Documente Academic
Documente Profesional
Documente Cultură
Effect of the weld groove shape and pass number on residual stresses
in butt-welded pipes
I. Sattari-Far*, M.R. Farahani
Faculty of Mechanical Engineering, Amirkabir University of Technology, Tehran, Iran
a r t i c l e i n f o a b s t r a c t
Article history: This study used finite element techniques to analyze the thermo-mechanical behaviour and residual
Received 7 August 2008 stresses in butt-welded pipes. The residual stresses were also measured in some welds by using the Hole-
Received in revised form Drilling method. The results of the finite element analysis were compared with experimentally measured
11 May 2009
data to evaluate the accuracy of the finite element modelling. Based on this study, a finite element
Accepted 15 July 2009
modelling procedure with reasonable accuracy was developed. The developed FE modelling was used to
study the effects of weld groove shape and weld pass number on welding residual stresses in butt-
Keywords:
welded pipes. The hoop and axial residual stresses in pipe joints of 6 and 10 mm thickness of different
Residual stresses
Elasto-plastic analysis groove shapes and pass number were studied. It is shown that these two parameters may have signif-
Groove shape icant effects on magnitude and distribution of residual stresses in welded pipes.
Pass number Ó 2009 Elsevier Ltd. All rights reserved.
Butt weld
Finite element method
0308-0161/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2009.07.007
724 I. Sattari-Far, M.R. Farahani / International Journal of Pressure Vessels and Piping 86 (2009) 723–731
weld groove shape on welding residual stress in but welded plate temperature of the radiation heat source. The radiation heat
[10]. Furthermore, there are very limited studies on the effect of transfer coefficient is expressed as:
weld pass number on welding residual stresses independent from
others geometrical and welding parameters. hr ¼ seF T 2 þ Tr2 ðT þ Tr Þ (6)
In this work, the development of welding residual stresses in
steel pipes is studied by using 3-D finite element method (FEM). where s is the Stefan–Boltzmann constant, e is the effective emis-
The FEM results are compared with experimentally measured data sivity and F is a configuration factor.
obtained from the hole-drilling method. Then, effects of the weld Solving Eq. (3) by considering the boundary conditions
groove shape on magnitude and distributions of residual stresses expressed in Eqs. (4) and (5) gives the temperature distribution in
are studied. The most common weld groove shapes of, V-groove, the body. This temperature field will then be applied to the
X-groove and U-groove types are studied here. In addition, effects mechanical model to calculate the welding residual stresses and
of weld pass number on welding residual stresses are studied by strains (distortions).
models considering three different pass numbers. In each study
other impacting parameters are kept fix. 2.2. Elasto-plastic mechanical analysis
2. Theoretical aspects The equilibrium and constitutive equations used here in con-
ducting the mechanical analysis are as follow [11].
Welding residual stresses are calculated here by using finite
element method. Theoretical considerations of the thermal and Equilibrium equations:
mechanical analyses are briefly described below.
sij; j þ rbi ¼ 0 (7)
2.1. Transient thermal analysis Here, sij is the stress tensor and bi is the body force. It is assumed
that the stress tensor is symmetrical, i.e. sij ¼ sji.
When a volume is bounded by an arbitrary surface, the balance
relation of the heat flow is expressed by: Constitutive equations:
vRx vRy vRz vTðx; y; z; tÞ The thermal elasto-plastic material model, based on the von
þ þ þ Q ðx; y; z; tÞ ¼ rC (1)
vx vy vz vt Mises yield criterion and the isotropic strain hardening rule, is
considered. Stress–strain relations are expressed as:
where Rx, Ry and Rz are the rates of heat flow per unit area, T(x,y,z) is
the current temperature, Q(x,y,z) is the rate of internal heat h i
generation, r is the density, C is the specific heat and t is the time. ½ds ¼ Dep ½d3 C th dT (8a)
The model can then be completed by introducing the Fourier heat
ep
flow as D ¼ De þ Dp (8b)
vT e p
where [D ] is the elastic stiffness matrix, [D ] is the plastic stiffness
Rx ¼ kx (2a)
vx matrix, [Cth] is the thermal stiffness matrix, ds is the stress incre-
ment, d3 is the strain increment and dT is the temperature
vT increment.
