Fatigue fracture behaviour and thermographic analysis of friction stir-welded AZ31

Introduction

Recently magnesium alloy has been applied in aerospace and automobile industries as their weight reduction and high specific properties. ¹ Therefore, welding process is urgently required to realise their application, especially for fatigue fracture behaviour under cyclic loading. However, when magnesium alloy is welded by the conventional fusion weld, it is difficult to get excellent mechanical and metallurgical properties of joints. ² Friction stir welded (FSW) that a solid joining method can effectively avoid the fusion welding defects (e.g. low distortion of the workpiece, minor loss in parent material properties and the lack of porosity) ³ and prove to be an enabling welding technique for joining Mg alloys.^4,5

AZ31 magnesium has been widely investigated on FSW, Previous studies have established the influence of micro and macro features on the joint strength, including shear bands, ¹ texture distribution,^6–8 material flow,^9,10 residual stress ¹¹ and mechanical properties.^5,12,13 The welding quality was generally perfect in the FSW joints.

Fatigue fracture is a major failure mode in metal structures and failure accidents, about 70–90% involving welded structures are caused by fatigue of welding joints. Welded structures are used in most buildings and construction projects, and fatigue damage accidents often result in catastrophe, which lead to lose of lives and properties. Recently, some studies have evaluated the fatigue behaviour of AZ31 Mg alloy FSW joints, including the behaviour of strain-controlled low-cycle fatigue, ¹⁴ non-uniform deformation ¹⁵ and S–N curves.^16,17 An extensive literature exists on the fatigue behaviour of FSW joints in Mg alloys; however, there are less attention about relationship between the fatigue fracture and cyclic deformation behaviour of FSW joints.

It is well known that most of the dissipated strain energy is converted into heat, which manifests that it is in the form of a change of temperature during fatigue test.^18,19 When material is subjected to cyclic loading, the temperature on the surface is varied, and when the heat transfer to surrounding environment equals with the value of heat generation in the loaded specimen, ²⁰ the temperature will no longer rise. Based on this phenomenon, the thermographic technique has been applied to detect the occurrence of fatigue damage. ²¹ The infrared thermographic is a nondestructive method that allows measuring the surface temperature of a specimen by means of an infrared thermal scanner during the tests. Yang et al. ²¹ used a high-speed and high-sensitivity infrared imaging system and observed five stages in the temperature evolution of reactor pressure vessel steels by cyclic loading. They believed that the effect which contains thermoelastic, inelastic and heat conduction could be used to explain and predict the temperature variation during the fatigue process.

In our previous work, the fatigue fracture behaviour of AZ31B magnesium alloy and its welding joints were investigated by using infrared thermographic method, including fatigue life prediction, ²² analysis of fatigue behaviour, ²³ mechanism of heat generation ²⁴ and rapid determination for fatigue parameters (fatigue limit, residual fatigue life, S–N curve). ²⁵ However, according to the authors’ knowledge, there is little study on the fatigue fracture of FSW AZ31 by infrared thermographic which was not presented in those works. Therefore, the aim of this study was to understand the fatigue fracture behaviour of FSW AZ31 and cyclic deformation behaviour of this FSW joints by infrared thermographic.

Experiment details

Five millimetre thick AZ31-H24 plate was used in this study. Before FSW, the plates were obtained by extruding at 490 K and then annealed at 573 K for 2 h to release the residual stress generated from the extruding. Its nominal chemical composition and mechanical properties are given in Tables 1 and 2, respectively. All plates were FSW along the rolling direction. The welding tool was fabricated from tool steel and consisted of a concave shoulder 20 mm in diameter and a cylindrical threaded pin of 5 mm in diameter with a pin length of 4.9 mm. Both the pin and the shoulder of the tool have smooth cylindrical shapes. Figure 1 shows a schematic illustration of the FSW process. FSW was conducted at a welding speed of 300 mm min⁻¹ and a rotation rate of 1200 rpm, with a tool tilt angle of 2.7°. Figure 2 shows the macrostructure of FSW joints. The specimens for the fatigue tests were sectioned perpendicular to the welding direction. The advancing side (AS) where the travelling and rotating directions of the tool are the same and the retreating side (RS) where the travelling and rotating directions of the tool are the opposite, which is written as the AS and RS, respectively, throughout this paper.

OPEN IN VIEWER

1

Schematic illustration of FSW process

OPEN IN VIEWER

2

Macrostructure of FSW joint

OPEN IN VIEWER

Table 1

 The chemical compositions (wt-%) of AZ31 magnesium alloy

Al	Mn	Zn	Ca	Si	Cu	Ni	Fe	Mg
3.02	0.63	1.10	0.04	0.10	0.01	0.001	0.005	Bal.

