Validation of Hot Corrosion and Fatigue Models in HOTPITS: K. S. Chan

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

K. S.

Chan1
Department of Material Engineering,
Southwest Research Institute,
San Antonio, TX 78238
e-mail: kchansatx@icloud.com
Validation of Hot Corrosion and

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
J. T. Burns
Department of Materials Science and Engineering,
University of Virginia, Fatigue Models in HOTPITS
Charlottesville, VA 22904
e-mail: jtb5r@virginia.edu HOTPITS is a set of physics-based modeling tools for treating Type II hot corrosion in
Ni-based superalloys. The methodology includes modeling the nucleation, growth, and coa-
M. P. Enright lescence of pits and microcracks as a random process, as well as the transition of pits to
Department of Material Engineering, micrcracks and the propagation of the resulting large crack to failure. In this investigation,
Southwest Research Institute, critical experiments were performed on coupon and low-cycle fatigue (LCF) specimens in
San Antonio, TX 78238 order to validate the hot corrosion and the fatigue models in HOTPITS. The pit nucleation,
e-mail: menright@swri.org growth, and coalescence models in HOTPITS including the assumption of a random
process are validated by the hot corrosion critical experiments performed at two salt con-
J. Moody tents. The LCF critical experiments, performed using a marker band protocol, validated the
Department of Material Engineering, stress concentration factor-based models used to predict the pit-to-crack transition in the
Southwest Research Institute, HOTPITS tool. [DOI: 10.1115/1.4045710]
San Antonio, TX 78238
e-mail: jonathan.moody@swri.org Keywords: hot corrosion, model validation, pit-to-crack transition, pit nucleation, pit
growth, pit coalescence, low-cycle fatigue
W. Goodrum
Elder Research Inc.,
Charlottesville, VA 22901
e-mail: william.goodrum@elderresearch.com

1 Introduction extended to treat multiple pits and microcracks. In addition, a set


of criteria have been developed for predicting the onset of active
Type II hot corrosion is a concern in advanced gas engine compo-
and dormant (non-active) states for hot corrosion [13,16]. During
nents [1–11] that are used in coastal areas, over deserts, and volcanic
active states, hot corrosion occurs by the nucleation, growth, and
regions [12] where high chloride bearing salts and fine sulfate
coalescence of hot corrosion pits. During dormant or non-active
bearing sands can be ingested in the engines. This concern is
states, the progression of hot corrosion ceases but fatigue cracks
driven by two trends: (1) higher service temperatures and (2) the
may nucleate, grow, and coalesce at previously formed hot corro-
potential use of low-grade fuels or biomass-derived fuels with rela-
sion pits. To incorporate these damage processes during Type II
tively high sodium, sulfur, and chloride contents [11]. Under the
hot corrosion, the pit initiation and growth models have been
sponsorship by NASA, Elder Research Inc. (ERI) and Southwest
coupled with a pit coalescence model, a fatigue crack nucleation
Research Institute (SwRI) have developed a set of physics-based
model, a microcrack coalescence model, and an enhanced pit-to-
modeling tools, called HOTPITS, for predicting hot corrosion in
crack transition model. These advanced pit evolution models involv-
Ni-based superalloys [13]. This prototype methodology has been
ing multiple pits and microcracks are highlighted in boxes with a
implemented into a commercial probabilistic life-prediction code,
yellow background in Fig. 1. These algorithms have been imple-
called DARWIN [14], in order to facilitate the life prediction
mented in the DARWIN [14] and utilized to perform two benchmark
and risk assessment of Ni-based gas turbine engine components
computations to assess the capabilities of HOTPITS to predict the
subject to service conditions that may be susceptible to the occur-
various hot corrosion damage processes in a fictitious Ni-based
rence of hot corrosion attacks.
superalloy disk subject to realistic service conditions and loading
HOTPITS is a physics-based hot corrosion prediction methodol-
histories [15,16]. Detailed descriptions of the HOTPITS methodol-
ogy which has been developed according to the schematics outlined
ogy and the results of the benchmark computations are presented
in Fig. 1 [15,16]. The key features of the probabilistic framework
in the paper by Chan et al. [16].
are (1) a sulfate deposition model for predicting the formation of
This is the third of a series of three articles on HOTPITS. The first
a sulfate layer on hot-section components based on an input from
article [15] documents the model development and computational
contaminant concentrations in the fuel and in the air as well as rel-
results on the treatment of nucleation and growth of a hot corrosion
evant engine conditions, (2) a pit initiation model for predicting pit
pit in a turbine disk. The second article [16] documents the model
density, and (3) a pit growth model for predicting pit size as a func-
development and computational results on the treatment of the
tion of time of hot corrosion. The application of these three models
nucleation, growth, and coalescence of multiple hot corrosion pits
to simulate the initiation and growth of a single hot corrosion pit and
and microcracks in a turbine disk. The aim of this paper is to
its transition from a microcrack that ultimately propagates to failure
report the validation of various hot corrosion and fatigue models
has been reported in an earlier publication [15]. In a subsequent
in HOTPITS by performing critical experiments and subsequent
publication [16], the pit initiation and growth models have been
comparison of model predictions against pertinent experimental
data. The experimental procedures are presented in Sec. 2, which
include (1) hot corrosion tests of coupon specimens and round-bar
1
Corresponding author. low-cycle fatigue specimens, (2) characterization of hot corrosion
Contributed by the Materials Division of ASME for publication in the JOURNAL OF
ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received September 24, 2019;
pits in the pre-corroded coupon and low-cycle fatigue (LCF) spec-
final manuscript received November 20, 2019; published online December 13, 2019. imens, and (3) low-cycle fatigue testing of pre-corroded specimens
Assoc. Editor: Peter W. Chung. from crack nucleation to crack growth through loading with

Journal of Engineering Materials and Technology JULY 2020, Vol. 142 / 031001-1
Copyright © 2020 by ASME
2.1 Preparation of Pre-Corroded Specimens. The coupon
specimens were pre-corroded by applying a mixture of Na2SO4
and MgSO4 to ME3 specimens using the procedure provided by
NASA [19]. In particular, a salt mixture of Na2SO4 (4.9836 g)
and MgSO4 (3.33 g) by weight was dissolved in deionized water
(28.3 g) to produce a salt solution containing 22.7% salt and
77.3% water by weight. The salt solution was evaporated dry at

