
0.289
2.000
lower limit
0.252
2.000
comparison
D1
D2
14.210 14.913
Table 1: tolerance zone for M16 for 6g and 6H [mm] out of DIN13
21
The geometry parts of the bolt thread and the nut thread are modeled in ABAQUS with help
of the CAE pre-processor and afterwards assembled with special interface conditions between
these parts (s. Figure 16). Both parts are meshed with axially symmetric shell elements with
quadratic shape (ABAQUS: CAX8R-elements).
22
The mechanical contact between bolt thread and nut thread can be established by the
interaction and contact properties of ABAQUS. There are two typical ways to assign the
interaction: the surface to surface contact and the node to surface contact. The surface to
surface interaction is a reliable and mostly used method for solving a contact problem
between two parallel surfaces with friction. The definition of the interaction requires the
assignment of the master surface and the slave surface. The five threads flanks of the nut are
defined as master surface and the five flanks of the bolt thread are defined as slave surface.
The explicit adjustments of the interface properties are described in 4.2.
The thread flanks are subjected to two mechanical loads: the normal load perpendicular to the
thread flank surface and the tangential force caused by friction. The coefficient of friction is
assumed with the value 0.12.
In order to reliable consideration of the contact, the mesh density of the slave surface should
be just as high as the mesh density of the master surface. Good results compared to
maintainable computing time can be obtained by 85 surface elements in contact. With the
quadratic shape, there are about 170 nodes per surface to surface interaction.
23
4.2 Boundaries
Figure 17 illustrates the external influences on the threaded connection. The alternating load
F(t) pulls the bolt. The translation of the nut along the screw axis is avoided at the edge.
There are three boundary conditions that must be assigned in Abaqus (s. Figure 18):
1) The boundary condition (U1=0) along the rotation axis in vertical direction defines the
rotational symmetric problem.
2) The boundary condition (U2=0) along the horizontal edge of the nut does not allow
the displacement in vertical direction. This condition results from the construction
shown in Figure 17.
3) The third boundary type is the interface between the bolt thread and the nut thread. It
is not listed in the menu item Boundaries inside the Step-Module in ABAQUS, but
mechanically it is a boundary condition that establishes the connection between the
two geometry parts. The contact definition is a surface-to-surface contact with
activated small sliding effect. The edge of the nut is the master-surface, the one of the
screw the slave surface. Adjust to avoid overclosure is activated. For the normal
behavior, a hard contact with Augmented Lagrange is assigned. For the tangential
behavior, Penalty with the friction coefficient of 0.12 is defined.
4. Abaqus Model of the Threaded Connection without Crack
24
Figure 18: boundaries of the bolt thread and the nut thread
25
4.3 Loads
A fatigue loading alternating between 68kN and 85kN pulls the bolt thread. Therefore, the
mid-load FM is 76.5kN and the amplitude is 8.5kN. As described in the introduction, the
model geometry is due to time-saving analysis reduced to a two-dimensional problem.
However, the surfaces of the axisymmetric parts are nevertheless loaded by a pressure [MPa]
instead of a line loads. The surface area AC results from the parameter d3 and is smaller than
the actual normal cut surface AG (s. Figure 19). The following equations are out of [9] p.12:
E3 =
Em =
E" =
j k = )/. G 66
0
0 +0
0+0
M = H . G 66
M = nP. Hn66
The decisive surface area is AC and will be used for the determination of the pressure below.
Figure 19: normal cut surface and stress area (s. [9] p. 12)
>61 =
D=
GH///2
E3
= n/ . n n op1
nGq2
= /. G
GHq2
In LEFM, the minimum load Fmin does not play a role in the quasi-static fracture mechanical
calculation. However, the stress ratio R is an important parameter that plays a decisive role for
the prediction of the life endurance of the screw thread.
4. Abaqus Model of the Threaded Connection without Crack
26
In the next step, the crack tip must be selected. There is also the possibility to predefine a
contour integral region. The direction is chosen by the q-vector, which is in general defined
in tangential direction to the crack flanks near the crack tip (Figure 21).
Finally, the characteristics of crack tip elements must be configured. The most important point
is the adjustment of the quarter point position emanating from the crack tip (midside node
27
The crack tip is surrounded of 32 40 collapsed quarter-tip elements. The crack tip
elements and the first 7 contours have the length 2.2 m (a0 / 90). The mesh around the crack
tip complies with the suggestion of chapter 3.1.2.
Figure 23 shows the deformed crack tip region loaded with F = 85kN. At the quarter of the
length is a kink. If there is no possibility using a module that determines the fracture
mechanical SIFs, they can be determined with the displacements.
4. Abaqus Model of the Threaded Connection without Crack
28
29
500
1000
1500
2000
2500
3000
3500
1,5
2
2,5
3
3,5
4
4,5
5
stress distribution along the screw axis
Diagram 1 shows the maximum tensile stress in the thread roots 1 5. Thread root 1 yields
the highest stress. It is the thread root, in which the initial crack must be initiated.
Figure 25: maximal principal stress above the first loaded threads flank
30
Diagram 2 shows the principal stress along the thread root subjected to the maximum tensile
load (F = 85kN). The path of the thread root starts at the upper runout to the flank (point A)
and ends at the lower one (point B) (s. Figure 25). The full range of the angle is 120 (-60
60). The maximum principal stress is at 20 and has the magnitude 3171.08 MPa.
