International Conference on Ships and Offshore Structures

ICSOS 2018
17 – 19 September 2018, Gothenburg, Sweden

An Empirical Formulation for Predicting Welding-Induced

Compressive Residual Stresses on Steel Stiffened Plate
Structures and Its Application
Myung Su Yia,b, Chung Min Hyunb and Jeom Kee Paika,c,d*
a
Department of Naval Architecture and Ocean Engineering, Pusan National University, Busan 46241, Korea
b
Ship & Offshore Research Institute, Samsung Heavy Industries Co., Ltd., Geoje, Gyeongsangnam-do 53261, Korea
c
The International Centre for Advanced Safety Studies (The Lloyd’s Register Foundation Research Centre of Excellence), Pusan
National University, Busan 46241, Korea
d
Department of Mechanical Engineering, University College London, London WC1E 7JE, UK
* Corresponding author. J.K. Paik (Email) [email protected]

Abstract
The aim of this paper is to derive an empirical formulation for predicting welding-induced
compressive residual stresses in welded steel plate panels. In addition to direct measurement database
for full-scale models of welded steel plate structures, elastic-plastic finite element method solutions are
obtained. The formula is derived by a curve fitting of the databases obtained from both direct
measurements and numerical computations as a function of the plate slenderness ratio and the weld leg
length. An applied example is presented to demonstrate the applicability of the proposed formula where
the welding-induced compressive residual stresses cause distortions due to a thermal buckling
phenomenon in thin plate panels of living quarter structures, and accurate predictions of the
welding-induced compressive residual stresses together with an advanced buckling design technology
can effectively reduce the thermal buckling induced distortions. Important insights and conclusions are
presented.

Keywords: Welding-induced residual stresses; steel stiffened plate structures; full-scale measurements; elastic-plastic thermal
stress finite element method analysis

1. Introduction
Plated structures such as ships and offshore structures are constructed by welding to attach support
members (e.g., longitudinal stiffeners and transverse frames) to the plating as shown in Figure 1. In this
process, initial imperfections develop in the form of deformations and residual stresses. Figure 2 shows
the types of welding-induced initial deformations or distortions in plated structures. Figure 3 shows the
distribution of welding-induced residual stresses that can develop in both longitudinal and transverse
directions. Such initial imperfections significantly affect buckling and ultimate strength of these
structures (Paik 2018). In some types of structures, e.g., living quarter structures of offshore

1
installations, which are less strength critical, the welding-induced initial deformations may essentially
require fairing works to remove them. It is therefore of paramount importance to predict the magnitude
and pattern of welding-induced initial imperfections. This paper focuses on the prediction of
welding-induced residual stresses on steel stiffened plate structures.
A number of related studies are available in the literature in association with the direct measurement
of welding-induced initial imperfections in steel or aluminium stiffened plate structures. For example,
Masubuchi (1980), Smith et al. (1988), Ueda (1999), Paik et al. (2006), Paik (2007c, 2008), Paik et al.
(2008, 2012), Vhanmane et al. (2008), Luís et al. (2009), Bruno et al. (2011), Khedmati et al. (2012),
and Teresa et al. (2013) provided measurement databases for the initial deformations of
aluminum-plated structures; Masubuchi (1980), Smith et al. (1988), Cheng et al. (1996), Ueda (1999),
and Kenno et al. (2010, 2017) obtained measurement databases of welding-induced residual stresses on
steel-plated structures; and Paik et al. (2006), Paik (2008), and Paik et al. (2008, 2012) obtained
measurement database of welding-induced residual stresses and softening on aluminum-plated
structures. Khan et al. (2011), Gannon et al. (2012a, 2012b, 2015), Eggert et al. (2012), Farajkhah et al.
(2016), Lillemäe et al. (2017) and Chen et al. (2018) studied the effects of welding-induced residual
stresses on the fatigue strength of steel-plated structures. A survey of the literature suggests that more
databases of direct measurements for welding-induced initial imperfections on full-scale stiffened plate
structure models are needed. Also, it is well realized that accurate predictions of welding-induced initial
imperfections are very useful.
The aim of the present paper is to derive an empirical formula for predicting the welding-induced
compressive residual stresses on steel plate panels. For this purpose, direct measurement databases of
welding-induced residual stresses on full-scale models of steel stiffened plate structures obtained by the
authors (Yi et al. 2018b) are used. In addition, elastic-plastic finite element method computations
obtained with varying the plate thickness and the weld leg length by the authors (Yi et al. 2018a) are
used. The formula is derived by a curve fitting of the databases obtained from both direct measurements
and numerical computations as a function of the plate slenderness ratio and the weld leg length.
An applied example is presented to demonstrate the applicability of the derived formula where thin
plate elements in living quarter structures deflect significantly during welding process as
welding-induced compressive residual stresses cause a thermal buckling phenomenon, and it is shown
that accurate predictions of the welding-induced compressive residual stresses on plate panels together
with an advanced buckling design technology can effectively prevent the thermal buckling.

Figure 1. A stiffened plate structure (Paik 2018).

