HYBRID REPAIR OF CRACKS ON PLATES UNDER THERMO-MECHANICAL LOADING: MODELLING AND OPTIMIZATION

Abstract

Cracked structures pose significant challenges in engineering applications. Cracks can lead to stress concentrations, increasing the risk of catastrophic failure and complicating maintenance efforts. In fields such as aerospace, civil, and mechanical engineering, cracks may arise from factors like material fatigue, environmental conditions, and manufacturing defects. Aircraft structures are frequently subjected to thermal variations due to operating conditions, which can exacerbate crack propagation and compromise structural integrity. While passive repair methods using composite patches have been extensively studied, recent advancements focus on active repair techniques, notably integrating piezoelectric (PZT) actuators for electro-mechanical crack control. However, studies often overlook thermal influences and the impact of adhesive disbonding on repair effectiveness. This study addresses these challenges by investigating the hybrid repair of Mode I cracked plates under thermo-mechanical loading, integrating adhesive disbonding effects and precise PZT patch placement. The research validated the proposed thermal stress intensity factor equation, conducts Finite Element Analysis (FEA) using ANSYS software, and assesses the influence of temperature variations and mechanical loads on hybrid repair using composite and PZT actuators. Additionally, various adhesive disbond scenarios in composite and PZT patches are examined to evaluate repair efficacy. Furthermore, this study employs novel Machine Learning (ML) techniques to optimize hybrid repair under thermo-mechanical loading, selecting five key parameters for ML optimization and training using FE simulation data. Results indicate that temperature fluctuations significantly impact repair efficiency and fatigue life. Importantly, strategic positioning of the PZT patch directly above or on the crack tip significantly enhances repair efficacy, preventing further propagation and optimizing stress mitigation. Disbonding, however, diminishes repair efficacy from 85% to 30% and accelerates crack propagation. In the study of optimization and prediction using Machine Learning, the findings indicate that the Squared Exponential Gaussian Process Regression (GPR) model demonstrated superior accuracy, effectively minimizing errors, and enhancing predictive performance. Particle Swarm Optimization identified Graphite/epoxy as the optimal material, achieving a minimum SIF of 0.163 with parameters: composite patch thickness of 0.633 mm, PZT patch size of 250 mm�, PZT thickness of 1 mm, and PZT distance of 0.75 mm. This study highlights the importance of considering thermal influences and adhesive disbonding in hybrid repairs, emphasizing the need for robust repair techniques under thermo-mechanical loading conditions. The results offer valuable insights into improving repair efficiency and predictive capabilities.