Aluminum components used in aerospace structures are commonly coupled with stainless steel fasteners. These through-hole geometries on the aluminum substrate cause a concentrated stress field. The high-stresses at the fastener sites can preferentially initiate coating damage allowing for moisture ingress, which can lead to the formation of a galvanic couple between the aluminum alloy and the stainless steel fastener. Corrosion damage is known to cause early initiation of fatigue cracks, thus severely reducing the total life of the component. This work aims to understand the relative impact and interaction of fastener hole geometry-induced stress concentrations and corrosion damage on the fatigue crack initiation behavior and total fatigue life. Specifically, by imparting various levels of corrosion severities at different locations within the macro-scale stress field, the relative impact of each on the fatigue process can be determined. This work demonstrated a dominant role of the macro-scale stress field on the ability of corrosion morphologies to initiate fatigue cracks. Specifically, crack formation was found to preferentially occur at high-stress regions in lieu of forming at lower-stress regions, regardless of corrosion severity, and corrosion severity in the through-hole had a significant, but nonpredictive, correlation with the total fatigue life. Critically, the findings of this work will inform the means by which coatings are evaluated and will serve as a controlled validation of experiments for fracture mechanics modeling.

7xxx-series aluminum alloys, such as AA7075-T6, are ubiquitous in structural aerospace applications due to their advantageous strength-to-weight ratios and acceptable fracture toughness values.1-3  In particular, AA7075 is used in geometrically complex components on aircrafts. These built-up structures are often joined with stainless steel fasteners that are wet-installed with sealant and then painted with protective coatings to mitigate corrosion damage.4  The circular geometry of the fastener sites acts as a stress concentrator, which locally increases the stresses at the fastener hole. These enhanced local stresses can lead to a preferential breakdown of the sealant and protective coatings, resulting in moisture ingress and the subsequent development of corrosion damage due to the formation of a galvanic couple between the aluminum substrate and the stainless steel fastener. The induced corrosion on the aluminum substrate is deleterious in that they further concentrate the stress on a local level and are hypothesized to locally enhance hydrogen uptake.5-8  These corrosion features, therefore, serve as preferred nucleation sites for fatigue cracks reducing the initiation life of fatigue to near-zero values.5  Both the extent of static (or fatigue) loading during ground-basing and the environment experienced during operation (e.g., active loading) will vary depending on the component and airframe operational details. However, modeling approaches will very often decouple the corrosion development and the active loading stages due to the relatively short time scales associated with active airframe operation (compared to ground-basing) and the cold-dry nature of a high-altitude operation.5,8-15  Regardless, corrosion-nucleated fatigue cracks can result in premature failure of structurally critical components at stresses well below the yield strength of the material.16 

The initial stages of fatigue (damage accumulation, crack initiation, and microstructurally small crack propagation) typically account for the majority of the fatigue life of a component experiencing high cycle fatigue.17  As such, the premature initiation of a crack caused by corrosion can significantly reduce the total fatigue life of the component.5,18  Managing this increased risk of component failure has in part been responsible for the increase in maintenance costs in the aging aircraft fleet.19  A recent teardown analysis found that 80% of fatigue cracks initiated in regions with corrosion damage.16  The high frequency of corrosion-nucleated fatigue cracks in aerospace structures justifies investigations to understand and develop mitigation and structural management approaches to account for the impact of corrosion on fatigue.

Fastener sites are of particular interest due to the high stresses at the fastener hole which act to increase the likelihood of coating failure and therefore facilitate the conditions necessary for the development of local corrosion. The stress concentrations from the geometry of the machined hole and the developed corrosion morphologies enhance the damage accumulation process at the beginning stages of fatigue. The use of coatings is the primary mitigation strategy used on airframe structures to preclude the deleterious effect of corrosion damage on structural integrity. Typically, such coating systems are designed not only to serve as a barrier to moisture ingress, but also to provide either chemical and/or galvanic inhibition to mitigate damage even when there is a defect in the coating.20  A “scribe test” is commonly used to evaluate the ability of a coating to provide protection when damaged.21-27  A scribe test consists of a sample with a coating that has been mechanically damaged (typically in an X-shape) to remove the coating in a precise area. This test is commonly used to investigate the ability to protect a fastener location, where there is a galvanic couple between a coated substrate and a dissimilar fastener.21-26  The analysis of such tests is typically limited to characterization of the damage on the coated surface, with little attention given to characterizing the conditions within the through-hole. While research has expanded upon the importance of investigating the through-hole corrosion,23,28  the relevance of corrosion morphologies within the through-hole to the subsequent fatigue behavior has not been evaluated.

Standard analysis of the surface corrosion damage via coarse metrics (e.g., area corroded, mean corrosion depth, etc.) provides important insights into the efficacy of the coating in mitigating surface corrosion. However, three aspects of this approach limit the applicability and relevance of such results to gain insights into how the coatings will impact the primary failure mechanism in most aerospace components, i.e., corrosion-nucleated fatigue. First, Al-alloys are well known to be susceptible to localized corrosion such as pitting, exfoliation, and intergranular corrosion.29  A detailed study of the evolution of corrosion morphologies at a galvanic couple demonstrated that a wide range of localized corrosion morphologies can be observed for galvanic couple conditions.30  Extensive literature has demonstrated that these localized corrosion features are critical to the fatigue crack initiation process.5,18,30-31  Second, coarse surface metrics fail to account for the potentially important role of damage location relative to macro-scale stress concentrators like through-holes. Previous research investigated the importance of the corrosion location relative to high- and low-stress concentration values of a notch under fatigue loads and showed that fatigue cracks did not form at corroded low-stress locations despite forming at equivalently corroded high-stress locations.32  Although this work demonstrated that both the geometry-induced, macro-scale stresses and local corrosion morphologies work synergistically to enhance the fatigue damage, it did not investigate various corrosion morphologies relevant to a galvanic couple, the impact of coatings, or a through-hole geometry. Third, coarse surface metrics will not capture the unique position-dependent corrosion behavior associated with typical galvanically coupled configurations (e.g., a fastener in a through hole). Specifically, prior work demonstrated that the net anodic charge associated with galvanic corrosion was 3.5-fold greater for a fastener geometry than for a flat geometric coupled multielectrode array exposure under the same conditions.28  Another study using a finite element model of a galvanically coupled stainless steel fastener and a AA7075 substrate found that the anodic charge density of the through-hole was largest at the opening of the through-hole, but was not as large as the anodic charge density of the scribes.33  In all, these coarse surface metrics commonly utilized in the analysis of scribe test corrosion are insufficient to characterize the relative impact of this corrosion damage, and thus the effectiveness of the coating of interest, on the corrosion nucleated fatigue behavior proximate to a galvanically coupled configuration.23,28 

