Mechanical testing equipment

Figure 15: (a) Polishing in the longitudinal direction of the specimen axis and (b) the direction of the abrasive film and the revolving direction of the specimen.

3.8. Mechanical testing equipment

Table 4 Show the mechanical testing equipment detail.

Description Software Conformance

INSTRON - 8800 Controller. Servo hydraulic universal testing machine.

27 | P a g e Figure 16: A SCHENCK/MTS retrofit testing

machine used during HCF testing.

Figure 17: An INSTRON 1342 universal testing machine used for tensile testing.

28 | P a g e 3.9. X-ray diffraction

In order to understand the effect of residual stress on fatigue life of the material under investigation, the residual stress was measured. For the purpose of using a non-contact method the residual stress of each specimen was determined using a PROTOXRD x-ray diffraction machine, see Fig. 19. Although the term stress measurement has come into common usage, stress is an extrinsic property that is not directly measurable. In x-ray diffraction residual stress measurement, the strain in the crystal lattice is measured, and the residual stress producing the strain is calculated, assuming a linear elastic distortion of the crystal lattice. The residual strain of the specimens was measured in the longitudinal direction of the specimen axis. Figure 18 shows the diffraction of a monochromatic beam of x-rays at a high diffraction angle (2θ) from the surface of a stressed sample for two orientations of the sample relative to the x-ray beam.

Figure 18: Show the principle of diffraction of a monochromatic beam of x-rays[21].

29 | P a g e Figure 19: Show the PROTOXRD machined used for the

residual stress measurement.

3.10. Surface roughness

The surface roughness was measured using a MarSurf PS1 portable surface roughness measurement apparatus, see Fig. 20. The roughness was measured in the longitudinal direction over a distance of 8mm – traverse of probe. Both the arithmetic mean roughness (R_a) and the mean peak-to-valley height (R_z) were measured. The calibrated apparatus conformed to DIN EN ISO 4287 : 1998; ISO 4287 : 1997; JIS B 0601 : 2001.

Figure 20: Show the Mahr – MarSurf PS1 Rz measurement.

30 | P a g e 3.11. Fractographic analysis

The fractured surface as well as the polished surface of the specimens were analysed in a JEOL-JSM-6510 scanning electron microscope in order to investigate the effect of the different stress and surface conditions.

Figure 21: The fractographic analysis was conducted in a JEOL – JSM-6510 Scanning Electron Microscope.

31 | P a g e 3.12. Gripping arrangement – high cycle fatigue

In order to optimise the alignment of the load train the grip end section of the specimen was specially designed to fit in a high precision collet chuck, see fig. 22.

Figure 22: Collet chuck gripping arrangement designed to optimise alignment of the load train.

3.13. Alignment and bending strain

The alignment of the load train was determined using a strain gauged specimen, see Fig.

23, installed in the load train during which the axial strain was measured and bending strain calculated.

In accordance with ASTM 1012-99, four strain gauges attached at the mid-length reduced section the specimen and placed at equally spaced positions around the specimen circumference were used during which the specimen is subjected to strain at loads within the elastic range of the material. Adjustment to the load train alignment was made until the maximum bending strain did not exceed the elastic limit of the material.

ASTM E1012-99 Standard Practice for Verification of Specimen Alignment under Tensile Loading: ISBN 92-826-9681-2; EUR 16138 EN.

32 | P a g e Figure 23: Cylindrical specimen with four

gauge sensors was used for the axial alignment measurement of the load train.

“A CODE OF PRACTICE FOR THE MEASUREMENT OF MISALIGNMENT INDUCED BENDING IN UNIAXIAL LOADED TENSION-COMPRESSION TEST PIECES”.

CALCULATIONS: Cylindrical specimens using four gauge sensors.

Axial strain

a = (e1 + e2 + e3 + e4)/4

Local bending strain, b1 = e1 - a

b2 = e2 - a b₃ = e₃ - a

b4 = e4 - a Maximum bending strain,

B 0.5 b b b b Percentage bending,

PB = (B/a ) x 100

33 | P a g e 3.14. Software and data acquisition

CATMAN data acquisition software and a HBM-MGC data logger was used to record strain during the alignment measurement of the load train. The output from the four strain gauges of the alignment specimen and the load cell was recorded independently at a rate of 100Hz. The data was plotted on an x-y scale and the bend stress calculated.