Ry ¼ ky (2b)
vy Since thermal elasto-plastic analysis is a non-linear problem, the
incremental calculation technique is employed here in solving the
vT problem. The incremental stress is obtained by using the full
Rz ¼ kz (2c) Newton–Raphson method [1].
vz
where kx, ky and kz are the thermal conductivities in the x, y and z 3. Experimental investigation
directions, respectively. Generally, the material parameters kx, ky, kz,
r and C are temperature dependent. Inserting Eq. (2) into Eq. (1) Experimental tests were conducted to collect data on residual
yields: stresses on the outside surfaces of butt-welded pipes. The experi-
mental results were used to verify the finite element model.
v vT v vT v vT vT
kx þ ky þ kz þ Q ¼ rC (3)
vx vx vy vy vz vz vt
3.1. Specimen preparation
Eq. (3) is the governing differential equation of the problem. The
general solution is obtained by applying the following initial and Two pipes, with outer diameter of 320 mm, thickness of 10 mm
boundary conditions: and length of 500 mm were prepared with V-grooved edges and
fitted by using six equal-distance tack welds from the start point of
Tðx; y; z; 0Þ ¼ T0 ðx; y; zÞ (4) welding. Dimensions of the weld preparation are shown in Fig. 3.
Annealing was carried out to relieve initial residual stresses in pipes
vT vT vT before welding.
kx Nx þky Ny þ kz Nz þ qs
vx vy vz Welds were carried out in a robotic TIG welding system with
þ hc ðT TN Þ þ hr ðT Tr Þ ¼ 0 ð5Þ advanced control system. This system controls the Power source,
Gripper chuck, Torch driving vehicle, Inert gas supplier and Auto-
where Nx, Ny and Nz are the direction cosines of the outward drawn matic wire feeder simultaneously. The schematic view of this
normal to the boundary, hc is the convection heat transfer coeffi- automatic circumferential welding system is shown in Fig. 1. Three
cient, hr is the radiation heat transfer coefficient, qs is the boundary weld beads were deposited to complete the weld and then one
heat flux, TN is the surrounding temperature and Tr is the pass without filler used to make a uniform weld surface. Argon
I. Sattari-Far, M.R. Farahani / International Journal of Pressure Vessels and Piping 86 (2009) 723–731 725
Table 1
Welding parameters for each pass.
200
theta=0
150
theta=90
100
theta=180
0
0 20 40 60 80 100
-50
-100
-150
-200
Distance from the weld center line (mm)
Fig. 6. Distribution of the hoop residual stress in the axial direction on the outer
surface of the pipe.
150
100
50
Axial Stress (Mpa)
0
0 20 40 60 80 100
-50
theta=0
-100
theta=90
-150
theta=180
-200 theta=270
-250
-300
Distance from the weld center line (mm)
Fig. 7. Distribution of the axial residual stress in the axial direction on the outer
Fig. 5. Finite element mesh used for the analysis of butt-welded joints. surface of the pipe.
I. Sattari-Far, M.R. Farahani / International Journal of Pressure Vessels and Piping 86 (2009) 723–731 727
pffiffiffi 0
6 3ff Q ð3x2 =a2 Þ ð3y2 =b2 Þ ð3x2 =c2 Þ 0 20 40 60 80 100
qf ¼ pffiffiffi e e e (11a)
abcf p p -50
pffiffiffi
6 3fr Q ð3x2 =a2 Þ ð3y2 =b2 Þ ð3z2 =c2 Þ -150
qr ¼ pffiffiffi e e e (11b)
abcr p p
-200
where x, y and z are the local coordinates of the double ellipsoid Distance from the weld center line (mm)
model aligned with the welded pipe; ff and fr are parameters which
Fig. 8. Experimental and FEM results of the hoop residual stresses on the outer surface
give the fraction of the heat deposited in the front and rear parts,
of the studied pipe.