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
110 °C and subsequently made into a paste by adding ethyl cellu-
lose (0.6 g) and terpineol (24.86 g), followed by grinding for
30 min in a blender. The salt paste was then applied to the
coupon specimens and baked dry in an alumina boat at a furnace
at 400 °C for 1 h. The weights of the specimens before salt applica-
tion and after baking were utilized to determine the amount of salt
content on individual coupon specimens. Several attempts were
made until the salt content on individual specimens was about
2.0 mg/cm2. The salt-covered coupon specimens were then
exposed to a hot corrosion temperature in air in a furnace ranging
from 600 °C to 800 °C for a duration ranging from 1 h to 50 h.
These conditions were chosen in order to generate experimental
data of pit depth, width, density, and spacing as a function of expo-
sure time and temperature. For illustration, Fig. 2(a) shows an
optical micrograph of the pre-corroded coupon specimen
ME3 C-1 with salt deposits on the surfaces after it was exposed
to 700 °C for 2 h. A total of 15 coupon specimens were pre-corroded.
The same procedure was utilized to coat the entire gauge section
including the shoulder areas of ten LCF specimens. Two salt concen-
trations were targeted: (1) 2 mg/cm2 and (2) 30 mg/cm2.
After hot corrosion exposure, the pre-corroded coupons were
studied for a detailed 3D characterization of the pit depth, width,
density, and center-to-center spacing using white light interferome-
try (WTI) using a system called Zygo (Zygo Corp., Middlefield, CT)
and scanning electron microscopy (SEM). First, salt residue was
Fig. 1 Schematics of the hot corrosion prediction methodology cleaned from all coupons using a cleaning procedure that involved
called HOTPITS. Adapted from Chan et al. [16]. (Color version immersing the specimens in an ultrasonic cleaner with boiling
online.)

prescribed load-spectra to produce marker bands on fatigue sur-


faces. The results of the critical experiments are compared with
those of model predictions in Sec. 3, followed by discussion in
Sec. 4, and conclusions in Sec. 5.

2 Critical Experiments
The materials used in the validation of the hot corrosion and fatigue
models were powder metallurgy disk superalloy ME3, which were
supplied by NASA in the form of 20 pieces of heat-treated blanks
and four small pieces of heat-treated end blocks from forged disks.
The nominal compositions of ME3 were (in weight %) 21Co,
13Cr, 3.5Al, 3.5 Ti, 3.7 Mo, 2.5 Ta, 2.0 W, 0.8 Nb, 0.005Zr,
0.03B, 0.05C, and Ni balance [17]. All materials were solution-
treated and aged to attain the supersolvus microstructure comprised
of γ + γʹ. Detailed descriptions of the heat-treatment procedures
were presented by Gabb et al. [17]. The same ME3 materials were
used previously by Gabb et al. [6,7] and Telesman et al. [18]. All
20 pieces of ME3 were sent to a commercial vendor for machining
the gauge section, inertial welding of the gauge section to IN718
end pieces, and final machining of the LCF specimens. The dimen-
sions of the LCF specimens were 140 mm in length, which included
the gage section, extended shoulders, and threaded ends. The gage
section was 17.91 mm in length and 6.35 mm in diameter.
One of the end blocks was selected for preparing the coupon
specimens to be used in the hot corrosion tests. Three thin slices
(19 mm × 133.4 mm × 3.2 mm) were machined from the block
using electro-discharge machining (EDM). Five coupon specimens
of a dimension of 25.4 mm × 12.7 mm × 3.2 mm were machined
from each of the three thin slices by EDM to produce a total 15
coupon specimens. Each of the coupon specimens was mechani- Fig. 2 (a) Optical micrograph shows the salt deposits on
cally polished using Grit papers 240, 400, and 600, followed by pol- the corroded surfaces in ME3 after thermal exposure in air at
ishing wheels with solutions of 6 µm, 3 µm, and 0.5–1 µm alumina 700 °C for 2 hrs and (b) optical micrograph shows the LCF speci-
powders in water sequentially. men after hot corrosion exposure at 704 °C in air for 2 hrs

031001-2 / Vol. 142, JULY 2020 Transactions of the ASME


distilled water for 12 h during which the distilled water was brought ≈2 mg/cm2 at 704 °C for 2 hrs) and four had severe corrosion pre-
to boil prior to the start of the cleaning, but boiling conditions were damage (≈30 mg/cm2 at 704 °C for 2 hrs). For each condition (pris-
not maintained throughout the cleaning. After the coupons were tine, mild corrosion, and severe corrosion) two tests were performed
cleaned, surface measurements were taken using the Zygo white at high stress and two tests at low stress. The low-stress tests were
light interferometer. Following the surface topographical measure- performed at a stress range of 855 MPa (maximum stress of
ments, several sections were cut from selected coupon specimens 950 MPa), corresponding to 0.55% strain. The high-stress tests
(corrosion protocols: 700 °C for 2 h and 700 °C for 24 h). The had a stress range of 1165 MPa (maximum stress of 1294 MPa),

