See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/315572540

Finite element analysis of welded structures

Conference Paper · May 2007

CITATIONS READS
0 2,783

1 author:

Sachin Awasare
Indian Register of Shipping
7 PUBLICATIONS   12 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Distortion control in welding View project

"DEVELOPMENT OF MATHEMATICAL MODEL IN MATLAB FOR BOTTOM IMPACT, BOWFLARE SLAMMING AND GREEN WATER PRESSURE COMPUTATIONS" View project

All content following this page was uploaded by Sachin Awasare on 11 June 2019.

The user has requested enhancement of the downloaded file.

 Finite element analysis of welded structures

Ramesh Sawant, Sachin Awasare and Asokendu Samanta♣

Long Term Strength Assessment Group
Research and Co-ordination Division
Indian Register of Shipping, Powai, Mumbai – 400072
♣
Phone: 91-22-30519400, E-mail: a.samanta@irclass.org
________________________________________________________________________

Abstract: In welding process, there are many parameters, which affect the distortion and residual
stress distribution in a welded plate. In this paper attempts have been made to determine
temperature distribution due to welding by finite element method in welded panel structures.
General purpose finite element software ANSYS is used for this purpose. The FE analysis is
carried out in two steps. A nonlinear transient thermal analysis is conducted first to obtain the
global temperature history generated during the welding process. A stress analysis is then
developed with the temperatures obtained from the thermal analysis used as loading to the stress
model. The scope of present study is primarily limited to predict the temperature distribution
developed in butt and fillet welded panel structures only and temperature profiles are presented.
Good agreement of temperature profiles with existing literatures validates the efficacy of the
present methodology.

Keywords: Welded structure; Finite element method; Transient thermal analysis; Temperature
distribution
________________________________________________________________________

1. Introduction close to the weld are subjected to different

rates of expansion and contraction
Welding joints are extensively used in developing a three-dimensional complex
fabrication industry including residual stress state. In particular, during
shipbuilding, offshore structures and steel the weld thermal cycle, material
bridges. The merits of these joints are mechanical properties change drastically,
high joint efficiency, water tightness and especially when material approaches
low fabrication cost. The ships are melting temperature. Therefore, due to the
fabricated from metals like steel or temperature dependence of material
aluminum. In particular, there are properties and the large deformation in
structural, material and welding factors. welding, material and geometrical non-
The structural parameters include linearity have to be taken into account.
geometry of the plates, thickness, width These high residual stresses in regions
and joint type. Among the material near the weld not only promote brittle
parameters mechanical and physical facture, fatigue, or stress corrosion
properties and type of filler metal were cracking but also distortion in the plate
considered. Welding process parameters that reduce the buckling strength of
include type of process employed, welded structure. Due to that, process is
welding procedure, current, voltage, arc needed to predict the temperature
travel speed and arc efficiency. distribution, residual stresses and
As a consequence of the non-uniform distortion in weld. To understand the
temperature distribution, parts of materials formation of residual stress, node

1
 Finite element analysis of welded structures

temperature history during the welding have a more refined mesh close to the
process must be calculated. weld line while in regions located away
However, accurate prediction of from weld-line a more coarse mesh is
residual stresses and distortions induced used. The material properties of the weld
by welding process is extremely difficult metal (WM), base metal (BM) and heat
because the thermal and mechanical affected zone (HAZ) are temperature and
behavior in welding include local high temperature-history dependent. Due to the
temperature, temperature dependence of lack of information on material properties
the material properties and a moving heat of WM and HAZ, both thermal and
source. mechanical material properties of WM
The method described here performs and HAZ are assumed to be the same as
non-linear transient thermal analysis using that of the BM in the analysis [2].
finite element technique to analyze the Following assumptions are made in the
thermo mechanical behaviour. It is thermal model
compared with experimental results from
different literatures. 1. All the thermal properties except
density are considered as a function of
2. Modeling procedure temperature.
2. Linear Newtonian convection cooling
Welding simulation in ANSYS consist is considered on all the surfaces except the
of two parts, 1) Calculation of thermal weld zone.
cycles (thermal analysis) 2) Calculation of 3. Cooling due to natural convection is
stresses (structural analysis). The thermal considered, whereas heat loss due to
analysis gives the temperature distribution radiation is neglected
in the welded panel whereas from 4. A dimensionless parameter (arc
structural analysis residual stresses and efficiency) is taken into account for losses
distortions are calculated. occur (due to loss of current, voltage etc)
A 3D model using solid elements is in calculating the heat input.
necessary to determine distribution of 5. Effects arising from phase change are
temperature through thickness and across taken into account. i.e. enthalpy changes
the plate in time. The out of plane during the phase change.
distortion is of a particular interest
because if large enough it may affect The heat is transfered from the arc of
buckling behavior of the plate. electrode to the plate, then due to
Applying heat flux and power density conduction it transfers to the plate and
on element faces simulate the moving heat eventual heat releases through boundary
source for surface and volume. The convection.
accuracy of the finite element method
depends upon the density of the mesh
3.Heat source calculation
used in the analysis [1]. The temperature
3.1.Surface heat flux
around the arc is higher than the melting
point of the material, and it drops sharply
For arc butt-welding process, the
in regions away from the weld pool.
moving heat load is applied as distributed
Therefore in order to obtain the correct
heat flux (W/m2) to the top surface of the
temperature field in the region of high
model. The region within which the heat
temperature gradients it is necessary to
is applied has circular shape assuming the