The overarching goal of this work is to better understand the synergistic effect of local corrosion morphologies and the stress-concentrating effect of a fastener through-hole on the fatigue behavior. By investigating the damage location relative to high- and low-stress regions, this work attempts to elucidate the efficacy of current coating evaluation protocols. Specifically, the goal is to determine if analysis of surface corrosion is over-emphasized when considering coating performance in the context of its impact on fatigue crack initiation. The effect of surface corrosion vs. through-hole corrosion on the total fatigue lives of precorroded samples will be investigated by varying the corrosion severities at these respective locations and fatiguing the samples under constant amplitude loading. Analysis of the total fatigue life and the initiation location for each of the various combinations of corrosion severity and macro-scale stress concentration will provide insight into the relative impact of each on the fatigue crack initiation behavior. Such insights will inform improvements in the testing protocol for coating systems used in built-up airframe components and, more generally, will inform engineering scale prognosis approaches for predicting the remaining life of a corroded component.

The goal of this research is to evaluate the effect of pre-existing corrosion severity and corrosion location on the fatigue response of the samples by varying the corrosion severity present in the fastener hole and on the surface. The corrosion damage on different samples is quantitatively ranked for each damage location, the fatigue performance is measured by the total fatigue life of the samples, and the crack initiation locations are identified. Using the quantitative groupings, the fatigue lives of the samples in each group were correlated to (1) the corrosion severity of the fastener hole and (2) the corrosion severity of the surface/scribe to determine the relationship between corrosion severity and the total fatigue life at each location relative to a stress concentration. Another variable that is investigated is the orientation of the scribes with respect to the loading axis, thus varying the location of the corrosion on the surface relative to the gradient of stress proximate to the hole. This scribe orientation study was done to determine if the fatigue responses were noticeably affected by the surface condition when the surface corrosion was inline with the highest local stresses. The locations of the primary and secondary fatigue cracks relative to their macro-scale location (i.e., either the surface/scribe or the through-hole) will be determined through scanning electron microscopy (SEM) of the fracture surface. This will also indicate if the severity of the corrosion has an effect on initiation location.

Sample Preparation

Twelve samples that are 127 mm long in the longitudinal (L) direction, 25.4 mm wide in the radial (R) direction, and 6.35 mm thick were excised from a 25.4 mm AA7075-T6 bar stock. The ends of the samples were mill finished and a 4.85 mm diameter hole was drilled at the center of each sample. The holes are deburred and chamfered at each end of the through-holes. Before exposure to the aggressive environment, all samples were coated with a tri-chrome conversion coating pretreatment meeting MIL-DTL-5541F, followed by a chromate epoxy primer that meets MIL-PRF-88582D Type I, Class 2. A finish coat of white polyurethane (that meets MIL-PRF-85285 Type I, Class H) was applied on all sides (white background area in Figures 1[a] through [c]). Special care was taken to ensure that the polyurethane topcoat was not applied within the fastener hole. The L-R surfaces on each of the samples were either: left untouched as shown in Figure 1(a), scribed through the coatings to the base metal to create a 45° (relative to the loading axis) scribe as shown in Figure 1(b), or scribed to create a 90°-/0°- scribe combination as shown in Figure 1(c). All scribes had a width of 0.46 mm and a length of 25.4 mm and were induced using a CNC mill. M5x 0.8 316SS socket head fasteners were then installed with #10 SAE 316SS washers. The fasteners and washers had two coating configurations: not coated (shown in Figures 1[a] and [b]) or coated (shown in Figure 1[c]) with Blockade GC, an inorganic/organic composite sol-gel coating (developed by LUNA Innovations) that is comprised of various silanes. Blockade GC acted to electrically insulate the fastener and washer from the aluminum matrix and therefore limit a galvanic couple from developing between the aluminum matrix and the stainless steel fastener and washer. The Blockade GC (commonly referred to as sol-gel) was shown in previous work to severely reduce the level of galvanic corrosion both within the fastener hole and on the surface of the panel.23  The sample configurations are summarized in Table 1.

Table 1.

The Test Matrix Describing the Coating Conditions on the AA7075-T6 Samples and the 316SS Fasteners

The Test Matrix Describing the Coating Conditions on the AA7075-T6 Samples and the 316SS Fasteners
The Test Matrix Describing the Coating Conditions on the AA7075-T6 Samples and the 316SS Fasteners
FIGURE 1.

Images of the white-coated AA7075-T6 samples against a dark background demonstrating the various scribe orientations with the variously painted 316SS fasteners in place. (a) Sample 2 with no scribe and a bare fastener bolted in place, (b) sample 4 with a 45°-scribe and a bare fastener bolted in place, and (c) sample 10 with a 90°/0° scribe with a Blockade GC coated (shown by the blue coating) fastener bolted in place. These images were taken right before they were placed in the salt spray environment.

FIGURE 1.

Images of the white-coated AA7075-T6 samples against a dark background demonstrating the various scribe orientations with the variously painted 316SS fasteners in place. (a) Sample 2 with no scribe and a bare fastener bolted in place, (b) sample 4 with a 45°-scribe and a bare fastener bolted in place, and (c) sample 10 with a 90°/0° scribe with a Blockade GC coated (shown by the blue coating) fastener bolted in place. These images were taken right before they were placed in the salt spray environment.

Close modal

The samples were then exposed to a continuous salt spray of 0.9 M NaCl for 504 h per ASTM B117 at ambient laboratory temperatures.34  The samples were corroded with fasteners in place, with the fastener heads facing upward, and tilted at 15° perpendicularly in order to simulate galvanic corrosion in an aggressive environment. This angle ensures that a consistent water layer is present during the testing.35  The scribes allow for the sample surface to be exposed to the salt spray, therefore allowing for corrosion to develop as shown in Figure 2. After 504 h, the samples were removed from the testing environment, the fasteners were removed from the samples, and the samples were chemically depainted. To remove the corrosion products, the samples were sonicated in 70% nitric acid for 5 min and then immediately sonicated in deionized (DI) water for 15 min to remove the nitric acid and prevent damage to the aluminum.36-37  The samples were then sonicated in acetone and then methanol for 15 min.

FIGURE 2.

Images of the AA7075-T6 samples after corroding and after the coating was removed demonstrating the various scribe orientations. (a) Sample 2 with no scribe, (b) sample 4 with a 45° scribe, and (c) sample 10 with a 90°/0° scribe. The samples all have through-holes with the same machined diameter (4.85 mm).

FIGURE 2.