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
cross sections were polished and examined using an SEM to corresponding to a 0.7% strain.
measure the actual pit diameter and depth measurements. Energy
dispersive spectroscopy (EDS) compositional analysis was also
performed on selected specimens to identify the chemical composi-
tions of oxides formed near the pits. Oxides of chromium and alumi- 3 Model Validation
num were identified using EDS. 3.1 Pit Characteristics—Model Validation Against Coupon
The white-light interferometry-based 3D characterization of the Specimens. The pit density measurements were first utilized to
pre-corroded surfaces was performed (using Zygo) for eight LCF compare against similar pit density measurements for ME3 reported
round-bar specimens. Eighteen scans were taken on each specimen by Birbilis and Buchheit [8]. It is noted that the hot corrosion data
at 20-deg increments, completing 360 deg around the entire gauge reported by Birbilis and Buchheit [8] were generated using a
section of each specimen. A specimen manipulator was designed sodium sulfite/calcium sulfate/petrolatum mixture, while those for
and utilized to ensure precise positioning and 20-deg rotations. this study were generated using a salt mixture of 60% Na2SO4
Strict positioning of the specimens was maintained before each and 40% MgSO4 by weight. The difference in the salt mixture
scan so that each individual scan was done straight along the longi- resulted in a 20× difference in the pit density. Figure 3 shows a com-
tudinal dimension of the gauge section. This allowed for accurate parison of the pit density data from Birbilis and Buchheit [8] against
20-deg increments between scans. Each scan was also assured to those obtained from the coupon specimens from this investigation.
cover 20 mm along the longitudinal dimension as this was found A logarithmic scale is used for the pit density in Fig. 3(a) in order to
to be the precise gauge length outside of which the longitudinal cur- show both sets of data in one plot. Previously, Birbilis and Buchheit
vature of each specimen began to increase rapidly. Calculations data [8] were used to obtain the model constants in the hot corrosion
(assuming a hemispherical pit) of the relevant angle of incidence model (HOTPITS) and the computed pit density is shown as the
within the 20-degree field of view demonstrated that an error in solid curve in Fig. 3(a). To fit the new set of pit density data, the
the depth measurement of <1.5% would result from the oblique
versus nadir perspectives of the line of sight measurements. Further-
more, the WLI analysis was validated by comparison with multiple (a)
cross-sectional images to ensure data fidelity.
Post-processing of each of the individual scans was performed
using MoutainsMap software (by Digital Surf) to ensure proper
registration of the images, no “double-counting” of damage fea-
tures, and from removal of the gauge section curvature to isolate
characterization of the corrosion pitting. A Visual Basic (VGA)
code was developed to perform filtering of the corrosion data. Spe-
cifically, size (>5 micrometers) and geometry (aspect ratios
between 0.5–12) thresholds were established to exclude false
pitting indications that could arise from topical changes associated
with the machining finish of the specimen. The thresholds were
informed by either: (1) measurement fidelity of the Zygo WLI
instrument, or (2) enforce correspondence between the WLI
results and empirical data from digital optical microscopy of the
cross-sectional views of the pits.

2.2 Fatigue Testing at Elevated Temperature. Low-cycle (b)


fatigue testing of pre-corroded specimens was performed in an
MTS testing frame set up for the high-temperature fatigue tests. A
high-temperature furnace (MTS Model 653) was used to heat indi-
vidual test specimens. The furnace contained a dual heating
element, which reduced temperature gradients over the gage
section of the specimen. A total of 12 tests were performed at
704C at 0.33 Hz and a baseline R value of 0.1. A programmed
loading sequence was used to periodically mark the fracture
surface at known cycle counts. This process has been previously
employed to clearly identify the crack formation location, calculate
the crack initiation life, and determine the microstructurally small
crack growth rates [20,21]. While the marker-band sequence was
refined throughout the testing, it generally consisted of a large,
fixed number of baseline R = 0.1 cycles, followed by alternating
sequences of a very small number of R = 0.6 and R = 0.1 cycles, fol-
lowed by another large, fixed number of baseline R = 0.1 cycles.
This process would continue to be repeated and would lead to
fine but visible fracture surface markers that could be clearly iden- Fig. 3 Computed pit density against experimental data: (a) com-
tified; full details of this analysis can be found elsewhere [22]. Of puted pit density versus exposure time against experimental
the 12 specimens tested, four were pristine samples (no corrosion data of Birbilis and Buchheit [8] and the current study and
damage) and eight were pre-corroded samples. Of the eight cor- (b) comparison of computed pit density as a function of temper-
roded samples, four had mild corrosion pre-damage (salt content ature against experimental data from this study

Journal of Engineering Materials and Technology JULY 2020, Vol. 142 / 031001-3
(a)

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
(b)

Fig. 5 Lognormal distribution of pit size/pit spacing ratio pre-


dicted by HOTPITS on the basis that the location where hot cor-
rosion pits form is described by a random process

(a)

Fig. 4 Hot corrosion pits observed on ME3 coupon specimen


tested at 700 °C after 2 hrs of exposure show non-uniform pit
spacing and (b) a comparison of the distributions of the pit size
to pit spacing ratio for ME3 pre-corroded at 700 °C for 2, 24,
and 50 hrs. Lognormal distributions (curves) of pit size are
observed in all three specimens.

parameter N*MO, which represents the number of potential pit (b)


nucleation sites (oxides), was reduced from 900 to 45, while
other model constants were kept with the same values. The corre-
sponding computed pit density curve is shown as the dashed line
in Fig. 3(a). The agreement between the model and the coupon
data is good for exposure times of 20 and 50 hrs, but over-prediction
is observed for exposure times from 1 to 5 hrs. Figure 3(b) presents
the pit density as a function of exposure temperature using N*MO =
2. In this case, the there is good agreement between the model and
the experimental data on the dependence of pit density on exposure
temperature.
Figure 4(a) illustrates the distribution of type II hot corrosion
pits observed on one of the coupon specimens tested at 700 °C
for 2 hrs. The hot corrosion pits are seen to exhibit various pit
depths, diameters, and center-to-center spacing. The center-to-
center pit spacing values were measured for individual pits on
exposed coupon specimens to generate the corresponding pit
spacing probability distribution for modeling pit coalescence.
The experimental values of pit size to pit spacing of three Fig. 6 Validation of pit depth and width measurements against
coupon specimens (C1, C13, and C14) were analyzed using HOTPITS computations: (a) pit depth and width as a function of
Minitab to determine the distributions of pit size/pit spacing temperature and (b) validation of the Ac parameter in the pit
ratios for individual pre-corroded ME3 specimens. The results growth equation based on pit width data as a function of
temperature
for these three specimens are presented in Fig. 4(b), which
shows that the pit size/pit spacing ratios are well described by a
lognormal distribution and that the pit size/pit spacing ratio typical distribution of the computed λ values is presented in
tends to shift to higher values as the time of exposure increases Fig. 5, which has been determined to be a lognormal distribution.
from 2 hrs to 24 hrs, and to 50 hrs. The HOTPITS simulation has been performed on the basis that
For comparison, the ratio, λ, of the size to spacing of a pit and its the location where a hot corrosion pit forms is a random process.
nearest neighbor during a hot corrosion simulation using the The nucleating pit is allowed to coalesce with an existing pit if
HOTPITS prototype was computed during pit evolution. A the spacing between the two pits meets the coalescence criterion.

031001-4 / Vol. 142, JULY 2020 Transactions of the ASME


Such a random pit initiation process leads to a lognormal pit size/ (a)
pit spacing distribution, as shown in Fig. 5, which is in accor-
dance with the experimental observations shown in Fig. 4(b).
Since none of the experimental data shown in Fig. 4(b) were
used in the HOTPITS simulation, the HOTPITS result shown
in Fig. 5 is a true model prediction that is validated by the exper-
imental data shown in Fig. 4(b). The implication is that the loca-

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
tion where hot corrosion pits form can be described as a random
process.

