Enhanced Wear Resistance of Epoxy Composites Through the Incorporation of Diatom Frustules: a Multi-Objective Optimization Approach

Abstract

The present work investigates the enhancement of wear resistance in epoxy composites through the incorporation of calcined and natural diatom frustules (CDFs and NDFs) as reinforcing fillers. The CDFs, pre-calcined at 1200 degrees C during manufacturing to improve structural integrity and eliminate organic matter, were supplied in processed form. Both CDFs and NDFs were subsequently wet-sieved (below 325 mesh) and dried at 120 degrees C for 2 h to ensure particle uniformity and moisture removal. Epoxy composites were prepared with 5-20 wt% frustule content. The fillers were ultrasonically dispersed in the epoxy matrix to improve uniformity and reduce agglomeration, followed by vacuum degassing and thermal curing. Wear performance was initially evaluated for all samples at a fixed 1000-cycle duration. Based on preliminary results, composites with 15 wt% and 20 wt% filler content which showed the highest wear resistance, were further tested under varying sliding distances corresponding to 300-1000 cycles to examine long-term behavior. Tests were conducted under dry sliding conditions using a block-on-ring tribometer at 50 N load. Using a systematic modeling-design-optimization framework, the study defines diatom weight fraction, sliding test duration, and frustule type as design variables. The experimental process was modeled through multiple nonlinear neuro-regression analyses, selecting the most realistic model based on Rtraining2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R_{{{\text{training}}}}<^>{2}$$\end{document}, Rtesting2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R_{{{\text{testing}}}}<^>{2}$$\end{document}, Radjusting2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R_{{{\text{adjusting}}}}<^>{2}$$\end{document}, and stability evaluations from 13 functional structures, with a second-order trigonometric nonlinear type model (SOTN) as the highest predictive performance. Stochastic optimization methods-including Modified Differential Evolution (MDE), Modified Nelder-Mead (MNM), Modified Simulated Annealing (MSA), and Modified Random Search (MRS)-were employed under three design scenarios to determine optimal wear parameters. The results revealed that epoxy composites containing 15 wt% NDFs exhibited the most substantial improvement, with a 95% reduction in specific wear rate (SWR) compared to neat epoxy and a 60% reduction relative to CDF-filled composites. The lowest optimized specific wear rate achieved was 1.086 x 10-5 mm3/N<middle dot>m. This work offers a comprehensive framework integrating material processing, statistical modeling, and stochastic optimization for the design of high-performance, wear-resistant epoxy composites.