Experimental and Numerical Analysis of Fatigue Life of Aluminum Al 2024-T351 at Elevated Temperature

: This paper presents the prediction of the fatigue life of aluminum Al 2024-T351 at room and elevated temperatures under uniaxial loading using ﬁnite element simulation. Structural parts such as fuselage, wings, aircraft turbines and heat exchangers are required to work safely at this working condition even with decreasing fatigue strength and other properties. The monotonic tensile and cyclic tests at 100 ◦ C and 200 ◦ C were conducted using MTS 810 servo hydraulic equipped with MTS 653 high temperature furnace at a frequency of 10 Hz and load ratio of 0.1. There was an 8% increase in the yield strength and a 2.32 MPa di ﬀ erence in the ultimate strength at 100 ◦ C. However, the yield strength had a 1.61 MPa di ﬀ erence and 25% decrease in the ultimate strength at 200 ◦ C compared to the room temperature. The mechanical and micro-structural behavior at elevated temperatures caused an increase in the crack initiation and crack propagation which reduced the total fatigue life. The yield strength, ultimate strength, alternating stress, mean stress and fatigue life were taken as the input in ﬁnite element commercial software, ANSYS. Comparison of results between experimental and ﬁnite element methods showed a good agreement. Hence, the suggested method using the numerical software can be used for predicting the fatigue life at elevated temperature. of applied similar to the experimental setup. The ﬁxed support A indicates the specimen end was clamped with all degrees of freedom ﬁxed and at point B the load was applied. In addition, stress-life was selected in the fatigue tool option with the load ratio set to 0.1.


Introduction
Aluminum alloy is one of the leading materials in aircraft, automotive and heavy industries due to its lightweight, high strength, good electrical conductivity, high elastic modulus and thermal resistance properties [1,2]. All these properties have an impact on the budgeting in terms of cost, performance and production. Furthermore, the designed components must meet the anticipated requirement and factor of safety of the parts for various working conditions. The study of fatigue is important because the components can experience catastrophic failure due to the condition of loading and unloading after a certain amount of time [3]. The fatigue behavior is quite important in order to estimate changes in the properties and cumulative damage of the material related to the life cycle [4,5]. Several coefficients such as size, shape, surface finish and fatigue strength are important to determine the characteristics of the structural components. These characteristics are used in plotting the stress-life curve or strain-life curve [6]. In addition, the collection of data from experimental tests [7], and analytical or numerical models [8] is the common practice to determine the fatigue properties. A standard machine testing for axial loading can have frequency of up to 20 Hz and the commonly applied frequency is 10 Hz [9]. A study on the damage accumulation proposed an approach that takes the amplitude from below the fatigue limit [10]. Moreover, a fatigue prediction without material constant is also possible [11] have been able to predict fatigue at elevated temperatures, but they are time consuming and require expensive setup and equipment. The unpolished surfaces or scratches contribute to faster crack initiation that starts from micro to macro scale [32]. Moreover, the crack initiation process is faster at elevated temperatures continued by rapid crack propagation rate than at room temperature [33].
The approach of using numerical software to solve problems of multiphysics solutions related to thermal, structural or fluid has been widely accomplished by other researchers. The advantage of using numerical software is that the strengths and weaknesses of the model are known before the final products manufactured. The commonly used numerical software programs for predicting fatigue life are ABAQUS, MSC NASTRAN, SolidWorks and ANSYS Workbench. Several studies were reported [34][35][36][37][38] predicting the fatigue life theoretically or by comparing with experimental results. ANSYS was used [39,40] in predicting the stress concentration areas. However, the design of the 3D model was by different computer aided design (CAD) software before being imported to the desired numerical software to run the fatigue analysis. This is due to experience in accustomed CAD software that reduces the time taken to design.