Images of the AA7075-T6 samples after corroding and after the coating was removed demonstrating the various scribe orientations. (a) Sample 2 with no scribe, (b) sample 4 with a 45° scribe, and (c) sample 10 with a 90°/0° scribe. The samples all have through-holes with the same machined diameter (4.85 mm).

Close modal

Fatigue Testing

The disassembled and cleaned samples were fatigue-loaded in tension (along the L-direction) under constant amplitude in a controlled moist air environment contained within an environmental cell. To avoid laboratory-based environmental fluctuations and ensure consistency between tests, the relative humidity (RH) within the cell was held to be >90% for the entire test by bubbling nitrogen gas through DI water at ambient lab temperatures. The samples were fatigued at a maximum stress (σMAX) equal to 100 MPa which is roughly 20% of the yield strength reported for AA7075-T6 alloys, a stress ratio (R) of 0.5, and a frequency of 20 Hz. A controlled load sequence was used to induce fracture surface markings (marker bands) to aid in the post-test analysis of the crack initiation and propagation behavior. The controlled load sequence was applied at a frequency of 10 Hz and was used to induce the marker bands consisting of 35 cycles at R = 0.1, 70 cycles at R = 0.5, then repeating this sequence four times to create a single marker band. By changing the r-ratio four times, the marker band consists of four, visually distinct, bands. The first marker band sequence was applied after 5,000 baseline cycles and subsequent marker sequences were applied after every 20,000 baseline cycles. The marker band load sequence resulted in a thin, identifiable fracture surface feature that traces the presence of the crack front. This approach has been previously applied to induce marks at known increments of cycles in order to pinpoint the initiation location of a crack, calculate the crack initiation life, quantify the microstructurally small crack growth rates, and track the crack path of multiple cracks. This approach has been shown to have minimal impact on the overall fatigue behavior.5-7,38-42 

Characterization

Optical micrographs were taken using a Hirox RH-8800 light microscope prior to fatigue loading to document the corrosion damage on the sample surface. After failure, fractography was performed on the fatigue surfaces to determine the initiation location, initiating feature, and image the through-hole corrosion damage. The marker bands were traced to determine where the cracks initiated and to discern which initiating feature is the first and primary crack initiation location, as shown in Figure 3. The through-hole corrosion damage state was further characterized via fractographs taken with an SEM of the border between the corroded and pristine material. Imaging was performed orthonormal to the fatigue surface using an Everhart-Thornley detector and a concentric backscatter (CBS) detector at an accelerating voltage of 10 kV and a spot size of 4. White light interferometry (WLI) was used to characterize the scribe corrosion conditions using the Zygo New View 7300 optical profiler. The scans were taken using 2.5× lens and multiple field of view measurements were stitched together to achieve the final topographic map of the surfaces.

FIGURE 3.

The Everhart-Thornley Detector SEM micrographs of the fracture surface (a) showing the marker bands before analysis (three are pointed out with white arrows) and (b) with demarcations over the marker bands that help in fractography efforts to determine the initiation location. The blue lines represent the super marker bands, the yellow lines represent the marker bands, and the red and yellow “X”s represent the fatigue initiation locations.

FIGURE 3.

The Everhart-Thornley Detector SEM micrographs of the fracture surface (a) showing the marker bands before analysis (three are pointed out with white arrows) and (b) with demarcations over the marker bands that help in fractography efforts to determine the initiation location. The blue lines represent the super marker bands, the yellow lines represent the marker bands, and the red and yellow “X”s represent the fatigue initiation locations.

Close modal

Corrosion Quantification

The goal of the corrosion characterization for this study is to rank the corrosion severity of various morphologies. Prior work has demonstrated that there is no single corrosion metric that governs the crack formation behavior;5,42  rather, initiation is governed by a complex interaction of the macroscale features, microscale features, and the underlying microstructure.41,43-46  As such, a robust array of metrics are used to capture the possible contributions of various features of the corrosion damage that literature has suggested could potentially impact the fatigue initiation process.5,18  The corrosion categories of interest are: (1) the maximum corrosion feature depth, (2) the mean corrosion depth, (3) the total corroded area, (4) the total extent of corrosion, and (5) a description of the distribution of large features. Of note, these categories are robust in the dimensionalities they characterize and are justified by the literature but are not sufficiently comprehensive to enable quantitative prediction of fatigue life. Metrics that provide an indication of the severity of each of these categories are quantified for the corrosion on the surface/scribe and in the through-hole for each sample. These metrics are used to rank the severity of the corrosion of each of the samples in the context of each of the corrosion categories. These individual ranks are then combined (via a process described below) to give an overall ranking of the severity of the corrosion of each sample, both on the surface and in the through-hole.

The corrosion metrics used for the rankings were established via either WLI or surface imaging of the samples. Due to the geometry of the samples, the corrosion features were quantified via analogous, but, in some instances, slightly different metrics depending on whether the corrosion was (1) on the surface (i.e., the coated area or the scribed area in Figures 1[a] through [c]), or (2) with-in the through-hole. In general, WLI is used to characterize the surface corrosion and cross-sectional image analysis is used to generate the corrosion metrics for the through-hole; both methods are used extensively in the literature to characterize corrosion morphologies and corrosion distributions.5,18,23,26,28,47-49  While the metrics obtained from these different techniques are slightly different, they both enable a quantitative description of the five broad categories of corrosion descriptors, thus enabling comparative rankings for each of the samples.

Surface/Scribe Corrosion Quantification

The WLI data collected on the corroded surface/scribe area were postprocessed using the Digital Surf application MountainsMap Universal, version 7.4. Figure 4 shows a field of view WLI scan of a scribe. The corrosion pit depth distribution data for the surface/scribe area excludes any pit surface areas and volumes with a radius less than 5 μm. A pit was determined through a watershed segmentation method defined in the ISO 25178-2 standard.50  Equations (1) through (5) were used to calculate quantitative metrics (from the WLI data) that describe the five corrosion description categories listed above. The surface and scribe areas were analyzed separately to get the corrosion depth data due to the scribe removing base metal from the sample before corrosion and therefore having a different uncorroded depth than the surface. While the surface was coated, undercutting of the coating during corrosion is common. The maximum corrosion depth (Equation [1]) is the largest depth of corrosion that occurred on either the surface/scribe area. The mean corrosion depth (Equation [2]) is the average of the surface corrosion depths. The total corroded area (Equation [3]) is the area (on the L-R surface) between the initial pristine surface and the border of the corrosion morphologies on the surface/scribe. Equation (4) is the total corroded volume on the surface/scribe. The mean of the max for the 20% largest individual pits, shown in Equation (5), is the average of the maximum measurements for the largest 20% corrosion pits. The equations are:
formula
formula
formula
formula
formula
where n is the number of corrosion depths measured, m is the number of corrosion pits defined through the watershed thresholding method outlined in ISO 25178-2, X is the measured corrosion depth dataset for a sample, A is the area of a corroded pit, V is the total corroded volume, X(20% Largest) is a subset consisting of the largest 20% of the corrosion pits, i is the index of summation which cycles through each corrosion pit depth distribution data point, j is the index of summation which cycles through each corrosion pits as defined through the watershed method outlined in ISO 25178-2, and n20% Largest is the number of corrosion pits that are the largest 20% of corrosion pits in the field of view WLI scan.
FIGURE 4.