2
 Figure 2. Types of welding-induced distortion in plated structures (Paik 2018).

Transverse stiffener
Weld bead

y
at a  2a t at
bt

Longitudinal stiffener
Comp.
rcy
b2b t

rty
Tens.
bt

Tens.
x
rtx rcx

Figure 3. Distribution of welding-induced residual stresses in stiffened plate structures (Paik 2018).

2. Investigated structures
Figure 4 shows a two-bay model of steel stiffened plate structures considered in this study, which
are associated with the direct measurements and finite element method analyses detailed in separate
papers (Yi et al. 2018a, 2018b). Longitudinal stiffeners have an angle type and transverse frames have
a T-bar type, as shown in Figure 5.
Tables 1 and 2 indicate the dimensions and the mechanical properties of the structures. In this study,
tensile coupon tests were not performed to define the material properties but Table 2 presents nominal
values of mechanical properties defined from steel mill certification of steel maker (POSCO steel mill).
The support members were attached by fillet welding using flux cored arc welding (FCAW)
technique, and the details of welding conditions are provided in Table 3 where one-pass welding was
applied.

3
 Transverse stiffeners

Longitudinal stiffeners
b/2
b

Symmetry
b/2

a/2 a a/2

Figure 4. The two-bay model of steel stiffened plate structures considered in this study.

a ba ba ba b b

t tt tt tt t t

hwy
wx hhwx
wy hwx hwx
hwy hhwxwy
twy ttwx
wy twx
twy twx
tfy twx
wy tfx
tfy tfxtfy twx tfx

bfy bbfxfy bbfxfy bfx

(a) Longitudinal stiffener (b) Transverse frame

Figure 5. Types of support members used for the object structures.

Table 1. Dimensions of the investigated structures.

Stiffener
a (mm) b (mm) t (mm) Lw (mm) hw (mm) tw (mm) bf (mm) tf (mm)
Type
6 Longitudinal
125 7 75 7
8 (Angle type)
3,200 800 6.5
10 Transverse
350 12 100 17
12 (T-bar type)

Note: Lw is the weld bead length (leg length), with the nomenclature defined in Figure 6.

4
 Lth
Lw
Lth : Design throat thickness
Lw : Weld bead length (leg length)

Figure 6. Nomenclature of the weld leg length.

Table 2. Nominal values of mechanical properties.

Kind of material E (GPa) Y (MPa) u (MPa)   f (mm/mm)

Mild steel (Grade A) 203 309 458 0.3 0.23

Note: E is the elastic modulus, Y is the yield strength, u is the ultimate tensile strength,  is
the Poisson ratio, and  f is the fracture strain (elongation).

Table 3. Welding conditions to attach the support members.

Lw Current Voltage Speed Speed Heat input

(mm) (A) (V) (CPM) (mm/s) (KJ/mm)
6.5 320 32 36 6.0 1.71

Trans. Stiffener
Trans. stiffeners
b/2

Longi.
Longi . Stiffener

Symmetry
stiffeners

Symmetry
b

Domain
b/2

of calculation
a/2 a a/2

Figure 7. The extent of the finite element analysis.

5
 Fix_y

Figure 8. Thermal loading and boundary conditions for a quarter model of the object structure (Yi et al.
2018a).

3. Numerical computations and direct measurements

The details of numerical computations and direct measurements of welding-induced residual
stresses are presented in separate papers (Yi et al. 2018a, 2018b), but a brief summary is provided here.

3.1 Numerical computations

To compute welding-induced residual stresses on steel stiffened plate structures, the
three-dimensional thermo-elastic-plastic finite element analysis method was applied using the
MSC/MARC program (MSC 2013). Figure 7 shows the extent of the analysis applied in the
computations. Figure 8 shows a typical mesh model used for the purpose of thermal analysis. Figure 8
presents the welding sequences that are divided into eight stages, which are the same as in the actual
welding process. The finite elements used are eight-node three-dimensional brick thermal-mechanical
finite elements, also called Solid 7 (full integration type) elements that address plasticity,
hyper-elasticity, stress stiffening, creep, large deflection, and large strain capabilities (MSC 2013). In
the finite element analyses, the plate thickness is varied at 6 mm, 8 mm, 10 mm, and 12 mm, and the
weld leg length is varied at 3.5 mm, 4.5 mm, 6 mm, 6.5 mm, and 8 mm.

3.2 Direct measurements

The X-ray diffraction technique was used to measure welding-induced residual stresses. Figure 9
shows a set-up for measuring such stresses. The welding-induced residual stresses, both in the
longitudinal and transverse directions, were measured at eight locations in each direction. All
measurements using XRD were performed on the opposite side to which the stiffener is welded. Figure
10 shows the measuring locations of the welding-induced residual stresses.

Figure 9. The set-up of the X-ray diffraction technique to measure the welding-induced residual
stresses.

6
 b/2 = 400 mm

Longitudinal direction
Transverse direction

b = 800 mm

b/2 = 400 mm

800 mm
a/2 = 1,600 mm a = 3,200 mm a/2 = 1,600 mm

30 mm

380 mm
50 mm
60 mm
180 mm
100 mm
180 mm
80 mm
630 mm
60 mm
1,580 mm
30 mm
20 mm

Figure 10. Measuring points of the welding-induced residual stresses both in longitudinal and
transverse directions.

