Professional Documents
Culture Documents
a r t i c l e
i n f o
Article history:
Received 28 March 2015
Received in revised form
22 September 2015
Accepted 23 September 2015
Available online 26 September 2015
Keywords:
Thermo-elasto-plastic stress analysis
Finite element analysis
Welding residual stress
Welding heat source
X-ray diffraction
Electro-gas welding
a b s t r a c t
A computational approach considering moving heat sources was introduced to predict the residual stress
distribution produced by electro-gas welding (EGW) joints. Considering the two-pole tandem of EGW, a
nite element analysis (FEA) was suggested to evaluate the thermal behavior of EGW applied to a joint of
ultra-thick plates. Based on a thermo-mechanical FEA, the proles of the residual stresses are investigated
for EH40 thermo-mechanical control process (TMCP) steel plates with a thickness of 80 mm. In order to
simulate the weaving motion of two weldment poles, a quasi-steady heat ux model is introduced based
on Goldaks double ellipsoidal heat source model. The X-ray diffraction methods were used to measure the
residual stress eld on the surface treated by chemical polishing. The residual stress proles determined
by the FEA and measurement showed quite good agreement, as regards the values both peaks and of the
prole. Residual stresses relieved by post-weld treatments, particularly by ultrasonic peening and toe
grinding, is also presented.
2015 Elsevier B.V. All rights reserved.
1. Introduction
Recently, the major shipyards have begun using thermosmechanically controlled ultra-thick plates with thicknesses of more
than 70 mm and yield strengths of approximately 460 MPa, in order
to meet the strength requirements near the upper deck and hatch
coaming area of very large container carriers. These ships transport containers from 10,000 to 18,000 twenty-foot equivalent units
(TEU) (Park et al., 2007; Han et al., 2009). The shear strakes or deck
plates of a large container carrier and FPSO vessels are fabricated
by joining the ultra-thick plates. Traditionally, welding processes
such as submerged arc welding (SAW) and ux-cored arc welding
(FCAW) are preferred for welding these thick plates. FCAW is used
more widely in many shipyards, particularly for butt welds during
the erection procedure and for vertical butt welds during the joining process of the side shell. FCAW has some benets over other
processes in that it is relatively easy to automate and requires less
heat input, both of which are key parameters for improving productivity during the fabrication process of large merchant ships.
FCAW always entails a relatively large number of welding deposits,
normally more than 15, for the thick-sectioned plates, eventually
350
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
Table 1
Chemical composition of EH40 (wt%).
C
Si
Mn
Ni
0.08
0.03
1.56
0.009
0.0013
0.01
Table 2
Mechanical properties of EH40.
Steel
Thickness (mm)
EH40 80
Yield Strength
(MPa)
Tensile Strength
(MPa)
Elongation (%)
470
582
25
Table 3
Welding conditions of tandem EGW.
Welding conditions
Electrode number
Current (A)
Voltage (V)
Speed (cm/min)
Heat input (kJ/mm)
Total heat input (kJ/mm)
1
380
42
2.8
34.47
68.95
2
380
42
2.8
34.47
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
351
Fig. 4. Measurement of the variations in lattice spacing from the diffraction angle.
(1)
lattice distance is calculated from Eq. (1), the strain can be obtained
accordingly.
Using XRD technique, longitudinal residual stresses (stresses
parallel to weld direction) were measured on the top surface of the
specimens. Fig. 5 shows the measured stress distributions on the
raw specimens without any surface polishing. Compressive (negative) peak stresses were found in the HAZ while there would be
a peak tensile stress in the arc welding process. It can be supposed that the stress prole on the surface was distorted by the
surface roughness due to the external forces produced by welding
equipment and subsequent works.
352
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
Fig. 5. Measured residual stress distribution with respect to the distance from the weld toe before the chemical polishing.
Table 4
Surface treatment by chemical polishing.
Thickness
Before treatment
After treatment
81 mm
80.7 mm
Surface
Therefore, a clean surface condition without rust and mechanical noise was required for accurate measurements, and then the
light grinding or polishing was necessary to produce a better surface. The scales/oxides on the surface of the weld joints were
removed using chemical etching without disrupting the residual
stresses. Table 4 shows the surface status of the treated specimens.
The residual stress distribution was also measured for the specimens under different conditions. The specimens were prepared
with and without toe treatments to check the residual stress
redistribution characteristics according to the different toe treatment techniques, such as toe grinding (TG) and ultrasonic peening
(UP). These toe treatment techniques are benecial to the fatigue
strength but sometimes the toe treatment also changes the residual stress distribution near a specimens surface. This can affect
surface crack formation during its initiation phase. Fig. 6 shows the
surface residual stress distributions of the as-welded, toe ground,
and ultrasonic peened specimens, respectively. In the case of the
as-welded specimen, the welding residual stress reached 300 MPa
at the location of the weld toe, and decreased gradually as it moved
away from the weld down to the zero level. In contrast, in case of
the toe treated specimen, the welding residual stresses began with
negative or near-zero values at the weld toe and converged to a zero
level with respect to the distance from the weld toe. The stresses
plotted in Fig. 6 are transverse stresses; i.e., components normal to
the weld line.
