Fatigue Behavior of a Dissimilar Aluminum Alloy Welding Joint With and Without Natural Defect

SHI Liang, WANG Chiquan, LIU Zhiyi, WANG Juntao, WANG Wukun

(1. China Aero-Polytechnology Establishment, Beijing 100028, China; 2. Kunming Precision Machinery Research Institute, Kunming 650118, China)

Abstract: In order to investigate the influence of natural defect on the fatigue behavior of 5A06/7A05 dissimilar aluminum alloys welding joint, fatigue tests of two types of specimens with and without defects were carried out systematically under stress amplitude control conditions (stress ratio R=0.1) at normal temperature in laboratory air condition. Furthermore, a new parameter, i e, fatigue defect effect factor (FDEF) was introduced to assess the effect of defect on fatigue strength. The fatigue failure analysis was conducted as well to compare the fatigue and fracture behavior of the two types of specimens. The results show that: (1) natural defects have a strong effect on the fatigue lives of welding joint, and the differences between the specimens with and without defects can reach 80 times under a same theoretical net sectional stress; (2) the FDEF parameter introduced is effective to deal with the defect effect, and the FDEF decreases along with the increase of fatigue life. The mean relative error between the experimental data and predicted fatigue strength based on the FDEF is 10.2%; (3) the macro fracture of both types of specimens have three typical zones, i e, fatigue source zone, crack propagation zone and final fracture zone, while there are more than one fatigue sources for specimens with natural defects.The overall pattern of crack propagation zone and fracture zone are quite similar, but the morphologies are different in details.

Key words: aluminum alloys; welding joint; natural defect; fatigue life; scanning electron microscope(SEM)

1 Introduction

Aluminum alloys are widely used in industry such as rail transit, vehicles, ships, building structures and aerospace due to their high comprehensive mechanical properties[1-3]. Among various aluminum alloys, 5xxx series (Al-Mg) aluminum alloy and 7xxx series (Al-Zn-Mg) aluminum alloy are two widely used materials.5xxx series aluminum alloys have the advantages of medium strength, corrosion resistance, good processability and welding performance[4], which is usually used in components such as skin, frame and beam.While 7xxx series high strength aluminum alloys are widely used as main bearing structures in industry[5].Since these different series of aluminum alloys have quite different mechanical properties, a combined use of them to meet different requirements is common in applications such as lighter vehicles, high speed ships,etc.[6-8]One challenge of using different material in one structure is to deal with the connection problem,and welding is one of the most widely used methods to join different materials. Widely used welding methods for aluminum alloys are as follows: argon arc welding including Tungsten Inert Gas (TIG)[9,10]and Melt Inert Gas (MIG)[11], Friction-Stir Welding (FSW)[12],etc. There are many factors influencing the quality and strength of welds such as energy input, base material,filler material, the design of the joint,etc, thus destructive or nondestructive testing are usually essential to guarantee the quality of welds.