3.3 Characterization of welding-induced residual stresses

Figure 11 plots the distributions of the welding-induced residual stresses in the investigated
structures that were obtained from the finite element method and direct measurements. It is shown that
finite element method solutions are in good agreement with direct measurements although the former
tends to underestimate compressive residual stresses compared with the latter. The heat-affected zone is
formed with a band width, in which the stress is approximately equal to the tensile yield stress because
the molten metal can expand freely, as a liquid, whereas after welding it quickly reverts to a solid and
the shrinkage that occurs during cooling involves “plastic flow”. It is observed that the magnitude of
tensile residual stresses in thicker plates is slightly greater than that in thinner plates.
The plate slenderness ratios in the longitudinal or transverse directions can be defined as follows:

b Y a Y
x  , y  (1)
t E t E
Figure 12 shows the changes of tensile residual stresses with changes in the plate slenderness ratios
and the weld leg length. Table 4 indicates the parameter of welding heat input according to the leg
length. The tensile residual stresses tend to decrease as the welding leg length increases and the plate
slenderness ratio increases.

6t_Longitudinal
Longitudinal DirectionDirection
for t = 6 mm 8t_Longitudinal Direction
Longitudinal Direction for t = 8 mm
350 350
Direct measurement Direct measurement
300 300
FE Analysis FE Analysis
(MPa)

(MPa)

250 250
stress[MPa]

stress[MPa]

200 200

150
ResidualStress

150
ResidualStress

100 100
Residual

Residual

50 50

0 0

-50 -50

-100 -100

-150 -150
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Distancefrom
Distance fromweld
weld center
centre line
line [mm]
(mm) Distancefrom
Distance fromweld
weld center
centre line
line [mm]
(mm)

7
 10t_Longitudinal
Longitudinal Direction Direction
for t = 10 mm 12t_Longitudinal Direction
Longitudinal Direction for t = 12 mm
350 350
Direct measurement Direct measurement
300 300
FE Analysis FE Analysis
(MPa)

(MPa)
250 250
stress [MPa]

stress [MPa]
200 200
Residual Stress

Residual Stress
150 150

100 100
Residual

50 50

0 0

-50 -50

-100 -100
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Distance
Distance fromweld
from weldcenter
centreline
line (mm)
[mm] Distancefrom
Distance fromweld
weld center
centre line
line [mm]
(mm)

Figure 11(a). Distribution of the welding-induced residual stresses in the longitudinal direction.

6t_Transverse
Transverse Direction Direction
for t = 6 mm 8t_Transverse
Transverse Direction
Direction for t = 8 mm
350 350
Direct measurement Direct measurement
300 300
FE Analysis FE Analysis
(MPa)

(MPa)

250 250
stress[MPa]

stress[MPa]

200 200
Residual Stress

ResidualStress

150 150

100 100
Residual

Residual

50 50

0 0

-50 -50

-100 -100
0 200 400 600 800 1000 1200 1400 1600 0 200 400 600 800 1000 1200 1400 1600
Distancefrom
Distance fromweld
weld center
centre line
line [mm]
(mm) Distancefrom
Distance fromweld
weld center
centre line
line [mm]
(mm)

10t_Transverse
Transverse Direction
Direction for t = 10 mm 12t_Transverse
Transverse DirectionDirection
for t = 12 mm
350 350
Direct measurement Direct measurement
300 300
FE Analysis FE Analysis
(MPa)

(MPa)

250 250
stress[MPa]
stress[MPa]

200 200
ResidualStress
ResidualStress

150 150

100 100
Residual
Residual

50 50

0 0

-50 -50

-100 -100
0 200 400 600 800 1000 1200 1400 1600 0 200 400 600 800 1000 1200 1400 1600
Distancefrom
Distance fromweld
weld center
centre line [mm]
(mm) Distancefrom
Distance fromweld
weld center
centre line
line [mm]
(mm)

Figure 11(b). Distribution of the welding-induced residual stresses in the transverse direction.

8
Table 4. Parameters of welding heat input according to leg length.

Lw Current Voltage Speed Speed Heat input

(mm) (A) (V) (CPM) (mm/s) (KJ/mm)
3.5 200 28 86 14.3 0.39
4.5 300 32 70 11.7 0.82
6.0 320 32 40 6.7 1.54
6.5 320 32 36 6.0 1.71
8.0 330 33 32 5.3 2.04

1.2

1.1
σrt,x / σY

βx=5.1640 (tp=6mm)
βx = 5.1640 (t = 6 mm)
0.9 βx = 3.8730
βx=3.8730 (tp=8mm)
(t = 8 mm)

βx = 3.0984
βx=3.0984 (tp=10mm)
(t = 10 mm)
βx = 2.5820
βx=2.5820 (tp=12mm)
(t = 12 mm)

0.8
3 4 5 6 7 8 9
Leg length, Lw (mm)

Figure 12(a). Tensile residual stress versus welding leg length in the longitudinal direction.