2
 Ansys India Conference, Bangalore – 2007

heat source is applied perpendicularly to time in butt and fillet welds. Element
the plate without any inclination. Birth & Death allows the user to
Assuming that the heat from the welding deactivate or activate specific elements
arc is applied at any given instant of time during course of an analysis. All elements
as a normally distributed heat flux, then must be created, including those weld
the rate of heat generation [3] is given by fillers to be ‘born’ in later stages of the
analysis.
3Q e ⎡ ⎛ r ⎞2 ⎤ A typical use of this feature to model
q su p ( r ) = ex p ⎢−3 ⎜ ⎟ ⎥
πr2 ⎢⎣ ⎝ r ⎠ ⎥⎦ (1) assembly procedures is deactivating all
elements associated with weld pool and
Where, r is the distance from the center again activating elements at various
point of the heat source to the point for loadsteps. Reducing their contributions to
which the heat flux is calculated on the stiffness matrices significantly does the
plate surface, r is the characteristic radial deactivation of elements. Similarly when
dimensional distribution parameter that elements are born, they are not actually
defines the region in which 95% of heat added to the model, but are simply
flux is deposited. reactivated, its stiffness mass, element
The total rate of energy input Qe loads return to its original values
depends upon the type of heat source i.e.
type of welding. For electric arc welding 5. Butt weld joint of steel

Qe = η V I (2) The weld sample (75×25 mm, two

plates butt joint) shown in Fig. 1 is
3.2. Power density selected for the verification purpose. This
problem was previously investigated by Li
The moving heat load applied in the et al. [1]. Due to symmetry, only one half
finite element model for fillet welding is of the plate is investigated.
taken as a hemispherical Gaussian Temperature dependent thermal
distribution of power density (W/m3). The properties used for this problem are
cross section of nodes is triangular on shown in Figs 2 – 6 [4, 5]. The
which power density function heat source parameters, I = 206 A, V = 31 V, η =
is applied. 0.95, v = 3.79 mm/sec are used for the
The moving heat load is applied on the present investigation. Efficiency and
volume of each element bounded by weld welding speed are assumed, as these are
lines. The power density distribution for not mentioned in the literatures.
volume source is given [3] by

⎡ ⎛ r ⎞2 ⎤
q sup ( r ) = 6 3Q e
e ⎢ −3⎜ ⎟ ⎥ (3)
r 3π
⎣⎢ ⎝ r ⎠ ⎦⎥
π

4. Element birth and death

The model in this study adopts the

technique of element ‘birth and death’ to Fig. 1 Schematic diagram of the butt-welding
simulate the weld filler variation with test plate

3
 Finite element analysis of welded structures

60 1800
55
Li et al. (2004)
1600
50 Present
Thermal Conductivity (Watt/m K)
0

45 1400
40
1200
35

Temperature ( C)
0
30 1000
25
800
20
15 600
10
400
5
0 200
0 200 400 600 800 1000 1200 1400 1600

Temperature ( C)
0 0
0 25 50 75 100 125 150 175 200
Time (Sec)
Fig. 2 Temperature dependent thermal
conductivity Fig. 6 The welding thermal cycle in the fusion
zone at the end of the welding point
16
Thermal expansion coefficient (10 / C)
-6 0

14

12

6. Fillet weld joint of mild steel

10

A stiffened T- joint is considered here for

8
0 200 400 600 800 1000 1200 1400 1600 verification purpose. This problem was
previously investigated by Camilleri et al.
0
Temperature ( C)

Fig. 3 Temperature dependent thermal

[6]. It consists of fillet welding of central
expansion coefficient
stiffener with dimension 0.15m×
10
0.5m×6mm thick flat bar attached to a
8
plate, 0.5m×0.5m×6.5mm thick, as shown
in Fig. 7. Fig. 8 shows the mesh divisions
Convection Coefficient (W/m K)
2

6 at fusion zone of fillet weld. The

parameters I = 256 A, V=24 V, η =0.95, v
=4.08 mm/s are used in this example.
4

2 Temperature dependent thermal properties

used for this example are shown in Figs 9-
0
0 250 500 750 1000 1250 1500 1750 2000 2250 2500 2750 3000 13.
o
Temperature ( C)

Fig. 4 Temperature dependent convection

coefficient
1800

1500
K )
0
S p e c ific H e a t (J /K g

1200

900

600

300

0
0 200 400 600 800 1000 1200 1400 1600
0
T e m p e ra tu re ( C )

Fig. 5 Temperature dependent specific heat Fig. 7 Schematic diagram of the fillet welding

4
 Ansys India Conference, Bangalore – 2007

12

Convection Coefficient (W/m K)

10

2
8

0
0 1000 2000 3000 4000
0
Temperature ( C)

