International Journal of Recent Technology and Engineering (IJRTE)

ISSN: 2277-3878, Volume-2, Issue-2, May 2013

Analysis of Residual Stresses and Distortions in

Girth-Welded Carbon Steel Pipe
Prabhat Kumar Sinha, Raisul Islam, Chandan Prasad, Mohd. Kaleem

Abstract-This article, the weld joint suffers various types of weld- Numerical modelling using the finite element method
induced residual stress fields (hoop and axial) and deformation (FEM) has been used by many researchers to predict
patterns (axial shrinkage, radial shrinkage). In this paper weld residual stresses in complex welded structures.
Three-dimensional finite element modeling of residual To date, most of the finite element analysis[22] for pipe
stresses in a girth-welded carbon steel pipe is presented with
girth welds use the inherent shrinkage modelling
an emphasis on modeling procedures for the global residual
stress characteristics. To precisely capture the distortions and technique that assumes an axisymmetric condition.
residual stresses, computational methodology based on three- This is due to the fact that the three-dimensional,
dimensional finite element model for the simulation of gas nonlinear FEM analysis of girth welds with a moving
tungsten arc welding in thin-walled pipe is presented. Butt-weld heat source is computational intensive and sometimes cost
geometry with single "V" for a 300 mm outer diameter cylinder prohibitive. The inherent shrinkage model is incapable of
of 3 mm thick is used. The complex phenomenon of arc welding predicting the transient residual stress distributions near the
is numerically solved by sequentially coupled transient, non- weld start and stop locations.
linear thermo-mechanical analysis. The accuracy of both the The complex nature of the welding process due to multi-
thermal and structural models is validated through experiments
field (thermal, mechanical, metallurgy etc.) interactions and
for temperature distribution, residual stresses and distortion. The
simulated result shows close correlation with the experimental intricate geometries in real world applications has made the
measurements. prediction of weld-induced imperfections, a truly difficult
and computationally intensive task. However, with the
Keywords: FEM; welding simulations; Distortions; Residual availability of 64 digit computers and refined FE tools,
Stresses; Girth Weld. welding engineers around the world are more biased
towards the computer simulations of complex welding
I. INTRODUCTION phenomenon instead of the conventional trial and error
The distribution of residual stresses in a girth welded pipe is approach on the shop floor. A significant simulation and
complex. Weld shrinkage in the circumferential direction experimental work focusing on circumferential welding is
induces both shearing and bending that result in stress available in the literature [6-13]. As the computer simulation
components in the circumferential direction (hoop of welding processes is highly computationally intensive
stress) and in the axial direction (meridian stress). and large computer storage and CPU time are required, most
Process and geometric related factors that influence of the previous research reduces computational power
residual stresses include welding heat input, pipe requirements by simplifying with assumptions such as
diameter, wall thickness and joint design [1-5].Brust, et rotational symmetry and lateral symmetry in numerical
al. [2], reported high tensile stresses in both axial and simulations [7-10]. These assumptions reduces the
hoop directions on the pipe inner surface using the computational demand at the cost of the accuracy of the
axisymmetric, inherent shrinkage model. However, results because the model was over simplified by limiting
discrepancies in stress magnitude were found between the solution domain to only a section of the whole do-main
the predicted results and the experimentally measured with forced symmetry assumptions which did not prevails.
data, particularly in the axial component in the outer Further, these simplified assumptions are not capable of
surface. capturing the considerable effects of weld start/stop and
weld tack modeling. In this regard an experimental work by
Jonsson and Josefson [16] and some three-dimensional finite
element (FE) studies [12, 13, 15, 24]; reported deviations
from rotational symmetry, especially at the beginning and
end of the welding cycle for circumferential joint in welding
Manuscript received on May, 2013. of pipes with lateral symmetry. Later, by using a full three-
Prabhat Kumar Sinha, Mechanical Engineering Department Shepherd dimensional model for multi-pass welding of pipes, Fricke
School of Engineering and Technology Sam Higginbottom Institute of et al. [16] concluded that residual stresses[21] are by no
Agriculture, Technology and Sciences (Formerly Allahabad Agriculture
Institute) Allahabad 211007, India. means axis-symmetric.
Raisul Islam, Mechanical Engineering Department Shepherd School of In the present analysis, the temperature fields and
Engineering and Technology Sam Higginbottom Institute of Agriculture, consequently the weld-induced residual stress fields and
Technology and Sciences (Formerly Allahabad Agriculture Institute) distortion patterns are investigated by numerical simulations
Allahabad 211007, India.
Chandan Prasad, Mechanical Engineering Department Shepherd based on FEM modeling. Tomodel the physics behind the
School of Engineering and Technology Sam Higginbottom Institute of gas tungsten arc welding (GTAW) process, a sequentially
Agriculture, Technology and Sciences (Formerly Allahabad Agriculture coupled,
Institute) Allahabad 211007, India.
Mohd. Kaleem, Mechanical Engineering Department Shepherd School of
Engineering and Technology Sam Higginbottom Institute of Agriculture,
Technology and Sciences (Formerly Allahabad Agriculture Institute)
Allahabad 211007, India.

