Failure Behavior of Heat-Affected Zones within HSLA-100

and HY-100 Steel Weldments

D. CHAE, C.J. YOUNG, D.M. GOTO, and D.A. KOSS

The deformation and fracture behavior of simulated heat-affected zones (HAZ) within HSLA-100
and HY-100 steel weldments has been studied as a function of stress state using notched and unnotched
axisymmetric tensile specimens. For the case of the HSLA-100 steel, the results for fine-grained, as
well as coarse-grain HAZ (CGHAZ) material, show that, despite large differences in the deformation
behavior when compared to base plate or weld metal, the failure strains are only weakly dependent
on the thermal history or microstructure. Ductile microvoid fracture dominates the failure of the
HSLA-100 steel with small losses of ductility occurring in the HAZ conditions only at high stress
triaxialities. In contrast, the HY-100 steel is susceptible to a large loss of ductility over all of the
stress states when subjected to a severe, single-pass simulation of a CGHAZ. The ductility loss is
greatest at the high stress triaxiality ratio in which case failure initiation occurs by a combination of
localized cleavage and ductile microvoid fracture.

I. INTRODUCTION and unnotched axisymmetric tensile specimens.[19,20] Those

THE performance of welded structures is usually limited results indicate a rapid decrease in failure strains with
by failure initiation within the heat affected zone (HAZ) of increasing stress triaxiality ratio, where stress triaxiality ratio
is defined as the ratio of mean stress to equivalent stress.
the base plate, especially the coarse-grained HAZ (CGHAZ)
Significant ductility was retained in the quenched-and-tem-
adjacent to the weld metal. In the case of high strength
pered base plate even at the highest stress triaxiality ratio.
steels, such as HY-100 and HSLA-100, designed for naval
In the present study, we extend the previous work to include
applications, several studies have addressed the issue of
toughness degradation within weldments by determining those HAZ microstructures that are most likely to limit the
failure of a welded structure based on either HY-100 or
mechanical properties, such as the Charpy V-notch impact
HSLA-100 steel: namely, a CGHAZ condition of both HY-
energy at low temperatures, the dynamic tear energy transi-
tion behavior from drop-weight tests, and the elastic-plastic 100 and HSLA-100 steels, as well as a fine-grained HAZ
fracture toughness response.[1–8] However, for structures (FGHAZ) microstructure of the HSLA-100 steel. In all cases,
made with high toughness steels, such as HY-100 and HSLA- we rely on a thermal simulation procedure to obtain the
100, and containing very small flaws, a local fracture representative microstructural conditions within the HAZ
approach is often the preferred method to predict structural using a Gleeble thermal simulator. As before,[19,20] notched
failure.[9–13] Such a design methodology does not rely on axisymmetric tensile specimens will be used to establish the
crack growth resistance or toughness parameters, but rather dependence of failure behavior on stress triaxiality for the
it depends on a strain-based failure criterion that accounts HAZ materials using the base plate failure response as a
for the fact that material properties, as well as multiaxial benchmark.
stress states within welded structures, vary on a local scale. Some important differences between the HY-100 and
The success of such an approach relies on accurate determi- HSLA-100 steels deserve comment. Both HY-100 and
nations of failure strains as a function of multiaxial stress HSLA-100 are high strength steels developed for naval
state[14–20] and for those microstructural conditions that limit applications. While these HY-100 and HSLA-100 steels
the performance of a welded structure. share similar deformation behavior (the yield strengths are in
In previous publications, we have examined the failure the 100 ksi range), they differ considerably in microstructural
behavior of HY-100 and HSLA-100 steels in the form of features, such as inclusions and second-phase particles, and
base plate and as a function of stress state using notched in their response to the thermal conditions imposed within
a HAZ of a weldment.[5–8,21] The HY-100 steel is character-
ized by rather large (approximately 50-␮m long), elongated,
lath-shaped MnS inclusions and finely distributed carbides.
D. CHAE, formerly Graduate Research Assistant with the Department
of Materials Science and Engineering, The Pennsylvania State University, HY-100 also has sufficient alloy and carbon content (0.16
is Research Engineer, Stainless Steel Research Group, Technical Research pct) to make it susceptible to a very high hardness within
Laboratories, Pohang Iron & Steel Co., Ltd., 1, Koedong-dong, Nam-ku, the HAZ under conditions of rapid cooling to room tempera-
Pohang, Kyungbuk 790-785, Republic of Korea. C.J. YOUNG, formerly ture. In contrast, the HSLA-100 steel has smaller (typically
with the Department of Materials Science and Engineering, The Pennsylva-
nia State University, is at 6701 King Court, Woodbridge, IL 60517. D.M. 1 to 5 ␮m) spherical inclusions, a much lower carbon content
GOTO, formerly Graduate Research Assistant with the Department of (0.06 pct) that results in comparatively few carbides in the
Materials Science and Engineering, The Pennsylvania State University, is microstructure, and, importantly, a resistance to the forma-
Engineer with Lawrence Livermore National Laboratory, Livermore, CA tion of high hardness regions within a HAZ. In both cases,
94551. D.A. KOSS, Professor, is with the Department of Materials Science
and Engineering, The Pennsylvania State University, University Park,
the volume fraction of sulfides, which usually serves as the
PA 16802. primary site for void nucleation during fracture, is low. Given
Manuscript submitted November 14, 2000. the differences in the inclusion characteristics and in the

