Experimental Studies into the Analysis Required for the Durability Assessment of 7075 and 6061 Cold Spray Repairs to Military Aircraft

: This paper presents an experimental study into the analysis required for the durability assessment of 7075 and 6061 cold spray repairs to military aircraft. To this end, it is ﬁrst shown that provided the bulk stress in a 7075 cold spray coating can be kept beneath approximately 150 MPa, then the coating should not crack. A range of examples are presented in which the interface between the coating and the substrate only fails subsequent to crack growth in the substrate. We also show that failure of cold spray repaired / coated panels can also be due to the nucleation and growth of cracks in the substructure immediately adjacent to the coated / repaired region. As such, when performing a durability analysis for a cold spray repair, the growth of such small naturally occurring cracks, both at the interface and immediately adjacent to the ends of the coating, need to be accounted for.

The United States Air Force (USAF) requirements for the durability and damage tolerance certification for cold spray repairs to aircraft structural parts are detailed in Structures Bulletin EZ-19-01 [23], and essentially reflect the requirements enunciated in MIL-Standard 1530D [24] and the US Joint Services Structural Guidelines JSSG2006 [25]. As noted in the recent review paper [9], which discussed a building block approach to certifying AM and cold sprayed (SPD) repaired parts, in order to meet the guidelines outlined in Structures Bulletin EZ-19-10, several questions need to be addressed, viz: (i) How can we accurately compute crack growth in an AM or SPD repaired part when there is a residual stress field?
(ii) How can we accurately compute crack growth in a cold spray repair where cracking can nucleate, either in the substructure being repaired, or at the intersection between the cold spray? (iii) How can we accurately perform a durability analysis for a cold spray repair if there are multiple collocated cracks? (iv) Can we relate da/dN versus ∆K equations, where da/dN is the crack growth rate per cycle and ∆K (= K max − K min ) is the difference between the maximum and minimum values of the stress intensity factor in a load cycle, needed to perform items (i) through (iii) to the corresponding curves determined for the conventionally manufactured material? Here, a is the crack length and N is the number of cycles.
To this end, [9,10] presented a methodology that was based on the Hartman-Schijve variant of the NASGRO crack growth equation [26], which is outlined in the Appendix A, that (to some extent, and provided that the crack growth studies needed for the durability assessment are known) enables each of these questions to be answered in a fashion that is consistent with the building block approach outlined in MIL-STD-1530D [24], and presented a number of example that illustrated this approach. However, the key is to know what cracking needs to be analysed as part of the durability assessment. In this context, it is clear that if the durability assessment does not address the actual cracking seen in the part being assessed, then the analysis will be of little use. Experience with cold spray repairs to fatigue critical parts in operational aircraft has shown that there can be several possible failure mechanisms. For example, in the SPD (cold spray) repair example presented in [9], which was a repair to intergranular corrosion at a fastener hole in a Royal Australian Air Force (RAAF) P3C Orion wing skin, the cracks nucleated from a material discontinuity at the bore of a fastener hole. This meant that the (nucleating) crack was exposed to the environment-in this case, laboratory air. Since, as shown in [13][14][15][19][20][21], cracks that determine the fatigue life in cold spray repairs can nucleate, and subsequently grow, from small sub mm internal material discontinuities in the substrate at the intersection of the cold spray coating with the substrate, such cracks may not be exposed to the environment. This is an important distinction since, as explained in [14], such cracks grow in a "vacuum-like" environment and, therefore, as per [27][28][29][30], grow much slower than if they could "see" the environment. As such, the growth of these interior cracks, which nucleate in the substrate at the intersection of the cold spray, will be much slower than interfacial cracks that can "see" the environment. The current paper presents a series of experimental studies that reveals the importance of this phenomena, i.e., failure due to cracking that nucleates from material discontinuities in the substrate at the interface between the cold spray and the substrate. At this stage, it should be noted that the growth of small naturally occurring lead cracks, i.e., the cracks that determine the operational life of a component [31][32][33], which see the environment often grow in a near exponential fashion [26,[31][32][33][34][35][36][37] and follow a da/dN versus ∆K curve that is essentially a Paris-like crack growth equation with a small fatigue threshold. Furthermore, for small cracks growing in laboratory air, there are minimal R ratio effects [26,30,[37][38][39][40][41][42]. (The relationship between the long crack and the small crack da/dN versus ∆K curves, for cracks that are growing in air, is discussed in [38].) However, as discussed in [14,43,44], for small cracks growing in a vacuum, there appears to be a well-defined (non-zero) threshold. Nevertheless, it appears that for such small interior cracks, the crack growth rate (da/dN) can be expressed as a function of (∆K − ∆K thr ) [43], where ∆K thr is the fatigue threshold.
Structures Bulletin EZ-19-01 [23] stresses the importance of analysis in the durability/economic life assessment of AM and AM repairs, such as cold spray (SPD) and laser-assisted deposition, and mandates the use of a minimum equivalent initial damage size (EIDS) of 0.254 mm. This size EIDS is taken from JSSG2006 [25]. A problem faced in the durability analysis needed for the certification of cold spray repairs is how can the necessary small crack da/dN versus ∆K curve be determined. It is tempting to use the approach proposed in [17], whereby a small sub mm notch is cut into a specimen. Unfortunately, there are numerous problems with this approach. In this context, it should be noted that the paper by Virkler and Hillbery [45] is acknowledged as the seminal paper that established the degree of scatter seen in da/dN versus ∆K tests performed to the fatigue test standard ASTM E647-13a [46]-see Figure 1. Other examples of this non-uniqueness, i.e., the variability in the da/dN versus ∆K curves for a conventionally manufactured material, are given in [47,48].
