摘要
有限元数值模拟的精度不但与本构模型的描述能力有关,而且与其材料参数的测定方法密切相关。该文以由Hill1948屈服函数和Swift等向强化模型组成的本构模型为例,比较了不同的材料参数测定方法对单向拉伸的轴向应力-轴向应变、横向应变-轴向应变试验数据的预测能力。参数测定方法采用了两种,一种是传统方法,即使用r值计算屈服函数系数,采用轧制方向单向拉伸应力-应变数据拟合强化模型参数;另一种是考虑单向拉伸横向应变演化的反向优化法,即使用不同方向单向拉伸轴向应力-轴向应变和横向应变-轴向应变试验数据,同时求解屈服函数和强化模型的材料参数。结果表明,当使用传统方法时,所得材料参数不能很好描述与轧制方向成45°方向的单向拉伸数据;当使用考虑单向拉伸横向应变演化的反向优化法时,所得材料参数能够较准确描述各个方向的单向拉伸力学性能。
The accuracy of numerical simulation based on finite element method is not only related to the constitutive model,but also closely depended on the parameters identification strategy.Taking the elasto-plastic constitutive model based on Hill1948 yield function and Swift isotropic hardening model as an example,the predicted directional Cauchy stresses and transverse strains were compared based on different parameters identification strategies.One identification strategies used was the traditional method with the anisotropic coefficients of yield function being calculated from constant r-values,and the parameters of hardening model being fitted from the uni-axial tensile Cauchy stress along rolling direction.Another strategy used was the inverse optimization method considering the transverse strain evolution in uni-axial tension with all the parameters of yield function and hardening model being inversely optimized from the experimental directional Cauchy stresses and transverse strains at the same time.The results show that,the traditional parameters identification strategy cannot well predict the Cauchy stress and transverse strain in uni-axial tension at 45° to the rolling direction.While,the inverse optimization method can accurately capture all the directional Cauchy stress and transverse strain due to the introduction of transverse strain evolution in parameters identification.
出处
《塑性工程学报》
CAS
CSCD
北大核心
2012年第1期77-80,共4页
Journal of Plasticity Engineering
基金
国家自然科学基金资助项目(11002105)
省部共建教育部重点试验室开放基金资助项目(10ZXZK03)
教育部博士点基金资助项目(200806981025)
关键词
本构模型
材料参数
数值模拟
反向优化
数字散斑
constitutive model
material model parameters
numerical simulation
inverse method
digital image correlation