3.2 Pit Characteristic − Model Validation Against


Low-Cycle Fatigue Specimen. The depth and width (i.e., dia-
meter) of individual corrosion pits on the surfaces of LCF speci-
mens were measured using the WLI technique are compared
against those measured on coupon specimens as well as those com-
puted based on HOTPITS using material constants fitted to the
coupon specimens in Fig. 6(a). The comparison revealed that
both the mean depth and width of pits on the surfaces of the LCF
specimens are slightly smaller than those on the coupon specimens (b)
at equivalent salt concentrations (about 2.0 mg/cm2). The scatters of
the pit depth and width values are quite large. The computed mean
values of the pit depth and width are well within the experimental
scatters. In earlier work on coupon specimens, the value of Ac
fitted to the data of Birbilis and Buchheit [8] was determined to
be 4.0. This value of Ac was found to underpredict the pit width
values on the pre-corroded coupon specimens in this program and
the value of Ac was increased to 7.0, Fig. 6(b). With the addition
of the pit depth and width measurements from the LCF specimens,
the mean of the two sets of pit width data now lies between the com-
puted curves for Ac = 4 and Ac = 7.0. Since the differences are small,
no change to the value of Ac was made.
The pit growth model in HOTPITS does not include salt concen-
tration explicitly, but is determined by the Na and S contents. To
validate this assumption, pit depth and pit diameter (i.e., width)
are plotted as a function of salt concentration in Figs. 7(a) and
7(b), respectively. The comparisons revealed that the mean pit
depth increased slightly with increasing salt concentration, as Fig. 7 Pit depth and diameter (width) of individual LCF speci-
mens as a function of salt concentrations: (a) pit depth and
shown in Fig. 7(a). The scatter, however, is so large that the
(b) pit diameter
increase may not be statistically significant. Similarly, the mean
pit diameter increases with increasing salt concentration with a
large scatter, as shown in Fig. 7(b).
agreement, but considerable scatter occurs in both sets of experi-
For both low and high salt concentrations, the normalized pit
mental data. The computed fatigue life curve is in good agreement
depth and diameter show frequency peaks near small values of pit
with both the NASA and UVA data. On the other hand, the pre-
depth and diameter. In addition, their distributions appear to
dicted crack nucleation life curves for kt of 1.4 and 2 are shorter
follow the lognormal distribution observed in the coupons speci-
than those observed in the pre-corroded LCF specimens.
mens. Plots of pit diameter normalized by the center-to-center pit
The discrepancy suggests that the kt value in the pre-corroded spec-
spacing are presented in Figs. 8(a) and 8(b) for low and high salt
imens may be lower than 1.4 or 2.0.
concentrations, respectively. As alluded to earlier, the lognormal
The validity of the Kitagawa diagram [23] for treating the growth
distribution was predicted by HOTPITS, as shown in Fig. 5. The
of fatigue cracks emanating from corrosion pits was examined by
lognormal distribution indicates that the location where hot corro-
plotting stress range versus pit depth and pit radius (pit diameter/
sion occurs is a random process; the random process occurs in
2) using pit size measurements from UVA and NASA, as shown
coupon specimens as well as in LCF specimens. Figure 9 presents
in Fig. 11. The figure indicates that the stress ranges employed in
a comparison of the predicted and observed frequency distribution
the UVA and NASA experiments were both higher than the antic-
of pit diameter normalized by pit spacing for M3-127. The compar-
ipated fatigue limit. As a result, the relevant parameter for fatigue
ison indicates very good agreement between the prediction and the
crack growth is the large-crack fatigue crack growth threshold,
experimental measurements for the pre-corroded LCF specimens.
ΔKth, which is depicted by the slanted line in Fig. 11. The figure
This validation further supports the notion that pit formation is gov-
also shows that for the UVA data, the pit depth data lie near to or
erned by a random process.
on the dashed line for the large-crack ΔKth and the pit radius data
all exceed the dashed line. This finding indicates that the stress
3.3 Crack Nucleation From Pits. The crack nucleation model intensity factors at the pits are either at the ΔKth or greater than
in HOTPITS was validated by comparing the predicted and mea- ΔKth; therefore, these pits can transition to become cracks and prop-
sured crack nucleation life and fatigue life (cycles-to-failure). Spe- agate to failure. In comparison, the NASA data also show the pit
cifically, the kt values in the range of 1.4 to 2.0, which were reported depths lie to the right of the dashed line and the pits can transition
by Telesman et al. [18] for hot corrosion pits in ME3, were used. to become cracks that propagate to failure. Attempts were made to
Other relevant model parameters are shown in Table 1. The com- perform fatigue life tests at lower stress ranges but the specimen did
puted crack nucleation and the fatigue life curves are compared not fail after 106 cycles. To speed up the test, the applied stress
against experimental data from this study (labelled as UVA data) range was raised in order to shorten the time of testing. These
and NASA data in the literature [7,18]. The summary plot in data are shown as black diamond data points in Fig. 11. The
Fig. 10 indicates that the NASA and UVA data are generally in results suggest that the pit-to-crack transition may be controlled

Journal of Engineering Materials and Technology JULY 2020, Vol. 142 / 031001-5
(a)

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
(b)

Fig. 10 Validation of the fatigue crack nucleation and failure


model against fatigue data of pre-corroded fatigue specimens
from this study (UVA data) and the literature (NASA data)

