Investigation of Weld Crack Mitigation Techniques with Advanced

Numerical Modeling and Experiment - Review

Yu-Ping Yang1, Suresh Babu2, Jeffrey Kikel3 and Frederick Brust4
1
Edison Welding Institute, Columbus, Ohio, USA (yyang@ewi.org)
2
The Ohio State University, Columbus, Ohio, USA (babu.13@osu.edu)
3
The Babcock & Wilcox Company, Barberton, Ohio, USA (jmkikel@babcock.com)
4
Engineering Mechanics Corporation of Columbus, Ohio, USA (bbrust@emc-sq.com)

Abstract and liquation cracking during welding of 2024 high

This paper reviews published methodologies of
mitigating weld cracking by altering thermal strain
and stress distributions during welding with
thermomechanical methods. Four thermomechanical
methods are introduced which include (a) trailing
cooling, (b) side rolling, (c) controlling heat input,
and (d) weld sequencing. These methods and
mechanisms were developed with advanced
numerical modeling and experiment. Each method
has its own advantage, potential applications, and
limitations which are also discussed.
Introduction
Hot cracking has been a subject of intensive studies
over the last few decades. In general, two basic
approaches are usually taken to control hot cracking:
(1) improving the material ductility of weld and heat-
affected zone and (2) improving the
thermomechanical conditions during welding. This
paper reviews published methodologies to design
these conditions using computational models. Studies
that consider alterations in the welding-induced
transient stress and strain through (a) trailing cooling,
(b) side rolling, (c) controlling heat input, and (d)
weld sequencing will be presented in aluminum,
nickel and iron-alloys.
strength aluminum alloy.
Controlling heat input [3] was used to mitigate
ductility-dip cracking and solidification cracking
during cladding and buttering a low alloy steel pipe
with Ni-Cr filler wire. Experimental mockup and
thermo-mechanical modeling were performed to
understand the mechanical reasons for the cracks.
Experimental and modeling results suggest that
reducing the heat input in regions susceptible to
cracking could be a viable mitigation technique
which has been proven during experiment.
Weld sequence [4] was studied with numerical
modeling to understand the cracking reasons in
bimetallic welds that join the hot leg to the reactor
pressure vessel nozzle in the V. C. Summer Nuclear
Power Plant. Numerical results showed that weld
sequence affected the stress distributions on the inner
surface of the hot leg. Tensile weld residual stresses,
in addition to service loads, contribute to the crack
and its growth rates.

Trailing cooling [1] and side rolling techniques [2]

were investigated to control solidification and
liquation cracking with thermal-elastic-plastic
modeling and experimental trials. Numerical
analysis helped understand the mechanism and found
the best process parameters of trailing cooling and
side rolling. Cooling devices and mechanical rolling
equipment were designed for experimental trials. Figure 0.1: Stress and Strain Distributions during
Experimental results show that both trailing cooling Welding
and side rolling can effectively prevent solidification
 Thermomechanical Conditions for Hot
Cracking

The necessary conditions for hot cracking are the

presence of tensile strains in the region that
undergoes the brittle temperature range (BTR) as
shown in Fig. 0.1. Figure 0.1 illustrates the stress and
strain distribution near a weld region. Before the
weld pool, a plastic compressive zone was formed
during thermal expansion and the constraint of
surrounding low-temperature area. During melting,
stress and strain in the weld pool is zero. The region
behind the weld pool subjects to tensile stress and
strain which could induce hot cracking. Further Figure 0.2: Thermomechancial Consditions
behind the weld pool, plastic tensile stress and strain Associated with Hot Cracking
exists, but no solidification cracking occurs since
material has recovered the full ductility after
solidification. Some materials could have dip
Trailing Heat Sink
ductility cracking (DDC). Dip ductility cracking can
also occur between weld pass due to reheating and re- Trailing cooling technique [1] was developed on
melting. aluminum alloy 2024. As shown in Fig. 1.1, a heat
sink was introduced behind the TIG welding torch to
In this study, mechanical strains were used as a generate an auxiliary compression zone within the
measure of the driving force for hot cracking instead region between the heating source and the heat sink.
of stresses, since at the BTR region, the transient Assuming a simple additive relationship between
stress level is usually low due to significantly welding-induced strain (w) and the strain (c)
reduced material yield strength at high temperature. generated by the heat sink source, the total
As depicted in Fig. 0.2, if the tensile strain rate (with mechanical strain at any moment in time becomes,
respect to temperature) exerting on the BTR region
becomes smaller than "the critical strain rate for w c
temperature drop (CST)" as depicted by the tangent
line (Curve B) to the ductility curve, or (3)
d
CST , (1) and Eqn (1) takes the form of (1)
dT
in theory, hot cracking can be avoided, as depicted by d w d c
line A. Decomposing d/dT, one obtains the CST
following: dT dT
(4)

