You are on page 1of 9

Weld Metal Ductility in Aluminum Tailor Welded Blanks

R.W. DAVIES, M.T. SMITH, H.E. OLIVER, M.A. KHALEEL, and S.G. PITMAN

The objective of the research described in this article was to characterize and numerically describe
the ductility of weld material in aluminum tailor welded blanks under uniaxial tension conditions.
Aluminum tailor welded blanks consist of multiple thickness and alloy sheet materials welded together
into a single, variable thickness blank. To evaluate the mechanical properties of the weld material in
these tailor welded blanks, a series of tensile specimens containing varying ratios of weld and
monolithic material in the gage area of the specimen were tested. These experimental results show
that increasing the amount of weld in the cross-sectional area of the specimen decreases the ductility
of the specimen and that the weld characteristics have a pronounced impact on ductility. Using the
experimental results and classical tensile instability and necking models, a numerical model was
developed to describe the ductility of the weld metal. The model involves basic material properties
and an initial imperfection level in both the weld and monolithic materials. The specimens studied
were produced from 1- to 2-mm AA5182-O aluminum alloy sheet material welded into blanks using
an autogenous gas tungsten arc welding process.

I. INTRODUCTION Prior research has focused on the problem of describing


the formability of aluminum tailor welded blank material in
THE automobile industry is continually working to stamping applications.[6–10] Most of this research has focused
develop and apply technology that reduces the cost and on performing simulative stamping tests and placing welds
weight of automobiles. The intent of these efforts is to mini-
within a standard size tensile specimen. These methods pro-
mize the energy consumption and environmental impact of
vide insight into the combined properties of the material,
vehicles. Tailor welded blanking is a technology that allows but they offer limited quantitative means of describing the
vehicle designers to reduce component weight and the num-
weld metal ductility individually. The research described in
ber of components in a structure without compromising the
this article produced a series of tensile specimens with the
final strength, stiffness, and durability of the component. weld oriented longitudinal to the tensile axis. The specimens
Tailor welded blanks consist of multiple sheet materials
were designed with varying amounts of cross-sectional gage
joined with welds to produce a single variable thickness
area, which resulted in varying amounts of specimen gage
and/or alloy stamping blank. Automobile manufacturers are area comprised of weld material. The experimental results
finding more and more applications for the use of steel
exhibit a trend that indicates increasing the amount of weld
tailor welded blanks in current vehicle designs. Currently,
in the specimen gage area will lead to a decrease in tensile
aluminum tailor welded blanks are being investigated for elongation. Using these experimental results and the classical
potential use in aluminum intensive vehicles to further
theories of tensile instability and necking presented by
reduce vehicle weight.
Hart[12] and Ghosh,[13] a model of weld material ductility is
A photograph of a prototype aluminum tailor welded developed. The numerical model is based upon the existence
blank is shown in Figure 1. The formability of aluminum
of an initial imperfection in the cross section of the
tailor welded blanks currently is an area of considerable
specimens.
interest to the automobile industry.[1–3] Although tailor
welded blank technology is effective at minimizing material II. EXPERIMENTAL PROCEDURE
use and thus vehicle weight, the stamping process is compli-
The tailor welded blank materials characterized in this
cated by the existence of welds in the blank.[4,5] Product
study were autogenously gas tungsten arc welded AA5182-
design and process reliability for components produced in
this way become critical issues when compared to designs O alloy in 1- to 2-mm combinations. The monolithic sheet
and processes based on monolithic sheet stamping. From a materials were manufactured through conventional direct
chilling and rolling techniques. All test materials were col-
process modeling standpoint, the weld produces a geometric
discontinuity that is more difficult to accurately model than lected from the same material lots, though the lot for each
monolithic sheet stamping. In addition, if a model does thickness was different. Saran et al.[11] and Pickering et
al.[6] described the butt-welding process used to produce the
accurately predict the stress and strain imposed on the weld
during stamping, the weld metal limits of formability are subject materials. Figure 2 illustrates the typical geometry
substantially unknown. and microstructure of the welds. The grain size of the weld
material is significantly larger than the grain size observed
in the monolithic sheets. In addition, the weld material exhib-
ited larger equiaxied grains in the center of the weld and
R.W. DAVIES, Senior Development Engineer, M.T. SMITH, Technical columnar grains leading to the monolithic sheet interface.
Group Manager, and M.A. KHALEEL and S.G. PITMAN, Staff Engineers, Microhardness investigation of the non-heat-treatable
are with Battelle Pacific Northwest National Laboratory, Richland, WA
99352. H.E. OLIVER, Team Leader of Project Engineering and Technology,
AA5182-O alloy showed neither a significant strength varia-
is with the Reynolds Metals Company, Chester, VA. tion nor a heat-affected zone in the weld region or sur-
Manuscript submitted August 17, 1999. rounding areas.