Fig. 8 Frequency of pit diameter (pit width) normalized by pit


spacing: low salt concentration and (b) high salt concentration
3.4 Pit-To-Crack Transition. In HOTPITS, the pit-to-crack
transition is determined by two criteria: (1) the fatigue crack
by the large-crack fatigue crack growth threshold, ΔKth. Figure 11 growth rate must exceed the pit growth rate and (2) the stress inten-
also shows schematically Type A transition (pits transitioning sity factor at the pit must exceed the large-crack fatigue crack
directly to cracks) and Type B transition (multiple pits coalesce to growth threshold, ΔKth. Strictly speaking, the first criterion is not
form a crack). These theoretical transition regimes were not tested applicable here since the hot-corrosion salt deposits have been
by the critical experiments due to high stress ranges chosen for removed from the specimen surfaces prior to fatigue of the pre-
the LCF tests. corroded specimens at elevated temperature. For inactive corrosion
pits, fatigue crack nucleation is expected at the bottom of the pits. At
the onset crack formation at a pit, the first criterion is better consid-
ered in terms of the fatigue crack nucleation rate rather than the
fatigue crack growth rate. Analysis of the crack nucleation at a pit
indicates that the first criterion is met at a critical pit depth, given
by Eq. (23) in Chan et al. [16], which has a value of 41.5 µm for
ME3 at 704 °C. In addition, the second criterion requires that the
stress intensity range at the pit must be greater the ΔKth value for
ME3 at 704 °C, which is about 9 MPa(m)1/2 [24]. To validate
these two criteria for treating pit-to-crack transition, the depth and
diameter (i.e., width) of all pits in individual specimens were ana-
lyzed to determine the number of pits that exhibited the following
characteristics: (1) a pit depth greater than the critical pit depth
(dcrit = 41.5 µm), (2) a pit diameter greater than 2dcrit (83 µm),
and (3) ΔK > ΔKth. A summary of these results is presented in
Table 1, which tabulates the salt concentration, stress range,
number of pits, pit density, mean and standard deviation (SD) of
pit depth, maximum pit depth, number of pits with depth greater
than dcrit, mean and SD of pit diameter, maximum pit diameter,
and the number of pits with diameter greater than 2dcrit for individ-
Fig. 9 Comparison of predicted frequency and measured nor-
ual specimens. The results in Table 1 indicate that none of the pit
malized pit diameter for M3-127 at high salt concentration depths exceed the critical values, but the pit diameters exceed
(≈2.0 mg/cm2). The good agreement provides a validation of 2dcrit in all LCF specimens except one (M3-128). The number of
the assumption that the location of the pit formation process is critical pits in individual specimens ranges from 0 to as high as
governed by a random process. 16. The number of critical pits in individual specimens appear to

031001-6 / Vol. 142, JULY 2020 Transactions of the ASME


with

1.13 − 0.07(d/a)1/2
F= 1/2
(2)
[1 + 1.46(d/a)1.64 ]

where d is the pit depth, and a is the pit radius. Table 2 presents the

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
computed maximum stress intensity range and number of pits with
ΔK > ΔKth with ΔKth = 9 MPa(m)1/2 [24]. Based on the ΔKth crite-
rion, the number of critical pits in individual specimens ranges from
0 to 6. Four of eight LCF specimens did not exhibit any pits with
ΔK > ΔKth while the remaining four LCF specimens satisfied the
ΔK > ΔKth criterion, as shown in Table 2. For illustration, Figs.
12(a)–12(c) present the frequency distribution of pit depth, pit dia-
meter, and ΔK value, respectively, for pits in M3-114. In this case,
there are six critical pits with ΔK > ΔKth. The maximum ΔK in
M3-114 was 11.61 MPa(m)1/2. For comparison, Figs. 13(a)–13(c)
present the frequency distribution of pit depth, pit diameter, and
ΔK value, respectively, for pits in M3-128. In this case, the
maximum ΔK computed based on the measured pit depth and dia-
meter was 6.1 MPa(m)1/2. Results of the frequency distribution of
pit depth, pit diameter, and ΔK values for all eight LCF specimens
are presented in Goodrum et al. [13].
A detailed examination of the SEM fractographs of pits on the
Fig. 11 Validation of the Kitagawa diagram [23] for treating the
propagation of fatigue cracks emanating from corrosion pits
fracture surfaces revealed that some of the pits that transitioned to
based on the large-crack fatigue crack threshold, ΔKth fatigue cracks were coalesced pits with ligaments. Figures 14(a)
and 14(b) present the SEM fractographs of pits in M3-114 and
M3-128, respectively, which show that multiple pits were involved
in the transition of pit to fatigue cracks. For these interacting pits,
depend on salt concentration. One specimen with low salt concen- the stress intensity range computed via Eq. (2) underestimated the
tration did not exhibit any critical pits, while all specimens with actual ΔK by a factor of 1.2 to 2, depending on the size of the over-
high salt concentration exhibited multiple critical pits. lapping ligament, as shown in Fig. 14. Using an interacting factor of
The stress intensity factors of individual critical pits were com- 1.25, the stress intensity ranges of the interacting pits were com-
puted on the basis of the pit depth and pit radius by treating the puted and are presented in Table 2. By considering interaction of
pit depth and pit radius as the crack depth and crack half-length overlapping pits in the ΔK computation, the number of critical
using the equation given by [25] pits was increased to at least one critical pit in each of the eight
LCF specimens, as shown in Table 2. Despite the uncertainty of
the interacting factor for the coalesced pits, the results strongly
ΔK = FΔσ(πd)1/2 (1) suggest that the pit-to-crack transition appears to be controlled by

Table 1 Summary of number of critical pits in individual LCF specimens based on critical pit depth or critical pit diameter

Pit
Salt No. Pit Pit depth Pit No. Pit Pit Pit No.
Con. ΔS, of density depth mean, SD, depth Max, > diameter mean, diameter diameter Max, >
Specimen mg/cm2 MPa Pits no/mm2 µm µm µm dcrit µm SD, µm µm dcrit

M3-155 1.8420 1165.0 143 0.40 6.14 2.69 18.2 0 12.49 17.41 120.8 3
M3-128 1.9209 1165.0 225 0.63 5.05 1.17 12.7 0 6.81 3.43 42.6 0
M3-157 2.2893 1055.0 204 0.57 5.63 2.02 16 0 9.68 10.38 107 1
M3-166 2.5787 1055.0 381 1.07 6.29 3.62 31.5 0 13.84 20.27 207.4 8
M3-127 27.4452 1165.0 148 0.41 6.87 3.22 20.5 0 12.92 19.41 122.3 4
M3-114 29.3924 1165.0 302 0.85 8.41 4.15 30.3 0 20.56 32.31 306.2 12
M3-152 29.4977 1055.0 425 1.19 9.01 5.12 30.2 0 22.15 24.91 188.8 16
M3-134 39.1548 1055.0 1590 4.45 6.93 2.76 22.65 0 11.78 11.52 132.9 9

