UC Davis
UC Davis Previously Published Works
Title
Assessment of tensile residual stress mitigation in Alloy 22 welds due to laser peening
Permalink
https://escholarship.org/uc/item/312761b6
Journal
Journal of Engineering Materials and Technology-Transactions of the ASME, 126(4)
ISSN
0094-4289
Authors
DeWald, A T
Rankin, J E
Hill, Michael R
et al.
Publication Date
2004-10-01
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
PROOF COPY 007404JYT
PROOF COPY 007404JYT
Adrian T. DeWald
Department of Mechanical and Aeronautical
Engineering,
University of California,
One Shields Avenue,
Davis, CA 95616
Laser Science and Technology,
Lawrence Livermore National Laboratory,
PO Box 808, Livermore, CA 94550
Jon E. Rankin
Laser Science and Technology,
Lawrence Livermore National Laboratory,
PO Box 808, Livermore, CA 94550
Michael R. Hill
*
e-mail address: mrhill@ucdavis.edu
Department of Mechanical and Aeronautical
Engineering,
University of California,
One Shields Avenue,
Davis, CA 95616
Matthew J. Lee
Hao-Lin Chen
Laser Science and Technology,
Lawrence Livermore National Laboratory,
PO Box 808, Livermore, CA 94550
Assessment of Tensile Residual
Stress Mitigation in Alloy 22
Welds Due to Laser Peening
This paper examines the effects of laser peening on Alloy 22 (UNS N06022), which is the
proposed material for use as the outer layer on the spent-fuel nuclear waste canisters to
be stored at Yucca Mountain. Stress corrosion cracking (SCC) is a primary concern in the
design of these canisters because tensile residual stresses will be left behind by the closure
weld. Alloy 22 is a nickel-based stainless steel that is particularly resistant to corrosion,
however, there is a chance that stress corrosion cracking could develop given the right
environmental conditions. Laser peening is an emerging surface treatment technology that
has been identified as an effective tool for mitigating tensile redisual stresses in the
storage canisters. The results of laser-peening experiments on Alloy 22 base material and
a sample 33 mm thick double-V groove butt-weld made with gas tungsten arc welding
(GTAW) are presented. Residual stress profiles were measured in Alloy 22 base material
using the slitting method (also known as the crack-compliance method), and a full 2D
map of longitudinal residual stress was measured in the sample welds using the contour
method. Laser peening was found to produce compressive residual stress to a depth of 3.8
mm in 20 mm thick base material coupons. The depth of compressive residual stress was
found to have a significant dependence on the number of peening layers and a slight
dependence on the level of irradiance. Additionally, laser peening produced compressive
residual stresses to a depth of 4.3 mm in the 33 mm thick weld at the center of the weld
bead where high levels of tensile stress were initially present. 关DOI: 10.1115/1.1789957兴
1 Introduction
The U.S. Department of Energy 共DOE兲 has been charged with
developing a facility for the safe storage of spent nuclear material
in a centralized location 关1兴. This effort has been named the Yucca
Mountain Project 共YMP兲 after the proposed site in the Nevada
desert where the material will be stored. The goal of this project is
to design a repository that will isolate hazardous nuclear waste
from the environment for thousands of years 关2兴. The combination
of the extremely long design life and the high consequence of
failure presents unique engineering challenges and has required
particularly careful study of every possible failure mode.
A highly corrosion-resistant material, Alloy 22 共UNS N06022兲
关3兴, will be used for the outer layer of the waste package to mini-
mize the possibility of failure due to corrosive environmental con-
ditions. After being loaded with radioactive material, the cylindri-
cal canisters will be sealed with a final closure weld, which will
leave behind a tensile residual stress from welding. One of the
most studied failure mechanisms for the waste package system is
stress corrosion cracking 共SCC兲. For SCC to occur, three condi-
tions must exist simultaneously. These conditions are a susceptible
material, a corrosive environment, and a state of tensile stress.
