Effects of process factors on tribological behaviour of epoxy composites including Al2O3 nano particles: a comparative study on multi-regression analysis and artificial neural network
Yusuf Şahin, Fatih Şahin
Abstract
Basalt fabric filled epoxy composites (BFRCs) and adding of nano-Al2O3 particles by conventional stirring and followed by moulding technique to fabric filled epoxy composites (n-BFRCs) were prepared to obtain higher rigidities and strengths for automotive and chemical industries. The tribological property of the composites was studied under dry condition with a Response Surface Methodology (RSM) to study influences of material’s type, load and surface roughness. In addition, multiple regression analysis (MRA) was developed for predicting the dry wear results and compared with artificial neural network (ANN). The results indicated that volume loss increased with increasing load and surface roughness, but decreased with increasing hardness. Improvements of nano-Al2O3 particles in basalt composites were about 16.9% in compared with BFRCs. All main factors were effective on the wear of the composites, but hardness and load were the most significant factors, followed by surface roughness. Furthermore, both MRA and ANN approach provided an effective methodology for prediction of the wear of composites, but ANN was found to be more effective tool for getting more accurate results than that of MRA, particularly based on the determination of coefficient and absolute relative errors.