An ABAQUS® plug-in for generating virtual data required for inverse analysis of unidirectional composites using artificial neural networks

Abstract

This paper presents a robust ABAQUS® plug-in called Virtual Data Generator (VDGen) for generating virtual data for identifying the uncertain material properties in unidirectional lamina through artificial neural networks (ANNs). The plug-in supports the 3D finite element models of unit cells with square and hexagonal fibre arrays, uses Latin-Hypercube sampling methods and robustly imposes periodic boundary conditions. Using the data generated from the plug-in, ANN is demonstrated to explicitly and accurately parameterise the relationship between fibre mechanical properties and fibre/matrix interphase parameters at microscale and the mechanical properties of a UD lamina at macroscale. The plug-in tool is applicable to general unidirectional lamina and enables easy establishment of high-fidelity micromechanical finite element models with identified material properties.

A review on micromechanical modelling of progressive failure in unidirectional fibre-reinforced composites

“…However, it was found that the average Young's modulus and strength of the interphase are around 5 and 9 times larger than those of the bulk resin matrix [44], and the interphase was modelled as a separate zone with the same constitutive and damage model but different mechanical properties from the matrix, which makes the model more complicated. Therefore, the parameter identification of the interphase of the carbon fibre-reinforced composite was conducted by the inverse strategy based on the experimental data, microstructural modelling method and Kriging metamodel [47] and artificial neural networks [48], including the identification of thickness, normal and shear stiffnesses of fibre/matrix interface and transverse stiffness of fibres, which are challenge to be determined experimentally. These sets of parameters was applied to predict the elastic and strength properties of CFRP composite yarn [49] and failure analysis of CFRP composites under multiaxial loadings [50,51].…”

Comprehensive inter-fibre failure analysis and failure criteria comparison for composite materials using micromechanical modelling under biaxial loading

“…The initial response of the cohesive model is assumed to be linear elastic governed by penalty stiffnesses (K nn -normal, K ss -shear longitudinal and K tt -shear transversal), which should be large enough to ensure displacement continuity. Here a machine learning based approach was used to determine the stiffnesses, 26,34 which can be found in Table 1. Damage onset is controlled by a quadratic interaction criterion, and the damage occurs when the criterion involving the sum of nominal stress ratios reaches one.…”

A micromechanics and machine learning coupled approach for failure prediction of unidirectional CFRP composites under triaxial loading: A preliminary study

“…Thus, a thickness of 5 µm for the RVE was selected in this study as a compromise between the accuracy of the results and the computing efforts, resulting in a RVE of 50 µm ×50µm× 5µm. An identified interface thickness using an artificial neural network from our previous work [31] was utilised in the construction of the 3D RVE model. More details will be described later in Section 4.1.…”

“…Due to the absence of some micro-parameters from experiments, a parameter identification plugin tool developed for ABAQUS in our previous work [31] was adopted to identify these parameters. Using the plugin, 1000 samples from Latin hypercube sampling of unit cell RVEs with different micro-parameters were generated and the corresponding macro-parameters were calculated, as illustrated in Fig 9Fig 9.…”