1.2

1.1
σrt,y / σY

βy = 20.6559
βy=20.6559 (t = 6 mm)
(tp=6mm)
0.9 β = 15.4919 (t = 8 mm)
βy=15.4919
y
(tp=8mm)
βy = 12.3935
βy=12.3935 (tp=10mm)
(t = 10 mm)
βy = 10.3280
βy=10.3280 (tp=12mm)
(t = 12 mm)
0.8
3 4 5 6 7 8 9
Leg length, Lw (mm)

Figure 12(b). Tensile residual stress versus welding leg length in the transverse direction.

9
4. Derivation of an empirical formulation
To derive the empirical formulation for predicting welding-induced residual stresses in a plate
element, their distribution is modeled, as shown in Figure 3, where it is composed of tensile residual
stress block and compressive residual stress block. In this case, an equilibrium condition must be
satisfied between tensile and compressive residual stress blocks, and thus the breadths of the tensile
residual stress zone can be defined as follows (Paik 2018):

 rcx  rcy
2bt  b , 2at  a (2)
 rcx   rtx  rcy   rty

Based on the databases obtained from both direct measurements and the finite element method
solutions, the breadths of the tensile residual stress zone can be formulated as a function of both the
plate slenderness ratio and the welding leg length as follows:

bt  c1  Lw  d1 (3)

where c1  0.6212   x2  5.9145   x  3.3894 and d1  1.9505   x2  16.234   x  18.711

at  c2  Lw  d2 (4)

where c2  0.0119   y2  0.5177   y  1.1372 and d2  0.0089   y2  0.1506   y  3.1284 .

Figure 13 shows the accuracy of Equations (3) and (4) in comparison with finite element method
solutions, where the accuracy of the coefficients of Equations (3) and (4) are confirmed in Figure 14.

80

60
bt (mm)

40

βx=5.1640(6t)
βx = 5.1640 (t = 6 mm)
20
βx=3.8730(8t)
βx = 3.8730 (t = 8 mm)
βx = 3.0984 (t = 10 mm)
βx=3.0984(10t)
βx = 2.5820 (t = 12 mm)
βx=2.5820(12t)
0
3 4 5 6 7 8 9
Leg length, Lw (mm)

Figure 13(a). Accuracy of Equation (3) for the tensile residual stress zone breadth in the transverse
direction.

10
 60

40
at (mm)

20 βy=20.6559
βy = 20.6559
(tp=6mm)
(t = 6 mm)
βy=15.4919
βy = 15.4919
(tp=8mm)
(t = 8 mm)
βy=12.3935
βy = 12.3935
(tp=10mm)
(t = 10 mm)
βy=10.3280
βy = 10.3280
(tp=12mm)
(t = 12 mm)

0
3 4 5 6 7 8 9
Leg length, Lw (mm)

Figure 13(b). Accuracy of Equation (4) for the tensile residual stress zone breadth in the longitudinal
direction.

12.0 -4.0

10.0 -6.0

y = -0.6212x2 + 5.9145x - 3.3894 -8.0

Coefficient, d1
Coefficient, c1

8.0
R² = 0.9855
-10.0 y = 1.9505x2 - 16.234x + 18.711
6.0 R² = 0.9768
-12.0
4.0
-14.0

2.0 -16.0
Coefficient,
Coefficient, cc1
1
Coefficient,
Coefficient, dd1
1
0.0 -18.0
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5

Slenderness ratio, βx Slenderness ratio, βx

Figure 14(a). Accuracy of coefficients in Equation (3).

12.0 0.0

10.0 -1.0

y = 0.0089x 2 - 0.1506x - 3.1284

Coefficient, d2
Coefficient, c2

8.0 -2.0
R² = 0.9628
6.0 -3.0
y= -0.0119x2 + 0.5177x + 1.1372
4.0 R² = 0.9997 -4.0

2.0 -5.0
Coefficient,
Coefficient, cc2
2
Coefficient,
Coefficient, d2
d2
0.0 -6.0
8.0 10.5 13.0 15.5 18.0 20.5 23.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 22.0

Slenderness ratio, βy Slenderness ratio, βy

Figure 14(b). Accuracy of coefficients in Equation (4).

Once the breadths of the tensile residual stress zone are defined from Equation (3) or (4), the
magnitude of the compressive residual stress block can be determined from the equilibrium equations,
Equation (2), as follows:

11
 2bt
 rcx    rtx (5)
2bt  b

2at
 rcy    rty (6)
2at  a

where the magnitude of tensile residual stress block may be taken as rtx  rty  Y .
Figure 15 shows the statistical error assessment of Equations (5) and (6). The mean value of the
deviations is 0.98 and the coefficient of variation is 0.01.

0.400
FEA Results Empirical Eq. (Eq./FEA)
0.104 0.103 0.984
0.350 0.152 0.150 0.984
0.214 0.214 1.001
0.237 0.233 0.984
0.303 0.298 0.986
0.300
Statics
Mean value=0.98447 0.081 0.081 0.998
Standard deviation=0.01059 0.122 0.120 0.986
0.250 0.177 0.177 1.000
(σrcx / σY) Eq.