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, SEPTEMBER 2001—2229

 Table I. Chemical Compositions of the Steels and 100S Wire Electrode Examined in the Current Study
Wt Pct C Mn P S Si Ni Cr Mo Cu Nb Ca
HY-100 0.16 0.26 0.008 0.009 0.22 2.62 1.32 0.25 0.14 — —
HSLA-100 0.06 0.86 0.011 0.003 0.26 3.45 0.75 0.60 1.56 0.034 0.0012
Weld electrode Mil-100S-1 0.08 max 1.25–1.80 0.012 0.008 max 0.20–0.55 1.40–2.10 0.30 max 0.25–0.55 — — —

anticipated response of these two steels to the thermal cycles

imposed during welding, we contrast the stress-state depen-
dence of the failure behavior between two simulated HAZ
conditions for each of these two steels. The failure behavior
of Mil-100S weld metal used in HSLA-100 weldments will
also be reported and compared to that of the base plate and
HAZ conditions.

II. EXPERIMENTAL PROCEDURE

A. Materials
The materials studied were HY-100 and HSLA-100 steel
base plates provided by the Caderock Division, Naval Sur- (a)
face Warfare Center (NSWC, West Bethesda, Maryland).
These two steels were in the form of commercial and hot-
rolled plates that had the chemical compositions listed in
Table I. The weld metal examined in this study was deposited
by multipass (5 passes), gas-metal arc welding using com-
mercially available Mil-100S-1 wire electrodes (manufac-
tured by ESAB Corp., Florence, South Carolina); see the
chemical composition in Table I.
The HY-100 steel base plate was in the form of a 25.4-
mm-thick plate that had been austenized at 900 ⬚C and
subsequently quenched and tempered at 593 ⬚C. The HSLA-
100 steel base plate was in the form of 19-mm-thick plate
that had been calcium treated for sulfide-shape control. That
plate was austenized at 904 ⬚C for 53 minutes and quenched
and tempered at 650 ⬚C for 53 minutes. It should be noted
that there is a small chemical difference in Ni content (b)
between this HSLA-100 and that used in a previous study.[19]
This HSLA-100 has a higher Ni content (3.45 pct vs 1.68
pct) than the previous HSLA-100 for which failure limit
diagram data were previously reported.[19] As will be shown
later in this study, these two versions of the HSLA-100
steel exhibit nearly identical failure behavior in the base-
plate condition.

B. Gleeble Simulation of HAZ Materials

The HAZ of a HSLA-100 steel weldment typically con-
sists of several regions with different microstructures,
depending on the location from the fusion line. A previous
study by Spanos et al.,[21] which was based on simulating
low heat input welding with little or no preheat of 2.25-in.-
(c)
thick HSLA-100 plate, identified two HAZ conditions that
exhibited very high hardnesses. Anticipating that these two Fig. 1—Gleeble thermal histories of (a) coarse-grain HAZ of HSLA-100
conditions might be most susceptible to a loss of ductility steel, (b) fine-grain HAZ of HSLA-100, the (c) partially reaustenitized
HAZ of HY-100 steel.
compared to the base plate, we used a Gleeble thermal
simulator to simulate the following two conditions: a
CGHAZ and a FGHAZ. For the HY-100 steel, the following Gleeble-heat treatments were performed at NSWC-Card-
two HAZ conditions were examined: (a) a CGHAZ and (b) erock using a Gleeble 1500 thermal cycle simulator. A rod-
a partially reaustenitized HAZ condition (PRHAZ). shaped specimen (16 mm in diameter ⫻ 115 mm in length)

2230—VOLUME 32A, SEPTEMBER 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A