d t As a result, hot cracking can be prevented if the
( 2) (2)
dT T reduction of the strain rate with respect to
t temperature is sufficient within the BTR region.
The first term /t in Eqn (2) becomes the standard
expression of strain rate determined by the
thermomechanical response during welding. The
second term T/t represents cooling rate typically
controlled by the heat flow characteristics of the
work piece under consideration. For a given
material, both the strain rate and the cooling rate can
be altered by either modifying welding parameters or
introducing local heating/cooling mechanisms, or
applying mechanical means.
 conventional TIG welding conditions was measured
at 35mm, while there was no indications of any
cracks for specimens welded with the trailing heat
sink (D = 18mm).

Figure 1.1: Illustration of Welding with a Trailing

Heat Sink

Since the actual reduction of the strain rate (with Figure 1.2: Temperature Distribution Welding with a
respect to temperature) is dictated by the complex Trailing Heat Sink
interactions between the instantaneous strain rate
(/t) and cooling rate (T/t) during welding as
indicated in Eqn (2). Advanced finite element
procedures were used to quantify and optimize the
effects of the trailing heat sink method on the
dynamic thermomechanical interactions during
welding.

Fig. 1.2 shows the predicted and measured

temperature distribution. With a heat sink, the
temperature cooling rate behind the weld pool has
been significantly increased. The cooling effect
depends on the cooling media and distance (D)
between the welding torch and heat sink. For a given
cooling media, the distance between welding torch
and heat sink is critical to control weld hot cracking. Figure 1.3: Effects of Cooling Distance (D) on the
Fig. 1.3 shows the transverse (perpendicular to Development of Transverse Strain
welding direction) plastic strain rate is reduced as the
distance between welding torch and heat sink is
reduced. When D is smaller than 23mm, hot
cracking can be eliminated.

Based on the transverse strain prediction, the distance

(D) between the welding torch and the trailing heat
sink was an important parameter in reducing the
propensity for hot cracking. With the optimal process
conditions being approximately established from the
detailed finite element simulations, welding trials
were performed for various values of D. It can be
seen that the smaller the cooling distance D, the
shorter the hot cracking length. As the cooling
distance became shorter than 18.5 mm, hot cracking
Figure 1.4: Photographs of Weld Samples
was completely eliminated. Fig. 1.4 shows the
photographs of two typical welding trial specimens Based on the results of this study, it can found that
without and with using the trailing heat sink method, trailing cooling could be an effective method to
respectively. The hot crack length under control cracking for materials not sensitive to cold
cracking such as aluminum alloys and austenite
stainless steel.