Field of view WLI scan of the scribe corrosion on sample 3. Thresholding the depth used in the analysis and cropping the data to areas of interest were utilized to analyze the scribe corrosion and surface corrosion separately and avoid quantifying any obvious scratches on the surface.

FIGURE 4.

Field of view WLI scan of the scribe corrosion on sample 3. Thresholding the depth used in the analysis and cropping the data to areas of interest were utilized to analyze the scribe corrosion and surface corrosion separately and avoid quantifying any obvious scratches on the surface.

Close modal

Through-Hole Corrosion Quantification

The through-hole corrosion damage was quantified using micrographs taken with a CBS detector in an SEM after failure. For each sample, a border between the corroded and noncorroded area was traced on the corresponding micrographs, as shown by Figure 5. The border was then used as input into a MATLAB program developed to analyze the corrosion damage of the cross-sectional micrograph.49  Using this program and an image editing software, quantitative metrics were developed to describe the five corrosion description categories listed above. The through-hole corrosion metrics are maximum corrosion depth shown by Equation (1), mean corrosion depth shown by Equation (2), total area corroded shown be Equation (3), mean large corrosion depth shown by Equation (6), and percent surface length corroded shown by Equation (7). The maximum corrosion depth (Equation [1]) is the largest depth of the through-hole corrosion and the mean corrosion depth (Equation [2]) is the average of all of the corrosion depths measured along the through-hole (sampled every 0.4 μm). The total corroded area (Equation [3]) is the area between the initial pristine surface and the border of the corroded surface. The mean large corrosion depth, shown in Equation (6), was calculated by averaging only the corrosion depths greater than one standard deviation from the mean corrosion depth. The percent corrosion length, Equation (7), was calculated by measuring the corroded length across the through-hole and dividing that by the fastener hole length excluding the chamfered edges. The equations unique to the through-hole analysis are shown below:
formula
formula
where is the number of corrosion depths that are one standard deviation greater than or equal to the average corrosion depth, Lcorr is the total corroded length along the through-hole, and LTotal is the total length along the through-hole.
FIGURE 5.

(a) Merged CBS-SEM micrograph of the fracture surface of sample 4 visualizing the corrosion cross section on the plane of failure. (b) The border of the corroded and uncorroded area traced from the above micrograph (the border was filled with a gray color to aid in visualization). Large sections of the border with no corrosion were cropped out while still ensuring an uncorroded trace of the border was kept for measuring the corrosion depth.

FIGURE 5.

(a) Merged CBS-SEM micrograph of the fracture surface of sample 4 visualizing the corrosion cross section on the plane of failure. (b) The border of the corroded and uncorroded area traced from the above micrograph (the border was filled with a gray color to aid in visualization). Large sections of the border with no corrosion were cropped out while still ensuring an uncorroded trace of the border was kept for measuring the corrosion depth.

Close modal

Through-Hole and Surface/Scribe Metric Comparisons

The through-hole and the surface/scribe corrosion metrics were chosen to emulate each other and characterize similar characteristics of the corrosion. Certain corrosion metrics are not exactly equivalent, but the metrics act to quantify a similar quality of the corrosion. The first two metrics calculated for each location (the maximum and mean corrosion depth) are identical in their mathematical form. However, the mean large corrosion metric for the through-hole and the average of the maximum corrosion depth from the largest 20% of the corrosion pits metric both act to average the largest corrosion pits depths albeit in different manners. The percent corrosion length and the total corroded area metrics for the through-hole emulate the total corroded area and the total volume corroded metrics for the surface/scribe. These metrics quantify the dimensions of the corrosion and vary due to the dimensionality captured by the characterization methods used (e.g., WLI is a three-dimensional characterization technique and the SEM images quantify two dimensions of the corrosion). The goal of this corrosion characterization is not to try to correlate different corrosion features to the crack formation location or life, rather to characterize a wide range of potentially relevant corrosion metrics for both the surface/scribe and through-hole to enable a rank-order of the level of severity. As such, the lack of direct correspondence between the metrics at each location will not compromise the analysis.

Corrosion Severity Ranking

To compare the corrosion severity to the fatigue life, the through-hole corrosion severity and the surface/scribe corrosion severity were separately ordinally ranked based on a value termed the Corrosion Sum Rank. For each of the five corrosion description categories, the metrics were used to sort the samples from largest to smallest. After sorting, each metric value was then normalized by the maximum value of the metric; this served to bring each of the five metrics to the same scale, coarsely capturing the extent of variation between samples and making the metrics unitless. The five normalized metrics were then summed together for each sample. The resulting value is the Corrosion Rank Sum and is the value used to provide the final overall corrosion severity ranking. The equation for the surface/scribe Corrosion Rank Sum metric is shown in Equation (8) and the equation for the through-hole Corrosion Rank Sum metric is shown in Equation (9). The Corrosion Rank Sum metric is on the scale of 5 to 0 where 5 is the most severe corrosion present within the samples tested and 0 represents no corrosion.
formula
formula
where b represents the sample number and B represents the entire sample set.

Corrosion Quantification

Figure 6 illustrates the distribution of the corrosion depths for both the surface/scribe and through-hole corrosion using cumulative frequency plots. Figure 6(a) shows the cumulative percentage of the max corrosion pit depths on the surface/scribe corrosion for each sample (except samples 1 and 2 as they did not have scribes nor surface corrosion). Figure 6(b) shows the cumulative percentage of the through-hole corrosion depths for each sample; specifically, these plots represent the depths at all corroded locations along the through-hole sampled every 0.4 μm. The cumulative frequency plots show the cumulative percentage of the corrosion depths with respect to the depths and allow for the comparison of the corrosion depths between each sample visually. Relevant features of these plots are: the percentage of depths for each sample at (1) 25%, (2) 50%, (3) 75%, and (4) the shape of the curves as the percentages increase. By analyzing the data at 25%, 50%, and 75%, the skewness of the data can be ascertained. The data in question has a mostly right skew. A right skew means that the large data points are so large that they offset, or skew, the distribution toward the right, making the distribution asymmetrical. This quality of the data can be seen by the cumulative distribution not being symmetric at 50% and the curves increasing more in the X axis above the 50% than below the 50%. Of note is that every sample at each location shows a unimodal distribution of the depths except the through-hole corrosion pit distribution for sample 4 which shows a bi-modal distribution. In all, the box plots allow for the corrosion depths in the through-hole and the max corrosion pit depths in the surface/scribe to be visually compared to one another.

