Lehigh University

Lehigh Preserve
Teses and Dissertations
1-1-2003
Welding simulation of ship structures using
coupled shell and solid volume fnite elements
Dongjin Kim
Lehigh University
Follow this and additional works at: htp://preserve.lehigh.edu/etd
Tis Tesis is brought to you for free and open access by Lehigh Preserve. It has been accepted for inclusion in Teses and Dissertations by an
authorized administrator of Lehigh Preserve. For more information, please contact hms207@lehigh.edu.
Recommended Citation
Kim, Dongjin, "Welding simulation of ship structures using coupled shell and solid volume fnite elements" (2003). Teses and
Dissertations. Paper 816.
Kim, Dongjin
Welding
Simulation of Ship
Structures Using
Coupled Shell and
Solid Volume
Finite Elements
January 2004
Welding Simulation of Ship Structures Using Coupled Shell
and Solid Volume Finite Elements
by
Dongjin Kim
A. Thesis
Presented to the Graduate and Research Committee
of Lehigh University
in Candidacy for the Degree of
Master of Science
in
Department of Mechanical Engineering and Mechanics
Lehigh University
December, 2003
Abstract
Table of Contents
1
2
2
4
5
Chapter 1- Introduction
1.1 Introduction to Welding Distortion
1.2 Introduction to Welding Heat Sources
1.3 Introduction to Simulation of Welding
1.4 Application of Research (Introduction to Research concerning ship hull
construction) 9
1.5 The objectives of present research 11
Chapter 2 - Comparison -between the coupled model with shell elements
and the 3D volume model of a fillet welded T-beam 12
2.1 Properties of material (AL-6XN) 12
2.1.1 Thermal properties 12
2.1.2 Mechanical properties 14
2.2 Model generation 16
2.3 Thermal Model (thermal source modeling) 19 -
2.4 Mechanical Model 21
2.5 Thermal Results 23
2.6 Mechanical Results 29
2.7 Conclusion about Comparison of both models 43
111
Chapter 3 - Comparison of coupled elements models with different boundary
conditions 45
3.1 Properties of material 45
3.2 Model generation 45
3.3 Thermal Model 46
3.4 Mechanical Model 46
3.5 Thermal Results 48
3.6 Mechanical Results 48
Chapter 4 - Welding simulation of a practical problem (the hull of ship) 66
4.1 Model generation 66
4.2 Thermal and Mechanical Model 67
4.3 Thermal Results 68
4.4 Mechanical Results 70
Chapter 5 - Conclusions 77
5.1 Future work 77
5.2 Conclusions 78
Appendix 79
HEAT.DAT 79
MECH1.DAT 83
Reference 87
Vita 89
IV
Abstract
The objective ofthis research is to develop an efficient approach to determine the
global deformations in ship structures resulting from thermal effects from the welding
process by using coupled shell and solid finite elements. The finite element method has
emerged as one of the most attractive approaches for computing residual stresses in
welded joints, but its application to practical analysis and design problems has been
hampered by computational difficulties that occur due to the enormous computational
size of any practical problem. These difficulties arose primarily in situations with three-
dimensional (3D) modeling of a welding process. Although two-dimensional (2D)
modeling has been used widely in residual stress problems, current studies have shown
that 2D analysis cannot render accurate residual stresses in many specific welding
problems. Therefore, it is most effective to use shell elements in conjunction with solid
elements. In the zone close to the heat affected zone, 3D modeling is repeated. However,
use of shell elements away from this region will decrease significantly the number of
elements in the welding model, reducing the computational size for the overall 3D model.
This study investigates the temperature, distortion and residual stress in afillet welded T-
joint, comparing those computed by the coupled elements of both volume and shell
elements, with those computed by volume elements only. In addition, the displacements in
welded T-joints with different constraints were compared. The practical problem of a
welded box beam used in ship hull design was simulated by using a coupled model of
shell and solid elements.
1
Chapter 1 - Introduction
1.1 Introduction to welding distortion
Residual deformations and stresses will be generated in a structure as a
consequence of the local plastic deformations introduced by the local temperature history
associated with welding, Le., rapid heating and subsequent cooling.
K. Masubuchi discussed the various types of welding-induced distortion and
residual stresses. This research resulted in a number of empirical relations and focused
entirely on welding distortion that remains after the completion of the weld and ignoring
the intermediate states. [1]
In a general welding problem, residual stresses are produced by plastic strains due
to tremendous thermal gradients, by material dilation during solid phase transformations,
and by plastic deformations caused by plastic strains and solid phase transformations. [2]
Near the weld pool, the temperature change due to welding is extremely rapid and the
temperature distribution uneven. In the region of the weld, the molten metal supports no
load and no strength of the solid, but high-temperature metal around the weld is
drastically reduced. As the temperature far from the weld is relatively low, the expansion
of metal near the weld is constrained and forced into high compression. Regions far from
the weld are forced into tension to balance the compressive stresses close to the weld.
When the part c<?ols, the material near the weld contracts and results are in high tension,
yielding occurs, while the regions far from the weld balance with compression.
Distortion due to welding can be divided with three categories. The first is
transverse shrinkage, or shrinkage perpendicular to the weld line. With this type of
2
distortion, the part often contracts uniformly along the weld. (See Figure I) The second
is longitudinal shrinkage parallel to the weld line, and the final category is angular
distortion about the weld line. With thin-walled structures, buckling is also an important
problem. [1]
----_.... _--_........
10) TRAmVERSE SHRINKAGE
---------
Ie) ROTATIONAl OISTORTION
10) LOIHIITUDINAL BENDING
DISTORTION
(bl ANGULAR CHAt/GE
hI) LONGIT\JDINAL SHRINKAGE
(I) BUCKLING DISTORTION
Figure 1 - Distortions in Welding
Until recently, most researches of weld distortion relations were empirical
because the analytical solution of welding distortion was too difficult to be practical.
While empirical relations based on experiments are useful for estimating distortion in
parts similar to those used for deriving the relations, these solutions are available only for
simple geometries. [3] All experimental methods have at least two disadvantages. First,
their application usually requires special equipment and personnel that are not usually
3
associated with welding research. Second, residual stresses can most often be measured at
discrete locations on a weld, usually close to the weld surface. A complete picture of the
residual stress distribution in a general weldment is practically impossible to obtain by
experimental techniques. [4]
1.2 Introduction to welding heat sources
Rosenthal had developed a solution for conduction from a moving heat source in
the late 1930s. This has been the most popular analytical method for calculating the
thermal history of welds. [5] However, as these models assume a point source and
therefore infinite temperature at the source, the model breaks down close to the weld
pool. One additional limitation of the Rosenthal solution, when applied to direct metal
deposition, is that it does not include any mass addition to the weld pool. [6]
To overcome most of these limitations, Pavelic et al [7] suggested a heat source
modeled with a Gaussian distribution of flux deposited on the surface of the workpiece in
1969. With this model, the concentration of the heat source can be varied by changing a
parameter called the concentration coefficient. Friedmen, Krutz, and Segerlind [8-9]
developed a variation of Pavelic's model that is expressed in coordinates that move with
the heat source. While these models are a significant improvement over Rosenthal's
model, it has been suggested that heat should be distributed to the molten zone to reflect
more accurately the digging action of arc. These models do not account for the rapid
transfer of heat throughout the fusion zone. To better represent high power density
sources, a hemispherical Gaussian distribution was developed. Unfortunately, this model
was still ill suited to deal with deep penetration welds that are not spherically symmetric.
4
[5] To account for this problem, Goldak, Chakravarti, and Bibby [10] proposed a
nonaxisymmetric three-dimensional heat source model. This model accommodates
shallow welds, deep welds, symmetrical welds, and asymmetrical welds, all of which
lead to more accurate models of the welding process. Thermal models of welding
processes have also improved, taking into account parameters such as weld torch width,
non-linearities due to variation of thermo-mechanical properties of material with
temperature, radiation heat transfer from the weld pool, temperature-dependent
convective heat transfer coefficients, and more. [2]
For the simulation of the arc welding process, a double ellipsoidal geometry of the
heat source is used in this research. This approach is numerically more stable and more
accurate than a point or line source, especially in the temperature range above 600C.
1.3 Introduction to simulation of welding
Interest in developing adequate analytical models of welding processes dates from
the late 1930s and 1940s. [11-12] To be sure, from the perspective of continuum
mechanics, the welding process can be viewed as transient boundary value problem. The
constitutive equations in this problem take into account the physics of heat transfer and
the mechanics of thermal dilatation, as well as the processes of change in material
microstructure and phase transformation. The boundary conditions model the welding
heat input, the surface heat losses, the mechanical restraints, and, most importantly, the
contact between the welded parts and the filler metal deposited. At a minimum, a
temperature-dependent elastic-plastic material model should be incorporated. Practical
solution of such a complex boundary value problem became possible only in the 1960s,
5
with the computer implementation of powerful numerical techniques, among which the
finite element method emerged as the most powerful. [13]
The computational demands of fully 3D welding models are so prohibitive that all
FE investigations of residual stresses in welds before the late 1980s were performed on
simplified two-dimensional (2D) models. In 2D models, only a plane perpendicular to the
direction of the weld is considered. The behavior out of plane can be taken as plane stress
(assuming zero out-of-plane stresses), plane strain (assuming zero out-of-plane strains),
generalized plane strain (assuming constant strain normal to the model plane), or
axisymmetric. [5] Argyris et. al. [14] computed the thermo-mechanical response using
2D models in a staggered solution strategy to combine and integrate the thermal and
mechanical computational steps. Rybicki et. al. [15] performed thermo-elasto-plastic
analysis on a 2D axisymmetric finite element model for a two-pass girth-butt welded pipe
problem, and verified the numerical results with the experimentally obtained temperature
history and residual stress distributions. Papaxoglu and Masubuchi [16] solved the
multipass GMAWprocess problem by performing uncoupled 2D heat transfer and stress-
strain analyses, incorporating the phase transformation strains.
Since investigators tried to avoid modeling in 3D, the computed residual stresses
were verified by comparison with experimental measurements. Certain discrepancies
between computed and experimentally measured residual stresses were reported. This led
to the belief that 2D models in certain situations are inadequate to quantify the residual
stresses accurately in a welded joint. [4] 2D models, as mentioned above, have been
particularly useful with their high efficiency and accuracy in determining the solution in
6
the analysis plane and reduced computational requirements. However, for welding
practices where tack welding or fixturing allow out-of-plane movement 2D analyses may
not be accurate. This seems to be particularly for distortion predictions. Furthermore,
longitudinal heat transfer, instability aspects and end effects (Le. due to initiation and
termination of the heat source) cannot be realized in two dimensional formulations. [17]
3D modeling of welds was first attempted by Tekriwal and Mazumder [18-19] and,
independently, by Karlsson and Josefson. [20] Their analyses confirmed the 3D nature of
the temperature and stress fields developed during welding, but at the same time
demonstrated the restrictions of the 3D calculations. The investigators were limited to
relatively coarse FE meshes to accommodate the analysis in the computer facilities
available to them. [4]
Most of the currently performed welding simulations, both 2D and 3D, are based
on small deformation theory and are limited to simpler structures and weld geometries or
focus only on the heat affected zone, ignoring the surrounding structure. [17]
Brown and Song [21] show that the interaction between the weld zone and the
structure can have a dramatic effect on the accumulated distortion in many cases, the
contribution of the structure dominates the state of distortion and stress, a state that is
much different from the on e predicted by a simulation of the weld zone alone.
