PSI - Issue 71

Akash S.S. et al. / Procedia Structural Integrity 71 (2025) 180–187

187

processing. The mean, standard deviation and number of fibers of the fiber diameter distribution in each SVE is given as input to the trainedANN model and the corresponding elastic properties are predicted by them. The predicted elastic properties are thenused to construct the probability density function to represent uncertainty. The ANN model is trained using data generated by running simulations on FE models of SVE cells with varying fiber diameters and number of fibers. The model is validated using the test data generated and was found to be accurate. The model was also validated using a new set ofdata obtained from the microstructure image [Figure 4(b)]. The results obtained using the model was found to be matching with the FE results with good accuracy as it can be observed in Figure 6. Contrary to the traditional methods involving FEsimulations, the ANN model saves computational time and requires less computational power after training. Hours ofFE simulations for uncertainty quantification is being replaced by seconds of computation time by the ANN model. The present method can be applied to any type of composites like cross-ply 2D woven composites. The basic principle behind the method, i.e., generating data for training through FE analysis of various SVEs representing the type of composite material accounting all uncertainties being studied and using the data to train and generate the optimized ANN model, should be preserved. Based on the microstructural images, appropriate image processing techniques should be adapted. Acknowledgements This research is a part of the undergraduate project of Mr.Akash S S. The authors would like to acknowledge the support received from IIST and CMSE/VSSC for the research. References Chahar, R. and Mukhopadhyay, T. 2023. Multi-fidelity machine learning based uncertainty quantification of progressive damage in composite laminates through optimal data fusion, Engineering Applications of Artificial Intelligence 125, 106647. C.T. Sun, R.S. Vaidya. 1996. Prediction of composite properties from a representative volume element, Composites Science and Technology 56 (2), 171-179 Dey, S., Mukhopadhyay, T., Spickenheuer, A., Gohs, U. and Adhikari, S. 2016. Uncertainty quantification in natural frequency of composite plates - an artificial neural network-based approach, Advanced Composites Letters 25(2), 096369351602500203. Dinesh Kumar, Yao Koutsawa, Gaston Rauchs, Mariapia Marchi, Carlos Kavka, Salim Belouettar.2020. Efficient uncertainty quantification and management in the early stage design of composite applications, Composite Structures 251, 112538 Dženan Hozić, Carl -Johan Thore, Christopher Cameron, Mohamed Loukil.2023. Material uncertainty quantification for optimized composite structures with failure criteria, Composite Structures 305, 116409 Fernandez, J., Chiachio, M., Chiachio, J., Munoz, R. and Herrera, F. 2022. Uncertainty quantification in neural networks by approximate bayesian computation: Application to fatigue in composite materials, Engineering Applications of Artificial Intelligence 107, 104511. Guillermo Requena, Georg Fiedler, Bernhard Seiser, Peter Degischer, Marco Di Michiel, Thomas Buslaps. 2009, 3D Quantification of the distribution of continuous fibres in unidirectionally reinforced composites, Composites Part A: Applied Science and Manufacturing 40 (2),152-163 Karnik, S., Gaitonde, V., Rubio, J. C., Correia, A. E., Abrao, A. and Davim, J. P. 2008, Delamination analysis in high-speed drilling of carbon fiber reinforced plastics (cfrp) using artificial neural network model, Materials Design 29(9), 1768-1776. O’Malley, Tom and Bursztein, Elie and Long, James and Chollet, Franc¸ois and Jin, Haifeng and Invernizzi, Luca and others, Keras Tuner, 2019 Omairey, S.L., Dunning, P.D. & Sriramula, S.2019. Development of an ABAQUS plugin tool for periodic RVE homogenisation. Engineering with Computers 35, 567–577. Sepahvand, Kian K. 2021. Deep Learning Based Uncertainty Analysis in Computational Micromechanics of Composite Materials Applied Mechanics 2, no. 3: 559-570. Xiaolei Yin, Wei Chen, Albert To, Cahal McVeigh, Wing Kam Liu. 2008. Statistical volume element method for predicting microstructure–constitutiveproperty relations, Computer Methods in Applied Mechanics and Engineering 197 (43–44), 3516-3529