Side Rolling

Side rolling [2] is another example to control hot

cracking with mechanical methods. Sometimes,
trailing cooling technique cannot be used such as
weldinghighstrengthsteel. Trailingcoolingcould
enhancethepossibilityofcoldcracking.Therefore, Figure 2.2: Finite Element Model and Predicted
siderollingwasdevelopedandtheconceptisshown Temperature Distribution
in Fig. 2.1. During welding, two rollers were Figure 2.3 shows a comparison of transverse strain
amounted on the both sides of weld to move
without and with side rolling. Without rolling, tensile
synchronously with the welding torch. By rolling
bothsidesoftheweld,acompressivestrain ccanbe strain is in the weld and HAZ. With side rolling, the
induced to counterbalance the weldinginduced tensile strain because compressive strain or small
tensile strain W exerting on weld metal and HAZ tensilde strain. Therefore, hot cracking can be
experiencing BTR. Therefore, the resultant strain reduced or eliminated.
actingontheweldandHAZwithintheBTRcanbe
reducedtopreventhotcracking.

Figure 2.3: Effect of Rolling Technique on

Transverse Strain
Finite element model was also used to optimize the
side rolling technique parameters. With the optimized
Figure 2.1: Schematic Diagram of Rolling Position parameters, experiment was conducted without and
with side rolling. Figure 2.4 shows that 52-mm
Finite element analysis was used to invesitgate the cracking is in the conventional welding and there is
mechanism and optimize the technique parameters on no cracks found in the welding with side rolling.
aluminum 2024 ally. The critical parameters are
pressure, Dy, and Dx shown in Fig. 2.1, Figure 2.2
shows the finite element model and predicted
temperature distribution. The model is half model
and left side is a symmetric plane. Behind the weld
torch and on the side, two rollers will move with
welding torch and apply pressure on the plate.
Contact between the roller and plate was modeled.
Figure 2.4: Photographs of Welding Joints
Based on the results of this study, it can found that
side rolling cooling could be an effective method to
control cracking for materials which cannot apply
trailing cooling.
Reducing Welding Heat Input
Although dissimilar materials have been successfully
welded for many applications in the nuclear industry,
the cladding of low alloy steel pipes with Ni-Cr filler
wire has remained a challenge. This challenge is
related to ductility-dip cracking (DDC) and weld
solidification cracking [3]. Initial characterization of
a mockup weld showed that cracks were pronounced
in the outer-bead regions while negligible cracking
was encountered in inner-bead region (see Fig. 3.1).
Note that the outer-bead and the inner-bead regions
were fabricated with similar processing conditions
and the same filler wire. Therefore, the difference in
the cracking tendency is expected to be due to the
different heat conduction paths (caused by
geometrical changes) and consequently, strain levels
and gradients in the overlay.
Figure 3.1: Crack Rigions
To understand the cracking tendencies in the current
application, a mockup of the low-alloy pipe was
designed and three-dimensional computational weld
simulation was performed to predict thermo-
mechanical strain histories during cladding and
buttering of the low-alloy pipe. The model was
calibrated with the experimental measurements of
temperatures and dimensional changes of the pipe
[3].
Experiment
Figure 3.2 shows the mockup design that includes the
two weld build-ups. Seventeen layers (two beads per
layer) were used to fill up the cladding area and three
layers covered the buttering area. The average
thickness of each cladding layer was 2.17 mm and
average thickness of buttering layer is 2.54 mm. The
cladding and buttering filler material was Alloy 600
and the pipe was made from steel 3NiCrMo.
Figure 3.2: Mockup Design
Weaved gas tungsten arc welding (TIG) process was
used in building the mockup. The cladding
parameters were welding current of 300A, voltage of
13V, and traveling speed of 2.75 mm/sec. while the
buttering parameters were welding current of 300A,
voltage of 16V, and traveling speed of 2.75 mm/sec.
The cladding and buttering preheating temperature
was 145 C and the maximum interpass temperature
was 176 C. During the cladding and buttering
operations, the pipe was clamped at four locations as
it rotated while the TIG torch was weaved.
The completed mockup is shown in Fig. 3.3. The cap
cladding weld starts at 277.5 and stops at 262.5
while buttering starts at 0 and stops at 180. After the
mockup was completed, it was sectioned to
determine if there was cracking and if so, locations.
Figure 3.4 identifies four regions on a macrograph of
the clad and buttered regions (this macrograph
corresponds to Fig. 1). Cracks were observed mainly
in the Region 1 and Region 3.

