Integrated experimental and machine learning analysis of Al-Si/CSA/MoS2 hybrid composites for enhanced mechanical and tribological performance

This study introduces a cohesive experimental and statistical modelling methodology to create lightweight Al–Si/CSA/MoS 2 hybrid composites with enhanced mechanical and tribological characteristics for automotive and marine applications. The composites were made using the stir casting method, which included coconut shell ash (CSA: 5–15 wt%) and molybdenum disulphide (MoS 2 : 1–3 wt%) as reinforcements. The developed materials were subjected to microscopical, tribological, and mechanical testing. A Face Centred Composite (FCC) model was used to investigate the effects of CSA, MoS 2 , load, and sliding velocity on wear rate (W R ) and coefficient of friction (CF). The best composite (Al/10 %CSA/3 %MoS 2 ) was 31.9 % raised hardness, a 25.9 % amplify in tensile strength, a 32.5 % decline in density, a 42.2 % reduced W R , and a 23.3 % decreased CF than Al–Si alloy. The FCC based statistical models that were created had R 2 values of more than 0.94, which shows reliable of predictions. Complementary Random Forest regression modeling corroborated the experimental findings, producing R 2 values of 0.894 for W R and 0.925 for CF, with sliding velocity and load recognized as the predominant influencing parameters. Microstructural characterization verified consistent reinforcement distribution, improved interfacial adhesion, and the synergistic strengthening-lubrication action of CSA and MoS 2 . The optimized hybrid composite has better wear resistance, load-bearing capacity, and self-lubricating performance, which makes it a good choice for parts, such as brake rotors, and cylinder liners. • Developed lightweight AMCs via stir casting to incorporate MoS 2 (1-3 wt. %) and CSA (5–15 wt. %) reinforcements, promising for use in marine and automotive parts. • Developed comprehensive microstructure–property correlations utilizing SEM/EDS, XRD, phase mapping, and fractography. • The effects of CSA, MoS 2 , load, and sliding velocity on wear rate and coefficient of friction were investigated using the FCC model and determined outperformed once. • Developed FCC-based RSM and Random Forest models to forecast wear rate and coefficient of friction with less than 10 % validation error. • Validated the robustness of both models by confirming ideal conditions through independent confirmation experiments.