As a full three-dimensional model is computationally expensive and unnecessary
In many temperature and stress calculations, Daniewicz [22] developed a hybrid
(experimental-numerical) approach that the weld joints are represented by "weld
elements" to simulate the shrinkage caused by welding, which is determined
7
experimentally. This approach does not deliver the desired accuracy due to the difficulty
in measuring weld shrinkage.
Michaleris et al developed a two-step numerical analysis technique for predicting
welding-induced distortion and assessing the structural integrity of large and complex
structures, that combines two-dimensional welding simulations with three-dimensional
structural analysis in a decoupled approach. [3-23] First, a two-dimensional welding
simulation is performed to determine the residual stress distribution. The model limited to
a portion of the structure that represented the mechanical restraints that were used. Then a
three dimensional structural (elastic) analysis performed using the stress distribution of
the welding simulations as. loading to determine if the structure would buckle and the
corresponding mode and/or magnitude of deformation. The advantage of a decoupled
approach is computational simplicity and efficiency. Complex 3D welding simulations
were not performed. It should be pointed out that Michaleris' approach has the difficulty
of applying the accurate weld load, obtained from 2D model. The difficulty is due to the
different mesh size, the limited region applied weld load to 3D model and the assumption
that residual stress distributions are generated by imposing a strain as load. Although this
method delivers reasonable results by using limited computer resources, a critical
buckling load can be only predicted using decoupled 2D welding simulation and 3D
eigenvalue buckling analyses. The effects of temperatures and distortions per time step
cannot be calculated over predicted and the temperatures around the weld pool, since the
two-dimensional model neglects conduction in the weld direction.
8
1.4 Application of Research (Introduction to Research concerning ship hull
construction)
Many ships of current construction are of the conventional hull type, i.e., basically
a single skin of steel plating stiffened orthogonally by stiffeners and transverse members.
The double hull is a relatively new development in the evolution of ship structural
systems. The fundamental difference from the conventional hull is that its twin skins of
steel plate are separated from each other by longitudinal girders that span between
transverse bulkheads along the length of the ship. The double hull offers some advantages
over conventional hulls, such as improved combat and collision resistance, fewer areas of
discontinuities and complex welded details, possibilities for automated fabrication
techniques, and simplified distribution systems.
Recent advances in steel making have resulted in the development of new steels
with improved material properties such as high yield strength, good weldability, good
ductility and high corrosion resistance. Application of AL-6XN non-magnetic
superaustenitic stainless steel in double hull ship structure could be essential for the
development of future double hull construction because of the characteristics that this
alloy begins to deform inelastically under relatively low stress level, but have more work
hardening capability and higher ultimate strength than many plain carbon steels. [24]
Most solutions concerned with ship hull fabrication are numerical, i.e., based
upon the finite element method. Because the determination of the ultimate limit state of
the overall hull girder is a complex problem involving large deflection, and elasto-plastic
behavior of the hull components. Much research has been carried out in developing a
simplified model due to the expense and time consuming process of full modeling. Lu
9
and Pang investigated the ultimate strength of a ship hull under axial compression
induced by vertical longitudinal bending without considering of strain hardening. [25]
Thereafter, Lu developed an analytical capability to predict the load-deformation
relationships and the ultimate strength of double hull cells under axial compression,
which incorporated strain hardening due to the highly ductile nature of the AL-6XN. [24]
For the investigation the effects of welding in ship construction, Dydo et al [26]
researched buckling distortion that is caused by compressive stresses between stiffeners.
2D thermal finite element models of the weld cross section in conjunction with a 2D
elastic-plastic mechanical finite element model were used to predict the longitudinal
stress induced along the weld. The equivalent axial load was then applied to the 3D
structural model and the Critical Buckling load was predicted.
In order to calculate the deformations per time step by local temperature history,
Ramasy [27] investigated predictive methods of assessing the effects of the welding
process using the finite element method to be used in welding simulation for ship
structures. This study provided information to help decide on mesh densities which are
feasible and sufficiently accurate.
Murugan [28] investigated the temperature and residual stress distributions per
time step in back step welding process to reduce the residual stresses in welded joints of
ship construction with 3D welding modeling of T-beam. However, these 3D welding
simulations were limited to the simple structure of T-beam and weld geometries and
focus primarily on the heat affected zone.
10
1.5 The objectives of present research
The prediction of residual stresses can be done on local models while the
computation of global deformations requires the modeling of the whole structure. The
mesh sizes involved with this kind of simulation are very important due the movement of
the heat source and the very high temperature gradient near the weldment. This may lead
to unreasonable CPU time. Due to this reason, the accurate analysis of global
deformation is currently lacking. In fact, comparisons between the coupled model with
shell and volume elements and the 3D model with only volume elements applied to the
same weld problem are practically nonexistent.
This study couples in the same analysis shell and volume elements. This method
enables one to simulate more efficiently the welding of thin structures as the solid volume
elements can be limited to an area close to the heat affected zone, with the rest of the
structure modeled using shell elements. Non-linear thermal and mechanical behavior is
available in shells. The computation is optimized in the sense that the additional degrees
of freedom associated with volume elements are largely eliminated. Compatibility
elements must be defined to transfer rotations between solid and shell elements. [29]
The purpose of this research is to develop a comprehensive 3D finite element
model and compute global deformation of the ship structure.
11
Chapter 2 - Comparison between the coupled model with shell elements
and the 3D volume model of a fillet welded T-beam
2.1 Properties of material (AL6XN)
AL6XN stainless steel (45Fe-25.7Ni-21Cr-6.3Mo) is one of the leading materials
being considered for Navy hull fabrication. Therefore, this material was used in the
following computations to compare results between shell and volume elements.
2.1.1 Thermal properties
Variation of thermal conductivity and volumetric specific heat with temperature
were considered in the thermal model. [24, 30]
Models including magnetohydrodynamic effects, thermo-solutal buoyancy
effects, and Marangoni or surface tension effects offer new insight into the formation of
the melt pool in welding. [31, 32] By artificially increasing the thermal conductivity
above the melting temperature, one can achieve a reasonable approximation for the
effects of convective mixing without much increase in complexity or solution times. To
compensate for weld pool convective heat transfer of AL-6XN material, a high
conductivity value, 160 W/m-K, was used in the weld pool. [33] Heat of fusion is the
energy required to change a solid at its melting temperature to liquid at the same
temperature. The release or absorption of latent heat of fusion was simulated by an
artificial increase in the value of specific heat over the melting temperature range. The
latent heat was taken as 2.1 x 10
9
J/m
3
; the melting range was 1320-1400C.
Thermal property data for AL6XN stainless steel used in the analysis are given in
Figure 2,3.
12
2000 1500 1000
Temperature (C)
500
0.18
0.16
0.14
~
~ 0.12
()
::1-
-cO 1
cEO.
8 E
Cii ~ 0.08
E-
Q; 0.06
.c
I-
0.04
0.02
0+----------,-----,--------,----------,
o
Figure 2 - Adjusted Thermal Conductivity of AL6XN
0.03
0.025
-

I
('/')
E 0.02
~
J
--
<t1 0.015
Q)
I
()
:E 0.01
()
Q)
0-
Cf) 0.005
2000 1500 1000
Temperature (C)
500
0+---------,------,-----------,---------,
o
Figure 3 - Adjusted Specific Heat of AL6XN
13
2.1.2 Mechanical properties
Temperature dependent values of properties such as modulus of elasticity,
Poisson's ratio, and yield strength, coefficient of thermal expansion (or thermal strain)
were provided to the model. The mechanical properties used in the study are given in
Figure 4 to 7.
When a material is deformed repeatedly, its mechanical properties may continue
to change. This behavior can be accounted for by using adapted strain hardening models
and isotropic strain hardening model was used in the analysis. Figure 7 obtained from
SYSWELD database approximately represents the behavior of AL6XN stainless steel.
250000
-
200000
..........
~
:g 150000
en
ctl
Q)
'0 100000
en
:::J
:::J
-g 50000
~
O + - - - - - - r - - - - - . - - - - . - - - - - - - , - - - - - . - - - . - - ~ . - . - - - - - ,
o 200 400 600 800 1000 1200 1400 1600
Temperature (C)
Figure 4 - Modulus of elasticity of AL6XN
14
1500 500 1000
Temperature (C)
50
O + - - - - - - - , - - - - - - - - - - - , r - - - - ~ - - - - - - ,
o
400
350
.-
8: 300
~
- 250
..c
..-
0>
C 200
~
..-
(J) 150
"C
. ~ 100
>-
Figure 5 - Yield Strength of AL6XN
0.03
400 600 800 1000 1200 1400
Temperature (C)
200
0-++""----,---,---...,-----,----,----,-----------,
o
0.025
c
'cu 0.02
....
..-
(J)
~ 0.015
E
....
Q)
~ 0.01
0.005
Figure 6- Thermal Strain of AL6XN
15
700
600
__ 500
co
~ 400
-
C/)
C/) 300
~
-+-'
C/) 200
--...
.----
.--..............
....... - ......
......... -
--+-20C
- -II- - 500C
1400C
0.2 0.15 0.1
Plastic Strain
0.05
100
O ~ ~ ~ ..... r - - ..........,.l---.......:-----------,
o
Figure 7- Temperature Dependent Strain Hardening
In a coupled shell-solid analysis using SYSWELD, there are many things
that are different from standard 3D analysis. Mechanical properties of shell elements
should give the thermal expansion coefficient instead of thermal strain in volume
elements. For the plastic part of the stress strain curve, volume elements give the values
of plastic strain and the difference between the stress at that strain and the yield stress
was provided to the model, while shell elements give the values of plastic strain and
stress at that strain. [34]
2.2 Model generation
AT-beam made of two plates of thickness 9 mm with a fillet weld between them
was taken up for the investigation. The width of the non-butting member of the T-beam
was 160mm and butting member was 122.5 mm.
16
Of the two fillet welds of T-beam
on either side of the butting member, first
a one sided fillet weld was considered.