FIGURE 6.

(a) Cumulative frequency plots of the surface/scribe max corrosion pit depths for each AA7075-T6 sample with corrosion damage on the surface. (b) Cumulative frequency plots of the through-hole corrosion depths. Both figures show the cumulative percentage distribution, where each symbol (square) represents a data point and the lines connect the points together and have an inset graph that shows a zoomed-in graph of the data between 0 and 100 depths.

FIGURE 6.

(a) Cumulative frequency plots of the surface/scribe max corrosion pit depths for each AA7075-T6 sample with corrosion damage on the surface. (b) Cumulative frequency plots of the through-hole corrosion depths. Both figures show the cumulative percentage distribution, where each symbol (square) represents a data point and the lines connect the points together and have an inset graph that shows a zoomed-in graph of the data between 0 and 100 depths.

Close modal

Table 2 shows the individual metrics for each sample at each location. The metrics show that while there is some variation between the corrosion ranking of the samples, there is a consistent trend of the corrosion severity in the samples. The overall corrosion severity rankings based on the Corrosion Rank Sum metric are shown in Table 3. In this paradigm, the closer the Corrosion Rank Sum is to 5, the higher the corrosion severity. The Corrosion Rank Sum value associated with the surface/scribe area for Samples 1 and 2 are 0 because there were no scribes and no surface corrosion. This absence of corrosion was confirmed via imaging the samples after they were exposed to the ASTM B117 salt spray corrosion protocol. Critically, Table 3 shows that the samples have varying corrosion severities between locations (e.g., surface/scribe vs. through-hole), which enables us to independently interrogate the relative impact of corrosion severity at different macro-scale locations. Using the Corrosion Rank Sum metric (CRS), the samples can be sorted into corrosion severity bins: severe (5≤CRS < 1), moderate (1≤CRS < 0.3), and minimal-to-no corrosion (0.3≤CRS < 0). Note that Samples 1, 2, 5, and 7 are categorized in different bins for the surface/scribe vs. the through-hole corrosion severities; all other samples remain in the same bins for both locations. Furthermore, each of these samples (1, 2, 5, and 7) increases to a higher corrosion bin when considering damage in the through-hole as opposed to the surface of the panel.

Table 2.

The Through-Hole and Surface/Scribe Corrosion Metrics for Each Sample as Calculated by Equations (1) Through (7).

The Through-Hole and Surface/Scribe Corrosion Metrics for Each Sample as Calculated by Equations (1) Through (7).
The Through-Hole and Surface/Scribe Corrosion Metrics for Each Sample as Calculated by Equations (1) Through (7).
Table 3.

The Surface/Scribe and the Through-Hole Corrosion Severity Ranking for the Test Matrix Sorted from Largest to Smallest Corrosion Rank Sum with the Associated Total Fatigue Lives for Each Sample

The Surface/Scribe and the Through-Hole Corrosion Severity Ranking for the Test Matrix Sorted from Largest to Smallest Corrosion Rank Sum with the Associated Total Fatigue Lives for Each Sample
The Surface/Scribe and the Through-Hole Corrosion Severity Ranking for the Test Matrix Sorted from Largest to Smallest Corrosion Rank Sum with the Associated Total Fatigue Lives for Each Sample

Fatigue Behavior

The total fatigue lives of all samples are reported in Table 3. The micrographs revealed that for all samples that the fatigue cracks initiated within the through-hole and at corrosion features no visible fatigue cracks initiated from corrosion on the surface or in the scribe. Figure 7 plots the fatigue lives sorted based on the bins of corrosion severity for the surface/scribe (Figure 7[a]) and the through-hole (Figure 7[b]) locations. The samples with both severe through-hole and surface/scribe corrosion damage samples (Samples 3, 4, and 8) exhibit low fatigue lives. The fatigue lives are within an order of magnitude of those of Samples 5 and 7 that have (1) moderate corrosion on the surface/scribe and (2) severe corrosion in the through-hole and Samples 1 and 2 that have (1) the minimal-to-no corrosion surface/scribe and (2) moderate through-hole corrosion damage. Samples 9 and 10, which have minimal-to-no surface/scribe and through-hole corrosion damage, have significantly higher fatigue lives than the other groups. Sample 6 is in the moderate corrosion bin for both the surface/scribe and the through-hole.

FIGURE 7.

(a) Surface/scribe and (b) through-hole corrosion severity binned and plotted by total fatigue life for the AA7075-T6 samples.

FIGURE 7.

(a) Surface/scribe and (b) through-hole corrosion severity binned and plotted by total fatigue life for the AA7075-T6 samples.

Close modal

It is useful to compare the fatigue lives of Samples 9 and 10 that have minimal-to-no surface/scribe and through-hole corrosion conditions with Samples 1 and 2 that have minimal-to-no surface/scribe corrosion and moderate through-hole corrosion severities. These two sets of samples have drastically different fatigue lives despite having similar surface corrosion severities. Furthermore, samples with any amount of through-hole corrosion damage greater than 0.3 have similar fatigue lives exhibit similar fatigue lives. The greatest distinction in average fatigue lives occurs between the severe scribe and through-hole damage group and the minimal-to-no through-hole and surface/scribe damage group with the greatest difference between the maximum fatigue lives, minimum fatigue lives, and average fatigue lives of the respective groups.

Samples 5, 6, 9, and 10 were all galvanically corroded with a fastener coated with Blockade GC. Table 3 shows that for these samples, the corrosion condition of the through-hole is much less severe than the samples exposed to the aggressive environment with noncoated fasteners. Samples 9 and 10 both had 90°/0° scribes and significantly longer fatigue lives than the other samples. Figures 6(a) and (b) show that the depths of the corrosion features are also, on average, smaller than the corrosion depths of the other samples.