The Instron and MTS mechanical testing software use a control frequency of 10 kHz, providing rapid response to events during tests combined with a high measured-value acquisition-rate. Together with the 24- bit resolution, this enables very precise measurements over the entire measurement range for maximum data accuracy. The synchronous 10 kHz measured-value acquisition-rate delivers precise measurement regardless of number of measurement channels.

3.15. Environmental conditions

The laboratory temperature was regulated at 20⁰C ± 3⁰C during which the temperature and humidity are continuously recorded in accordance with ISO 17025 laboratory quality system requirements.

3.16. Test standards and procedures 3.16.1 Tensile test:

ISO 6892-1-2016 Metallic materials — Tensile testing — Part 1: Method of test at room temperature.

The material tensile properties of AL7449-T651 were confirmed by conducting tensile tests in accordance with ISO 6892-1-2016.

Procedure:

Specimens from the same sample plate and roll direction as the fatigue specimens, were subjected for tensile testing. The test was carried out in strain control at a rate of 0.5%

until 2% strain after which the displacement rate was increased to 2.5mm/min until

34 | P a g e 3.16.2 High cycle fatigue

ASTM E466 – 2015 Standard Practice for Conducting Force Controlled Constant Amplitude Axial Fatigue Tests of Metallic Materials.

Procedure:

Special care was taken during handling of the specimens to measure the diameter and during installation in the collet chuck gripping arrangement to prevent any surface damage prior testing.

The fatigue test was carried out in constant amplitude load control using a sinusoidal wave form at a frequency of 15Hz. In order to ensure optimum load application the proportional, integral and derivative PID algorithm function of the controller was optimized prior testing. Peak – valley monitoring of the load and displacement was applied during testing. The same R-ratio of 0.1 was applied for both stress ranges and each surface roughness condition.

The table below show the parameters for the high cycle fatigue test.

Table 6: Fatigue stress application

35 | P a g e From the axial applied stress the min./max, mean or amplitude stresses were regarded as controlled independent variables.

3.18. Limitations

Due to the round profile of the specimens in the gauge section as well as the measurement application with regards to the residual stress and surface roughness, measurements were limited to a small area of the total circumferential area of the specimen. However the sample of measurement represents the statistical mean value.

Figure 24: Show an approximation of the residual stress and roughness measurement surface area.

36 | P a g e

Chapter 4: Results

4.1 Introduction

The material properties and the data gathered during the specimen preparation, mechanical testing and fractographic processes were analysed and the effect on fatigue life of aluminium alloy was determined. The material composition and mechanical properties were characterized by means of chemical analysis and tensile testing; see tables 7 & 8 respectively. The residual stress and surface roughness results of each specimen were analysed and the effect on fatigue life determined with regards to the applied axial stress during dynamic loading. The fractured surfaces of the fatigue test specimens were analysed in a scanning electron microscope in order to compare the fatigue initiation process between the different surface roughness and residual conditions. The results are discussed separately in more detail.

4.2 ALA7449-T7651 material characterization

37 | P a g e 4.3 Analysis of load train alignment data

The load and strain data captured during the load application on the strain gauged specimen was analysed and the bending strain calculated in order to determine the percentage bending strain the specimen was subjected to. In Fig. 25 the strain vs load data plot is represented. Table 9 show the final bending stress.

Figure 25: Strain vs load plot

Axial strain,

a = (e1 + e2 + e3 + e4)/4

Local bending strain, b1 = e1 - a b2 = e2 - a b3 = e3 - a

b4 = e4 - a Maximum bending strain,

B = 0.5 b − b + b − b Percentage bending,

PB = (B/a) x 100 Maximum bending stress,

σ = E x ε

38 | P a g e Table 9: Bending tensile strain & stress in load train

Axial strain: a_ave (µε) a = 1974.031 Max. bending strain: B (µε) B = 63.370

Bending strain: (%) 3.2

Max. bending stress: MPa 4.43

Class 2

4.4 Analysis of surface roughness measurement

The results of the Ra and Rz roughness measurements, representing the four grit sizes are presented in figures 26 and 27 respectively. The mean roughness indicating a decrease in both the Ra and Rz with an increase in the grit size, however the scatter between the minimum and maximum roughness is significantly lower at the 2500 grit than at the 360 grit. Moreover the linear plots of Ra and Rz versus grit size appear to be similar however the slope of the two plots varies exponentially, see Fig. 28.

