Академический Документы
Профессиональный Документы
Культура Документы
Signature : .
Name of Supervisor : P.M. DR. SARIFFUDDIN BIN SAAD
Date : 23rd Mac 2005
DETERMINATION OF CTOD AND CMOD OF TUBULAR
T-JOINT USING THE FINITE ELEMENT SOFTWARE
COSMOS/M
MARCH 2005
ii
I declare that this thesis entitled Determination of CTOD and CMOD of Tubular T-
Joint using the Finite Element Software COSMOS/M is the result of my own
research except as cited in the references. The thesis has not been accepted for any
degree and is not currently concurrently submitted in candidature of any other
degree.
Signature : .
Name : OOI SEOK TENG
Date : 23rd MARCH 2005
iii
ACKNOWLEDGEMENT
I am also very thankful to Associate Professor Dr. Abdul Kadir Marsono for
his guidance and advice on my modeling using the COSMOS/M software.
ABSTRACT
In recent years, more and more attention has been paid to the study of fracture
behavior of cracked tubular joints. Crack exist to some extent in nearly all tubular
joints, either as a result of manufacturing fabrication defects or localized in service.
This undergraduate project is concerned with the investigation of the applicability of
COSMOS / M in solving the fracture mechanics parameter of tubular joint i.e. the
crack tip opening displacement, CTOD and crack mouth opening displacement,
CMOD of the tubular T-joints in the linear elastic region. The COSMOS/M finite
element software was used to obtain the results which were then compared with
relevant experimental results from the literature. Shell elements were used to model
the tubular joints with surface cracks; where line spring elements were embedded
into shell elements at the chord wall near the brace chord intersection to stimulate
cracks. In this approach, the crack geometry was actually not modeled and the line
spring elements are used to provide local flexibility of the two sides of a crack along
the edges of the shell elements. The finite element analyses were carried out in
different values of applied tension load at the brace end in order to obtain the
corresponding CTOD and CMOD. The finite element numerical results were
compared with the experimental test results obtained from the literature. It was found
that the CTOD results at various loads levels are quite close to the corresponding
experimental results. However, the CMOD results obtained in this work are quite
different compared to the experimental results at various levels of elastic loading.
vi
ABSTRAK
TABLE OF CONTENTS
PREFACE i
DECLARATION ii
DEDICATION iii
ACKNOWLEDGEMENT iv
ABSTRACT v
ABSTRAK vi
TABLE OF CONTENT vii
LIST OF TABLES x
LIST OF FIGURES xi
LIST OF SYMBOLS xiv
1 INTRODUCTION
1.1 Introduction 1
3 METHODOLOGY
3.1 Preliminary Survey 7
3.2 Literature Review 7
viii
4 LITERATURE REVIEW
4.1 Part I: experimental data base and brief summary of the
results. 12
4.2 Part II: R6 analysis. 14
4.3 Part III: UK BS7910 methodology. 15
4.4 Part IV: Application of the ETM 97/1. 16
4.5 Part V: Screening method by required toughness and
plastic stability considerations. 17
4.6 Part VI: Application of WES 2805-1997. 18
4.7 Tubular Joints
4.7.1 Introduction 20
4.7.2 Simple Welded Joints 20
5 THEORY REVIEW
5.1 Fracture Mechanics Approach 27
5.2 Linear Elastic Stress Field in Cracked Bodies
5.2.1 Crack Deformation Modes 27
5.2.2 The Energy Criterion 28
5.2.3 The Stress Intensity 31
5.3 Elastic-Plastic Stress Field in Cracked Bodies
5.3.1 Irwins Model 33
5.3.2 The Dugdales Model 34
5.3.3 The Crack Tip Opening Displacement (CTOD) 36
5.4 The Finite Element Method
5.4.1 Introduction 37
5.4.2 Concept of Finite Element Method
ix
REFERENCES 84
x
LISTS OF TABLES
LISTS OF FIGURES
LISTS OF SYMBOLS
E - Young Modulus
- Stress
- Poisson Ratio
- Degree
G - Strain Energy Release Rate
CTOD - Crack Tip Opening Displacement
CMOD - Crack Mouth Opening Displacement
K - Stress Intensity Factor
U - Energy
CHAPTER 1
INTRODUCTION
1.1 Introduction
It is now well accepted that all welded structures contain flaws, and that these
do not necessarily affect structural integrity or service performance. This is implicitly
recognised by most welding fabrication codes which specify weld flaw tolerance
levels based on experience and workmanship practice. However, these flaw
acceptance levels cannot provide quantitative measures of structural integrity, for
instance how 'close' a particular structure containing weld flaws is to the failure
condition. In addition, flaws can develop during service due to e.g. corrosion and
fatigue and the tolerance of the component regarding these needs to be known.
Early in the war, welded merchant vessels experienced difficulties in the form
of fractures, which could not be explained. The fractures, in many cases, manifested
themselves with explosive suddenness and exhibited a quality of brittleness, which
was not ordinarily associated with the behavior of a normally ductile material such as
2
ship steel. It was evident that the implications of these failures on welded ships might
be far-reaching and have a significant effect upon the war effort. They are::
(i) 4,694 welded steel merchant vessels were built by the Maritime Commission
in the United States and considered in this investigation;
(ii) 970 of these vessels suffered casualties involving fractures;
(iii) 24 vessels sustained a complete fracture of the strength deck;
(iv) 1 vessel sustained a complete fracture of the bottom;
(v) 8 vessels were lost, 4 broke in two and 4 were abandoned after fracture
occurred, 4 additional vessels broke in two, but were not lost;
(vi) the highest incidence of fracture occurred under the combination of low
temperatures and heavy seas;
(vii) every fracture examined started in a geometrical discontinuity or notch
resulting from unsuitable design or poor workmanship.
For over a century, researchers have studied the fracture and fatigue
properties of engineering materials, the main objective being to be able to predict the
fracture/failure characteristics or the fatigue life of a component. The fracture
properties of different materials undergoing different load patterns associated with
different environments are generally obtained by experimental testing. Knowing
these fracture properties, it is possible for engineers and designers to assess, with the
help of fracture mechanics, just how components are likely to behave while being
loaded.