Table 2 Summary of number of critical pits based on ΔKth

Specimen Salt Con. mg/cm2 ΔS, MPa Maximum ΔK, MPa(m)1/2 No. > ΔKth Interacting Pits ΔK, MPa(m)1/2 No. > ΔKth da/dN, mm/cycle
a
M3-155 1.8420 1165.0 8.76 0 10.95 1 1.10E-04
M3-128 1.9209 1165.0 6.1 0 7.63a 0 2.35E-05
M3-157 2.2893 1055.0 7.45 0 9.31a 1 3.00E-05
M3-166 2.5787 1055.0 9.48 1 – 1 7.12E-05
M3-127 27.4452 1165.0 9.13 1 – 1 5.34E-05
M3-114 29.3924 1165.0 11.61 6 – 6 3.00E-05
M3-152 29.4977 1055.0 9.23 3 – 3 4.40E-05
M3-134 39.1548 1055.0 8.51 0 10.64a 1 4.7E-05

FA = 1.25
a

Journal of Engineering Materials and Technology JULY 2020, Vol. 142 / 031001-7
Table 3 A comparison of ΔK of pits based on WLI and SEM measurements

WLI Pit WLI Pit Pit


Specimen Δσ, MPa dia, µm dia/2. µm depth, µm a/c F ΔK, MPa(m)1/2

M3-157 1055 110 55 5 0.0909 1.0934 4.57


M3-152 1055 189 94.5 23 0.2434 1.0243 9.19

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
M3-134 1055 86 43 23 0.5349 0.8741 7.84
1055 80 40 14 0.3500 0.9694 6.78
1055 73 36.5 15 0.4110 0.9375 6.79
1055 42 21 15 0.7143 0.7893 5.72
1055 80 40 13 0.3250 0.9825 6.62
1055 30 15 9 0.6000 0.8422 4.72
1055 8 4 6 1.5000 0.5330 2.44
Mean 6.07
SD 1.98

SEM Pit SEM Pit Pit


Specimen Δσ, MPa dia, µm dia/2. µm depth, µm a/c F ΔK, MPa(m)1/2
M3-157 1055 102 51 21 0.4118 0.9371 8.03
M3-152 1055 189 94.5 28 0.2963 0.9973 9.87
M3-134 1055 97 48.5 23 0.4742 0.9048 8.11
1055 70 35 14 0.4000 0.9433 6.60
1055 102 51 18 0.3529 0.9679 7.68
1055 75 37.5 21 0.5600 0.8616 7.38
1055 76 38 13 0.3421 0.9735 6.56
Mean 7.75
SD 1.12

Fig. 12 Frequency distributions of pit depth, pit diameter, and Fig. 13 Frequency distributions of pit depth, pit diameter, and
ΔK values for M3-114: (a) pit depth distribution, (b) pit diameter ΔK values for M3-114: (a) pit depth distribution, (b) pit diameter
distribution, and (c) ΔK distribution distribution, and (c) ΔK distribution

031001-8 / Vol. 142, JULY 2020 Transactions of the ASME


Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
Fig. 14 Pit geometry near fatigue crack nucleation/propagation site for selected LCF specimens: (a) M3-114 and (b) M3-128

the large-crack ΔKth. The dcrit criterion was not obeyed in the LCF presented in the last column of Table 2 . These results are compared
specimens, probably due to the fact that hot corrosion was not active against the large-crack da/dN data of ME3 704 °C in Fig. 16, which
during fatigue testing of the LCF specimens since most, if not all, of shows the critical pits propagate at rates comparable to those of the
the salts were removed prior to fatigue testing. large cracks at ΔK ranges above the large-crack threshold at ΔKth =
The pitting and fatigue mechanisms simulated by HOTPITS are 9 MPa(m)1/2 [24]. This finding validated the pit-to-crack transition
consistent with the hot corrosion mechanisms in ME3 reported by criteria implemented in HOTPITS.
Gabb et al. [7] and Telesman et al. [18], who demonstrated that A comparison of the ΔK values at the largest pits before and after
the stress concentration factor of the corrosion pits evolves as the fatigue loading of selected LCF specimens were also performed.
pits undergo significant changes in morphology due to individual For this comparison, the pit diameter and depth measurements
growth and coalescence with other pits. Nucleation of fatigue before fatigue loading were obtained from white light interferome-
cracks at coalesced surface pits was also observed in this investiga- try (WLI), while SEM measurements were used after fatigue
tion, as reported by Goodrum et al. [13]. For illustration, Fig. 15 loading. This type of comparison was done for M3-157, M3-152,
shows fatigue nucleation at irregularly shaped surface pits formed and M3-134, as presented in Table 3. The comparison revealed
by the joining of two closely spaced smaller semi-ellipsoidal pits. that the pit depths and pit diameters determined from SEM were
Multiple fatigue cracks are seen to nucleate and propagate from generally higher than those obtained by WLI. As a result, the ΔK
the coalesced pits. These microcracks subsequently linked together values at the pits were higher when computed based on the SEM
to form a large fatigue crack that propagated to fracture. The pit depth and pit diameter values. It should be noted that pit interac-
observed pitting and fatigue processes, which involved interaction tions were not considered in this set of ΔK computations. Based on
and coalescence of multiple hot-corrosion pits that subsequently these ΔK calculations, it can be see that the largest pit in M3-152
transition to microcracks under fatigue loading, validated the hot was sufficiently large to exceed the ΔKth (ΔKth = 9.0 MPa√m)
corrosion damage and fatigue crack nucleation and growth pro- before fatigue testing. In contrast, the ΔK values associated with
cesses implemented in HOTPITS. the largest pits in M3-157 and M3-134 were not sufficiently high
The fatigue crack growth rate, da/dN, for the critical pits were to exceed the ΔKth before fatigue testing. Thus, these pits need to
measured for some of the LCF specimens. These results are grow, coalesce, or nucleate fatigue crack at the pit tip in order to
become a propagating fatigue crack. For both M3-157 and
M3-134, the ΔK values were about 8 MPa√m or higher without

Fig. 15 SEM fractograph illustrates the formation and propaga-


tion of fatigue cracks from coalesced pits on a pre-corroded ME3
specimen fatigue tested at 704 °C. The micrograph depicts the
nucleation of multiple fatigue cracks from coalesced hot- Fig. 16 Critical pits in pre-corroded ME3 propagated at rates
corrosion pits. The multiple fatigue cracks subsequently coa- comparable to large-crack at stress ranges above the large-crack
lesced to from a propagating large fatigue crack. threshold in ME3 [24] at 704 °C

