Optimal design of thin-layered composites for type IV vessels: Finite element analysis enhanced by ANN
FanDing Li,XueDong Chen,Peng Xu,ZhiChao Fan,Quan Wang,ChengSi Lyu,Qian Zhang,Hao Yu,HengAn Wu
DOI: https://doi.org/10.1016/j.tws.2023.110752
IF: 5.881
2023-04-26
Thin-Walled Structures
Abstract:For a type IV hydrogen storage vessel, the complexity of the mechanical system is a result of the increased number of thin-layered composites and geometrical details at dome parts, which significantly prolongs the computation time for traditional numerical analysis and optimal design. In this work, a framework couples machine learning (ML) and finite element (FE) analysis is proposed for a 70 MPa type IV vessel. Providing an approach that allows the winding parameters optimization process to include the corresponded feasible dome geometry. High-fidelity FE models are established and different failure modes are included. The artificial neural network (ANN) is developed and coupled with the FE models, where the irregularity of directional material distribution on both cylindrical and dome regions is adequately considered during the optimization process. Winding parameters extended by practical production modifications such as transition areas and layer ending adjustments are introduced to the modified input layer of ANN, to ensure the integrality of the complex geometrical details. The computational cost is greatly reduced with satisfying accuracy compared to FE analysis, where the optimized lay-up scheme could be obtained with a prediction error of less than 2% on the damage state function. The associated simulation results show an obvious improvement in mechanical response under designed burst pressure. Particularly, the prototypical vessels are further produced to carry out comparative experiments , which indicate that the burst pressure of the type IV vessel has been improved from 145 MPa to 157.74 MPa. This is also consistent with the numerical prediction of burst pressure using Hashin failure criteria and progressive damage analysis.
engineering, mechanical,mechanics, civil