METALLURGICAL AND MATERIALS TRANSACTIONS A U.S. GOVERNMENT WORK VOLUME 31A, NOVEMBER 2000—2755
NOT PROTECTED BY U.S. COPYRIGHT
thickness and fully gripped the 1- and 2-mm sheets up to
the edge of the weld. Figure 4 illustrates the design used
for the fifth specimen, which was a miniature specimen that
consisted solely of weld material in the gage area. These
specimens were electron discharge machined from the center
of the weld material, parallel to the length of the weld. The
specimens were hand polished to the final test thickness. In
addition to the specimens prepared from the tailor welded
blanks, monolithic specimens were prepared for each of the
five specimen designs. Tensile tests for all the specimens
shown in Figure 3 were conducted using a constant crosshead
velocity of 12.7 mm/min. For the miniature specimen, a
crosshead velocity of 0.292 mm/min was used. All mechani-
cal testing of the weld and monolithic specimens were con-
ducted at room temperature.
Fig. 1—Photograph illustrating an aluminum alloy tailor welded blank after The uniaxial tensile results are reported in the form of
a typical stamping operation. The blank is butt welded from three separate true stress and true strain. For this study, the definition of
flat sheets and then formed as shown. The door inner stamping consists true strain is given by
of AA5182-O in varying thickness. (Photo courtesy of Reynolds Metals
Company and Ogihara America Corporation.) «t 5 ln (1 1 e) [1]
and the definition of true stress is given by
st 5 se(1 1 e) [2]
where e represents the engineering strain in the specimen,
and se represents the engineering stress in the specimen.
The axial strains developed during tensile testing of the
specimens illustrated in Figure 3 were measured using an
extensometer that recorded strain from test initiation through
specimen failure. The extensometer was placed on the speci-
men in such a way that it recorded strain directly on the
weld bead. The miniature specimens shown in Figure 4
were tested without an extensometer. Since there was no
extensometer used for these miniature tensile tests, the gage
length was measured via an optical comparator on each
specimen and the initial gage length was used to calculate
Fig. 2—Cross section through the weld looking normal to the axis of the the engineering strain. In these cases, the engineering strain
weld length. Portions of monolithic sheet material are shown on either side was calculated by dividing the displacement of the tensile
of the weld. Specimen was anodized to expose structure. machine crosshead by the measured initial length.

The mechanical properties of two separate weld popula- III. EXPERIMENTAL RESULTS
tions were investigated. The two populations were desig- Figures 5 and 6 are illustrations of representative true
nated first generation and second generation, and they were stress vs true strain tensile results for first-generation and
produced using an initial welding process for the first-gener- second-generation weld materials, respectively. The figures
ation population and a welding process believed to be more illustrate the effect the specimen design can have on the
refined for the second-generation population. The primary true stress vs true strain curves. The first-generation weld
difference between the two weld populations was that the specimens in Figure 5 show a significant drop in specimen
first-generation welds had approximately 20 times the num- ductility as the percentage of weld in the gage area increases.
ber of internal weld voids than the second-generation mate- The second-generation welds in Figure 6 show a less pro-
rial (as determined by radiography). All tests were conducted nounced degradation of ductility as the percentage of weld
more than 1 month after the welding operations were com- in the gage area increases.
pleted, and the specimens were stored at room temperature This observation led to the generation of the composite
(nominally at 23 8C). plot of specimen true strain at localization vs the percent of
The uniaxial tensile specimens consisted of five specimen specimen gage area comprised of weld metal shown in Fig-
designs. Figure 3 shows four of the specimens, which con- ure 7. True strain at localization was defined as the axial
sisted of both monolithic and weld material, in varying true strain developed in the specimen at peak true stress
amounts, in their gage areas. These specimens were designed immediately prior to fracture. Three sets of data are shown
with the weld positioned relative to the tensile axis such in Figure 7. The data set illustrates the monolithic material
that the areas of the 1- and 2-mm monolithic sheets were true strain at localization experienced using the various ten-
equal in cross section on either side of the weld. The mechan- sile specimen designs. The second and third data sets illus-
ical testing of the specimens required unique grips to uni- trate the effects of increasing weld percentage for the first-
formly hold the specimens. The grips consisted of stepped, and second-generation welds. Figure 7 shows that the first-
hardened steel pads that compensated for the differences in generation weld material appears to have a more detrimental

