Fatigue Behavior of Al 7075-T6 Plates Repaired with Composite Patch under the Effect of Overload

Abstract

Repair of aeronautical structures by composite patch bonding has shown its effectiveness in several studies during the last few decades. This repair technique leads to a retardation in the propagation of repaired cracks via load bridging across the patch throughout the adhesive layer, interfacing it with the repaired structure. The purpose of this study is to analyze the behavior of patch-repaired cracks present in thin plates made of aluminum alloy 7075-T6 and subjected to a single tensile overload. The sequence of application of overload on the fatigue behavior was also studied. Fatigue tests were conducted on Al 7075-T6 notched specimens where crack growth and number of cycles to failure were monitored for different patching/overload scenarios. A detailed fractographic study was performed on failed specimens to analyze the micromechanical behavior of the crack growth related to each scenario. The obtained results showed that the application of the overload before bonding the patch leads to an almost infinite fatigue life of the repaired plates.

1. Introduction

Operational aeronautical structures are, most often, subjected to fatigue loadings with variable amplitude. These differences in the amplitude level of fatigue loadings directly influence the crack propagation rates in aeronautical structures [1,2,3]. A clear, striking example of the effects of the variation in loading amplitudes on the crack growth rate is shown through the abrupt application of an overload during a constant amplitude load fatigue test [4,5,6]. The application of an overload causes a delay in the crack growth, resulting in a “stop” of the propagation of the crack after the overload and during a very large number of the original lower constant amplitude loading cycles. This behavior is essentially due to the formation of a large plastic zone and residual stress in the vicinity of the crack tip. It has been shown in the literature that there is a threshold overload rate below which the retardation effect will not exist [7]. In plane conditions, especially plane stress, higher plasticity around the crack front leads to a blunting of the crack tip, reported by several authors who studied overloaded specimens made of aeronautical aluminum alloys, such as 2024-T3 or 7075-T6 [8,9,10]. Bichler and Rippan [11] showed that the crack blunting increases the effective stress intensity factor (SIF), which causes an initial acceleration in the crack propagation, but the closure level increases rapidly as the crack grows because of the wedging action of the overload.

Repairing aeronautical cracked structures under fatigue loading by composite patch is one of the promising techniques to improve the fatigue life of these structures. This technique is based on bonding a high-performance composite material plate to the cracked region, thereby allowing load transfer from the cracked structure to the composite patch through the adhesive layer. This transfer makes it possible to reduce the stresses intensity around the crack front leading to the reduction of the crack propagation rate and consequently the improvement of the fatigue life of the damaged structure [12,13,14,15,16]. It has been shown in the literature that the effectiveness of composite patch repairs depends on a large number of parameters that can be classified as follows: (i) mechanical properties of repaired material, composite material, and adhesive type [17,18,19]; (ii) geometrical properties of the composite [20,21,22] (iii); and nature of the fatigue loading [23,24,25]. In addition, other phenomena could drastically influence the repair process and efficiency, such as the adhesive disband and the thermal residual stresses due to the adhesive curing. The design of a composite patch is a pretty difficult task because of the interaction of all these parameters on the performance of the repair.

The bonding of a composite patch on a fractured structure generates a crack retardation growth amount proportional to the initial crack length of the patch bonding. Albedah et al. [26] showed that the crack must be repaired as soon as it is detected. The history of fatigue loading has a crucial effect on the effectiveness of the repair. Thus, the amplitude of fatigue loading and the load ratio directly affect the residual life of the repaired structure using the composite patch. When a repaired structure is subjected to an overload cycle, two retardation effects interact, with the composite patch load bridging and the plastic zone created in front of the crack tip caused by the overload. The study of this interaction of the mentioned retardation techniques (Patch + Overload) has not been sufficiently analyzed in the literature. In recent studies, Mohammed et al. [27] searched some aspects of the interaction effects for Al 2024-T3 and Al 7075-T6 plates. They showed that if the patch is bonded before the application of the overload, the effect of the overload is attenuated by the presence of the composite patch. Conversely, and still according to Mohammed et al. [27], if the patch is bonded after the application of the overload, the two retardation effects are combined and the repaired plate experienced almost an infinite fatigue life. The explanation of these findings requires a considerable effort, which led us to an extensive use of the fractographic measurements and analyses to better understand the mechanism of interaction between the overload and those of composite patch.