Fatigue

Figure 7 demonstrates that the through-hole corrosion severity better correlates to the total fatigue life than the surface/scribe corrosion bins. This can be broadly considered by calculating the average fatigue life of samples grouped in the minimal-to-no, moderate, and severe corrosion severity bins. The average fatigue life decreases as the through-hole corrosion severity increases, to a lesser extent this is also true for the surface/scribe corrosion bins. However, there is a high degree of variability observed for the minimal-to-no corrosion condition in the surface/scribe case (Figure 7[a]); specifically, despite having the same surface/scribe corrosion morphologies sample 1 exhibits the lowest life and Samples 9 and 10 exhibit the longest lives. This difference is most likely due to the difference in the through-hole corrosion severity between Samples 1 and Samples 9 and 10. The relationship between through-hole corrosion severity and fatigue life is further illustrated by ranking the corrosion severity of each of the Samples 1 to 10 (where 1 represents the sample with the most severe corrosion and 9 or 10 represents the least severe corrosion) based on their CRS values (Table 3) for the surface/scribe (Figure 8[a]) and the through-hole (Figure 8[b]). These rankings are then plotted versus the total fatigue life of the samples. Of note, this ordinal ranking was done in lieu of directly plotting the CRS value to preclude scaling issues seen with other corrosion metrics when plotted against fatigue life.5,18  A simple regression analysis is performed and the resulting trendline and R2 value are reported on the plots. Critically, the through-hole corrosion severity ranking exhibits a relatively strong correlation (R2 = 0.65) with the total fatigue life (Figure 8[b]), whereas there is little to no correlation between the surface/scribe corrosion ranking and the total fatigue live (R2 = 0.13, as shown by Figure 8[a]). This is consistent with prior efforts that demonstrated that locations at a relatively low-stress concentration (KT = 0.8) did not cause fatigue crack initiation in AA7075-T651, but corrosion of the same severity did cause early failure at high-stress locations, specifically where KT = 2.1 and 3.1.32  This research further demonstrates and supports the idea that the location of corrosion relative to the stress gradient affects the propensity of corrosion to cause early initiation of fatigue.

FIGURE 8.

Graphs of the (a) surface/scribe corrosion and (b) through-hole corrosion rank vs. total fatigue life with the sample ID next to each symbol. Simple trendlines are seen on both graphs and were used to calculate the coefficient of determination. Simple trendlines are seen on both graphs and were used to calculate the coefficient of determination. The corrosion rank metric is a ranking of the samples by their corrosion severity (as calculated by the CRS metric) where 1 represents the sample with the most severe corrosion and 9 or 10 represents the least severe corrosion.

FIGURE 8.

Graphs of the (a) surface/scribe corrosion and (b) through-hole corrosion rank vs. total fatigue life with the sample ID next to each symbol. Simple trendlines are seen on both graphs and were used to calculate the coefficient of determination. Simple trendlines are seen on both graphs and were used to calculate the coefficient of determination. The corrosion rank metric is a ranking of the samples by their corrosion severity (as calculated by the CRS metric) where 1 represents the sample with the most severe corrosion and 9 or 10 represents the least severe corrosion.

Close modal

The total fatigue lives of samples with through-hole CRS greater than 0.3 are comparable to each other. This is shown in Figure 7(b) where Samples 1 through 8 all have fatigue lives within an order of magnitude of the average fatigue life of the moderate and severe corrosion severity bins. These results align with previous research which found that corroded samples have a sharp reduction in fatigue life that levels off with increasing corrosion depth.5  Previous research has shown that corrosion reduces the fatigue life by reducing the initiation life to near-zero values.5,18,30,41  The near-zero initiation life for corroded samples is due to the enhancement of the local stress state proximate to the corrosion features, which interacts with local microstructure features and, possibly, a locally H-charged region adjacent to the corrosion damage to cause crack initiation.18  None of the fatigue cracks that occurred on the tortuous fracture plane initiated at surface/scribe corrosion morphologies. All of the fatigue initiation sites visible through SEM observation occurred within the through-hole at corrosion morphologies of varying length scales. This suggests that the corrosion morphologies in the through-hole are reducing the total fatigue life through the early initiation of fatigue cracks. The fatigue cracks initiated on corrosion morphologies within the through-hole even when the samples had 90°/0° scribes that line up the corroded scribe with the highest stress concentrating plane on the surface of the sample. It should be noted that the samples had chamfers that created a perturbation in the stress gradient away from the hole surface. Regardless, none of the fatigue crack initiation locations occurred on the chamfers. This observation held true even when corrosion occurred on the chamfers as shown in Figure 1 for sample 4. In all, the fractography demonstrates that the component geometry-induced stresses have a strong effect on the initiation location and that corrosion in the through-hole has a higher propensity to initiate cracks than surface corrosion.

Although the CRS metric is used to bin the corrosion into severity groups, the metric cannot be used to predict the fatigue life accurately. While several macro-scale corrosion metrics have been proposed and have had success in engineering-scale predictions of the fatigue life,11,18,51-55  more detailed studies have demonstrated that such metrics fail to rigorously correlate with the fatigue crack initiation location.5,30,56  Rather, the crack initiation location and life are postulated to be dependent on a complex interaction of the macro-scale stress concentration, the local microstructure, and the micro-scale corrosion geometry.5,18,30  Recent work has shown that fatigue initiation locations can be predicted using fatigue indicator parameters, but these parameters require extensive modeling and experimental data to predict experimental corrosion initiation locations.41  As such, the variability in the CRS correlation is fully expected due to the coarse nature of the corrosion characterization, the failure to account for microstructural features, and variations in the global stress field.

Impact on Coating Evaluation Protocols

The findings of this research have strong implications for the current common practice coating evaluation protocol when those results are to be applied to built-up structure, particularly for cases with galvanic interactions between the fastener and the substrate. Coating performance is often judged by the severity of corrosion on the surface and the scribe,21-22,25  but in-service fatigue cracks that led to accidents were found to initiate at higher rates within fastener holes and at stress concentrating features than at other locations on the mainframe of the aircraft.57-59  Futhermore, recent research has shown that the cervice of a rivet/fastener may complicate the ability of chromate in protective coatings applied to the surface to inhibit corrosion developement.60  Fractography of the specimens in this research showed that all of the fatigue cracks visible through SEM inspection of the fatigued surface failed within the fastener hole at corrosion damage. This result suggests that corrosion damage at the surface does not control the early fatigue crack initiation when a hole (or another strong stress concentrator) is present. Furthermore, the corrosion quantification and ranking scheme of the surface/scribe corrosion does not correlate well to total fatigue life, but the through-hole corrosion quantification and ranking scheme generates a meaningful, albeit nonpredictive, correlation to fatigue life. This result shows that the corrosion condition on the surface is not sufficient to cause meaningful early initiation of fatigue cracks due to the relatively low macro-scale stress condition at the surface. This observation is consistent with prior research that found that the corrosion location relative to the macro-scale stress profiles has a strong effect on the initiation location and the total fatigue life of a corroded specimen.32  Small amounts of corrosion at stress concentrating features will result in drastic decreases in fatigue life.11,32  This effect is independent of the surface corrosion severity and the orientation of the scribes relative to the loading axis. As such, in order to properly evaluate the ability to mitigate corrosion damage that will deleteriously impact the fatigue behavior it is critical to understand how the coating impacts the corrosion damage in the through-hole.