was electrically heated with a rate of ⬃130 ⬚C/s to the peak values: R/␳ ⫽ 2 (“D-notch specimen”), 1 (“A-notch speci-
temperature. As depicted in Figures 1(a) and (b) for the men”), 0.5 (“B-notch specimen”), and 0.25 (“E-notch speci-
HSLA-100 steel, the peak temperatures for simulating a men”). In the case of the Gleeble-treated material, the
CGHAZ and a FGHAZ were 1250 ⬚C and 900 ⬚C, respec- minimum cross-sectional area of the notch was located at
tively. These temperatures were based on those required for the middle of the work-zone material. The tensile bars were
the initiation of austenization on heating (Ac1 ⬵ 720 ⬚C) and tested with the stress axis oriented in the long transverse
to complete austenization upon heating (Ac3 ⬵ 810 ⬚C) as direction of the plate; this is an orientation known to result
determined from the dilatometry at the heating rate of in the minimum ductility of the HY-100.[20] The tensile speci-
⬃130 ⬚C/s. The peak temperatures (1250 ⬚C and 900 ⬚C) mens were tested at an initial strain rate of 10⫺3/s to deter-
employed in the current HAZ simulations were all above mine the failure behavior. We used the experimental
the temperature for complete austenization of the HSLA procedure developed earlier[20] in which the onset of failure
steel (Ac3 ⬇ 810 ⬚C, as determined from dilatometry). In is determined using a criterion in which the loss of stress-
addition, producing the CGHAZ requires that the peak tem- carrying capacity of the material due to internal damage is
perature be high enough to induce the abnormal grain separated from that due to specimen necking by careful
growth. The temperature at which the abnormal grain growth examination of the characteristics of the load-displace-
occurs has been reported to be approximately 1100 ⬚C.[22] ment slope.
Therefore, the peak temperature, 1250 ⬚C, which is well In the case of the weld metal, tensile testing was performed
above 1100 ⬚C, was chosen for simulating a CGHAZ in the with the circumferentially-notched specimens directly ma-
HSLA-100. Based on the Spanos et al. study,[21] a fast cool- chined from the weldment in the as-deposited condition.
ing rate of approximately 60 ⬚C/s was chosen; no preheat Subsized circumferentially-notched tensile specimens (2R ⫽
was employed. 4.47 mm) were used in the E-notch and D-notch specimen
In the case of the HY-100 steel, a peak temperature of configurations. The notch center in the subsized specimen
1250 ⬚C was chosen for the CGHAZ case after previous was located in the middle of the deposited weld metal to
experience showed that a peak temperature of ⬇1300 ⬚C focus the deformation in a small material volume of the as-
resulted in local melting within the microstructure.[23] As deposited weld metal. It is important to recognize that a
with the HSLA-100 case, a “severe,” single-cycle heat treat- separate study of specimen size effects on failure of HY-
ment was chosen for the CGHAZ based on a heating rate 100 steel indicates no specimen size effect between speci-
of ⬇130 ⬚C/s followed by a rapid cool at a rate of ⬇60 mens with minimum diameters of 4.47 and 7.62 mm, which
⬚C/s. Examples of the single-cycle thermal profiles measured are the two conditions in question here.[24]
during the Gleeble heat treatments are shown in Figure 1(a). Compression testing was also performed to obtain the
The PRHAZ condition was produced using the two cycles stress-strain behavior using cylindrical specimens with a
shown in Figure 1(c). It was heated at ⬇130 ⬚C/s to 760 ⬚C height and diameter equal to 6.35 mm. Prior to placing the
(Ac1 ⬍ Tpeak ⬍ Ac3), cooled to 180 ⬚C, and then subjected specimens on the platens, molybdenum disulfide lubricant
to a second cycle to 650 ⬚C (Tpeak ⬍ Ac1) followed by a was applied on the specimen end-faces to decrease frictional
cooling rate of about 20 ⬚C/s. effects. In order to obtain a total strain value of ⬃0.5, the
The uniformity of the microstructures of the Gleeble- test was performed in two increments to decrease frictional
treated materials was probed using microhardness measure- effects. Following the first strain increment to a strain value
ments on the sectioned work-zone material. For both the of ⬃0.25, the test was interrupted, the samples were relubri-
HSLA-100 and the HY-100 steels, these hardness data indi- cated, and the second increment started until a total strain
cate that the microstructures transverse to the specimen value of ⬃0.5 was achieved on the specimens.
diameter were fairly uniform after the Gleeble treatments, Having measured the specimen diameter contraction at
and the lengthwise microstructure starts to change only near failure initiation and knowing the stress-strain response of
the copper jaws, well removed from the failure location the material, we subsequently used finite element analysis
within a test specimen. to determine the local stress-strain conditions at the center
Microstructural characterization using light microscopy of the minimum cross-sectional area of the notch where
was performed on specimens etched with either 2 pct nital failure initiated. Using procedures described previously,[19,20]
solution to reveal the lath microstructure of the martensite axisymmetric finite element analyses were performed for
in particular, or with Marshall’s reagent to reveal the prior each specimen geometry to determine the stress and strain
austenite grain size. In the latter case, in order to reveal states at failure initiation at the center of the minimum cross-
the prior austenite grain boundaries, the specimens were sectional area of the notch. The finite element analysis code,
tempered at 500 ⬚C for 48 hours.[22] ABAQUS, was used to determine the mean stress, ␴m , equiv-
alent stress, ␴eq , and equivalent plastic strain, ␧eq , in order
to identify the failure initiation condition in terms of the
C. Mechanical Test Procedure
stress triaxiality ratio, ␴m/␴eq , and the equivalent plastic
To determine the influence of stress states on the failure failure strain, ␧f,eq at the failure initiation site. Fractography
initiation behavior, circumferentially-notched tensile speci- was performed after tensile testing using scanning electron
mens were tested using four different notch acuities, each microscopy (SEM).
having a different radius of curvature, ␳, but the same initial
minimum notch diameter constant (2R ⫽ 7.62 mm). The III. RESULTS AND DISCUSSION
outer diameter of the notched specimens was equal to 15.24
mm, while the smooth bar specimens had a diameter of 7.62 A. Microstructural Response
mm and a gage length of 25 mm. The notch geometry, as The rationale for the heat treatments imposed on the
characterized by an R/␳ ratio, had the following range of HSLA-100 originated from a study by Spanos et al., who