Common types of welding defects include porosity, lack of penetration, undercut, incomplete fusion,etc. It has been proved that the presence of defects has normally a negative effect on the mechanical properties of welding joint. Wuet al[13]characterized the interaction between the pores and the fatigue damage in 7075-T6 joints. They found that the gas porosity inside 7075-T6 hybrid welds plays a significant role in the fatigue crack nucleation and propagation process.Similar conclusions were drawn by Liuet al[14]when carrying out the fatigue test of 6005A cold metal transfer welded joint. Although the defects may affect the failure of welding joint in different manners, a primary agreement has been reached that the fatigue resistance of welded aluminum alloy is quite related to the number, size and morphology of defects. Gauret al[15]found that pores of sizes comparable to or greater than mean grain sizes were often the cause of fatigue crack initiation. Chenet al[16]investigated the typical defects for 5 456 aluminum alloy using friction stir welding.They analyzed the outer factors on the typical defects and mechanical properties of the joint and found that the oxide layer,i e, Al2O3particles from the initial butt surface during FSW is dispersed at the grain boundary which are actually the major cause of failure of the joint. Kahet al[17]investigated the weld defect in both friction-stir welding and fusion welding of aluminum alloys. They found that there are two types of porosity in laser welding of aluminum alloys: metallurgical porosity and keyhole porosity. Keyhole instability is the main cause of bubble initiation especially in deep penetration welding. Since the defects have a detrimental effect on the fatigue life of welding joint, researchers have tried to analyze these effects quantitatively.Buffièreet al[18]identified a relationship between the defect size and the sphericity of defect. They found that gas pores are usually smaller and with a higher sphericity ratio with respect to natural shrinkages. Nicolettoet al[19]found that the pore morphology can influence the stress distribution by finite elements simulation and X-ray computed tomography, and the fatigue resistance reduction associated with the presence of defects is however strongly dependent on their size, the reduction being higher for a larger defect. Bracquartet al[20]investigated the influence of the defect size on the high cycle fatigue behavior of polycrystalline aluminum with different grain sizes experimentally. They found that fatigue crack initiation from a defect is found to be strongly impacted by the crystallographic orientation of the surrounding grain, and crack initiation preferably occurs in crystals being favorably oriented for plastic slip. Other materials with defects of different sizes and shapes such as brass[21]and pure iron[22]have also been studied, in which it is observed that when a defect is introduced, the decrease in fatigue limit compared to the corresponding smooth specimen was lower for a higher grain size. Due to the fact that defects affect the mechanical property, many studies have focused on preventing or reducing the welding defect. Kahet al[17]found that keyhole instability can be reduced by using effective welding parameters and vacuum conditions, while metallurgical porosity can be reduced by increasing weld speed, which results in insufficient time for hydrogen to accumulate due to rapid cooling and solidification. Praveenet al[23]found that choosing the right welding wire and reducing the heat input of welding parts can reduce the formation rate of cracks.It is found that the heat input required for pulsed MIG welding is low, which can effectively reduce the formation of hot cracks, but cannot prevent the formation of pores. Garciaet al[24,25]investigated the MIG welding of TiCp/1010Al and SiCp/A359, they found that when 2024Al aluminum alloy is used as welding wire, the metallurgical bonding between Al and TiC and SiC can be effectively slowed down by indirect arc welding.At the same time, the decomposition of reinforcement phase can be restrained and the generation of pore can be reduced. It can be concluded that the formation of defects in weld is usually complex, and a systematical study on the effect and formation of welding defect is essential in modern industry.

In the present work, an experimental study was carried out to investigate the influence of natural defect on the fatigue behavior of dissimilar aluminum alloys 5A06/7A05 welding joint. This study concentrated on fatigue lives in the presence of defects. Indeed, the influence of the defect is most important in the fatigue life reduction, as it acts as a stress concentrator and a preferential crack initiation site. Its influence becomes lower once a crack is initiated and starts to propagate.Furthermore, to better understand the factors governing the initiation of a fatigue crack, scanning electron microscope (SEM) techniques were used to evaluate the mechanism of failure with the comparative method.

2 Experimental

2.1 Description of material and specimens

The materials used in this study were 7 mm thick 5A06 and 7A05 aluminum alloys in O and T6 conditions respectively. The material and conditions chosen for this study are currently adopted in a sandwich cylindrical shell subjected to external load[26]. The sandwich cylindrical shell consists of outer shell and inner shell which is made of 5A06-O aluminum alloy and 7A05-T6 aluminum alloy, respectively. Confined space is pro-duced between the two shells by introducing supporting ribs to obtain a good bulking property, and the outer and inner shells are connected by welding. The tensile strength of 5A06-O and 7A05-T6 is 315 and 515 MPa respectively. The nominal composition and mechanical properties for each material are listed in Table 1.

Table 1 Chemical composition of 7A05-T6 and 5A06-O base metal/wt%

Fig.1 Schematic of the fatigue specimen: (a) dimension; (b) sample

The dissimilar welding joint of 5A06/7A05 was produced by Tungsten Inert Gas (TIG) welding which has the advantage of high thermal power, energy concentration and good protection effect, and the aluminum alloy wire ER5356 with a diameter of 1.6 mm was used as the welding material. In order to select the optimal welding parameters a set of TIG joints were produced by adjusting the vertical force, the tool travel speed and the tool rotation speed in the range of 4000-5500 N, 50-120 mm/min and 800-1000 rpm, respectively.The protective gas used high-purity argon with a purity of 99.999% and a gas flow rate of 24 L/min. Welding current, welding voltage and welding speed were 260 A,25 V and about 200 mm/min, respectively. More details of the welding process can be found in Reference [26].

It has been indicated that welding defects have a strong effect on the fatigue performance of welding joint as well as the real structure[13-25]. In order to investigate the effect of defects on the fatigue property of welding joint, the primary task is to obtain the defect information. Nondestructive testing (NDT) was carried out after the welding process finished, and the phased array ultrasonic testing (PAUT) method based on the standard ASTM E2700-09[27]was used for the inspection of the welding joint. According to the relevant regulations of ASTM E2700-09 standard, a PAUT instrument Omniscan-mx2 produced by Olympus Corporation was adopted for the inspection. Based on the NDT results, two types of specimen were designed and processed, one type was welding joint specimen without defect named as WJ specimen, and the second type was welding joint specimen with natural defect named as WJND specimen. Both WJ and WJND specimens were prepared having a width of 10 mm and a thickness of 5 mm in the test section as shown in Fig.1(a).Specimens were cut perpendicular to the welding direction from the sandwich cylindrical shell as shown in Fig.1(b).