Since Alloy 22 has been shown in recent studies to be vulnerable
to SCC under extreme conditions 共acidic solution, elevated tem-
perature, and weld residual stresses兲关4兴, and the second condition
cannot be controlled with absolute certainty, it is necessary to
eliminate the tensile residual stresses in the waste-package welds.
Laser peening is one of a few candidate technologies being exam-
ined as a method to mitigate tensile residual stresses and reduce
the possibility for SCC.
Laser peening is an emerging surface-treatment technology that
was developed in the 1970s at Battelle Columbus Laboratories
关5–11兴, but its entrance to the commercial marketplace has been
protracted due to limitations in laser technology. Like other simi-
lar surface treatments, laser peening is used to generate a com-
pressive stress on the surface of a part that has been shown to
inhibit failures caused by failure mechanisms, including fatigue
and SCC 关12兴. While other surface treatment techniques, such as
shot peening, are only capable of producing compressive stress
down to depths of a few tenths of a millimeter 关13兴, depths of
compressive residual stress for laser-peened components are typi-
cally on the order of 1 or 2 mm 关14兴. In this work, depths of up to
and beyond 4 mm are demonstrated, which is important in the
design of the canisters because compressive surface stresses must
remain as the outer layer of the canister becomes thinner over
time due to general corrosion 关3兴.
Surface preparation is necessary prior to laser peening. First, a
protective layer is applied to the surface; this is called the ablative
layer because its surface is ablated off during treatment. Typical
ablative layer materials include opaque tape or paint 关15兴. Next, a
transparent inertial tamping layer is applied over the ablative
layer, which acts to confine the expansion of the high-pressure
plasma to be generated by a laser pulse 关16兴共Fig. 1兲. Typical
materials for the confinement layer include water and glass 关17兴.
After these two surface layers are in place, the laser peening pro-
cess can be carried out.
Laser peening uses a pulsed, high-power laser to generate high-
*
Corresponding author.
Contributed by the Materials Division for publication in the J
OURNAL OF ENGI-
NEERING MATERIALS AND TECHNOLOGY. Manuscript received by the Materials
Division July 21, 2003; revision received February 26, 2004. Associate Editor: S.
Mall.
Journal of Engineering Materials and Technology OCTOBER 2004, Vol. 126 Õ 1
Copyright © 2004 by ASME
PROOF COPY 007404JYT
PROOF COPY 007404JYT
PROOF COPY 007404JYT
pressure plasma on a small region of the part surface. The laser is
fired at the material and the photons in the laser beam pass
through the transparent tamping layer and are absorbed by the
opaque ablative layer, which forms a high-pressure plasma. The
expansion of the plasma is confined by the inertial tamping layer,
causing a significant pressure to develop on the surface of the part.
The duration of the laser pulse is on the order of 20 ns, and the
short duration pressure pulse causes a shock wave to travel
through the material that leaves plastically deformed material in
its wake. The deformed material must remain in geometric com-
patibility with the bulk material, and an equilibrium state is ob-
tained with the laser-peened area in a state of compressive re-
sidual stress. Laser peening is applied in a spot by spot manner
with typical spot dimensions ranging from around 1 mm on a side
关18兴 up to 1 centimeter 关14兴, with spot shapes being round or
rectangular. Multiple layers of laser peening are commonly used
to help ensure that there is uniform coverage and to increase the
depth of the compressive residual stress 关19,20兴. A layer of laser
peening refers to a nominal 100% coverage of the treatment area
with a slight overlap between successive spots. In most cases the
ablative layer is replaced between peening layers.