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, SEPTEMBER 2001—2231

used dilatometry to investigate the microstructural changes
in a HSLA-100 steel that was thermally cycled to simulate
the HAZ of a thick, 57-mm base plate.[21] These investigators
imposed single-pass thermal histories in which two cooling
rates, 5 ⬚C/s and 60 ⬚C/s (between 800 ⬚C and 500 ⬚C) were
employed to simulate the microstructures induced by high
heat input and low heat input welding, respectively. For
each cooling rate, four typical HAZ microstructures were
produced by changing the peak temperatures. Microhardness
measurements that were performed on each microstructural
condition showed that the highest hardness region occurs in
the FGHAZ rather than in the CGHAZ if the heat input is
low (corresponding to the cooling rate of 60 ⬚C/s). The
presence of untempered martensite with a very small packet
size was suggested as the reason for the high hardness in
the FGHAZ. Spanos et al. suggest that, among the low heat
input microstructures, the FGHAZ may be the region of the
highest susceptibility to hydrogen-induced cracking.[21] We
also speculated that this high hardness condition might show
inferior fracture resistance, despite its fine grain size.
Assuming that the low heat input condition is most likely
to produce those HSLA-100 microstructures (both coarse-
grained and fine-grained) least resistant to fracture, the pres-
ent study examines the failure resistance of the low heat input
(rapid cooling rate) microstructures. Therefore, CGHAZ, as
well as FGHAZ conditions with low heat input (the cooling
rate of 60 ⬚C/sec), were chosen as the model HAZ conditions.
In short, the current study examines the extreme case where
the simulated HSLA-100 HAZ is likely to show the highest
susceptibility to fracture.
In the subsequent discussion, we also present limited Fig. 2—Microhardness profile along the “arrow path” across 100S weld
microstructure and hardness observations based on a metal and the HAZ within a HSLA-100 multipass (five passes) weldment.
multipass weldment of a HSLA-100 steel plate that is smaller
in thickness (19 mm) than that used in the Spanos et al.
study (57 mm). These observations are presented to compare
weldments of the HSLA-100 and HY-100 steels, respec-
our simulated, single-pass HAZ (the so-called worst case
tively. For the HSLA-100, the grain sizes and the appear-
condition) with an actual multipass HAZ and to show that
ances of the microstructures of the simulated HAZ materials
the microstructures and hardness are, in fact, fairly similar and the weldment HAZ regions are similar despite the antici-
in these two cases. pated slowing cooling rates in the weldment (Figure 3).
As is well known, the hardness values for weldments of Furthermore, while the as-received base plate (quenched and
these two steels depends on the proximity to the fusion line, aged) has a predominantly lathlike microstructure with the
an example of which is shown in Figure 2 for the HSLA- average prior austenite grain size of 16 ␮m, the simulated
100 steel. For this particular case, the 19-mm-thick steel HAZ materials consist of untempered martensites. Although
plate was welded with no preheat at 4.7 to 5.8 KJ/mm, which the grain sizes varied somewhat with location within the
correponds to levels of heat input such that cooling rates particular region of the HAZ, the average prior austenite
should be slower than the 60 ⬚C/s used in the Gleeble simula- grain size of 67 ␮m within the CGHAZ was much larger
tions. Nevertheless, hardnesses of ⬇350 to 360 Hv are than that (8 ␮m) of the FGHAZ.
observed in the CGHAZ region, and ⬇330 to 350 Hv are Figure 4 shows that the microstructure obtained in the
measured in the FGHAZ material. That CGHAZ value is Gleeble-treated HY-100 is fairly similar to that within the
greater than the 315 Hv for the Gleeble-simulated CGHAZ; coarse-grain region of the multipass weldment, again despite
the simulated FGHAZ has a similar hardness (330 Hv) to the somewhat slower cooling rate and tempering anticipated
that of the FGHAZ within the weldment. In the HY-100 in the multipass case. The prior austenite grain sizes do
steel with a base-plate hardness of 260 Hv, the simulated differ somewhat (92 ␮m for the Gleeble simulated structure
CGHAZ material had very high hardnesses of ⬇440 Hv, and 112 ␮m for the weldment CGHAZ). The simulated
which compare to a hardness of 420 Hv within the CGHAZ CGHAZ HY-100 is essentially untempered martensite, and
region of a preheated multipass (26 passes) weldment of a as such, it has a higher hardness (⬇440 Hv) when compared
25-mm-thick HY-100 steel plate. The increased hardness of to the slightly tempered weldment CGHAZ (⬇420 Hv).
the CGHAZ simulated HY-100 is a result of the severe, Light micrographs, as well as SEM fractographs taken from
single-pass heat treatment imposed with a high cooling rate the simulated HAZ materials, showed no sign of local melt-
(Figure 1(a)). ing of the inclusions. This result also suggests that the short
Figures 3 and 4 compare the microstructures of the simu- times encountered during the Gleeble heat-treatment should
lated HAZ materials with those of the HAZ regions in the not significantly change the inclusion characteristics.

