Neural network analysis for erosive wear of hard coatings deposited by thermal spray: Influence of microstructure and mechanical properties

An artificial neural network (ANN) analysis is used to obtain a model to predict the rate of erosive wear of hard coatings deposited by two different kinds of thermal spray techniques. High Velocity Oxygen Fuel (HVOF) and Flame Spray FlexiCords (FS/FC) techniques were used under various operational conditions. Different microstructural features that control the mechanical and the tribological performances of three groups of deposits: tungsten carbide, chromium carbide and metallic alloy coatings, were analyzed. The ANN technique involves database use to predict erosive wear evolution, having a large number of variables like deposition process, impingement angles and velocity of the erosive particles, porosity, roughness, microhardness and fracture toughness. Commercially available powders were used as feedstock for coatings deposited by HVOF. Commercial cord wires were used in the FS/FC coating deposition. The slurry erosion testing was performed using a laboratory made pot-type slurry erosion tester, at impact velocity of 3,61 m/s and 9,33 m/s combined with impact angle of 30° and 90°. From the results, it was observed that the microhardness and fracture toughness, as a combination factor, have the greatest influence on erosive rate followed by porosity. Samples coated with WC-CoCr cermet coating with fine WC carbides exhibit higher erosion resistance as compared with the other conventional cermet and metallic alloy coatings, mainly because of its homogenous microstructure and improved properties like low porosity, high microhardness and high fracture toughness. The numerical results obtained via neural network model were compared with the experimental results. The agreement between the experimental and numerical results is considered a good aspect.