This paper presents a novel, meshfree-based micromechanical model for homogenising the elastic properties and analysing the deformation and microscopic strains/stresses of twill woven composites. The proposed model was based on a minimum unit cell (mUC) whose internal features such as the cross-sectional shape and waviness of yarns were described by using sophisticated functions. The boundary conditions imposed on the mUC were derived by applying an equivalence approach, which converts the standard form of periodic boundary conditions into a generic set of fixed and relative displacement constraints. Theoretical formulations were developed to implement the micromechanical model within the framework of the moving kriging (MK)-based element-free Galerkin (EFG) method. An in-house computer program implementing the proposed model was developed for analysing a typical twill woven composite. Good agreements were found between the meshfree-based predictions and the reference results, highlighting the proposed model capable of homogenising twill woven composites and meanwhile avoiding the commonly required pre-processing tasks such as building an explicit geometry model and generating identical meshes on the mapping surfaces to enforce boundary conditions. Three case studies were also performed to identify the sensitivities of the predicted results to three numerical parameters, i.e. the total number of field nodes, the total number of background cells, and the support domain scaling factor. The results of these studies suggest that the numbers of field nodes and background cells used must be sufficiently large, while the support domain scaling factor in an appropriate range (e.g. 2.0–3.25) to achieve convergent results.

本文提出了一种新颖的，基于无网格的微力学模型，用于均化弹性特性并分析斜纹编织复合材料的变形和微观应变/应力。提出的模型基于最小单位单元（mUC），该单元的内部特征（例如横截面形状和纱线的波纹度）通过使用复杂功能进行描述。施加在mUC上的边界条件是通过应用等效方法得出的，该方法将周期性边界条件的标准形式转换为固定和相对位移约束的通用集合。开发了理论公式以在基于移动克里格（MK）的无元素Galerkin（EFG）方法的框架内实现微机械模型。开发了一个内部计算机程序来执行该模型，以分析典型的斜纹编织复合材料。在基于无网格的预测与参考结果之间发现了很好的一致性，突出了所提出的模型能够使斜纹编织复合材料均质化，同时避免了通常需要的预处理任务，例如建立显式几何模型并在映射表面上生成相同的网格加强边界条件。还进行了三个案例研究，以确定预测结果对三个数值参数的敏感性，即现场节点总数，背景单元格总数和支持域缩放因子。这些研究的结果表明，使用的场节点和背景单元的数量必须足够大，