3,17E+03
3000,00
2500,00
2000,00
1500,00
1000,00
500,00
-50
-40
-30
-20
0,00
-10
0
10
20
30
40
50
60
angle []
principal stress
maximum
31
The plausibility of the results of the inclination =20 can additionally be verified with
fracture mechanical methods. After the insertion of a crack perpendicular to the edge with the
length a0=200m, the decisive location is manually searched by varying the angle of this
crack (s. Figure 26). The result of the J-Integral of each angle step is represented in a curve in
Diagram 3. The angle of approximately 18-20 (maximum: 19) is decisive for the following
analysis.
J-Integral
maximum J-Integral
7,900
7,800
7,700
7,600
7,500
7,400
7,300
7,200
7,100
7,000
6,900
0
10
15
20
25
30
angle []
J-Integral
32
FRANC2D is a specialized finite element program for the simulation of curvilinear crack
propagation in two-dimensional structures. FRANC2D 4.0 (2015) is in this thesis mainly used
for the prediction of the crack path. In version 4.0, the back-calculation function determining
the amount of load cycles by integration is not available anymore. However, this function can
be replaced by an Excel sheet. FRANC2D can handle the following planar problem types:
plane strain
plane stress
axially symmetric
The problem can be solved by the application of the axially symmetric problem type that is
necessary for the consideration of the further radial stress component out of plane. The
mathematical formulation of the axially symmetric finite element is similar to the formulation
of plane strain problems [13].
CASCA 3.2 (2007) is a very simple pre-processor for modeling the geometry and the finite
element mesh. In CASCA, the mesh can be generated with eight noded quadrilateral (Q8
elements) and six noded triangular (T6 elements) serendipity elements with quadratic shape.
CASCA was especially written for FRANC2D and was as well as FRANC2D developed at
the Cornwell University.
33
34
5.2 Geometry
ABAQUS is a bit more user-optimized for the geometry input, but due to the fact that merely
the geometry of one M16 bolt thread is required, the decision of the more reasonable method
is not that relevant.
The problem of FRANC2D is that it cannot consider contact conditions at all. After trying on
the own and a longer search in the users guide and the literature, there seems to be no
reasonable result for this problem. The single possible way in FRANC2D could be the
definition of the interface area between the bolt and nut flanks as nonlinear cohesive crack (s.
Figure 27). The results after a trial with this method are not useful. Not even if there is just
one contact area defined as cohesive crack.
35
5.3 Boundaries
Due to the fact that the definition of frictional contact between screw and nut thread is not
possible, the problem has to be solved otherwise. Instead of arranging the displacement
boundaries on the top of the nut (s. chapter 4.2), they are arranged on the top of the screw (s.
Figure 28). The equivalent loads in the interfaces between the nut and screw flanks are
determined with ABAQUS (s. chapter 4.5) and applied on the thread flanks in the FRANC2D
model.
36
5.4 Loads
The loads act instead of the finite element analysis with ABAQUS on the thread flanks and
function as substitute loads to replace the not working contact conditions in FRANC2D. A
normal pressure orthogonal to the surface and a shear pressure tangential to it are applied on
the flanks (s. Figure 29).
37
The shapes of the stress components along the first thread flank are taken from the analysis
with ABAQUS (stress components CPRESS and CSHEAR1). These two stress components
are shown in Diagram 4 and Diagram 5. The two stress curves have a similar shape and the
magnitude of the tangential stress is about 12% of the magnitude of the normal pressure. That
is consistent with the predefined shear parameter = 0.12.
y = -552,57x + 945,35
0,2
0,4
0,6
0,8
y = -66,309x + 113,44
1,00E+02
5,00E+01
0,00E+00
0
0,2
0,4
0,6
0,8
38
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,1
2. loaded thread
4. loaded thread
5. loaded thread
3. loaded thread
1. loaded thread
2. loaded thread
3. loaded thread
4. loaded thread
5. loaded thread
N [MPa]
658,27
391,46
297,27
259,90
346,33
T [MPa]
78,99
46,98
35,67
31,19
-41,56
The contact force magnitudes decrease from thread flank to thread flank (s. Diagram 6),
except thread flank 5. Thread flank 5 is subjected to higher loads than thread flank 3 and 4.
For an effortless transport of the pressure magnitudes into Franc2D, the curves are averaged
(s. Table 2) and flattened to a linearized pressure (s. Diagram 4 and Diagram 5).
Thread flank 5 has compared to thread flank 1 4 a negative frictional force (s. Table 2). The
nut flank slides towards thread root 5. The other thread flanks of the nut veer away from the
thread roots. The sliding effect is a result of the elastic widening of the nut (s. Figure 30). The
figure shows that the nut is widening at the first loaded thread flanks. At the last loaded thread
flanks, the nut thread slides in direction to the screw axis.
39
Figure 31 illustrates the principal stress distribution in thread root 1. The five thread flanks are
subjected to the linearized equivalent loads. The principal stress in the thread root determined
with FRANC2D is 3182.55MPa. The result of the principal stress in ABAQUS is
3171.08MPa. Therefore, the accuracy is about 0.36% and consequently very well.
40
The location of the crack and the inclination are determined in chapter 4.5. The initial length
a0 = 200m is given in the job definition and the inclination 0 = 20.