Offshore structures used for oil and gas extraction have the common function
of providing a safe, dry working environment for the equipment and personnel who
operate the platform. Currently, four distinct types of structure are used. The most
common one is the Jacket, a term which originates from the steel template placed in
3
the shallow water swamps of Louisiana, as a guide for the piled foundations which
supported drilling equipment.
As a powerful and versatile numerical analysis tool, the finite element (FE)
method has been used widely for the analyses of different types of welded circular
tubular joint. In recent years, more and more attention has been paid to the study of
fracture behavior of cracked tubular joints. Crack exist to some extent in nearly all
tubular joints, either as a result of manufacturing fabrication defects or localized in
service. Finite element analysis using software e.g. COSMOS/M [1]can be used as an
attractive solution to carry out the fracture mechanics analyses on tubular joint rather
than to test the actual joints due to its massive size and also the associated testing
costs.
4
CHAPTER 2
The purpose of the study is to identify and delineate how COSMOS / M can
be applied in fractures mechanics study of tubular joint. The program produces
displacements, strains, stresses, forces, and error estimates as results of the analysis
under a variety of loading conditions.
(a) to model and analyse the T-tubular joint based on the experiment data from
referreed journal using the Finite Element Method (FEM): COSMOS / M;
(c) to obtain the load applied versus crack tip opening displacement and load
applied versus crack mouth opening displacement of the tubular T-joint using
the COSMOS/M FE software system.
5
The scopes of the study involve the analysis on Tubular T joint based on the
fracture mechanics that involve only tensile loading of the brace (tension force only).
The fracture mechanics analysis will be carried out using the finite element software
COSMOS/M [1]: finite element analyses.
CHAPTER 3
METHODOLOGY
Preliminary Survey
Literature Review
Theory Review
Data Collection
The literature review in this study is divided into two major components,
namely Fracture Mechanics Analysis and Introduction to Finite Element Analysis.
This chapter is divided into two main parts which are the Theory of Fracture
Mechanics in the Linear Elastic and Elastic Plastic Fracture Mechanics, the Finite
Element Analysis / Method (FEA / FEM).
8
The data are obtained from the experimental results from the Journal named:
The fracture behaviour of a welded tubular joint an ESIS TC1.3 round robin on
failure assessment methods [2].
Figure 3.1 and Figure 3.2 show the geometry and dimensions respectively of
the T-joint used in this study. Tensile loading was applied to the end of the T brace
and reacted at both ends of the chord.
A starter notch for the cracks was introduced in the saddle region of the joints
by electric discharge machining at a distance of 2 mm from the weld root position.
The angle between the notch plane and the loading direction was chosen as 66o to
ensure a straight extension of the fatigue crack. The geometry of the weld at the
saddle point is shown in more detail in Figure 3.3, which was taken from an original
component.
3.4.2 Material
The material used in the experiments was a cold deformed 450 YS TMCP-
steel. The tubes were cold prestraining as it was manufactured from plates by
bending and butt-welding. The Youngs modulus was determined as 210 GPa.
9
The collected data need to be converted into a standard format. Using the
Finite Element Analysis software named COSMOS/M, the data are then being
tabulated to produce result. New user needs to explore the functions of the software
to set-up database. The COSMOS/M software was used to generate a computer
model of the T-joint described above. Figure 3.4 shows a typical mesh of the T-joint.
Figure 3.1: Geometry, crack position, and loading geometry of the T-joints (schematic) [2]
CHAPTER 4
LITERATURE REVIEW
4.1 Part I: experimental data base and brief summary of the results. [2]
The journal described the experimental work that was being carried out to
determine the fracture behaviour of welded tubular T-joints made of high strength
TMCP- steel. The experimental information obtained from the experiments on the
joints are supposed to serve as reference solutions to check various fracture
mechanics assessment approaches.
Four T-joints were tested by tensile loading of the brace. Their geometry and
dimensions are presented in Figure 3.1 and Figure 3.2 respectively. A starter notch
for cracks was introduced in the saddle region of the joints by electric discharge
machining at a distance of 2 mm from the weld root position. The material used in
the experiments was a cold deformed 450 YS TMCP-steel. The tubes were cold
prestraining as it was manufactured from plates by bending and butt-welding.
The Youngs modulus was determined as 210 GPa. Multiple J- and CTOD-R
curves were determined on these small bend specimens. The J-integral was
calculated from the crack mouth opening displacement (CMOD) while the crack tip
opening displacement (CTOD) values were determined according to ESIS P2-92 and
ASTM E 1290-93. Additionally, crack tip profiles were obtained by a sectioning
technique, but for the centre line of the specimens only. There are fours types of
measurements used in the T-joints. There are:
13
There are six different types of flaw assessment methods were applied to the
T-joints experiments. The methods investigated include:
i. finite element analyses,
ii. R6 analyses,
iii. UK BS 7910 (PD 6493) methodology,
iv. engineering treatment model (ETM 97/1),
v. screening method according to Schindler et. al,
vi. design curve according to WES 2805-1997.
14
The R6 procedures in this paper used the method for primary loading, as
relevant to the application to the T-joint. The R6 procedures involved the calculation
of two parameters; measure of the proximity to linear elastic fracture and a measure
to proximity to plastic collapse.
The T-joint geometry and loading arrangement are shown in Figure 3.1 and
Figure 3.2, which includes the overall components dimensions. The R6 calculations
were performed to evaluate the limiting value of the central load on the brace of the
T-joint. For the purposes of the stress intensity factor calculation, the stress
distribution normal to the crack plane at a load of 480 kN had been used.
It was found that the problem was collapse-dominated through the calculation
of stress intensity factor at both the surface and deepest points on the defect.
Therefore, the precise definition of K was not important for determination of the load
carrying capacity of the T-joint. Results are shown in Figure 4.2 and the relevant
region of the failure assessment diagram is shown in close-up in Figure 4.3. From the
diagram, the limiting load F = 2.026 GN and 10 mm deep defect lying just on the
cut-off.