0.215 0.210 0.977

0.262 0.262 1.002
0.200
0.069 0.067 0.965
0.103 0.100 0.966
0.154 0.149 0.970
0.150
0.188 0.185 0.984
0.235 0.232 0.987

0.100 0.058 0.056 0.975

0.086 0.085 0.983
0.132 0.129 0.978
0.050 0.155 0.152 0.983
Nominal stress
Normalized stressvalue
value 0.206 0.205 0.996
Reference line
Reference line Mean value 0.984447
0.000
Standard
0.000 0.050 0.100 0.150 0.200 0.250 0.300 0.350 0.400 deviation
0.01059

(σrcx / σY) FEA

Figure 15(a). Statistical error assessment of Equation (5).

0.160
FEA Results Empirical Eq. (Eq./FEA)
0.031 0.031 1.015
0.140 0.043 0.043 1.012
0.079 0.077 0.980
0.101 0.107 1.063
0.126 0.127 1.003
0.120
Statics
Mean value=1.008564 0.028 0.027 0.973
Standard deviation=0.031873 0.038 0.039 1.019
0.100 0.070 0.066 0.942
(σrcy / σY) Eq.

0.094 0.098 1.042

0.114 0.113 0.991
0.080
0.025 0.025 0.982
0.035 0.035 1.012
0.062 0.061 0.982
0.060
0.081 0.085 1.047
0.102 0.101 0.992

0.040 0.022 0.022 1.000

0.032 0.033 1.037
0.059 0.058 0.972
0.020 0.077 0.080 1.044
Nominal
Normalizedstress
stressvalue
value
0.094 0.100 1.061
Reference line
Reference line
Mean value 1.008564
0.000
Standard
0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160 deviation
0.031873

(σrcy / σY) FEA

Figure 15(b). Statistical error assessment of Equation (6).

12
5. Application of the empirical formulations to prevention of thermal buckling
In the shipbuilding industry, thin plate panels are often used to build less strength-critical structures,
such as the living quarters of offshore platforms, as shown in Figure 16, where the plate thickness is
less than 10 mm. In these plate panels, the welding process may cause the thermal bucking
phenomenon. Figure 17 shows a typical example of thermal buckling in thin plate panels, where the
plate thickness is 6 mm.
In the following, it is demonstrated that this study’s empirical formulations for predicting
welding-induced residual stresses are useful for designing thin plate panels that otherwise may
thermally buckle during the welding process. The target plate panels are part of the living quarters of an
offshore platform where flat bar stiffeners are typically used. These plate panels often buckle from the
welding process alone and the doorframes become significantly twisted. Considerable fairing work is
then needed to correct the deflected plate panels so that the doorframes can be fitted. The empirical
formulations developed in this study to predict welding-induced compressive residual stresses can be
applied along with plate buckling design technology for the purpose of preventing such a thermal
buckling phenomenon.
Figures 18(a) and 18(b) show the original versus modified designs of the target plate panel, where
an angle type stiffener was used in the modified design. Table 5 indicates the dimensions of the original
versus the modified plate panel. Table 6 indicates the mechanical properties of material used for the
plate panel. Figure 19 shows the maximum deflection of the original versus the modified designs for
the target plate panel. Figure 20 shows the fairing work to remove the welding-induced deflections for
the original versus the modified designs of the target plate panel. Clearly, the modified design is
superior to the original design in terms of preventing thermal buckling due to welding.

Figure 16. Living quarters of an offshore platform.

Figure 17. Thermal buckling occurred in thin plate panels with a thickness of 6 mm.

13
 FB Stiffener (125x12 FB)

900

100
3700

840 x 2060 x R50 : 125x12 FB

700
85

Figure 18(a). Typical (original) design of the target plate panel.

L2 Stiffener (125x8+100x12 L2)

900

100
3700

840 x 2060 x R50

: 125x12 FB

700
85

Figure 18(b). Modified design of the target plate panel.

Table 5. Dimensions of the original design of the plate panel.

a b t Lw hw tw bf tf
Stiffener Type
(mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm)
Longitudinal
125 12 0 0
Original (FB type)
3,700 900 6 6
Design Transverse
125 12 0 0
(FB type)
Longitudinal
125 8 100 12
Modified (L2 type)
3,700 900 6 6
Design Transverse
125 12 0 0
(FB type)
14
Table 6. Mechanical properties of material used for the plate panels.

Kind of material E (GPa) Y (MPa) u (MPa)   f (mm/mm)

Mild steel (Grade A) 200 309 385 0.32 0.23

Original design Modified design

(FB Stiffener) (L2 Stiffener)
Guide
ruler Guide
ruler

Max. deflextion Max. deflection

= 10 mm = 2 mm

Wall plate = 6mm Wall plate = 6mm

Figure 19. Distortion of the doorframes due to welding in the original versus modified designs for the
target plate panel.

Original design Modified design

(FB Stiffener) (L2 Stiffener)

vs.

Fairing

Figure 20. Fairing work of the doorframes to remove the welding-induced deflections for the original
versus modified designs of the target plate panel.

