A data-driven Bayesian optimisation framework for the design and stacking sequence selection of increased notched strength laminates

Autor: Andrew Rhead, T.R.C. Chuaqui, Carl Scarth, Richard Butler
Jazyk: angličtina
Rok vydání: 2021
Předmět:
Zdroj: Chuaqui, T R C, Rhead, A T, Butler, R & Scarth, C 2021, ' A data-driven Bayesian optimisation framework for the design and stacking sequence selection of increased notched strength laminates ', Composites Part B: Engineering, vol. 226, 109347 . https://doi.org/10.1016/j.compositesb.2021.109347
DOI: 10.1016/j.compositesb.2021.109347
Popis: A novel Bayesian optimisation framework is proposed for the design of stronger stacking sequences in composite laminates. The framework is the first to incorporate high-fidelity progressive damage finite element modelling in a data-driven optimisation methodology. Gaussian process regression is used as a surrogate for the finite element model, minimising the number of computationally expensive objective function evaluations. The case of open-hole tensile strength is investigated and used as an example problem, considering typical aerospace design constraints, such as in-plane stiffness, balance of plies and laminate symmetry about the mid-plane. The framework includes a methodology that applies the design constraints without jeopardising surrogate model performance, ensuring that good feasible solutions are found. Three case studies are conducted, considering standard and non-standard angle laminates, and on-axis and misaligned loading, illustrating the benefits of the optimisation framework and its application as a general tool to efficiently establish aerospace design guidelines.
Databáze: OpenAIRE