2232—VOLUME 32A, SEPTEMBER 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A

Fig. 3—Light micrographs of HSLA-100 steel showing (a) the as-simulated CGHAZ, (b) the CGHAZ from the weldment, (c) the as-simulated FGHAZ,
and (d ) the FGHAZ from the weldment.

The Mil-100S weld metal microstructure in the as-depos- that the HY-100 steel is chemically banded such that the
ited multipass weld of a HSLA-100 base plate consists pri- inclusions are concentrated in bands roughly 26 ␮m wide
marily of acicular ferrites with an inclusion volume fraction and spaced 54 ␮m apart.[26] A detailed description of the
of the weld metal (Vf ⬇0.00072 ⫾ 0.00010). Although this density and size/spatial distributions of MnS inclusions
inclusion content is similar to that of the HSLA base plate within the HY-100 steel used in this study may also be
(Vf ⬇0.00065 ⫾ 0.00010), the size distribution of the inclu- found elsewhere.[27]
sions was quite different between the weld metal and the
base plate. In the case of the base plate, while most of the
inclusions have an average size of ⬇1 ␮m in diameter, the B. Deformation Response
volume fraction of these small inclusions (⬇1 ␮m) is less
than 10 pct of the total volume fraction. In fact, the dominant Given the significant differences in the thermal treatments
contribution to the volume fraction of inclusions in the base imposed on these two steels, we compare in Figures 5 and
plate is from inclusions greater than 10 ␮m in size. In the 6 the stress-strain responses of the HSLA-100 and HY-100
case of the weld metal, the small inclusions (⬇1 ␮m) domi- steels over the range of microstructural conditions induced
nate the total volume fraction such that a large population by the simulated HAZ treatments. For the case of the HSLA-
of small inclusions (⬇1 ␮m) constitute 70 pct of the total 100, Figure 5 indicates that the undermatched 100S weld
volume fraction. Inclusions greater than 5 ␮m in size were metal is much softer than the base metal (about 30 pct lower
not detected on either the metallographic sections or the yield strength), while the HAZ materials have significantly
fracture surfaces of failed 100S weld metal specimens. higher yield and flow stresses. As expected from previous
There is also a significant difference in inclusion micro- research[21] (and also as indicated by the hardness measure-
structures in the two base-plate materials. The inclusion ments), the FGHAZ condition of the HSLA-100 shows the
volume fraction of the HSLA-100 base plate was ⬵0.00065 highest yield strength. For example, the FGHAZ condition
and consisted of predominantly spherical inclusions with a has a yield stress 30 pct greater than that of the base plate
“bull’s-eye” type[25] microstructure due to the presence of and 85 pct greater than that of the weld metal. While the
elements, such as Ca. In contrast, despite its higher level of 100S weld metal has the lowest yield stress, it also exhibits
sulfur, the HY-100 steel has a lower volume fraction of the highest strain-hardening exponent (n ⫽ d ln ␴/d ln ␧ ⫽
inclusions (⬵0.00015), but these were dominated by large, 0.12), which compares to n ⫽ 0.077 for the base plate and
lath-shaped MnS elongated in the rolling direction. We note the low value of n ⫽ 0.035 for the FGHAZ. Thus, based

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, SEPTEMBER 2001—2233

 Fig. 6—Compressive stress-strain responses of HY-100 steel base plate,
the simulated CGHAZ material, and the PRHAZ material.

strength for the base plate. In contrast, the PRHAZ condition

shows a stress-strain response similar to that of the base
plate except for slightly higher strain hardening. It should
be recalled that the single cycle/rapid cool heat treatment
imposed on the HY-100 is likely to simulate the most severe
heat treatment a HY-100 HAZ might experience within a
multipass weldment. Thus, the difference in flow stresses
between the HY-100 base plate and the simulated CGHAZ
condition in Figure 6 is expected to represent the extreme
of the range of deformation responses expected within a
Fig. 4—Light micrographs of HY-100 steel showing (a) CGHAZ region HY-100 weldment.
from the weldment and (b) the as-simulated CGHAZ.