2756—VOLUME 31A, NOVEMBER 2000 METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 3—Drawings of four different tensile specimens with varying widths.

Fig. 4—Detailed drawing of the miniature tensile specimen.

effect on the tensile ductility than does the second-generation


weld material. There also was a considerable amount of Fig. 5—Uniaxial tensile results for first-generation tailor welded blank
scatter present in the true strain at localization given identical weld material. The data represent specimens of various width and percentage
specimen design within the same sample population. of weld in the gage area.
Another important observation is that no significant
change in the yield strength or strain-hardening characteris-
tics of the specimens was observed among the specimen The model chosen for application to the present problem
designs. Since changing the ratio of monolithic to weld of predicting the tensile instability of the specimens was
material in the gage area appears to have little impact on presented by Hart[12] and later elaborated by Ghosh.[13] The
the yield strength or hardening characteristics, it was con- model is based upon a commonly used method of assuming
cluded that using identical yield strength and constitutive and quantifying an initial imperfection level in the gage area
relations to carry out numerical modeling was sufficiently of the tensile specimens. This concept of an initial geometric
accurate for the given materials. imperfection is represented in Figure 8. The terminology
associated with the model can also be seen in Figure 8,
which theorizes that two separate regions exist in the gage
IV. DAMAGE MODEL
area. The first region is the homogeneous region, which
The quantitative description of the degradation of material consists of a uniform gage area. The second region is an
ductility in the weld was the focus of the modeling effort. area of imperfection that, for the purposes of the model, is

