Image-Based Predictive Modeling of Cavitation Wear Evolution for Service-Life Estimation

Cavitation erosion is a complex tribological degradation process that significantly affects the durability and reliability of engineering materials operating in aggressive liquid environments. Accurate prediction of damage evolution is therefore essential for optimizing maintenance strategies and extending service life. This study presents a non-destructive methodology for predicting cavitation resistance by integrating surface image analysis, principal component analysis (PCA), and strength degradation modeling. The methodology was demonstrated using a sintered talc-based refractory ceramic containing natural zeolite as a representative material exposed to cavitation according to the ASTM G32 ultrasonic vibratory test. High-resolution surface images acquired after successive exposure intervals were analyzed to quantify damage development through morphological descriptors of cavitation pits. PCA was employed to reduce data dimensionality and identify the descriptors that most effectively characterize the evolution of surface degradation. Area, perimeter, and maximum pit diameter were identified as the dominant parameters governing damage progression, while less informative descriptors were excluded from further analysis. Based on the quantified surface deterioration, a predictive model describing compressive strength degradation was established, enabling estimation of the critical exposure time corresponding to a predefined residual strength level. The results demonstrated a nonlinear evolution of cavitation wear and showed excellent agreement between image-derived parameters and the predicted mechanical response. The proposed methodology provides an efficient framework for non-destructive monitoring of cavitation-induced degradation and supports predictive assessment of material performance without extensive mechanical testing. The approach can be applied to a wide range of engineering materials subjected to tribological damage mechanisms such as wear, corrosion, thermal shock or their combined action, offering a valuable tool for service-life prediction and reliability assessment.