Regression and ANFIS-based models for predicting of surface roughness and thrust force during drilling of biocomposites

Biocomposites are increasingly being applied in the automotive industry. For assembly purpose, drilling process is often performed in these biocomposites parts to produce holes. The machining behavior of biocomposites is affected by the matrix, the reinforcements (nature and size), by the cutting conditions and the mechanical properties of biocomposites. Some of the newly developed biocomposites are made of a polypropylene/polyolefin blend matrix, which has low thermal conductivity and low softening temperature. Therefore, the selection of cutting parameters should be done with care to avoid thermal-induced damages occurred during drilling biocomposites. Also, due to the heterogeneous and anisotropic properties of biocomposites, it is essential to develop models for predicting the effects of these factors on the machining process performance indicators. In this research work, a full factorial design is used to investigate the effects of machining parameters (feed rate, drill diameter, spindle speed, and reinforcement type) on machinability indicators during the drilling of the three new biocomposites: Polypropylene/Polyolefin matrix reinforced with biocarbon particles and/or miscanthus fibers. The regression and ANFIS-based models are established for predicting thrust force (F_t) and surface roughness (R_a) based on empirical data. The results obtained show that, for each tested biocomposite, the proposed ANFIS models have better predictive accuracy for thrust force and surface roughness than the regression-based models. Furthermore, the drilling parameters and reinforcements used in these biocomposites considerably affect their machinability.