Figure 32 shows the process of the crack initiation. FRANC2D inserts a non-cohesive crack
by removing the elements along the chosen initial crack (b). It meshes the crack tip by a
rosette of eight triangular isoparametric quarter point elements with quadratic trial function
(T6-elements) (c). After that, FRANC2D remeshs the remaining region using a quad-tree
algorithm that produces well-formed T6-elements (d) (s. [11]). The process of the quasistatic crack propagation runs in the same way.
41
Figure 33 shows the initial crack. The crack tip is surrounded by 32 triangular quarter point
elements. The length of these triangular elements is 2.2 m (a0 / 90). The mesh around the
crack tip complies with the suggestion of chapter 3.1.2.
42
For the axisymmetric elements, the software Franc2D just offers the Displacement
Correlation Method (s. Figure 34). Therefore, it is very important to verify the results with the
fracture mechanical module of ABAQUS.
Diagram 7 shows a comparison between the three offered methods for the prediction of the
crack path. The FRANC2D model with the initial crack of chapter 5.5 was used. The paths are
illustrated in Cartesian coordinates. There is no difference between the three methods. This
result accords with the graph [3] p. 119, because KII << KI.
Y [mm]
5,10
5,60
6,10
6,60
X [mm]
initial crack
MCTS
Gmax
Smin
43
In this case study, the path of one crack inside the screw thread is predicted with help of the
two FE programs. The FE analysis with help of ABAQUS provides the equivalent loads for
the proximate analysis with FRANC2D. After that, FRANC2D predicts the crack path with
quasi-static methods. After a certain crack propagation with FRANC2D, the global
coordinates of this crack path are exported and afterwards appended to the undeformed
ABAQUS model. The analysis of the new ABAQUS model provides new equivalent loads for
the further crack path prediction. This change between these two FE programs repeats every
0.3 1.0 millimeter (s. Table 3). a is the quasi-static propagation length. After 3.0
millimeters, the fatigue analysis in FRANC2D continues with the current loads until the bolt
thread is almost cleaved into two parts (a = 6.75mm). The critical SIF range Kc is later
considered in the integration of the Paris law.
1
2
3
4
5
6
start [mm]
0.2
0.5
1.0
1.5
2.0
3.0
end [mm]
0.5
1.0
1.5
2.0
3.0
6.75
a [m]
50
50
50
50
100
200
For each step, the number of load cycles are calculated with help of the Paris law that is
detailed described in chapter 6.4. The number of load cycles are summed up and the result is
the number of cycles NP that caused the crack propagation. The number of cycles NI that is
necessary to develop the tension driven macro-cracks, can be added to NP and the result is the
life expectancy of the screw thread (s. [15]). The number of load cycles NI can be found in
literature.
44
Y [mm]
11,70
11,50
11,30
11,10
10,90
10,70
10,50
0,0
1,0
2,0
3,0
4,0
5,0
6,0
7,0
X [mm]
Path 1
Path 2
initial crack
Figure 35 shows the FE mesh of FRANC2D with complete through-thickness crack. The
crack paths imagined in Diagram 8 are the results of the complete analyses until ultimate
fracture. It is recognizable that the crack paths after a propagation of 1-2mm tend to follow
the perpendicular direction to the loading. This confirms the assumption that mode 1 has the
most controlling influence on the crack path.
6. Case Study 1: Screw Thread with 1 Crack
45
Path 1 is the crack path with consideration of the load redistribution between the thread
flanks caused by the crack growth. Path 1 is the decisive path for the integration with the
Paris law afterwards, because it rather conforms to the real circumstances.
Path 2 is the crack path, that is determined with the equivalent loads of the uncracked
threaded connection and does not consider the load redistribution.
0,5
1,0
1,5
2,0
2,5
3,0
thread flank 2
thread flank 4
thread flank 5
thread flank 3
Diagram 9 illustrates the development of the contact loads on the several thread flanks caused
by the crack propagation. The contact load is normalized and referred to the contact load of
the screw threads, when the screw thread is undefected (a = 0.0mm). The load on the first
thread flank decreases and is redistributed on the other thread flanks. The contact loads
applied on thread flank 2 decrease after 1.0mm crack growth. Also, the loads on thread flank
3 seem to reach the maximum at 3.0mm. The loads seem to redistribute on the nearest thread
flanks, but later to redistribute on the lower thread flanks.
46
1700,00
1500,00
1300,00
1100,00
900,00
700,00
500,00
0,0
0,5
1,0
1,5
2,0
2,5
3,0
3,5
thread root 3
thread root 4
thread root 5
47
deviation [%]
3,00
2,00
1,00
0,00
-1,00 0,0
0,5
1,0
1,5
2,0
2,5
3,0
-2,00
-3,00
-4,00
-5,00
Diagram 11: accuracy of the SIFs between ABAQUS and FRANC2D (Case Study 1)
Diagram 11 shows the relative error between the SIF results of FRANC2D and ABAQUS.
The SIF results of ABAQUS are compared with SIFs of FRANC2D. It is recognizable that
the relative error is positive in a short phase. However, the relative error is up to -3% during a
longer section. Perhaps, the reason is the different stress distributions caused by the different
methods of applying the loads. This is an acceptable deviation in the context of comparing the
result of two FE programs. The SIFs determined by FRANC2D is later used for the fatiguelife estimation and Diagram 11 shows that the results are on the safe side.