In this study, the fractographic analysis is used to study the fatigue failure of specimens made of Al 7075-T6 material, repaired with bonded carbon fiber composite patches, and subjected to constant amplitude loading including a single overload cycle. The aim of the study is to explain by the fractographic approach the interaction between the retardation effects due to the overload and the presence of a composite patch.

2. Material and Experimental Procedure

The substrate material investigated in this study was 2-mm thin sheet of Al 7075-T6 alloy provided by Aerometals. The chemical composition of the Al 7075-T6 is provided in

Table 1. Chemical composition of the aluminum alloys.

Wt.%
-	Al	Cr	Cu	Fe	Mg	Mn	Ti	Si	Zn	Other
Al 7075-T6	89.3	0.23	1.6	0.5	2.5	0.3	0.2	0.4	5.3	0.15

. The samples were machined into dimensions 150 × 50 mm, as shown in Figure 1a, by abrasive waterjet to avoid formation of any machining residual stresses at the v-notch. The notch was machined perpendicular to the rolling direction to facilitate initiation and mode I crack propagation. The cracked substrate was repaired by an eight layered carbon/epoxy patch using a structural adhesive Araldite 2015. The curing parameters and procedure of both carbon/epoxy and Araldite can be found elsewhere [27]. The schematic of the crack repair process is presented in Figure 1b, where the adhesive is applied on to the cracked plate and bonded with a composite patch. As the adhesive was cured at room temperature, there were no thermal residual stresses induced in the assembly. The crack was repaired only on one side in order to monitor its growth from the free surface.

Tension-tension fatigue tests were carried out on a servo-hydraulic testing system, Instron made (model 8801) at an ambient temperature under load-controlled mode. The load ratio (F_max/F_min) of the tests applied was 0.1, and a constant frequency of 20 Hz was maintained during the tests. All the tests were repeated at least twice to maintain the consistency and accuracy of the results. The deviation in the results was found to be less than 10%, which can be regarded as acceptable experimental scatter. Typical test results were included in the analysis. Constant amplitude (CA) load tests were performed at a maximum load F_max = 7 kN until the total failure of the specimen. The overload ratio (OLR) is defined as the ratio of the maximum peak of the overload (F_OL) to the maximum constant amplitude (F_max) load. The overload was applied under load control mode by raising the load to the desired overload value, decreasing to the minimum value, and returning to the same mean load as that prior to the overload. A single tensile OL with an OLR of 2 was used in the present investigation. The following five cases were considered in this study to conduct a comprehensive analysis of crack propagation due to OL (see Figure 2):

Case I (CAL): The specimen is subjected to CAL until the failure. In this case, there was no presence of OL or patch. It is referred as the reference/baseline case.

Case II (CAL+P): The specimen is subjected to CAL at a maximum load of 7 kN until a natural crack of a length a = 3 mm is reached. At this stage, a composite patch of the same width as the specimen is bonded on the top of the crack. Once the repair is ready (adhesive fully polymerized), CAL is continued until final failure.

Case III (CAL+OL): A single tensile OL of 14 kN is applied once the natural crack develops to a 3-mm length from the notch tip under a maximum load of 7 kN. After the OL is applied, the specimen is subjected to the CAL until failure.

Case IV (CAL+P+OL): In this case, the patch is bonded at the same crack length of a = 3 mm developed under the CAL. A single OL is applied, and CAL is carried on until the total failure of the specimen.

Case V (CAL+OL+P): This is similar to case IV except that the OL was applied before bonding the patch.