192
 Analysis of Residual Stresses and Distortions in Girth-Welded Carbon Steel Pipe

A number of thermal analyses based on the number of

elements as indicated in Fig. 3 are conducted to see the
corresponding effects of maximum temperature attained
during the analysis. This maximum temperature attained is
then considered as the basis for mesh refinement. As shown
in Fig. 3, it is clear that no further increase in maximum
temperature is achieved if the number of elements is
increased beyond 54720. The mesh with number of elements
indicated by the arrow in Fig. 3 is then considered for the
analysis in order to get the mesh independent results. Tack
Legends : BM = Base Metal WM = Weld MetalSH = Strain Hardening
lengths positions in the FE models are in accordance with
Fig. 1. Thermo-mechanical properties of low carbon steel as a function of the physical weld sample. Theoretically, two cylinders
temperature used in the present study. should be regarded as separate parts in the model setup,
because they are independent until the moving heat source
Full 3D model is employed. Transient, non-linear thermal (welding torch) passes over and joins these together. Both
solution based on heat conduction, convective and radiative the cylinders are modeled as a single model in FE modeling
boundary conditions is solved in the first part to obtain the as the cylinders are tack welded to each other and
nodal temperature history. The temperature fields are further mechanically restrained prior to welding.
utilized as thermal loads in subsequent elastoplastic
structural analysis to obtain the transient and residual stress
fields and distortions.

II. FE ANALYSIS
2.1 FE model

For the girth weldingof two pipe with "V" groove [23], a
full 3D FE model along with finite element statistics
developed in ANSYS® is shown in Fig. 2. The element type
in thermal analysis is SOLID70 (linear 8-node brick element
with one degree of freedom, i.e., temperature at each node)
and in structural analysis is SOLID45 (linear 8-node brick Fig. 3. Mesh sensitivity analysis based on maximum temperature
element with three degrees of freedom at each node: attained.
translations in the nodal x, y, and z directions.). Further
details about the selected elements may be found in [17]. 2.2 Thermal analysis technique
High temperature and flux gradients are anticipated in and
A high non-uniform temperature field is generatedduring the
around the fusion zone (FZ) and heat affected zone (HAZ);
welding process resulting in residual stresses in the welds.
therefore, a relatively fine mesh is used within a distance of
The transient temperature distribution is a function of total
10 mm on both sides of the weld line (WL). Away from the
heat applied and heat distribution patterns within the domain
HAZ the element size increases with an increase in the
and is highly sensitive to weldinduced residual stresses. A
distance from WL.In the weld direction, the element size is
detailed and accurate thermal analysis with appropriate
kept constant equal to 1.96 mm. Within the anticipated HAZ
boundary conditions such as heat transfer by conduction,
dimension of 10 mm on each side of the WLin transverse
heat losses due to convection and radiation and heat input
direction, the element size of 1 mm is used.
from the welding torch along with the effects of filler metal
The element size away from the weld region increases with
deposition, is of paramount importance for the
the increase in distance. In the thickness direction there are
determination of realistic temperature profiles. The
total three elements, 1 mm each to facilitate for “V” groove
governing equation for transient heat transfer analysis
modeling.Two tack welds on the start, i.e., 0º and middle,
during welding process is given by Eq. (1).
i.e., 180º of the weld are modeled, each of which is
comprised of 4 elements (7.85 mm) in circumferential 𝝏𝑻
𝝆𝒄 𝒙, 𝒚, 𝒛, 𝒕 = 𝛁. 𝒒 𝒙, 𝒚, 𝒛, 𝒕 + 𝑸(𝒙, 𝒚, 𝒛, 𝒕)…...(1)
direction and 4 and 2 elements (4mm and 2 mm, 𝝏𝒕
respectively) in two layers in thethickness direction. The
used mesh is based on a mesh sensitivity analysis performed where ρ is the density (kg m −3 ) of the material, c is the
for successive mesh refinements. specific heat capacity (J kg −1 K −1 ), T is the current
temperature, q is the vector of heat flux, Q is the rate of
internal heat generation (W m −3 ), tis the time (s), ∇is the
spatial gradient operator and x, y, zare the coordinates in the
reference plane. The non-liner isotropic Fourier heat flux
constitutive equation given by Eq. (2) is employed.