OPEN IN VIEWER

Table 2

The mechanical properties of AZ31 magnesium alloy

Tensile stress Rm/MPa	Yield stress R0.2/MPa	Elongation δ/%	Shrinkage on cross-section Z/%
242.3	140.3	19.06	14.80

The welded joint specimens were intercepted, and the shapes and dimensions are all shown in Fig. 3. The specimens were prepared by mechanical milling and polishing, thereafter, all the surfaces of specimens were polished with grinding papers up to grit 2000 so as to remove the roughness by wire-electrode cutting and sample surface processing defects, and have a consistent and smooth surface. Before the fatigue testing, the specimen was covered with a thin opaque black paint layer in order to increase the thermal emissivity of the specimen surface.

OPEN IN VIEWER

3

The welded joint specimen

In this work, Fatigue tests were conducted by using a PLG-200D high frequency fatigue testing machine with a stress ratio of 0.1. The fatigue frequency of the magnesium alloy specimens varied from 105 to 108 Hz. The microstructure of FSW joints were examined using optical microscopy (OM). Under the condition of same stress, the strain at the AS of one specimen and that at the RS of another specimen were measured separately using an extensometer with a gage length of 25 mm.

During fatigue process, thermography detection was performed by using a Vario CAM hr camera with a 320 × 240 focal plane array recording the surface temperature in the ambient temperature at an infrared camera speed of 50 Hz. The temperature sensitivity was better than 0.08 K at 303.13 K. The IRBIS^®3 software was used to provide a series of 384 × 288 matrix of the 2D temperature data as a function of time and space on each thermography, which helps us thoroughly examine the entire surface of the specimen at each point and the temperature variation over time.

Results and discussion

Fatigue strength

S–N curves presented in Fig. 4 showed fatigue property of friction stir butt welded AZ31 magnesium alloy joint in high-cycle fatigue test. All data were expressed in bi-logarithm scale by fitting the data using the least-square method and the regression line in Fig. 4 represents the S–N curve at 50% survival probability. The FSW joint showed a rising fatigue life with decreasing the maximum stress. On the S–N curve, the FSW joints showed a shorter fatigue life at the maximum stress of 60–150 MPa and exhibited a fatigue strength of 50.8 MPa which can be considered as endurance limit (based on 10⁷ cycles). All the specimens tested at the maximum stress higher than 50.8 MPa were broken before they reached 10⁷ cycles. The specimens tested at the maximum stress bellow 50.8 MPa exceeded 10⁷ cycles and were considered as run-outs. The point corresponding to the run-out specimens were denoted by arrows in Fig. 4. The S–N curve of FSW joints decreases continuously with increasing number of cycles in the broad interval from 5 × 10³ to 10⁷ cycles and can be well approximated by the equation. (1) where N is a number of cycles to failure, is maximum stress. Equation (1) is based only on the data corresponding to the broken specimens.

OPEN IN VIEWER

4

S–N curves for the fatigue performance of FSW AZ31 magnesium alloy joint for 5.0 × 10³ – 1.0 × 10⁷ cycles

Macroscopic analysis of AZ31 FSW joints fatigue

Figure 5 shows macro-image of the joints that experienced fatigue deformation. At stress amplitudes, because of the different frictional heating on two sides of a single specimen, inducing that fatigue specimens fractured along the Heat-affected zone (HAZ) of AS, eventually. The same phenomenon was also reported for an Al–Mg–Sc alloy in which fatigue cracks were likely to initiate at the AS of the stir zone. ²⁶