It has long been known [26,37,38,[47][48][49][50][51][52] that the degree of scatter associated with small naturally occurring cracks is significantly greater than that seen for long cracks. As such, the test protocol used in [17], which involved a limited number of tests (two) on artificial notches with lengths that varied from 0.15 to 0.2 mm, should not necessarily be thought of as generating the small crack growth data needed to perform the durability analysis required for the certification of cold spray parts, or cold spray repairs. This is reflected by the fact that in the near-threshold region, the da/dN versus ∆K curves presented in Figure 5a in [17] for what are termed long and short cracks in annealed cold spray 6061 are essentially identical, and lie within the scatter band that could be expected for the growth of long cracks [45,47]. Furthermore, since these tests were performed in a laboratory environment, the da/dN versus ∆K curves reported in [17] are also unlikely to be representative of those associated with cracks that nucleate beneath the surface of the cold spray (SPD) where the crack does not see the environment.
Aerospace 2020, 7, x FOR PEER REVIEW  3 of 19 It has long been known [26,37,38,[47][48][49][50][51][52] that the degree of scatter associated with small naturally occurring cracks is significantly greater than that seen for long cracks. As such, the test protocol used in [17], which involved a limited number of tests (two) on artificial notches with lengths that varied from 0.15 to 0.2 mm, should not necessarily be thought of as generating the small crack growth data needed to perform the durability analysis required for the certification of cold spray parts, or cold spray repairs. This is reflected by the fact that in the near-threshold region, the da/dN versus ΔK curves presented in Figure 5a in [17] for what are termed long and short cracks in annealed cold spray 6061 are essentially identical, and lie within the scatter band that could be expected for the growth of long cracks [45,47]. Furthermore, since these tests were performed in a laboratory environment, the da/dN versus ΔK curves reported in [17] are also unlikely to be representative of those associated with cracks that nucleate beneath the surface of the cold spray (SPD) where the crack does not see the environment. To illustrate the (potential) scatter associated with the growth of small/short cracks in conventionally manufactured 6061, it is pertinent to note the crack growth histories reported in [53,54]. Whereas [53] specifically stated that the crack growth was exponential, and that the crack rate da/dN was proportional to the crack length, a similar statement was not made in [54]. However, on plotting the crack growth history associated with the fastest crack reported in [54], it can be seen that crack growth was essentially exponential-see Figure 2. Hence, as a first approximation, the crack growth rate was essentially proportional to the crack length. As such, it follows that for these tests, the crack growth rate da/dN should be able to be approximated by a simple Paris-like power law with a low fatigue threshold and, as per [26,36,55,56], an exponent of (approximately) 2. As such, unlike the crack growth curves given in [17], the 6061 small/short crack growth data presented in [53,54] is consistent with the USAF Durability Design Handbook [34] and the USAF approach to assessing the economic life of military aircraft [35]. To illustrate the (potential) scatter associated with the growth of small/short cracks in conventionally manufactured 6061, it is pertinent to note the crack growth histories reported in [53,54]. Whereas [53] specifically stated that the crack growth was exponential, and that the crack rate da/dN was proportional to the crack length, a similar statement was not made in [54]. However, on plotting the crack growth history associated with the fastest crack reported in [54], it can be seen that crack growth was essentially exponential-see Figure 2. Hence, as a first approximation, the crack growth rate was essentially proportional to the crack length. As such, it follows that for these tests, the crack growth rate da/dN should be able to be approximated by a simple Paris-like power law with a low fatigue threshold and, as per [26,36,55,56], an exponent of (approximately) 2. As such, unlike the crack growth curves given in [17], the 6061 small/short crack growth data presented in [53,54] is consistent with the USAF Durability Design Handbook [34] and the USAF approach to assessing the economic life of military aircraft [35]. With this in mind, the purpose of the present paper is to shed light on the growth of small cracks in tests on specimens with cold spray (SPD) deposits on AA2024-T3, AA7075-T6, AA7075-T7351 and AA7050-T7451 substructures. A range of examples are presented in which the interface between the SPD and the substrate only fails subsequent to crack growth in the substrate. This failure mechanism is similar to those seen in [13][14][15][19][20][21], in that they involve the nucleation and subsequent growth from small material discontinuities in the substructure, at the intersection between the cold spray and the substructure, followed by delamination at the SPD to substrate interface. We also show that failure of cold spray repaired/coated panels can be due to the nucleation and growth of cracks in the substructure immediately outside of the coated/repaired region. As such, when performing a durability analysis for a cold spray repair, the growth of such small naturally occurring cracks, both at the interface and immediately adjacent to the ends of the coating, need to be accounted for. In the case of through thickness cracks in thin skins with cold spray coating, it should be noted that closedform analytical solutions for through thickness cracks repaired using cold spray were first given in [11].