q =−k∇T ..…(2)

Where k (W m −1 K −1 ) is the temperature dependent

thermal conductivityThe most widely acceptable double
ellipsoidal heat source model, presented by Goldak et al.
Fig. 2. (a) 3D FE mesh based on sensitivity analysis. (b) “V” [18] as shown in Fig. 4, is used to present the heat generated
groove, tack weld and root opening in FE model.

193
 International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-2, Issue-2, May 2013

by the welding torch for the heat input distribution tothe Table 1. Chemical composition of low carbon steel. % Chemical
welds. The very small variations due to curvature of the Composition
cylinders are not taken into account. The model gives the
%ChemicalComposition
Gaussian distribution for circumferential welding and has
C 0.130 V 0.056
excellent features of power and density distribution control
in the weld pool and HAZ. The spatial heat distribution in a Cr 0.026 Cu 0.015
moving frame of reference can be calculated with the Mn 1.418 S 0.007
governing equations Eq. (3) and Eq. (4) [18]. S 0.012 P 0.023
Si 0.346 Sn 0.002
𝒙 𝟐 𝒚𝟐 𝒛 𝟐
𝟔 𝟑𝜼𝑸𝒇𝒇 −𝟑 + +
𝒂𝒇 𝟐 𝒃𝟐 𝒄𝟐
𝒒𝒇 = 𝒆 .……(3)
𝝅 𝝅𝒂𝒇 𝒃𝒄 The spacers were removed after the tack welds were cooled
𝒙 𝟐 𝒚𝟐 𝒛 𝟐 to room temperature. To minimize the initial stress effects
𝟔 𝟑𝜼𝑸𝒇𝒓 −𝟑 + +
𝒒𝒓 = 𝒆 𝒂𝒓 𝟐 𝒃𝟐 𝒄𝟐 ………(4)
𝝅 𝝅𝒂𝒓 𝒃𝒄 prior to welding due to tack welds, the areas in and around
the tack welds were treated with post weld heating to 573K
Where, Q =VI and ff+ fr= 2 (300°C). Conventional gas torch heating with both infrared
and touch probe thermocouples measurement was utilized.
af is the length of the front ellipsoidal (m) ar is the length However, the stress data after the tack welds and after the
of rear ellipsoidal (m)b is the width of heat source (m) cis post weld heating was not recorded and the effects are
the depth of heat source (m) ff is the fraction of heat in front ignored. Further, as these cylinders were linearly seam
ellipsoidal fr is the fraction of heat in rear ellipsoidal Q is welded after roll forming of sheet metals, the effects of the
the total heat input (watts) V is the welding voltage (volts) I linear seam weld was also not considered in this work. The
is the welding current (amperes) furnace stress relieving prior to circumferential tack welds is
a reasonable justification for not considering these effects.
III. EXPERIMENTAL VERIFICATION The nodal temperature distributions from the thermal
The appropriate way to ensure the reliability of thenumerical analysis are used as a basic input for the sequel structural
simulations and to extend the utilization of the research analysis. Therefore, experimental data correlation for the FZ
work for shop floor applications is by conducting full-scale and HAZ dimensions or alternatively some nodal
experiments with proper instrumentation for data temperature verification for accurate pre-dictions of
measurement. For arc welding experiments, automatic subsequent stress fields and distortion patterns is a
welding setup with minimum human intervention and skill prerequisite. The latter technique by using the
is considered as mandatory for the proper validation of thermocouples and data acquisition system with computer
numerical results due to the variations associated with the interface is adopted for the thermal model validation in the
skill of the operators and rotary synchronization problems. present study. Thermocouples at four different locations
Similarly, the careful data acquisition during the were placed and temperature recorded through the data
experiments is of critical importance and demands a proper logger after every 10 seconds of time interval for
data measurement and analysis system. In the present comparison with FE results. A quantitative comparison of
research, to ensure the reliability of the FE models, GTAW measured and predicted transient temperatures at
experiments on two thin-walled pipe with similar geometric thermocouple locations is presented in Fig. 5. The
and welding process parameters from the finite element thermocouples TC1, TC2 and TC3 show close agreement
models are conducted. Low carbon steel equivalent to AH36 with the FE data. Whereas, thermocouple TC4 shows
with chemical composition as shown in Table 1 having comparatively higher variation of predicted and measured
slight variations in chemical composition from the material temperatures. Overall, the results are within the maximum
model used in the simulation is utilized. Similar variation of 8% and are reasonably accept-able. Residual
approximations were made in the past by [22] with stresses are measured at some specified points for
comparable measured and predicted results. In addition to comparison through the predicted results for structural
the FE parameters, argon with 99.999% purity was used as model validation.
shielding gas with flow rate of 15 liters/min. Commercially
available high-tech, fully automatic SAF GTAW welding
equipment along with rotary positioners and welding
fixtures was used to reflect the desired structural boundary
conditions. Single pass welding equipment along with rotary
positioners and welding fixtures was used to reflect the
desired structural boundary conditions. Single pass butt-
weld geometry is used with single "V" groove having
included angle of 90o and 2 mm root opening as shown in
Fig. 4. The welding specimen consists of two 150 mm outer
diameter and 3 mm wall thickness cylinders. Two tack Fig. 4. Butt-weld joint geometry (not to scale).
welds starting from 0° (weld start position) equally spaced
at 180°, each with length of ~ 8 mm were placed. These tack A center hole drilling strain gauge method is used to
welds were also used to create a root opening prior to measure the residual hoop and axial stresses at specified
welding by insertion of additional spacers of 2 mm at some locations, i.e., points P1 to P3 on cylinder outer surface and
appropriate locations during tack welding. P4 to P6 on cylinder inner surface. Refer to the gauge
location from P1 to P6 in Fig. 6. The details of the hole
drilling residual strain measurement method can be found in