The results of this study and prior work11,32  show that corrosion in the fastener hole is very detrimental to the fatigue life, more so than surface corrosion. This suggests an enhanced effort to ensure that coatings mitigate through-hole corrosion in addition to surface corrosion. This specific recommendation is constrained in relevance to a bolt hole under tension, however, this work supports the general statement that higher priorities should be put on protecting high-stress regions as compared to lower stress regions. Companion work is being performed to validate these assumptions for and develop LEFM-based life predictions methodologies for locations that are not the predetermined fatigue “hot-spots” identified for structural airframe components.61  This work will integrate the impact of corrosion and various geometry locations (e.g., stress concentrated regions), but is outside the scope of this paper.

The effect of macro-scale corrosion severity and location (surface vs. through-hold corrosion) proximate to a hole on the total fatigue life analyzed through constant amplitude fatigue tests and corrosion severity quantification. The following conclusions were drawn from these analyses.

  • All fatigue cracks initiated in the through-hole regardless of the disparity of the moderately corroded surface condition to the almost pristine through-hole condition (i.e., formed in the through-hole even if the surface/scribe had severe corrosion and the through-hole had minimal-to-no corrosion). Even the most severe surface corrosion condition did not initiate fatigue cracks.

  • The through-hole damage condition negatively affects the total life with increasing damage, but the sensitivity of the total fatigue life to the corrosion damage decreases with increasing damage.

  • The scribe orientation has no discernable effect on the total fatigue life nor the fatigue crack formation location.

Trade name.

We gratefully acknowledge the support from the Office of Naval Research under the contract ONR: N00014-14-1-0012 with Mr. William Nickerson as the Scientific Officer.