There are significant absolute requirements on the specifica-
tions of the laser used for laser peening. The laser must be capable
of producing an irradiance 共power per unit area兲 on the order of
1–10GW/cm
2
共dependent on Hugonoit elastic limit of the treated
material兲 with a pulse duration on the order of 10–30 ns. In ad-
dition to these requirements, it is also highly desirable, from a
practicality standpoint, that the laser system has a high energy per
laser pulse and a high repetition rate 共high average power兲. These
parameters are important because they directly influence the total
processing time for a given part. In general, an ideal irradiance
and pulse duration are identified for a given material, which
means that a laser with a higher total energy per pulse will be able
to operate at the ideal irradiance and pulse duration with a larger
spot size, reducing the total number of spots necessary to cover
the prescribed peening area. Furthermore, a higher repetition rate
will allow for rapid application of successive laser spots leading to
an additional decrease in the total processing time. The require-
ments, along with the desirable parameters, vastly limit the num-
ber of capable laser-peening systems in existence.
The laser-peening treatment for this study was applied at
Lawrence Livermore National Laboratory 共LLNL兲. The laser sys-
tem employed at LLNL is capable of generating up to 20 J of
energy at a repetition rate of 6 Hz 关21兴, which means that an area
of approximately 0.17 m
2
can be covered per hour at an irradiance
of 10 GW/cm
2
and a pulse duration of 25 ns. The laser system at
LLNL also includes a stimulated Brillouin scattering 共SBS兲 phase
conjugation device, which provides uniform wavefront control
关22兴. The SBS ensures that a uniform energy distribution exists
within the laser spot, which allows the laser to run at a high
repetition rate without damaging the optics.
Previous studies have shown that the pulse duration, irradiance,
and number of peening layers are the parameters that have the
most significant impact on the residual stress introduced by laser
peening 关19兴. The first objective of this paper is to summarize a
brief parametric study of the effect of these parameters on the
residual stress state generated in small Alloy 22 specimens. The
complete parametric study is still in progress, but some of the
important preliminary results are presented in this paper. The sec-
ond objective of this study is to demonstrate the effect of laser
peening on the residual stress in a thick Alloy 22 butt-welded
plate. Residual stress measurements were made using mechanical
release methods 共slitting method 关23,24兴 and contour method
关25兴兲 after problems were encountered using diffraction methods.
2 Methods
Laser-Peening Parameters for Alloy 22. The amount of
published information regarding the effects of various laser-
peening parameters on the residual stress generated in a specific
material is very limited considering that the technology has ex-
isted for more than 30 years. The few published accounts that are
available focus on materials commonly subjected to cyclic loading
in fatigue-limited applications, such as titanium 关20兴 and alumi-
num 关14兴. When an application arises requiring the treatment of an
unconventional material like Alloy 22, the peening parameters are
selected using a combination of published literature regarding
similar materials, experience, and parametric studies in small cou-
pons of base material. As stated above, previous research suggests
that the pulse duration, irradiance, and the number of layers have
the most significant impact on the residual stress developed by
laser peening. Experiments have shown that the plastically af-
fected depth is controlled by the pulse duration 关26兴, with a long
pulse inducing a greater depth of compressive residual stress, and
by the number of peening layers, with more layers producing
deeper compressive residual stress 关19兴.
A parametric study investigating the effect of the number of
peening layers and the irradiance on the residual stress state gen-
erated in Alloy 22 was performed. Residual stress measurements
were made on small blocks peened with a varying number of
layers 共2, 4, 10, and 20兲 at constant irradiance and pulse duration,
and small blocks peened with a varying irradiance 共7, 10, and
13 GW/cm
2
) at a constant number of layers and pulse duration. In
each case a nominally square shaped laser spot was used with a
side length of approximately 3.0 mm. Each laser spot overlapped
its neighbors by 10% of the spot length in both directions. A
Fig. 1 Description of laser-peening process: „a… workpiece is covered with a protective
ablative layer and an inertial confinement layer, a pulsed, high-energy laser is fired at the
part, and „b… a region of high-pressure plasma is generated, which causes a shock wave to
travel through the material.