The calculated limiting load is close to but less than maximum loads applied
experimentally of between 2.107 2.397 GN for four joints with crack depths in the
range of 9.00 11.46 mm. It shows that the limiting load is governed by plastic
collapse and the collapse load is insensitive to the crack size.
The journal is about the analyses that have been carried out to give a safe
prediction of the ultimate strength of a series of cracked tubular joints as part of an
ESIS round robin exercise. The basic data for the round robin assessment is given in
Part 1 of this series of papers.
The calculations have been based partly on the procedures given in British
Standard BS7910:1997 and partly on the experience of the author gained in several
programmes of research into ultimate strength of cracked tubular joints. The analysis
been carried out for the target case of initial surface defects of 10 mm depth and
46.5mm length at the weld toe at the saddle position of tubular T-joints in TMCP
steel even though the actual defect depths in the four tests carried out are reported to
have been 9,11.46, 9.94 and 10.67mm.
Expectation for the analyses would be that the tubular joints would be
expected to exceed the HSE characteristic design strength and approach the HSE
mean strength with only very limited ductile tearing crack extension. (Tubular joints
> HSE characteristic design strength) & (Tubular joints HSE mean strength).
Tearing would commence at the ends (surface) of the crack before the deepest point
with initiation of tearing at not less than 1000kN.
There are also some assumptions being made in the analyses; a notional R-
curve for more extensive tearing to explore the accuracy of predictions on this basis,
even though well outside any validity limits. The R-curve was terminated at ductile
crack extensions of 7 mm at the ends of the crack and when tearing had just
commenced at the deepest point. The author predicted that the maximum load in the
brace of about 2000 kN at 1mm tearing and the limit load of 3420 kN at 7 mm
tearing, approaching the nominal yield load of the brace. The actual tubular joint test
based on the experimental result is that the joint reached a maximum load of 2500kN
with very limited tearing.
16
A special feature of the ETM is that it prefers the 5 as definition for the
crack tip opening displacement as an alternative to the plastic hinge definition of BSI
and ASTM. The definition is illustrated in Figure 4.5. The CTOD determined by a
sectioning technique at the deepest point of the surface crack by its nature is close to
the 5 definition, but it was also shown to be comparable with the plastic hinge
definition.
The information needed for an ETM application to the T-joints includes the
geometry and dimensions of the component. The components were four welded
tubular T-joints containing surface cracks in their saddle region. They were loaded in
axial tension. The crack shape developed during precracking was almost an ideal half
ellipse.
The yield load, FY, of the cracked component was determined in two different
ways; by applying an analytical yield load solution and by using a finite element
analysis based on shell element. The related yield strength values were 430 and 460
MPa, respectively resulted in upper and lower bounds of the yield loads. The yield
loads estimated analytically and the shell finite element results were nearly identical.
The elastic stress distribution in the crack plane, determined without the
presence of the crack. The result illustrates the phenomenon that in the elastic
17
loading regime the hot spot region at the saddle point was loaded predominantly by
bending even though the remote load at the brace was pure tension.
The journal described the first screening method that considered a structure
with respect to its defect sensitivity. The method enabled one to identify fracture
critical parts and to predict roughly the behaviour of crack-like defect.
18
First of all, by considering the loading behaviour of the joint by simple rigid
plastics models according to the basic theory of strength of materials. From the
calculation, it was found out that the first yielding that manifested on the global load
vs. load-line displacement curve occured at a load of 1.23 MN.
When the load F was further increased, a plastic zone will spread out from A
into the chord (see Figure 4.6). One hand along the circumferential weld toe and the
other one perpendicular to the tube axis deeper into the chord. Therefore, two modes
of plastic collapse could occur when these zone were saturated. From the calculation,
plastic pull out of the brace occurs before plastic hinge formation in the chord.
Therefore, the representative plastic limit load Fp was 1.93 MN.
Even though the result of a simple model is- as expected conservative but
the predictions made in the present paper were done just on the basis without
knowing the further experimental results as documented. Comparison between the
predictions in the present paper and the measurements given shows a relatively good
agreement. As a conclusion, the cost-effective engineering failure assessment
technique presented, which does not required any expensive stress calculations, is a
powerful tool for an initial analyses of the failure behaviour of structures.
notched geometries, is the sum of the strain portions due to the boundary force, 1,
the welding residual stresses, 2 and strain concentration at the notch, 3. The WES
2805 procedure had been developed for brittle fracture. Brittle fracture did, however,
not exclude extended plasticity and even ductile crack growth. In the upper ductile-
to-brittle transition range which covered most of the brittle failures in engineering
practice, the cracked ligaments usually experienced large scale yielding and some
stable crack growth before failing by cleavage.
The T-joint investigated in the paper behaved in a ductile manner. They were
not broken but unloaded after a certain amount of stable crack growth. The elastic
stress concentration factor, Kt was determined experimentally. The WES analysis
was based on two Kt values 3 and 3.5 in order to cover for experimental scatter. In
both cases (T-joints B and E), the WES analysis yielded accurate or slightly
conservative results. In contrast to this, the crack tip loading was underestimated for
T-joints F and G, i.e., the results were significantly non-conservative. This was
because the analysis does not take into account the effect of stable crack growth on
the measured CTOD. In figure 4.7, the influence of increasing stable crack growth on
the conservatism of the WES predictions was illustrated.
Offshore structures used for oil and gas extraction have the common function
of providing a safe, dry working environment for the equipment and personnel who
operate the platform. Jacket, a term, which originates from the steel template placed
in the shallow water swamps of Louisiana, as a guide for the piled foundations,
which supported drilling equipment. The substructure, referred to as the 'jacket', is a
three-dimensional space frame made from large tubular steel members. The jacket,
which takes the loadings from the topside and the sea environment, is piled to the
seabed. These piles must also be able to resist tension as the hydrodynamic forces on
the structure have a tendency to cause overturning.
Simple welded joints are those formed by welding two or more tubular
members in a single plane without overlapping of brace members and without the use
of gussets, diaphragms, stiffeners or grout. Unlike a pipe joint, the chord wall is left
intact with the hidden plug regions enclosed by the braces. The geometric and other
notations given to simple joints are shown in Figure 4.8.