When the longer direction of the plate panel is taken as the x direction, the plate slenderness
ratios are calculated from Equation (1) as follows:
x  5.89 , y  24.24 (7)

15
 The welding leg length applied for building the structures is Lw  6.0mm , and thus the breadths
of the heat-affected zones can be defined from Equations (3) and (4) as follows:
bt  50.13 mm , a t  38.62 mm (8)

Therefore, the compressive residual stresses developed in the plate panels can be estimated from
Equations (5) and (6) as follows:
rcx  38.74 MPa , rcy  6.59 MPa (9)

Under biaxial compressive loading, the elastic buckling strength interaction relationship is given as
follows (Paik 2018):

 2
 x 
1
 y 
     1 (10)
  xE ,1    yE ,1 

where,

1   2  1 for 1  a / b  2

a
3
a
2
a 
1  0.0293    0.3364    1.5854    1.0596 
b   b   b   
3 2  for a / b  2
a a a 
 2  0.0049    0.1183    0.6153    0.8522 
 
b  
b  
b 

In Equation (10),  xE ,1 and  yE ,1 are the buckling strengths of the plate under uniaxial axial

compressive load in the x and y directions, respectively, which are defined by

 2E  t   2E  t 
2 2

 xE ,1  k x   ,  yE ,1  k y   (11)
12 1  2   b  12 1  2   b 

In Equation (11), kx and k y are the buckling coefficients of the elastic buckling strengths under

uniaxial compressive loading in the x and y directions, respectively, which can be obtained by
taking account of the rotational restraint effects associated with torsional rigidity of support members
(stiffeners) as follows (Paik 2018):

kx  kx1  kx 2  kx 0 , k y  k y1  k y 2  k y 0 (12)

where,
kx =buckling coefficient of the plate elastically rotation-restrained at both the long and short edges,
k x1 =buckling coefficient of the plate elastically rotation-restrained at the long edges and simply
supported at the short edges,
k x 2 =buckling coefficient of the plate elastically rotation-restrained at the short edges and simply
supported at the long edges,
k x 0 =buckling coefficient of the plate simply supported at all edges,
k y =buckling coefficient of the plate elastically rotation-restrained at both the long and short edges,
k y1 =buckling coefficient of the plate elastically rotation-restrained at the long edges and simply
supported at the short edges,
k y 2 =buckling coefficient of the plate elastically rotation-restrained at the short edges and simply
16
supported at the long edges, and
k y 0 = the buckling coefficient of the plate simply supported at all edges.
The buckling coefficients in Equation (12) are a function of the rotational restraint parameters that
are dependent on the torsional rigidities of the support members, where the rotational restraint
parameters are defined as follows (Paik 2018):
GJ L GJ S
L  , S  (12)
bD aD

E Et 3
3
hwx twx  b fx t 3fx 3
hwy twy  b fy twy
3

where G  , D , J  , and J  .
2 1    12 1   2 L
 3
S
3
For the original design,  L  1.19 and  S  0.38 , and thus the buckling coefficients of plate
panels, taking into account the effect of rotational restraints, are calculated as follows:
kx  kx1  kx2  kx0  6.12  4.19  4.0  6.31 , k y  k y1  k y2  k y0  1.38  1.13  1.12  1.39 (13)

The buckling strengths of the original plate under uniaxial compressive residual stresses are
calculated from Equation (11) as follows:
xE,1  50.74 MPa , yE,1  11.20 MPa (13)

In contrast, the values of the rotational restraint parameters for the modified design are calculated
as  L  1.71 and  S  0.38 because a L2 type stiffener is now used instead of the flat bar used for
the original design. Therefore, the buckling coefficients of plate panels, taking into account the effect
of rotational restraints, are now calculated as follows:
k x  k x1  k x2  k x0  6.30  4.20  4.0  6.50 , k y  k y1  k y2  k y0  1.49  1.13  1.12  1.50 (14)

The buckling strengths of the modified plate under uniaxial compressive residual stresses are
calculated from Equation (11) as follows:
xE ,1  52.23 MPa , yE,1  12.02 MPa (15)

1.01.0

0.80.8

Original design
0.60.6
Modified design

0.40.4

0.20.2

0.0
0.2 0.4 0.6 0.8 1.0
0.0 0.2 0.4 0.4 0.4 1.0
 rcx /  xE ,1
Figure 21. Elastic buckling strength interaction relationship of the plate panel under combined
welding-induced compressive residual stresses in the x and y directions.

Equations (13) and (14) show that the elastic buckling strengths are increased by 2.9% and 7.3% in
17
the x and y directions, respectively. Figure 21 ascertains that the modified design of the target plate
panel can prevent thermal buckling because the welding-induced compressive residual stress
components in the x and y directions can be controlled inside the buckling strength interaction
relationship, while the original design presents them outside of that relationship, thus causing thermal
buckling. It is interesting to note that the plate panels of living quarter structures are
non-strength-critical members and the welding-induced initial deformations can be effectively reduced
by accurate predictions of welding-induced residual stresses and application of the advanced buckling
design technology.