This study presented the prediction of fatigue life of aluminum Al 2024-T351 at elevated temperatures of 100 • C and 200 • C under uniaxial loading using ANSYS Workbench. To the author's knowledge, there is still no study conducted using this numerical software to predict the fatigue life at elevated temperatures. Several numerical software programs can predict fatigue life at room temperature. The author tried to implement the same procedure using the test data from the elevated temperature to the numerical software. The latest literature [34] showed the effectiveness of fatigue life prediction using ANSYS Workbench. The numerical software is user friendly that enables the author to complete the study. The fatigue life prediction of components that operate under elevated temperature can be conducted in the future using this study as a reference. The monotonic tensile and cyclic tests with different percentages of loadings were conducted to obtain the yield strength, ultimate strength, elastic modulus and life cycle as input to the numerical software. The results were also compared with a cyclic test at room temperature. It was expected that the results between experimental and numerical simulation be in good agreement to display the efficiency of the software in predicting fatigue life at elevated temperature.

Experimental Procedures
The material aluminum Al 2024-T351 was selected for the test due to good fatigue strength, corrosion resistance and lightweight properties. This material is widely used in transportation such as in the automotive and aerospace field. The lightweight properties require less force to move or fly a vehicle that leads to higher fuel efficiency and lower manufacturing cost. This strong and flexible material is ideal for the design of frames, aircraft fuselage, wings and other component structures. The specimen was cut from an aluminum plate using water jet technology with a dimension of 200 mm × 20 mm × 3 mm rendering to ASTM E8M standard. The tensile tests were conducted at room and elevated temperatures-100 • C and 200 • C, correspondingly.
A grip pressure of 4.83 MPa was applied to the test specimens at both ends. The alignment of grip was reset before running the test. The monotonic tensile tests were conducted at a crosshead speed of 1 mm/min using 100 kN MTS 810 servo hydraulic (MTS Systems Corporation, Eden Prairie, MN, USA) at room temperature conditions, as shown in Figure 1. For tests at elevated temperature, the MTS 810 was equipped with MTS 653 high temperature furnace (MTS Systems Corporation, Eden Prairie, MN, USA) as in Figure 2 with a capability to reach a maximum temperature of 1000 • C. The entire gauge area of the specimen was covered with MTS 653. The tests were performed to acquire the mechanical properties such as yield strength and ultimate strength.
Metals 2020, 10, x FOR PEER REVIEW 3 of 14 as in the automotive and aerospace field. The lightweight properties require less force to move or fly a vehicle that leads to higher fuel efficiency and lower manufacturing cost. This strong and flexible material is ideal for the design of frames, aircraft fuselage, wings and other component structures. The specimen was cut from an aluminum plate using water jet technology with a dimension of 200 mm × 20 mm × 3 mm rendering to ASTM E8M standard. The tensile tests were conducted at room and elevated temperatures-100 °C and 200 °C, correspondingly. A grip pressure of 4.83 MPa was applied to the test specimens at both ends. The alignment of grip was reset before running the test. The monotonic tensile tests were conducted at a crosshead speed of 1 mm/min using 100 kN MTS 810 servo hydraulic (MTS Systems Corporation, Eden Prairie, MN, USA) at room temperature conditions, as shown in Figure 1. For tests at elevated temperature, the MTS 810 was equipped with MTS 653 high temperature furnace (MTS Systems Corporation, Eden Prairie, MN, USA) as in Figure 2 with a capability to reach a maximum temperature of 1000 °C. The entire gauge area of the specimen was covered with MTS 653. The tests were performed to acquire the mechanical properties such as yield strength and ultimate strength.  The temperature was measured using digital thermometer TM902C (Reland Sung, Fujian, China). Four thermocouples, T1, T2, T3 and T4 were placed in the high temperature furnace. Three as in the automotive and aerospace field. The lightweight properties require less force to move or fly a vehicle that leads to higher fuel efficiency and lower manufacturing cost. This strong and flexible material is ideal for the design of frames, aircraft fuselage, wings and other component structures. The specimen was cut from an aluminum plate using water jet technology with a dimension of 200 mm × 20 mm × 3 mm rendering to ASTM E8M standard. The tensile tests were conducted at room and elevated temperatures-100 °C and 200 °C, correspondingly. A grip pressure of 4.83 MPa was applied to the test specimens at both ends. The alignment of grip was reset before running the test. The monotonic tensile tests were conducted at a crosshead speed of 1 mm/min using 100 kN MTS 810 servo hydraulic (MTS Systems Corporation, Eden Prairie, MN, USA) at room temperature conditions, as shown in Figure 1. For tests at elevated temperature, the MTS 810 was equipped with MTS 653 high temperature furnace (MTS Systems Corporation, Eden Prairie, MN, USA) as in Figure 2 with a capability to reach a maximum temperature of 1000 °C. The entire gauge area of the specimen was covered with MTS 653. The tests were performed to acquire the mechanical properties such as yield strength and ultimate strength.  The temperature was measured using digital thermometer TM902C (Reland Sung, Fujian, China). Four thermocouples, T1, T2, T3 and T4 were placed in the high temperature furnace. Three The temperature was measured using digital thermometer TM902C (Reland Sung, Fujian, China). Four thermocouples, T1, T2, T3 and T4 were placed in the high temperature furnace. Three thermocouples, T1, T2 and T3 were attached to the specimen and T4 hanging in the middle area, as shown in Figure 3.