Fatigue Tests on 7075 Powder SPD Doublers on 2024-T3 Specimens
A series of constant amplitude fatigue tests, at a load ratio of R = 0.1, were performed on 3.175 mm thick 2024-T3 dogbone specimens, with a geometry as shown in Figure 3. The tests were performed in an MTS 100 kN servo-hydraulic machine in accordance with the American Society Testing and Materials (ASTM) tests standard E 739-80 [57]. The effect of two different coating thicknesses, namely 0.1 and 0.5 mm, was evaluated and tests at a number of load levels were performed. The resultant stress life curve is shown in Figure 4 together with data obtained from tests on panels without a cold spray coating, and for one specimen that was not coated, which was grit blasted prior to testing at a stress in the working section of 225 MPa, which corresponds to a load of approximately 30.00 kN. The grit blasting (surface preparation) process used was the same as that used for all cold spray applications to Royal Australian Air Force aircraft [16], and the application parameters are as outlined in [11]. Indeed, this combination of surface treatment and application parameters is used in all of the cold spray examples discussed in this paper. Since it is common With this in mind, the purpose of the present paper is to shed light on the growth of small cracks in tests on specimens with cold spray (SPD) deposits on AA2024-T3, AA7075-T6, AA7075-T7351 and AA7050-T7451 substructures. A range of examples are presented in which the interface between the SPD and the substrate only fails subsequent to crack growth in the substrate. This failure mechanism is similar to those seen in [13][14][15][19][20][21], in that they involve the nucleation and subsequent growth from small material discontinuities in the substructure, at the intersection between the cold spray and the substructure, followed by delamination at the SPD to substrate interface. We also show that failure of cold spray repaired/coated panels can be due to the nucleation and growth of cracks in the substructure immediately outside of the coated/repaired region. As such, when performing a durability analysis for a cold spray repair, the growth of such small naturally occurring cracks, both at the interface and immediately adjacent to the ends of the coating, need to be accounted for. In the case of through thickness cracks in thin skins with cold spray coating, it should be noted that closed-form analytical solutions for through thickness cracks repaired using cold spray were first given in [11].

