Numerical Methods and Applications - 9th International Conference, NMA 2018, Borovets, Bulgaria, August 20-24, 2018, Revised Selected Papers
Numerical Methods and Applications - 9th International Conference, NMA 2018, Borovets, Bulgaria, August 20-24, 2018, Revised Selected Papers
复制标题
数值方法与应用 - 第九届国际会议，NMA 2018，保加利亚波罗维茨，2018 年 8 月 20-24 日，修订选录论文
DOI：
10.1007/978-3-030-10692-8_46
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Houghton M
中科院分区：
文献类型：
--
作者：
Houghton M
对代表现实世界材料的随机纤维网络进行建模，得到一个具有高条件数的大型稀疏线性矩阵系统。对于实际材料中大量随机纤维的力学性能，目前的离晶格网络并不是一个现实的模型。在本文中，我们提出了二维和三维模型中采用的数值方法，这些方法改善了现有非晶格模型中所见的计算时间限制。具体来说，我们给出了在不同预条件选择下迭代求解的二维随机光纤网络的性能比较，然后给出了三维模型的一些初步结果。
Modelling a random fibre network representative of a real world material leads to a large sparse linear matrix system with a high condition number. Current off-lattice networks are not a realistic model for the mechanical properties of the large volume of random fibres seen in actual materials. In this paper, we present the numerical methods employed within our two-dimensional and three-dimensional models that improve the computational time limitations seen in existing off-lattice models. Specifically, we give a performance comparison of two-dimensional random fibre networks solved iteratively with different choices of preconditioner, followed by some initial results of our three-dimensional model.