Journal of Engineering Materials and Technology JULY 2020, Vol. 142 / 031001-9
considering pit interaction associated coalesced pits. Since pit inter- (1) The pit nucleation, growth, and coalescence models in
action can increase the local ΔK by a factor of at least 1.25, the ΔK HOTPITS are validated by hot corrosion critical experiments
values for the coalesced pit in M3-157 and M3-134 could be as high performed to characterize the depth, width, density, and
as 10 MPa√m, far exceeding the ΔKth of 9 MPa√m required for center-to-center spacing of hot corrosion pits.
the pit-to-crack transition. This finding points to the need to (2) The assumption used in HOTPITS to treat the locations
include pit interactions in the ΔK computation. where hot corrosion pits nucleate as a random process was
validated by the crucial experiments performed on coupon

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
and LCF specimens. Both sets of experiments show the pit
4 Discussion depth to center-to-center spacing ratios are described by log-
normal distributions, validating the independent prediction
One of the key findings of this investigation is the validation that
by HOTPITS.
the locations where hot pits form occur through a random process.
(3) Pit coalescence is an important part of the hot corrosion
The consequence of the random process is that the pit size/
process. The coalesced hot corrosion pits serve as stress con-
center-to-center spacing ratios of the hot pits are predicted by
centration sites where fatigue cracks may nucleate and prop-
HOTPITS to obey a lognormal distribution, which was validated
agate away from the pits.
by both coupon and LCF pre-corroded specimens, as shown in
(4) The crack nucleation fatigue model in HOTPITS was vali-
Figs. 4, 5, and 9. Beside the correct distribution, the pit depth and
dated by testing pre-corroded LCF specimens at 704 °C at
the pit width kinetics as well as their temperature dependence are
0.33 Hz and monitoring crack nucleation and growth
also validated to within experimental scatters, as shown in Figs. 3
through marker band loading at various stress ratios.
and 6.
(5) The existence of a critical pit size (width) at the pit-to-crack
A novel innovation of HOTPITS is the treatment of the nucle-
transition was validated through a detailed analysis of the pits
ation, growth, and coalescence of multiple pits at random locations.
observed in fatigued, pre-corroded LCF specimens.
The coalescence of multiple pits are governed by the ratio of
(6) The transition of a critical pit to become a propagating crack
ligament width to center-to-center spacing. These assumptions are
at ΔK > ΔKth was confirmed by critical experiments per-
supported and confirmed by fractography performed on the pre-
formed on pre-corroded LCF specimens.
corroded LCF specimens which show coalesced pits on the pre-
(7) The use of a set of critical pit size and the large-crack thresh-
corroded LCF specimens, as shown in Figs. 14 and 15. Additional
old criteria for determining the transition of pit-to-crack was
micrographs of these coalesced pits on pre-corroded LCF speci-
validated by the critical experiments performed in this inves-
mens can be found in Goodrum et al. [13] and in Jamieson [22].
tigation.
These micrographs also provide evidence that the coalesced pits
acted as stress concentration sites where fatigue cracks are nucle-
ated during fatigue testing. The models for predicting the stress con-
centration factors for coalesced hot corrosion pits and the Acknowledgment
cycles-to-crack nucleation are validated by the experimental data The authors would like to acknowledge the financial support of
to the degree within the experimental scatter, as shown in Fig. 10. the NASA STTR Program through Contract No. NNX15CC33C
The large variability in the cycle-to-crack nucleation may be attrib- and the encouragement by Dr. Jack Telesman, Program Monitor.
uted to the variations of the coalesced pit geometry and its influence The authors are thankful to Dr. Tim P. Gabb for providing
on the stress concentration factor. Prior work has shown that kt can the ME3 materials, Dr. Jack Telesman for providing the ME3
range from 1.36 to 2.85 [18] for hot corrosion pits as the pit geom- fatigue data, Ms. Susan L. Draper for providing the hot corrosion
etry evolves with corrosion times. Future work is required to iden- procedure, and Dr. James A. Nesbitt for providing guidance on
tify the source of discrepancies and the approach for treating the the pit measurements using white light interferometry. The
variability of pit geometry. clerical assistance of Ms. Loretta Mesa of Southwest Research
The results of this study provide strong evidence for the existence Institute (SwRI®) in the preparation of the paper is also
of a critical pit size (depth or width) beyond which the crack nucle- acknowledged.
ation rate exceeds the pit growth so the nucleated fatigue crack can
outrun the pit. Of all the pits examined, the critical pit dimension is
obeyed in the width direction, probably due to the fact that the coa-
lesced pits exhibit a width that is large than the depth. In addition, References
the Kitagawa plot [23] in Fig. 11 provides strong evidence that the [1] Goebel, J. A., Pettit, F. S., and Goward, G. W., 1973, “Mechanisms for the Hot
pit-to-crack transition occurs when the ΔK at the pit exceeds the Corrosion of Nickel-Based Alloys,” Metall. Trans., 4(1), pp. 261–278.
[2] Chiang, K. T., Pettit, F. S., and Meier, G. H., 1983, High Temperature Corrosion,
large-crack threshold (i.e., ΔK > ΔKth). This pit-to-crack transition R. A. Rapp, ed., National Association of Corrosion Engineers, Houston, TX, pp.
criterion is validated by a detailed examination of the local ΔK of 519–530.
critical pits in individual specimens, once the local geometry of [3] Pettit, F. S., and Meier, G. H., 1984, “Oxidation and Hot Corrosion of
the coalesced pits are taken into account. Furthermore, measure- Superalloys,” Superalloys 84, Gell, M., Kortovich, C. S., Bricknell, R. H.,
Kent, W. B. and Radavich, J. F., eds., TMS, Warrendale, PA, pp. 651–687.
ments of local crack growth rates of fatigue cracks emanating [4] Rapp, R., 2002, “Hot Corrosion of Materials; a Fluxing Mechanism?,” Corros.
from the critical pits showed that they clustered near the large-crack Sci., 44(2), pp. 209–221.
threshold region of the baseline da/dN curve, as shown in Fig. 16. [5] Groh, J. R., and Duvelius, R. W., 2001, “Influence of Corrosion Pitting on Alloy
Thus, both the critical pit size and the large-crack threshold criteria 718 Fatigue Capability,” Superalloys 718, 625, 706 and Various Derivatives,
L. A. Loria, ed., TMS, Warrendale, PA, pp. 583–592.
for the pit-to-crack transition in HOTPITS were validated by the [6] Gabb, T. P., Telesman, J., Kantzos, P. T., Smith, J. W., and Browning, P. F., 2004,
critical experiments. “Effects of High Temperature Exposures on Fatigue Life of Disk Superalloys,”
Superalloys 2004, K. A. Green, T. M. Pollock, H. Harada, T. E. Howson, R. C.
Reed, J. J. Schirra, and S. Watson, eds., TMS, Warrendale, PA, pp. 269–274.
[7] Gabb, T. P., Telesman, J., Hazel, B., and Mourer, D. P., 2010, “The Effects of Hot
5 Conclusions Corrosion Pits on the Fatigue Resistance of a Disk Superalloy,” J. Mater. Eng.
Perform., 19(1), pp. 77–89.
The aim of this investigation is to validate the hot corrosion and [8] Birbilis, N., and Buchheit, R. G., 2008, “Measurement and Discussion of Low-
fatigue models in HOTPITS, which is a set of physics-based mod- Temperature Hot Corrosion Damage Accumulation Upon Nickel-Based
eling tools for treating Type II hot corrosion in Ni-based superal- Superalloy René 104,” Metall. Mater. Trans. A, 39(13), pp. 3224–3232.
loys. This methodology has been implemented in DARWIN, a [9] Luthra, K. L., and Shores, D. A., 1980, “Mechanism of Na2SO4 Induced
Corrosion at 600°–900°C,” J. Electrochem. Soc., 127(10), pp. 2202–2210.
commercial probabilistic life-prediction and risk assessment code [10] Gokoglu, S. A., and Santorp, G. J., 1987, High Temperature Alloys for Gas
for gas-turbine engine components. The conclusions reached as Turbines and Other Applications, W. Betz, R. Brunetaud, D. Coutsouradis, H.
the result of this investigation are as follows: Fischmeister, T. B. Gibbons, I. Kvernes, Y. Lindblom, J. B. Marriott, and