The fillet width was 8 mm. The length of
the T-beam was taken to be 128 mm. The
size of the T-beam was chosen to be small
to reduce the computational time required
for the analysis. The sketch of the T-beam
taken for investigation is shown in Figure 8. Figure 8- Shape of Plates
For full volume model of T-beam, 8 noded linear hexahedron volumetric elements
(H8 elements) in SYSWELD were used to construct the geometry. 4 noded linear
quadrilateral elements (Q4 elements) were used on the surface to apply the convective
and radiative boundary conditions. The fillet weld bead that arises due to deposition of
metal was divided into finer meshes using P6 elements. The exposed surface of the weld
bead was divided using Q4 elements to apply the boundary conditions. The total number
of nodes in the model was 8606, and the number of elements in the mesh was 11704.
For shell-volume coupled model, 4 noded linear quadrilateral elements (Q4
elements) were used to apply shell elements with the thickness of 9mm. A convection
coefficient is defined for the two sides of the shell elements. In mechanical analysis, the
number of degrees of freedom for shell elements (ux, uy, uz, rx, ry, rz) and for three-
dimensional elements (ux, uy, uz) is different. In order to obtain compatibility between
the movement of the plates modeled with 3D and 2D elements, special elements called
17
"compatibility" or "transition elements" are added in the connection section area. A
relation is therefore created on the nodes of the 3D section to ensure that global
displacement remains perpendicular to the mean plane of the shell element to which if is
connected. The three-dimensional mesh at the connection with the shell elements has to
be composed of 2 layers corresponding to the half-thickness of the shell, to correctly
apply relations of compatibility. [35] The total number of nodes in the model was 8246,
and the number of elements in the mesh was 11158.
The meshed views of the solid T-beam and the mixed element model are shown in
Figure 9, 10 respectively.
Figure 9 - mesh of full-volume model
18
Figure 10- mesh of shell-volume coupled model
2.3 Thermal Model (thermal source modeling)
The heat transfer model best suited for arc-welding applications is the Goldak, or
double ellipsoid, source. (See Figure 11) Goldak's source corrects the Rosenthal model's
point source assumption by distributing power through a volume of specified size and
shape.
This size and shape is adjusted through a number of Gaussian parameters, each
independently controlling the width, forward length, rearward length, and depth of
19
heating. By manipulating these parameters, the heat source can be changed to reflect a
very wide variety of welding conditions.
Figure 11 . Goldak Source
The formulation of the Goldak model is shown in Equations 2.1 and 2.2. By using
different parameters for the front and rear ellipsoids, it is possible to specify an
asymmetric distribution of power. Here, fr and J,. are the fractions of power sent to the
front and rear ellipsoids, respectively. Parameters a, b, c/' and c;- determine the shapes of
the ellipsoids as shown in Figure 11. Finally, Qis the total power that enters the part,
given by Q =rjVI , where rj is the arc efficiency, V is the voltage, and I is the current.
[10]
(2.1)
(2.2)
20
Goldak's heat source model was used in this study. The code has a provision to
apply this heat source model to the welding problems. In the model, the heat source was
made to traverse along the length of T-beam at a 45 inclination to deposit the fillet weld.
Heat transfer to the ambient takes place by convection and radiation. Both
convection and radiation heat transfer were considered in the model. The convective heat
transfer coefficient of 15 W/m
2
-K was used. The emissivity value used in the analysis
was 0.5. To simulate material deposition during welding, the 'activation/deactivation'
function of SYSWELD was employed. In the thermal analysis, the elements were
activated a little in front of the heat source to avoid numerical problems. The thermal
analysis was carried out up to 501Os, till the T-beam cooled down to room temperature,
after welding.
The welding parameters are as follows: Voltage: 35 V, Current: 250 A,
Arc efficiency: 75 %
The case considered was continuous forward welding over the 128 mm full length
of T-beam. The weld speed was 5 mrn/s and the duration of weld deposition was 25.6s.
Cooling phase started at 25.6s and the analysis was completed after 501Os.
2.4 Mechanical Model
The transient temperature distribution file obtained from thermal analysis was
given as input to the mechanical model. The boundary conditions or restraints applied to
the T-beam during mechanical analysis are shown in Figure 12.
21
The points A, Band C of Figure 12 were used to estimate the progress in
proportion to displacement with weldment. The displacement between point A, a start
point of shell elements in the coupled model, and weldment is 29mm, point B is 73.5mm
and point Cis 118mm. The direction of arrows means the direction fixed.
Figure 12 Restraints of Model
As in thermal analysis, elements representing filler metal were activated in
mechanical analysis whenever required. In the mechanical analysis, the elements were
activated slightly behind the centre of the heat source. The stress analysis was carried out
up to 5010s as in the case of thermal analysis. At 5000s, the restraints of the T-beam were
reset to an unrestrained condition, resulting in "spring back" and redistribution of stresses
and deflection in the T-beam.
22
2.5 Thermal Results
Thermal results for both models are shown in Figure 13 to Figure 18. The
temperature and stress contours were obtained using the post processing module of
SYSWELD.
CONTOURS
Temp
Time 10
Comput .Ref G1
Min = 20
Max =1969.
-
_ 197.247
Il!!!iII 37 4.4 94
_ 551.741
_ 728.988
~ 906.236
,".:< I 1083.48
CJ 1260.73
c:J 1437.98
I';:""Y,I 1615.22
_ 1792.47
Figure 13 - Volume Model at 10 seconds
23
CONTOURS
Temp
Time 20
Comput .Ref 8li
Min = 20
Max = 1939.
-
_ 194.472
_ 368.944
_ 543.416
_ 717.888
. 892.361
~ 1066.83
c::::J
c:J 1241.3
c:J 1415.78
jc,,,,!,j 1590.25
_ 1764.72
Figure 14 - Volume Model at 20 secouds
CONTOURS
Temp
Time 250
Comput .Ref 81
Min = 2 3 . 1 3 ~
Max = 157.1
-
_ 35.3115
IIiliIiI 47.4 903
IIIII!I 59.6691
_71.8479
lEI 84.0268
I , , ~ j 96.2056
c::J 108.384
c::J 120.563
p)C",q 132.742
_ 144.921
Figure 15 - Volume Model at 250 seconds
24
CONTOURS
Temp
Time 10
CompuLRef GI
Min;; 20
Max;; 1964.
-
_ 196.741
11III 373.482
.. 550.223
11\III1 726.964
mDn 903.705
""""! 1080.45
c::J 1257.19
c::J 1433.93
""'"",,,1 1610.67
_ 1787.41
Figure 16- Coupled Model at 10 seconds
CONTOURS
Temp
Time 20
Comput.Ref Gil
Min;; 20
Max;; 1939.
-
_ 194.462
_ 368.923
_ 543.385
.. 717.847
~ 892.309
,.,1066.77
r:=J 1241.23
r:=J 1415.69
b i i ~ , . 1 1590.16
_ 1764.62
Figure 17- Coupled Model at 20 seconds.
25
CONTOURS
Temp
Time 250
Comput.Ref 81
Min = 23.37;
Max = 160.2
-
_ 35.8243
_ 48.2713
... 60.7183
.. 73.1653
JBil 85.6123
11 98.0593
c:J 110.506
c:J 122.953
..... 135.4
",,,,.,,,,1
_ 147.847
Figure 18- Coupled Model at 250 seconds
The results of temperature at point A, Band C of Figure 12 were compared in
order to estimate the progress in proportion to displacement with weldment. (Figures 19,
20,21)
In Figure 19, the peak point of temperature in 3D volume model is 133.5C at
130seconds and the peak point in coupled model is 130JoC at 140seconds. The peak
temperature of both models at A point is almost same, while the temperature of coupled
model decreases slowly compare to the temperature of volume model. This is due to the
difference ofheat transfer between shell and volume elements.
In Figure 20, the peak temperature in volume model is 51.0C at 516seconds and
the peak point in coupled model is 56.3C at 652seconds. The rate of difference is 9.4%
26
and the delayed time of peak point is 136seconds. The temperature at point B of coupled
model decreases slowly compared to the temperature of volume model at point A.
The tendency of temperature at point C is similar to that at point B (Figure 21).
Although the absolute values of temperature are small, the difference of temperature
between both models increases a little and the delayed time of peak point is also longer
than point B. The difference of temperature between volume model and coupled model at
peak point is 9.2C and that is large as compared to 5.2C of point B.
__: volume model
------: coupled model
----
--------
160
140
6120
o
..........
~ 100
:::J
+-'
~ 80
Q.)
0-
E 60
~
40
20
o
o 1000 2000 3000 4000
Time (5)
5000 6000
Figure 19- Temperature at Point A
27
__: volume model
------: coupled model
60
50
-.
0
0
'-'40
Q)
....
:::l
-
30
Q)
0..
E 20
~
10
0
0 1000 2000 3000 4000
Time (8)
5000 6000
Figure 20- Temperature at Point B
1000 2000 3000 4000 5000 6000
Time (8)
60
-.50
0
0
'-'40
~
:::l
-
30
Q)
0..
E 20
~
10
0
0
...........
r ,
I .... ,
I .....
I ' ....
I ' ....
I. .....
__: volume model
------: coupled model
........
.... -
.... -.................. -
Figure 21- Temperature at Point C
28
The reasons for the differences in temperature between the volume model and
coupled shell/solid model are
i) the difference in heat transfer between the shell and volume elements at the
connecting line (When shell elements are connected with volume element, the
middle node of both volume elements is connected with the node of shell
element.)
ii) the difference in heat loss through heat convection and radiation (In the
volume model, the heat loss can occur through edges as well as both of the
front and rear parts of plate, while the heat loss in the shell model occurs at
only through the front and rear faces of the plate.)
iii) the difference element in the meshes of shell and volume elements(size,
calculating method, averaging method of result, ...etc.).
2.6 Mechanical Result
The thermal results in the previous section were used to create the mechanical
models. The results for distortion are of more interest than the results of stresses at shell
elements, because the purpose of this research was to develop a comprehensive 3D finite
element model in order to compute global deformation of assembled ship structures. The
resulting displacements for both models are shown in Figures 22 to Figure 25. The
contours in these figures represent the displacements in the x-direction.
29
CONTOURS
Ux
Time 16
Deformed shape
Comput.Ref Glob
Min = -0.0205;
Max = 1.61834
-
_ 0.12842
11III 0.277411
_ 0.426403
.. 0.575395
BiiI 0.724386
~ ~ : ~ ~ ~ ~ ~ t J
::::J 1.17136
!'f'C":d 1.32035
_ 1.46934
Figure 22- Volume Model at 16 seconds (deformed shape x 10)
CONTOURS
Ux
Time 5010
Deformed shape
Comput.Ref Glob;
Min = -0.1739;
Max =4.39886
-
_ 0.241738
_ D.65745
_ 1.07316
_ 1.48887
~ 1.90459
' .. 'c,! 2.3203
:::::::J 2.736 01
:::::::J 3.151 72
3.56744
l"k,,.'H
_ 3.98315
Figure 23- Volume Model at 5010 seconds (deformed shape x 10)
30
CONTOURS
Ux
Time 16
Deformed shape _
Comput.Ref GlobE
Min =-0.04756
Max. = 1.61103
-
_ 0.103219
_ 0.254001
_ OA04782
.. 0.555564
liB 0.706345
I",-! 0.857127
c:::I 1.00791
c:::I 1.15869
1.30947
I",;;;,,:!