6. Concluding remarks
This paper aims to derive an empirical formulation for predicting welding-induced residual stresses
in steel plated structures. The databases for welding-induced residual stresses on full-scale models of
steel stiffened plate structures were obtained from direct measurements and finite element method
computations. Based on this study, the following conclusions can be drawn.
(1) Elastic-plastic thermal stress finite element method computations are in good agreement with
direct measurements, although the former tends to slightly underestimate welding-induced residual
stresses in comparison with the latter.
(2) As well recognized, the distributions of welding-induced residual stresses are composed of both
tensile residual stresses in the heat-affected zone and compressive residual stresses in the middle of
the plate.
(3) The magnitude of the tensile residual stresses in the heat-affected zone reaches the yield stress of
the material. It tends to increase as the plate thickness increases.
(4) As the weld leg length increases, the breadth of the tensile residual stress block increases while
the magnitude of the tensile residual stresses in the tensile residual stress block tends to decrease.
(5) In this study, the breadths of the tensile residual stress block in the length and breadth directions
of plate elements were formulated as a function of the plate slenderness ratio and the weld leg length.
From the equilibrium between tensile and compressive residual stresses, the magnitude of
welding-induced compressive residual stresses can be determined once the breadths of the tensile
residual stress block are defined where the magnitude of tensile residual stresses is assumed to equal
the yield stress of steel.
(6) As an applied example, the developed formulations were applied to prevent the thermal buckling
of the thin plate panels that are part of the living quarters of an offshore platform.
(7) It is considered that the developed formulations can be useful for predicting the magnitude and
pattern of welding-induced residual stresses in welded plate panels.

References

Bruno, G., Yordan. G. & Guedes Soares, C. (2011). Effect of weld shape imperfections on the
structural hot-spot stress distribution, Ships and Offshore Structures, 6(1-2): 145-159.
Chen, B.Q., Hashemzadeh, M. & Guedes Soares, C. (2018). Validation of numerical simulations with
X-ray diffraction measurements of residual stress in butt-welded steel plates, Ships and Offshore
Structures, 13(3): 273-282.
Cheng, J.J.R., Elwi, A.E., Grodin, G.Y. & Kulak, G.L. (1996). Material testing and residual stress
measurements in a stiffened steel plate. In: Strength and Stability of Stiffened Plate Components.
Ship Structure Committee, SSC-399, Washington, DC.

18
Eggert, L., Fricke, W. & Paetzhold, H. (2012). Fatigue strength of thin-plated block joints with typical
shipbuilding imperfections. Welding in the World, 56(11-12): 119-128.
Farajkhah, V., Liu, Y. & Gannon, L. (2016). Finite element study of 3D simulated welding effect in
aluminium plates, Ships and Offshore Structures, 12(2): 196-208.
Gannon, L.G., Liu, Y., Pegg, N.G. & Smith, M.J. (2012a). Effect of three-dimensional
welding-induced residual stress and distortion fields on strength and behaviour of flat-bar stiffened
panels, Ships and Offshore Structures, 8(5): 565-578.
Gannon, L.G., Pegg, N.G., Smith, M.J. & Liu, Y. (2012b). Effect of residual stress shakedown on
stiffened plate strength and behaviour, Ships and Offshore Structures, 8(6): 638-652.
Gannon, L., Liu, Y., Pegg, N. & Smith, M.J. (2015). Nonlinear collapse analysis of stiffened plates
considering welding-induced residual stress and distortion, Ships and Offshore Structures, 11(3):
228-244.
Kavanagh, B.F. & Glenn Bird, S.J. (1996). Surveying principles and applications (4 ed.). Prentice Hall,
pp. 257-264.
Kenno, S.Y., Das, S., Kennedy J, Rogge, R. & Gharghouri, M. (2010). Distributions of residual
stresses in stiffened plates with one and two stiffeners, Ships and Offshore Structures, 5(3):
211–225.
Kenno, S.Y., Das, S., Rogge, R.B. & Gharghouri, M. (2017). Changes in residual stresses caused by an
interruption in the weld process of ships and offshore structures, Ships and Offshore Structures,
12(3): 341–359.
Khan, I. & Zhang, S. (2011). Effects of welding-induced residual stress on ultimate strength of plates
and stiffened panels, Ships and Offshore Structures, 6(4): 297-309.
Khedmati, M.R., Pedram, M. & Rigo, P. (2012). The effects of geometrical imperfections on the
ultimate strength of aluminium stiffened plates subject to combined uniaxial compression and
lateral pressure, Ships and Offshore Structures, 9(1): 88-109.
Lillemäe, I., Liinalampi, S., Remes, H., Itävuo, A. & Niemelä, A. (2017). Fatigue strength of thin
laser-hybrid welded full-scale deck structure, International Journal of Fatigue, 95: 282-292.
Luís, R.M., Soares, C.G. & Nikolov, P.I. (2009). Collapse strength of longitudinal plate assemblies
with dimple imperfections, Ships and Offshore Structures, 3(4): 359-370.
Masubuchi, K. (1980). Analysis of welded structures: Residual stresses, distortion and their
consequences. Pergamon Press, England & Oxford, U.K.
Matsui, S. (1983). Control of weld distortion in thin-skin welded structures, Journal of Welding and
Joining, 53(5): 58-66.
MSC (2013). Online User’s Guide. MSC Software Corporation, California, USA
Paik, J.K. (2007a). Practical techniques for finite element modeling to simulate structural
crashworthiness in ship collisions and grounding (Part I: Theory), Ships and Offshore Structures,
2(1): 69-80.
Paik, J.K. (2007b). Practical techniques for finite element modeling to simulate structural
crashworthiness in ship collisions and grounding (Part II: Verification), Ships and Offshore
Structures, 2(1): 81-85.
Paik, J.K. (2007c). Characteristics of welding induced initial deflections in welded aluminum plates,
Thin-Walled Structures, 45: 493-501.
Paik, J.K. (2008). Mechanical collapse testing on aluminum stiffened panels for marine applications.
Ship Structure Committee, SSC-451, Washington, DC.

