Robustness of deep-drawing finite-element simulations to process variations
Robustness of deep-drawing finite-element simulations to process variations
复制标题
DOI：
10.1007/s12289-022-01695-3
复制
发表时间：
2022-05
影响因子：
2.4
通讯作者：
Kelin Chen;A. Breunig;J. Ha;B. Kinsey;P. Groche;Y. Korkolis
中科院分区：
文献类型：
--
作者：
Kelin Chen;A. Breunig;J. Ha;B. Kinsey;P. Groche;Y. Korkolis
数值模型的稳健性为有效补偿导致偏离标称条件的扰动铺平了道路。如果数值模拟将用于确定闭环过程控制调整以确保最终零件的质量，这一点至关重要。这项工作通过使用 3D 伺服压力机对 AA1100-O 毛坯进行深拉深研究，详细介绍了建立和验证数值工艺模型的过程。特别令人感兴趣的是深拉仿真模型对不同工艺变化和非设计条件的鲁棒性。实验在 3D 伺服压力机上进行，用作传统压力机，并配备弹簧压边圈。从实验中，获得了冲头力-位移以及局部特征，即凸缘拉紧和壁薄化。创建了两种类型的绘图过程有限元模型，一种使用壳，另一种使用实体单元。相应地，毛坯的塑性各向异性使用 Yld2000-2d (2D) 和 Yld2004-18p (3D) 屈服函数进行建模。通过将模拟的冲头力-位移响应、凸缘拉紧和厚度变化与实验结果进行比较，反算毛坯和模具之间的摩擦系数。数值和材料模型的稳健性通过毛坯几何形状的工艺变化（即毛坯中心和椭圆形毛坯的初始偏移）得到证实。然而，模型并未捕获由于压边力变化而导致的法兰起皱。对模型进行修改，即通过向毛坯引入适当的几何缺陷，使其能够预测法兰起皱。这项工作研究了数值模型对不同类型过程变化的鲁棒性，这在基于模型的控制分析中至关重要。
Robustness of numerical models paves the way for efficient compensation of perturbations resulting in deviations from the nominal conditions. This is critical if the numerical simulations will be used to determine closed-loop process control adjustments to assure the final part quality. This work details the procedure to establish and validate numerical process models, through an investigation of deep-drawing of AA1100-O blanks using 3D Servo Press. Of particular interest is the robustness of the deep-drawing simulation models to different process variations and off-design conditions. The experiments are performed on a 3D Servo Press, used as a conventional press, and equipped with a spring-loaded blank holder. From the experiments, the punch force–displacement as well as local features, i.e., flange draw-in and wall-thinning, are obtained. Two types of finite element models of the drawing process are created, one using shell and the other using solid elements. Correspondingly, the plastic anisotropy of the blanks is modeled using the Yld2000-2d (2D) and Yld2004-18p (3D) yield functions. The friction coefficient between the blank and tooling is inversely identified by comparing the simulated punch force–displacement response, flange draw-in and thickness variations with the experimental ones. The robustness of the numerical and material models is confirmed by process variations on the geometry of the blanks, i.e., an initial offset of blank center and elliptical blanks. However, the wrinkling of the flange due to variation of the blank holder force is not captured by the model. A modification to the model, i.e., by introducing appropriate geometric imperfections to the blank, enables it to predict the flange wrinkling. This work investigates the robustness of numerical models to different types of process variations, which is vital in model-based control analyses.