From the geometries of the braces and the chord meeting at the tubular
intersection, the following geometric ratios may be defined:
diameter ratio
gap parameter
When the chord member is not thick enough to prevent excessive local
deformation, the chord wall may be thickened in the region of the joint by the
addition of a can. This will allow the static strength and/or fatigue strength of a joint
to be altered, without unnecessarily increasing the self-weight of the entire member.
Figure 4.1: Principle of the determination of the CTOD- a curves on the T-joints [3]
23
CHAPTER 5
THEORY REVIEW
(a) Mode I: The Opening Mode, where opposing crack surfaces move directly
apart. The forces are perpendicular to the crack (the crack is horizontal and
the forces are vertical), pulling the crack open.
(b) Mode II: The Edge Sliding Mode, where crack surfaces move more over each
other perpendicular to the crack tip. The forces are parallel to the crack. One
force is pushing the top half of the crack back and the other is pulling the
bottom half of the crack forward, both along the same line. This creates a
shear crack: the crack is sliding along itself. It is called in-plane shear
because the forces are not causing the material to move out of its original
plane.
(c) Mode III: The Shear Mode, where crack surfaces move over each other
parallel to the crack tip. The forces are perpendicular to the crack (the crack is
in front-back direction, the forces are pulling left and right). This causes the
material to separate and slide along itself, moving out of its original plane
(which is why its called out-of-plane shear). The forces could also be pushing
left and right and the same effect would occur. But the forces have to be
moving in opposite directions in order to grow the crack.
unstable if the stored energy released at fracture is greater than the creation of
surface energy due to new surfaces. In other words, the energy approach states that
crack extension (i.e. fracture) occurs when the energy available for crack growth is
sufficient to overcome the resistance of the material. The material resistance may
include the surface energy, plastic work, or other type of energy dissipation
associated with a propagating crack.
To apply his theory, we may consider the following model. Consider the plate
in the absence of the crack as shown in Figure 5.3 (a) when the material is uniformly
stressed and fixed remotely (constant displacement). The energy per unit volume is
known to be / 2E, where E is Young modulus. Next, introduce a crack of length
2a into the plate. There will be a general relaxation of the material above and below
the crack, and some strain energy will be released. In order to obtain an approximate
solution, let us assume that the released zone is in the form of a triangle, height a
above and below the crack shown shaded in Figure 5.3 (b). The relaxed volume of
the shaded zone is 2a2B, where B is the plate thickness. Hence, the energy, U
released per unit thickness is given by:
This is in good accord with Griffiths accurate solution for plane stress:
2
U 2
U = a 2 , = a (5.2)
2E a E
0 < a ac, energy input into the system is required to increase the crack length but
when a > ac, the crack will grow without the input of additional work. Therefore,
strain energy within the system provides the work needed for crack growth, and ac is
defined as the critical length at which crack growth takes place.
The value of (U/a) defines the strain energy release rate for an incremental
crack extension. Differentiating Equation (5.2), leads to:
U 2 a
G = = (Plane stress) (5.3)
a E
2 a
G = (1 2 ) (Plane strain) (5.4)
E
The critical value of G at which crack growth takes place provides a measure
of what is usually known as materials fracture toughness that is its resistance to
brittle fracture. It could be regarded as providing a measure of the materials
effective surface energy. In principle, the critical value of G may be measured by
increasing the load in a quasi-static test on a cracked specimen, and noting the
critical (tensile) stress, c, at which crack growth takes place. The critical values of G
for the plane stress and plane strain condition are given the symbols Gc and GIc
respectively. It is assumed that only Mode I crack growth is being considered.
Replacing variables in Equation (5.3) and Equation (5.4) by critical values of Gc
(plane stress), GIc (plane strain) and c gives:
2
a
Gc = c
(Plane stress) (5.5)
E
31
2
a
G Ic = c
(1 2 ) (Plane strain) (5.6)
E
For a particular mode of crack tip surface displacement, the leading term is
always of the same general form. Individual stress components and displacement are
proportional to K/r1/2, where r is the distance form the crack tip and K is the stress
intensity factor. Subscripts I, II, III are used to indicate mode.
lim
K = y ( 2 r ) 1 / 2 (5.7)
r o
The in plane stresses, which are identical for plane stress and plane strain are:
K 3
x = cos 1 sin sin (5.8)
( 2 r ) 1/ 2
2 2 2
K 3
= cos 1 sin sin (5.9)
( 2 r )
y
2 2 2
1/ 2
32
K 3
xy = cos 1 sin cos (5.10)
( 2 r ) 1/ 2
2 2 2
K I = a (5.11)
Failure occurs when KI = Kc. in this case, KI is the driving force for fracture
and Kc is a measure of material resistance. As with Gc, the property of similitude
should apply to Kc. that is Kc is assumed to be a size-independent material property.
Comparing Equation (5.5) and (5.11) results in a relationship between KI and G:
2
KI
G= (5.12)
E
Thus the energy and stress intensity approaches to fracture mechanics are
essentially equivalent for linear elastic materials.
33
According to the elastic stress field solutions, a stress singularity exists at the
tip of an elastic crack. There is always a region around the tip of a crack in a metal,
where plastic deformation occurs, and hence a stress singularity cannot exist. The
plastic region is known as the crack tip plastic zone.
Figure 5.5(a) shows the magnitude of the stress y in the plane = 0. Until a
distance rp* from the crack tip, the stress is higher than the yield stress ys. Take rp*
as the first estimate of the extent of the plastic zone.
2a
2
K1 K1
y = = ys rp = =
*
or (5.13)
2 ys 2 ys
2 2
2rp
*
It is clear that the force produced by the stress shown in the shaded area in
Figure 5.5(a), acting over length r will produce further yielding. In fact the whole
stress curve must be shifted so that equilibrium is maintained.
Irwin argued that the occurrence of plasticity around the crack tip may result
in the displacement become larger and the stiffness is lower than in the elastic case.
