Volume 48, No 3, 2026, Pages 474-495
Download full text in PDFOptimizing Solid Particle Erosion Wear Behaviour of the Flax Fiber Reinforced Epoxy Composites: Experimental Study Using RSM and ANFIS Approach
Authors:
S.M. Vinu Kumar 
,
N. Manikandaprabu ,
R. Hemanth ,
E. Sakthivelmurugan ,
N. Santhosh
DOI: 10.24874/ti.2026.09.25.12
Received: 19 September 2025
Revised: 12 November 2025
Accepted: 4 December 2025
Published: 15 September 2026
Abstract:
Flax fiber reinforced epoxy (F-E) composites were prepared by hand layup technique followed by curing in compression moulding machine and were designated as 20F-E, 30F-E, and 40F-E, based on their fiber content. These laminates were subjected to solid particle erosion wear test in accordance with ASTM G76. With the help of Taguchi’s design of experiments (L27 orthogonal arrays), tests were conducted for the different combinations of input parameters namely, impact velocities (72, 100, and 129 m/s), impingement angles (60, 75, and 90 degree), and fiber contents (20, 30, and 40 wt. %). Measured output responses of the study were erosion wear rate (EWR) and erosion efficiency (η). Results showed that the minimum EWR of 0.000165 g/g was observed for 40F-E composites under impact velocity 72 m/s, and impingement angle of 90°. Response surface methodology (RSM) and Adaptive Neuro-Fuzzy Inference System (ANFIS) models were developed for the EWR and their coefficient of determination (R2) values were found to be 0.9827 and 0.9999, respectively. ANOVA reveals that impact velocity was the most significant factor influencing the EWR. Furthermore, validation test confirms that EWR of the ANFIS results were closer to experimental values than RSM model and hence proved its prediction accuracy. Worn damages like, crater formation, deep grooves, micro cutting, ploughing with partial deformations were predominant and exposed by field emission scanning electron microscope (FESEM).
Keywords:
Erosion efficiency, Flax fiber, Epoxy composites, Analysis of variance, Response surface method, Adaptive neuro-fuzzy inference system