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 31A, NOVEMBER 2000—2757


the axial loads in the two regions are equal, and the cross-
sectional areas of the two regions differ, the stress levels in
the two regions are not equal. Writing an equation for axial
load equilibrium during the tensile test yields
si (1 2 f ) Aoe2«i 5 sh Aoe2«h [3]
where si , «i and sh , «h represent the stresses and strains
inside and outside the imperfection region, respectively, and
Ao represents the original cross-sectional area in the homoge-
neous region. The variable f is the nondimensional represen-
tation of the size of the initial imperfection. A value of f 5
0 represents no imperfection, and a value of f 5 0.20 repre-
sents a 20 pct reduction in cross-sectional area in the gage
area relative to the homogeneous region.
Obtaining a numerical solution of Eq. [3] requires an
associated constitutive law to describe the relationship
between stress and strain. The constitutive relation chosen
Fig. 6—Uniaxial tensile results for second-generation tailor welded blank
weld material. The data represent specimens of various width and percentage
for the current work is
of weld in the gage area. s 5 K«n«˚ m [4]
where n and m are the strain-hardening coefficient and strain-
rate sensitivity, respectively, and K refers to a strength coeffi-
cient. Substitution of this constitutive law into Eq. [3],
obtained from load equilibrium, yields
(1 2 f )e2«i«ni«˚ mi 5 e2«h«nh«˚ mh [5]
The strain-rate sensitivity of AA5182 aluminum alloy is
very near zero at room temperature and has been shown to
vary between slightly negative to slightly positive depending
on the strain rate.[14] For the purposes of the current work,
the strain-rate sensitivity was assumed to be zero. Applying
this assumption reduces Eq. [5] to
(1 2 f )e2«i«ni 5 e2«h«nh [6]
The relationship shown in Eq. [6] is the fundamental relation
between the initial damage level, the constitutive law, and
the strain both inside and outside of the imperfection region
Fig. 7—Summary plot illustrating the relationship between the percentage according to the model. The application of Eq. [6] involves
of weld in the gage area of a uniaxial tensile test and the achievable true
strain at localization.
choosing values for both n and f that represent the materials
of interest and numerically solving the equation to obtain
the relationship between «i and «h. Once the relationship
between strain in the imperfection and homogeneous regions
are known, specimen failure is defined at the point where
further uniaxial elongation results in no significant additional
plastic strain in the homogeneous region.
The primary interest of the current work was quantifying
the level of imperfection that exists within the weld material
of the subject tailor welded blanks. Using Eq. [6] and the
strain-hardening exponent of n 5 0.32, which was experi-
Fig. 8—Illustration of the concept of an initial geometric imperfection in mentally determined for the 1-mm monolithic sheet, a series
the specimen gage area. of numerical solutions were obtained. These results are pre-
sented in Figure 9. The results illustrate the effects of the
initial imperfection size on strain inside and outside the
equated to an area of reduced cross section that is the reason imperfection site. This family of curves shows the predicted
for, and the site of, strain localization and failure during levels of strain that will accumulate during a tensile test in
testing. This initial imperfection is also commonly described the homogeneous and imperfection region at various levels
as an initial damage. of initial imperfection size. Each curve asymptotically
The process of model development and solution involves approaches a maximum strain in the homogeneous region,
three major observations and steps. The fundamental obser- and the level of the maximum homogeneous strain degrades
vation is that, during the quasistatic analysis, the axial load quickly with the introduction of increasingly large geometric
supported across the homogeneous and imperfection cross- imperfections. The effect of the initial imperfection size on
sectional areas of the tensile specimen must be equal. Since the maximum predicted homogeneous strain is illustrated in

2758—VOLUME 31A, NOVEMBER 2000 METALLURGICAL AND MATERIALS TRANSACTIONS A


V. APPLICATION OF DAMAGE MODEL
The damage model presented in the current work assumes
that the undamaged weld material behaves in uniaxial ten-
sion identically to the monolithic material, except that weld
material possesses an increased level of geometric imperfec-
tion or damage. This assumption, together with the model
discussed above and the experimental results, was used to
quantitatively determine the level of imperfection in the
subject specimens that produces a correlation between the
experimental results and the model.
The first issue was establishing the level of damage in
the monolithic material that approximates the experimental
results of the monolithic material. The average value of true
strain at localization for the 1 mm monolithic specimens
of all specimen designs was 0.24. Iteratively solving the
preceding numerical model to determine the level of initial
imperfection that predicts this true strain at localization
yielded a value of f 5 0.01. This damage level in the mono-
lithic material agrees with damage levels proposed by other
investigators[13] and also shows that the monolithic materials
have relatively low imperfection levels.
Establishing the level of damage that exists solely in the
weld material was accomplished using the results of the
miniature tensile tests. These specimens contained only weld
material in their gage area and therefore offered the most
Fig. 9—Illustration of the strain inside vs outside the imperfection site for
various levels of initial imperfection as predicted by the model. The curves representative and exclusive test of the weld material. The
shown were produced using a strain-hardening exponent of n 5 0.32 and average true strain at localization for the first-generation
a strain-rate sensitivity of m 5 0. weld material was 0.10. Iteratively solving for the level of
imperfection required in the model to predict this level of
homogeneous strain yielded a value of f 5 0.14. The average
true strain at localization for the second-generation weld
material was 0.18. Iteratively solving for the level of imper-
fection required in the model to predict this level of homoge-
neous strain yielded a value of f 5 0.04.
The prediction of the true strain at localization of the
specimens that contained both monolithic and weld material
required a method to establish total damage based on two
independent damage levels. The method used was a linear
combination of the two damage levels based on the percent-
age of the materials that exist in the gage area of the speci-
men. The equation used to establish the composite level of
damage in these partial weld material specimens was