4.4.1 R_a Surface roughness measurement

Figure 26: R_a surface roughness.

4.4.2 R_z Surface roughness measurement

Figure 27: R_z surface roughness.

39 | P a g e 4.4.3 Mean roughness R_a & R_z

Figure 28: Mean roughness Ra & Rz

4.5 Analysis of residual surface stress

The mean residual stress results from each grit size are presented in Fig. 29. As expected the compressive residual stress decreased with an increase in grit size.

However the maximum stress was recorded at the 800 grit. The mean peak-valley roughness Rz in Fig. 30 and the arithmetic mean Ra roughness in Fig. 31 show a similar trend at the 800 grit.

4.5.1 Residual stress vs grit-P

Figure 29: Residual stress vs. grit P

40 | P a g e 4.5.2 Residual stress vs roughness- R_z

Figure 30: Residual stress vs roughness-Rz

4.5.3 Residual stress vs roughness- R_a

Figure 31: Residual stress vs roughness-R_a

4.6 Data analysis of fatigue testing results

The fatigue testing results are summarized in table 10 which shows the average values of the diameter, residual stress, surface roughness as well as the fatigue life. The average represents four specimens each with regards to the diameter, grit, residual stress and surface roughness. The two axial applied stress amplitudes represent the average of two specimens each per grit size. A combination of eight different test conditions were investigated which includes four grit sizes and two axial applied stresses. A total of sixteen specimens representing two specimens per condition were tested to failure and analysed. The mean peak-valley roughness R_z and the arithmetic mean R_a roughness are both included for the purpose of the investigation.

41 | P a g e Table 10: Summary of fatigue testing results

Diameter applied versus the fatigue cycles to failure. The power fit curve indicates the increase in fatigue life from 360 MPa to 320 MPa for each of the four surface conditions.

Figure 32: Show the high cycle fatigue test results for both axial stress levels.

42 | P a g e 4.6.1 Residual stress vs fatigue cycles at 360 MPa

Fig. 33 shows the effect of residual stress on fatigue life at 360 MPa applied axial stress.

It is inconclusive whether residual stress does affect fatigue life of the material when subjected to a maximum axial stress of 360 MPa. A similar effect is observed in the plots from surface roughness as well as the grit versus cycles at 360 MPa. However it is important to note that residual stress is directly affected during the process of surface roughness preparation.

Figure 33: Residual stress vs fatigue at 360 MPa

4.6.2 Residual stress vs fatigue cycles at 320 MPa

The fatigue results at 320 MPa show a significant increase in fatigue life for all surface conditions, which confirms the effect of cyclic stress amplitude and asymmetry on fatigue life, see Fig’s. 34 to 40.

Figure 34: Residual stress vs fatigue at 320 MPa

43 | P a g e 4.6.3 Grit P vs fatigue cycles at 320 MPa

Figure 35: Surface roughness grit P vs fatigue at 320 MPa

4.6.4 Grit P vs fatigue cycles at 360 MPa

Figure 36: Surface roughness grit P vs fatigue at 360 MPa

4.6.5 Roughness R_z vs fatigue cycles at 360 MPa

Figure 37: Roughness R_z vs fatigue at 360 MPa

44 | P a g e 4.6.6 Roughness R_z vs fatigue cycles at 320 MPa

Figure 38: Roughness Rz vs fatigue at 320 MPa

4.6.7 Roughness R_a vs fatigue cycles at 360 MPa

Figure 39: Roughness R_a vs fatigue at 360MPa

4.6.8 Roughness R_a vs fatigue cycles at 320 MPa

Figure 40: Roughness Ra vs fatigue at 320 MPa

45 | P a g e 4.7 Fractographic surface analysis

The surface roughness as well as the fractured surface of specimens from each condition was analysed using a scanning electron microscope SEM. Particular attention was paid to crack initiation and the origin at the micro structure of each fractured surface.

