Periodic Reporting for period 1 - RESTREIG (Development of A Method for Analysis of Creep Behaviour of Welded Rotating Components of High Temperature Applications Based on Eigenstrain Theory)
Berichtszeitraum: 2018-06-19 bis 2020-06-18
Creep is a time dependent deformation that occurs in metals and alloys after prolonged exposure to stress, which is below or above yield strength, at elevated temperatures. Failures related to creep appears with the tertiary stage of this three-stage process, but they initiate and develop at the early stages. Creep beginning temperature depends on the material composition and operating stress. The service life of components of high temperature and pressure applications usually ends with a creep related failure. Creep behaviour is an important parameter for lifetime assessment of materials. ASME Boiler and Pressure Vessel Code determine allowable stresses formation to be 1% creep expansion, or deformation, in 100,000 hours of service life. This property of creep damage necessitates careful investigation of creep behaviour of materials that are used in the design of aerospace components and other industrial parts which are gas turbines, super-heaters, boilers and reactors.
This summary presents a summary of recently developed computational and numerical tools to reconstruct residual stress fields and analyze creep in nickel superalloy welds used in aerospace engineering components. This approach combines experimental data with eigenstrain theory to reconstruct stress fields at the macroscopic scale and provided reliable means for numerical prediction of creep behavior of welded components under complex loading conditions. Experimental data in the form of profilometry scans was interpreted using a range of iterative eigenstrain methods that included the adaptation of the contour method and artificial intelligence models for eigenstrain-creep analysis. The integration of principles of artificial intelligence with eigenstrain models allowed highly accurate results to be obtained which were validated by comparison with experimental data obtained using independent techniques such as neutron diffraction. The use of artificial intelligence models is discussed for residual stress reconstruction and creep behavior prediction in annular aeroengine parts manufactured using inertia friction welding. To extend the range of experimental data included in consideration, the height Digital Image Correlation (hDIC) technique was introduced that utilizes information regarding triaxial displacements obtained from profilometry, allowing deeper and more reliable analysis to be conducted. The hDIC technique was validated using operando tensile testing data.
This summary presents a summary of recently developed computational and numerical tools to reconstruct residual stress fields and analyze creep in nickel superalloy welds used in aerospace engineering components. This approach combines experimental data with eigenstrain theory to reconstruct stress fields at the macroscopic scale and provided reliable means for numerical prediction of creep behavior of welded components under complex loading conditions. Experimental data in the form of profilometry scans was interpreted using a range of iterative eigenstrain methods that included the adaptation of the contour method and artificial intelligence models for eigenstrain-creep analysis. The integration of principles of artificial intelligence with eigenstrain models allowed highly accurate results to be obtained which were validated by comparison with experimental data obtained using independent techniques such as neutron diffraction. The use of artificial intelligence models is discussed for residual stress reconstruction and creep behavior prediction in annular aeroengine parts manufactured using inertia friction welding. To extend the range of experimental data included in consideration, the height Digital Image Correlation (hDIC) technique was introduced that utilizes information regarding triaxial displacements obtained from profilometry, allowing deeper and more reliable analysis to be conducted. The hDIC technique was validated using operando tensile testing data.
The Chebyshev functions combined with the Gauss function, used for creating eigenstrain fields, is represented between -1 and 1 interval in Figure 1. In this study bead on plate design is applied on Inconel 740H specimen where weld bead lies along one third of the length of the plate as illustrated in Figure 2. The processed surface topography data after smoothing is illustrated in Figure 3. Distribution of two data sets show that post-weld heat treated specimen has lower stress relaxation.
Multi-Component Iterative Eigenstrain-Contour Model: The eigenstrain reconstruction of residual stresses within the 3D body is accomplished by determining two components of eigenstrain. Through-thickness averaged profile plots of the multi-component iterative solution are given in Figure 4. Distribution and magnitudes of eigenstrain components in as-welded and heat-treated model are illustrated in Figures 5.
Multi-Component Multi-Dimensional Iterative Eigenstrain-Contour Model: Second component of displacement information that define distortion is included in the eigenstrain reconstruction process. Figure 6 represents eigenstrain distribution in as-welded and heat-treated cases. Linear finite element solutions were performed using the eigenstrain distribution calculated using the coefficients of components. Results of these calculations are given in Figures 7. Through thickness averaged residual stress distributions along the transversal direction are illustrated in Figure 8.
Artificial Intelligence and Eigenstrain-Contour Model: Parameters of the eigenstrain-contour method are determined using an artificial agent. Results of the eigenstrain-contour model calculations using optimum model parameters are given in Figure 9. Figure 10 illustrates profile distributions of comparison of calculated and measured displacements and residual stresses on Plane. Results of neutron strain scanning measurements and their comparison with model calculations given in terms of residual strains in Figure 11.