48
deviation [%]
4,00
3,00
2,00
1,00
0,00
0,0
0,5
1,0
-1,00
1,5
2,0
2,5
3,0
Diagram 12: deviation of the SIFs before and after software change
Diagram 12 illustrates the deviation of the SIF results before changing the FE program. In
vertical direction, the peaks show the deviation of the SIFs before updating these loads and is
normalized to the SIFs after the update of the loads. The maximum deviation of about 5% is
at 0.5mm crack propagation and the SIFs are mostly influenced by the load rearrangement.
The deviation is quite small along the whole fatigue-life and therefore the integration gives
reliable results. Better results can be achieved with an additional change between the FE
programs between 0.2mm and 0.5mm.
49
///
k = P. HG /ZG j
///
.n
= G. n/ H) /Z
The following equations provide the necessary variables for the integration with Paris law.
KII is very tiny in comparison to KI. However, as a matter of form, the von Mises SIF range
is used for the integration. However, the assumption K = KI is allowed.
=
D = /, G
+ #
+ H.
The following calculations are described in chapter 2.6 and determine the crack closure effect.
Crack closure (Newman):
= /.
E/ = /, G H /, ) I + /, /H I
E = /, ) H /, /P I
61
L
= /. /n/n
E =
E/ E E = . ) nnr
@=
,=>
E =
E/ + E
,61
61
L
MN
OI
= /. )H PP
= /. ))Gn)n
= AB C /. G; /. PrHH = /. G
50
The result of Newman shows that the crack closure effect has no influence on the life
expectancy of the screw.
Crack closure (Schijve):
W=
=>
61
= /. HH + /.
D + /.
D = /. Gr > /. G
However, the solution of Schijve shows the opposite. To be on the safe side, it is better to
ignore the crack closure effect.
= r// 266Z
`
= PG/266Z
Figure 36 illustrates the transition between stable and instable crack growth. Instable crack
growth occurs, if the maximum SIF reaches the critical SIF (s. [6] p. 109). Alternatively,
special literature offers critical SIF ranges Kc.
51
K-a-curve
7000,00
K [Nmm^(-3/2)]
6000,00
5000,00
4000,00
3000,00
2000,00
785,89
1000,00
0,00
0,00
1,00
2,00
3,00
4,00
3.4mm
5,00
6,00
a [mm]
0,25; 276,95
Kc
Diagram 13 shows the relationship between the SIF range K and the crack growth a. The
curve starts at a0 with a SIF range > 0. The range smoothly increases with the further
propagation. This effect is illustrated in Figure 36. The critical SIF range of 780Nmm-3/2 is
exceeded after 3.4mm crack growth. After that, instable crack growth begins. The exact
critical crack length ac is 3.38 mm using the linear interpolation between the quasi-static
steps.
52
Figure 37 illustrates the procedure of the integration. The area under the curve must be
determined. The subdivision of the crack growth length is fine enough caused by the tiny
quasi-static crack length steps. The determination of the crack growth rate is quite simple with
help of the Paris law:
01
0S
= 3
1U
In order to determine at, it is very important to care about the circumstance that during the
quasi-static FE analysis different crack length steps are used. The crack growth rates are
determined between the quasi-static crack length steps (between at,before and at,after).
Therefore, it is on the one hand important to consider the transitions between these quasistatic lengths and on the other hand to care about the time points a0 and ac:
1U =
1/ =
1U,_;<=
1U,1<U;
2 1 = `Uy/ w
+ 1U,1<U;
;
1U
1` =
j01O0Sk
1U,_;<=
53
a-N-curve
6,00
5,00
a [mm]
4,00
3.4 mm
3,00
2,00
1,00
0,00
0
10000
20000
30000
40000
50000
60000
N [cycles]
50660 cycles
a-N-curve
Kc
Diagram 14 shows the relationship between the crack growth a and the amount of load cycles
NP. The tangential gradient at each point of this curve shows the crack growth rate da/dn that
increases by proceeding crack propagation. This curve starts with the initial crack length a0 =
0.2mm with the number of cycles 0. The curve is incomplete, because it does not contain the
number of load cycles for the crack initiation NI. The crack propagation per load cycle
increases dependent on the further propagation. After exceeding ac, the incline of the curve is
rapidly increasing until it is nearly vertical. This means that the crack growth rate builds up
towards infinity and the crack growth is instable. The crack in fact propagates, but the
structure cannot bear further load cycles anymore. The procedure ends up in ultimate fracture.
The critical crack length ac = 3.4mm is reached after Nc = 50660 load cycles (s. Diagram 14).
54
In [6] p. 78, Irwins crack length correction for plane strain is suggested. This correction of
the crack length adds a plastic crack propagation length that considers the elastic-plastic stress
redistribution near the crack tip.
z{ = 900~
1>R =
(s. https://de.wikipedia.org/wiki/Streckgrenze)
K
M =
/.
r r. n
M = /. )r66
r//
55
Fatigue tests with low pre-load forces lead to the result that screw threads can develop more
than one fractured surface. The fractured surfaces either develop on neighbored thread roots
(s. Figure 38 (1)) or independent from each other in the most loaded thread root (s. Figure 38
(2)).
In this work, a rotationally symmetric model is used. Consequently, this model considers a
rotating crack that excludes the investigation of the independent crack (2). This model
includes this crack. The investigation of independent cracks is possible with threedimensional models and half-elliptical cracks.