Metals 2020, 10, x FOR PEER REVIEW 4 of 14 thermocouples, T1, T2 and T3 were attached to the specimen and T4 hanging in the middle area, as shown in Figure 3. The cyclic tests were performed according to ASTM E466 standard at a frequency of 10 Hz, sinusoidal waveform and load ratio of 0.1. Each test was loaded with a specific mean load and for tests at room temperature, the test started at the maximum load value of 90% of the yield and decremented by 5% for the next tests until 70% of the yield strength. Similarly, the maximum load for 100 °C started from 90% to 75% of the yield strength; the maximum load decreased by 5% in each test. For tests at 200 °C, the maximum load started from 90% to 70% of the yield strength; the maximum load decreased by 10% in each case. The results were plotted in the form of stress-life curves to estimate the cycles to failure after definite loading. The loads applied were shown in Table  1. The cyclic test can only be conducted after certain inputs such as mean load, amplitude and frequency are known. The load ratio, Lr can be calculated by the given Equation (1) [41].
The range of load, LR can be calculated by the difference between maximum load, Lmax and minimum load, Lmin in the following Equation (2).
The amplitude load, La is half of range of load, LR is given by Equation (3). The cyclic tests were performed according to ASTM E466 standard at a frequency of 10 Hz, sinusoidal waveform and load ratio of 0.1. Each test was loaded with a specific mean load and for tests at room temperature, the test started at the maximum load value of 90% of the yield and decremented by 5% for the next tests until 70% of the yield strength. Similarly, the maximum load for 100 • C started from 90% to 75% of the yield strength; the maximum load decreased by 5% in each test. For tests at 200 • C, the maximum load started from 90% to 70% of the yield strength; the maximum load decreased by 10% in each case. The results were plotted in the form of stress-life curves to estimate the cycles to failure after definite loading. The loads applied were shown in Table 1. The cyclic test can only be conducted after certain inputs such as mean load, amplitude and frequency are known. The load ratio, L r can be calculated by the given Equation (1) [41].
The range of load, L R can be calculated by the difference between maximum load, L max and minimum load, L min in the following Equation (2).
The amplitude load, L a is half of range of load, L R is given by Equation (3).
Moreover, the mean load, L m is important as higher load means the lower fatigue life of the tested components. The Equation (4) can be used to calculate the mean load.

Numerical Procedures
The use of numerical software is to determine an approximate solution to the specific problem. When the experimental results are available, any numerical results can be verified. The selected computer aided design (CAD) software to design the 3D model of the specimen was SolidWorks 2012. The dimension was set to 200 mm in length, 20 mm width and 3 mm thickness as shown in Figure 4. A top plane was selected for the sketching of the model. The thickness was generated by using a boss-extrude feature from the sketch. The complete 3D model is as shown in Figure 5. The 3D model was saved as IGES format for the ease of importing from ANSYS Workbench 16.1 (ANSYS Inc., Canonsburg, PA, USA).