_ 1.46025
Figure 24- Coupled Model at 16 seconds (deformed shape x 10)
CONTOURS
Ux
Time 5010
Def Dr med sh ape _
CompuLRef GlobE
Min =-0.23156
Max =4.37173
-
_ 0.186921
.. 0.6D5401
.. 1.02388
.. 1.44236
~ 1.86D84
1:,<,1 2:27932
c::J 2.6978
c::J 3.11628
3.53476
he'"'''''\
_ 3.95324
Figure 25- Coupled Model at 5010 seconds (deformed shape x 10)
31
The results of distortion in the x-direction at point A, B, C were compared in
order to estimate transient distortion relative to displacement from weld fusion zone. (See
Figures 26, 27, 28, 29, 30, 31) The values of distortion of both volume model and
coupled model at each point per time step show a good agreement. The difference of
deflection at point A after being cooled is 0.06 mm (the distortion of volume model is
1. 16mm and that of coupled model is 1. lOmm), that at point B is 0.05 mm (the distortion
of volume model is 2.60mm and that of coupled model is 2.55mm) and that at point C is
0.04 mm (the distortion of volume model is 4.04mm and that of coupled model is
4.00mm). There is no difference of distortion relative to displacement from weld fusion
zone.
-- --------------------
__: volume model
------: coupled model
1.4
1.2
E
E
1
'-"
- 0.8 c
Q)
E
0.6
Q)
u
co
0.4
0-
CJ)
0
0.2
0
- 0.2
1000 2000 3000
Time (5)
4000 5000 6000
Figure 26- Deflection (ux) of Point A (0 s to 5010 s)
32
-------------- ",.,..",,--
1.4
1.2
E
E 1
'-'
-
a3 0.8
E
Q) 0.6
u
ro 0.4
c..
(/)
(5 0.2
o
- 0.2
20 40
Time (8)
__: volume model
------: coupled model
60 80
Figure 27- Deflection (ux) of Point A (0 s to 70 s)
__: volume model
------: coupled model
3
---
2.5
E
E
2
'-'
....
c
E-1.5
Q)
u
1
ro
c..
(/)
0.5
0
0
1000
- 0.5
2000 3000 4000
Time (8)
5000 6000
Figure 28- Deflection (ux) of Point B (0 s to 5010 s)
33
3
2.5
E
E
2
--
-
c
Q)
1.5
E
Q)
u
1
ro
Q.
(J)
0.5
0
0
-0.5
20 40
__: volume model
------: coupled model
60 80
Time (5)
Figure 29- Deflection (ux) of Point B (0 s to 70 s)
4.5
4
E 3.5
E 3
--
C 2.5
Q)
E 2
Q)
u 1.5
ro
Q. 1
(J)
o 0.5
o
- 0.5
1000 2000 3000
Time (5)
: volume model
------: coupled model
4000 5000 6000
Figure 30- Deflection (ux) of Point C (0 s to 5010 s)
34
4.5
4
E 3.5
E
3
.....-
-
c
2.5 Q)
E
2
Q)
c.>
1.5 co
0-
1
rn
0
0.5
0
-0.5
20 40
Time (5)
__: volume model
------: coupled model
60 80
Figure 31- Deflection (ux) of Point C (0 s to 70 s)
In the mechanical models of welding presented, longitudinal stresses (y-direction)
show the largest stress contours compared with those of other directions. The results of
longitudinal stresses for both models are compared and are shown in Figure 32 to Figure
35. At the case of the coupled model, the stress contours of shell and solid elements are
displayed separately because the stress results of shell elements can not be showed with
the solid elements in SYSWELD. In addition, the values of stress in the connected nodes
are the added values of both results from solid and shell elements.
35

Longitudinal
direction
CONTOURS
Sigma 22
Time 16
Comput .Ref 81
Min =
Max = 50 B.2
- -300
-
_ -233.333
.. -166.667
-100
-
tmJ -33.3333
33.3333
c:::J 100
c:::J 166.667
\;;,:" ,,,:! 233.333
_300
Figure 32- Volume Model at 16 seconds
CONTOURS
Sigma 22
Time 5010
CompuLRef 81
Min =
Max = 740.1
- -300
-
_ -233.333
.. -166.667
-1 DD
..
-33.3333
1,'1 33.3333
c:::J 100
c:::J 155.667
233.333

_300
Figure 33- Volume Model at 5010 seconds
36
CONTOURS
Sigma 22
Time 16
Comput.Ref 81
Min = - 3 2 5 . ~
Max =508.2
- -300
-
IBI -233.333
_ -166.667
-1 00
..
-33.3333
le.,1 33.3333
c:J 1DO
c:::J 166.667
h,W1,H 233.333
_300
Figure 34- Coupled Model at 16 seconds
37
CONTOURS
Sigma 22
Time 5D10
Comput.Ref 81
Min = - 2 5 4 . ~
Max = 740.1
--300
-
... -233.333
.. -166.667
-10 D
-
IiliiJ -33.3333
1'>,',;1 33.3333
c:::J 100
c:::J 166.667
!0:i,,,",,,! 233.333
_300
Figure 35- Coupled Model at 5010 seconds
38
The stresses in the volume elements obtained from the coupled model are almost
same with those obtained from the volume model entirely composed of solid volume
elements. The results of longitudinal stresses at points A, B, C were compared in order to
estimate the stress evolution in relation to displacement from the welded zone. (See
Figures 36, 37, 38, 39, 40, 41) The absolute values of longitudinal stresses at shell
elements are relatively small, when compared to equivalent stresses in the volume
elements of welded part. However, the values of stress for both the volume model and
coupled shell model at each point per time step show a little difference though the
tendencies of these stresses are similar.
100
50
__: volume model
------: coupled model
o
1000 2000 3000 4000 5000 6000
(J) - 50
(J)
~
U5 -100
-150
-200
,----------------------------
Time (8)
Figure 36- Longitudinal stress of Point A (0 s to 5010 s)
39
100
50
-- co
Q.
0
.........
en
en
(])
- 50
s...
.....
(f)
-100
-150
__: volume model
------: coupled model
110.
60 80 100 120
"
,
"
....
....
....
....
.... _-
....... _--
Time (5)
Figure 37- Longitudinal stress of Point A (0 s to 100 s)
---
---------
-----------
__: volume model
------: coupled model
50
40
-- 30
co
Q.
~
.........
en
20
en
(])
s...
.....
10
(f)
0
-10
1000 2000 3000
Time (5)
4000 5000 6000
Figure 38- Longitudinal stress of Point B (0 s to 5010 s)
40
-------
-------
--
__: volume model
------: coupled model
50
40
-ro
30
0.
~
'-"
(J) 20
(J)
Q)
s..
U5 10
0
-10
20 40 60 80 100 120
Time (5)
Figure 39- Longitudinal stress of Point B (0 s to 100 s)
6000 5000
__: volume model
------: coupled model
4000 2000 3000
----------
----
--
-- ---
,.
1000
30
25
20
'"
15 //
ro I
0. 10 I
~ I
'-" 5 I
~ 0 +-1;.-1-------,,---,------,--------,-------,r-------,
~
U5 -5
-10
-15
-20
-25
Time (5)
Figure 40- Longitudinal stress of Point C (0 s to 5010 s)
41
__: volume model
------: coupled model
100 120 80 60 40 20
'I
\
.....
'" \
\
\
\
\
I
\
\
I
I
I
\
I
\ ,-"",,'"
------
o
-2
-4
......... - 6
co
0.. _ 8
~
--
Ul - 10
Ul
~ -12
C15
-14
- 16
- 18
-20
Time (8)
Figure 40- Longitudinal stress of Point C (0 s to 100 s)
The difference of residual stress at point A after cooling is 36.6Mpa (the residual
stress of volume model is -145.2Mpa and that of coupled model is -108.6Mpa), that at
point B is 3.0Mpa (the residual stress of volume model is -1.2Mpa and that of coupled
model is 1.8Mpa) and that at point C is 7Mpa (the residual stress of volume model is
18.9Mpa and that of coupled model is 25.9Mpa). Although the values of the nodes in the
connecting region show the difference of 36.6Mpa, this difference is small, when
compared to the stresses in the volume elements of welded part. (See Figure 41)
In Figure 41, the longitudinal stresses after cooling according to displacement
from the weld fusion zone are presented. The residual stresses of both models show the
similar results.
42
Point C
__: volume model
------: coupled model
Point B
1
20 100 80
600
500
400
300
.-....
co
Q.
200
--
en
Point A
en
100
~
-
en
0
-100
-200
-300
Displacement from weld fusion zone (mm)
Figure 41- Longitudinal stresses in relation to displacement from welded zone
The most different value within the results of residual stresses shows in the
connecting region between the shell and solid elements.
2.7 Conclusion about Comparison of both models
The results of the coupled model with shell and volume elements and the 3D
model with only volume elements applied to the practical welding problem of T-beam
were compared in this chapter.
The Followings are the major observations of the present comparison:
43
i) The difference in temperature between both models slightly increases and the
delayed time to reach the peak temperature is also taken longer in proportion
to displacement from weld fusion zone. The temperature for shell elements
cools down slowly. This difference, however, can be ignored because the
absolute values of temperature at shell elements are very small.
ii) The values of deformation at coupled model are almost same with those at
volume model. Therefore, this coupled model of shell and volume elements
can be considered as an effective method to estimate the global distortion of a
large structure such as ship construction.
iii) The values of residual stress for both the volume model and coupled shell
model show a little difference though the tendencies of these stresses are
similar. Especially, the temperature at connecting region should be considered
carefully. This difference also can be ignored due to the fact that the absolute
values of stresses at shell elements are relatively small.
44
Chapter 3 - Comparison of the coupled elements models with different
boundary conditions
3.1 Properties of material
AL6XN stainless steel is used in this chapter and the same thermal and
mechanical properties of chapter 2 are applied.
3.2 Model generation
The vertical height of the butting member of the T-beam is 78mm. All dimensions
of T-beam are same with what was used in chapter 2 except the vertical height of the
butting member. Both sides fillet welding was considered.
The total number of nodes in the model was 8201, and the number of elements in
the mesh was 11116. The meshed views of the T-beam are shown in Figure 42.
Figure 42 - mesh of model
45
3.3 Thermal source modeling
The double ellipsoid source that is the same source with chapter 2 was applied
and all of the parameters in this heat source were same but both sides fillet weld.
After the first weld deposition time (25.6s), a cooling time of 60s was allowed
before the second fillet weld deposition was started on the other side of the vertical plate.
