Neural surface roughness models of CNC machined Glass Fibre Reinforced Composites

CNC machining of parts from pre-made Glass Fibre Reinforced Composites (GFRCs) blocks started gaining ground. However, wrong cutting conditions result in poor surface quality, delaminations or other damaging effects. In this work, a computational tool is developed to help improve machinability of these parts by accounting for surface quality. Artificial Neural Network models trained with data obtained through Taguchi-style designed experiments predict surface roughness obtained. GFRC blocks made from D.E.R.321 epoxy resin, CHEM.93-1-74, PC12 stabiliser and Woven Roving (500 gr/m(2) and 800 gr/m(2)) were CNC machined. Microscopy and image analysis studies enrich the ANN models with machined material macro-structural characteristics.