Fig. 6. Measured residual stress distribution with respect to the distance from the
weld toe after the chemical polishing.
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
353
the high temperatures induced by the high input of the EGW, was
modeled realistically with a mechanical contact condition against
the specimen. Because the complicated thermal contact behavior
between the sliding copper shoe and the specimen is difcult to
model, it was assumed that the heat transfer into the sliding copper
shoe occurred without resistance. To minimize the CPU time during
the analysis, the mesh size was controlled in such a way that the ne
mesh with a 5 mm length was applied to both the base materials
near the weld line and the sliding copper shoe. An increasing mesh
size was applied gradually to the region away from the weld line.
The size of the entire FE model was 320 mm long, 630 mm wide, and
80 mm thick. The length and width of the model was chosen so that
the residual stress distribution would not be affected signicantly
354
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
(Kim et al., 2000). The spatial distribution of the heat source might
be modeled by using the Gaussian surface heat ux model or the
double ellipsoidal heat source model (Nguyen, 2004). The latter
provides a better representation of the real physics in the case of
a moving arc for a butt weld (Kim et al., 2000). The weaving characteristics of the moving heat source should be considered in the
numerical simulation of the EGW process. In this study, a quasisteady movement of the heat source was assumed, as shown in
Fig. 8.
Fig. 9 shows the double ellipsoidal heat source model used in this
study. The heat source distribution combines two different ellipses,
i.e. one in the front quadrant of the heat source and the other in
the rear quadrant (Goldak et al., 1984). The power densities of the
double-ellipsoid heat source, qf (x, y, z) and qr (x, y, z), describing
heat ux distributions inside the front and rear quadrant of the heat
source can be expressed by the following equations.
qf (x, y, z) =
(2)
(3)
qr (x, y, z) =
Q =V I
(4)
where qf and qr represent the weld ux rates per unit volume for
the front and rear weld pools, respectively, a, b, cf , and cr denote the
geometric parameters dening the size and shape of the ellipses of
heat ux shown in Fig. 9. Q is the heat power, is the efciency, V
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
355
is the voltage, and I is the current. The fractional factors ff and fr are
dened as given by Eq. (5). It is assumed that the total energy input
rate Q splits into fractions ff and fr in the front and the rear quadrant
of the heat ux deposited; then, ff + fr = 2 (Joshi et al., 2013).
ff =
2
2
, fr =
1 + cf /cr
1 + cr /cf
(5)
The efciency () is a quantity that represents the ratio of the power
transferred from the arc to the base metal to the entire power generated by the arc. The efciency is difcult to estimate due to the
dependence on the welding conditions and molten depth, so an
assumed value of 0.65 was used in the analysis (Armentani et al.,
2007). The geometrical parameters a, b, cf , and cr were adjusted to
356
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
be 0.9, and the lm coefcient (h) was set to be a function of temperature. The heat ux losses on the surfaces of the welded plate
due to the convection and radiation are given by:
h = 24.1 104 r T 1.41
(6)
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
357
of EH40 TMCP material. The curves were assumed based on multilinear isotropic hardening properties suggested by Chang and Teng
(2003) and Masubuchi (1980). A nonlinear isotropic hardening
model with a von Mises yield criterion was applied for the plastic
behavior of the base metal. The mechanical and thermal properties
of the weld metal were assumed to be the same as those of the base
metal. The material properties of the sliding copper shoe were set
to be those of copper (Simon et al., 1992) Fig. 11.
5. Result of nite element analysis
5.1. Temperature distribution
As it is shown in Section 4.2, the heat source model is dened
by a function of position and time together with a number of
parameters, which affect the heat ux distribution. The param-
358
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
Fig. 16. FZ and HAZ estimated from the temperature distribution of FEA.
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
359
360
S.-Y. Hwang et al. / Journal of Materials Processing Technology 229 (2016) 349360
Table 5
Summary of numerical analysis of EGW.
Geometry
Welding Condition
Material Properties
Acknowledgement
This research was supported by the Special Education Program
for Offshore Plants of the Ministry of Trade, Industry and Energy
Affairs (MOTIE) of the Korean government, and by an Inha University research grant. The authors gratefully acknowledge this
support.
Analysis Option
References
Table 6
Parameters of heat source for FEA.
a
cf
cr
ff
fr
25 mm
25 mm
10 mm
10 mm
shows the stress distribution with respect to the specimens thickness direction. The larger tensile stresses are present near the
mid-thickness area for both transverse and longitudinal stress components.
