Study On Numerical Simulation And Optimization Techniques For Auto Panel Forming Processes

Due to limitations of the technique, the leadtime and the development cost in the design phase, an acceptable forming solution is not easy to obtain with a large number of trial-and-error runs. Numerical simulation technology, a quickly developing engineering analysis method, has laid a foundation for the automation of the forming process design. It has been an indispensable tool to speed up the development of auto body.Most of the existing research on auto panel forming is focused on verifying whether product defects appear. However, comprehensive and systematic work has rarely been done on how to improve the performance figures of auto panel in the design phase. Optimization of forming process is a very difficult problem due to high nonlinearity and huge computational cost of numerical simulations. Therefore, there are notable engineering significance and academic value to seek for a numerical simulation based optimization method for auto panel forming process.Firstly, the background and the significance of this dissertation are elaborated. Then, the basic contents of design of auto panel forming process are summarized, and analysis methods of auto panel forming are compared in detail. Furthermore, based on the state-of-the-art of relative domain about numerical simulation and engineering optimization techniques at home and abroad, the objective and technical route of this research are proposed.In chapter 2, the forming process of auto panel, the stress-strain states of the deforming blank and the roles of some key parameters of sheet metals are analyzed. Based on these analyses, the technique of determining forming limit diagram (FLD) is studied deeply, and is treated as the main criterion of estimating quality of auto panel forming. Then, the main factors of resulting in forming defects, such as fractures, wrinkles and springbacks are analyzed. The correlations between part defects and some design parameters are also obtained. In the end, the forming characteristics of a novel type of sheet metal, tailor-welded blanks (TWBs), are discussed.In chapter 3, several techniques involved in numerical simulations of auto panel forming are studied, and an efficient simulation flowchart of auto panel forming process is schemed. In this flow, the formability of drawing parts and the shape of blanks is firstly predicted by the inverse finite element approach. Then the forming process and part defects are investigated using the dynamic explicit finite element method, so the path to improving the forming quality is found out. The approach is proved by some examples.In chapter 4, based on the principle of engineering optimization, some state values, such as material consumption, forming quality and residual deformations are selected to establish mathematical models of various optimal objectives. The method that takes sheet metal forming limit curve (FLC) as the main design constraint is given. The selection principle of design variables in optimization is specified, and the concept and feasibletechniques to solve optimization problems are analyzed summarily.In chapter 5, aiming at highly nonlinear nature of the auto panel forming, Response Surface Methodology (RSM) is utilized to construct sequential response surfaces to approximately describe the objective the constraint functions, and the global optimum solution can be obtained with some successive iterations. Studies have indicated that this method could eliminate computational noise and help to seek for the optimal forming solution faster while guaranteed the accuracy.The drawback of RSM is that many numerical simulations are needed for optimization with higher order approximate models or many design variables. In chapter 6, in order to improve the efficiency of optimization, Space Mapping (SM) technique utilizing surrogate models is integrated with RSM. Through a parameter mapping, the majority of the optimization is performed on the coarse model and the fine model is used to correct the error in the coarse model. This approach is illustrated by some examples.In the conclusion of this dissertation, the future work is put forward.