Numerical Procedures
The use of numerical software is to determine an approximate solution to the specific problem. When the experimental results are available, any numerical results can be verified. The selected computer aided design (CAD) software to design the 3D model of the specimen was SolidWorks 2012. The dimension was set to 200 mm in length, 20 mm width and 3 mm thickness as shown in Figure 4. A top plane was selected for the sketching of the model. The thickness was generated by using a boss-extrude feature from the sketch. The complete 3D model is as shown in Figure 5. The 3D model was saved as IGES format for the ease of importing from ANSYS Workbench 16.1 (ANSYS Inc., Canonsburg, PA, USA).  The mechanical properties such as yield strength, ultimate strength, mean stress, alternating stress and fatigue life acquired from experimental tests were assigned in the engineering data of the

Numerical Procedures
The use of numerical software is to determine an approximate solution to the specific problem. When the experimental results are available, any numerical results can be verified. The selected computer aided design (CAD) software to design the 3D model of the specimen was SolidWorks 2012. The dimension was set to 200 mm in length, 20 mm width and 3 mm thickness as shown in Figure 4. A top plane was selected for the sketching of the model. The thickness was generated by using a boss-extrude feature from the sketch. The complete 3D model is as shown in Figure 5. The 3D model was saved as IGES format for the ease of importing from ANSYS Workbench 16.1 (ANSYS Inc., Canonsburg, PA, USA).  The mechanical properties such as yield strength, ultimate strength, mean stress, alternating stress and fatigue life acquired from experimental tests were assigned in the engineering data of the The mechanical properties such as yield strength, ultimate strength, mean stress, alternating stress and fatigue life acquired from experimental tests were assigned in the engineering data of the software. Moreover, a fine mesh with SOLID 187 10-noded element was selected due to its suitability in covering the uneven surface areas in order to predict results closer to the experiments. The mesh model can be observed in Figure 6a with the generated number of nodes of 6181 and the number of elements of 1008. The boundary conditions shown in Figure 6b were applied similar to the experimental setup. The fixed support A indicates the specimen end was clamped with all degrees of freedom fixed and at point B the load was applied. In addition, stress-life was selected in the fatigue tool option with the load ratio set to 0.1.
Metals 2020, 10, x FOR PEER REVIEW 6 of 14 software. Moreover, a fine mesh with SOLID 187 10-noded element was selected due to its suitability in covering the uneven surface areas in order to predict results closer to the experiments. The mesh model can be observed in Figure 6a with the generated number of nodes of 6181 and the number of elements of 1008. The boundary conditions shown in Figure 6b were applied similar to the experimental setup. The fixed support A indicates the specimen end was clamped with all degrees of freedom fixed and at point B the load was applied. In addition, stress-life was selected in the fatigue tool option with the load ratio set to 0.1.

Fatigue Life of Aluminium Al 2024-T351
The working temperature changes the properties of the material, especially in terms of its strength [42]. The yield strength at 100 °C had a decrease of 6%, and 7% at 200 °C when compared to the controlled room temperature. On the other hand, the average ultimate strength had a small decrease of 3.75% at 100 °C and 25% at 200 °C when compared to the controlled room temperature. The data are shown in Table 2. The flat specimen under tensile loading experience shear rupture upon failure [43]. The grains are affected by the level of loading, type and temperature. At room temperature, the visibilities of the grains were not clear and there was an increase in the length as well. In addition, the grain growth at high temperature caused it to be compressed and elongated due to tensile load applied. The length increases, but the width decreases [44]. During tests at high temperature and loading, the precipitates from the coarse grain may not visibly or dissolve completely.

Fatigue Life of Aluminium Al 2024-T351
The working temperature changes the properties of the material, especially in terms of its strength [42]. The yield strength at 100 • C had a decrease of 6%, and 7% at 200 • C when compared to the controlled room temperature. On the other hand, the average ultimate strength had a small decrease of 3.75% at 100 • C and 25% at 200 • C when compared to the controlled room temperature. The data are shown in Table 2. The flat specimen under tensile loading experience shear rupture upon failure [43]. The grains are affected by the level of loading, type and temperature. At room temperature, the visibilities of the grains were not clear and there was an increase in the length as well. In addition, the grain growth at high temperature caused it to be compressed and elongated due to tensile load applied. The length increases, but the width decreases [44]. During tests at high temperature and loading, the precipitates from the coarse grain may not visibly or dissolve completely. The stress-life curve for the aluminum at room and elevated temperatures is presented in Figure 7. As anticipated the number of cycles increases when the stress applied deceases. The average fatigue life at 90% was 45,343 cycles. As the load decreased, the fatigue life increased by 16% to 85%. Moreover, the fatigue life at 80% was 88,215 cycles. There was about a 49% increase in the fatigue life when the difference between the loading was only 10%. When the load was further decreased by 10%, the difference increased to 82% showing an average of 252,827 cycles at 70% of the yield strength.