Amono fmono 1 Aweld fweld


f5 [7]
Ao

where Amono and Aweld represent the cross-sectional areas of


the monolithic and weld materials in the gage area, respec-
Fig. 10—An illustration of the effects of imperfection size on the strain
tively, and fmono and fweld represent the level of damage in
in the homogeneous area as predicted by the model. The curve shown was the monolithic and weld materials, respectively. Figure 11
produced using a strain-hardening exponent of n 5 0.32 and a strain-rate presents a family of curves that show the effect of various
sensitivity of m 5 0. levels of weld material imperfection on the model predicted
true strains at localization. For all curves in Figure 11, the
monolithic damage level was assumed to be f 5 0.01.
To compare the experimental results presented previously
Figure 10, which shows that an initial imperfection size of and the model predictions of the material performance, the
0.01, or 1 pct of the specimen cross section, yields a predicted experimental results were plotted with the results obtained
maximum homogeneous strain of approximately 0.24. Fig- from the numerical model. Figure 12 is a plot of the first-
ure 10 also shows that introducing an initial imperfection generation weld material experimental results and the model
of 0.2, or 20 pct of the specimen cross section, reduces the results, and Figure 13 is a plot of the second-generation
predicted maximum homogeneous strain to approximately experimental results and the model results. In both cases,
0.075. the level of imperfection in the monolithic material was f 5

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 31A, NOVEMBER 2000—2759


Fig. 14—Tensile results of miniature tensile specimens taken from the
Fig. 11—Summary plot illustrating the relationship between the percentage monolithic and weld material compared to the results of a standard ASTM
of weld in the gage area of a uniaxial tensile test and the achievable true A517 (12.7-mm gage width) tensile test. Miniature weld material specimen
strain at localization according to the model. shown was selected from a region with no detectable weld voids.

0.01. The level of imperfection in the first- and second-


generation materials was f 5 0.14 and 0.04, respectively.
Since the levels of imperfection in both the monolithic and
weld material were determined from the same experimental
data, it is not remarkable that the model predictions of aver-
age true strain at localization correlate well at the 0 and 100
pct weld in the gage area of the specimens. The significance
of the data shown in Figures 12 and 13 is the correlation
between the model results and the experimental data between
the conditions of 0 and 100 pct weld material in the gage area.

VI. DISCUSSION
A. Material Properties and Associated Assumptions
Establishing a model to describe the flow characteristics of
multiple materials within a single tensile specimen requires
either varying constitutive relations and compatibility condi-
Fig. 12—Summary plot illustrating the ability of the proposed damage tions or some simplifying assumptions. The current model
model to predict the achievable true strain at localization of the first- formulation assumes the yield strength, strain-hardening
generation welds. coefficient, and strain-rate sensitivity are equal in the mono-
lithic sheets and subject weld materials. The limitations of
this assumption are clear, particularly if one attempts to
apply the existing model to a steel tailor welded blank that
may have regions of relatively strong and brittle martensitic
structure in the weld material. Uniaxial tensile results com-
paring the true stress vs true strain for miniature monolithic,
miniature weld, and standard full-sized (ASTM E8) mono-
lithic materials are shown in Figure 14 for the subject
AA5182 materials. The miniature specimen data in this fig-
ure were collected from a region of a weld material that
had no detectable porosity, as observed via radiography and
microscopy. In Figure 14, it can be seen that the three differ-
ent stress vs strain curves agree well with the assumption
of identical constitutive relations. The existence of weld
voids and irregular external weld geometry is believed to
be the primary influence on the true strain at localization
for the subject welds.

B. Effects of Varying Experimental Specimen Design


Fig. 13—Summary plot illustrating the ability of the proposed damage
model to predict the achievable true strain at localization of the second- The experimental results of the uniaxial data exhibit sig-
generation welds. nificant variation in the true strain at localization in the

2760—VOLUME 31A, NOVEMBER 2000 METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 15—Illustration of the effects of varying the tensile specimen design
on the achievable true strain at localization. All data shown are from
monolithic material specimens prepared to the drawings in Figs. 3 and 4.