C. Failure Behavior
Material failure in this study is defined as that condition
in which the material damage is sufficiently severe such that
the stress-carrying capacity is measurably degraded. Thus,
using the procedure described elsewhere,[19,20,26] failure initi-
ation was determined experimentally as that point at which
the load-diametric contraction curve showed an abrupt drop
during the tensile test. This test procedure had the sensitivity
to detect roughly a 2 pct load loss due to damage accumula-
tion, which is consistent with a 3 pct area fraction of micro-
voids present on the fracture surfaces of specimens that were
strained to failure and subsequently fractured by cleavage
at liquid nitrogen temperatures.[20] In addition to the above
determination of failure strains, specimen fracture strains,
␧f , were also calculated for the HY-100 steel using initial
d0, and final diameter, df , measurements taken from the
broken tensile specimens and using the expression: ␧f ⫽ 2
ln (d0/df).
The failure limit diagrams of base-plate materials, as well
Fig. 5—Compressive stress-strain responses of HSLA-100 steel base plate, as simulated HAZ materials (a CGHAZ and a FGHAZ), are
simulated CGHAZ and FGHAZ materials, and 100S weld metal. presented in Figure 7 for HSLA-100 and in Figure 8 for
HY-100. The base-plate data are a combination of results
from both high Ni and low Ni base-plate materials; within
on these simulated heat treatments, a wide range of stress- experimental error, these two versions of HSLA-100 have
strain responses are anticipated within a HSLA-100 steel identical failure limits, which is consistent with their similar
weldment. inclusion contents. As shown in Figure 7, both HSLA-100
As shown in Figure 6, the simulated CGHAZ of the HY- HAZ materials exhibit failure behavior comparable to each
100 steel has a much higher yield stress than the base plate, other over the entire range of imposed stress triaxialities
approximately 1250 MPa as opposed to the 700 MPa yield investigated. Furthermore, both HAZ materials exhibit

2234—VOLUME 32A, SEPTEMBER 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A

 strains, the previous results are consistent with the hypothesis
that the ductile fracture resistance of these two steels is
controlled by the inclusion content. Matrix microstructure
and the associated deformation behavior appear to play only
minor roles in determining the tensile ductility over a range
of stress states.
The magnitude of the failure strains within the simulated
HAZ materials (Figure 7) suggests failure of the HSLA-100
by a ductile microvoid fracture process. As shown in Figure
9, the dimpled fracture surface features confirm ductile frac-
ture as a result of the coalescence of equiaxed microvoids.
Dimples having a wide range of size distribution but includ-
ing those up to 50 ␮m in diameter were found on the fracture
surfaces of both the HSLA base plate and the HAZ materials.
In contrast, the dimples on the fracture surface of the weld
metal were quite small (ⱕ10 ␮m in diameter), and they
have a relatively uniform size distribution. Such behavior
Fig. 7—Failure limit diagrams for the simulated CGHAZ and FGHAZ is also consistent with the role of the inclusions in controlling
conditions of a HSLA-100 steel weldment. The base-plate and weld-metal fracture. Only small inclusion (⬍5 ␮m) are present in the
behaviors are shown for comparison. weld metal microstrucre, while the base plate and HAZ
conditions contain inclusions with a range of sizes up to 20
␮m in diameter; these differences in inclusion sizes are
readily evident in the sizes of the particles at the bases of
the dimples in Figure 9. Finally, it is significant that we did
not find any evidence of cleavage in the central region of
the HSLA-100 specimens where failure initiates. The only
instance of cleavage being detected occurred near the shear
lip on the fracture surface of the D-notch of the CGHAZ
material, and such a location would be associated with frac-
ture propagation, not failure initiation.
Taken as a whole, the HSLA-100 steel failure limit data
and fracture surface observations indicate that, although
there are differences in the stress-strain response between
the base plate and HAZ conditions due to the thermal history
(Figure 5), a ductile fracture process persists, and the
resulting failure strains are only weakly dependent on the
microstructures. Such an inference is consistent with the
premise that, barring any change in fracture mechanism
and given the observed small void nucleation strain,[28] the
Fig. 8—Failure limit diagrams for the base plate and two simulated HAZ ductile fracture process in this steel is controlled primarily
conditions within a HY-100 steel weldment. Note that, for the base plate, by inclusion content, which remains similar among these
(⫹) denotes failure initiation due to equiaxed void growth and coalescence,
while (⫻) denotes failure initiation due to the void-sheet process.[19,20] heat-treat conditions. It should also be recognized that in
alloys where void nucleation is difficult and requires a sig-
nificant strain, increasing the material yield strength, such
somewhat smaller equivalent plastic strains to initiate failure as within a HAZ, should enhance void nucleation. Unlike our
at high stress triaxialities. In short, the HAZ conditions of situation, microstructure changes that increase yield strength
HSLA 100 remain ductile but with a somewhat reduced should contribute to a decreased ductility in this latter case.
level of ductility at high stress triaxialities. However, the In contrast, as shown in Figure 8, the Gleeble-simulated,
failure limit data between HAZ materials and the base-plate CGHAZ HY-100 steel shows a definite loss of failure strains
overlap at lower stress triaxialites (␴m/␴eq ⬍ 1) and are (i.e., the strain at the onset of material failure) for all stress
within the experimental scatter. triaxialities. These data are confirmed by measurements of
As is also shown in Figure 7, the failure curve for Mil- the fracture strains based on the measurements of the speci-
100S weld metal is nearly identical with that of the base men diameter after fracture. In this case, the CGHAZ mate-
plate. The similar failure behavior occurs despite the fact rial exhibits fracture strains roughly 40 pct of those of the
that the 100S alloy is significantly softer than the base plate base plate. At low stress triaxialities (␴m/␴eq ⱕ 1), the failure
and contains smaller inclusions. All inclusions in the 100S remained ductile and was characterized by dimpled fracture
are less than 5 ␮m in size, while those inclusions that domi- surfaces. As in previous observations,[19,20,26] the fracture
nate the volume fraction values in the HSLA-100 have sizes surface of the low stress triaxiality specimens was dominated
greater than 10 ␮m. Recalling that the inclusion content of by the coalescence of equiaxed voids (Figure 10(a)), while
the weld metal (Vf ⬇0.00078 ⫾ 0.00010) is comparable to the void-sheet mode of void coalescence was evident in the
that of the HSLA-100 steel base plate (Vf ⬇0.00065 ⫾ higher triaxiality (␴m/␴eq ⬇ 1) specimens (Figure 10(b)).
0.00010) and assuming similarly small void nucleation Such behavior is consistent with the failure behavior of