Fatigue Tests on 7075 Powder SPD Doublers on 2024-T3 Specimens
A series of constant amplitude fatigue tests, at a load ratio of R = 0.1, were performed on 3.175 mm thick 2024-T3 dogbone specimens, with a geometry as shown in Figure 3. The tests were performed in an MTS 100 kN servo-hydraulic machine in accordance with the American Society Testing and Materials (ASTM) tests standard E 739-80 [57]. The effect of two different coating thicknesses, namely 0.1 and 0.5 mm, was evaluated and tests at a number of load levels were performed. The resultant stress life curve is shown in Figure 4 together with data obtained from tests on panels without a cold spray coating, and for one specimen that was not coated, which was grit blasted prior to testing at a stress in the working section of 225 MPa, which corresponds to a load of approximately 30.00 kN. The grit blasting (surface preparation) process used was the same as that used for all cold spray applications to Royal Australian Air Force aircraft [16], and the application parameters are as outlined in [11]. Indeed, this combination of surface treatment and application parameters is used in all of the cold spray examples discussed in this paper. Since it is common practice for aerospace skins to be surface treated so as to minimise the effect of corrosion, Figure 4 also presents the stress life curves given in [58] for tests on chromic anodized 2024-T3.
Aerospace 2020, 7, x FOR PEER REVIEW 5 of 19 practice for aerospace skins to be surface treated so as to minimise the effect of corrosion, Figure 4 also presents the stress life curves given in [58] for tests on chromic anodized 2024-T3.  A key point to be noted in Figure 4 is that below a stress of approximately 150 MPa, there was no failure. There was also no cracking in the 7075 coating, or delamination of the coating from the practice for aerospace skins to be surface treated so as to minimise the effect of corrosion, Figure 4 also presents the stress life curves given in [58] for tests on chromic anodized 2024-T3.  A key point to be noted in Figure 4 is that below a stress of approximately 150 MPa, there was no failure. There was also no cracking in the 7075 coating, or delamination of the coating from the A key point to be noted in Figure 4 is that below a stress of approximately 150 MPa, there was no failure. There was also no cracking in the 7075 coating, or delamination of the coating from the 2024-T3 substrate. To illustrate this, Figure 5 presents a picture of the stress range of the bulk stresses field (∆σ 1 + ∆σ 2 + ∆σ 3 ), which is the first invariant of the stress tensor, obtained using infra-red thermography, in a 7075 coating of a specimen that was tested at a stress corresponding to approximately 150 MPa, that did not fail after more than 3,000,000 cycles. Here, ∆σ 1 , ∆σ 2 , and ∆σ 3 are the ranges of the three principal stresses, namely σ 1 , σ 2 , and σ 3 , in a cycle. (The test was stopped after approximately 3,000,000 cycles.) This picture reveals that there was no cracking in the coating.
Aerospace 2020, 7, x FOR PEER REVIEW 6 of 19 2024-T3 substrate. To illustrate this, Figure 5 presents a picture of the stress range of the bulk stresses field (Δσ1 + Δσ2 + Δσ3), which is the first invariant of the stress tensor, obtained using infra-red thermography, in a 7075 coating of a specimen that was tested at a stress corresponding to approximately 150 MPa, that did not fail after more than 3,000,000 cycles. Here, Δσ1, Δσ2, and Δσ3 are the ranges of the three principal stresses, namely σ1, σ2, and σ3, in a cycle. (The test was stopped after approximately 3,000,000 cycles.) This picture reveals that there was no cracking in the coating. To further illustrate this, Figure 6 presents a picture of the bulk stress field at 2,293,500 cycles in a specimen tested at a stress of approximately 169 MPa. This test was terminated shortly thereafter due to failure of the pump. This picture also reveals that there was no cracking in coating.  To further illustrate this, Figure 6 presents a picture of the bulk stress field at 2,293,500 cycles in a specimen tested at a stress of approximately 169 MPa. This test was terminated shortly thereafter due to failure of the pump. This picture also reveals that there was no cracking in coating.
Aerospace 2020, 7, x FOR PEER REVIEW 6 of 19 2024-T3 substrate. To illustrate this, Figure 5 presents a picture of the stress range of the bulk stresses field (Δσ1 + Δσ2 + Δσ3), which is the first invariant of the stress tensor, obtained using infra-red thermography, in a 7075 coating of a specimen that was tested at a stress corresponding to approximately 150 MPa, that did not fail after more than 3,000,000 cycles. Here, Δσ1, Δσ2, and Δσ3 are the ranges of the three principal stresses, namely σ1, σ2, and σ3, in a cycle. (The test was stopped after approximately 3,000,000 cycles.) This picture reveals that there was no cracking in the coating. To further illustrate this, Figure 6 presents a picture of the bulk stress field at 2,293,500 cycles in a specimen tested at a stress of approximately 169 MPa. This test was terminated shortly thereafter due to failure of the pump. This picture also reveals that there was no cracking in coating.    8 show the failure surface associated with specimen number S24, which had a 0.5 mm thick 7075 coating and which failed at 114,344 cycles, and specimen S22 which failed at 122,521 cycles. In both cases, we see that cracks in the 2024-T3 nucleated at multiple locations at the intersection between the 7075 coating and the 2024-T3 substructure. The growth of these (multiple) cracks subsequently led to delamination of the coating from the 2024-T3. Also note the intact nature of the interface on the opposing (uncracked) surface. This phenomenon, i.e., failure due to the nucleation and growth of multiple cracks in the 2024-T3 at the interface between the 2024-T3 and the coating, was seen in all cases.
Aerospace 2020, 7, x FOR PEER REVIEW 7 of 19 Figures 7 and 8 show the failure surface associated with specimen number S24, which had a 0.5 mm thick 7075 coating and which failed at 114,344 cycles, and specimen S22 which failed at 122,521 cycles. In both cases, we see that cracks in the 2024-T3 nucleated at multiple locations at the intersection between the 7075 coating and the 2024-T3 substructure. The growth of these (multiple) cracks subsequently led to delamination of the coating from the 2024-T3. Also note the intact nature of the interface on the opposing (uncracked) surface. This phenomenon, i.e., failure due to the nucleation and growth of multiple cracks in the 2024-T3 at the interface between the 2024-T3 and the coating, was seen in all cases.  A similar phenomenon, i.e., that the cold spray (SPD) coating could withstand significant cracking in the 2024-T3 skin, was seen also in tests on 7075 powder SPD repairs to fuselage lap joints Figures 7 and 8 show the failure surface associated with specimen number S24, which had a 0.5 mm thick 7075 coating and which failed at 114,344 cycles, and specimen S22 which failed at 122,521 cycles. In both cases, we see that cracks in the 2024-T3 nucleated at multiple locations at the intersection between the 7075 coating and the 2024-T3 substructure. The growth of these (multiple) cracks subsequently led to delamination of the coating from the 2024-T3. Also note the intact nature of the interface on the opposing (uncracked) surface. This phenomenon, i.e., failure due to the nucleation and growth of multiple cracks in the 2024-T3 at the interface between the 2024-T3 and the coating, was seen in all cases.  A similar phenomenon, i.e., that the cold spray (SPD) coating could withstand significant cracking in the 2024-T3 skin, was seen also in tests on 7075 powder SPD repairs to fuselage lap joints A similar phenomenon, i.e., that the cold spray (SPD) coating could withstand significant cracking in the 2024-T3 skin, was seen also in tests on 7075 powder SPD repairs to fuselage lap joints seen in Boeing B737 and B727 aircraft [12]. In [12], it was found that cracking and delamination in the 7075 coating only arose after the cracks in the 2024-T3 fuselage skin had reached a length of approximately 6 mm. The cracking in the coating arose as it attempted to carry the load shed from the 2024-T3 skin [12]. Furthermore, as reported in [12], the cracking in the coating often lay on a different plane to that of the cracking in the 2024-T3. In other words, the crack from the 2024-T3 did not grow into the cold spray.
This phenomenon was also seen in the current fatigue test program-see Figures 9 and 10, where we see both delamination of the 7075 coating from the 2024-T3-which arose as a result of failure in the 2024-T3 substrate, and the fact that the failure of the coating was frequently associated with a different plane to that of the cracking in the 2024-T3 substrate.
Aerospace 2020, 7, x FOR PEER REVIEW 8 of 19 seen in Boeing B737 and B727 aircraft [12]. In [12], it was found that cracking and delamination in the 7075 coating only arose after the cracks in the 2024-T3 fuselage skin had reached a length of approximately 6 mm. The cracking in the coating arose as it attempted to carry the load shed from the 2024-T3 skin [12]. Furthermore, as reported in [12], the cracking in the coating often lay on a different plane to that of the cracking in the 2024-T3. In other words, the crack from the 2024-T3 did not grow into the cold spray. This phenomenon was also seen in the current fatigue test program-see Figures 9 and 10, where we see both delamination of the 7075 coating from the 2024-T3-which arose as a result of failure in the 2024-T3 substrate, and the fact that the failure of the coating was frequently associated with a different plane to that of the cracking in the 2024-T3 substrate.   seen in Boeing B737 and B727 aircraft [12]. In [12], it was found that cracking and delamination in the 7075 coating only arose after the cracks in the 2024-T3 fuselage skin had reached a length of approximately 6 mm. The cracking in the coating arose as it attempted to carry the load shed from the 2024-T3 skin [12]. Furthermore, as reported in [12], the cracking in the coating often lay on a different plane to that of the cracking in the 2024-T3. In other words, the crack from the 2024-T3 did not grow into the cold spray. This phenomenon was also seen in the current fatigue test program-see Figures 9 and 10, where we see both delamination of the 7075 coating from the 2024-T3-which arose as a result of failure in the 2024-T3 substrate, and the fact that the failure of the coating was frequently associated with a different plane to that of the cracking in the 2024-T3 substrate.