Thus, the two welds art not deposited simultaneously. The thermal analysis was carried
out up to 50lOs.
3.4 Mechanical Model
The boundary conditions applied to the T-beam during mechanical analysis are
shown in Figures 43, 44, 45 separately. The BC I is the same BC in previous chapter, and
the z-direction displacement is fixed. In the BC IT, the x-direction displacement of both
edges in butting plate additionally. The BC III constrained the two end points of the
middle line at non-butting member in order to make the displacement of four edges in
this plate free. The direction of arrows means the direction fixed.
The points A, Band C in Figure 44 and lines DD', EE', FF' in Figure 45 were
used to compare the results for each boundary condition. The displacement between point
A and weld fusion zone is 11.5mm, point B is 29mm, and point C is 73.5mm and the
lines are in the middle part of T-beam.
46
Figure 43 - Boundary Conditiou (Be) I ofModel
Figure 44 - Be II of Model
47
.
.
48
3.5 Thermal Results
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
FF'
DD'
At 5000
s
, the restraints of the T-beam were remove<! using the 'release' option of
EE' .-
Figure 45 - Be III of Model
Thermal results for the three models are identical, and are shown in Figure 46,
the SYSWELD code
Figure 47 and Figure 48.
CONTOURS
Temp
Time 16
CompuLRef 81
Min =20
Max = 19D9
-
_ 191.767
.. 363.533
_ 535.3
_ 707.066
IPD 878.833
I'".! 105D.6
c=I 1222.37
c=I 1394.13
hi';"'" 1565.9
_ 1737.67
Figure 46 - Contour of temperature at 16 seconds
CONTOURS
Temp
Time 95.4999
Comput.Ref 81
Min = 20.73
Max =2039
-
_ 204.275
_ 387.819
.. 571.363
.. 754.907
ram 938.451
",.,,1
112
1.99
c=I 1305.54
c=I 1489.08
p"",.:"'1 1672.63
_ 1856.17
Figure 47 - Contour of temperature at 95 seconds
49
CONTOURS
Temp
Time 200
CompuLRef 811
Min = 47.36E
Max = 367.9
-
_ 76.5137
_105.661
_ 134.808
.. 163.955
~ 193.1 02
222.25
~ c" 'c.' ,:1
:::::J 251.397
:::::J 280.544
t;.>;je;\'d 3D 9.691
_ 338.838
Figure 48 - Contour of temperature at 200 seconds
The results of temperature at point A, Band C of Figure 43 per time step were
displayed in order to estimate the temperature evolution in relation to displacement with
welded zone. (See Figure 49)
In Figure 48, the temperatures of point A, the point closest to the weld fusion
zone, shows the highest values and two peak points influenced by both sides welding
having started at different times. The temperatures of points B and C hardly show the
influence of both sides being welded with a time delay between depositions of the
separate :fillet welds, but the values of temperature are higher than those of one sided
weld in previous chapter.
50
1200 1000
__: Point A
___: PointB
------: Point C
."
400 600 800
Time (5)
,
,
."
200
;-'-- ......
( ......
/ .......
I
I
("
I
I
,
" f .,'
....
o
500
450
_400
o
~ 3 5 0
~ 300
::J
+J
~ 250
Q)
0.200
E
~ 150
100
50
o
Figure 49 - Temperature at Point A, B, C (Os to 1000s)
3.6 Mechanical Results
The thermal results were used to create mechanical models with different
constraints. The results of displacement for the three models are shown in Figure 50 to
Figure 56 and the contours of result represent the normal displacement, the magnitude of
the displacement vector.
51
CONTOURS
norm U
Time 16
Deformed shape X
Comput.Ref Global
Min = 4.88142e-
Max = 2.53303
-
_ 0.230276
IIIiI 0.460551
_ 0.690827
.. 0.921103
mDI 1.15138
1.38165
81.61193
c:J 1.84221
,,,,,"','" 2.07248
_ 2.30276
Figure 50- Normal displacement ofBC I at 16 seconds (deformed shape x 10)
CONTOURS
norm U
Time 101
Deformed shape X
Comput.Ref Global
Min = 6.19116e-
Max = 4.05585
-
_ 0.368713
6lIIDI 0.737427
_ 1.10614
Ii!IIIIiI 1. 47485
1.84357
1,",,12.21228
c:J 2.58099
c:J 2.94971
3.31842
_ 3.68713
Figure 51- Normal displacement ofBC I at 101 seconds
52
CONTOURS
norm U
Time 5010
Deformed shape
Comput.Ref G l o b ~
Min = 0.037353
Max = 4.37212
-
_ 0.431423
_ 0.825493
.. 1.21956
_1.61363
lEiI'fil 2.0077
c:::::J 2.40177
c:::::J 2.79584
c:::::J 3.18991
1-;,-,,01 3.58398
_ 3.97805
Figure 52- Normal displacement ofBC I at 5010 seconds
CONTOURS
norm U
Time 99.4999
Deformed shape X
Comput .Ref Global
Min = 2.87276e-
Max = 3.34659
-
_ 0.304235
_ 0.608471
... 0.912706
.. 1.21694
lei 1.52118
1,',>1 1.82541
c:::J 2.12965
c:::J 2.43388
I.";,,,,! 2.73812
_ 3.04235
Figure 53- Normal displacement ofBCnat 99.5 seconds
53
CONTOURS
norm U
Time 50] 0
Deformed shape
Comput.Ref Globe
Min = 0.033392
Max =3.64998
-
_ 0.362173
IlIIII 0.690954
_ 1.01973
_1.34851
1.6773
mlmI 2.00608
"".>1
I:=J 2.33486
I:=J 2.66364
2.99242

_ 3.3212
Figure 54- Normal displacement ofBC n at 5010 seconds
CONTOURS
norm U
Time 99.4999
Deformed shape X
Comput.Ref Global
Min =2.42872e-
Max =3.9125
-
_ 0.355682
I11III 0.711363
.. 1.06705
_ 1.42273
Il!I1l!il 1. 77B41
1",,1 2.13409
I:=J 2.48977
I:=J 2.84545
3.20114

_ 3.55682
Figure 55- Normal displacement ofBCill at 99.5 seconds
54
CONTOURS
norm U
Time 5010
Deformed shape X
Comput.Ref Global
Min = 4.21491e-
Max =3.23099
-
_ 0.293755
IlII!IIIl 0.587487
_ 0.88121
.. 1.17493
!ImJ 1.46865
1,.,:,1 1.76238
c=J 2.0561
c=J 2.34982
2.64354
1'",'.:,,1]
_ 2.93727
Figure 56- Normal displacement ofBC mat 5010 seconds
In order to compare the distortion of T-beam, the angular change and longitudinal
bending distortion was tabulated.
The results of displacement after being cooled are shown in Table 3-1. The
angular change a, (3, y and longitudinal bending distortion d in Figure 57 are the degrees
and displacement ofline DD', EE', FF' in Figure 44.
55
(a) Angular change
--..
Transverse
direction
________~ " \ 0
------ I - - - - - - - - - ~
'Y
\)-1
I
I
I
I
I
I
I
I
I
+
.r--
a -----
~ - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - l - ~ - - - - - - - - - - - - - - - - - -
--..
Longitudinal
direction
(b) Longitudinal bending distortion
Figure 57-Distortion in Welding
Boundary
Distortion
Condition
~
d a y
I 1.88 1.91 0.18 0.065mm
II 1.50 2.05 -0.41 0.064mm
III 2.00 1.82 0.07 0.062mm
Table3-1 - Distortion after removing the restraints
The final results of distortions in the cross-section after removing the restraints on
the T-beam are slightly different depending on the boundary conditions, while the
56
longitudinal bending distortions have almost same values. Especially, the final result of
the model with BC II is different as compared to BC I and III. (See Figures 58, 59, 60, 61,
62) The interesting part of the model with BC IT is that the displacement shows a large
change of values at the instant that restraint is released. This is because the BC I, III
basically allowed the free-movement of vertical plate while the constraint of BC II was
fixing the distortion of model during welding and cooling. When the constraint was
removed, the suppressed distortion occurred abruptly.
After the first welding and 60 seconds cooling, the angle changes of BC I, II are
very different with those of BC III due to the fact that both of BC I and II constraint the
four comers of non-butting member and BCITI does the two end points of the middle line
at non-butting member. The angle changes of three models, however, are converged to
similar values when the other side welding is finished. Although the difference of
distortion reduced after welding of both side, the results were varied greatly by boundary
condition.
57
2.5
2
....................................................................................................
ID 1.5
~
e>
Q) 1
"'0
--
o
0.5
----------------------------1
_: B.C. I
___: B.C. II
------: B.C. III
o
- 0.5
1000 2000 3000 4000 5000 6000
Time (s)
Figure 58- a Angle change (0 s to 5010 s)
250 200
_:B.C.I
: B.c. II
------: B.C. III
150
.........................................
,
.1' '/-------------
100
I ..
,
.
50
. r--------
: I.
,
,
.
,
,
,
.
2.5
2
-
1.5
Q)
Q)
l-
e>
Q)
1
"'0
--
0
0.5
0
-0.5
Time (s)
Figure 59- a Angle change (0 s to 200 s)
58
-. ..
_: B.c. I
___: B.C. II
------: B.C. III
2.5
2
.-
1.5
Q)
Q)
l-
e>
1
Q)
"0
--
C!l.
0.5
0
-0.5
1000 2000 3000
Time (5)
4000 5000 6000
Figure 60- PAngle change (0 s to 5010 s)
,
.
'I :
-----'1 :
,..--------/ "
I :
r :
.
2.5
2
.-
1.5
Q)
Q)
l-
e>
Q)
1
"0
--
C!l.
0.5
0
-0.5
.. .. .. .. .. .. .. .. .. .. .. .. .. ..
50 100
Time (5)
150
_:B.C.I
___: B.C. II
------: B.C. III
200 250
Figure 61- PAngle change (0 s to 200 s)
59
2
1.5
.......
1
Q)
~
0>
0.5
Q)
"0
-
>-
0
- 0.5
- 1
2
: B.C. I
___: B.C. II
------: B.C. III
.... ~ ..
I ~ - - - - - - - - - - - - - - - - - - - - - - - - - - - - - I
l' 2000 4000! 6000
Time (8)
Figure 62- y Angle change (0 s to 5010 s)
..
1.5
.......
Q)
~ 1
0>
Q)
"0
;: 0.5
o
- 0.5
................ -_ ..
#,'/ "". ...... - - - - - - - - -
50
: B.C. I
___:B.c. II
------: B.C. III
......................................
1O ~ - - - - -1-5e- - - - -ZOO
Time (8)
250
Figure 63- y Angle change (0 s to 200 s)
60
The results of longitudinal stresses that have the largest stress values at three
models are shown in Figure 64 to Figure 66.