4.7.1 Fractographic analysis of surface roughness

Images of the surface roughness close to the fatigue fracture are presented in figures 41 to 44, showing the effects of the different P grit abrasive films used during polishing.

Figure 41: P360-grit - 0.401 µm Ra, 2.922 µm Rz

Figure 42: P800-grit - 0.343 µm R_a, 2.575 µm R_z

Figure 43: P1200-grit - 0.301 µm Ra, 2.185 µm Rz

46 | P a g e Figure 44: P2500-grit - 0.196 µm Ra, 1.179 µm Rz

4.7.2 Fractographic analysis of fatigue fractures

The fractured surface of the specimens from each condition was analysed in the SEM in order to investigate the fatigue process. Images of the fatigue cracks were taken from the same angle to compare the effect of the axial stress amplitude as well as the position of crack initiation see Fig’s 45 to 52. Images of the fracture process and crack initiation site are presented in Fig’s 53 to 60.

Figure 45: 360 MPa axial stress – 360P-grit

47 | P a g e Figure 46: 320 MPa axial stress – 360P-grit

Figure 47: 360 MPa axial stress – 800P-grit

Figure 48: 320 MPa axial stress – 800P-grit

48 | P a g e Figure 49: 360 MPa axial stress – 1200P-grit

Figure 50: 320 MPa axial stress – 1200P-grit

Figure 51: 360 MPa axial stress – 2500P-grit

49 | P a g e Figure 52: 320 MPa axial stress – 2500P-grit

Figure 53: Show precipitates at the surface.

Figure 54: Intermetallics precipitation close to the surface.

50 | P a g e Figure 55: Brittle cleavage cracks.

Figure 56: Striations during fatigue crack growth.

Figure 57: Show the fatigue crack initiation and crack growth.

51 | P a g e Figure 58: Crack initiation at the surface.

Figure 59: Persistant slip bands at the 2500 grit surface.

Figure 60: Intermetallic obstructing the PSB’s.

52 | P a g e 4.8 Summary

The characteristic material properties from the tensile test results are very consistent which indicate an isotropic micro structure of the material used during the investigation.

However the fractographic analysis revealed some intermetallic precipitation at crack initiation sites.

The results from the load train alignment measurement show very low axial bending stress conforming to minimum requirements for fatigue testing applications.

In general the effect from both the surface roughness and the residual stress appear to be enhancing the fatigue life from 800 to 2500 P-grit size. The fatigue data represented in a stressmax vs cycle (S-n) plot show significant increase in fatigue life at 320MPa maximum axial applied load, however at 360MPa the effect of the surface condition was inconclusive. The fractographic analysis of the micro structure show intermetallic precipitates very close to the surface which is detrimental during the initiation process if micro cracks.

53 | P a g e

Chapter 5: Discussion

5.1 Introduction

The initial indications of the fatigue test results show that asymmetry and stress amplitude had a substantial influence on the fatigue life of the material used during this investigation. However at 360 MPa the effect of surface roughness and residual stress on fatigue life was unfounded mainly due to the effect of adequate plastic strain amplitude which reduces the crack initiation process and increase the crack growth process in relation to total fatigue life.

It is important to note that at low stress amplitudes the majority of the fatigue lifetime is spent in the nucleation stage of microcracks, whereas at high amplitudes nucleation is usually completed within a small fraction of the fatigue life[2], [13], [18]. Moreover it is important to understand that a number of factors are influential during the nucleation stage. Research shows that the nucleation stage depends mainly on the amplitude and asymmetry of cycling, the shape of the specimen, material parameters, environment, temperature, and surface layer[17].