19
Paik, J.K. (2018). Ultimate limit state analysis and design of plated structures. Second Edition,
Chichester, UK.
Paik, J.K., Andrieu, C. & Cojeen, H.P. (2008). Mechanical collapse testing on aluminum stiffened plate
structures for marine applications, Marine Technology, 45(4): 228-240.
Paik, J.K. & Frieze, P.A. (2001). Ship structural safety and reliability, Progress in Structural
Engineering and Materials, 3: 198-210.
Paik, J.K. & Kim, D.K. (2012). Advanced method for the development of an empirical model to predict
time-dependent corrosion wastage, Corrosion Science, 63: 51-58.
Paik, J.K., Kim, K.J., Lee, J.H., Jung, B.G. & Kim, S.J. (2017). Test database of the mechanical
properties of mild, high-tensile and stainless steel and aluminum alloy associated with cold
temperatures and strain rates, Ships and Offshore Structures, 12(sup1): 1-27.
Paik, J.K., Kim, B.J., Sohn, J.M., Kim, S.H., Jeong, J.M. & Park, J.S. (2012). On buckling collapse of a
fusion-welded aluminum stiffened plate structure: An experimental and numerical study, Journal of
Offshore Mechanics and Arctic Engineering, 134: 021402.1-8.
Paik, J.K., Lee, J.M., Hwang, J.S. & Park, Y.I. (2003a). A time-dependent corrosion wastage model for
the structures of single- and double-hull tankers and FSOs and FPSOs, Marine Technology, 40(3):
201-217.
Paik, J.K. & Melchers, R.E. (2008). Condition assessment of aged structures. CRC Press, New York.
Paik, J.K. & Pedersen, P.T. (1996). A simplified method for predicting the ultimate compressive
strength of ship panels, International Shipbuilding Progress, 43: 139-157.
Paik, J.K. & Thayamballi, A.K. (2007). Ship-shaped offshore installations: Design, building, and
operation. Cambridge University Press, Cambridge.
Paik, J.K., Thayamballi, A.K., Park, Y.I. & Hwang, J.S. (2003). A time-dependent corrosion wastage
model for bulk carrier structures, International Journal of Maritime Engineering, 145(A2): 61-87.
Paik, J.K., Thayamballi, A.K., Ryu, J.Y., Jang, J.H., Seo, J.K., Park, S.W., Seo, S.K., Andrieu, C.,
Cojeen, H.P. & Kim, N.I. (2006). The statistics of weld induced initial imperfections in aluminum
stiffened plate structures for marine applications, International Journal of Maritime Engineering,
148(Part A1): 19-63.
Paik, J.K. & Yi, M.S. (2006). Experimental and Numerical Investigations of Welding-Induced
Distortions and Stresses in Steel Stiffened Plate Structures. The Korea Ship and Offshore Research
Institute, Pusan National University, Busan, South Korea.
Smith, C.S., Davidson, P.C., Chapman, J.C. & Dowling, P.J. (1988). Strength and stiffness of ships’
plating under in-plane compression and tension, Transactions of the Royal Institution of Naval
Architects, 130: 277-296.
Teresa, M. & Craig, F. (2013). Effect of weld-induced imperfections on the ultimate strength of an
aluminium patrol boat determined by the ISFEM rapid assessment method, Ships and Offshore
Structures, 9(2): 218-235.
Ueda, Y. (1999). Computational Welding Mechanics (A volume of selected papers in the
commemoration of the retirement from Osaka University). Joining and Welding Research Institute,
Osaka University, Osaka, Japan.
Vhanmane, S. & Bhattacharya, B. (2008). Estimation of ultimate hull girder strength with initial
imperfections, Ships and Offshore Structures, 3(3): 149-158.

20
Wang, J., Han, J., Domblesky, J.P., Li, W., Yang, Z. & Zhao. Y. (2015). Predicting distortion in butt
welded plates using an equivalent plane stress representation based on inherent shrinkage volume,
Journal of Manufacturing Science and Engineering, 138(1): 011012.
Xstress 3000 Manual (2010). Stresstech Inc., Finland.
Yi, M.S., Hyun, C.M., Cho, S.H. & Jang, T.W. (2009). A study of correction effect in T-bar according
to cooling method, Proceeding of the 2009 Spring Annual Meeting of Korea Welding Society (in
Korean).
Yi, M.S., Hyun, C.M. & Paik, J.K. (2018a). Three-dimensional thermo-elastic-plastic finite element
method modeling for predicting weld-induced residual stresses and distortions in Steel
stiffened-plate structures, World Journal of Engineering and Technology, 6:176-200
Yi, M.S., Hyun, C.M. & Paik, J.K. (2018b). Full-scale measurements of welding-induced initial
imperfections in steel-stiffened plate structures, International Journal of Maritime Engineering,
Accepted for publication.

21