monolithic specimens. Some variation may be expected,


however. The current variation levels approaching 0.15 true
strain appear to be more than normally anticipated. The
cause for this may be understood by evaluating the true strain
at localization of the monolithic specimens as a function of
specimen design. Figure 15, an illustration of the true strain
at localization for the monolithic material as a function of
the specimen gage area, shows that the design of the speci-
men plays a significant role in the achievable true strain at
localization during tensile testing of monolithic materials.
This variation as a function of specimen design increases
the amount of variation that one might expect when trying
to determine the weld material properties with varying and
unique specimens. Since there is no standard that covers the
design of tailor welded blank tensile specimens, the unique
design that individual investigators use may result in varia-
tion in the reported properties.
The weld method and tensile specimen design may also
have a significant impact on the mechanical properties
reported for tailor welded blank weld materials. Laser weld-
ing, electron beam welding, gas tungsten arc welding, and
the variations of these methods used to produce tailor welded
blanks may all produce varying weld geometry. Each weld-
ing process may also invalidate the assumptions of identical Fig. 16—Micrographs of first-generation weld specimen fracture surface.
The micrographs show the overall weld region (center) and areas showing
constitutive laws in the weld and monolithic regions, or lack of fusion (top) and complete fusion (bottom). Approximately 20 pct
may produce a heat-affected zone near the weld. For these of the weld’s cross section shows lack of fusion.
reasons, the most effective way to analyze the mechanical
properties of the weld materials in tailor weld blanks in
uniaxial tension may be to perform testing using miniature C. Origins of Weld Metal Imperfections
specimens. The use of miniature specimens (containing only
In an effort to quantify the effects of different types of
weld material) avoids the need to somehow deconvolute imperfections, an investigation of the origin of imperfections
weld material properties from specimens that contain three was made. The traditionally held conception is that small
different materials with unique mechanical properties. Use microstructural or minute geometric imperfections on the
of miniature specimens may also offer the advantage of order of f 5 0.005 or 0.01 exist in every sheet material as
allowing researchers to accurately assess the effects of weld a practically unavoidable manufacturing legacy. However,
porosity on the weld material properties. Larger specimens this traditional perspective is based on analysis of wrought
effectively use monolithic materials to bridge the weld mate- monolithic materials, which have low levels of imperfection,
rial and associated defects that ultimately may prove detri- and this preconception of nearly immeasurable geometric
mental to the material, thereby overestimating the ductility imperfection is not valid for weld materials, which have
of the weld material. imperfection levels that approach f 5 0.15. In the case of

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 31A, NOVEMBER 2000—2761


material. The number and size of these voids varies signifi-
cantly with different weld process methods, and the number
of these voids appears to have a dominant effect on the weld
material true strain at localization. The primary differences
between the first- and second-generation materials evaluated
in this work were the number and size of the internal voids
in the material. The weld voids in the first-generation welds
are illustrated on the fracture surfaces in Figures 16 and 17.
Figure 16 shows that approximately 20 pct of the cross-
sectional area of the weld consists of a series of voids and
material exhibiting lack of fusion. The figure also shows
regions of the weld that exhibit good material fusion and
ductile fracture characteristics. Figure 17 is an illustration
of a fracture surface of a first-generation weld material from
a miniature tensile specimen. This miniature specimen also
shows voids on the fracture surface, along with regions of
ductile fracture. The fact that these voids are on the fracture
surface and are visibly larger in size as the true strain at
localization decreases suggests that voids are significant con-
tributors to strain localization and failure in aluminum tailor
welded blanks. Moreover, the existence of the voids on the
fracture surface lend creditability to the numerical modeling
of the tensile tests using a method based on the reduced
weld material ductility resulting from pre-existing geomet-
ric imperfections.