In Figure 5.5(b), the physical crack size is replaced by a longer crack of size,
aff = a + , the physical crack size plus a correction and the elastic stress
distribution (y) at the tip of the effective crack is given. Once again, the stress at the
tip of the effective crack is again limited to the yield stress ys. Therefore, in order to
replaced the lost by cutting the area A as shown in Figure 5.5(b), must be large
enough to carry the load so that the area A is equal to area B. The distance, can
written as:
K a + 2 (a + )
ys = = or = rp
*
(5.14)
2 2 2 ys
2
34
Since is small compared to the crack size, it can be neglected and rp*. By
equalizing the area A and area B with being neglected,
The size of the plastic zone rp is found to be twice as large as the first
estimate, rp*, also known as Irwins plastics zone correction. Assuming for the time
being that the plastic zone has a circular shape, the situation can be represented as in
Figure 5.5(c), where the effective crack extends to the centre of the plastic zone. If
the plastic zone correction is applied consistently, a correction to K is also necessary:
(
K = C a + r p
*
) (5.16)
4
COD = 2 = a2 x2 (5.17)
E
Where x = a at the crack tip. If plasticity occurs, crack tip blunting takes place and
the crack tip opening displacement (CTOD) may be different from zero, whereas for
x = a the prediction from Equation (5.16) is that CTOD = 0.
CTOD =
4
E
(a + r ) p
* 2
a
2 4
E
2arp
*
=
4 K2
E ys
(5.18)
the crack length is increased from 2a to 2 (a+), where is the extent of the plastic
zone.
2 2 a K 2
= = (5.19)
8 ys 8 ys
2
By using the plastic zone correction of Dugdale, the CTOD equation becomes:
8 ys a
= CTOD = log sec (5.20)
E 2 ys
After series of expansion of the log sec yields the following formula for CTOD (and
as long as the is small compared to ys):
2 a G1
CTOD = = (5.21)
E ys ys
In general,
K 1 (1 2 )
2
G1
CTOD = = Plane strain situations (5.22)
ys E ys
2
G1 K1
CTOD = = Plane stress situations (5.23)
ys E ys
36
4
COD = 2 = a2 x2 (5.24)
E
As being discussed in Section 5.3.1 regarding the crack tip plastic zone, the
displacement of crack surfaces equation always being used together with the plastic
zone correction rp*.
(a) By applying a plastic zone correction of Irwin, the COD equation become:
COD =
4
E
(a + r )p
* 2
x2 (5.25)
where a + rp* is the effective crack size and where the origin of the coordinate system
is at the center of the crack. Since rp* << a, it turns out that
4
CTOD =
*
2arp (5.26)
E
A displacement of the origin of the coordinate system to the crack tip yields the
general expression for crack opening:
4
COD = 2 a eff r (5.27)
E
CTOD equation follows from r = rp* and aeff = a. Substitution of rp* = 2a / 2ys2
yields:
2
4 K1
CTOD = (5.28)
E ys
37
(b) By applying the plastic zone correction of Dugdale, the CTOD equation
become:
2
G1 K1
CTOD = = Plane stress situations (5.29)
ys E ys
The finite element method (FEM) is a numerical technique, which solves the
governing equations of a complicated system through a discretization process. The
system of interest can be either physical or mathematical. The governing equations
can be given in differential form or be expressed in terms of variational integrals.
FEM has been successfully used to solve very complex solid mechanics, fluid
mechanics, heat transfer and electro-magnetic problems.
The engineer of today can concentrate on the objective of the analysis, and
then obtain answers that normally would have required a prototype. Indeed, modern
graphic portraits of the model inside the computer make one believe it is real. The
FEM is used as it:
(a) can analyze solids with complicated Structural Configurations (shapes),
(b) can consider complicated Loadings,
(c) can consider complicated Boundary Conditions.
38
The general idea of the finite element method is illustrated in the Figure 5.8,
where u is used to represent the displacements or deformation effects. When
problems such as heat and fluid flow are discussed, they will involve distribution of
temperature and fluid heats and their gradients.
Before an analysis is carried out, the whole system has to be divided into a
number of individual subsystems or components, whose behaviour is readily
understood. The basic units of the discretized subsystems are called finite elements,
which should neither overlap nor have gaps between each other. The body is divided
into a number of smaller regions as shown in Figure 5.8 (a) called finite elements. A
consequence of such subdivision is that the distribution of displacement is also
discretized into corresponding sub zones, Figure 5.8 (b). The subdivided elements
are now easier to examine as compared to the entire body and distribution of u over
it.
different planning and operations depending upon the physical nature and the
mathematical modeling of the problem.
The solution obtained will be in terms of the unknown only at the nodal
points. Figure 5.9 shows the final solution is a combination of solutions in each
element patched together at the common boundaries. It is further illustrated by
stretching a cross section along A-A. The computed solution is not necessarily the
same as the exact continuous solution shown by the solid curve.
the stress or velocity must also be defined. This can be done by expressing this
relationship with the strain, called stress-strain law.
Use of either of the two foregoing methods will leads to equations describing
the behaviour of an element, which are commonly expressed as
where,
41
Until now, the properties of the body or structures have being considered but
without constraints as shown in Figure 5.10 (a). In the case of engineering bodies, the
surrounding or the constraints are the boundary conditions. Boundary conditions are
the physical constraint or supports that must exist to withstand the structure or body
in space uniquely as shown in Figure 5.10 (b). To reflect the boundary conditions in
the finite element approximations of the body, the final modified assemblage
equations are expressed as
[ K }{r } = {R } (5.33)
K 11 r1 + K 12 r2 + ... + K 1n rn = R1 ,
K 21 r1 + K 22 r2 + ... + K 2 n r n = R 2 ,
.
.
K n1 r1 + K n 2 r2 + ... + K nn rn = R n , (5.34)
The aim is to reduce the results obtained from the finite element calculations
to a form that can be readily used for analysis and design. Results can be obtained on
the form of printed output from the computer. The critical sections of the body be
selected, and the values for displacement and stresses along them are plotted.