Test Specimen Geometry and Dimensions
To continue this study, we tested SPD (cold spray) repairs to a simulated corrosion rework in a 7075-T7351 skin. The geometry of the test specimen is as shown in Figure 11, and a typical repaired specimen is shown in Figure 12. A total of 16 specimens were tested-see Table 1. The tests were performed at a maximum load of 38.1 kN and R = 0.1. This corresponds to a maximum remote stress of approximately 240 MPa. The cold spray coatings evaluated in this study were 7075, 6061, and ceramic particles in a 6061 aluminium matrix, which in this study is referred to as AMC.

Test Specimen Geometry and Dimensions
To continue this study, we tested SPD (cold spray) repairs to a simulated corrosion rework in a 7075-T7351 skin. The geometry of the test specimen is as shown in Figure 11, and a typical repaired specimen is shown in Figure 12. A total of 16 specimens were tested-see Table 1. The tests were performed at a maximum load of 38.1 kN and R = 0.1. This corresponds to a maximum remote stress of approximately 240 MPa. The cold spray coatings evaluated in this study were 7075, 6061, and ceramic particles in a 6061 aluminium matrix, which in this study is referred to as AMC.

Test Specimen Geometry and Dimensions
To continue this study, we tested SPD (cold spray) repairs to a simulated corrosion rework in a 7075-T7351 skin. The geometry of the test specimen is as shown in Figure 11, and a typical repaired specimen is shown in Figure 12. A total of 16 specimens were tested-see Table 1. The tests were performed at a maximum load of 38.1 kN and R = 0.1. This corresponds to a maximum remote stress of approximately 240 MPa. The cold spray coatings evaluated in this study were 7075, 6061, and ceramic particles in a 6061 aluminium matrix, which in this study is referred to as AMC.    Prior to fatigue testing, lock-in thermography (infra-red) stress pictures were taken both for the unrepaired specimens and for the specimens with 7075, 6061 and AMC SPD coatings. A typical infra-red bulk stress picture is shown in Figure 13. These figures revealed that the application of the coatings resulted in the specimens having a near uniform stress field. In other words, the coating had (essentially) removed the stress concentration due to the simulated corrosion cut.  Prior to fatigue testing, lock-in thermography (infra-red) stress pictures were taken both for the unrepaired specimens and for the specimens with 7075, 6061 and AMC SPD coatings. A typical infrared bulk stress picture is shown in Figure 13. These figures revealed that the application of the coatings resulted in the specimens having a near uniform stress field. In other words, the coating had (essentially) removed the stress concentration due to the simulated corrosion cut.

Fatigue Test Results
The test lives associated with these specimens are listed in Table 1. Table 1 reveals that the baseline specimens had the longest test lives. Of the repaired specimens, the non-heat-treated 6061 doubler specimens had the longest fatigue lives. The mean fatigue lives of specimens with non-heattreated 7075 coatings, heat-treated 7075HT, AMC and AMC LT-NGB were approximately 77.6%, 38%, 43.3%, and 60% of that seen by specimens with a non-heat-treated 6061 SPD coating (doubler).

Fatigue Test Results
The test lives associated with these specimens are listed in Table 1. Table 1 reveals that the baseline specimens had the longest test lives. Of the repaired specimens, the non-heat-treated 6061 doubler specimens had the longest fatigue lives. The mean fatigue lives of specimens with non-heat-treated 7075 coatings, heat-treated 7075HT, AMC and AMC LT-NGB were approximately 77.6%, 38%, 43.3%, and 60% of that seen by specimens with a non-heat-treated 6061 SPD coating (doubler). Heat treatment of the 7075 coating significantly reduced the fatigue life. Typical failure surfaces are shown in Figures 14-18.
All of the specimens failed as a result of the nucleation of multiple cracks in the 7075-T7351 substrate at the intersection between the coating and the substrate. This failure mechanism is identical to that seen in the previous studies.
Aerospace 2020, 7, x FOR PEER REVIEW 11 of 19 Heat treatment of the 7075 coating significantly reduced the fatigue life. Typical failure surfaces are shown in Figures 14-18. All of the specimens failed as a result of the nucleation of multiple cracks in the 7075-T7351 substrate at the intersection between the coating and the substrate. This failure mechanism is identical to that seen in the previous studies.     Heat treatment of the 7075 coating significantly reduced the fatigue life. Typical failure surfaces are shown in Figures 14-18. All of the specimens failed as a result of the nucleation of multiple cracks in the 7075-T7351 substrate at the intersection between the coating and the substrate. This failure mechanism is identical to that seen in the previous studies.     Heat treatment of the 7075 coating significantly reduced the fatigue life. Typical failure surfaces are shown in Figures 14-18. All of the specimens failed as a result of the nucleation of multiple cracks in the 7075-T7351 substrate at the intersection between the coating and the substrate. This failure mechanism is identical to that seen in the previous studies.     Heat treatment of the 7075 coating significantly reduced the fatigue life. Typical failure surfaces are shown in Figures 14-18. All of the specimens failed as a result of the nucleation of multiple cracks in the 7075-T7351 substrate at the intersection between the coating and the substrate. This failure mechanism is identical to that seen in the previous studies.     Here it should be noted that, despite the stress in the 7075 coating exceeding 150 MPa, failure did not initiate in the coating.