From the fatigue tests, the life summary of the different cases was evaluated, and the crack propagation was evaluated as a function of the crack length. Comprehensive fractography of specimens after failure were observed under Scanning Electron Microscope (SEM) model Joel JSM-6610LV. Images of the fractured surface of failed specimens were taken for different loading cases with high and low magnification (see detail in Figure 3).

3. Results and Discussion

As stated earlier, the aim of this work was to study the interaction between crack retardation due to the application of the overload and that due to the presence of bonded composite patch for specimens made of aluminum alloy 7075-T6. The fractography was used as the principal approach to analyze crack growth behavior. Nonetheless, before looking at combined effects of the two techniques, it is necessary first to quantify in terms of fatigue crack propagation the two retardation effects independently and then combined in different orders.

3.1. Fatigue Crack Propagation

In order to analyze the crack propagation with the application of the overload and with the presence of the composite patch, we measured the evolution of the crack length with respect to the number of loading cycles to failure for different scenarios: CAL to failure, CAL+OL+CAL to failure, CAL+OL+P+CAL to failure, and CAL+P+OL+CAL to failure. These measurements made it possible to plot the fatigue crack growth curves (a = f(N)), from which we derived the fatigue crack growth (FCG) rate curves (da/dN = f (a)) for different point of application of the overload.

Figure 4 presents the fatigue crack growth curve for different loading cases: CAL, CAL+OL, CAL+P+OL, and CAL+OL+P on a semi-log scale. We can notice that in the case where the patch is bonded after the overload, the fatigue life is close to 3 million cycles; indicating that the combined retardation effect (overload + patch) in this case is very significant. In opposition, the retardation effect is practically negligible in the cases of overload without patch (CAL+OL) and overload after patch bonding (CAL+P+OL), as shown in Figure 5 on a semi-log scale. To better elucidate the results of Figure 5, we present in

Table 2. Summary of fatigue life extension.

Configuration	Retardation Cycles	Number of Cycles to Failure	Extension in Fatigue Life with Respect to CAL	Extension in Fatigue Life with Respect to CAL (%)
CAL	-	50,140	-	-
CAL+P	33,960	185,480	3.70	269.92
CAL+OL	209,40	79,360	1.58	58.28
CAL+P+OL	141,820	372,700	7.43	643.32
CAL+OL+P	2,720,380	3,292,460	65.67	6466.53

the details of the number of cycles to failure and the number of retardation cycles for each loading scenario compared with the reference case (CAL). We can see in Table 2 that the fatigue life is increased by 1.50 times when applying an overload of 14 kN (14 MPa) at a crack length of 3 mm. This increase is 7.43 times when the patch is bonded just before the overload. The improvement of the fatigue life rises to 65.67 times when the patch is bonded after the overload application.

We derived from the lifetime curves the values of the crack growth rate (da/dN), and we plotted the generated curves as a function of the crack length for different scenarios of the overload application. Figure 6 presents the variation of the FCG rate (da/dN) as a function of the crack length for the following cases: constant amplitude loading (CAL), overload without repair (CAL+OL), and overload before patch bonding (CAL+OL+P). It can clearly be seen that the crack growth rate (da/dN) is greater for constant amplitude loading (CAL) without the presence of overload or patch. We also noted that at the time of the application of the overload, the crack propagation rate drops for the cases of overload without patch (CAL+OL) and overload before patching (CAL+P+OL), and the decrease of da/dN is more significant in the case combining the patching and overload. However, once the overload is applied, we notice that the crack growth rate is more significant for the case for an overload after patching (CAL+P+OL) compared to an overload without patch (CAL+OL). These results explain the weak difference of the fatigue life between those two cases. This slight fatigue life difference can be explained by the bridging effect that the patch makes against the overload causing a smaller plastic region in front of the crack tip compared to the unpatched cases [28]. This reduces the effect of the overload on the crack growth retardation and hence on the fatigue life. It is also visible in Figure 6 that there is a brief acceleration of the crack after the application of the overload, which precedes the retardation in both cases, CAL+OL and CAL+P+OL, confirming the finding of Bichler and Rippan [11].