The fatigue life at 90% and at 100 • C was 21,679 cycles. There was an increase of 65% when the load decreased to 85% of the yield strength. However, there was only a 5% increase in the fatigue life as the load reduced to 80%. The fatigue life at 75% increased by 33% compared to 80% displaying an average of 96,688 cycles. The stress-life curve for the aluminum at room and elevated temperatures is presented in Figure 7. As anticipated the number of cycles increases when the stress applied deceases. The average fatigue life at 90% was 45,343 cycles. As the load decreased, the fatigue life increased by 16% to 85%. Moreover, the fatigue life at 80% was 88,215 cycles. There was about a 49% increase in the fatigue life when the difference between the loading was only 10%. When the load was further decreased by 10%, the difference increased to 82% showing an average of 252,827 cycles at 70% of the yield strength. The fatigue life at 90% and at 100 °C was 21,679 cycles. There was an increase of 65% when the load decreased to 85% of the yield strength. However, there was only a 5% increase in the fatigue life as the load reduced to 80%. The fatigue life at 75% increased by 33% compared to 80% displaying an average of 96,688 cycles. At 200 °C, the fatigue life was 14,272 cycles. When the load was reduced by 10% the fatigue life increased by 63% displaying 38,427 cycles. Further decrease in the percentage of loading by 10% increased the fatigue life by 60%. The fatigue life at 70% has an increase of 85% compared to 90% of the yield strength. It can be observed that at 90% of the yield strength, the fatigue life was reduced by half at 100 °C and 69% at 200 °C compared to the room temperature. Additionally, at 80%, the fatigue life difference was 27% and 56% at 100 °C and 200 °C, respectively. The difference at 70% for 200 °C was 62% and for 100 °C the percentage was expected to be lower compared to the room temperature.
When a material is subjected to a repeated loading and unloading, formation of micro-crack arises in the stress concentration area. The surface area is the most common place that the crack initiation starts [45,46]. Next, the crack propagates until reaching the critical length that causes failure of the material. The letter A represents the stable region and letter B the unstable region. The surface failure of aluminum at room temperature can be observed in Figure 8 with different percentages of loading. The visibility of the fracture areas with smooth and shiny plus uneven appearance [47] can be observed as well. In addition, the size of cavities in the unstable region is larger than the stable region. The bigger particle size in the cavities means shorter fatigue life. The formation of void is a sign that damage is experienced by the specimen and contributed by separation of grain boundaries or particle fracture [48]. Moreover, the crack growth pattern is not consistence due to different orientation of the grains. The fatigue life can also be estimated using the striation length [49]. At 200 • C, the fatigue life was 14,272 cycles. When the load was reduced by 10% the fatigue life increased by 63% displaying 38,427 cycles. Further decrease in the percentage of loading by 10% increased the fatigue life by 60%. The fatigue life at 70% has an increase of 85% compared to 90% of the yield strength. It can be observed that at 90% of the yield strength, the fatigue life was reduced by half at 100 • C and 69% at 200 • C compared to the room temperature. Additionally, at 80%, the fatigue life difference was 27% and 56% at 100 • C and 200 • C, respectively. The difference at 70% for 200 • C was 62% and for 100 • C the percentage was expected to be lower compared to the room temperature.