2.2 Experimental procedure

Fatigue tests of welding joint specimens were conducted using a 100 kN servo-hydraulic system (MTS 370.10) according to the ASTM E466 standard[28]. All fatigue specimens were loaded cyclically under different stress amplitude ranges. The loading was constant amplitude, and the stress ratio was set to 0.1. The load frequency is 10 Hz. Lines were drawn at the appropriate symmetrical positions at both ends of the specimen before the test to facilitate the symmetrical clamping of the test piece. Each sample was properly preserved and fracture morphology was protected after the test completed. At least three fatigue specimens were tested at identical stress amplitude to ensure the accuracy of the fatigue data.

3 Results and discussion

3.1 Fatigue life results of specimens with and without defect

The fatigue life testing results of WJ specimens are shown in Table 2, and 13 specimens under three levels of stress were tested. The peak load, load radio,net area, theoretical net sectional stressσnet, and fatigue lifeNfare listed. Based on the test results anS-Ncurve is fitted by use of the power function which can be written as:

wheremandCare fitting parameters,SandNare maximum stress and fatigue life. TheS-Ncurve as well as the test data is shown in Fig.2, in which the stress is equal to the maximum theoretical net sectional stress, and the fitting parameters arem=4.63 andC=3.16×1015, respectively.

Table 2 Fatigue life testing results of welding joint without defect

The fatigue life testing results of welding joint with natural defects are shown in Table 3, and 16 specimens with natural defects under three levels of stress were tested. The parameters listed in Table 3 are the same as those in Table 2. The test data is shown in Fig.3. It can be seen from Fig.3 that the fatigue lives varies greatly under a certain load level due to the existence of natural defects. For example, the fatigue lives of WJND specimens varies from 2 480 to 199 000 under the same theoretical net sectional stressσnetof 120 MPa, whose difference reaches 80 times. These results demonstrate that the natural defects have a strong effect on the fatigue life of welding joint, which will be analyzed in details in next section.

Table 3 Fatigue life testing results of welding joint with natural defect

Fig.3 S-N curve of WJND specimens with natural defects

Fig.4 Typical macro morphology of specimens after fracture: (a)WJ specimen; (b) WJND specimen

The typical macro morphology of both WJ and WJND specimens after fracture is shown in Fig.4. It can be seen that the fracture plane of WJ specimens is clean and smooth, while the WJND specimens have obvious natural defects in the fracture zone, which belongs to incomplete fusion. The incomplete fusion belongs to a two-dimensional defect due to lack of union between weld metal and parental metal.

3.2 Fatigue strength assessment

It has been indicated that the natural defect (i e, incomplete fusion) has a detrimental effect on the fatigue life of welding joint. In order to quantify its effect, the fatigue strength assessment was carried out and a new parameter is introduced to deal with the effect of defect. Since the incomplete fusion defect area cannot bear the external load due to the physical nature of lack of union, the effective stress can be introduced based on the continuum damage mechanics (CDM).The notion of continuum damage mechanics was proposed first by Kachanov[29]when he tried to predict the brittle creep rupture time of metals under constant tension. With the development of CDM, researchers have proposed different variables such as modulus of elasticity to describe the damage and then describe the mechanical behavior of damaged material and structure. The mechanical behavior of a damaged material is usually described by using the notion of the effective stress, and a concept of effective stress in the case of uniaxial tension is written as[29]:

whereσeis the effective stress related to the undamaged area of the section,Fis the load,A0is the initial area of the undamaged section, andAis the lost area due to damage. The value can be interpreted as the effective area of the section. This concept is based on the assumption that the rate of damage growth is determined primarily by the level of the effective stress. According to Formula (2), the effective stress is calculated as shown in Table 4, in which the defect area (i e, damaged area) is measured by image processing technique based on the fractography of each WJND specimen.It can be seen from Table 4 that the effective stress is quite larger than the theoretical net sectional stress. TheS-Ncurve of the welding joint with natural defect as well as the testing data is shown in Fig.5, in which the stress equals to the effective stress for the WJND specimens, and the fitting parametersmandCequals 6.99 and 2.92×1019, respectively.