194
 Analysis of Residual Stresses and Distortions in Girth-Welded Carbon Steel Pipe

[20]. A quantitative comparison of residual stresses from

experiments with predicted data, showing good agreement,
can be found in Fig. 6. From the qualitative comparison of
nodal temperatures and residual stresses from Fig. 5 and Fig.
6, it is evident that predicted results agreed well with the
experimental data; thus the developed models have been
experimentally validated.

Fig. 7. Temperature profiles at four different time steps during the

welding process.

Thermocouplelocation TC1 TC2 TC3 TC4

DistancefromW L(mm) 10 15 10 20
Distancefromweldstart(0) 30 30 90 90
Fig. 5. Comparison of computed and measured transient temperature
profiles at four different locations on pipe outer surface.

Pointreference(1to6)
Straingaugelocation P1 P2 P3 P4 P5 P6
DistancefromWL(mm) 10 15 20 10 10 10
Anglefromweldstart(0) 30 30 30 45 125 225
Fig. 6. Computed and measured residual stress values for different
locations at pipeouter surface.

IV. ASPECTED RESULTS

4.1 Thermal analysis
Fig. 7(a-d) shows the temperature distributions for four
different times during the welding process. As anticipated,
the peak temperatures are observed at the heat source
location (i.e., close to WL). Steep temperature gradients are
observed ahead of the heat source, showing the least
significance of heat flow ahead of the welding torch. The
gradients behind the torch show the cooling phenomenon
after peak temperature achieved, as the torch moves ahead
from certain point. Fig. 7(d) shows the temperature
distribution when the weldments are cooled to almost
uniform temperature, followed by some more time steps Fig. 8. Axial temperature distributions for four different cross-
further simulating the cooling phase. sections at different time steps from the weld start.