2 Õ Vol. 126, OCTOBER 2004 Transactions of the ASME
PROOF COPY 007404JYT
PROOF COPY 007404JYT
PROOF COPY 007404JYT
summary of the parametric study is presented in Table 1. An ad-
ditional residual stress measurement was performed on an un-
peened specimen to determine the amount of residual stress
present in the blocks prior to laser peening.
The in-plane dimensions of the Alloy 22 blocks used for the
parametric study were 38⫻ 38 mm and the thickness was 20 mm.
The blocks were removed from unused sections of a butt-welded
plate using a wire electric discharge machine 共EDM兲. The laser
peening was applied in a rectangular pattern that covered roughly
30⫻ 30 mm of the top surface with a pattern of 11⫻ 11 spots. The
blocks were peened using an ablative layer of 120
m thick alu-
minum tape 共replaced between layers兲 and a tamping layer of
approximately 1 mm thick flowing water.
The residual stress in the blocks was measured using the slitting
method 共also known as the crack-compliance method兲, which was
first published by Vaidyanathan and Finnie in 1971 关23兴 and re-
cently reviewed by Prime 关24兴. This method uses metallic foil
strain gages to measure the strain released during incremental cuts
into the depth of the material. The recorded strain versus depth
data are used to solve for the initial residual stress normal to the
plane of the cut through elastic inverse methods 关27兴.
The residual stress was measured as a function of the depth at
the middle of the top surface. On the laser-peened blocks, foil
strain gages with a grid length of 0.787 mm were applied to the
top surface roughly 2.54 mm from the location of the cut and on
the back surface directly below the center of the cut. A layer of
silicone was applied over the strain gages to protect them from
moisture during the experiment. The cutting was performed on a
wire EDM with 0.25 mm diameter brass wire in increments of
0.13 mm to a total depth of 1.02 mm, 0.25 mm to a total depth of
2.54 mm, 0.51 mm to a total depth of 8.64 mm, and 1.02 mm to a
final depth of 18.80 mm. Released strain was read with a com-
mercial Wheatstone bridge instrument and recorded by hand after
each slitting increment. Precise measurements of the position of
the strain gages relative to the position of the cut were made after
the completion of the experiment using photogrammetry. A simi-
lar procedure was followed for the unpeened block except that
smaller depth increments were used due to the expected shallow
depth of residual stress 共increments of 0.025 mm to a total depth
of 0.25 mm, 0.051 mm to a total depth of 0.51 mm, 0.076 mm to
a total depth of 0.90 mm, 0.102 mm to a total depth of 1.40 mm,
0.127 mm to a total depth of 2.30 mm, and 0.25 mm thereafter兲.
For a more thorough description of the theory and application of
the slitting method please consult Refs. 关24兴 and 关27兴.
In order to select the best set of laser-peening parameters for
Alloy 22, an objective system of judging the residual stress pro-
files is required. The primary figure of merit for this laser-peening
application is the depth where the tensile residual stress reaches
20% of the material yield strength (S
y
⫽ 372 MPa for Alloy 22
heat treated plate, 20% S
y
⫽ 74.4 MPa), which was identified by
YMP personnel as the level of tensile residual stress significant to
their life prediction calculations. The secondary figures of merit
for this application are the depth of compressive residual stress
共zero stress crossing兲 and the magnitude of near-surface residual
stress. For quantitative comparison, the near-surface residual
stress was determined at 0.15 mm below the surface, which was a
near-surface value where results were available for all measure-
ments. The slitting results will be summarized based on these
figures of merit.
Residual Stress Determination in Peened and Unpeened Al-
loy 22 Welds. Residual stress was also determined in as-welded
and laser-peened sections of a quality-controlled 33 mm thick
welded plate. This weld was an Alloy 22, multipass, double-V
groove, butt-welded 共GTAW兲 plate prepared by a commercial pro-
vider. The original welded plate had a length of 812 mm and a
width of 200 mm. The plate was cut into four nominally identical
200 mm long sections, two of which were used for this experi-
ment 共Fig. 2兲. Since the plate was made by continuous welding, it
is assumed that the weld residual stress in each specimen is simi-
lar. Therefore, measuring the residual stress in two specimens, one
with and one without laser peening, will demonstrate the effect of
laser peening on the welded plate.