The tetrahedron shown in Figure 5.13 (a) is the basic finite element for three-
dimensional problems corresponding to the triangle. It has four primary external
nodes. If necessary, secondary external nodes or internal nodes for each of the
elements can be introduced. By analogy, with the two-dimensional elements, a
hexahedron can be constructed from several tetrahedral. Other types of three-
dimensional elements are shown in Figure 5.13.
The significant advances made in finite element technology, coupled with the
rapid developments in computer hardware and software, provided the foundation
from which many general-purpose finite element programs have evolved. After many
years of research and development, a wide variety of finite element programs are
currently being used in government and industry for the solution of a wide variety of
practical problems.
Some of the finite element software systems available in the market, which
are widely being used in the structural analysis are ANSYS, ABAQUS, COSMOS /
M, GTSTRUDL, LUSAS and many more.
With STAR, users can address 2D and 3D linear static problems. The
programs large element library and its many other modeling features allow users to
set up the designs for quick and accurate analyses. Among the extensive capabilities
of STAR is the support of various types of constraint equations, and bonding of
surfaces with incompatible meshes. STAR also supports other features such as
contact, geometric stiffening, P-type elements and adaptivity, asymmetric loading,
and much more.
45
48
49
CHAPTER 6
6.1 Methodology
6.1.1 Introduction
The success of a finite element method for the modeling and analysis of a
structure is based largely on the basic procedures used. The basic steps in the finite
element analysis are:
i. Create the problem geometry
ii. Mesh the defined geometry with appropriate type of element (s).
iii. Apply boundary conditions (constraints) on the finite element model.
iv. Define the loads on the model.
v. Define the material properties.
vi. Submit the completed finite element model for analysis.
vii. Interpret and analyze the results.
Basically, the process can be divided into three main sections, which are the
preprocessing, analysis and post processing. The preprocessing refers to the
operations to perform such as defining the model geometry, mesh generation,
applying loads and boundary conditions, and other operations that are required prior
to submitting the model for analysis. The analysis refers to the phase of specifying
the analysis option and executing the actual analysis. Lastly the post processing
refers to the manipulation of the analysis results for easy understanding and
interpretation in the graphical environment.
53
In this modeling, the module used for analysis is The Linear Static Analysis
Module (STAR). This module uses the linear theory of structures, based on the
assumption of small displacements, to calculate structural deformation. STAR calls
the STRESS sub module to calculate stresses. The STRESS sub module calculates
element and nodal stresses for most elements based on the results from STAR. The
Flow Chart below represents the flow for Linear Static Analysis in the basic system.
54
Figure 3.1 shows the geometry and dimensions of the T-joint used in this
study. Figure 3.2 was shoes the localities of crack at the brace- chord intersection.
Tensile loading was applied to the end of the T brace and reacted at both ends of the
chord.
A starter notch for the cracks was introduced in the saddle region of the joints
by electric discharge machining at a distance of 2 mm from the weld root position.
6.1.2.2 Material
The material used in the experiments was a cold deformed 450 YS TMCP-
steel. The tubes were cold prestraining as it was manufactured from plates by
bending and butt-welding. The Youngs modulus was determined as 210 GPa.
In this dissertation, there are 3 types of geometric entities being used in the
modeling job. They are keypoints (PT), curves (CR) and also surfaces (SF).
Keypoints are the most primitive entity in GEOSTAR, and the lowest in hierarchy.
They are part of all other hierarchies. Keypoints can be created by snapping to an
active grid on plane, or by specifying coordinates for a point in space.
PT, 1,0,0,0
Key point 1was defined at the origin of the global
coordinate system.
PT, 2, 0.07625,0,0
PT, 3, 0, 0.07625,0
PT, 4,0,0,0.14925
PT, 5, 0, 0.14925,0
PT, 6, 0, 0,-0.14925
(See figure 6.1) Figure 6.1: Key points
56
CRARC, 1, 3,2,1,0.07625
An arc of radius (76.25mm) was defined and
located between key points 3 and 2 with
key point 1 towards its center of curvature
CRARC, 2, 5,4,1,0.14925
CRARC, 3, 5,6,1,0.14925
(See figure 6.2) Figure 6.2: Curves arcs
SFEXTR, 1, 1, 1, Z, 0.8385
A surface was generated by extruding curve 1 by 838.5mm in the z- direction
SFEXTR, 2, 2, 1, X, 0.760
SFEXTR, 3, 3, 1, X, 0.760
(See Figure 6.3)
PT,12,0,0.07625,0.1283292
CRARC,12,12,5,1,0.14925
PTONCR,12,0.032378743
A key point was defined and located on curve 12
PTONCR,12,0.033026318
PTONCR,12,0.064757486
(See figure 6.4) Figure 6.4: Key point on curves
57
CRARC, 13,12,13,1,0.14925
CRARC, 14,13,14,1,0.14925
CRARC, 15, 14, 15, 1, 0.14925
SF4CR, 4,26,16,19,17,0
A surface was defined where it was bounded
by the curves 26, 16, 19, and 17
SF4CR, 5,27,17,20,18,0
CRONSF, 19, 25, 2, 0.001,0
CRONSF, 17, 23, 2, 0.001,0
SF4CR, 6,24,31,26,30,0
CRONSF, 21, 27, 2, 0.001,0
Figure 6.7: Surface 4 curves
SF4CR, 7,25,30,27,32,0
(See figure 6.7)