Figure 7 presents the variation of the FCG rate as a function of the crack length for the following cases: CAL, CAL+OL, and CAL+OL+P. It is clear that bonding the patch after the application of the overload leads to a considerable drop in the speed of the crack propagation. The FCG rate (da/dN) decreases from 1 × 10⁻⁷ m/cycle to 5 × 10⁻¹⁰ m/cycle, giving a reduction rate of 1400 times. This reduction in the crack propagation rate causes a near-stop of the crack for about 2.7 × 10⁶ cycles. It can be confirmed that for this case (overload before the patch bonding), there is a positive combined effect between the two retardation sources (patch and overload). Indeed, the application of the overload causes a fairly significant plastic area, leading to a closure of the crack lips and a blunting of its front inducing a crack growth retardation. Besides, the bonding of the patch after the application of the overload reduces the stresses around the crack front by bridging a part of the fatigue loading causing those stresses through the composite patch and the adhesive layer. The sequence of the two retardation effects lead to an arrest of the propagation of the cracks during a very large number of cycles, which gives a practically infinite fatigue life. We can conclude that the study of the overload in conjunction with a bonded composite repair revealed quite complex phenomena that need to be better explained. For this reason, we opted for fractographic analysis to explain the observed behaviors of crack propagation dependency on the sequence of patching/overloading.

3.2. Fractographic Analysis

In this section, the analysis of the fracture surface of failed specimens using SEM observations is presented for the three cases: overload without repair (CAL+OL), overload after repair (CAL+P+OL), and overload before repair (CAL+OL+P).

3.2.1. Overload without Repair

Figure 8a shows the fracture surface of the unrepaired specimen failed under CAL+OL. Multiple crack initiation sites can be observed as marked in the Figure 8. This is because the material is less ductile than other aluminum alloys such 2024 T3. Moreover, the crack front at the point of application was found to be straight. Figure 8b differentiates the obvious region between the notch and the overload zone. It can be observed that there is an evident mark at the point of application of overload due to the severe plastic damage. It is interesting to note that the plastic wake zone of about 150 μm was induced due to the application of overload.

In order to better relate the fractography of CGR due to overload, we focused the images on the vicinity of overload. Figure 9a depicts the overload region at a higher magnification. It can be observed that the overload zone comprises of several shallow dimples and fine striations, indicating mixed type of fracture (ductile + brittle). However, a large number of shallow dimples indicate the plastic deformation caused by the overload and dominant ductile fracture. Conversely, in Figure 9b (taken at 9 mm from the point of application of overload), mixed fracture with dominant brittle failure can be observed. This is similar to the typical failure of brittle aluminum alloy, which indicates the diminishing effect of overload. Furthermore, several fine striations were observed which indicate the signature fatigue failure. Figure 9c presents the final fracture region of the unrepaired sample. A typical brittle (cleavage) can be observed in this region. After the critical length, the specimen failed abruptly, leading to a brittle type of fracture.

3.2.2. Overload after Repair

Figure 10a presents a fractograph of failed specimen in the region of overload application. The patch was also bonded in the same location just before the overload. In this figure, there was no clear indication of the overload due to the stress transfer by the patch. This confirms that the bonding of patch attenuates the effect of overload in the alloy. The presence of the patch reduces the stress around the crack front and consequently the plasticity due to the overload application. The advantage of increase in fatigue life due to retardation effect caused by overload was impaired by the presence of the patch. This complements the explanation of the low performance of patch repair subjected to a significant overload (Figure 5).

Figure 10b presents the fracture surface in the vicinity of overload. In contrast to the type of fracture observed in CAL+OL, brittle failure features were observed in overload after repair configuration. This is again due to the reduction of overload effect caused by the presence of the patch. Moreover, typical fatigue striations and several microcracks were observed in the overload region, indicating the fatigue cycles due to retardation and damage due to overload, respectively. In addition, compressive features are also observed and explained by the auxiliary bending moment created by a single-sided composite patch. The compressive features were mostly observed on the free surface rather than the patched surface due to the fact that the inner curvature experiences compression. Indeed, when a single-sided composite patch is bonded to a plate, the neutral axis of this plate is shifted, and the application of axial load creates a bending moment that can reduce considerably the repair efficiency.