031001-10 / Vol. 142, JULY 2020 Transactions of the ASME


D. B. Meadowcroft, eds., Reidel Publishing Co., Dordrecht, Germany, pp. 1117– [17] Gabb, T. P., Garg, A., Ellis, D. L., and O’Connor, K. M., 2004, “Detailed
1126. Microstructural Characterization of the Disk Alloy ME3,” NASA/
[11] Leyens, C., Wright, I. G., and Pint, B. A., 1999, “Hot Corrosion of Nickel-Based TM-2004-213066, NASA Glenn Research Center, Cleveland, OH.
Alloys in Biomass-Derived Fuel Simulated Atmosphere,” Elevated Temperature [18] Telesman, J., Gabb, T. P., Yamada, Y., and Draper, S. L., 2016, “Fatigue
Coatings: Science and Technology III, ed. J. M. Hampikian and N. B. Dahotre, Resistance of a Hot Corrosion Exposed Disk Superalloy at Varied Test
eds., TMS, Warrendale, PA, pp. 79–90. Temperatures,” Mater. High Temp., 33(4–5), pp. 517–527.
[12] Encinas-Oropesa, A., Drew, G. L., Hardy, M. C., Leggett, A. J., Nicholls, J. R., [19] Draper, S. L., 2015, Private communication, NASA Glenn Research Center,
and Simms, N. J., “Effects of Oxidation and Hot Corrosion in a Nickel Disc Cleveland, OH.

Downloaded from https://asmedigitalcollection.asme.org/materialstechnology/article-pdf/142/3/031001/6482197/mats_142_3_031001.pdf by Visvesvaraya Technonological Univ-Belgaum (VTU) Consortia user on 11 June 2020
Alloy,” in Superalloy 2008, R. C. Reed, K. A. Green, P. Caron, T. P. Gabb, [20] Donahue, J. R., and Burns, J. T., 2016, “Effect of Chloride Concentration on the
M. G. Fahrmann, E. S. Huron, and S. A. Woodard, eds., TMS, Warrendale, Corrosion-Fatigue Crack Behavior of an Age-Hardenable Martensitic Stainless
PA, 2008, pp. 609–618. Steel,” Int. J Fatigue, 91, pp. 79–99.
[13] Goodrum, W., Chan, K. S., Enright, M. P., Burns, J. T., and Mills, D. M., 2018, [21] Burns, J. T., Larsen, J. M., and Gangloff, R. P., 2011, “Driving Forces for
“Development of Physics-based Modeling Tools for Life-prediction and Localized Corrosion to Fatigue Crack Transition in Al-Zn-Mg-Cu,” Fatigue
Durability Assessment of Advance Materials,” Phase II Final Report, Fract. Eng. Mater. Struct., 34(10), pp. 745–773.
NNX15CC33C, ERI, Charlottesville, VA. [22] Jamieson, A., 2018, “Hot Corrosion Effects on the Fatigue Behavior of a
[14] Southwest Research Institute, 2013, “DARWIN® User’s Guide”, San Antonio, Nickel-Based Superalloy ME3 (Rene 104),” MS thesis, University of Virginia,
TX. Charlottesville, VA.
[15] Chan, K. S., Enright, M. P., Moody, J. P. and Fitch, S. H. K., 2016, [23] Kitagawa, H., and Takahashi, S., 1976, “Applicability of Fracture Mechanics
“Physics-based Modeling Tools for Predicting Type II Hot Corrosion in to Very Small Cracks or the Cracks in the Early Stages,” Proceedings of the
Ni-Based Superalloys,” Superalloy 2016, M. Hardy, E. Huron, U. Glatzel, B. Second International Conference on Mechanical Behavior of Materials,
Griffin, B. Lewis, C. Rae, V. Seetharaman, and S. Tin, eds., Wiley, Hoboken, American Society for Metals, Metals Park, OH, Aug. 16–20, pp. 627–631.
NJ, pp. 917–925. [24] Gabb, T. P., Telesman, J., Kantzos, P. T., and O’Connor, K., 2002,
[16] Chan, K. S., Enright, M. P., Moody, J., Thomas, C., and Goodrum, W., 2019, Characterization of the Temperature Capabilities of Advanced Disk Alloy
“HOTPITS: The DARWIN Approach to Assessing Risk of Hot ME3,” NASA/TM-2002-211796, (NASA Glenn Research Center, Cleveland, OH.
Corrosion-Induced Fracture in Gas Turbine Components,” Eng. Fracture [25] Lindley, T. C., McIntyre, P., and Trant, P. J., 1982, “Fatigue Crack Initiation at
Mechanics, (Under Review). Corrosion Pits,” Metals Technology, 9(1), pp. 135–142.

Journal of Engineering Materials and Technology JULY 2020, Vol. 142 / 031001-11

You might also like