Repair of Simulated Corrosion in a 7075-T6 Panel
It is pertinent to compare the results of the previous study with those seen in a study [11,12] on cold spray repairs to corrosion damage in AA7075-T6. The geometry of the specimens tested [11,12] is shown in Figure 19. As can be seen in Figures 20 and 21, for a specimen subjected to a remote maximum load of 35 kN, which failed at 640,000, crack nucleation occurred at multiple locations in the 7075-T6 substructure at the intersection between the 7075 coating and the 7075-T6, and also on the opposing face that did not have a 7075 coating/repair. The growth of these cracks also subsequently led to delamination of the coating. However, crack initiation and growth also occurred on the surface of the specimen where there was no coating-see Figures 20 and 21. The sizes of the nucleating defects at these two opposing sites were similar.   Here it should be noted that, despite the stress in the 7075 coating exceeding 150 MPa, failure did not initiate in the coating.

Repair of Simulated Corrosion in a 7075-T6 Panel
It is pertinent to compare the results of the previous study with those seen in a study [11,12] on cold spray repairs to corrosion damage in AA7075-T6. The geometry of the specimens tested [11,12] is shown in Figure 19. As can be seen in Figures 20 and 21, for a specimen subjected to a remote maximum load of 35 kN, which failed at 640,000, crack nucleation occurred at multiple locations in the 7075-T6 substructure at the intersection between the 7075 coating and the 7075-T6, and also on the opposing face that did not have a 7075 coating/repair. The growth of these cracks also subsequently led to delamination of the coating. However, crack initiation and growth also occurred on the surface of the specimen where there was no coating-see Figures 20 and 21. The sizes of the nucleating defects at these two opposing sites were similar. Here it should be noted that, despite the stress in the 7075 coating exceeding 150 MPa, failure did not initiate in the coating.

Repair of Simulated Corrosion in a 7075-T6 Panel
It is pertinent to compare the results of the previous study with those seen in a study [11,12] on cold spray repairs to corrosion damage in AA7075-T6. The geometry of the specimens tested [11,12] is shown in Figure 19. As can be seen in Figures 20 and 21, for a specimen subjected to a remote maximum load of 35 kN, which failed at 640,000, crack nucleation occurred at multiple locations in the 7075-T6 substructure at the intersection between the 7075 coating and the 7075-T6, and also on the opposing face that did not have a 7075 coating/repair. The growth of these cracks also subsequently led to delamination of the coating. However, crack initiation and growth also occurred on the surface of the specimen where there was no coating-see Figures 20 and 21. The sizes of the nucleating defects at these two opposing sites were similar.   Here it should be noted that, despite the stress in the 7075 coating exceeding 150 MPa, failure did not initiate in the coating.

Repair of Simulated Corrosion in a 7075-T6 Panel
It is pertinent to compare the results of the previous study with those seen in a study [11,12] on cold spray repairs to corrosion damage in AA7075-T6. The geometry of the specimens tested [11,12] is shown in Figure 19. As can be seen in Figures 20 and 21, for a specimen subjected to a remote maximum load of 35 kN, which failed at 640,000, crack nucleation occurred at multiple locations in the 7075-T6 substructure at the intersection between the 7075 coating and the 7075-T6, and also on the opposing face that did not have a 7075 coating/repair. The growth of these cracks also subsequently led to delamination of the coating. However, crack initiation and growth also occurred on the surface of the specimen where there was no coating-see Figures 20 and 21. The sizes of the nucleating defects at these two opposing sites were similar.