SF4CR, 8,53,16,57,38,0
SF4CR, 9,57,19,54,40,0
SF4CR, 10,54,20,58,41,0
SF4CR, 11,39,58,18,55,0
SF4CR, 12,61,52,59,21,0
SF4CR, 13,24,59,49,56,0
SF4CR, 14,25,56,50,60,0
SF4CR, 15,60,47,55,32,0
(See figure 6.11)
Figure 6.11: Surfaces on crack region (2)
60
CRCOMPRESS, 1, 65
PTONCR, 8,0.024013159
PTONCR, 8,0.037171221
CRARC, 65, 15, 5, 1,0.14925
CRONSF, 23, 43, 2, 0.001,0
CRONSF, 27, 44, 2, 0.001,0
CRLINE, 68,5,43
CRLINE, 69, 43, 44
CRONSF, 23, 27, 2, 0.001,0
SF4CR, 18,68,65,64,66,0
SF4CR, 19,66,70,67,69,0 Figure 6.13: Surfaces on chord (1)
(See figure 6.13)
CRINTSS, 1, 2, 2, 1,0.001
PTONCR, 77, 0.149628061
PTONCR, 77, 0.190622051
PTONCR, 77, 0.23161604
PTCOMPRESS, 1, 50
PT, 51, 0.76,0,0
PT, 52,0,0,0.14925
PTCOMPRESS, 1, 51
PTONCR, 77, 0.149628061
CRARC, 78,12,51,4,0.14925
CRARC, 79,51,48,4,0.14925
CRARC, 80,48,49,4,0.14925
CRARC, 81,49,47,4,0.14925
CRONSF, 50, 16, 2, 0.001,0
CRONSF, 51, 16, 2, 0.001,0
CRONSF, 48, 18, 2, 0.001,0
CRONSF, 49, 20, 2, 0.001,0
CRONSF, 20, 45, 2, 0.001,0
CRARC, 85, 10, 45, 50, 0.14925
CRLINE, 86, 47, 10
SF3CR, 22, 61, 78, 82, 0
Figure 6.15: Surfaces between chord
SF4CR, 23, 18, 82, 79, 83, 0 and brace (1)
SF4CR, 24, 19, 83, 80, 84, 0
(See Figure 6.15)
PTONCR, 4, 0.152389127
PTONCR, 4, 0.194139573
PTONCR, 4, 0.235890019
PT, 56, 0, 0, 0.8385
CRLINE, 91, 7, 12
CRARC, 92, 7, 53, 56, 0.07625
CRARC, 93, 53, 54, 56, 0.07625
CRARC, 94, 54, 55, 56, 0.07625
CRARC, 95, 55, 8, 56, 0.07625
CRONSF, 50, 53, 1, 0.001, 0
CRONSF, 51, 53, 1, 0.001, 0
CRONSF, 48, 54, 1, 0.001, 0
CRONSF, 49, 55, 1, 0.001, 0
CRONSF, 47, 8, 1, 0.001, 0
SF4CR, 28, 78, 91, 92, 96, 0
SF4CR, 29, 79, 96, 93, 97, 0
SF4CR, 30, 80, 97, 94, 98, 0 Figure 6.16: Surfaces on brace (1)
6.1.4 Meshing
M_SF, 8, 11, 1, 4, 2, 2, 1, 1
Surfaces 8, 9, 10, and 11were meshed with 4-noded
elements considering 2 elements along each
major side of these surfaces.
ACTDMESH, SF, 1
The default meshing flag was activated
66
SFSYM, 1, 106, 1, Y, 1, 0
Additional surfaces and their meshes were created by the symmetric reflection of
surfaces 1 to 106 about the XZ plane.
Surfaces at the other side of the joint will need to be removed and remodel as rigid
joint without crack.
SFDEL, 164, 175, 1
SFDEL, 192, 195, 1
SFDEL, 139, 142, 1
SFDEL, 111, 122, 1
SFDEL, 123, 124, 1
SFDEL, 176, 177, 1
(See figure 6.24)
Figure 6.24: Delete surfaces without crack
A gap is left blank around the area where crack tip located, will be modeled
through the command NPTPUSH. This command pushes a node to a key point and
replaces the coordinates of the node with that of the key point and preserves the
The Tubular T-joint was being fixed at both ends. Therefore, all of the nodes
at these particular curves were restrainted from moving in all direction.
Due to the unequal length of each curve at the boundary of the brace end
where tension load was applied, a surface was created to carry the tension load in the
form of pressure load. Each node that coincided the boundary nodes was merged in
order to obtain a rigid connection.
The tension loading applied on the brace and were varied and there are shown
in Table 6.1.Area of tension loading at brace end, A = 0.018265416m2
Shell elements were being used to model the joints with the thickness of 20
mm.
EGROUP, 1, SHELL4, 2, 0, 0, 0, 0, 0, 0, 0
RCONST, 1, 1, 1, 6, 0.02, 0, 0, 0, 0, 0
PICK_MAT, 1, STEEL, SI
Initially the stiffness of the spring elements was set equal to the value of
Youngs modulus.
EGROUP, 2, SPRING, 0, 2, 0, 0, 0, 0, 0, 0
RCONST, 2, 2, 1, 1, 210E9
PICK_MAT, 2, STEEL, SI
74
Shell elements are used for tubular joints with surface cracks, incorporated with
line spring elements being embedded in to shell elements to stimulate cracks. In
this approach, crack geometry is actually not modeled and the line spring elements
are used to provide local flexibility of the two sides of a crack along edges of shell
elements.
Since the analysis was carried out in the linear elastic form, the R_STATIC
command was used. The A_STATIC command (See figure 6.34) specifies details of
the linear static analysis to be performed by the R_STATIC command.
R_STATIC
76
CHAPTER 7
7.1 Results
The finite element analysis were carried out using different values of applied
tension loads at the brace end in order to obtain the corresponding fracture mechanics
parameter of tubular joint i.e. the crack tip opening displacement, CTOD and crack
mouth opening displacement, CMOD with the applied load.
In each analysis, the tension load was converted into pressure load to ease the
analysis due to unequal length of the curves modeled. In each analysis, the value of
CTOD obtained was 3 mm from the actual crack tip, which is the nearest nodes to
the tip modeled. Meanwhile the CMOD values were obtained from the center line of
the surface crack.
Z3
Z1
Z2
Z4
Center Line
3mm
CTOD, mm
Different
2
Load, MN Pressure, N/m COSMOS/M Experimental (%)
0.25 1.3687E+07 0.006 0.0057 5.3
0.50 2.7374E+07 0.012 0.011 9.1
0.75 4.1061E+07 0.0175 0.017 2.9
1.00 5.4748E+07 0.023 0.024 4.2
1.25 6.8435E+07 0.029 0.031 6.5
1.50 8.2122E+07 0.035 0.037 5.4
2.00 1.0950E+08 0.047 0.043 9.3
Table 7.1 shows the results of CTOD obtained from COSMOS/M in this
project and also the corresponding experimental CTOD values taken from the
literature [2] at various load levels. It can be seen that there is a maximum difference
The CTOD results are then nicely plotted as shown in Figure 7.1. The best
line was drawn for each set of results. It can be seen that both lines are very close to
each other.