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, SEPTEMBER 2001—2235

Fig. 9—SEM images of the fracture surfaces of high stress triaxiality ratio D-notch specimens (␴m/␴eq ⫽ 1.4) for (a) HSLA-100 base plate, (b) simulated
CGHAZ of HSLA-100, (c) simulated FGHAZ HSLA-100, and (d ) 100S weld metal.

HY-100 steel in which two ductile failure mechanisms are of HSLA-100 and HY-100 steels that show significantly
present: a void coalescence process in which relatively equi- higher notch toughnesses at elevated temperatures in the
axed voids grow to impingement and a void-sheet process HSLA-100.[3,4] Also, previous results indicate improved
that links elongated inclusion-initiated voids by a shear resistance to hydrogen cracking of HSLA-100 weldments
instability.[19,20,26] compared to HY-100.[5,6,8] Thus, while a principal objective
At the highest stress triaxiality ratio (␴m/␴eq ⬇ 1.2), failure in developing HSLA-100 steel was a reduction in hull fabri-
of the CGHAZ HY-100 steel shows a pronounced departure cation costs,[5,6,8] HSLA-100 weldments should also perform
from the fracture mechanism observed in the lower stress better than their HY-100 counterparts under conditions lim-
triaxiality conditions. As shown in Figure 10(c), along with ited by tensile failure under multiaxial stress states.
microvoid coalscence, cleavage fracture is now clearly visi-
ble in regions spanning several grain diameters. Thus, frac-
ture initiation in this case occurs as a result of a mixture of IV. SUMMARY
cleavage and ductile fracture with the fraction of cleavage
fracture in the failure initiation region being approximately For the case of the HSLA-100 steel, when the failure
13 pct. This level of cleavage appears to be large enough behavior of the two simulated HAZ conditions is contrasted
to interfere with the void-sheet process normally seen in with those of the base plate, as well as the Mil-100s weld
HY-100 at high stress triaxialities; rather, the ductile fracture metal, our results indicate small losses of failure strains in
present occurs by coalescence of equiaxed voids, similar to the simulated HAZ conditions only at high stress triaxialities.
that in low stress triaxiality conditions. Since a significant The similarity in ductility behavior occurs despite the pres-
level of cleavage occurs in the failure initiation region, ence of pronounced differences in deformation behavior
clearly perturbing the fracture process, we conclude that (yield stresses ranging from 585 to 1080 MPa), as well as
localized cleavage is responsible for the low ductility exhib- microstructures (untempered martensite to acicular ferrite)
ited by the CGHAZ HY-100 steel at the highest stress among these four conditions. Importantly, fractography con-
triaxiality. firms that failure always occurs by a ductile fracture process
Finally, a comparison of the failure limit diagrams involving the coalescence of equiaxed microvoids. These
between the simulated HAZ conditions of these two steels results are thus consistent with the ductile failure of this
(Figures 7 and 8), as well as the associated fractography type of steel being dominated by the volume fraction of
(Figures 9 and 10), indicates the the HSLA-100 steel weld- inclusions, at least for the present case where void nucleation
ment should exhibit a superior fracture resistance when com- is easy. Thus, given the similar inclusion shape and content,
pared to a HY-100 weldment. Such a result is consistent the HAZ materials, as well as the weld metal, exhibit failure
with the Charpy V-notch transition curves of the CGHAZs strains comparable to those of the base plate, despite the

2236—VOLUME 32A, SEPTEMBER 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A

 in failure strains in the CGHAZ HY-100 under conditions
of high stress triaxiality.

ACKNOWLEDGMENTS
The authors are very grateful to J. Blackburn, D. Meldrom,
and R. DeNale in performing the Gleeble heat treatments.
Comments and discussions with E. Czyryca and J. McKirgan
are very much appreciated. We also appreciate very much
the encouragement of George Yoder and financial support
of the Office of Naval Research. One of the authors (DC)
acknowledges the support of the POSCO Scholarship So-
ciety.