Damage Tolerance of SPD Repaired Specimens
The paper by Jones et al. [11] was one of the first to illustrate the damage tolerance of cold spray coatings and repairs. To this end, [11] presented the results of constant amplitude tests on a 350 mm (long) by 60 mm wide 1.27 mm thick 2024-T3 clad aluminum alloy single-edge notch (SENT) test specimen with a 2 mm long edge crack. The maximum and minimum stresses in the tests were σmax = 181 MPa and σmin = 18.1 MPa respectively. Two tests were performed, viz a baseline specimen that did not have a 7075 coating, and a specimen that had a 1 mm thick full-width 7075 coating on both sides of the 2024-T3 specimen. Whilst the baseline specimen (no coating) failed at approximately 35,000 cycles, the 7075 coated panel was stopped test after approximately 70,000 cycles with no evident delamination damage or cracking in the 7075 coating, or crack growth in the 2024-T3 skin. (It should be noted that in [11] only the result up to 60,000 cycles was reported. An additional 10,000 cycles were subsequently completed to ensure a life factor of two.) As reported in [11], there was no crack growth in either the 2024-T3 or the 7075 coating and no delamination of the coating from the 2024-T3.
To further illustrate that cracking in the coating and delamination can be controlled by the appropriate choice of the thickness of the coating, [12] presented the results of a study into a 7075-T6 specimen with a 4.76 mm (3/16 inch) universal dome head fastener that contained a 1.27 mm long initial crack, in a 30 mm wide 4 mm thick and 400 mm long panel that was sprayed with a 1 mm thick 7075 coating. The specimens were subjected to a USAF C-141 military transport spectrum with a peak remote stress in the spectrum of 200 MPa. To avoid cracking in the coating, the thickness of the coating was designed to keep the maximum stress in the coating beneath 133 MPa. The tests were stopped after approximately 19,000 flights, as there was no apparent cracking or delamination in the coating, or crack growth in the 7075-T6 skin. This result is consistent with the observation reported Section 2 that below a stress of approximately 150 MPa, there should be no cracking in a 7075 coating.
The damage tolerance of an appropriately designed and applied coating, specifically its ability to resist cracking and delamination, was further illustrated in [11] that studied the ability of a 10 mm wide and 1 mm thick 7075 SPD strip, with a triangular cross section, to retard the growth of a 3.2 mm long edge crack in a 1.27 mm thick 2024-T3 aluminum alloy specimen-see Figure 22. The 7075 strip was deposited a distance of 5 mm from the edge of the specimen-see [11] for more details. The specimen was subjected to a cyclic stress, in the working section, of approximately 53 MPa and R = 0.1-see [11]. The stress images given in [11] revealed that the stresses in the 7075 strip were significantly less than 150 MPa, i.e., less than the endurance level of the 7075 coating. Furthermore, despite the crack continuing to grow beneath the 7075 strip, there was no evidence of cracking or delamination of the 7075 strip and failure initiated in the 2024-T3 at the ends of the strip-see Figure   Figure 21. Close-up view of cracking on the side opposite to the repair/coating, and subsequent delamination both between the coating and the 7075-T6 substrate.

Damage Tolerance of SPD Repaired Specimens
The paper by Jones et al. [11] was one of the first to illustrate the damage tolerance of cold spray coatings and repairs. To this end, [11] presented the results of constant amplitude tests on a 350 mm (long) by 60 mm wide 1.27 mm thick 2024-T3 clad aluminum alloy single-edge notch (SENT) test specimen with a 2 mm long edge crack. The maximum and minimum stresses in the tests were σ max = 181 MPa and σ min = 18.1 MPa respectively. Two tests were performed, viz a baseline specimen that did not have a 7075 coating, and a specimen that had a 1 mm thick full-width 7075 coating on both sides of the 2024-T3 specimen. Whilst the baseline specimen (no coating) failed at approximately 35,000 cycles, the 7075 coated panel was stopped test after approximately 70,000 cycles with no evident delamination damage or cracking in the 7075 coating, or crack growth in the 2024-T3 skin. (It should be noted that in [11] only the result up to 60,000 cycles was reported. An additional 10,000 cycles were subsequently completed to ensure a life factor of two.) As reported in [11], there was no crack growth in either the 2024-T3 or the 7075 coating and no delamination of the coating from the 2024-T3.
To further illustrate that cracking in the coating and delamination can be controlled by the appropriate choice of the thickness of the coating, [12] presented the results of a study into a 7075-T6 specimen with a 4.76 mm (3/16 inch) universal dome head fastener that contained a 1.27 mm long initial crack, in a 30 mm wide 4 mm thick and 400 mm long panel that was sprayed with a 1 mm thick 7075 coating. The specimens were subjected to a USAF C-141 military transport spectrum with a peak remote stress in the spectrum of 200 MPa. To avoid cracking in the coating, the thickness of the coating was designed to keep the maximum stress in the coating beneath 133 MPa. The tests were stopped after approximately 19,000 flights, as there was no apparent cracking or delamination in the coating, or crack growth in the 7075-T6 skin. This result is consistent with the observation reported Section 2 that below a stress of approximately 150 MPa, there should be no cracking in a 7075 coating.
The damage tolerance of an appropriately designed and applied coating, specifically its ability to resist cracking and delamination, was further illustrated in [11] that studied the ability of a 10 mm wide and 1 mm thick 7075 SPD strip, with a triangular cross section, to retard the growth of a 3.2 mm long edge crack in a 1.27 mm thick 2024-T3 aluminum alloy specimen-see Figure 22. The 7075 strip was deposited a distance of 5 mm from the edge of the specimen-see [11] for more details. The specimen was subjected to a cyclic stress, in the working section, of approximately 53 MPa and R = 0.1-see [11]. The stress images given in [11] revealed that the stresses in the 7075 strip were significantly less than 150 MPa, i.e., less than the endurance level of the 7075 coating. Furthermore, despite the crack continuing to grow beneath the 7075 strip, there was no evidence of cracking or delamination of the 7075 strip and failure initiated in the 2024-T3 at the ends of the strip-see Figure 23. This indicated that the 7075 deposition was more damage tolerant than the baseline 2024-T3 material. This test program was the first to highlight that the durability assessment of cold spray repairs needs to consider cracking in the substrate adjacent to a repair that can arise due to a cold spray deposition.
Aerospace 2020, 7, x FOR PEER REVIEW 14 of 19 23. This indicated that the 7075 deposition was more damage tolerant than the baseline 2024-T3 material. This test program was the first to highlight that the durability assessment of cold spray repairs needs to consider cracking in the substrate adjacent to a repair that can arise due to a cold spray deposition.