Figure 3.3: Completed Mockup
Figure 3.4: Cladding and Buttering Regions
Thermo-Mechanical Modeling
The processing conditions encountered by the clad
and buttered regions were represented by a three-
dimensional (3D) thermo-mechanical finite element
model. Fixturing conditions during cladding and
buttering were translated into appropriate boundary
conditions. Thermal and thermo-mechanical analyses
were performed using ABAQUS and user
subroutines [1], which describe the heat flux and
control data transfer. The cladding and buttering
parameters (current, voltage, and travel speed) and
the temperature-dependent physical properties of
base (pipe) metal and filler metal were used as inputs
in the analyses. The model was calibrated by
comparing the temperature-time profiles at
monitoring locations in the mockup with those at
closest nodal locations.
Finite Element Model
A schematic of the finite element distribution on the
model of the mockup is shown in Fig. 3.5. At three
locations, where the tendency for cracking was high,
detailed thermo-mechanical histories were predicted.
The primary reason for selecting few locations was to
save computational time. The movement of the torch
was simulated using a heat-source model; only the
arc between 0 and 180 was considered here a finer
mesh was used. A coarse mesh was used between
180 and 360, where material was added at the
appropriate temperature to account for the thermal
mass and structural stiffness.
Figure 3.5: Finite Element Model Details
Thermo-Mechanical Strain Distributions
With the calibrated thermo-mechanical model, a full
analysis of the cladding and buttering operations was
performed. Figure 3.6 shows the final predicted, (a)
hoop strain distributions, (b) through-thickness strain
distributions, and (c) axial strain. The distributions
show that hoop strains are tensile, through-thickness
strains are mostly compressive, and axial strains are
mostly tensile. More tensile strains were observed in
the hoop direction than the axial direction.
(a) Hoop direction

(b) Thickness direction
(c) Axial direction
Figure 3.6: Final Strain Distributions
Crack Mitigation Studies
Two heat inputs, normal and reduced heat input, in
the outer bead were studied to find the effect of heat
input on thermo-mechanical strains. The normal heat
input for cladding is 3900W and for buttering is
4800W. The reduced heat input for cladding is
3315W and for buttering is 4080W. The travel speed
was kept the same. Figure 3.7 shows the area in
which the heat input was reduced (by about 15%).
Figure 3.7: Area where Heat Input Was Proposed to
Be Reduced
To find out the effect of reduced heat input on the
weld bead shape, an experiment was conducted with
the normal heat input and the reduced heat input. For
the case with the reduced heat input, only welding
current was reduced from 300A to 255A. Figure 3.8
shows the bead shape with the normal heat input and
the reduced heat input.
Figure 3.8: Bead Shapes with the Normal Heat Input
and the Reduced Heat Input
For the case with the normal heat input, the bead size
in traveling direction was 18.8 mm and the bead size
in transverse direction was 18.3 mm. So the average
bead size in traveling and transverse direction was
18.55 mm. For the case with the reduced heat input,
the bead size in traveling direction was 17.1mm and
the bead size in transverse direction was 16.6mm. So
the average bead size in traveling and transverse
direction was 16.85 mm.
Therefore, the average (half) weld pool size reduced
from 9.275 mm to 8.425 mm by the reduced heat
input. Since the in-plane bead size (top view) is
mainly controlled by the weaving, it only changed
9.4%. Because the weld penetration is mainly
controlled by the welding current, an assumption was
made that the weld penetration would be reduced by
15%. Therefore, the depth was reduced from 2.9 mm
to 2.5 mm (2.9mm 85%).
Thermal analysis was performed to investigate the
effect of bead-size reduction on the fusion zone
prediction. The reduced bead size in both traveling
direction and transverse direction is 8.4mm and the
depth is 2.5mm. Figure 3.9 shows the predicted
fusion zone for the outer bead with the normal heat
input and reduced heat input. With the reduced heat
input, the predicted weld fusion zone is smaller than
that with normal heat input.
Figure 3.9: Predicted Fusion Zone with the Reduced
Heat Input