For this investigation the test temperature was kept constant at 20⁰C (± 3⁰C) during all testing and the specimens were protected against corrosion prior to testing. Also the geometry of all the specimens was the same and in accordance with ISO 1099-17 dimensions. Therefore only the material parameters and the surface layer were regarded as independent variables that may have influenced the fatigue life at 320 MPa peak alternating stress. Even though the specimens were all prepared from the same sample plate, the chemical composition of Al 7449-T7651 alloy show a high concentration of intermetallic elements added for strength, which are known to have a detrimental effect on the nucleation of microcracks[23].

The decrease in peak alternating stress from 360 MPa to 320 MPa shows a significant increase in fatigue life for all surface conditions, which confirms the effect of cyclic stress amplitude and asymmetry on fatigue life, see Fig. 61.

54 | P a g e Figure 61: Shows the difference in fatigue life at 360 MPa and 320 MPa.

Test results of the specimens, tested at 320 MPa indicate that a decrease of surface roughness resulted in an increase in the fatigue life as expected. This increase in fatigue life is particularly evident from R_a 0.343µm to R_a 0.301µm, which represents the compressive residual stress was most likely the main reason for the increase in fatigue life from R_a 0.343µm to R_a 0.301µm. The scatter in the fatigue life data at 0.301µm (P1200 grit) is most likely the result of intermetallic precipitation responsible for crack initiation near the surface of one of the specimens, see Fig.60.

The test results further show that a surface roughness more than 0.2 µm R_a has a detrimental effect on fatigue life. Below 0.2 µm Ra the roughness effect on fatigue life is less evident mainly because of lower external surface and residual stress concentrations.

Subsequently the process of slip irreversibility that creates localized plastic deformation within the PSB’s which result in the formation of extrusions and intrusions at the surface appear to be more dominant during microcrack initiation.

310

55 | P a g e Figure 62: Shows the trend of a decrease in fatigue life when roughness increases.

The trend of decreasing compressive residual stress relative to a decrease in surface roughness can be attributed to the sequential polishing process from P360 to P2500 grit which decreases the diameter of the specimen, see Fig. 63.

Figure 63: Shows the substantial decrease in compressive residual stresses between 0.301µm and 0.196µm – Ra from -188 MPa to -77 MPa.

5.2 Summary

It is evident that surface roughness and compressive residual stress affect the fatigue life because of the specimen preparation that shows a decrease in diameter which resulted in a decrease in surface roughness as well as compressive residual stress. The test results confirmed the theory that a decrease in surface roughness and an increase in compressive residual stress enhance fatigue life at lower stress amplitudes[17]. It is also concluded that the effect of surface roughness was more predominant than residual

56 | P a g e stress in total. However the methods used during the investigation to determine residual stress and particularly the surface topography was found to be insufficient for characterization of surface stress factors used in finite modelling and fatigue life prediction. This is discussed further in chapter 6.

Chapter 6: Conclusions & Recommendations

The effectiveness of the arithmetic mean roughness (Ra) and the mean peak-to-valley height (Rz) was found to be questionable in order to accurately detect the maximum valley depth on the surface layer where microcrack nucleation is most likely to initiate.

The surface residual and roughness measurements covering a small localized area of the total surface in the longitudinal direction of the specimen was found to be inadequate and not representative enough.

Also for semi-empirical fatigue life assessment methods based on linear elastic fracture mechanics that uses the surface stress concentration factor K_t , the asperity root radius of the surface topography is required which is not possible to measure using traditional stylus-based methods.

The fractographic analysis of the fractured surfaces clearly indicated that fatigue failure initiated at the surface defects as a result of surface roughness, intermetallic precipitation or a combination of both.

For the purpose of determining the effect of the surface layer on the fatigue crack initiation process it is recommended that residual stress measurement to be conducted after final polishing and should represent the full circumference of the gauge section of round specimens. It is also advisable to use optical methods like white light interferometry or perhaps tomography for surface topography measurement which is

For the purpose of determining the effect of the surface layer on the fatigue crack initiation process it is recommended that residual stress measurement to be conducted after final polishing and should represent the full circumference of the gauge section of round specimens. It is also advisable to use optical methods like white light interferometry or perhaps tomography for surface topography measurement which is

In document The effect of surface finish on fatigue life of aluminium alloy during uniaxial fatigue loading (Page 39-83)