CONTOURS
Sigma 22
Time 5010
Comput.Ref 81
Min = -225.:
Max = 714.7
- -300
-
.. -233.333
.. -166.667
-100
.-
1M! -33.3333
he,.';:;! 33.3333
c::J 100
c::J 166.667
233.333
"';,'1,,,{1
_30D
Figure 64- Longitudinal stresses ofBC I at 5010seconds
61
CONTOURS
Sigma 22
Time 5010
Comput.Ref 81
Min =
Max = 715J!
- -300
-
_ -233.333
_ -166.667
-} 00
..
IleI -33.3333
I.",'--.! 33.3333
c::J 100
c::::I 166.667
233.333
_3DO
Figure 65- Longitudinal stresses ofBC II at 5010seconds
62
CONTOURS
Sigma 22
Time 5010
CompuLRef 81
Min = -231.-
Max =
- -300
-
Il\'IIl1I -233.333
_ -166.667
-100
-
-33.3333
1",1 33.3333
c::::J 100
c::=:J 166.667
233. 333
_300
Figure 66- Longitudinal stresses ofBe ill at 5010seconds
The results of longitudinal stresses at point A, Band CofFigure 43 per time step
were displayed at Figures 67, 68, 69.
63
6000
_:B.C.I
___: B.C. II
------: B.C. III
4000 5000
-------
----
- --'--""--......

200
150
100
50
-
[ 0

';;; -50
en
-100 t
.....
(J)
-150
-200
-250
-300
Time (5)
Figure 67- Longitudinal stress of Point A (0 s to 5010 s)
40
20
0
-
-20
eel
Q.

-- -40
en
en
CD
....
.....
-60 (J)
-80
-100
-120
1000 2000 3000 4000 5000
-------

_:B.C.I
___: B.C. II
------: B.C. III
6000
Time (5)
Figure 68- Longitudinal stress of Point B (0 s to 5010 s)
64
200
150
100
--
: B.C. I
---
: B.C. II
- 50 ------: B.C. III eu
Q.
~
- 0
en
en
3000 4000 5000 6000
Q)
....
.....
-50
CI)
-100
-150
-200
Time (s)
Figure 69- Longitudinal stress of Point C (0 s to 5010 s)
The results of longitudinal stresses with different constraints are similar, though the
results of BC II show a bit of difference due to the restraints of the vertical plate.
Therefore, it could be analogized that the most influenced factor of stresses at welding
simulation is the contours of temperatures per time steps and the boundary conditions
applied in this chapter did not affect so much.
65
Chapter 4 - Welding simulation of a practical problem (the hull of ship)
4.1 Model generation
A single cell section of a double hull welded box beam was simulated. This cell
with a cross section shape of a cellular box is fabricated using AL-6XN steel plates. All
dimensions of the cell were based on a prototype double hull section. The dimensions are
shown in Figure 70.
26 1/16"
T
5/16"
I I
"11III
,.
"IIIl
,.
5/16" 5/16"
..... ~ ... ..
...
r
4
"
... ... ll'
I I
.....I ~ ~ .. 25_3/4_"-----.h+- 5/16"
~ 273/4" ~
Figure 70- Cross section of box cell
66
The length of box cell used in the finite element models is 3 feet 2 and 3/4 inches.
It is believed that this length is sufficient to obtain satisfactory results that could estimate
a states of distortion for longer beams and still allow reasonable computation times and
storage.
One-quarter of this cell was meshed due to the assumption of symmetry
conditions. This assumption implies that the welding proceeds on all four edges
simultaneously. The total number of nodes in the model was 10231, and the number of
elements in the mesh was 15868. The meshed views of this model are shown in Figure
69.
4.2 Thermal and Mechanical modeling
The applied heat source is a double ellipsoid source and both sides on the vertical
plate arc fillet welded. In the model, the heat source was made to traverse along the
length of T-beam at a 45 inclination to deposit the fillet weld metal. Some of the
welding parameters are given in the Table 4-1.
Convective heat transfer coefficient 15 W/m
2
-K
Emissivity 0.5
Voltage 35 V
Welding parameters Current 250 A
Arc efficiency 75 %
Table4-1 - Distortion after removing the restraints
After the first weld deposition to 197s, the model was cooled for 5000s and the
second weld deposition of the other side was started. At 5000s, the restraints of the T-
67
beam were reset to an unrestrained condition. The thermal analysis was carried out up to
SOlOs. The boundary conditions applied to this cell box during the mechanical analysis
are shown in Figure 71.
Figure 71 - Boundary Condition of Cell box (114 Model)
4.3 Thermal Results
Thermal results for this model are shown in Figure 72, Figure 73 and Figure 74.
68
CONTOURS
Temp
Time 50.0999
Comput.Ref Glo
Min 20
Max = 277BJ
-20
-
_ 173.333
.. 326.667
_4BO
I&lI 633.333
7B6.667
C:::t 940
c=J 1093.33
1246.67
=1400
Figure 72 - Contour oftemperature at 50.1 seconds
CONTOURS
Temp
Time 150.001
Comput.Ref Glo
Min = 20
Max = 2B36.
-20
-
.. 173.333
_ 326.667
.. 4BO
IEi 633.333
1'_'1 7B6.687
0
940
CJ 1093.33
1248.87

_1400
Figure 73 - Contour of temperature at 150 seconds
69
CONTOURS
Temp
Time 1000
Comput.Ref Gle
Min = 20.0B3
Max = 118.51
-
_ 29.030B
11II37.9781
_ 48.9254
.. 55.8728
~ 64.B201
1,,;c',i ( 73.7 875
r=J B2.714B
r=J 91.8822
100.809
=109.557
Figure 74 - Contour of temperature at 1000 seconds
4.4 Mechanical Results
The results of displacement for the symmetric cell are shown in Figure 75 to
Figure 77 and the contours of result represent the magnitude ofthe displacement vector.
70
CONTOURS
norm U
Time 3D
Deformed shape
Comput .Ref Glob
Min::; 0.0 00 9 9 ~
Max::; 0.79001
=0.0727219
IiIII 0.144451
_ 0.216181
.. 0.28791
m!fii 0.35964
0.431369
~ 0.503098
r::::::J 0.574828
0.646557
,,",*:;,4
_ 0.718287
Figure 75 - Normal displacement at 30 seconds
CONTOURS
norm U
Time 102.6
Deformed 9hape
Comput.Ref G l o b ~
Min::; 0.002112
Max::; 1.68112
-
_ 0.15475
IIlII 0.307387
_ 0.460024
.. 0.612661
~ 0.765298
le,.1 0.917935
r::::::J 1. 07057
r::::::J 1. 22321
1.37585
1:,,";"-<1
_ 1.52848
Figure 76 - Normal displacement at 102.6 seconds
71
CONTOURS
norm U
Tlme5D1D
Deformed shape
CompuLRef Glob
Min = D.D 2273:
Max =9.1389!5
-
_ D.85148
_ 1.68D23
_ 2.5D897
.. 3.33772
~ 4.16647
1'CO '.:"" 1 4.99521
r:::::I 5.82396
CJ 6.65271
7.48145
1 , ; ~ w : " , 1
_ 8.31D2
Figure 77 - Normal displacement at 5010 seconds
The results oflongitudinal stresses that have the largest stress values for this
model are shown in Figure 78 to Figure 80.
In Figure 80, high tensile residual stresses are produced in areas near the weld
and the longitudinal stresses in areas away from the weld are compressive.
72
Figure 78 - Longitudinal stress at 30 seconds
73
CONTOURS
Sigma 22
Time 3D
Comput .Ref 81
Min = -26 D.!
Max = 3 8 7 . ~
- -300
-
_ -233.333
.. -166.667
-1 DD
-
IBiJ -33.3333
1"'",,,1 33.3333
t:=I
1DD
t:=I 166.667
!;'W\i@ ~ ~ ~ . 3 3 3
-
Figure 79 - Longitudinal stress at 102.6 seconds
74
CONTOURS
Sigma 22
Time ID2.6
Comput.Ref 13:
Min = -279.
Max = 433.i
- -3DD
-
... -233.333
_ -166.667
-IDD
-
ftBlI -33.3333
1.'.'::;,:1 33.3333
c:::J 1DD
c:::J 166.667
233.333
_3DD
CONTOURS
Sigma 22
Time 5010
Comput.Ref 81
Min =
Max = 639.4
- -300
-
_ -233.333
.. -166.667
-100
.-
imJ -33.3333
33.3333
""';'S""!
::=lIDO
:::J 166.667
233.333
_300
Figure 80 - Longitudinal stress at 5010 seconds
In Figure 81, the residual stresses after cooling according to displacement from the
weld fusion zone are presented.
75
600
500
400
-
~ 300
-
en
en
200 Q)
....
+-'
C/)
100
0
-100
50 100 150 200 250 300 350
Displacement from weld fusion zone (mm)
Figure 81- Longitudinal stresses in relation to displacement from welded zone
The boundary condition that was used in this chapter may not precisely replicate
the state of stress in the "actual" problem due to the fact of that one-quarter symmetry of
a cell box is assumed. However, it is expected that this will provide a reasonable estimate
of the final distortion. In addition, the model presented here are too small to adequately
simulate the complete behavior of an entire hull structure. These results, however, show
that a practical problem in welded ship construction can be simulated by using the
coupled model of shell and volume elements.
76
Chapter 5 - Conclusions and Future Work
5.1 Future Work
The greatest limitations in modeling a large structure are computation time and
storage requirements. The advantage of plate and shell elements is that the elements can
be fairly coarse and still deliver reliable results. Therefore, to use solid elements in
regions of high thermal gradients and plates and shells elsewhere offers a significant
reduction in solution times and storage requirements without a painful loss of accuracy.
In the simulations considered here, only simple models such as T-beam and single
cellular box beam of the double hull were considered. Eventually, an entire hull ship,
with more complicated geometry, should be simulated using this technique. (See Figure
82)
Figure 82- Design of double hull in ship
77
When a three-dimensional mesh is generated to accurately simulate welding, the
entire region that will be welded must be very refined. Therefore, any model that has a
large welding region presents enormous computational problems in terms of CPU time
and storage.
5.2 Conclusions
The temperature, distortion and residual stresses of the coupled model with shell
and volume elements and the 3D model with only volume elements in a welded T-joint
were investigated in this study. The all results for a coupled shell/solid volume model
showed the same tendencies and similar values obtained from an equivalent solid volume
model. Therefore, a coupled model can be considered as an effective method to estimate
the global distortion of a large structure, such as ship double hull structures.
Finally, models of both sides welded T-joints with different constraints were
simulated and compared. According to the boundary conditions, the results of distortions
varied greatly and the stress concentration occurred at the place that was fixed. It is
important to recognize that boundary condition constraints play a critical role in
determining distortion for welded structures.
78
Appendix
Listed here are the main input files for asingle cell section of adouble hull welded box
beam simulation discussed the chapter 4. The second mechanical input file for solving
the released constraints is not presented because this file is almost same with the first
mechanical file except the constraints part.
