The Library
Investigations of fast quality prediction and rivet/die optimization for self-piercing riveting joints
Tools
Zhao, Huan (2021) Investigations of fast quality prediction and rivet/die optimization for self-piercing riveting joints. PhD thesis, University of Warwick.
|
PDF
WRAP_Theses_Zhao_2021.pdf - Submitted Version - Requires a PDF viewer. Download (9Mb) | Preview |
Official URL: http://webcat.warwick.ac.uk/record=b3759992
Abstract
Self-piercing riveting (SPR) technique has achieved rapid development over the last three decades, and has already become one of the major connection approaches in the automotive industry. The selection of suitable rivets and dies for new SPR joints is always a big challenge because of the high requirements on engineers’ joint design experience and the heavy costs of numerous experimental tests. Therefore, to shorten the design cycle and to quickly identify the suitable rivets and dies for new joints, this thesis carried out in-depth research in the development of fast joint quality prediction tools and rivet/die optimization strategies.
Firstly, in order to overcome the high investment in time and money of the great number of experimental SPR tests, a finite element analysis (FEA) model of the SPR process was established to collect sufficient joint quality data for the training and testing of fast quality prediction models. Then, fast quality prediction tools were developed using the multiple regression analysis, the shallow neural network (SNN) and the deep neural network (DNN) respectively, and their performances were evaluated and validated through experimental SPR tests. To obtain the desired quality for a specific single joint, a strategy that combined the developed SNNs with the genetic algorithm (GA) was proposed to automatically optimize the rivet and die parameters. Meanwhile, to simplify the selection of rivet/die for multiple new sheet combinations, two novel approaches suitable for inexperienced engineers were also proposed with the DNN. The first approach took the joint robustness into account and achieved automatic selection of rivet/die for multiple new sheet combinations. The Monte Carlo method was employed to evaluate the robustness of designed SPR joints. The second approach was developed based on application range maps of different rivet/die combinations, and was also approved effective to quickly determine the minimum rivet/die combinations for multiple sheet combinations.
In addition, to deepen understanding of the SPR process and to facilitate the selection of rivet and die, analysis of interaction effects between rivet, sheet and die parameters on the SPR joint quality was conducted with the developed regression models and the SNNs. The formation mechanisms of SPR joints with varying joining parameters were also numerically investigated with the FEA model. Moreover, two graphic user interfaces (GUIs) respectively integrating the developed fast quality prediction models and the automatic rivet/die selection approach were also developed to facilitate their practical applications in the industry sector. Overall, the results from this thesis are beneficial for simplifying the design process of new SPR joints, and have a great prospect in the automotive industry.
Item Type: | Thesis (PhD) | ||||
---|---|---|---|---|---|
Subjects: | T Technology > TA Engineering (General). Civil engineering (General) T Technology > TL Motor vehicles. Aeronautics. Astronautics T Technology > TS Manufactures |
||||
Library of Congress Subject Headings (LCSH): | Rivets and riveting, Riveted joints, Joints (Engineering), Finite element method | ||||
Official Date: | September 2021 | ||||
Dates: |
|
||||
Institution: | University of Warwick | ||||
Theses Department: | School of Engineering | ||||
Thesis Type: | PhD | ||||
Publication Status: | Unpublished | ||||
Supervisor(s)/Advisor: | Liu, Xianping, 1957- ; Han, Li | ||||
Sponsors: | University of Warwick. School of Engineering ; Jaguar Land Rover (Firm) | ||||
Format of File: | |||||
Extent: | xxi, 177 leaves : illustrations | ||||
Language: | eng |
Request changes or add full text files to a record
Repository staff actions (login required)
View Item |