Figure 3.11: Hoop Strain Comparison between

Two Heat Inputs

Based on the results of this study, the following

conclusion can be drawn:

Figure 3.10: Predicted Fusion Zone with the Hoop strain is the highest in tension by
Reduced Heat Input comparing with axial strain and strain
through thickness. The outer bead has higher
Thermo-mechanical analysis was performed to tensile thermo-mechanical strains than the
predict the thermo-mechanical strain with normal and inner bead in hoop direction. High hoop
reduced heat input. Figure 3.11a is a hoop strain strain was observed in the buttering region
comparison between the inner bead and the outer near the top surface.
bead of cladding layer 16 with the normal heat input. Reducing heat input is a valid method
Figure 3.11b is a hoop strain comparison between the
toward reducing the cracking tendency. To
inner bead and the outer bead of cladding layer 16
eliminate cracking, the heat input in the
with the reduced heat input. Since the heat input in
outer bead of the buttering region should be
the inner bead was not changed, the predicted hoop
further reduced and the heat input in the
strain of the inner bead for the reduced heat input is
inner bead should be also reduced.
close to the hoop strain for the normal heat input.
However, the hoop strain in the outer bead with the
reduced heat input is much smaller than that with the
normal heat input. These results indicate that the
thermo-mechanical strain can be reduced by reducing Changing Welding Sequence
the heat input so that the crack tendency could be
reduced by reducing heat input for the outer beads.
Changing welding sequence can change weld stress
distributions which could be used in reducing
cracking tendency. This example shows here is not
for hot crack mitigation, but it serves as an example
how welding sequence changes impact the weld
stress distribution.
This study [4] was performed to help support the
assessment of the cracking found in the A reactor
pressure vessel nozzle to hot-leg pipe bimetal weld
(Fig. 4.1) in the Virgil C. Summer nuclear plant. The
hot leg weld is a bimetallic weld joining a SA-508
(Class 2) reactor vessel nozzle with a Type 304
stainless steel pipe using an Inconel weld procedure
(Fig. 4.2). The hot leg pipe carries reactor-heated
water to the steam generator. It is then re-circulated
by the pump back through the cold leg. Both the hot
and cold leg stainless steel pipes are joined to the
reactor vessel nozzles via bimetallic welds. The
cracking of concern occurs in the Inconel weld only
on the inner surface. Tensile weld residual stresses, in
addition to service loads, contribute to the cracking.
distributions (1100F (593C)) and heating areas (10
in.). The heating lasts for 3 hours which was modeled
via creep analysis. Step 4 to preheat nozzle and
buttered region to 66C in preparation for weld
modeling. Step 5 is to deposit weld passes from
inside to a depth of 17.8 mm as shown in upper
picture of Fig. 4.5. At that time, defects were found
in the weld and the weld has to ground out. For
holding the two parts together, a bridge was left as
shown in the lower picture of Fig. 4.5.

Figure 4.3: Mesh for Cladding on the Carbon Steel

Figure 4.1: Components in a Nuclear Reactor

An axis-symmetric finite element model was used in

the analysis (Fig. 4.2). The distance from the center
of the weld to the pressure vessel is about 0.5m and Figure 4.4: Post-Weld Heat Treatment Modeling
the distance from the weld center to the steam after Cladding
generator is about 7 meters. The finite element model
included the entire 7.5m of pipe system. The
boundary conditions used in the finite element model
included fixed displacements at the pressure vessel
and steam generator. Lump-pass welding simulation
technique was used in the analysis since more than 50
weld passes were used in the weld.