OPEN IN VIEWER

5

Macro-image of FSW AZ31 joint after fatigue deformation

Microstructures of FSW joints

In the cross-sections of Mg alloys FSW joints, four microstructural zones could be identified: nugget zone (NZ), thermo-mechanically affected zone (TMAZ), HAZ and BM, which are shown in Fig. 1. The different microstructures can be found in Fig. 6. It should note that four regions are distinguished based entirely on the grain size. The average grain size of BM was determined to be 130 µm by a linear intercept method. The average grain sizes of HAZ at AS, HAZ at RS, TMAZ at AS, TMAZ at RS and NZ were 89.9, 71.5, 31.99, 31.15 and 16.03 µm, respectively. At a higher resolution, the microstructure of BM mainly consists of twins or deformed grains, as shown in Fig. 6a. This grain structure and size of the BM may originate from plastic deformation during the warm rolling process. ⁶ The HAZ microstructure lay in between those of TMAZ and BM. In comparison with TMAZ, the HAZ (Fig. 6b and c) contained a greater fraction of coarse grains and a small amount of twins due to the dominating effect of frictional heating, which resulted in a lower value of yield strength (YS). In the TMAZ (Fig. 6d and e), less plastic deformation and elevated temperature resulted in dynamic recrystallisation, and the microstructure is characterised by fine equiaxed and recrystallised grains. However, relative to NZ (Fig. 6f), the grains in the TMAZ were larger in size, indicating that grains may originate from incomplete dynamic recrystallisation during the FSW. The primary reason for the region suffers from insufficient mechanically stirring and frictional heat by the interaction of pin with base material compared to the stir zone. The same phenomenon was also reported for an 7050-T7451 aluminium alloy in which fatigue cracks were likely to initiate at the AS of the stir zone. ²⁷ Microstructure analysis showed that a recrystallised fine-grained microstructure was generated in NZ after FSW. In the NZ middle, the basal slip is less easily activated. Accordingly, dynamic recrystallisation process accompanying plastic flow, which will increase the density of grain boundaries. Furthermore, it is considered that a high density of grain boundaries can act as an energy barrier that hinders crack initiation and propagation of the crack.

OPEN IN VIEWER

6

Optical microstructures of various zones on the cross-section of FSW AZ31 joint: a BM, b HAZ at AS, c HAZ at RS, d TMAZ at AS, e TMAZ at RS, f NZ

As described above these regions experience different degree of deformation and temperature and this gives rise to the observed variation in YS due to variation in material flow pattern around the tool. The frictional heating is mainly contained the interacting of the shoulder and workpiece, pin and workpiece. The material under the condition of softening is compelled to relative flow to the reverse welding direction. The schematic of the FSW process is shown as Fig. 1. It can be observed that the different plastic flow of specimens on the two sides leads to different thermal cycle effects. At the AS, rotating directions of the tool had opposite direction with metal flow, while rotating directions of the tool had same direction with metal flow at the RS, resulted in the frictional heating of AS was higher than RS during friction stir welding. Accordingly, the grains of HAZ at the AS were coarser than that HAZ at the RS (89.9 vs. 71.5 µm). Furthermore, it is well known that YS of a material at room temperature is dependent on the grain size according to the following Hall–Petch relationship. ²⁸ (2) where is the YS of the material, d is the grain size, is the ‘friction stress’ representing the overall resistance of the crystal lattice to dislocation movement (or the YS corresponding to a material with infinitely large grain size, which is similar to that of a single crystal), and k is a constant called ‘locking parameter’ reflecting the relative hardening contribution of grain boundaries as obstacles to the slip of dislocations across the grain boundaries.^12,29 Grain boundaries were main obstacle to the slip of dislocation, the microstructure with a smaller grain size would have a higher resistance to localised plastic deformation due to the presence of a greater number of grain boundary area that led to a higher YS. Since a lower value of YS of HAZ at the AS would be expected. That is why the fracture path of specimen was around the HAZ at the AS.

Cyclic stress–strain responses

Hysteresis loops

Figure 7 shows the typical hysteresis loops for the 1st cycle, the 25th cycle, and 2500th cycle at the stress of 110 MPa. Hysteresis loops became smaller in width with increasing total cycles, showing a little cyclic hardening compared to the first-cycle loop. It is seen that the hysteresis loops from the AS were quite different from those from the RS. For the FSW joints during cyclic deformation, hysteresis loops from the AS with the stain amplitude were higher than RS, therefore, plastic deformation which was bigger for AS compared with RS, which corresponded well to the result of lower value of YS at the AS.

OPEN IN VIEWER

7

The typical hysteresis loops for the 1st cycle, the 25th cycle and 2500th cycle at the stress of 110 MPa

Cyclic deformation response

Figure 8 shows variation of the strain with the cycles during cyclic deformation at the stress of 110 MPa for the friction stir-welded AZ31 joints. First, the strain amplitude increased with the increasing number of cycles. Second, the strain amplitudes of RS were lower than AS, which could be attributed to the different of microstructures of FSW joints.

OPEN IN VIEWER

8

The stain with the number of cycles during cyclic deformation: AS and RS

At lower numbers of cycles, plastic strain amplitude of FSW joints increased quickly as cyclic deformation proceeded, indicating appear plastic deformation. RS showed plastic deformation phenomenon at No. of cycles ranging from 1 to 2.5 × 10³ cycles, however, the AS of welded joint showed plastic deformation phenomenon at cycles ranging from 1 to 5 × 10³ cycles, and then the plastic strain amplitudes were nearly constant during the entire cyclic deformation process, indicating that cyclic hardening appeared. Patel et al. ³⁰ considered that the cyclic hardening resulted from the increase of the dislocation density during micro-plastic deformation, as well as twin-dislocation interactions in some larger grains. Actually, the plastic strain amplitudes of AS is relatively higher than RS, which also corresponded well to the result of lower value of YS at the AS, hence, the fatigue fracture path of specimen was around the AS.