Artificial Intelligence and Eigenstrain-Creep Model: Principles of artificial intelligence were used for the second time for simulation of post weld heat treatment process. The proposed method uses eigenstrain fields determined by eigenstrain-contour model and simulates PWHT process. Comparison of residual stresses with calculated after the PWHT simulation with experimental contour method solutions are given in Figure 12. Distribution of through thickness averaged longitudinal residual stresses determined by the model shows good agreement with experimental results. To understand the effect of ambient temperature and holding time on PWHT Inconel Alloy 740H, analyses performed using the optimum eigenstrain-creep model parameters determined for the current specimen and results are illustrated in Figure 13.
Height Digital Image Correlation for Triaxial Displacement Information: Surface topography data from reference and 3 steps of plastic range are illustrated in Figure 14. Plastic-A is the start of plastic stage, Plastic-B is the ultimate tensile stage, and Plastic-C is the last step before break. Triaxial displacements calculated during the 3 steps of plastic range of tensile test are illustrated in Figure 15.
Multi-Component Iterative Eigenstrain-Contour Model: The eigenstrain reconstruction of residual stresses within the 3D body is accomplished by determining two components of eigenstrain. Through-thickness averaged profile plots of the multi-component iterative solution are given in Figure 4. Distribution and magnitudes of eigenstrain components in as-welded and heat-treated model are illustrated in Figures 5.
Multi-Component Multi-Dimensional Iterative Eigenstrain-Contour Model: Second component of displacement information that define distortion is included in the eigenstrain reconstruction process. Figure 6 represents eigenstrain distribution in as-welded and heat-treated cases. Linear finite element solutions were performed using the eigenstrain distribution calculated using the coefficients of components. Results of these calculations are given in Figures 7. Through thickness averaged residual stress distributions along the transversal direction are illustrated in Figure 8.
Artificial Intelligence and Eigenstrain-Contour Model: Parameters of the eigenstrain-contour method are determined using an artificial agent. Results of the eigenstrain-contour model calculations using optimum model parameters are given in Figure 9. Figure 10 illustrates profile distributions of comparison of calculated and measured displacements and residual stresses on Plane. Results of neutron strain scanning measurements and their comparison with model calculations given in terms of residual strains in Figure 11.
Artificial Intelligence and Eigenstrain-Creep Model: Principles of artificial intelligence were used for the second time for simulation of post weld heat treatment process. The proposed method uses eigenstrain fields determined by eigenstrain-contour model and simulates PWHT process. Comparison of residual stresses with calculated after the PWHT simulation with experimental contour method solutions are given in Figure 12. Distribution of through thickness averaged longitudinal residual stresses determined by the model shows good agreement with experimental results. To understand the effect of ambient temperature and holding time on PWHT Inconel Alloy 740H, analyses performed using the optimum eigenstrain-creep model parameters determined for the current specimen and results are illustrated in Figure 13.
Height Digital Image Correlation for Triaxial Displacement Information: Surface topography data from reference and 3 steps of plastic range are illustrated in Figure 14. Plastic-A is the start of plastic stage, Plastic-B is the ultimate tensile stage, and Plastic-C is the last step before break. Triaxial displacements calculated during the 3 steps of plastic range of tensile test are illustrated in Figure 15.
The inverse problem of eigenstrain is the determination of eigenstrain field that cause formation of residual stresses that match with experimentally determined deformations. In this research, the use of optical profilometry data for the purpose of eigenstrain reconstruction process was presented. Large amount of data from a surface of wire-cut allowed determination of volumetric distribution of residual stresses. In the final stage of current research, the use of optical profilometry techniques for the purpose of determination displacements that can be used for eigenstrain-contour and eigenstrain-creep model is presented.
The iterative solution of eigenstrain-contour model showed us the main source of welding residual stress is the eigenstrains formed in and around the weld zone. The inclusion of second component of experimental data showed that it is possible to model residual stress distribution and weld distortions in the whole body accurately. However, the functions used for the determination of eigenstrain distribution required determination of corresponding parameters manually by a trial and error process. In order to deal with this problem, principles of artificial intelligence were implemented with eigenstrain-contour method. Results showed that starting from a randomly determined model parameters, the artificial agent determines the optimum model parameters that provide perfect match of model results with experimental profilometry measurements. The same approach is applied for the simulation of PWHT process. The implementation of principles of artificial intelligence with eigenstrain-creep model allowed determination of creep properties specific to the specimen accurately.
The iterative solution of eigenstrain-contour model showed us the main source of welding residual stress is the eigenstrains formed in and around the weld zone. The inclusion of second component of experimental data showed that it is possible to model residual stress distribution and weld distortions in the whole body accurately. However, the functions used for the determination of eigenstrain distribution required determination of corresponding parameters manually by a trial and error process. In order to deal with this problem, principles of artificial intelligence were implemented with eigenstrain-contour method. Results showed that starting from a randomly determined model parameters, the artificial agent determines the optimum model parameters that provide perfect match of model results with experimental profilometry measurements. The same approach is applied for the simulation of PWHT process. The implementation of principles of artificial intelligence with eigenstrain-creep model allowed determination of creep properties specific to the specimen accurately.