56
01
D
3 '
01
0S
= 3
1 (6
01`
01`
013
1`q
D
3 a
1`q
1`q
3 a
,3 1`q
D
D
,` 1`q
,` 1`q
,` 1`q
,` 1`q
01`
013
a
6
1`q
D
01`
,3 1`q
1`q
1`q
57
It is noteworthy that the crack growth speed of Crack 2 is dependent on the ratio between the
SIFs of Crack 1 and Crack 2 that is smaller than 1.0 during the whole process. The
consequence is an over-proportional decrease of the crack growth of Crack 2 with the ratio to
the power of m. Step by step the quotient between the two SIFs becomes smaller (
, ,
, ,
Y ). The result is that Crack 2 is propagating with tiny, hardly recognizable steps and
finally stops growing. Crack 1 propagates completely through the structure. The ratio between
the stress distribution in the several thread roots (s. Diagram 10 chapter 6.2) leads to the
assumption that Crack 2 will stay belong a critical dimension. This will be shown in detail in
chapter 7.5.
58
Figure 38 illustrates the two cracks. As expected, the propagation of Crack 2 results in a
standstill.
59
Y [mm]
9,80
9,75
9,70
9,65
9,60
6,6
6,6
6,7
6,7
6,8
6,8
X [mm]
Crack 2 path)
initial crack
The crack path is shown in Diagram 15. The initial crack a0 = 200 m merely continues
growing. The crack turns away from the direct path to the screw axis. However, the
propagation is too small for a reliable statement.
Y [mm]
11,70
11,50
11,30
11,10
10,90
10,70
10,50
0,0
1,0
2,0
3,0
4,0
5,0
6,0
7,0
X [mm]
Path 1 (Case Study 2)
initial crack
Diagram 16 shows the crack paths predicted in Study Case 1 and 2. There is no appreciable
deviation between the predicted crack paths.
60
0,5
1,0
1,5
2,0
2,5
3,0
thread flank 2
thread flank 4
thread flank 5
thread flank 3
Diagram 18 illustrates the principal stress redistribution in the thread roots of the three
remaining thread roots without crack. The curves are normalized to therespective principal
stress of the uncracked structure. Also, the fifth thread root must yield the highest stress of
1315.3 MPa at 3.0mm. However, thread root 3 must yield 1339.7 MPa at a = 0.0mm.
0,5
1,0
1,5
2,0
2,5
3,0
thread root 4
thread root 5
61
deviation [%]
0,5
1,0
1,5
2,0
2,5
3,0
Diagram 19: accuracy of the SIFs Kv between ABAQUS and FRANC2D (Case Study 2)
The deviation between the determined SIFs is low or is even lower than the deviation of Case
Study 1 (s. Diagram 19).
In contrast, the change of the equivalent loads plays a more important role (s. Diagram 20)
than in Case Study 1 (s. Diagram 11). Before the update of the interface loads on the
FRANC2D model, the deviation is up to 6% at a = 0.5mm. The accuracy is lower than in
Case Study 1. Nevertheless, the results are acceptable and still on the safe side.
deviation [%]
5,00
4,00
3,00
2,00
1,00
0,00
0,0
0,5
1,0
1,5
2,0
2,5
3,0
Diagram 20: deviation of the SIF results before and after software change
62
deviation [%]
3,00
2,00
1,00
0,00
-1,00 0,0
1,0
2,0
3,0
4,0
5,0
6,0
-2,00
-3,00
-4,00
-5,00
The comparison between the SIFs of the two case studies is illustrated in Diagram 21. The
SIFs of Case Study 2 are referred to Case Study 1. The deviation of the results is about 1% 2% (excluding two outlier values). In the section between 0.2 mm and 2.0 mm the deviation is
negative, which shows a tendency that the fatigue strength Np is higher in Case Study 2.
Between 3.0 mm and 6.0 mm, the deviation is positive and very low. However, there is no
influence on the fatigue strength, because ac is exceeded. The presence of a further crack in
the structure changes the stress distribution and therefore influences the stress distribution
around Crack Tip 1. The strength of this effect is difficult to estimate. However, the deviation
is too small for a reliable statement.
63
K-a-curve
8000,00
7000,00
K [Nmm^(-3/2)]
6000,00
5000,00
4000,00
3000,00
2000,00
786.85
1000,00
0,00
0,00
1,00
2,00
3,00
a [mm]
0,20; 266,77
4,00
3.4 mm
5,00
6,00
Kc
Diagram 22 shows the relationship between SIF range and crack length. The crack exceeds the
critical SIF range Kc of 780 Nmm-3/2 after 3.4mm crack growth. The critical crack length ac
is the same as in Case Study 1. The uncertainty of ac is 200m, because the quasi-static steps
between 3.0 mm and 6.75 mm crack length are predefined with 200m (s. Table 3). However,
the exact critical crack length ac is 3.38 mm using the linear interpolation between the quasistatic steps.
64
K [Nmm^(-3/2)]
80,00
60,00
40,00
20,00
0,00
0,20
0,21
0,22
0,23
0,24
0,25
0,26
0,27
a [mm]
SIF range curve with of Crack 2
Diagram 23 illustrates the K-a-curve of Crack 2 in the structure. The SIFs slightly decrease
during the increase of the SIFs of Crack 1. The crack propagation ends in a standstill of the
crack after 62m (without initial crack length). There is no consideration of the threshold SIF
range Kth. Kv drops below the threshold value and after 22 m crack growth, there would
be in practice no crack propagation anymore (s. chapter 2.5).