Figure 4.5: Welding to 0.7inch and Ground out the

Figure 4.2: Bimetallic Welding Geometry and Finite
Element Mesh near the Weld
The numerical analysis involves several steps. Step 1
is to preheat nozzle to 66C in preparation for
cladding layer application. Step 2 is to model
application of cladding layer over nozzle end with the
model shown in Fig. 4.3. Step 3 is to conduct a local
post weld heat treatment (PWHT) on the cladding
area. Figure 4.4 shows the preheating temperature
Defects
There are two welding sequences which can be used
to finish the hot-leg welding from the bridge.
Sequence 1 is to weld inside and then weld outside.
Sequence 2 is to weld outside and then weld inside.
Finite element analyses were conducted to compare
the stress distribution on the inner surface. Figure 4.6
shows the hoop stress comparison between the two
welding sequences. The left-hand two figures show
the hoop stress distributions after finishing inside and
after completing the weld with sequence 1. The
right-hand two figures show the hoop stress
distributions after finishing outside and after
completing the welding with sequence 2. The results
show that sequence 1 has lower stress on the inner
surface than sequence 2. This reason of lower stress
in the sequence 1 is that the outside weld shrinkage in
hoop direction applies compression on the finished
inside weld. This results in high tensile stress outside
and low tensile stress or compressive stress inside.
The same principle should apply to sequence 2. But
the inside weld is small compared to the outside weld
and cannot produce enough shrinkage to reduce the
tensile stress outside.
Figure 4.6: Hoop Stress Comparison between Two
Sequences
The shrinkage theory can also be used to explain the
lower stress in sequence 1 than sequence 2 in the
axial direction as shown in Fig. 4.7. Figure 4.7
shows the final residual stress distribution for the two
sequences. For the sequence 1, the outside weld
shrinkage after welding applies compression on the
inside weld in the axial direction. Therefore the
stress on the inner surface has lower tensile stress or
compressive stress.
Figure 4.7: Axial Stress Comparison between Two
Sequences
Based on the results of this study, it can found that
changing weld sequence can alter the stress
distributions on the inner surface of a pipe
component. This could be used in controlling stress
distributions to prevent cracking during welding.
Summary
Stress and strain conditions associated with hot
cracking was discussed and four thermomechanical
methods were reviewed to demonstrate how to
control hot cracking from thermomechanical point of
view. Experiment and numerical modeling tool was
used in the development to understand the physics
and optimize process parameters.
Trailing cooling and side rolling can be used for thin
materials to prevent hot cracking. For materials not
sensitive to cold cracking such as aluminum alloys
and austenite stainless steel, trailing cooling could be
selected and side rolling can be used for most
materials in principle, but it requires a complicate
mechanical rolling equipment design. Reducing heat
input and changing weld sequence can be practical
methods to control hot cracking or other-type of
cracking such as stress corrosion cracking Primary
Water Stress Corrosion Cracking (PWSCC).
References
1. Y. P. Yang, P. Dong and J. Zhang, "A Hot-Cracking
Mitigation Technique for Welding High-Strength
Aluminum Alloy", Welding Journal, Vol. 79, No. 1,
2000, pp.9-s to 17-s.
2. Y. Yang, et al., Prevention of welding hot cracking
of high strength aluminum alloys by mechanical
rolling, Proceedings of the Conference on Trends
in Welding Research, American Society for Metals,
Pine Mountain, GA, 1998, pp. 700705.
3. Y.P Yang, S. Babu, S. Vaze, J. Kikel and D.
Dewees, Crack Mitigation during Buttering and
Cladding of A Low Alloy Steel Pipe, Proceedings
of the 8th International Conference on Trends in
Welding Research, Pine Mountain, Georgia; June
2-6, 2008.
4. F. W. Brust, P. M. Scott, and Y. P. Yang, Weld
Residual Stresses and Crack Growth in Bimetallic
Pipe Welds", Proceeding of 17th International
Conference on Structural Mechanics in Reactor
Technology (ed. S. Vejvoda),
http://www.iasmirt.org/SMiRT17/G08-1.pdf,
Praque, Czech Republic, August, 2003, Brno
University of Technology, Brno, 2003, Section G,
CR-ROM.