Thermographic analysis in friction stir-welded AZ31

Figure 9a is the inferred thermograph of intrinsic dissipation which was produced by the plastic deformation caused difference of temperature on the surface of magnesium alloy. It was shown that temperature of P2 is higher than the point of P1 and P3 (P1 = 33.8°C, P2 = 35.6°C, P3 = 33.6°C). As shown in Fig. 9b (=110 MPa), when cyclic loading is above the fatigue limit, the temperature evolution mainly undergoes five stages: initial increase (stage 1), steep reduction (stage 2), steady state (stage 3), abrupt increase (stage 4), and final drop (stage 5), which are the typical temperature evolution for AZ31 magnesium alloy. The process of fatigue fracture mainly exhibits macroscopic plastic deformation. Stage 1 causes local plastic deformation and yielding phenomenon. The temperature increased with increasing the No. of cycles in this stage. Temperature rises since the energy generated is greater than heat transferred out of specimen, which on the surface of specimen reaches up to the peak soon. Stage 2 presents the steep reduction of temperature due to strain hardening after limited plastic deformation. Subsequently, the temperature decreases until that reaches a relatively constant value. The stage 3 shows a constant value of temperature that is still affected by plastic deformation; however the plastic deformation is smaller than that in stage 1. The temperature maintains stable under the transfer balance between the specimen and surrounding environment. In stage 4, macro-cracks are formed gradually. The temperature increases rapidly for comparatively very small number of cycles. There is a local plastic deformation at the crack tip and the plastic work generated during this deformation is mostly converted to heat. As the crack expands before failure occurs, the deformation area decreases to the area around creak tip, temperature rises comparatively lower than that in stage1. The stage is believed as a warning of an imminent fracture. Then, the specimen's temperature drops down after breaking down occurs, as shown in stage 5. Figure 9b also shows the evolution of temperature with number of cycles of magnesium alloy in different points. Detailed reasons will be discussed.

OPEN IN VIEWER

9

a Infrared thermography on the surface of magnesium alloy specimen in fatigue test and b temperature evolution under load of 110 MPa

Figure 10a shows a thermograph of the fatigue specimen in a fatigue test. The brighter colour represents the higher temperature, and the darker colour represents lower temperature. Under the stress of 110 MPa, Fig. 10b exhibits the distribution of temperature in different stages (1–5) on the surface of FSW AZ31 magnesium alloy joints. It also shows that the temperature is quite different along the line L1. Besides, the temperature of HAZ at AS is higher than other zones from the stage 1 to stage 5 in Fig. 10b.

OPEN IN VIEWER

10

a The distribution of temperature on the surface of FSW AZ31 magnesium alloy joints in a fatigue test and b the distribution of temperature in different stages

According to the principle of thermodynamics, higher-order terms is going to be negative, leading to the following formula for this dependence. ³¹ (3) where is the thermal stress concentration factor, is system volume and is heat capacity under the condition of constant strain.

Strain tensor consists of spheric tensor and deviation tensor in genera. ³² (4) where , where the is volumetric strain, i, j ϵ {1,2,3}.

In view of equations (3) and (4), we find that is a correlation relationship between volumetric strain and temperature changes, and just consider the volumetric strain ³³ : (5) where , ɑ is constant for a given object.

That means that volumetric strain in proportion to temperature changes. Namely, volumetric strain in proportion to infrared radiation changes theoretically. From what has been mentioned above we can come to the conclusion that a lower value of YS at the AS, so its capability of resisting deform is poor. Under the condition of the same loading, large plastic deformation for AS of FSW joints compared with the RS during fatigue processes, which led to the temperature of HAZ at AS is higher than other zones. The distance of the fracture region to the NZ of welded joints corresponds to the position with the peak of temperature. These results further explain the fracture path of specimen was around the HAZ at the AS.

Conclusions

The present study on the fatigue fracture behaviour and thermographic analysis of FSW AZ31 was performed. The following conclusions are presented:

FSW AZ31 joints exhibited a shorter fatigue life at the maximum stress of 60–150 MPa and exhibited a fatigue strength of 50.8 MPa can be considered as the endurance limit (based on 10⁷ cycles).