Optimizing Thermal-Elastic Properties of C/C–SiC Composites Using a Hybrid Approach and PSO Algorithm

• Published in: Materials Vol. 9; no. 4; p. 222
• Main Authors: Xu, Yingjie, Gao, Tian
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 23.03.2016; MDPI
• Subjects: Aerospace engineering; Algorithms; Back propagation; Carbon; Composite materials; Heterogeneity; Mathematical models; Microstructure; Neural networks; Optimization; Scanning electron microscopy; Silicon carbide; Swarm intelligence; C/C–SiC composites; thermal-elastic properties; BP neural network; micromechanical modeling; particle swarm optimization algorithm
• ISSN: 1996-1944; 1996-1944
• DOI: 10.3390/ma9040222

Carbon fiber-reinforced multi-layered pyrocarbon–silicon carbide matrix (C/C–SiC) composites are widely used in aerospace structures. The complicated spatial architecture and material heterogeneity of C/C–SiC composites constitute the challenge for tailoring their properties. Thus, discovering the intrinsic relations between the properties and the microstructures and sequentially optimizing the microstructures to obtain composites with the best performances becomes the key for practical applications. The objective of this work is to optimize the thermal-elastic properties of unidirectional C/C–SiC composites by controlling the multi-layered matrix thicknesses. A hybrid approach based on micromechanical modeling and back propagation (BP) neural network is proposed to predict the thermal-elastic properties of composites. Then, a particle swarm optimization (PSO) algorithm is interfaced with this hybrid model to achieve the optimal design for minimizing the coefficient of thermal expansion (CTE) of composites with the constraint of elastic modulus. Numerical examples demonstrate the effectiveness of the proposed hybrid model and optimization method.