,U
= ) op16 =
n. H 266Z
The crack length is verified with help of the deduced equation out of chapter 7.1:
01`
1`q
,` 1`q
,` 1`q
01`
1`q
The simplest verification of the results is to use the SIFs of every quasi-static step and to
compare the crack growth steps of FRANC2D with the crack growth steps predicted with help
of the equation above. This approach is justified, because of the nearly constant SIFs along
the crack path of Crack 2. The sum of these predicted steps dacrack2 is 67m and therefore 9%
higher than the FRANC2D result. However, this deviation is not relevant, because the crack
is quite short. The difference is so short, that the FE program can hardly consider it.
65
a-N-curve
6,00
5,00
a [mm]
4,00
3.4mm
3,00
2,00
1,00
0,00
0
10000
20000
30000
40000
50000
N [cycles]
a-N-curve
60000
51403
Kc
The following calculation considers the Irwin correction with the curve:
The determination of Np is described in chapter 6.4. The Irwin correction length of 0.49mm
can be taken out of chapter 6.4, because the SIFs just marginally differs. The result of NP is
1.5% higher than the result of Study Case 1. Here, the deviation of the SIFs illustrated in
Diagram 21 plays a decisive role. Nevertheless, the deviation of the results compared to the
results of Case Study 1 is very minor.
66
This case study is poorly academic. The aim is to investigate the influence of further cracks
in the screw thread on the life expectancy. The screw thread contains cracks in all five thread
roots. A further assumption accrues: in order to enlarge the crack growth lengths of Crack 2
5, the essential Paris constant m is lowered from 2.63 to 2.0 that is quite low for metals. The
aim of these assumptions is to investigate the influence of further and longer cracks on Crack
1. With the objective to compare the life expectancy with the results of Case Study 1, the
constant m is later put back to 2.63 again in context of the integration. The justification of this
modification of the Paris constant m is the missing influence on Crack 1 using quasi-static
methods (s. 7.1). At first, Crack 1 propagates with the predefined step ai, then the
propagation of Crack 2 5 is determined with the help of the SIFs.
Table 1 shows the quasi-static crack lengths and the crack length, at which the software
changes. Due to time-saving calculation, the iteration is rougher. Table 5 shows the
inclinations of maximum principal stress in the several thread roots. Crack 2 5 are all
initiated with 40 in the FE models.
start [mm]
end [mm]
a [m]
0.2
0.5
50
0.5
1.0
150-200
1.0
2.2
140
2.2
3.0
50
3.0
6.75
100
Crack
1
2
3
4
5
inclination
[]
20
38
40
40
42
67
Y [mm]
11,70
11,50
11,30
11,10
10,90
10,70
10,50
0,0
1,0
2,0
3,0
4,0
5,0
6,0
7,0
X [mm]
Path 1 (Case Study 3)
initial crack
68
Diagram 25 shows the trajectory of Crack 1 of Case Study 1 and Case Study 3. The difference
between the paths is insignificant.
Y [mm]
9,75
9,70
9,65
9,60
9,55
9,50
6,5
6,6
6,6
6,7
6,7
6,8
6,8
X [mm]
initial crack
Crack Path 2
Crack Path 3
Crack Path 4
Crack Path 5
Diagram 26 illustrates the development of Crack 2-5 with Paris constant m = 2.0. Due to
visual comparison, the paths are displaced to the same crack tip (Crack Tip 2). The
development of the cracks is different. The crack lengths decrease with the number of thread
flank. The crack lengths reflect the principal stress magnitude at the beginning of the crack
growth (s. Diagram 10 in chapter 6.2).
All cracks turn away from the direct path to the screw axis. However, the curvature of the
cracks is different. This is caused by the different mode 2 SIFs. The decisive crack
propagation of Crack 2 - 5 occurs during the first 1.0mm growth of Crack 1. Therefore, the
redistribution of the stress state inside the structure does not influence the crack propagation
of Crack 2 5.
69
0,5
1,0
1,5
2,0
2,5
3,0
thread flank 2
thread flank 4
thread flank 5
thread flank 3
Diagram 27 illustrates the load redistribution that is comparable to Diagram 9 and Diagram
17. The curves do not essentially differ from the curves in chapter 6.2. The tendency is
comparable (s. Diagram 9). Noteworthy are the results at a = 0.2 mm. When there is not
difference in the crack length, the load distribution on the thread flanks is identical with the
load distribution at a = 0.0 mm. After the first crack propagation, the curves have the same
shape like in Case Study 1.
70
deviation [%]
3,00
1,00
-1,00 0,0
0,5
1,0
1,5
2,0
2,5
3,0
-3,00
-5,00
Diagram 28: deviation of the SIFs Kv between ABAQUS and FRANC2D (Case Study 3)
The deviation of the results illustrated in Diagram 28 are not comparable to the results of the
other case studies. There is an offset of 1%. Maybe the number of cracks in the structure
causes problems in the FE calculation. The difference is that the deviation is not in sections
3%, but continuous. The deviation of NP after the integration is higher than in Case Study 1,
but on the safe side.
deviation [%]
4,00
3,00
2,00
1,00
0,00
0,0
-1,00
0,5
1,0
1,5
2,0
2,5
3,0
71
In contrast, the deviation of the results before load redistribution is comparable with Case
Study 1 (s. Diagram 29).
deviance [%]
2,00
1,00
0,00
-1,00
0,0
1,0
2,0
3,0
4,0
5,0
6,0
-2,00
-3,00
-4,00
-5,00
The comparison between the SIFs is shown in Diagram 30. The SIFs of Case Study 3 are
referred to the SIFs of Case Study 1. The deviation of the results is especially in the section
between 0.2 mm and 2.0 mm higher than in Case Study 2. The deviation is in this section
negative. Between 2.0 mm and 3.2 mm, the deviation is positive. However, the influence of
the SIFs is in this section not decisive. After 3.2 mm crack growth, the deviation is quite low.