One of the sample weld specimens was laser peened at LLNL.
Peening was performed with a pulse duration of 25 ns, an irradi-
ance of 10 GW/cm
2
, and 10 peening layers. These parameters
were selected before the completion of the parametric study due to
the accelerated schedule under which this effort was performed.
Laser peening was applied to a region at the center of the speci-
men, measuring 100 mm in the transverse direction and 76 mm in
the longitudinal direction 共Fig. 2兲.
The longitudinal component of residual stress in the peened and
unpeened weld specimens was measured using the recently devel-
oped contour method 关25兴. An illustrative description of the con-
tour method will be given in two dimensions 共for simplicity兲, but
the measurement principle applies equally in three dimensions.
The principle behind the contour method is that when a part con-
taining residual stress is cut in half along a straight line 关Fig. 3共a兲兴
the newly created free surface will deform as the stresses normal
to the surface are released by cutting 关Fig. 3共b兲兴. The deformations
of the cut surface can be used to uniquely determine the initial
residual stress acting normal to the cut plane using Bueckner’s
superposition principle 关Fig. 3共c兲兴 provided the stress release was
elastic. The contour method, therefore, consists of three steps: 1兲
cutting of the part at the location where residual stress is to be
determined, 2兲 measurement of the cut surface profile, and 3兲
calculation of the precut residual stress.
From an experimental standpoint, one must cut the part in half
with a very controlled method that does not significantly alter the
existing residual stress field 共wire EDM typical兲. The cut for this
measurement was performed on a submerged wire EDM with 0.25
mm diameter brass wire and finish cut settings to minimize the
roughness along the surface of the cut. The specimen was securely
clamped to a thick aluminum backing plate during the cutting
process to minimize the amount that the part was able to move as
residual stresses were released. The clamping helps to satisfy two
assumptions made during the contour analysis: that the stress re-
lease is elastic 共as clamping reduces stress concentration at the cut
tip兲, and that the plane of the cut does not grossly deviate from a
straight line.
Table 1 Specimen matrix for laser peening parameter study
Specimen
number
Irradiance
(GW/cm
2
)
Number of
layers
Pulse duration
共ns兲
07-10 7 10 25
10-10 10 10 25
13-10 13 10 25
10-02 10 2 25
10-04 10 4 25
10-20 10 20 25
Fig. 2 Geometry of 33 mm thick butt-weld specimen „laser-
peened area shaded…
Journal of Engineering Materials and Technology OCTOBER 2004, Vol. 126 Õ 3
PROOF COPY 007404JYT
PROOF COPY 007404JYT
PROOF COPY 007404JYT
The measurement of the cut surfaces was performed at LLNL
using a high-precision coordinate measuring machine 共CMM兲
running in drag mode 共i.e., the CMM probe remains in contact
with the part as it moves across the surface taking data at specified
increments, in contrast with pecking mode where the probe is
lifted off between each measurement兲. A 2.0 mm diameter silicon
nitride tip was used for the measurements, which limits the impact
of surface roughness on the data since the large diameter ball
contacts only the high points of the rough EDM surface. Both
halves of the cut were measured with a point spacing of approxi-
mately 0.5 mm in the transverse and depth directions (x and y
directions on Fig. 2兲 over the entire cut surface. A region consist-
ing of about 30 mm to either side of the weld center was measured
with a point spacing of 0.25 mm to better capture the surface
profile in the area of expected stress gradients. The point spacing
described above resulted in approximately 25,800 data points for
each of the two surfaces. The deformations from opposite sides of
the cut were averaged to remove anti-symmetric effects of shear
stresses present along the plane of the cut and nonlinearities in the
cutting path 关25兴.