195
 International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-2, Issue-2, May 2013

Fig. 8(a-d) shows the axial temperature distributions for

different cross-sections at different time steps from the weld
start towards the time progress. It is clear that the
temperature distribution at a section is steep as the arc
crosses the section, e.g., in case of Fig. 8(a) the section is
located at an angle of 45° from the weld start position, i.e.,
0°. A welding torch traveling at a speed of 3 mm/s around a
circumference of 300 x πreaches the section after 39.27 s.
Hence maximum temperature as expected is observed at the
torch position. The temperature falls slowly as the torch
crosses the section. A preheating action of the section due to
the forward heat flow through the torch just before the torch
arrival at a section is also shown and is more dominant Fig. 11. Residual axial stresses (MPa) on inner surfaceat different
incase for the sections oriented at 225° and 315°as depicted cross sections from the weld start position.
in Fig. 8(c) and Fig. 8(d), respectively. Fig. 9 further
The compressive residual axial stresses near the WL
elaborates on the temperature distribution by using Goldak’s
diminish to zero after 17 mm on both sides of the WL.
heat source model.
Beyond this, a stress reversal from compressive to tensile is
observed. These low magnitude tensile stresses again
approach a zero value almost 68 mm away from the WL.
After 68 mm from the WL a constant axial stress value near
to zero is obvious from Fig. 10. The high tensile stresses
near the WL approach to zero and then reverse to lower
compressive residual stresses at 17 mm, again increasing to
almost constant value of zero at 68 mm on both sides of WL
observed for cylinder inner surfaces at different cross
sections from weld start position as shown in Fig. 11.The
general residual axial stress profile shows close analogy
with the previous research [7, 10, 12, 14, 15]. The variation
in magnitude (higher or lower residual stresses) in the
present research work is attributed to the different material
Fig. 9. Temperature contour plots. properties, i.e., mechanical properties like yield strength for
base and weld filler metals along with other parameters such
4.2 Residual stress analysis 4.2.1 Axial residual stress as weld geometry and heat source parameters. Some
fields observations of the significance pertaining to axial stress
fields from Fig. 10 and Fig. 11 are as follows:
For circumferentially welded cylinders, stress normal to the • Due to symmetry across the WL, the stresses are also
direction of the weld bead is the axial stress. Compressive symmetric. In and around the FZ, the high magnitude stress
and tensile axial stress fields are observed in and near the fields, i.e., compressive and tensile for outer and inner
weld region on the outer and inner surfaces of the cylinders, surfaces, respectively, are observed.
respectively. This is attributed to different temperature • Axial stresses are weakly dependent on the circumferential
profiles on the inner and outer surfaces of the cylinders. location. Fig. 10 and Fig. 11 clearly reveal that axial residual
Varying shrinkage patterns through the wall thickness on the stresses on outer and inner surfaces at four different cross
inner and outer surfaces due to different temperature sections (50°, 90°, 150° and 250°) are almost of the same
gradients; results in tensile and compressive residual stress magnitude and trend. This is in absolute agreement with the
fields on inner and outer surfaces, respectively, near the previous research [7]. Except the weld start and its vicinity,
weld line (WL). Fig. 10 shows axial stress distributions on the axial stress around the circumferential direction almost
the cylinder’s outer surface at different cross sections from has a homogeneous distribution.
the weld start position. The convention used is that Axial-50 • At the outer surface of cylinders near the WL, there is a
represents the axial stress at 50° from the weld start position prominent bulge indicating stress variations beneath the
and so on. weld crown (refer Fig. 10).
4.2.2 Hoop residual stress fields.
Stresses parallel to the direction of the weld bead are known
as hoop stresses. The residual hoop stresses are developed
due to the radial expansion andcontraction during the
heating and cooling sequence of welding process. Residual
hoop stress distributions for outer and inner surfaces of
cylinders along the axial directions in different cross
sections from the weld start position are shown in Fig. 12
and Fig. 13, respectively. On the inner and outer surfaces, in
and around the vicinity of the WL, large tensile and
compressive hoop stresses, respectively, are developed.
Similar trends for stress reduction and stress reversal are
Fig.10. Residual axial stresses (MPa) on outer surface at different observed for hoop residual stresses as for axial residual
cross sections from the weld start position