REFERENCES
1. T.W. Montemarano, B.P. Sack, J.P. Gudas, M.G. Vassilaros, and H.H.
Vanderveldt: J. Ship Production, 1986, vol. 2, pp. 145-62.
2. P. Deb, K.D. Challenger, and A.E. Therrien: Metall. Trans. A, 1987,
vol. 18A, pp. 987-99.
3. M.R. Krishnadev, J.T. McGrath, J.T. Bowker, and S. Dionne: Welding
Metallurgy of Structural Steels, TMS, Warrendale, PA, 1987, pp.
137-55.
4. M.R. Krishnadev, J.T. Bowker, J.T. McGrath, V.K. Vasudevan, and
K.D. Challenger: in Recent Trends in Welding Science and Technology,
ASM INTERNATIONAL, Materials Park, OH, 1990, pp. 799-803.
5. E.J. Czyryca, E.J. Link, R.E. Wong, R.J. Aylor, T.W. Montemarano,
and J.P. Gudas: Nav. Eng. J., 1990, May, pp. 63-91.
6. E.J. Czyryca: in Metallurgy of Vacuum-Degassed Steel Products, R.
Pradham, ed., TMS, Warrendale, PA, 1990, pp. 469-85.
7. M.R. Krishnadev, W.L. Zhang, and J.T. Bowker: Int. Symp. on Low-
Carbon Steels for the 90’s, TMS, Warrendale, PA, 1993, pp. 501-09.
8. E.J. Czyryca: Key Eng. Mater., 1993, vols. 84–85, pp. 491-520.
9. D.M. Norris, J.E. Reaugh, B. Moran, and D.F. Quinones: J. Eng.
Mater. Technol., 1978, vol. 100, pp. 279-86.
10. A. Pineau: Proc. 5th Int. Conf. on Fracture, D. Francois, ed., Pergamon
Press, 1981, vol. 2, pp. 533-77.
11. F.M. Beremin: Metall. Trans. A, 1983, vol. 14A, pp. 2277-87.
12. G. Rousselier: Nucl. Eng. Des., 1987, vol. 105, pp. 97-111.
13. J. LeMaitre: Eng. Fract. Mech., 1986, vol. 25, pp. 523-37.
14. J.W. Hancock and A.C. MacKenzie: J. Mech. Phys. Solids, 1976, vol.
24, pp. 147-69.
15. A.D. MacKenzie, J.W. Hancock, and D.K. Brown: Eng. Fract. Mech.,
1977, vol. 9, pp. 167-88.
16. J.W. Hancock and D.K. Brown: J. Mech. Phys. Solids, 1983, vol. 31,
pp. 1-24.
17. J.W. Hancock and R.D. Thomson: Mater. Sci. Technol., 1985, vol. 1,
pp. 684-90.
18. G.R. Johnson and W.H. Cook: Eng. Fract. Mech., 1985, vol. 21, pp.
Fig. 10—SEM images of the fracture surfaces of HY-100 in the simulated
31-48.
CGHAZ condition failed at (a)␴m/␴eq ⫽ 0.70, (b) ␴m/␴eq ⫽ 1.05, and (c)
19. D.M. Goto, D. Chae, and D.A. Koss: Progress in Mechanical Behavior
␴m/␴eq ⫽ 1.25.
of Materials, Fleming Printing Co., Victoria, BC, 1999, pp. 150-55.
20. D.M. Goto, V. Jablokov, and D.A. Koss: Metall. Mater. Trans. A,
1999, vol. 30A, pp. 2835-42.
significant differences in inclusion sizes, microstructures, 21. G. Spanos, R.W. Fonda, R.A. Vandermeer, and A. Matuszeski: Metall.
Mater. Trans. A, 1995, vol. 26A, pp. 3277-93.
and hardnesses. 22. A.D. Wilson: Master’s Thesis, The Pennsylvania State University,
In contrast to the HSLA steel, the failure of HY-100 University Park, PA, 1996.
steel is quite sensitive to a severe, single-pass coarse-grain 23. D. Goto: Ph.D. Thesis, The Pennsylvania State University, University
simulated HAZ condition in which a hard (1250 MPa yield Park, PA, 1997.
strength), untempered martensite is present. While ductile 24. C. Young and D.A. Koss: The Pennsylvania State University, Univer-
sity Park, PA, unpublished research, 2000.
fracture is retained at low to intermediate stress triaxialities, 25. A.D. Wilson: in Inclusions and Their Influence on Material Behavior,
the failure strains decrease significantly in the CGHAZ con- ASM INTERNATIONAL, Materials Park, OH, 1989, pp. 21-34.
dition. Moreover, at the highest stress triaxiality ratio, this 26. D. Chae, D.A. Koss, A.L. Wilson, and P.R. Howell: Metall. Mater.
condition induces localized cleavage fracture in the HY-100. Trans. A, 2000, vol. 31A, pp. 995-1005.
27. A.B. Geltmacher and R.K. Everett: Scripta Mater., 1999, vol. 40, pp.
Although the associated fracture remains primarily ductile, 567-71.
the presence of this alternate mode of fracture accelerates 28. D. Chae: Ph.D. Thesis, The Pennsylvania State University, University
the fracture process such that there is a pronounced decrease Park, PA, 1999.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, SEPTEMBER 2001—2237