Figure 10c presents the fracture surface away from the overload zone (11 mm from the overload zone). At this position, the fracture is found to be mainly brittle, but some dimples were observed, indicating a mixed type of fracture (brittle + ductile). The crack growth led to the increase of the plastic strain around the crack region. In Figure 10d, we present the fractography from the final failure region. A clear dominance of the brittle fast fracture was observed though some dimples were present. However, we also observed some micro-cracks, which can be explained by the high strength of the aluminum alloy 7075-T6.

3.2.3. Overload before Repair

In this section, we analyze the fracture surface of failed specimen subjected to overload before repairing with a composite patch. This loading case gave practically very high fatigue life. The fractographic analysis can explain some complex behaviors of the fatigue crack growth in this case. Figure 11a presents a SEM image of fracture surface of failed specimen near the overload zone with low magnification (×25). We can clearly see the damaged area created by the overload application to be wider in the vicinity of the patched edge [28,29]. This means that the patch bonding after its application does not act negatively on the retardation effect due to the overload. Contrarily, there was a strong cumulative effect between the patch and the overload leading to an almost infinite fatigue life.

Figure 11b shows a SEM image of the fracture surface at 16 mm of the notch (13 mm from the overload zone) near the free edge of the repaired specimen (unpatched edge). However, near the patched edge and again at 16 mm from the notch (Figure 11b), we can see fatigue striations with relatively significant width, which means that the fracture is mixed with brittle domination.

Figure 11c,d present the fracture surface. The failure in the two cases is a mixed type (ductile + brittle), and compressive features are observed and explained by the auxiliary bending moment visible in Figure 11c,d. The summary of the failure mechanisms depending on the test configurations is presented in Figure 12. The individual and combined effect of patching and overload can be clearly distinguished from the fractographic analysis.

Finally, the obtained results showed that the number of cycles to failure is exceptionally high in the case where the overload is applied before the patch bonding (OL+P) compared to the opposite case (P+OL). In the former case (OL+P), a failure cycle number N_f exceeded three million cycles, which is attributed to difference between the failure mechanisms due to the time of application of overload between the two cases that determines the fatigue life of the repaired plate. Indeed, when the overload is applied after the patch bonding (P+OL), its effect is largely attenuated by the bonding of the patch, which absorbs part of the stresses around the crack front. The plastic zone around this front caused by the overload in this case is less consistent, and the retardation effect due to the overload is therefore reduced, leading to secondary cracks and more ductile failure. When the patch is bonded after the overload (OL+P), the effect of this overload will be significantly large/maximum. The plastic zone around the crack front will be consistent, and the retardation of the crack propagation due to the overload will be significant. In addition to the effect of the overload, the patch bonding will further delay the crack propagation leading to higher fatigue life (see Figure 5) by more stress absorption and abrupt brittle failure. The overload before patching also causes auxiliary bending moment due to the single-side patch, which exhibited both shear and compressive fracture features.

4. Conclusions

It has been shown in this study that the efficiency of a composite patch repair of cracked Al 7075-T6 plates is greatly affected when the fatigue loading is subjected to a tensile overload peak. The sequence of the overload application has considerable influence on the performance of the repair. If the overload is applied after bonding the patch, the overload not only causes a limited crack growth retardation but also damages the patch repair, reducing its bridging capacity and hence its efficiency. Contrarily, when the overload is applied before bonding the patch, both actions (patching and overloading) complement each other to generate a significantly increased crack retardation effect such that the repaired plate has an almost infinite fatigue life. The fractographic analysis showed that the application of the overload before patch bonding led to extreme failure conditions for the repaired plate. High shear stresses caused by the auxiliary bending moment due to the unsymmetrical single-sided patching exhibited both shear and compressive fracture features.