6. Conclusion
3-D nite element model of tandem EGW was developed to
simulate the temperature eld, and the residual stress distribution induced by the tandem EGW process. Meanwhile, the residual
stresses computed by FE model were veried by the experiment.
In the simulation, a quasi-steady-state heat source based on a double ellipsoidal heat source model was used to consider the heat
input of two electrodes of EGW. Although a relatively simple heat
source model was used to calculate the residual stress eld induced
by EGW, and a good prediction was obtained in the present work.
The proposed process was evaluated by a comparison between the
measured and calculated residual stress distributions. The overall
correspondence between the results is good, but some discrepancy
is observed, because of surface condition of the specimens affected
by ceramic backing material and sliding copper shoe.
Before the residual stress measurements, the scales/oxides on
the surface of the weld joints were removed using chemical polishing. The residual stress characteristics of the EGW specimens were
evaluated by XRD measurements of the residual stresses for the
toe-treated specimen and the ultrasonic peened specimen. Longitudinal tensile residual stress was much higher than the yield
strength. A typical feature of welding residual stress in the EGW
joint was that the region with high tensile stress was large, which
was similar with that induced by other arc welding processes such
as FCAW and CO2 welding.
Armentani, E., Esposito, R., Sepe, R., 2007. The effect of thermal properties and
weld efciency on residual stresses in welding. J. Achiev. Mater. Mfg. Eng. 20
(1), 319322.
Bang, H.S., Bang, H.S., Joo, S.M., 2009. Analysis of mechanical behavior and fracture
toughness KIC in EGW welded joints for high strength EH36-TMCP ultra thick
plate. J. Comput. Struct. Eng. Inst. Korea 2 (6), 565572.
Chang, P.H., Teng, T.L., 2003. Numerical and experimental investigations on the
residual stresses of the butt-welded joints. J. Comput. Mater. Sci. 29 (4),
511522.
Chun, K.S., Song, H.C., Im, S.W., 2007. Measurement of welding residual stresses by
X-ray diffraction method. In: proceedings of the annual autumn meeting. J.
Soc. Nav. Arch., 208213.
Goldak, J., Bibby, M., Chakravarti, A., 1984. A new nite element model for welding
heat sources. Metall. Trans. 15B, 299305.
Gery, D., Long, H., Maropoulos, P., 2005. Effects of welding speed, energy input and
heat source distribution on temperature variations in butt joint welding. J.
Mater. Process. Technol. 167 (23), 393401.
Joshi, S., Hildebrand, J., Aloraier, A.S., Rabczuk, T., 2013. Characterization of
material properties and heat source parameters in welding simulation of two
overlapping beads on a substrate plate. Comput. Mater. Sci. 69, 559565.
Joshi, S., Semetay, C., Price, J.W.H., Nied, H.F., 2010. Weld-induced residual stresses
in a prototype dragline cluster and comparison with design codes. J.
Thin-Walled Struct. 48 (2), 89102.
Kim, I.S., Kim, I.K., Moon, C.J., Jeong, Y.J., An, Y.H., 2000. A study on analysis of heat
transfer and residual stress on the weld zone using FEM. J. Korean Soc.
Machine Tool Eng. 9 (5), 96104.
Masubuchi, K., 1980. Analysis of Welded Structures. Pergamon Press, Oxford, New
York.
Nguyen, N.T., 2004. Thermal Analysis of Welds. WIT Press, Southampton UK.
Park, C.S., Son, Y.R., Lee, J.S., 2007. A study on the application of vertical welding to
the shipbuilding with high deposition rate. J. Soc. Nav. Arch. Korea 44 (5),
482487.
Price, J.W.H., Ziara-Paradowska, A., Joshi, S., Finlayson, T., Semetay, C., Nied, H.,
2008. Comparison of experimental and theoretical residual stresses in welds:
the issue of gauge volume. Int. J. Mech. Sci. 50 (3), 513521.
Rosenthal, D., 1941. Mathematical theory of heat distribution during welding and
cutting. Weld. J. 20 (5), 220234.
Rykalin, R.R., 1974. Energy source for welding, Houdrement lecture. Int. Inst.
Weld., 123.
Sasaki, K., Suda, K., Motomatsu, R., Hashiba, Y., Ohkita, S., Imai, S., 2004.
Development of two-electrode electrogas arc welding process. Nippon Steel
Tech. Rep. 90, 6774.
Simon, N.J., Drexler, E.S., Reed, R.P., 1992. Properties of copper and copper alloys at
cryogenic temperatures. NIST Monogr., 176177.
Teng, T.L., Chang, P.H., 2004. Effect of residual stresses on fatigue crack initiation
life for butt-welded joints. J. Mater. Process. Tech. 145, 325335.
Wahab, M.A., Painter, M.J., 1997. Numerical models of gas metal arc welds using
experimentally determined weld pool shapes as the representation of the
welding heat source. Int. J. Press. Vess. Pip. 73, 153159.