Solid Particle Erosion and Predictive Modeling of Epoxy Composites Reinforced With Diatom Frustules

ABSTRACT This study examines the solid particle erosion wear behavior of epoxy composites reinforced with 5–20 wt% calcined diatom frustules (DFs), a sustainable bio‐silica filler containing 89 wt% SiO 2 . Erosion tests were performed according to ASTM G76‐07 using garnet erodent particles at six impingement angles (30°–90°) and a velocity of 32 m/s. All composites exhibited ductile erosion characteristics, with the maximum material loss observed at 30° oblique incidences. Increasing the DFs content significantly decreased the erosion resistance at oblique angles (30°–45°), with the 20 wt% composite showing a reduction of up to 48% compared to the neat epoxy. At normal impingement (90°), erosion rates of neat and filled composites were comparable, indicating the limited role of DFs addition under perpendicular impact. The reduction in erosion resistance is attributed to the brittle, silica‐rich nature of the fillers. X‐ray diffraction analysis confirmed enhanced crystallinity and improved filler dispersion with higher DF loading. To predict and optimize erosion response, Response Surface Methodology (RSM) and Artificial Neural Networks (ANN) were employed. RSM revealed significant interactions between angle and filler fraction ( R 2 = 0.9892), while ANN achieved excellent generalization performance ( R > 0.99) using limited data. These results provide new insights into the erosion mechanisms of bio‐silica composites and demonstrate the effectiveness of statistical–computational modeling for material design optimization in abrasive environments.