The Research On Springback Prediction And Its Application Technology In Sheet Forming

Springback is a common physical phenomenon in the metal sheet forming. The springback problem, which make the shape and dimension diverge from design requirements, direct affects the quality of forming product, including the appearance quality, the assembly quality and the final product performance. How to predict the shape of products after forming, and design the required die/punch surface to compensate springback is a difficult problem in the metal sheet forming industry. It is essential to use numerical methods to predict the springback of complex products. Finite element method (FEM) is a widely used numerical method in sheet forming simulation, and is approved to predict springback successfully, but the springback prediction result is affected by many factors, and the accuracy of springback prediction is not satisfactory, deep research on the springback is still necessary.Element sizes, material parameters in simulation model and equivalent drawbead model are three main factors influencing the accuracy of the springback prediction. Focusing on these factors, methods to improve the accuracy of springback prediction in the metal sheet forming are studied. The main work is listed as follows:(1) The size of elements of blank passing through the die/punch shoulder in flanging process is studied, and it is verified that the element size should be enough small when the element passing the die/punch shoulder to get reasonable springback prediction of flanging process. Many researchers haven't reached an agreement on the number of element passing throng the shoulder, but it is generally accepted the element size can be larger if the radius of the shoulder is larger. Author has found this opinion is not suitable when it is used on the flanging springback simulation. In this paper, models of three flanging process are built, and springback predicted by models with different element size are compared with the experimental results. It is shown that springback predicted by the models using small element size of blank in simulation of flanging processes with different die radius are consistent with the experimental results.(2) The influence of element size on springback in U-shape forming is investigated, and why the springback is always under predicted by FEM is explained, which is a problem many researchers have been puzzled with. Experiments of Ushape forming are carried out. Simulations of U shape forming are divided into two phases: at the process and at the end of forming process. The influence of element size on springback prediction is discussed individually for these two phases. Results show that at the process of forming simulation, "stress relaxation" ,which low down the stress, incurs at the bending part of the blank, and the "stress relaxation" is more obvious when the element size is larger. The stresses, which determine the springback, are affected by the "stress relaxation", and thus the springback prediction is affected by the element size. In this situation, springback is smaller in models with big element size for more "stress relaxation" incurs. At the end of forming simulation, penetration is unavoidable between the blank and die/punch. The contact force caused by the penetration has same effect of the force in shape correction at the bending process with die bottom, which will reduce the springback. For the force in shape correction is varied with different element size, so as the springback predicted. In this situation, springback is smaller in models with big element size for big force in shape correction is applied. In this paper, the methods to low down the influence of element size on springback prediction are also suggested.(3) A new inverse method to identify parameters in material models is suggested. The accuracy of material model parameters, which is determined by the description accuracy of mechanical properties of sheet, has great influence on the accuracy prediction of springback. The material identification process can be time consuming if simulation model of experiment is complex. An inverse method, which combined FEM software ANSYS LS-DYNA with modified Levenberg Marquart (LM) optimization algorithm, is utilized to identify material parameters based on tensile test with rectangular specimens. It is time efficient and converges fast. Results show that the material parameters in Barlat 1989 and Barlat 1991 identified by this method describe the mechanical properties well. Using the same inverse method, a new approach to identify the true stress-strain curve based on the blank deformation in the necking process in tensile test. This method can get the true stress-strain curve at large strain region and avoid the shortcoming of normal method, which can get the curve at small strain region only.(4) The influence of drawbead on springback is investigated, and a modified equivalent drawbead model is proposed. The influence of using equivalent drawbead model on springback prediction is not clear. Based on drawbead test and flanging test, the influence of drawbead on springback is studied. Results show that after passingthrough the drawbead, the blank become hardened, and thickness of blank is reduced, and stress is left in the blank, which affect the springback of parts. Even the equivalent model used now can describe the drawbead restraint force accurately; it isn't qualified to predict springback accurately. To avoid the shortage of the equivalent drawbead model, a modified equivalent drawbead model is suggested, which can describe the material hardening, thinning, and stress left accurately besides the drawbead restraint force. Theory method, experiment method and simulation method to calculate parameters in the modified equivalent model are suggested, and the application of this model on simple springback problem is also suggested. Results show the modified equivalent model can improve the accuracy of springback prediction.