Table 4 The defect area and effective stress of WJND specimens

Fig.5 S-N curve of the WJND as well as WJ specimens

It can be seen from Fig.5 that the fatigue lives of the WJND specimens reduce a lot compared with the WJ specimens at the same stress level. In fact, the existence of incomplete fusion defects not only will reduce the bearing area but also lead to a severe stress concentration near the edge of defect, which may accelerate the crack initiation. The stress concentration induced by the defect is quite similar to that caused by geometrical mutation such as hole, fillet,etc. The reduction of fatigue life induced by the stress concentration is characterized by the fatigue notch factorKf[30], whose definition is the ration of the fatigue strength of a smooth specimens and the fatigue strength of a notched one under the same experimental conditions and the same number of cycles :

where,σsis the fatigue strength of smooth specimens,andσnis the fatigue strength of notched specimens. The fatigue notch factor is usually determined experimentally, which plays a very important part in the estimation of fatigue life and fatigue strength of structures.

Inspired by the definition of fatigue notch factor,a new parameter fatigue defect effect factor (FDEF)Kd(Nf) is defined as:

whereσtis the fatigue strength of specimens without defects, andσdis the fatigue strength of specimens with defects. It can be seen that the FDEFKd(Nf) is usually greater than 1 since the existence of defect always reduces the fatigue life, and it should be varied with the material, loading mode and defect information. Based on the experimental results, the FDEF of the dissimilar aluminum 5A06/7A05 welding joint with incomplete fusion defect is calculated as shown in Fig.6. It can be seen that the FDEF decreases along with the increase of fatigue life, and the FDEF is approaching to 1 with the fatigue life achieving 107cycles, which demonstrates that when the fatigue life reaches infinite, the fatigue strength is close to equality for specimens with and without defects.

Fig.6 The fatigue defect effect factor (FDEF) of dissimilar aluminum alloy 5A06/7A05 welding joint

Fig.7 Comparison of fatigue strength of specimens between the predictions and experimental results

Once the FDEF is obtained, the fatigue strength of specimens with defects can be predicted by the fatigue strength of specimens without defects, and by using the formulaσd=σt/Kd, the fatigue strength predicted based on the FDEF compared with the experimental results is shown in Fig.7. The mean relative error between the fatigue strength obtained from experimental data and calculated based on the FDEF is 10.2%, which demonstrates that the FDEF is effective to capture the defect influence.

3.3 Fatigue failure analysis

In order to obtain the fatigue failure mechanism and analyze the effect of incomplete fusion defect on the fatigue performance, macro and micro observations of the two types of specimen are conducted, the JSM-6010 LA Scanning Electron Microscope (SEM) is used for the analysis, and typical WJ and WJND specimens are chosen for the observation and comparison.

Fig.8 Macro fracture of the fractured specimens: (a) WJ specimen;(b) WJND spcimen

Fig.8 shows the whole macro fracture of the specimen with and without defect. The two fatigue failure fractures have three typical zones,i e, the fatigue source zone, crack propagation zone and the final fracture zone. Once the fatigue crack initiated, it would propagate continually under the imposed load until fracture. The locations of the fatigue source and crack propagation zone are highlighted by arrows. It can be seen from Fig.8 (a) that the fatigue crack initiated at the surface of the specimen for the WJ specimen, and only one fatigue source is observed. While Fig.8(b) shows that the fatigue crack initiated at multi-sources, and the fatigue crack initiated at the internal of specimen close to the incomplete fusion zone. Generally, the fatigue crack of the welding joint usually initiate at two kinds of specific zones[31],i e,material surface and inner defect of material. The phenomenon that cracks initiated at the material surface is mostly due to the high stress zone located at the surface or sub-surface of specimens.In addition, the machining marks such as cutter marks or extra scratches may exist at the surface of specimen.It is also demonstrated that the surface in a plate stress stage may help the start of the plastic flow, while the plastic flow is one of the main reason inducing crack initiating. However, there is no material that is completely uniform, and discontinuity may exist in the internal zone of material, especially for the welding joint.

Fig.9 SEM images showing the fractograph of crack initiation zone: (a) WJ specimen; (b) WJND specimen

The SEM images of crack initiation zone is shown in Fig.9. It can be seen that a cleavage-like river pattern appearance is evident in the crack initiation zone in Fig.9(a). According to the study of Beachem and Wark[32]this faceted appearance may result from a glide plane de-cohesion mechanism. Fig.9(b) shows that the crack initiation sites located in the inner zone of the specimen where incomplete fusion defect existed, the incomplete fusion defect produced a larger stress concentration than welding pores[33], as the actual loading was significantly larger than the nominal loading. It can be seen that oxide film existed in the incomplete area,and it is proved that the film is consist of an oxide of Al by energy spectrum analysis. Since elements Al and O have strong affinity, if the cleaning of surface is inadequate, it is unavoidable to form a thin oxide film in the welding zone. Although the existence of oxide film may be the main cause of incomplete fusion defect, the cause of incomplete fusion can also be inappropriate current, voltage, travel speed, gas flow rate,etc.[33]The comprehensive cause of the incomplete fusion, which is not the focus of our study, can be investigated deeply in the future. There is no doubt that the existence of incomplete fusion reduces the bearing area of the specimen, which would result in a severe stress concentration near the edge of defect zone.

