Response surface method and neural computation for the analysis and prediction of erosion response of glass-polyester composites filled with waste marble dust

Abstract Erosion wear response of waste marble dust filled glass-polyester composites is studied using response surface method (RSM) and neural computation. Marble dust is a construction/industrial waste generated from the processing of marble producing rocks. In the present investigation, convention hand lay-up technique is used for the preparation of hybrid composites consisting of 40% of glass fibers and 0, 16 and 32 wt% of waste marble dust respectively. The erosion behavior of the hybrid composites is investigated using an erosion tester as per ASTM G76. The experimental results are successfully analyzed using an analysis tool based on response surface method. Striking velocity, filler content and impingement angle in that sequence among the test parameters are found significant affecting the erosion loss of the composites. The morphologies of the eroded composite surfaces are analyzed and the predominant wear mechanisms are identified. Finally, an analysis and prediction tool working on artificial neural networks (ANN) is gainfully implemented for the analysis of experimental results and prediction of erosion rate for a wide combination of control factors within the experimental domain.