respectively. Here it is assumed that ff is 1.5 and fr is 0.5. This is based
on the fact that the temperature gradient in the front leading part is
steeper than in the tailing edge. Q is the power of the welding heat
source. It can be calculated by knowing the welding current and 4.3. Mechanical analysis
voltage and considering the arc efficiency. The robotic welding used
in this study gave precise recording of these parameters during In the mechanical analysis, the temperature history obtained
welding. The arc efficiency of the TIG welding process is assumed to from the thermal analysis is input into the structural analysis as
be 70% according to the catalogue of the welding machine used. a thermal loading. Thermal strains and stresses are then calculated
The parameters a, b and c in Eq. (11a,11b) are related to the at each time increment. First order elements were used in a non-
characteristics of the welding heat source. These parameters are linear large displacement mechanical analysis. It is assumed that no
adjusted to create a desired melted zone according to the welding solid-state phase transformation occurs for stainless steels. During
conditions in each model. The moving heat source is modeled by the welding process, the total strain rate can be decomposed into
a user subroutine in the ANSYS code. three components as follows:
In ANSYS, latent heats of phase changes in thermal analysis can
be handled using thermal enthalpy method. In this paper, equiva- 3 ¼ 3e þ 3p þ 3th (14)
lent specific heat method is adopted to deal with latent heat. The components on the right hand side of Eq. (14) correspond to
Enthalpy can be expressed as: elastic strain, plastic strain and thermal strain, respectively.
Z The welding residual stresses and strains (distortions) are the
H ¼ r cðTÞdT (12) accumulated results at the final stage of the calculation, when the
whole model is cooled down. The materials are assumed to follow
The temperature difference of solid phase and liquid phase is DT, the von Mises yield criterion and flow rule. Isotropic bilinear strain
so the equivalent specific heat is considered to be C* ¼ H/rDT, as hardening is assumed in the calculations.
proposed in Ref. [14]. The equivalent specific heat is assumed By studying the development of stresses in the whole model, it
a constant in the region between solid phase and liquid phase. is observed that the stresses at the start point of the weld and its
The thermal boundary conditions include the radiation and vicinity are different from the other locations. In Figs. 6 and 7,
convection to the environment from all sides of the welded pipe distributions of the hoop and axial residual stresses in the axial
except the symmetry surface and the area upon which the heat is
applied. Radiation losses are dominating for higher temperatures
near and in the weld zone, and convection losses for lower 150
temperatures away from the weld zone. A user subroutine was
100
developed to simulate the combined thermal boundary condition.
The total temperature-dependent heat transfer (w/m2) coefficient 50
is given by [4].
Axial Stress (Mpa)
0
0 20 40 60 80 100
ah ¼ 0:0668T for 0 < T 500
(13) -50
ah ¼ 0:231T 82:1 for T > 500
-100
where T ( C) is the temperature. Above the melting point, the
thermal conductivity is assumed to be doubled to simulate the -150 EXP FEM
convective stirring effect in a molten material [14].
-200
By comparing the temperature versus time at different points in
the circumferential direction (q ¼ 90 , 180 , 270 ) and the same -250
distance from the weld centre line on the inner surface, it could be
-300
concluded that the temperature filed was very steady when the
Distance from the weld center line (mm)
welding torch was moving around the pipe. From this analysis it
was observed that the maximum temperature at the weld pool was Fig. 9. Experimental and FEM results of the axial residual stresses on the outer surface
around 2000 C. of the studied pipe.
728 I. Sattari-Far, M.R. Farahani / International Journal of Pressure Vessels and Piping 86 (2009) 723–731
300
250 V
200
X1
100
X2
Fig. 10. Schematics of three different weld groove shapes studied here. 50
350 200
V V
300
X1 150 X1
250
200 U 100 U
Hoop Stress (Mpa)
150
X2 50 X2
100
0
50
0 5 10 15 20 25 30 35 40 45 50
0 -50
0 5 10 15 20 25 30 35 40 45 50
-50
-100
-100
-150 -150
Distance from the weld center line (mm) Distance from the weld center line (mm)
Fig. 11. Distribution of the hoop residual stress in the axial direction on the inner Fig. 13. Distribution of the hoop residual stress in the axial direction on the outer
surface of the 10-mm pipes in the four-pass weld. surface of the 10-mm pipes in the four-pass weld.
I. Sattari-Far, M.R. Farahani / International Journal of Pressure Vessels and Piping 86 (2009) 723–731 729
100 350
V
50
250
X1
0
Axial Stress (Mpa)
-200 -150
X2
-250 -250
Distance from the weld center line (mm) L (mm)
Fig. 14. Distribution of the axial residual stress in the axial direction on the outer Fig. 16. Distribution of the axial residual stresses through the wall thickness of the 10-
surface of the 10-mm pipes in the four-pass weld. mm pipes in the four-pass weld.
6-mm thick welded pipes has no significant effect on the magni- weld centre line, the hoop stresses at the other points are lower
tude of the hoop residual stresses. than the V and U models.
