Analysis and prediction of erosion behavior of short kenaf fiber‐based hybrid composites using response surface method integrated with neural networks

Abstract

AbstractThere has been a huge demand in industries for the design and development of bio‐degradable and low‐cost wear resistant composite materials since long. In response to this, the present work includes an exhaustive analysis for the solid particle erosion characteristics of a new class of hybrid polyester composites reinforced with a natural fiber (kenaf) and filled with a construction waste (marble dust) in different proportions. The test trials on the composite samples are conducted using a DUCOM high temperature erosion test rig as per ASTM G76. The experimentation and analysis are conducted as per the L30 model of response surface method (RSM). The results revealed that the incorporation of marble dust enhanced the resistance to erosion wear of the fiber‐polyester composites and among the control factors, the contributions of striking velocity and angle of impingement towards the erosion rate are significant. A predictive model working on the concept of neural networks (ANN) is used for the prediction of erosion rates at different levels and combinations of the significant control factors. The mechanisms of erosion loss are identified from the morphologies of the worn surfaces taken using a scanning electron microscope.