196
 Analysis of Residual Stresses and Distortions in Girth-Welded Carbon Steel Pipe

stresses. The trend for the hoop stress also agreed well with • On the outer surface, the low magnitude hoop stress (-76
the previous research [7, 10, 12, 14, 15]. The quantitative MPa to 77 MPa) remains almost constant. Some exceptions
variation is again attributed to different welding parameters, atweld start/end and tack weld locations (0° and 180°) are
heat source parameters and material properties in the present observed. At these locations, almost a zero hoop stress is
and referred research work, respectively. Some important observed. Slight variation in magnitude and trend is
aspects are as follows: observed for hoop stresses on the inner surface. The stress
• Hoop residual stresses are also symmetric due to varies from -70 MPa to 140 MPa, with some exceptions on
symmetry across the WL. weld start/end and tack weld locations. Again, at tack weld
• Near the FZ high tensile stresses (130 MPa on outer and locations, stress values close to zero are observed.
313 MPa on inner surface) are predicted. Away from the • On the outer surface, a compressive axial stress profile is
HAZ region (about 17 mm from WL), both on outer and shown in Fig. 14. The compressive stress varies from 114
inner surfaces a compressive residual stress of 200 MPa MPa to 361 MPa. A low magnitude, stable stress profile
is observed. from weld start to weld end with some exceptions in and in
• Hoop stresses are dependent on the circumferential vicinity of weld start/end and tack locations is observed. The
location from weld start to weld end. From Fig. 12 and significant effects of weld start are observed for axial stress
Fig. 13 it is obvious that hoop residual stresses on outer on inner surface. Also at tack weld points, the effect is
and inner surfaces at three different cross sections (50°, prominent. The stress varies from -200 MPa to about 340
90°, and 250°) vary in magnitude, with almost similar MPa in magnitude. A dip of about 472 MPa and 150 MPa is
trend. This is also in close agreement with the previous shown at the weld start of 0° and tack weld locations of 180°
research [7]. respectively.
4.2.4 Stress contour plots
From previous discussion it is evident that due to weld
start/end and tack welds, the residual stress varies
significantly along the entire periphery. Fig. 10 and Fig. 11
for cylinder outer and inner surfaces, respectively, presents
axial residual stresses at four different cross sections from
weld start position at 0o. No presentable variation is
observed because the data shown is away from the weld
start/end and tack weld orientations. Similarly, slight
variation in hoop residual stress patterns is observed from
Fig.12. Residual hoop stresses (MPa) on outer surface at different cross Fig. 12 and Fig. 13 for outer and inner surfaces,
sections from the weld start position respectively. Again, the data at weld start/end and tack weld
location(s) ismissing in this case. In order to get a better
insight of the stress variation along the hoop co-ordinates,
hoop residual stress fields on outer and inner surfaces are
shown in Fig. 15 and Fig. 16, respectively.From Fig. 15, on
outer surfaces the stress pattern on the whole periphery is
strongly affected by the weld start/end and tack weld at
180°. Highly fluctuating stress patterns along the entire
periphery, transverse to welddirection (axial direction) are
obtained. Pronounced localized stress reduction in and
around the weld start/endand tack weld locations is
Fig.13. Residual hoop stresses (MPa) on inner surfaceat different cross shown.However, these effects are slightly less significant at
sections from the weld start position the weld end location. Transverse to weld direction andaway
from the weld line, stress reversal is shown along the entire
4.2.3 Axial and hoop residual stress fields along the
periphery with some exceptions at weldstart/end and tack
circumference
Comparison of axial and hoop residual stress distribution for weld positions.Hoop residual stress fields from Fig. 16 on
outer and inner surfaces, on a circumferential path at the inner surface reveals that the effect
WLis shown in Fig. 14. Again, the stress profiles are
generally in agreement with the previous research. Some
important observations are:

Fig. 14. Axial and hoop residual stress fields on pipe outer and inner Fig. 15. Hoop residual stress fields on pipe outer surface .
surfaces on a circumferential path at the WL