Figs. 15 and 16 show a comparison between the hoop and axial
residual stresses through wall thickness of the 10-mm pipe at the
5.2. 10-mm Thick pipe model weld centre line in the four-pass weld. The magnitudes of residual
stresses in the middle of pipe thickness for the X2 and V models are
In these models two pipes with 10 mm thickness were studied the minimum and maximum, respectively. The axial residual
with three different groove shapes that are schematically shown in stresses in this region are tensile for the V models, while they are
Fig. 10. These joints were welded in four passes. For the X-groove compressive for the X2 model.
shape, two different pass sequences are studied, denoted as X1 and
X2 as shown in Fig. 10. 6. Study of weld pass number
Figs. 11 and 12 show the comparison between the hoop and
axial residual stresses of these four models at the inner surface of Effect of weld pass number on residual stresses distribution is
the pipes. Because of very little difference between stress distri- studied for two different pipe thicknesses of 6 mm and 10 mm.
butions in each pipe part, only stresses in one half of the welded Heat input magnitude for each pass was adjusted to get a melted
pipes are shown. zone similar to the expected form, as shown in Fig. 3.
It is observed that all four models have almost the same peak in
the hoop stresses. The compressive hoop stress in the V-model is 6.1. 6-mm Thick pipe model
greater than others. It is also observed that the maximum of the
axial stresses in the weld zone of these different shapes occur in the A welded pipe of 6 mm thickness was studied with three
X2 model. different pass numbers (2, 4 and 6 passes), as shown in Fig. 17.
Figs. 13 and 14 show the results of the hoop and axial residual Comparison between the hoop residual stresses of these three
stresses of these four models at outer surface of pipes. Except for pass numbers in the inner surfaces of the pipes is shown in Fig. 18. It
a little increase in the hoop stresses for the X1 and X2 models in is observed that, the peak of the tensile hoop residual stresses at
inner surface of the weld decreases with increasing pass number.
As a result of comparison among these three models, it is
350
observed that pass number has a minor effect on the magnitude
300
Hoop Stress (Mpa)
V
250
X1
200
U
150 X2
100
0 1 2 3 4 5 6 7 8 9 10
L (mm)
Fig. 15. Distribution of the hoop residual stresses through the wall thickness of the 10-
Fig. 17. Schematics of different weld pass numbers.
mm pipes in the four-pass weld.
730 I. Sattari-Far, M.R. Farahani / International Journal of Pressure Vessels and Piping 86 (2009) 723–731
350 axial residual stress in the outer surface of the 4-pass model is
about 30% higher than in the other two models.
300
Figs. 21 and 22 show a comparison between the hoop and axial
2-Pass
250 residual stresses through wall thickness of the 10-mm pipe of
Hoop Stress (Mpa)
200
4-Pass
150 100
100 6-Pass 50
50
0
Fig. 18. Distribution of the hoop residual stresses in the axial direction on the inner -200 13-Pass
surface of the 6-mm pipes of different pass number.
-250
and distribution of residual stresses on the inner surface along the
Distance from the weld center line (mm)
axial direction.
Furthermore, it is observed that pass number in these models Fig. 20. Distribution of the axial residual stress in the axial direction on the outer
has no significant effect on the magnitude of the residual stresses surface of the 10-mm pipes.
on the outer surface of the welded pipe.
350
6.2. 10-mm Thick pipe model
250 250
4-Pass
200
4-Pass
200
150 9-Pass
150 9-Pass 100
Axial Stress (Mpa)
Axial Stress (Mpa)
50 13-Pass
100
13-Pass 0
0 1 2 3 4 5 6 7 8 9 10
50 -50
-100
0
0 5 10 15 20 25 30 35 40 45 50 -150
-5 0 -200
-250
-100
Distance from the weld center line (mm) L (mm)
Fig. 19. Distribution of the axial residual stress in the axial direction on the inner Fig. 22. Distribution of the axial residual stresses through the wall thickness of the 10-
surface of the 10-mm pipes of different pass number. mm pipes of different pass number.
I. Sattari-Far, M.R. Farahani / International Journal of Pressure Vessels and Piping 86 (2009) 723–731 731
different pass numbers at the weld centre line. The magnitude of At the weld centre line in thick pipes (10 mm thick) the
residual stresses in the middle of pipe thickness increased when magnitude of residual stresses in the middle of pipe thickness
the pass number decreased. The axial residual stresses in this increased when the pass number decreased.
region are tensile for the 4-pass model, while they are compressive
for the 9- and 13-pass models.
References