Conclusions
The examples presented in this study illustrate that provided the stress in a 7075 SPD (cold spray) coating can be kept beneath approximately 150 MPa, then the coating should not crack or delaminate. On the other hand, we have seen that the failure in tests on 7075 cold spray coated 2024-T3 specimens is often as a result of the nucleation of multiple small cracks in the substrate at the intersection between the coating and the substrate. This failure mechanism is also seen in 7075 cold spray repairs to simulated corrosion damage in both 7075-T6 and 7075-T7351 specimens, and in 6051 cold spray repairs to simulated corrosion damage in 7075-T7351 specimens. In this context, we have seen that: (a) The post heat treatment of the 7075 powder coatings significantly reduced the fatigue life of repaired 7075-T7351 specimens. It is hypothesised that this may be due to its effect on the residual stresses in the repaired specimen. (b) When compared to the use of 7075 and 6061 powders, the use of AMC powder to repair simulated corrosion in the 7075-T7351 specimen tests led to lower fatigue lives.   23. This indicated that the 7075 deposition was more damage tolerant than the baseline 2024-T3 material. This test program was the first to highlight that the durability assessment of cold spray repairs needs to consider cracking in the substrate adjacent to a repair that can arise due to a cold spray deposition.

Conclusions
The examples presented in this study illustrate that provided the stress in a 7075 SPD (cold spray) coating can be kept beneath approximately 150 MPa, then the coating should not crack or delaminate. On the other hand, we have seen that the failure in tests on 7075 cold spray coated 2024-T3 specimens is often as a result of the nucleation of multiple small cracks in the substrate at the intersection between the coating and the substrate. This failure mechanism is also seen in 7075 cold spray repairs to simulated corrosion damage in both 7075-T6 and 7075-T7351 specimens, and in 6051 cold spray repairs to simulated corrosion damage in 7075-T7351 specimens. In this context, we have seen that: (a) The post heat treatment of the 7075 powder coatings significantly reduced the fatigue life of repaired 7075-T7351 specimens. It is hypothesised that this may be due to its effect on the residual stresses in the repaired specimen. (b) When compared to the use of 7075 and 6061 powders, the use of AMC powder to repair simulated corrosion in the 7075-T7351 specimen tests led to lower fatigue lives.

Conclusions
The examples presented in this study illustrate that provided the stress in a 7075 SPD (cold spray) coating can be kept beneath approximately 150 MPa, then the coating should not crack or delaminate. On the other hand, we have seen that the failure in tests on 7075 cold spray coated 2024-T3 specimens is often as a result of the nucleation of multiple small cracks in the substrate at the intersection between the coating and the substrate. This failure mechanism is also seen in 7075 cold spray repairs to simulated corrosion damage in both 7075-T6 and 7075-T7351 specimens, and in 6051 cold spray repairs to simulated corrosion damage in 7075-T7351 specimens. In this context, we have seen that: (a) The post heat treatment of the 7075 powder coatings significantly reduced the fatigue life of repaired 7075-T7351 specimens. It is hypothesised that this may be due to its effect on the residual stresses in the repaired specimen. (b) When compared to the use of 7075 and 6061 powders, the use of AMC powder to repair simulated corrosion in the 7075-T7351 specimen tests led to lower fatigue lives.
We have also seen that failure of cold spray repaired/coated panels can be due to the nucleation and growth of cracks in the substructure immediately adjacent (outside) to the coated/repaired region.
It follows from these studies that a durability analysis of a cold spray repair needs to be able to account for the growth of such small naturally occurring cracks, particularly the growth of the lead cracks both in the substructure outside the repair and in the substructure at the intersection between the repair/coating and the substructure, under the flight load spectra of interest. In this context, to meet the requirements delineated in Structures Bulletin EZ-SB-19-01 [22] and MIL-STD-1530D [23], the analysis tools must first be shown to be able to account for the growth of small naturally occurring cracks, in the unrepaired structure, under this flight load spectrum.
It should also be noted that in several of these studies, the cold spray repair did not significantly enhance the fatigue life of the structure. How to overcome this shortcoming will be explained in a subsequent paper.
Author Contributions: Conceptualisation and initial analysis, R.J.; experimental testing, D.P.; specimen manufacture and review of the final manuscript, N.M.; initial draft preparation, R.K.S.R.; overview of the paper and its relationship to aircraft certification, N.P. All authors have read and agreed to the published version of the manuscript.
Funding: This work was funded by the US Navy, Naval Air Systems Command (NAVAIR).

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A. The Hartman-Schijve Variant of the Nasgro Crack Growth Equation
The Nasgro crack growth equation [59] takes the form: where D, m, p, A and ∆K eff,thr are constants, and ∆K eff is an effective stress intensity factor, which, as per Elber [60], can be expressed in terms of a function U(R), viz: where ∆K is the range in the stress intensity factor seen in a load cycle, and U is a function of the R ratio. (In [60], the term ∆K eff was introduced to enable the various R ratio-dependent da/dN versus ∆K curves to collapse onto a single da/dN versus ∆K eff curve.) Setting m = p and q = p/2, we see [26] that Equation (A1) reduces to the Hartman-Schijve crack growth equation, viz: da/dN = D (∆K eff − ∆K eff,thr )) p /(1 − (K max /A)) p/2 (A3) For cracks that grow from small naturally occurring material discontinuities, there is often little crack tip shielding [26,[37][38][39]46,53,54,61], so that ∆K eff asymptotes to ∆K [26,53,54] and Equation (A3) reduces to da/dN = D (∆κ) p2 (A4) where the crack driving force ∆κ is as defined by Schwalbe [62], viz: A feature of this formulation is that it has been shown to be able to accurately perform durability analyses for unrepaired airframes, as well as for cold spray repaired and lased added deposition (LAD) repairs [1,9,26,37,39,[63][64][65].