A finite element model of the part was used to convert the
measured deformations back to residual stress. Since the surface
deformations are small 共typically on the order of tens of microns兲
it is equivalent to apply the inverse of the measured surface to an
initially flat surface or to flatten a surface that contains the mea-
sured contour. From a practicality standpoint, however, it is much
simpler to generate a model with an initially flat surface 关25兴.
Surface profile data were numerically reduced, and reduced
data were fit to a smooth surface to eliminate surface roughness
effects. The data from each surface were first translated and ro-
tated until they were aligned with the same coordinate system. To
ensure that measurements were made near the edges of the sur-
face, each CMM line scan was started and completed off the edge
of the surface so that the edge would be explicitly apparent when
the data were examined. Since data were collected beyond the
coupon edges, the data set had to be filtered to remove points that
were not actually part of the surface. Since the cross section of the
surface was not perfectly rectangular, simply trimming points with
a rectangular mask was not a viable option. An effective method
for systematically removing the data points that were not part of
the surface was developed. This method operated only on data
within 8.0 mm of the surface perimeter, dividing it up into 20
small patches 共each roughly 20 mm⫻ 8 mm), and then individu-
ally fitting the patches with a first-order Fourier surface 共nine
terms兲. All of the data points that were not within 7
m of this
fitted surface were then removed, and the remaining data were
again fit to a first-order Fourier surface. This process of fitting the
small patches of data and removing points continued until no
points remained that were outside the specified 7
m tolerance
from the surface fit. Once the data from off the edges were re-
moved, the deformed surfaces from each half of the cut were
individually fit to a Fourier surface to smooth out the influence of
surface roughness. For both coupons, a sixth-order Fourier surface
共169 terms兲 was used because this order was sufficient to produce
a plateau in the root mean square 共RMS兲 error between the data
and the fit. The two Fourier surfaces were then interpolated at a
set of common locations 共finite element node locations兲, and the
two data sets were averaged to get the average surface deforma-
tion due to residual stress release.
A finite element model of half of the weld specimen was gen-
erated consisting of 55,560 eight-node, linear brick elements with
incompatible modes for enhanced bending performance. The
model was made to represent the geometry of the welded plate
after it has been cut in half 共Fig. 4兲. The number of elements on
the top surface and throughout the middle of the surface was
increased to allow for better spatial resolution of the resulting
residual stress field and to help ensure that the solution was con-
verged. The mesh on the bottom surface is even more refined to
get additional spatial resolution within the peened region. The
negative of the deformed contour from the average surface was
applied to the finite element model as a displacement boundary
condition at nodal locations on the cut surface of the finite ele-
ment model. An equilibrium step was taken, and the resulting
stress normal to the surface was calculated, which is the estimate
of residual stress prior to sectioning.
The results from the finite element model provide the residual
stress acting normal to the plane of the cut over the entire cut
surface. A contour plot of residual stress over the weld cross sec-
tion is an effective way to visualize the resulting residual stress.
Contour plots of residual stress were prepared for both the as-
welded and laser-peened specimens to show how laser peening
affects the residual stress. While the contour plot is helpful in
visualizing the results, it is difficult to draw any quantitative con-
clusions about the effects of laser peening from a contour plot. For
this reason, line plots of the residual stress versus depth from the
surface were prepared for both specimens at three different loca-
tions 关center of weld bead (x⫽ 102 mm), weld bead toe (x
⫽ 111 mm), outside of weld (x⫽ 132 mm)].
Fig. 3 Contour method principle: „
a
… a body containing un-
known residual stress is cut in half, „
b
… the free surface de-
forms as the stresses normal to the plane of the cut are re-
leased, and „
c
… applying the opposite of the deformations back
to the part recovers the initial residual stress state.
Fig. 4 Finite element model of half of the sample weld speci-
men
4 Õ Vol. 126, OCTOBER 2004 Transactions of the ASME
PROOF COPY 007404JYT
