Implementation of machine learning strategies for determination of finite width correction factors for orthotropic plates containing central hole

Design of aviation products made of composite materials rely on damage tolerance strength allowables. A common industry practice is the usage of open hole specimens to determine the allowed strain levels for the design. However, if the design introduces a case of an edge distance that is shorter than that of the specimen that was used to determine the strength allowables, a finite width correction factor should be applied to the damage tolerance strength allowables. Such factor depends not only upon the edge distance, but also upon the layup. To this end, a machine learning based methodology is proposed in this study to obtain the finite width correction factor for any given set of layup and geometrical properties. Such approach can efficiently provide an accurate prediction of the finite width correction factor with relatively small amount of test or simulation data. The methodology also includes a procedure to determine the number of required training data points. Three different machine learning algorithms were used in this study, all providing very similar predicted regressions of the entire domain studied. Finite width correction factors carpet plots were produced for IM7/8552 composite, allowing the designer to easily obtain the required correction factor based on the layup and geometrical characteristics.