VII. SUMMARY
The ductility of autogenously welded AA5182-O alumi-
num tailor weld blanks was experimentally studied using a
variety of longitudinal tensile specimen designs. The results
of these studies show that increasing the percentage of weld
Fig. 17—Fracture surface of miniature tensile specimen. Specimen shows material in the cross-sectional area of the specimen decreases
regions of voids and lack of fusion (bottom) and full fusion regions with the ductility of the specimen and that the weld characteristics
ductile fracture characteristics. have a pronounced impact on ductility. Using the experimen-
tal results obtained, a level of weld metal imperfection was
established. This information was integrated with classical
the weld material, exterior surface irregularity and internal tensile instability and necking methods to develop a numeri-
voids or porosity probably are the primary contributors to cal model to quantitatively describe the ductility of the weld
geometric imperfection. metal in the tailor welded blanks. The results of the work
The surface roughness of the weld material varied show that the level of weld material porosity significantly
depending on the side and location of the weld evaluated. impacts the ductility of the weld material, and that the numer-
The surface roughness of the welds and the monolithic sheets ical model is capable of accurately describing the weld mate-
were evaluated using a standard stylus-type surface rial effect on the properties of longitudinal tensile specimens.
roughness tester (Mitutoyo SJ-201, Japan), with five sam-
pling lengths over a total length of 2.5 mm. The average
surface roughness (Ra) of the monolithic sheet was 0.8 mm, ACKNOWLEDGMENTS
whereas the average surface roughness of the weld material
was 8.7 mm. The maximum surface roughness (Ry) of the The Pacific Northwest National Laboratory is operated
monolithic sheet was 8 mm, and the maximum surface by Battelle for the United States Department of Energy. This
roughness of the weld material was 46 mm. Using the Ry work was supported by the United States Department of
of the weld material and dividing by the weld material Energy Office of Transportation Technology.
thickness yields a potential initial imperfection of nearly f
5 0.02, which results from the surface irregularity on one
side of the weld alone. Though the surface irregularity may REFERENCES
be significant, other factors may have a more substantial
impact. 1. B. Irving: Welding J., 1994, Dec., pp. 47-50.
An additional and likely dominant contribution to the 2. Automotive Eng., 1997, Dec., pp. 33-35.
imperfection in the weld material is the internal voids or 3. Automotive Eng., 1997, Dec., pp. 29-31.
4. J.M. Story, S. Heinemann, and S. Naefeler: Light Metal Age, 1998,
porosity that may form during solidification of the weld Oct., pp. 40-47.
metal. These voids, which are internal to the material, result 5. M.F. Shi, K.M. Pickett, and K.K. Bhatt: SAE Technical Paper Series,
in reduced cross-sectional areas in the gage area of the SAE paper 930278, SAE INTERNATIONAL, Warrendale, PA, 1993.

2762—VOLUME 31A, NOVEMBER 2000 METALLURGICAL AND MATERIALS TRANSACTIONS A


6. E.R. Pickering, M.A. Glagola, R.M. Ramage, and G.A. Taylor: SAE Mining, Metallurgy, and Petroleum, Montreal, Quebec, 1998. pp.
Technical Paper Series, SAE paper 950722, SAE INTERNATIONAL, 409-23.
Warrendale, PA, 1995. 10. Yoshida, Y. Yamasaki, S. Kabasawa, and M. Ono: NKK Tech. Rev.
7. M.C. Stasik and R.H. Wagoner: in Aluminum and Magnesium for No. 72, 1995, pp. 28-35.
Automotive Applications, J.D. Bryant, ed., TMS, Warrendale, PA, 11. M.J. Saran, E.R. Pickering, M.A. Glagola, and R.T. VanDyke: Materi-
1996, pp. 69-83. als and Body Testing, SAE Int. Body Engineering Conf. Expos., SAE
8. F.I. Saunders and R.H. Wagoner: Metall. Mater. Trans. A, 1996, vol. INTERNATIONAL, Warrendale, PA, 1995, pp. 120-24.
27A, pp.2605-16. 12. E.W. Hart: Acta Metall., 1967, vol. 15, pp. 351-55.
9. P. Martin, R. Zhang, B. Altshuller, M.J. Finn, and M.J. Worswick: 13. A.K. Ghosh: Acta Metall., 1977, vol. 25, pp. 1413-24.
Proc. Int. Symp. on Light Metals 1998, Calgary, AB, Canada, August 14. K. Higashi, T. Mukai, K. Kaizu, S. Tsuchida, and S. Tanimura: J.
16–19, 1998, M. Sahoo and C. Fradet, eds., Canadian Institute of Phys. IV, 1991, vol. 1, pp. 347-52.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 31A, NOVEMBER 2000—2763

You might also like