Fig. 8 Meshing in the weld region Fig. 12 Temperature dependent convection

140
coefficient
120 10000
Thermal Conductivity (Watt/m K)
0

100
8000
80

Enthalpy (MJ/m )
3
60 6000

40
4000
20

0 2000
0 500 1000 1500 2000
O
Temperature ( C)
0
0 1000 2000 3000 4000
Fig. 9 Temperature dependent thermal o
Temperature ( C)
conductivity
1600

1400 Fig. 13 Temperature dependent enthalpy

1200
Specific Heat (J/Kg K)

Thermocouple positions for

0

1000

800 measurement of thermal history are shown

600 in Fig.14. For present analysis the results
400 considered are at the row that is 250 mm
200 away from the start of the weld. The
0
0 500 1000 1500 2000 points for present validation are taken as
Temperature( C)
O
WR1 23,WR1 28,WR1 38, NSR1
Fig. 10 Temperature dependent specific heat 20,NSR1 25 and NSR1 35 as shown in
16 Fig 14.
Thermal expansion co-efficient(10 / C)
-6 o

15

14

13

12

11

10

0 1000 2000 3000 4000 5000

O
Temperature ( C)

Fig. 11 Temperature dependent thermal

expansion coefficient.
Fig. 14 Thermocouple positions

5
 Finite element analysis of welded structures

Temperature profiles measured at

260
240 Camilleri et al (2006)

points, which are away from fusion zone 220

200
Present

of fillet weld, are plotted in Figs 15-20. 180

Temperature ( C)
160

O
The maximum temperature and the 140
120

temperature profiles in respective 100

examples are matching well with the

80
60

existing results. The slight variations in 40

20

results are due to the assumptions 0

0 50 100 150 200 250

(welding efficiency, electrode diameter) Time (Sec)

that are made in the example. In addition, Fig. 17 Thermal cycle at WR1 38 mm
linear extrapolation of material properties
to high temperatures may also be the 350
Camilleri et al (2006)
source of slight difference. It is also to be 300 Present

mentioned that temperature dependent 250

Temperature ( C)
thermal properties are taken from the

O
200

graph, as exact data are not available. The 150

variation of the temperature profile is 100

observed may be due to all these reasons. 50

0
0 50 100 150 200 250

Time (Sec)

Fig. 18 Thermal cycle at NSR1 20 mm

350

Camilleri et al (2006)
300 300
Present
Camilleri et al (2006)
250 Present
250
Temperature ( C)
O

200
200
Temperature ( C)
O

150

150
100

50 100

0
0 50 100 150 200 250
50
Time (Sec)

0
0 50 100 150 200 250
Fig. 15 Thermal cycle at WR1 23 mm Time (Sec)

Fig. 19 Thermal cycle at NSR1 25 mm

250
300
Camilleri et al (2006)
Camilleri et al (2006)
Present Present
250 200

200
Temperature ( C)
Temperature ( C)
O

150

150

100
100

50
50

0
0 50 100 150 200 250
0
Time (Sec) 0 50 100 150 200 250

Time (Sec)

Fig. 16 Thermal cycle at WR1 28 mm Fig. 20 Thermal cycle at NSR1 35 mm

6
 Ansys India Conference, Bangalore – 2007

7. Conclusions [4] Peng-Hsiang Chang, Tso-Liang Teng,

2004,Numerical and experimental
A 3-D finite element analysis has investigations on the residual stresses
been carried out using ANSYS software of the butt-welded joints,
Computational Materials Science,
to determine temperature profile due to
Vol 29, pp.511–522
welding. Element death and birth option [5] Tsirkas S.A., Papanikos P.,
has been used to simulate the welding Kermanidis Th., 2003, Numerical
process. Maximum temperatures in both simulation of the laser welding
the cases are matching well with the process in butt-joint specimens,
existing results. A slight variation of the Journal of Materials Processing
temperature profile is observed. This is Technology, Vol 134, pp.59-69
due to the assumptions (welding speed, [6] Camilleri D., Mollicone P. and Gray
welding efficiency, electrode diameter) T G F, 2006 Alternative simulation
that are made in the present analysis, as all techniques for distortion of thin plate
the data are not available in the literatures. due to fillet-welded stiffeners,
Modelling And Simulation In
The present analysis will be extended to
Material science And engineering,
determine the residual stress developed Vol 14, pp.1307-1327
due to welding.

Nomenclature

I = Welding current (A)

Qe= Rate of energy input (W)
V = arc voltage (V)
η = arc efficiency
v = Welding speed (mm/s)

References

[1] Li Yajiang, Wang Juan, Chen Maoai

and Shen Xiaoqin, 2004,Finite
element analysis of residual stress in
the welded zone of a high strength
steel, Bulletin of Material Science, vol
27, No. 2, pp. 127-132
[2] Wu A., Syngellakis S. and Mellor
B.G., 2001, Finite element analysis of
residual stresses in a butt-weld.
Proceedings of the Seventh
Postgraduate conference in
Engineering Materials, 37-38
[3] Goldak J., Chakravarti A., and Bibby
M., 1984, A New Finite Element
Model for Welding Heat Sources,
Metallurgical
Transactios B, 15B: pp.299-305

View publication stats