均质化(气候)
材料科学
微观结构
分子动力学
周期边界条件
原子堆积因子
体积分数
纤维
复合材料
生物系统
边值问题
机械
数学
数学分析
物理
量子力学
生物
生态学
生物多样性
核磁共振
作者
Marisa Bisram,Jannat Ahmed,Adrian Hood,Jian Cao
标识
DOI:10.1016/j.compscitech.2022.109849
摘要
Composites have a spectrum of mechanical behavior resulting from uncertainty and variation in the microstructure shape and position. With microstructural simulation of fiber architectures, complex numerical algorithms are needed to enforce boundary constraints such as fiber-to-fiber overlap or periodic boundary conditions. Variations are not typically modeled because randomization of unique aggregate shapes can be challenging and current methods for doing so do not achieve high fiber volume fraction (FVF) when packing; however, this can cause inaccurate predictions in simulations. To address this challenge, a novel approach is developed using Molecular Dynamics (MD), achieving high FVFs up to 85% with rapid speed, easy user implementation, and a potential for capturing non-uniform fiber geometry. Two types of packing are presented to demonstrate the scope of possibilities with this approach: a post randomization molecular dynamics (PRMD) method for uniform fibers, and a mixed aggregate randomization using varying radii circles. Comparisons between periodic hexagonal, randomized, and variable radii packing fiber microstructures are performed to determine the significance of randomization and non-uniform aggregate shape for mechanical property homogenization and damage prediction in a unidirectional laminate. It was found that homogenization between the three microstructures do not vary from randomization; however, using the molecular dynamics approach randomizing fiber position and aggregate shape resulted in significant differences in predicting the transverse damage initiation, reducing the error by approximately 25% from the periodic method.
科研通智能强力驱动
Strongly Powered by AbleSci AI