Application of Deep Learning for Stress Prediction Under Fan Blade Out Loads in Turbofan Engine

Abstract Ensuring the structural integrity of critical engine components in turbofan engines under extreme conditions such as fan blade out (FBO) is paramount for aerospace safety and performance. FBO events, characterized by the sudden detachment of a fan blade due to high-stress conditions, can generate extreme loads on the order of several thousand g’s, posing significant risks to engine integrity and potentially leading to catastrophic failures. To check the structural integrity of critical engine components under extreme conditions, finite element analysis (FEA) was utilized. However, FEA can be slow for a large number of outputs since it solves complex equations. Therefore, in this study deep neural network (DNN) a cutting-edge approach was used to predict the stress, strain, and deformation of bearing support in turbofan engine under FBO loads instead of solving equations with using FEA. The proposed approach DNN is trained with the results of extremum points of FEA and the structural responses for the remaining time points were predicted. To confirm the accuracy of the deep learning model, a comparison is made between DNN predictions and conventional FEA results. As a result, deep learning model exhibited exceptional accuracy, with prediction deviations between 0.19% and 7.4% compared to conventional FEA results. This close alignment underscores the model’s reliability and potential as a faster, efficient alternative to traditional methods by significantly reducing computational time while maintaining high precision. The findings demonstrate the deep learning model’s ability to accurately capture complex stress and strain patterns induced by FBO events, suggesting its integration could enhance the safety and reliability of gas turbine engines. This research marks a significant advancement in applying artificial intelligence (AI) techniques to critical aerospace engineering challenges, promising improved performance and safety outcomes during severe operational events like FBO.