HEAT.DAT (Thermal model data)
NAME
SEARCH DATA 101
COMPATIBILITY PENALTY 200*6
MIXING SOLID SHELL
SOLID ELEMENT GROUP $PART$
SHELL ELEMENT GROUP $SHELL$
RETURN
DEFINITION
T-JOINT CONTINUOUS WELDING
OPTION THERMAL METALLURGY SPATIAL
RESTART GEOMETRY
MATERIAL PROPERTIES
ELEMENTS GROUP $PART$ / C=-10001 KX=-10002 KY=-10002 --
KZ=-10002 RHO=l MATE=l
ELEMENTS GROUP $BEAD1$ / C=-10001 KX=-10002 KY=-10002
KZ=-10002 RHO=l STATE=-5 MATE=l
ELEMENTS GROUP $SHELL$ / C=-10001 KX=-10002 KY=-10002
KZ=-10002 RHO=l MATE=l H=7.9375
CONSTRAINTS
ELEMENTS GROUP $SKINPART$ / KT=l VARIABLE=10
ELEMENTS GROUP $SHELL$ / KT=l LOWER UPPER VARIABLE=10
LOAD
1 WELDING/ NOTHING
ELEMENTS GROUP $PART$ / QR=l VARIABLE=-100
ELEMENTS GROUP $BEAD1$ / QR=l VARIABLE=-100
ELEMENTS GROUP $SHELL$ / TT=20 LOWER UPPER
ELEMENTS GROUP $SKINPART$ / TT=20
TABLE
10001 / 1 20 0.004030 500 0.004836 1200 0.005239 1293 0.00529 --
1300 0.0054 1310 0.0060 1325 0.02525 1330 0.02600 1333 0.02625
1388 0.02625 1392 0.02600 1395 0.02525 1400 0.0066 1405 0.0058 --
1420 0.0053599 1600 0.0053599
10002 / 1 20 0.0137 100 0.0137 500 0.0250 1283 0.0382 1306 0.042
1327 0.05048 1397 0.152 1412 0.156 1445 0.1600 1570 0.1600 --
1700 0.1600
79
10 / FORTRAN
function f{t)
c radiation losses f = sig * e * (t + to) (t**2 + to**2)
e = 0.5
sig = 5.67*-8
to = 20.
to = 20. + 273.15
tl = t + 273.15
a = t1 * t1
b to * to
c a + b
d tl + to
d = d * c
d = d * e
d = d * sig
c convective losses = 15 W/m2
f = d + 15.
g=1.0*-6
f=f*g
return
END
Heat Source Definition
100 / FORTRAN
FUNCTION F{X)
DIMENSION X(5)
xa = X [1] ;
ya = X[2] ;
za = X [3] ;
time X[4] ;
c initial position of heat source in the new frame
xc 7.14375;
yc = 1.0;
zc = 11.1125;
c Translation
xa xa xc
ya = ya - yc
za = za - zc
c rotation matrix
al = 0.7071;
a2 = 0.0;
a3 -0.7071;
bl = 0.0;
b2 = 1.0;
b3 = 0.0;
cl = 0.7071;
c2 = 0.0;
c3 0.7071;
c rotation
aa a1 * xa
bb = a2 * ya
cc a3 * za
xx = aa bb cc + +
aa = b1 * xa
bb = b2 * ya
80
cc = b3 * za
yy aa bb cc + +
aa = c1 * xa
bb = c2 * ya
cc = c3 * za
zz = aa bb cc + +
c Q=U*I
uu = 35.0;
ii = 250.0;
q = uu*ii;
q q*6.0;
q q*1.7320508; Ilq=q*sqrt(3)
q = ql3 .1415927;
q q/1.7724539; II q/sqrt(pi)
c EFFICIENCY
e=0.75;
q=q*e;
c WELDING VELOCITY mm/s
wv=-5.0
c PARAMETERS OF WELDING POOL
a= 4.5; IIHalf width of weld pool (from T-Joint welded piece)
b= 4.9; IIDepth of weld pool (estimation using AWS Doc.A3.0-94)
Yf= a * 0.75; IILength of weld pool in front of center (Goldak's
paper)
Yr= a * 1.5; IILength of weld pool behind center (Goldak's paper)
c Fitting power base to mesh
Qc = 1.0; II Energy for a density = 1.
Qe = 1./Qc; II density of energy
c Proportion of heat in front and rear (Non sYmmetric Gaussian
distribution)
c Note: Qf + Qr = 2
Qf = 0.6; II Fraction of energy in front of HS (Goldak's paper)
Qr = 1.4; II Fraction of energy in the rear of HS (Goldak's paper)
c POSITION OF HEAT SOURCE CENTER
tim1=0.0
tim2=time-tim1
center = wv*tim2;
center = wv*time;
if (yy .GT. center) Qg=Qr;
if (yy .LE. center) Qg=Qf;
if (yy .LE. center) cc=Yf;
if (yy .GT. center) cc=Yr;
Qg=Qg*Qe;
Qg=Qg/a;
Qg=Qg/b;
Qg=Qg/cc;
c CALCULATION HEAT SOURCE BY GOLDAKS FORMULA
rx=xx;
rx=rx*rx;
rx=-rx;
rx=rx*3.0;
s=a*a;
rx=rx/s;
81
rx=exp (rx) ;
ry=center-yy;
ry=ry*ry;
ry=-ry;
ry=ry*3.0;
e=cc*cc;
ry=ry!e;
ry=exp (ry) ;
rz=zz;
rz=rz*rz;
rz=-rz;
rz=rz*3.0;
dc=b*b;
rz=rz!dc;
-rz=exp(rz);
coef=ry*rx;
coef=coef*rz;
f=coef*Qg;
f = f*q;
RETURN
END
5 / FORTRAN
FUNCTION F(X)
DIMENSION X(4)
XX X(l);
YY X(2);
ZZ = X(3) ;
TT X(4) ;
VY =-5.0;
C OUTPUT PARAMETERS
C F=l ELEMENT ACTIVATION
C F=-l ELEMENT DEACTIVATION
C F=O NO EFFECT
F=l
VYT=VY*TT
VYT=VYT-0.01
IF(YY.LT.VYT) f=-l
RETURN
END
RETURN
SAVE DATA 102
SEARCH DATA 102
RENUMBER ITERATION 50
RETURN
SAVE DATA 102
SEARCH DATA 102
TRANSIENT NON-LINEAR EXTRACT 0
BEHAVIOUR METALLURGY 2 FILE META.DAT
ALGORITHM BFGS IMPLICIT 1 ITERATION 250
PRECISION ABSOLUTE NORM 0 FORCE 1*-10 DISPLACEMENT 1
82
METHOD ITERATIVE NONSYMMETRICAL
INITIAL CONDITIONS
NODES / TT 20
ELEMENTS GROUP $PART$ / P 1 0
ELEMENTS GROUP $SHELL$ / P 1 0
ELEMENTS GROUP $BEAD1$ / P lOIS -1
ELEMENTS GROUP $BEAD2$ / P lOIS -1
TIME INITIAL 0.0
0.25 STEP 0.125 / STORE 1
SEARCH DATA 102
ASSIGN 19 TRAN102.TIT
TRANSIENT NON-LINEAR EXTRACT 0
BEHAVIOUR METALLURGY 2 FILE META.DAT
ALGORITHM BFGS IMPLICIT 1 ITERATION 200
PRECISION ABSOLUTE NORM 0 FORCE 1*-10 DISPLACEMENT 1
METHOD ITERATIVE NONSYMMETRICAL
INITIAL CONDITION RESTART CARD LAST
TIME INITIAL RESTART
1 STEP 0.2 / STORE 1
6 STEP 0.25 / STORE 1
201 STEP 0.3 /STORE 1
210 STEP 1 / STORE 1
260 STEP 2 / STORE 1
500 STEP 5 / STORE 1
800 STEP 20 / STORE 1
5000 STEP 100 / STORE 1
5004 STEP 0.5 / STORE 1
5010 STEP 1 / STORE 1
RETURN
SAVE DATA 102
DEASSIGN 19
83
MECH.DAT (Mechanical model data)
SEARCH DATA 102
DEFINITION
T - WELD JOINT
OPTION SHELL SPATIAL MULTI" THERMOELASTICITY
RESTART GEOMETRY
MATERIAL PROPERTIES
ELEMENTS GROUP $PART$ / E=-10001 YIELD=-10004 LX=-10003 LY=-10003
LZ=-10003 MODEL=3 NU=-10002 SLOPE= -10008 PHAS=2 TF=1400
ELEMENTS GROUP $BEAD1$ / STATE=-4 E=-10001 YIELD=-10004 LX=-10003
LY=-10003 LZ=-10003 MODEL=3 NU=-10002 SLOPE=-10008 PHAS=2 TF=1400
ELEMENTS GROUP $BEAD2$ / STATE=-6 E=-10001 YIELD=-10004 LX=-10003
LY=-10003 LZ=-10003 MODEL=3 NU=-10002 SLOPE=-10008 PHAS=2 TF=1400
ELEMENTS GROUP $SHELL$ / H=7.9375 INTE=902 TYPE=4 E=-10001 YIELD=-
10004 --
LX=-10021 LY=-10021 LZ=-10021 MODEL=3 NU=-10002 SLOPE=-10023 PHAS=2
TF=1400
ELEMENTS GROUP $SKINPART$ / TYPE=5
ELEMENTS GROUP $SKINBEAD1$ / TYPE=5
ELEMENTS GROUP $SKINBEAD2$ / TYPE=5
ELEMENTS GROUP $ELEM_TRAN_SH$ / E=200000000 SHAPE=l TYPE=9
CONSTRAINTS
NODES 27101 / UX UY
NODES 27100 / UX
NODES 27102 / UX
NODES 27103 / UX
NODES 27104 / UX
NODES 27105 / UX
NODES 27106 / UX
NODES 27107 / UX
NODES 27108 / UX
NODES 27109 / UX
NODES 27110 / UX
NODES 27111 / UX
NODES 27112 / UX
NODES 27205 / UZ
NODES 27206 / UZ
NODES 27207 / UZ
NODES 27208 / UZ
NODES 27209 / UZ
NODES 27210 / UZ
NODES 27211 / UZ
NODES 27212 / UZ
NODES 27213 / UZ
NODES 27214 / UZ
NODES 27215 / UZ
NODES 27216 / UZ
NODES 27217 / UZ
LOAD
1 WELDING/ NOTHING
84
TABLE
10001 / 1 24 195000 93 189000 204 180000 316 171000 427 161000
538 152000 982 90000 1093 72000 1200 45000 1260 41000 1300 --
20000 1320 10000 1350 50
10002 / 1 0 0.29 900 0.30
10003 / -10006 -10006
10006 / 1 20 0.0000000 100 0.001224 300 0.004396 400 0.006080 500
0.007872 600 0.009686 700 0.011628 800 0.013728 1200 0.023128
1250 0.024477 1300 0.025728 1320 0.026520
10021 / -10022 -10022
10022 / 1 20 0.0000153 100 0.0000153 200 0.0000155 300 0.0000157 400
0.000016
500 0.0000164
10004 / -10007 -10007
10007 / 1 21 365 93 325 149 290 204 270 260 255 316 235 371 230 427
230 --
482 220 538 215 982 70 1093 39 1200 31 1260 28 1300 20 1320 10 1350 1
10008 / -10009 -10009
10009 / 7 20 10010 500 10011 1450 10012
10010 / 1 0 0 0.01 23 0.02 60 0.04 102 0.06 136 0.08 168 0.10 198 0.15
270
10011 / 1 0 0 0.01 23 0.02 42 0.04 70 0.06 92 0.08 109 0.10 126 0.15
156
10012 / 1 0 0 0.01 0.5 0.02 0.5 0.04 0.5 0.06 0.5 0.08 0.5 0.10 0.5
0.15 0.5
10023 / -10024 -10024
10024 / 7 20 10025 500 10026
10025 / 1 0 365 0.01 388 0.02 425 0.04 467 0.06 501 0.08 533 0.10 563
0.15 635
10026 / 1 0 218 0.01 241 0.02 260 0.04 288 0.06 310 0.08 327 0.10 344
0.15 374
4 / FORTRAN
FUNCTION F(X)
DIMENSION X(4)
XX=X(l);
YY X(2);