It is not clear, if the number of cracks have an influence on the life expectancy of the screw
thread, because the deviation is not that high that a reliable statement is possible. Perhaps a
calculation with m = 1.0 can result in clearer deviations.
72
K-a-curve
8000,00
K [Nmm^(-3/2)]
7000,00
6000,00
5000,00
4000,00
3000,00
2000,00
786.85
1000,00
0,00
0,00
1,00
2,00
3,00
a [mm]
0,20; 267,89
4,00
3.4 mm
5,00
6,00
Kc
Diagram 31 shows the K-a-curve of Crack 1. The crack exceeds the Kc of 780 Nmm-3/2
after 3.4mm crack growth. The critical crack length ac is the same as in Case Study 1. The
exact critical crack length ac is 3.38mm using the linear interpolation between the quasi-static
steps.
73
K-a-curve
0,23; 125,84
140,00
120,00
100,00
K [Nmm^(-3/2)]
80,00
60,00
40,00
20,00
0,00
0,20
0,22
0,24
0,26
0,28
0,30
0,32
0,34
0,36
0,38
0,40
a [mm]
Crack 2
Crack 3
Crack 4
Crack 5
Diagram 32 shows the SIF range curve of Crack 2 - 5. The SIFs inside the structure seem to
be not influenced by the progression of Crack 1. The SIFs stay quite constant and after
reaching standstill, they decrease. The crack length of Crack 2 is higher, because of the
change of the Paris constant. Crack 3 5 just grow in this investigation, because the
threshold value Kth = 126.5 Nmm-3/2 is not exceeded. The SIF range goes below the
threshold value at 23 m crack propagation.
Table 6 compares the crack lengths determined by FRANC2D with the verification by the
equation used in chapter 7.1. The determination is except Crack 5 very accurate. Crack 5 is
neglectable, because the crack propagation is marginal.
Crack 2
Crack 3
Crack 4
Crack 5
FRANC2D [m]
193,02
157,87
136,21
61,38
equation [m]
189,74
159,80
134,45
77,91
[%]
-1,70
1,22
-1,29
26,94
74
a-N-curve
6,00
5,00
a [mm]
4,00
3.4 mm
3,00
2,00
1,00
0,00
0
10000
20000
30000
40000
50000
60000
N [cycles]
51419 cycles
a-N-curve (Case Study 3)
Kc
Diagram 33 shows the a-N-curve of Case Study 3. The following calculation includes the
Irwin correction:
The number of load cycles NP is 2.1% higher than the result of Case Study 1, though one
would expect a lower value. The fact that the deviation is positive, shows that there is no
influence of the 4 added further cracks, which are even artificially extended by the lower
value of m.
So, the result of the academic investigation of the influence on the life expectancy is that there
is no influence at all. Increasing the number of software changes can give more reliable
results.
75
The objective of this work was to investigate the fracture mechanical behavior of a threaded
M16 connection subjected to cyclic tensile loading. At first, the propagation of a single crack
was investigated (Case Study 1). For this purpose, finite element analyses were performed
with help of an axisymmetric 2D model of the threaded connection with ABAQUS. Crack 1
was initiated perpendicular to the surface in thread root 1 with an inclination of 20. This
inclination was determined on the one hand by locating the maximum principal stress, on the
other hand by varying the inclination and evaluating the J-integral. The normal and tangential
loads in the interface between the thread flanks of the two mechanical components were
exported to Excel. The stress components of the five thread flanks were linearized and applied
on the appropriate screw thread flanks of the 2D model in FRANC2D. FRANC2D was used
for the numerical simulation of the crack propagation within the screw thread based on linear
elastic fracture mechanics. The crack propagation in the screw thread effects a rearrangement
of the loads on the thread flanks. Hence, the crack path was updated in the ABAQUS model
geometry and the current equivalent loads on the thread flanks in FRANC2D were determined
for a further crack propagation. This interplay between the two FE programs was repeated six
times until the crack has caused the ultimate fracture. The fatigue strength analysis was
performed with quasi-static methods and the stress intensity factors were verified with the
internal module of ABAQUS while interacting with FRANC2D.
An additional objective of this work was to investigate the influence of further cracks in the
screw thread on the life endurance of the screw thread. Two cases were investigated: a further
crack in the neighboring loaded thread flank (Case Study 2) and cracks in all the five loaded
thread flanks (Case Study 3). The coexisting cracks were also initiated at the location of
maximum principal stress and maximum J-integral with a length of 200m. The inclination is
between 38 (Crack 2) and 42 (Crack 5). The result of the multiple crack growth in the
structure is that Crack 2 5 in fact propagate, but stop growing quite early caused by the
much lower SIFs and the power m. Merely the Paris parameter m and the stress intensity
factors Ki influence the propagation of the coexisting cracks. Not even the load redistribution
from the first loaded thread flank to the further thread flanks effects a further crack
propagation of these cracks. Whereas the leading crack achieves the instable crack growth
with a length of 3.4 mm, Crack 2 just achieves a length of 262 m (Case Study 2).