When a material is subjected to a repeated loading and unloading, formation of micro-crack arises in the stress concentration area. The surface area is the most common place that the crack initiation starts [45,46]. Next, the crack propagates until reaching the critical length that causes failure of the material. The letter A represents the stable region and letter B the unstable region. The surface failure of aluminum at room temperature can be observed in Figure 8 with different percentages of loading. The visibility of the fracture areas with smooth and shiny plus uneven appearance [47] can be observed as well. In addition, the size of cavities in the unstable region is larger than the stable region. The bigger particle size in the cavities means shorter fatigue life. The formation of void is a sign that damage is experienced by the specimen and contributed by separation of grain boundaries or particle fracture [48]. Moreover, the crack growth pattern is not consistence due to different orientation of the grains. The fatigue life can also be estimated using the striation length [49]. The surface failure at 100 °C can be observed in Figure 9 ranging from 90% to 75% of the yield strength. The visibility of the stable crack growth region is very small due to catastrophic failure during loading. Meanwhile, in Figure 10, surface failures from 90% to 70% of the yield strength can be seen. The stable region surfaces are smaller compared to those at room temperature. Consequently, at elevated temperature, the elastic modulus and surface energy are lower, causing faster crack initiation and crack propagation [50]. The increased grain size causes the particle density to decrease. As a result, the fatigue life decreases as temperature increases [51]. Similarly, the fracture of particles leads to formation of voids [52]. In addition, the increase in rate of oxidation also contributes to the decrease in fatigue life and affects the mechanical properties as well [53]. Therefore, the total fatigue life at elevated temperature is affected by mechanical and micro-structural behavior that contributes to crack initiation and crack propagation [54]. The plastic cracking mechanism happened when exposed to elevated temperature. The plastic zone in the micro and macro cracks during loading leads to plastic deformation at the elevated temperature.  The surface failure at 100 • C can be observed in Figure 9 ranging from 90% to 75% of the yield strength. The visibility of the stable crack growth region is very small due to catastrophic failure during loading. Meanwhile, in Figure 10, surface failures from 90% to 70% of the yield strength can be seen. The stable region surfaces are smaller compared to those at room temperature. Consequently, at elevated temperature, the elastic modulus and surface energy are lower, causing faster crack initiation and crack propagation [50]. The increased grain size causes the particle density to decrease. As a result, the fatigue life decreases as temperature increases [51]. Similarly, the fracture of particles leads to formation of voids [52]. In addition, the increase in rate of oxidation also contributes to the decrease in fatigue life and affects the mechanical properties as well [53]. Therefore, the total fatigue life at elevated temperature is affected by mechanical and micro-structural behavior that contributes to crack initiation and crack propagation [54]. The plastic cracking mechanism happened when exposed to elevated temperature. The plastic zone in the micro and macro cracks during loading leads to plastic deformation at the elevated temperature. The surface failure at 100 °C can be observed in Figure 9 ranging from 90% to 75% of the yield strength. The visibility of the stable crack growth region is very small due to catastrophic failure during loading. Meanwhile, in Figure 10, surface failures from 90% to 70% of the yield strength can be seen. The stable region surfaces are smaller compared to those at room temperature. Consequently, at elevated temperature, the elastic modulus and surface energy are lower, causing faster crack initiation and crack propagation [50]. The increased grain size causes the particle density to decrease. As a result, the fatigue life decreases as temperature increases [51]. Similarly, the fracture of particles leads to formation of voids [52]. In addition, the increase in rate of oxidation also contributes to the decrease in fatigue life and affects the mechanical properties as well [53]. Therefore, the total fatigue life at elevated temperature is affected by mechanical and micro-structural behavior that contributes to crack initiation and crack propagation [54]. The plastic cracking mechanism happened when exposed to elevated temperature. The plastic zone in the micro and macro cracks during loading leads to plastic deformation at the elevated temperature.      Figure 11 shows the example of numerical results from the software ANSYS. The maximum stress is located at the center of the specimen. At this area, the failure occurs and the fatigue life is the minimum.