1.
Peters
M.
,
Leyens
C.
,
Mater. Sci. Eng.
3
(
2009
):
p
.
1
11
.
2.
Campbell
F.C.
, “
Chapter 2: Aluminum
,”
in Manufacturing Technology for Aerospace Structural Materials
(
Amsterdam, Netherlands
:
Elsevier Science
,
2006
).
3.
Starke
E.A.
,
Staley
J.T.
,
Prog. Aerosp. Sci.
32
(
1996
):
p
.
131
172
.
4.
Calle
L.M.
, “
Corrosion Control in Space Launch Vehicles
,”
in
Corrosion Control in the Aerospace Industry
(
Amsterdam, Netherlands
:
Elsevier Ltd
,
2009
),
p
.
195
224
.
5.
Co
N.E.C.
,
Burns
J.T.
,
Int. J. Fatigue
103
(
2017
):
p
.
234
247
.
6.
Burns
J.T.
,
Larsen
J.M.
,
Gangloff
R.P.
,
Int. J. Fatigue
42
(
2012
):
p
.
104
121
.
7.
Burns
J.T.
,
Larsen
J.M.
,
Gangloff
R.P.
,
Fatigue Fract. Eng. Mater. Struct.
34
(
2011
):
p
.
745
773
.
8.
Koch
G.
,
Hagerdorn
E.L.
,
Berens
A.P.
, “
Effect of Preexisting Corrosion on Fatigue Cracking of Aluminum Alloys 2024-T3 and 7075-T6
,” CC Technologies Laboratories Inc, Final report,
1995
, http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA430616.
9.
Sankaran
K.K.
,
Perez
R.
,
Jata
K.V.
,
Mater. Sci. Eng. A
297
(
2001
):
p
.
223
229
.
10.
Molent
L.
,
Eng. Fract. Mech.
137
(
2015
):
p
.
12
25
.
11.
McMurtrey
M.D.
,
Bae
D.
,
Burns
J.T.
,
Fatigue Fract. Eng. Mater. Struct.
40
(
2017
):
p
.
605
622
.
12.
Duquesnay
D.L.
,
Underhill
P.R.
,
Britt
H.J.
,
Int. J. Fatigue
25
(
2003
):
p
.
371
377
.
13.
SCHÜTZ
W.
, “
Corrosion Fatigue of Structural Components
,” RTO Applied Vehicle Technology Panel Symposium (Corfu 1998-10-07) (
Neuilly-sur-Seine, France
:
Research and Technology Organization
,
1999
).
14.
Hoffman
M.E.
,
Hoffman
P.C.
,
Int. J. Fatigue
23
(
2001
):
p
.
1
10
.
15.
Burns
J.T.
,
Bush
R.W.
,
Ai
J.H.
,
Jones
J.L.
,
Lee
Y.
,
Gangloff
R.P.
,
Eng. Fract. Mech.
137
(
2015
):
p
.
34
55
.
16.
Shoales
G.A.
,
Fawaz
S.A.
,
Walters
M.R.
, “
Compilation Of Damage Findings From Multiple Recent Teardown Analysis Programs
,”
in
ICAF 2009, Bridging the Gap between Theory Operational Practice
(
Dordrecht, Netherlands
:
Springer
,
2009
),
p
.
187
202
.
17.
Ritchie
R.O.
,
Lankford
J.
,
Mater. Sci. Eng.
84
(
1986
):
p
.
11
16
.
18.
Burns
J.T.
,
Kim
S.
,
Gangloff
R.P.
,
Corros. Sci.
52
(
2010
):
p.
498
508
.
19.
Kiley
G.T.
, “
The Effects of Aging on the Costs of Operating and Maintaining Military Equipment
,” Congressional Budget Office (US CONGRESS),
2001
, https://apps.dtic.mil/sti/pdfs/ADA393973.pdf.
20.
Shaw
B.A.
,
Kelly
R.G.
,
Electrochem. Soc. Interface
15
(
2006
):
p
.
24
26
.
21.
Matzdorf
C.A.
,
Nickerson
W.C.
,
Rincon Troconis
B.C.
,
Frankel
G.S.
,
Li
L.
,
Buchheit
R.
,
Corrosion
69
(
2013
):
p
.
1240
1246
.
22.
Wang
S.S.
,
Boerstler
J.T.
,
Frankel
G.S.
,
Corrosion
74
(
2018
):
p
.
499
508
.
23.
Marshall
R.S.
,
Kelly
R.G.
,
Goff
A.
,
Sprinkle
C.
,
Corrosion
75
(
2019
):
p
.
1461
1473
.
24.
Feng
Z.
,
Frankel
G.S.
,
Corrosion
70
(
2014
):
p
.
95
106
.
25.
Feng
Z.
,
Frankel
G.S.
,
Matzdorf
C.A.
,
J. Electrochem. Soc.
161
(
2014
):
p
.
C42
C49
.
26.
Boerstler
J.T.
, “
Corrosion Degradation of Coated Aluminum Alloy Systems through Galvanic Interactions
” (
Ph.D. diss.
,
The Ohio State University
,
2018
).
27.
Hunter
C.N.
,
Osborne
J.H.
,
Taylor
S.R.
,
Corrosion
56
(
2000
):
p
.
1059
1070
.
28.
Rafla
V.N.
,
Scully
J.R.
,
Corrosion
75
(
2019
):
p
.
587
603
.
29.
Andreatta
F.
,
Terryn
H.
,
de Wit
J.H.W.
,
Electrochim. Acta
49
(
2004
):
p
.
2851
2862
.
30.
Co
N.E.C.
,
Burns
J.T.
,
Corrosion
72
(
2016
):
p
.
1215
1219
.
31.
Larrosa
N.O.
,
Akid
R.
,
Ainsworth
R.A.
,
Int. Mater. Rev.
63
(
2018
):
p
.
283
308
.
32.
Brown
N.
,
Clark
B.
,
Kramer
P.
,
Co
N.E.C.
,
Friedersdorf
F.
,
Burns
J.T.
,
Corros. Rev.
37
(
2019
):
p
.
507
520
.
33.
Marshall
R.S.
,
Goff
A.
,
Sprinkle
C.
,
Britos
A.
,
Kelly
R.G.
,
Corrosion
76
(
2020
):
p
.
476
484
.
34.
ASTM B117-19
, “
Standard Practice for Operating Salt Spray (FOG) Apparatus
” (
West Conshohocken, PA
:
ASTM International
,
2003
).
35.
Katona
R.M.
,
Tokuda
S.
,
Perry
J.
,
Kelly
R.G.
,
Corros. Sci.
175
(
2020
):
p
.
108849
.
36.
ASTM G1-03
, “
Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens
” (
West Conshohocken, PA
:
ASTM International
,
1985
).
37.
Ferrer
K.S.
,
Kelly
R.G.
,
Corrosion
57
(
2001
):
p
.
110
117
.
38.
Spear
A.D.
,
Fai
S.
,
Lind
J.F.
,
Suter
R.M.
,
Ingraffea
A.R.
,
Acta Mater.
76
(
2014
):
p
.
413
424
.
39.
Barter
S.A.
,
Molent
L.
,
Goldsmith
N.
,
Jones
R.
,
Eng. Fail. Anal.
12
(
2005
):
p
.
99
128
.
40.
Barter
S.A.
,
Molent
L.
,
Wanhill
R.J.H.
, “
Marker Loads For Quantitative Fractography of Fatigue Cracks in Aerospace Alloys
,”
in
Bridging the Gap Between Theory Operational Practice
(
Dordrecht, Netherlands
:
Springer
,
2009
),
p
.
15
54
.
41.
Nicolas
A.
,
Co
N.E.C.
,
Burns
J.T.
,
Sangid
M.D.
,
Eng. Fract. Mech.
220
(
2019
):
p
.
106661
.
42.
Co
N.E.C.
,
Burns
J.T.
,
Int. J. Fatigue
150
(
2021
):
p
.
106301
.
43.
Weiland
H.
,
Nardiello
J.
,
Zaefferer
S.
,
Cheong
S.
,
Papazian
J.
,
Raabe
D.
,
Eng. Fract. Mech
. (
2009
):
p.
709
714
.
44.
McDowell
D.L.
,
Dunne
F.P.E.
,
Int. J. Fatigue
32
(
2010
):
p
.
1521
1542
.
45.
Castelluccio
G.M.
,
McDowell
D.L.
,
Mater. Sci. Eng. A
(
2014
):
p.
34
55
.
46.
Chan
K.S.
,
Int. J. Fatigue
32
(
2010
):
p
.
1428
1447
.
47.
Moran
A.
,
Jennings
J.
,
Nee
H.
,
Pearson
S.
,
Clark
B.
,
Lillard
R.S.
,
Corrosion
75
(
2019
):
p
.
484
498
.
48.
Young
P.S.
,
Payer
J.H.
,
Corrosion
71
(
2015
):
p
.
1278
1293
.
49.
Tattersall
C.
, “
Modeling and Analysis of Geometrically Complex Corrosion
” (
Ph.D. diss.
,
University of Virginia
,
2018
).
50.
ISO 25178-2:2012
, “
Geometrical Product Specifications (GPS)—Surface Texture: Areal—Part 2: Terms, Definitions and Surface Texture Parameters
” (
Geneva, Switzerland
:
ISO
,
2012
).
51.
Kondo
Y.
,
Corrosion
45
(
1989
):
p
.
7
11
.
52.
DuQuesnay
D.L.
,
Underhill
P.R.
,
Britt
H.J.
,
Int. J. Fatigue
25
(
2003
):
p
.
371
377
.
53.
Van Der Walde
K.
,
Hillberry
B.M.
,
Int. J. Fatigue
30
(
2008
):
p
.
106
118
.
54.
Gruenberg
K.M.
,
Craig
B.A.
,
Hillberry
B.M.
,
Bucci
R.J.
,
Hinkle
A.J.
,
Int. J. Fatigue
26
(
2004
):
p
.
629
640
.
55.
Kim
S.
,
Burns
J.T.
,
Gangloff
R.P.
,
Eng. Fract. Mech.
76
(
2009
):
p
.
651
667
.
56.
Co
N.E.C.
,
Brown
D.E.
,
Burns
J.T.
,
JOM
70
(
2018
):
p
.
1168
1174
.
57.
Findlay
S.J.
,
Harrison
N.D.
,
Mater. Today
(
2002
):
p
.
18
25
.
58.
Campbell
G.S.
,
Lahey
R.T.C.
, “
A Survey of Serious Aircraft Accidents Involving Fatigue Fracture Vol. 2 Rotary-Wing Aircraft
,”
National Aeronautical Establishment, Ottawa
,
1983
.
59.
Campbell
G.S.
,
Lahey
R.T.C.
, “
A Survey of Serious Aircraft Accidents Involving Fatigue Fracture Vol. 1 Fixed-Wing Aircraft
,”
National Aeronautical Establishment
,
Ottawa
,
1983
.
60.
Charles-Granville
U.-E.
,
Glover
C.
,
Scully
J.R.
,
Kelly
R.
,
J. Electrochem. Soc.
168
(
2021
):
p.
121509
.
61.
Miedlar
P.C.
,
Berens
A.P.
,
Gunderson
A.
,
Gallagher
J.P.
, “
Analysis and Support Initiative for Structural Technology (ASIST)
” (
Dayton, OH
:
University of Dayton Research Institute
,
2002
).