197
 International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-2, Issue-2, May 2013

For the welding process parameters and tack weld geometry

in the present research, the axial shrinkage is maximum
(0.263 mm) at 78° and axial deflection is maximum (0.162
mm) at 330° near the weld end. From weld start position at
0º to 150º the FE results are under predicted as compared to
experimental data, from 17% to 44% with an average under
prediction of 24%. From 132º to 359º again low predicted
results are observed with variation of 45% to 57% at an
average under prediction of 45%. Although the variation is a
bit on the larger side, the results are in good qualitative
agreement from weld start to weld end. Fig. 18 shows the
variability in axial shrinkage at four different sections from
Fig. 16. Hoop residual stress fields on pipe inner surface. the WL, at cylinder outer surface. As the coordinate axis is
located on the WL, the shrinkage on the restraint-free end is
of weld start and weld end are not as much pronounced as shown to be positive. Near the WL(10 mm from weld line
for the cylinder outer surface. However, the effects of tack towards free restraint free end), maximum axial shrinkage of
weld location at 90° are quite significant. Over the entire 0.17 mm is indicated for axial sections at 50° and 150° from
periphery from weld start at 0° to weld end at 360°, hoop the weld start position. The axial shrinkage decreases away
residual stress oscillates significantly in the transverse from the WLtowards the free end to a minimum value of
direction on both sides of the weld line. Away from the 0.05 mm. Similarly, on the constrained end, a maximum
weld, the fluctuation trend is progressively stabilized. axial shrinkage of 0.2 mm is observed at a distance of ~10
mm from WLfor all the sections under investigation. The
4.3 Distortion analysis axial shrinkage decreases continuously away from the WL
4.3.1 Axial and radial deflection and a minimum shrinkage of almost zero is shown at the
restrained end.
Fig. 17 present the comparative (predicted and
experimentally measured) axial tilt of the restraint-free face V. CONCLUSIONS
of the cylinder after the cooling of the weld-ments. Keeping
in view the thickness of the cylinder face, it was difficult to Computational methodology and techniques based on finite
accurately measure the face tilt of the cylinder element analysis for the prediction of temperature profiles
experimentally. With some appropriate arrangements and and subsequent weld-induced residual stress fields and
welded cylinders still in the welding positioner, a digital dial distortion patterns in GTA welded thin-walled pipe of low
indicator was precisely positioned at an average diameter of carbon steel are developed and implemented successfully
the cylinder (in this case 297 mm) and the data was recorded with close correlation to the experimental investigations.
by rotating the positioner. To minimize the data acquisition Detailed results and discussion pertaining to residual stress
error, five different readings after the tack weld and after fields are presented. Further, the author's also present some
cooling of the weld-ments to room temperature were taken. data pertaining to residual deformations.
An average value is plotted in Fig. 17 for comparison with The following are the significant conclusions from the
predicted results. Axial face tilt ranging from -0.34mm to results presented.
+0.23mm is observed, as the dial indicator tracks the face (1) In the girth-welded pipes, axial residual stresses in
from 0° to 360°. The extent of axial shrinkage is dependent the weld vicinity are compressive in the outer surface
on a number of factors including welding process and tensile in the inner surface. Hoop stresses aretensile
parameters, tack weld dimensions and orientation. in both inner and outer surfaces.
(2).Hoop residual stresses are sensitive to the angular
location from the weld start position.
(3) Pipe wall thickness has a significant effect on the
residual stress distributions in the girth-welded pipes.
Thicker pipe tends to result in lower axial residual
stresses, but higher hoop residual stresses.
(4) Maximum axial and radial deflection is observed near
the weld line. The axial shrinkage decreases continuously
away from the WL and a minimum shrinkage of almost zero
shown at the restrained end. However, on the restraint free
Fig. 17. Measured and predicted axial deformation (face tilt) end some deflection (face tilt) is observed.
sections from the of the cylinder face (5) On the inner surface, the weld start effect is more severe
for both axial and hoop stresses and is dominant in the weld
start direction. A significant effect of tacks on the axial
stress on the inner surface is observed at angular positions of
0° and 180° from the weld start point, whereas the effect of
tacks on hoop stresses is not as prominent. The author has
concluded that the stress distribution is no more axis-
symmetric for a single pass butt circumferential weld with
initial tacks. However, if the weld start/end effects are
ignored, hoop stresses are almost uniform.
Fig.18. Axial shrinkage at four differentWL on pipe outer surface.