Numerical Results
Metals 2020, 10, x FOR PEER REVIEW 9 of 14 Figure 11 shows the example of numerical results from the software ANSYS. The maximum stress is located at the center of the specimen. At this area, the failure occurs and the fatigue life is the minimum.   Figure 12 shows the comparison of the stress-life curve for the aluminum at room temperature. The increase in stress decreases fatigue life as expected. At 90% of the yield strength, the experimental fatigue life was 45,343 cycles and the numerical value was 47,213 cycles. There was only a small difference of about 4%. At 85%, the difference increased to 13.5%. The experimental life at 80% was 88,125 cycles and numerical life was 100,172 cycles. The difference was 14%. Moreover, the difference between experimental and numerical results decreased by 1% at 75% of the yield strength. At 70%, the difference was reduced to 9%. The highest difference between experimental and numerical results was at 80% and the lowest was at 90%. The stress-life in Figure 13 shows the comparison of experimental and numerical results at 100 °C. The highest difference was 27% at 90% of the yield strength. In terms of numbers, the experiment displayed 21,679 cycles and numerical 29,548 cycles. The lowest difference was at 85% with only a 1% difference. The difference at 80% and 75% were at 10% and 4%, respectively.  The stress-life in Figure 13 shows the comparison of experimental and numerical results at 100 • C. The highest difference was 27% at 90% of the yield strength. In terms of numbers, the experiment displayed 21,679 cycles and numerical 29,548 cycles. The lowest difference was at 85% with only a 1% difference. The difference at 80% and 75% were at 10% and 4%, respectively. The stress-life in Figure 13 shows the comparison of experimental and numerical results at 100 °C. The highest difference was 27% at 90% of the yield strength. In terms of numbers, the experiment displayed 21,679 cycles and numerical 29,548 cycles. The lowest difference was at 85% with only a 1% difference. The difference at 80% and 75% were at 10% and 4%, respectively. The comparison of experimental and numerical results at 200 °C is presented by the stress-life curve in Figure 14. The results showed the fatigue life of 14,272 cycles and 16,620 cycles for the experimental and numerical work, respectively, at 90% of the yield strength. The difference between the two methods was 14%. At 80%, the difference decreased by 2%. The lowest difference The comparison of experimental and numerical results at 200 • C is presented by the stress-life curve in Figure 14. The results showed the fatigue life of 14,272 cycles and 16,620 cycles for the experimental and numerical work, respectively, at 90% of the yield strength. The difference between the two methods was 14%. At 80%, the difference decreased by 2%. The lowest difference was at 70% with only a difference of 3%. Nevertheless, the gap decreased between the trend lines at 100 • C and 200 • C using ANSYS as the stress decreased. The difference of experimental results is due to several factors such as scratches, unpolished surfaces and micro-structural behavior. Therefore, all the results show a good trend line comparison which proves a good agreement.

Conclusions
The fatigue testing of aluminum has been conducted at elevated temperatures with a frequency of 10 Hz and load ratio of 0.1. The influence of this temperature has caused a decrease in fatigue life, yield strength and ultimate strength. The yield and ultimate strength decrease as the temperature increases. The increase in the crack growth rate and grain size has caused the particle density to decrease. The surface failure was separated into a stable and unstable region. The stable region was smaller when the load applied was higher. Factors such as grain size, particle density, loading and temperature have influence in the fatigue life of the material. The fatigue life at 90% of the yield strength decreased by 50% when exposed to a temperature of 100 °C compared to the results at room temperature. A further decrease in fatigue life was observed when the temperature was increased to 200 °C. The comparison of experimental and numerical results at room temperature showed that the highest difference was 14% and the lowest was 4%. The comparison at 100 °C showed that the highest difference was 27% and the lowest was 1%. The lowest difference at 200 °C was 3% and the highest was 14%. A good agreement between experimental and numerical results was observed for all the stress-life curves. The numerical software, ANSYS Workbench successfully proved its effectiveness in predicting the fatigue life at room and elevated temperatures.

Conclusions
The fatigue testing of aluminum has been conducted at elevated temperatures with a frequency of 10 Hz and load ratio of 0.1. The influence of this temperature has caused a decrease in fatigue life, yield strength and ultimate strength. The yield and ultimate strength decrease as the temperature increases. The increase in the crack growth rate and grain size has caused the particle density to decrease. The surface failure was separated into a stable and unstable region. The stable region was smaller when the load applied was higher. Factors such as grain size, particle density, loading and temperature have influence in the fatigue life of the material. The fatigue life at 90% of the yield strength decreased by 50% when exposed to a temperature of 100 • C compared to the results at room temperature. A further decrease in fatigue life was observed when the temperature was increased to 200 • C. The comparison of experimental and numerical results at room temperature showed that the highest difference was 14% and the lowest was 4%. The comparison at 100 • C showed that the highest difference was 27% and the lowest was 1%. The lowest difference at 200 • C was 3% and the highest was 14%. A good agreement between experimental and numerical results was observed for all the stress-life curves. The numerical software, ANSYS Workbench successfully proved its effectiveness in predicting the fatigue life at room and elevated temperatures.

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