The fatigue crack propagation zone of WJ and WJND specimens are shown in Fig.10. It can be seen from Fig.10 that both WJ and WJND specimens have obvious fatigue striations in the propagation zone, the river pattern, which is formed by the cleavage steps,can be seen clearly. Its formation mechanism is as follows[34]: when cleavage micro cracks propagate in a grain or propagate through a grain to adjacent grains,they will cause cleavage cracks to break off on different solidification surfaces, and the cleavage step will be formed at the intersection of these cleavage cracks.Therefore, cleavage steps are gradually formed with the increase of the number of cycles. It can be seen that the direction of fatigue crack propagation square is perpendicular to the direction of fatigue strip, which is consistent with the general law of fatigue crack propagation.However, it can be seen from Fig.9(b) that with the crack propagation, this cleavage-like appearance faded away and vanished. Instead, widespread striations and secondary cracks can be seen.

Fig.10 SEM images showing the fractograph of crack propagation zone: (a) WJ specimen; (b) WJND spcimen

Fig.11 SEM images showing the fractograph of fracture zone: (a)WJ specimen; (b) WJND spcimen

Fig.11 shows the final fracture zone of both WJ and WJND specimens. It can be seen that the morphologies of the final fracture zones of the two types of specimens are similar: 1) All final fracture zones contain quasi-cleavage facets and a large number of dimples; 2) The second phase particles, which play an important role in the formation of dimples, can be observed at the bottom of the dimples. However, according to the morphological characteristics of dimples, the morphologies of dimples in the fracture surface of the two types of samples are different. Compared with the WJ specimens, Fig.11 (b) shows that the tear ridges around the dimples of WJND specimens are thinner,and the number and density of the dimples are larger than that of WJ specimens. It is due to the incompleteness of the welding process, the second phase particles are easily formed and aggregated in the welding joint which leads to a large increase in the number and density of dimples in the fracture.

4 Conclusions

The fatigue property of a dissimilar aluminum alloy 5A06/7A05 welding joint was investigated, and specimens with and without natural defects were processed and experimented. A new fatigue defect effect factor (FDEF) parameter was introduced to assess the effect of defect on fatigue strength, and the fatigue failure analysis was conducted to compare the fatigue and fracture behavior of the two types of specimens. From the experimental and analytical investigations, the following conclusions can be drawn:

a) Natural defects (i e, incomplete fusion) have a strong effect on the fatigue lives of 5A06/7A05 dissimilar aluminum alloy welding joint, and the differences between the specimens with and without defects can reach 80 times under a same theoretical net sectional stress. The effect degree of welding defect such as incomplete fusion is related to its geometrical size, the larger the geometric size, the greater the reduction of fatigue life.

b) A new parameter fatigue defect effect factor(FDEF) was introduced inspired by the method dealing with notch effect on the fatigue life, and was obtained by use of the experimental data. The results show that the FDEF decreases along with the increase of fatigue life, and the FDEF is approaching to 1 with the fatigue life achieving 107cycles. The fatigue strength of WJND specimens are predicted by use of FDEF, and the mean relative error is 10.2%, which demonstrates that the FDEF is effective to deal with the effect of natural defect.

c) There are many distinctive differences to compare the welding joint with and without defect on the fracture phenomena: although the fatigue failure fractures of both WJ and WJND specimens have three typical zones,i e, the fatigue source zone, crack propagation zone and the final fracture zone, the location and number of fatigue source is different, and the fatigue crack initiated at the surface of specimens for WJ specimens while at the inner zone near the edge of defects for WJND specimens. In addition, there are more than one fatigue source for specimens with natural defects;both WJ and WJND specimens have obvious fatigue striations in the propagation zone, the river pattern,which is formed by the cleavage steps. The direction of fatigue crack propagation square is perpendicular to the direction of fatigue strip, which is consistent with the general law of fatigue crack propagation; the morphologies of the final fracture zones of the two types of specimens are similar: all final fracture zones contain quasi-cleavage facets and a large number of dimples.But the tear ridges around the dimples of WJND specimens are thinner, and the number and density of the dimples are larger than that of WJ specimens, which may be due to the incompleteness of the welding process.