78
2.50
2.00
Load at Brace, MN
1.50
1.00
COSMOS/M
0.50
EXPERIMENTAL
0.00
0 0.01 0.02 0.03 0.04 0.05
Crack Tip Opening Displacement, mm
CMOD, mm
Pressure, Different
Load, MN N/m2 COSMOS/M Experimental (%)
0.25 1.3687E+07 0.016 0.025 36.0
0.50 2.7374E+07 0.033 0.050 34.0
0.75 4.1061E+07 0.05 0.080 37.5
1.00 5.4748E+07 0.066 0.100 34.0
1.25 6.8435E+07 0.082 0.120 31.7
1.50 8.2122E+07 0.098 0.150 34.7
2.00 1.0950E+08 0.131 0.200 34.5
79
The results of the CMOD at various load levels obtained using COSMOS/M
together with the corresponding experimental results taken from the literature [2] are
tabulated in the Table 7.2. Looking at both sets of results, it can be seen that there is
a maximum difference of 37.5 %
These results are then plotted as shown in Figure 7.2. The best line is drawn
for each of the COSMOS/M as well as for the experimental data. It can be seen that
as the applied load on the brace is increased, the CMOD results using the
COSMOS/M and experimental method can be seem to be deviate between each
other.
2.00
1.50
1.00
COSMOS/M
0.50
EXPERIMENTAL
0.00
0 0.05 0.1 0.15 0.2 0.25
Crack Mouth Opening Displacement, mm
The figure above (Figure 7.6 and Figure 7.7) shown that there is not much
deformed shape felt by the tubular T-joint when tension load was applied into the
brace end. The joints displaced as if it is being pull out from the chord. The Figure
7.7 shows the detail deformed shape of the crack region. It can be seen that the crack
propagate when load is applied.
81
7.2 Discussion
In the experiment tests reported in literature [2], the value of crack mouth
opening displacement, CMOD was measured. This was because the measurement of
the crack tip opening displacement, CTOD was difficult and virtually impossible to
be obtained in a routine test. Therefore, by measuring CMOD, the CTOD might be
inferred. However, in the present work using finite element analysis, it is possible to
measure the CTOD due to the capabilities of the software to tabulate a result at any
location on the model depending on the nodes generated. in this project, the CMOD
was measured as an additional comparison between the COSMOS/M and
experimental result.
There are some differences between the experimental result and the
COSMOS/M result. The biggest percentage of different between the experimental
data and COSMOS/M analysis were 9.3% for the CTOD reading and 37.5% for the
CMOD reading.
The difference might be due to the fact that in this project, the CTOD and the
CMOD obtained using the linear elastic analysis whereas the experimental results
were obtained with some plasticity occurring at the crack tip. The analyses were
carried out using one type of mesh only. The use of optimum meshing may generate
more accurate crack driving force, M.
The use if line spring element embedded into the shell elements to stimulate
the crack may not correctly model the actual crack like in the experiment. Since the
approach is simple and computationally inexpensive, it may be useful for some
simple component; it may not be accurate enough for more complex components like
cracked tubular joints.
82
CHAPTER 8
8.1 Conclusion
Finite element linear elastic was carried out using COSMOS/M software to
obtain the crack tip opening displacement and crack mouth opening displacement of
tubular T-joints.
A graph of load applied versus crack tip opening displacement, CTOD and
graph load applied versus crack mouth opening displacement, CMOD is plotted to
compare the result from the COSMOS/M with the experimental data from journal.
The comparison between the numerical method and experimental result is good for
the CTOD case whereas it is not so good for the CMOD case. For the CTOD case, a
maximum difference of 9.3% was obtained whereas for the CMOD case, the
maximum difference was 37.5%.
83
8.2 Recommendation
i. It is proposed that the actual crack is modeled on the chord at the junction
between the chord and the brace rather than using line spring elements.
ii. The welding part between the chord and the brace can also be taking into
consideration in order to generate a better result as the present work did not
take into consideration of welding effect.
LIST OF REFRENCES
(1) COSMOS/M User Guide Manual (Volume 1, Volume 2 & Volume 3) (1998),
COSMOS/M 2.0
(3) Marshal GW, Ainsworth RA. The fracture behaviour of a welded tubular
joint an ESIS TC1.3 round robin on failure assessment methods. Part II: R6
analysis. Engineering Fracture Mechanics 2002:69:1111-8.
(4) The fracture behaviour of a welded tubular joint An ESIS TC1-3 round
robin on failure assessment methods. Part III: UK BS7910 methodology.
(5) Zerst U, Primas R, Schindler H-J, Heerens J, Schwalbe K-H. The fracture
behaviour of a welded tubular joint an ESIS TC1.3 round robin on failure
assessment methods. Part IV: Application of the ETM 97/1. Engineering
Fracture Mechanics 2002:69:1129-48.
(6) Schindler HJ, Primas R, Veidt M. The fracture behaviour of a welded tubular
joint an ESIS TC1.3 round robin on failure assessment methods. Part V:
Screening method by required toughness and plastic stability considerations.
Engineering Fracture Mechanics 2002:69:1149-60.
(7) Zerst U, Miyata T. The fracture behaviour of a welded tubular joint an ESIS
TC1.3 round robin on failure assessment methods. Part VI: Application of
WES 2805-1997. Engineering Fracture Mechanics 2002:69:1161-9
(8) L.P.Pook (2000), Linear Elastic Fracture Mechanics for Engineers: Theory
and Application, WIT Press.
85
(9) David Broek (1986), Elementary Engineering Fracture Mechanics, 4th ed.,
Martinus Nijhoff Publishers.
(14) Cao JJ, Yang GJ, Parker JA, Burdekin FM. Crack modeling in FE analysis of
circular tubular joints. Journal Engineering Fracture Mechanics 1998:537-553