PSI - Issue 57

Khashayar Shahrezaei et al. / Procedia Structural Integrity 57 (2024) 711–717 K. Shahrezaei et al. / Structural Integrity Procedia 00 (2023) 000–000

714

4

Details about the model and bc and justification of RVEs can be found in previous publications by the authors Eliasson et al 2022.

3. Case Study - Micromechanical Simulations

The case study represents a micromodel of a CFRP material. A challenge for composite materials is that their be havior is heavily influenced by the choice of manufacturing process. Voids are one of the most common manufacturing defects (Mehdikhani et al., 2019) and voids also have a pronounced e ff ect on the fatigue strength and long-term prop erties of composite material (De Almeida and Neto, 1994; Maragoni et al., 2021). The local porosity of a composite material a ff ects the local material sti ff ness (Eliasson et al., 2022b), and the shape of the voids can distinctively impact the various moduli (Huang and Talreja, 2005). Porosity also has a detrimental e ff ect on sti ff ness reduction develop ment under fatigue loading (Sisodia et al., 2015), which plays an important part in damage development (Eliasson et al., 2022b), hence for fatigue life prediction (Liu et al., 2018). A locally varying material sti ff nesswill a ff ect fatigue life and fatigue damage development. Monitoring the sti ff ness loss over time is a common method to evaluate fatigue damage (Eliasson et al., 2022a). To understand the influence of void shape and void size for the CFRP material, void data is extracted from a microscopic analysis. A Representative Volume Element (RVE) is utilized to simulate the local influence of porosity using extracted experimental data. RVEs are commonly used in MSM frameworks. From the RVE simulations the e ff ective macrome chanical properties can be extracted utilizing computational homogenization. The computationally expensive RVE simulations are replaced by the cheap-to-evaluate metamodels. The e ff ective material properties that are extracted are Young’s moduli, E xx , E yy , E zz , and the shear moduli, G xy , G xz , G yz . These are the output of interest for the metamodels and further fatigue life prediction. The RVE model is generated based on experimental microgeometrical data dis tributing fibers and varying the void shape and size (Figure 2), with regards to experimental data. The RVE size is fixed to 200 µ m and the void is assumed to have an elliptical shape. Details about the RVE model, boundary condi tions, and modeling justifications of RVE can be found in previous publications by the authors Eliasson et al. (2022a). The RVE models are run in the Finite Element (FE) software Abaqus. The elliptical void is parameterized using the input factors: angle , major axis , and ratio , and these are the uncertain input parameters. The angle variation is be tween 0 ◦ to 90 ◦ , the major axis is < 75 µ m, and the ratio is obtained by dividing the major axis by the minor axis, being > 1. An automated simulation process is set up using LHS to create the preliminary ED. For the GSA the angle has a uniform distribution and the major axis and ratio have a lognormal distribution (Figure 1) extracted from the experimental data of the CFRP material.

(a)

(b)

Fig. 2: (a) Micrograph with the identification of the elliptical fit for a void in the material, and (b) the FE model with a parameterized void.