76
Case Study 3 is poorly academic with a lowered Paris constant m. The aim is to enlarge
Crack 2 -5 to see, if this then effects the life expectancy. The cracks achieve a length of
between 393 m (Crack 2) and 262 m (Crack 5). In order to compare the results, the
determination of the amount of load cycles was performed with the given Paris constant m =
2.63. The calculations show that both the path trajectory and the life expectancy of the leading
Crack 1 is not influenced by the coexisting cracks.
Using the critical value of Kc = 3900 Nmm-3/2, one gets a length ac = 3.4 mm for Crack 1 and
the amount of load cycles NP between 50660 (Case Study 1) and 51419 (Case Study 3). Using
Irwins correction, the crack length is 2.9mm and NP is between 48706 (Case Study 1) and
49724 load cycles (Case Study 3). The deviation of NP of the three case studies is maximum
2.1%. The life expectancy is effected only very slightly by further cracks, if at all. This is
deducted from the fact that the calculations gave even larger life expectancies for the
numerical calculations with extra cracks.
This work has the systematic weakness that the FE models are two-dimensional.
Consequently, the assumption is a rotating crack around the screw axis. A three-dimensional
model can consider the typical crack initiation of half-elliptical shape. Another advantage
over the 2D-model is that further aspects like the thread pitch and the crack initiation on the
opposite side can be considered.
The results of this work can be improved by using more iterations in which the FE program
change occurs. Between the crack lengths 0.2 mm and 1.00 mm, the stress intensity factor
range has a high influence on the life expectancy. The deviation of up to 6% between 0.2 mm
and 0.5 mm crack length can be avoided by more often applying the current equivalent loads
on the thread flanks.
77
Appendix
Normal contact pressure (Case Study 1):
a = 0,0mm:
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,1
2. loaded thread
4. loaded thread
5. loaded thread
3. loaded thread
y = -552,57x + 945,35
1000
500
0
0
0,2
0,4
0,6
0,8
y = -3,7566x + 386,32
1000
500
0
0
0,2
0,4
0,6
0,8
Appendix
y = 139,52x + 217,45
1,00E+03
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
y = 171,9x + 170,66
1000
500
0
0
0,2
0,4
0,6
0,8
y = 392,17x + 152,4
1,00E+03
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
Appendix
II
a = 0,2mm:
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,1
2. loaded thread
4. loaded thread
5. loaded thread
3. loaded thread
y = -659,04x + 969,67
1000
500
0
0
0,2
0,4
0,6
0,8
y = -23,918x + 407,12
1000
500
0
0
0,2
0,4
2. loaded thread
Appendix
0,6
0,8
III
y = 132,36x + 229,27
1,00E+03
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
y = 156,02x + 184,13
1000
500
0
0
0,2
0,4
0,6
0,8
y = 397,45x + 156,38
1000
500
0
0
0,2
0,4
0,6
0,8
Appendix
IV
a = 0,5mm:
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,1
2. loaded thread
4. loaded thread
5. loaded thread
3. loaded thread
y = -899,77x + 1039,6
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
y = -64,819x + 445
1000
500
0
0
0,2
0,4
0,6
0,8
Appendix
y = 117,95x + 252,26
1000
500
0
0
0,2
0,4
0,6
0,8
y = 154,58x + 197,89
1,00E+03
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
y = 410,86x + 162,45
1000
500
0
0
0,2
0,4
0,6
0,8
Appendix
VI
a = 1,0mm:
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,1
2. loaded thread
4. loaded thread
5. loaded thread
3. loaded thread
y = -1171,3x + 1113,2
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
y = -133,01x + 488,97
1000
500
0
0
0,2
0,4
0,8
Appendix
0,6
VII
y = 94,616x + 285,94
1,00E+03
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
y = 150,16x + 220,64
1000
500
0
0
0,2
0,4
0,6
0,8
y = 432,87x + 172,16
0,2
0,4
0,6
0,8
Appendix
VIII
a = 3,0mm:
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,1
2. loaded thread
4. loaded thread
5. loaded thread
3. loaded thread
y = -1322,1x + 1129,2
1,00E+03
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
y = -414,48x + 559,75
0,2
0,4
0,6
0,8
Appendix
IX
y = -5,7584x + 357,38
1000
500
0
0
0,2
0,4
0,6
0,8
y = 143,08x + 281,33
1,00E+03
5,00E+02
0,00E+00
0
0,2
0,4
0,6
0,8
1500
1000
500
0
0
0,2
0,4
0,6
0,8
Appendix
Bibliography
[1]
[2]
[3]
Course material "CEE 7700 Engineering Fracture Mechanics" Cornell University 2011.
[4]
[5]
[6]
[7]
[8]
[9]
D.-I.
S.
Beyer,
Course
material
"Verbindungstechnik"
"Tragfhigkeit
von
[19] J.-H. Kim und G. H. Paulino, Mixed-Mode crack propagation in functionally graded
materials.
Declaration of Authorship
Signature:
Unterschrift: