A Framework for Predicting Corrosion Resistance in Aluminum Alloy Composites

Thesis Abstract

Corrosion resistance in aluminum alloy composites remains a critical challenge impacting their durability and application in aerospace, automotive, and marine industries. Despite extensive research on metallurgical compositions and surface treatments, predicting corrosion behavior with high accuracy in diverse operational environments remains unresolved. This study aims to develop a comprehensive predictive framework for assessing corrosion resistance in aluminum matrix composites reinforced with alumina and silicon carbide particulates. Specific objectives include identifying key material and processing parameters influencing corrosion, formulating a predictive model using statistical and machine learning techniques, and validating the framework through experimental corrosion testing. The research adopts a mixed-methods approach, integrating quantitative experimental data with qualitative insights from material characterization processes. A cross-sectional population comprising 150 aluminum alloy composite samples, fabricated via stir casting with varying reinforcement percentages (5%, 10%, and 15%) and different surface treatments, serves as the basis for the study. Stratified random sampling was employed to select samples based on reinforcement type, percentage, and fabrication parameters, ensuring representative variability. Data collection involved the use of electrochemical impedance spectroscopy (EIS), potentiodynamic polarization tests, and surface analysis via scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS). The framework's development is underpinned by the Theory of Materials Degradation and the Fracture Mechanics Theory, providing a robust conceptual foundation linking material microstructure to corrosion mechanisms. Data analysis encompasses advanced statistical methods, including multiple regression analysis, to identify significant predictors of corrosion resistance, and machine learning algorithms—particularly random forest models—to enhance predictive accuracy. The model's performance is evaluated through metrics such as R-squared, mean squared error (MSE), and cross-validation techniques. Additionally, analysis of variance (ANOVA) tests determine the significance of differences among various reinforcement levels and surface treatments. Expected findings indicate that reinforcement type, percentage, and surface treatment parameters significantly influence corrosion resistance, with the developed machine learning framework outperforming traditional statistical models in predictive accuracy. The study anticipates establishing quantifiable relationships between microstructural features—such as phase distribution, interfacial bonding, and porosity—and corrosion behavior, thereby enabling reliable predictions based on fabrication and material parameters. The validated framework aims to serve as a decision-support tool for engineers and materials scientists, facilitating the selection of optimal composite configurations for corrosion-prone environments. This research contributes substantial theoretical and practical advancements to materials engineering by integrating the principles of corrosion science, machine learning, and materials characterization into a unified predictive model. It fills existing knowledge gaps concerning the microstructure-property linkages and provides a validated, scalable framework applicable across various aluminum composite formulations. The study's main conclusion underscores the importance of tailored reinforcement strategies and surface modifications for improving corrosion resistance, supported by a robust, data-driven predictive model. Based on the findings, the study recommends adopting the framework for pre-fabrication assessments in manufacturing settings and prompts further investigation into real-time corrosion monitoring integrated with machine learning models. Future research suggestions include extending the model to other metal matrix composites, exploring additional environmental variables such as chloride ion concentration, and incorporating in-situ corrosion testing to refine model accuracy. Overall, this research establishes a significant step towards predictive materials design, promising enhanced durability and sustainability for aluminum alloy composites in diverse industrial applications.