Monday  |  Milwaukee  |  10:40 AM–11:00 AM

#19730, Bayesian Data Fusion to Calibrate CFRP Material Properties Based on Experiments that Utilise a Modified Arcan Fixture

Design and development programs for high-performance composite structures rely on integrated physical testing, model predictions, and model validation across the length scales, including at the coupon, subcomponent, and full structural scales. Recent progress in the development of high-fidelity simulation frameworks for composite materials and structures, commonly based on the Finite Element (FE) method, have the potential to significantly accelerate the current design development approaches by reducing the reliance on time-consuming and costly physical tests. The integration of advanced virtual testing requires new methodology for the systematic integration/fusion of experimental and numerical results. Therefore, Bayesian analysis is proposed to statistically calibrate models of composite structures based on experimental and FE model data. In the present paper, the Bayesian approach is demonstrated using specimens manufactured from multidirectional aerospace grade carbon fibre reinforced polymer (CFRP) laminates with an open hole. The specimens were subjected to shear, combined tension-shear, and compression-shear loading within the linear part of the load response using a Modified Arcan Fixture (MAF), and full-field displacement data was obtained using Digital Image Correlation (DIC). It is demonstrated that the Bayesian analysis, informed by only three MAF configurations, can produce accurate values of the CFRP elastic properties using the data from the multidirectional 8-ply quasi-isotropic [+45/90/-45/0]s specimens. Uncertainties in the applied loading associated with the MAF are also accounted for and shown to have little effect on the outcome. It is demonstrated that the Bayesian approach can identify updated (posterior) material elastic properties for future use in structural models as well as providing uncertainties associated with material properties and experimental loading conditions. The methodologies presented pave the way for smarter, integrated testing and modelling of complex composite structures at the higher length scales of the testing pyramid.

Sinan Xiao   University of Bath
Tobias Laux   University of Bristol
Karim Anaya-Izquierdo   University of Bath
Janice Dulieu-Barton   University of Bristol