Options
A combined experimental, data-driven, and numerical approach to deform thin Ti6Al4V sheets using laser peen forming
Citation Link: https://doi.org/10.15480/882.13757
Publikationstyp
Doctoral Thesis
Date Issued
2024
Sprache
English
Author(s)
Advisor
Referee
Title Granting Institution
Technische Universität Hamburg
Place of Title Granting Institution
At Schwarzenberg Campus 1, 21073 Hamburg
Examination Date
2024-11-19
Institute
TORE-DOI
Citation
Technische Universität Hamburg (2024)
The need for complex-shaped, thin-walled structures has surged in the aerospace industry, necessitating resilient and dependable manufacturing techniques. Traditional sheet metal
forming (SMF) methods often face challenges such as wrinkling, shearing, and springback, affecting the quality and precision of formed parts. This thesis explores the optimization of the laser peen forming (LPF) process, a modern SMF technique that uses high-intensity, ultra-short laser pulses to deform materials with minimal surface damage. The research focuses on applying LPF in the aerospace sector to shape sheets to specific geometries and correct existing deformations, particularly using titanium alloy Ti6Al4V.
This work is driven by the need to reduce part rejections due to non-conformity to target geometries, tooling and production costs, and CO2 emissions, aligning with industry trends
towards sustainability and automation. As a result, LPF in the present work, is investigated as a method to produce definite geometries as well as to correct the existing geometries. The study is structured around three main approaches: experimental investigations, data-driven methodologies, and numerical simulations.
Experimental investigations focus on identifying optimal LPF process parameters to achieve desired deformations while maintaining surface integrity. Data-driven methodologies
employ artificial neural networks (ANN) to predict deformations based on process parameters, enhancing the potential for autonomous forming processes. Numerical simulations using finite element methods (FEM) complement the experimental work, providing insights into the deformation mechanisms and optimizing process parameters. A simplified numerical simulation workflow is developed to generate data that accurately represents experimental deformations for various peening patterns.
The findings demonstrate that LPF can be integrated into automated manufacturing systems, offering precise control over deformation, high accuracy, and repeatability. The developed
process planning approach with ANN predictions produces desired deformations in treated regions. This approach is successfully demonstrated on three benchmark cases involving thin Ti6Al4V sheets: unidirectional deformation, bidirectional deformation, and the modification of existing deformations in pre-bent specimens using LPF.
The cellular automata neural network (CANN) approach developed in this study utilizes a convolutional neural network (CNN) to accurately predict peening patterns based on
the deformations in the specimen after LPF. This approach enhances the LPF process, providing a reliable tool for achieving precise deformations in various applications involving complex peening patterns.
The research presented in this thesis advances the understanding and application of LPF for thin-walled Ti6Al4V structures. Experimental investigations identified optimal LPF process parameters that achieve the desired deformation while maintaining surface integrity. Numerical simulations using the eigenstrain method validated these findings and demonstrated the feasibility of applying LPF to more complex geometries. Additionally, a data-driven approach utilizing an ANN was developed for process planning, enabling the prediction of deformations for various LPF process parameters. Furthermore, a CNN-based approach presented the applicability of LPF to achieve target shapes on flat specimens by predicting the peening patterns. This integrated methodology, combining experimental, data-driven, and numerical techniques, highlights the potential of LPF as an innovative, autonomous forming process for aerospace applications, addressing both practical implementation and theoretical understanding.
forming (SMF) methods often face challenges such as wrinkling, shearing, and springback, affecting the quality and precision of formed parts. This thesis explores the optimization of the laser peen forming (LPF) process, a modern SMF technique that uses high-intensity, ultra-short laser pulses to deform materials with minimal surface damage. The research focuses on applying LPF in the aerospace sector to shape sheets to specific geometries and correct existing deformations, particularly using titanium alloy Ti6Al4V.
This work is driven by the need to reduce part rejections due to non-conformity to target geometries, tooling and production costs, and CO2 emissions, aligning with industry trends
towards sustainability and automation. As a result, LPF in the present work, is investigated as a method to produce definite geometries as well as to correct the existing geometries. The study is structured around three main approaches: experimental investigations, data-driven methodologies, and numerical simulations.
Experimental investigations focus on identifying optimal LPF process parameters to achieve desired deformations while maintaining surface integrity. Data-driven methodologies
employ artificial neural networks (ANN) to predict deformations based on process parameters, enhancing the potential for autonomous forming processes. Numerical simulations using finite element methods (FEM) complement the experimental work, providing insights into the deformation mechanisms and optimizing process parameters. A simplified numerical simulation workflow is developed to generate data that accurately represents experimental deformations for various peening patterns.
The findings demonstrate that LPF can be integrated into automated manufacturing systems, offering precise control over deformation, high accuracy, and repeatability. The developed
process planning approach with ANN predictions produces desired deformations in treated regions. This approach is successfully demonstrated on three benchmark cases involving thin Ti6Al4V sheets: unidirectional deformation, bidirectional deformation, and the modification of existing deformations in pre-bent specimens using LPF.
The cellular automata neural network (CANN) approach developed in this study utilizes a convolutional neural network (CNN) to accurately predict peening patterns based on
the deformations in the specimen after LPF. This approach enhances the LPF process, providing a reliable tool for achieving precise deformations in various applications involving complex peening patterns.
The research presented in this thesis advances the understanding and application of LPF for thin-walled Ti6Al4V structures. Experimental investigations identified optimal LPF process parameters that achieve the desired deformation while maintaining surface integrity. Numerical simulations using the eigenstrain method validated these findings and demonstrated the feasibility of applying LPF to more complex geometries. Additionally, a data-driven approach utilizing an ANN was developed for process planning, enabling the prediction of deformations for various LPF process parameters. Furthermore, a CNN-based approach presented the applicability of LPF to achieve target shapes on flat specimens by predicting the peening patterns. This integrated methodology, combining experimental, data-driven, and numerical techniques, highlights the potential of LPF as an innovative, autonomous forming process for aerospace applications, addressing both practical implementation and theoretical understanding.
Subjects
Laser peen forming (LPF)
Artificial neural networks
CANN
Dimensional analysis
Process planning
DDC Class
670: Manufacturing
Funding(s)
LuFo VI-1 Program, Project PEENCOR
Funding Organisations
Loading...
Name
PhD_Thesis_Siva_Teja_Sala.pdf
Size
36.32 MB
Format
Adobe PDF