198
 Analysis of Residual Stresses and Distortions in Girth-Welded Carbon Steel Pipe

REFERENCES [24] Chin-Hyung Lee, Kyong-Ho Chang, Jeong-Ung Park, Three-

dimensional finite element analysis of residual stresses in dissimilar
[1]. Rybicki, E. F., Schmueser, D. W., Stone-sifer, R. W., Groom, J. steel pipe welds, Nuclear Engineering and Design256, (2013),160–
J., Mishler, H. W. 1978. A finite-element model for residual 168
stresses and deflections in girth-butt welded pipes. Journal of
Pressure Vessel Technology, Vol. 100, pp. 256-262.. [2].
Brust, F. W., Stonesifer, R. B. 1981. Effect of weldingarameters Prabhat Kumar Sinhais a M.Tech in CAD. He is
on residual stresses in BWR piping systems. EPRI NP-1743, having overall 6 year of teaching experience.His
Project 1174-1, Final Report major interests are in Auto Mobile Engg., Principles
[3]. Karlsson, C. T. 1989. Finite element analysis of temperatures of Machine Design,FEM, Foundry Technology,
and stresses in a single-pass butt-welded pipe-influence of mesh Industrial Engg. Mgt.Published various papers in
density and material modeling. Engineering Computations, Vol. international journals.
6, pp. 133-141..
[4]. Fujita, Y., Nomoto, 1., Hasegawa, H. 1980. Deformations and
residual stresses in butt welded pipes and shells. Nav. Archit. Raisul Islam is a M.Tech(Production & Industrial
a.OceanEngng. (Soc. of Nay. Archit. of Jap.) 18, pp. 164-174, Engineering) Pursuing from SHIATS,Alld-211007. He
and IIW-Doc. X-963-80 is having overall 2 year of teaching experience.His
[5]. Josefson, B. L. 1983. Stress redistribution [12] Y. Dong, J. Hong, major interests are in Manufacturing
C. Tsai and P. Dong. Finite Element modeling of residual stresses in Science,ProductionEngg., Fabrication.
austenitic stainless steel pipe girth welds, Welding Journal, Weld
Research Supplement,442 (1997) 449-444.
[6] A. Yaghia, T. H. Hydea, A. A. Becker, W. Suna and J. A. Williams,
Residual stress simulation in thin and thick-walled stainless steel
pipe welds including pipe diameter effects, International Journal of Chandan Prasad : M.tech(Production & Industrial
Pres-sure Vessels and Piping,83 (11-12) (2006) 864-874. Engineering) PursuingFrom SHIATS,Alld-211007
[7] D. Deng and H. Murakawa, Numerical simulation of temperature His major interests are in design,
field and residual stress in multi-pass welds in stainless steel pipe manufacturingScience, Industrial Relation. Published
and comparison with experimental measurements, Computational a papers in (IJMET).
Material Science,37 (3) (2006) 269-277
[8] B. Brickstad and B. L. Josefson, A parametric study of residual
stresses in multi-pass butt-welded stainless steel pipes, International
Journal of Pres-sure Vessels and Piping,75 (1) (1998) 11-25.
[9] E. F. Rybicki, D. W. Schmueser, R. W. Stonesifer, J. J. Groom and
H. W. Mishaler, A Finite Element model for residual stresses and Mohd. kaleem :M.tech(Production & Industrial
deflections in girth-butt welded pipes, ASME Journal of Pressure Engineering) PursuingFrom Sam Higginbottom
Ves-sel Technology,100 (1978) 256-262. Institute of Agriculture Technology and
[10] E. F. Rybicki, P. A. McGuire, E. Merrick and J. Wert, The effect of sciences,Allahabad 211007
Pipe thickness on residual stresses due to girth welds. ASME Journal
of Pres-sure Vessel Technology, 104 (1982) 204-209
[11] E. F. Rybicki and R. B. Stonesifer, Computation ofresidual stresses
due to multi-pass welds in piping system. ASME Journal of Pressure
Vessel Technology, 101 (1979) 49-54.
[12] Y. Dong, J. Hong, C. Tsai and P. Dong. Finite Element modeling of
residual stresses in austenitic stainless steel pipe girth welds,
Welding Journal, Weld Research Supplement,442 (1997) 449-444.
[13] R. I. Karlsson and B. L. Josefson, Three- dimensional Finite
Element analysis of temperature and stresses in single-pass butt-
welded pipe. ASME Journal of Pressure Vessel Technology,112
(1990) 76-84.
[14] M. Jonsson and B. L. Josefson, Experimentally determined transient
and residual stresses in the butt-welded pipes, Journal of Strain
Analysis,23 (1) (1988) 25-31.
[15] L. Karlsson, M. Jonsson, L. E. Lindgren, M. Näss-tröm and L.
Troive, Residual stresses and deforma-tions in a welded thin-walled
pipe, Proc. ASME Pressure Vessel and Piping Conf. (Hawaii, July
1989)PVP-173 (1989) 7-11.
[16] S. Fricke, E. Keim and J. Schmidt, Numerical weld modeling-a
method for calculating weld-induced residual stresses, Nuclear
Engineering and Design,206 (2-3) (2001) 139-150.
[17] ANSYS-10.0 user manual.
[18] J. Goldak, A. Chakravarti and M. Bibby, A new Finite Element
model for welding heat source. Metallurgical Transactions B. 15B
(1984) 299-305
[19] L. F. Anderson, Residual stresses and deformations in steel
structures, PhD. Thesis, Technical University of Denmark,(2000)
[20] Vishay Group, Measurement of residual stresses by the hole drilling
strain gage method, Technical Note No.TN-503.
(www.vishay.com/brands/measurements_group/guide/tn/tn503/503i
ndex.htm).
[21] P.J. Bouchard, D. George, J.R. Santisteban, G. Bruno, M. Dutta, L.
Edwards, E. Kingston,D.J. Smith, Measurement of the residual
stresses in a stainless steel pipe girth weld containing long and short
repairs, International Journal of Pressure Vessels and Piping, 82,(4),
( 2005),299–310
[22] W.C. Jiang,B.Y. Wang,J.M. Gong,S.T. Tu, Finite element analysis
of the effect of welding heat input and layer number on residual
stress in repair welds for a stainless steel clad plate, Materials &
Design, 32(5), (2011), 2851–2857
[23] Chin-Hyung Lee, Jeong-HoonBaek, Kyong-Ho Chang, Bending
capacity of girth-welded circular steel tubes, Journal of
Constructional Steel Research,75 (2012), 142–151

199