ZZ = X(3);
TT = X(4);
VY =-5.0;
C OUTPUT PARAMETERS
C F=l ELEMENT ACTIVATION
C F=-l ELEMENT DEACTIVATION
C F=O NO EFFECT
F=l
VYT=VY*TT
VYT=VYT+0.1
IF (YY.LT.VYT) f=-l
85
RETURN
END
RETURN
SAVE DATA 106
SEARCH DATA 102
SEARCH TRAN 102
TEMPERATURE METALLURGY TRANSIENT SHELL CARD 0 TO 800 STEP 1
SEARCH DATA 106
TRANSIENT NON-LINEAR STATIC EXTRACT 0
BEHAVIOUR PLASTIC METALLURGY 2
ALGORITHM BFGS IMPLICIT 1 ITERATION 200
PRECISION ABSOLUTE NORM 0 FORCE 1 DISPLACEMENT 1*-3
METHOD NONSYMMETRICAL TEST 1 ITERATIVE
INITIAL CONDITION
ELEMENTS GROUP $BEAD1$ / IS -1
ELEMENTS GROUP $BEAD2$ / IS -1
TIME INITIAL 0
0.25 STEP 0.125 / STORE 2
RETURN
SAVE DATA TRAN 106
SEARCH DATA 106
ASSIGN 19 TRAN106.TIT
TRANSIENT NON-LINEAR STATIC EXTRACT 0
BEHAVIOUR PLASTIC METALLURGY 2
ALGORITHM BFGS IMPLICIT 1 ITERATION 300
PRECISION ABSOLUTE NORM 0 FORCE 1 DISPLACEMENT 1*-3
METHOD NONSYMMETRICAL TEST 1 ITERATIVE
INITIAL CONDITION RESTART CARD LAST
TIME INITIAL RESTART
1 STEP 0.2 / STORE 3
6 STEP 0.25 / STORE 4
210 STEP 1 / STORE 10
260 STEP 2 / STORE 10
500 STEP 5 / STORE 6
800 STEP 20 / STORE 3
5000 STEP 100 / STORE 3
RETURN
SAVE DATA 106
DEASSIGN 19
86
References
[1] K. Masubuchi, "Analysis of Welded Structures-Residual Stresses, Distortion, and
Their Consequences" Pergamon Press, 1980.
[2] Y.V.L.N. Murthy, G. Venkata Rao, P. Krishna Iyer, "Numerical Simulation of
Welding and Quenching Processes Using Transient Thermal and Thermo-Elasto-
Plastic Formulations" Computers & Structures, vol. 60, No.1, pp. 131-154, 1996.
[3] P. Michaleris, A. DeBiccari, "Prediction of Welding Distortion" Welding Journal,
. vol. 76(4): 172-180s, 1997
[4] S. Sarkani, V. Tritchkov, G. Michaelov, "An efficient approach for computing
residual stresses in welded joints" Finite Elements in Analysis and Design, vol.35, pp.
247-268,2000.
[5] Bora Yildirim, "Nonlinear Thermal Stress/Fracture Analysis of Multilayer Structures
using Enriched Finite Elements" Lehigh University, 2000.
[6] R Dykhuizen, D. Dobranich, "Scaling Analysis for LENS process" Sandia National
Labs, 1998
[7] V. Pavelic, R Tanbakuchi, a.A. Uyehara, and P.S. Myers, Welding Journal Research
Supplement, vol. 48, pp.295-305, 1969
[8] E. Friedman, Journal Pressure Vessel Technology, Trans. ASME, vol. 97, pp.206-213,
1975
[9] F. W. Krutz, L.J. Segerlind, Welding Journal Research Supplement, vol. 57, pp.211-
216, 1978
[10] J. Goldak, A. Chakravarti, M. Bibby, "A New Finite Element Model for Welding
Heat Sources" Metallurgical Transactions B, Volume 15B, pp. 299-305, June 1984.
[11] N.S. Boulton, H.E, Lance Martin, "Residual Stresses in Arc-welded Plates" Proc.
Inst. Mech. Engrs., vol. 133, pp 295-347, 1936
[12] D. Resenthal, "Mathematical theory and heat distribution during welding and
cutting" Welding J. vol. 20 (5), pp 220-234, 1941
[13] K. Masubuchi, F.B. Simmons, RE. Monroe, "Analysis of thermal stresses and metal
movement during welding" Technical Report NASA-TM-X-61300, N68-37857,
Battele Memorial Institute, RSIC-820, Redstone Scientific Information Center, 1968
[14] 1.H. Argyris, 1. Szimmat, and K.J. Will am, "Computational Aspects of Welding
Stress Analysis" Computer Methods in Applied Mechanics and Engineering, vol. 33,
pp. 635-666, 1982
[15] E.F. Rybicki, D.W. Schmueser, R.B. Stonesifer, J.J. Groom, and H.W. Mishler, "A
Finite-Element Model for Residual Stresses and Deflections in Girth-butt welded
pipes" Journal of Pressure Vessel Tech., vol. 100, pp 256-262, 1978
[16] V.I. Papazoglou, K. Masubuchi, "Numerical Analysis of Thermal Stresses during
welding including phase transformation effects" Journal of Pressure Vessel Tech., vol.
104, pp 198-203, 1982
[17] a.A. Vanli; P. Michaleris, "Distortion Analysis of Welded Stiffeners" Journal of
Ship Production, vol. 17(4), pp 226-240, 2001
[18] P. Tekriwal. 1. Mazumder, "Finite element analysis of three-dimensional transient
heat transfer in GMA welding" Welding 1. vol. 67, pp 150-156, 1988
87
[19] P. Tekriwal, J. Mazumder, ''Transient and residual thermal strain-stress analysis of
GMAW" 1. Material Technology, vol.113, pp 337-343, 1991
[20] R.I. Karlsson, B.L. Josefson, "Three-dimensional FE analysis of temperatures and
stresses in a single-pass butt-welded pipe" Journal of Pressure Vessel Tech., vol. 112,
pp 77-84, 1990
[21] S.B. Brown, H. Song, "Finite element simulation of welding of large structures"
Journal of Engineering for Industry, vol. 114, pp. 441-451, 1992
[22] S.R. Daniewicz, M.D. McAninch, B. McFarland, and D. Knoll, "Application of
distortion control technology during fabrication of large offshore structures"
AWS/ONRL International Conference on Modeling and Control of Joining Processes,
Orlando, Fla., 1993
[23] M.V. Deo, P. Michaleris, 1. Sun "Prediction of Buckling Distortion of Welded
Structures" Science and Technology of Welding and Joining, 2002
[24] L.W. Lu, 1.M. Rides, P. Therdphithakvanij, S. Jang, 1. Chung, "Compressive
Strength of AL-6XN Stainless Steel plates and Box columns" ONR N00014-99-0887,
ATLSS report No.02-04, Lehigh University, Bethlehem, PA, 2002
[25] L.W. Lu, A.A. Pang, "Structural Instability Failure Modes: Survey of Existing
Models and Methods of Analysis" Final Report for phase 1.1, TDL L91-01, ATLSS,
Lehigh University, Bethlehem, PA, 1992
[26] J.R. Dydo, W.Cheng, "Stainless steel buckling and angular distortion study" Naval
surface warfare center Project No. 42372-GDE, 1999
[27] A.C. Ramsay, "A study of finite element meshing methods to be used in welding
simulation for ship structures" Recent advances in welding simulation, IMechE, pp
99-117, 2000
[28] S. Murugan, "Temperature and Residual Stress Distributions in Back-Step Welding"
Lehigh Report, 2002
[29] G. Mangialenti, T. Carter "New capabilities of the software product SYSWELD for
the prediction of welding residual stresses and distortions", Recent advances in
welding simulation, IMechE, pp 45-53, 2000
[30] Technical Data blue sheet for stainless steel AL-6XN alloy, Allegheny Ludlum
Corporation, Pittsburgh, PA
[31] S. Marugan, P.V. Kumar, T.P.S. Gill, B. Raj, and M.S.C. Bose, "Numerical
Modeling and Experimental Determination of Temperature Distribution During
Manual Metal Arc Welding", Science and Technology of Welding and Joining, vol.
4, number 6, pp. 357-364, 1999
[32] C.L. Chan, 1. Mazumder, M.M. Chen, "Effect of surface tension gradient driven
convection in a laser melt pool: Three-dimensional perturbation model", Journal of
Applied Physics, vol. 64, number 11, pp 6166-6174, 1988.
[33] C. K. Lueng, R. 1. Pick, D. H. B. Mok, Weld. Res. Counc. Bull., vol. 356, pp. 1-10,
1990
[34] SYSWELED 2003, SYSWELD 2003 Reference Manual, ESI Group, 2003
[35] SYSWELD 2003, "Shell/solid coupling" SYSWELD 2003 Example Manual, ESI
Group,pp.74-84,2003
88
VITA
Dongj in Kim was born in Pusan, South Korea on 26
th
of March 1971. He is the
second son of Chunchu Kim. He graduated from high school in 1989 and started his
undergraduate education in mechanical engineering at Pusan National University. He
completed his undergraduate studies in 1996 and started to work for the company,
POSCO (Pohang Iron and Steel Co., Ltd.). After 6 years work for his company, he came
to Lehigh University in 2002. He expects to get his Master's degree in December 2003.
89
END OF
TITLE