Neural Network Analysis for Deposition of Nickel Alumnide Coatings on Steel by Plasma Spraying

Abstract

Nickel-aluminides are used in thermal spray applications mostly as bond-coat materials, where their function is two-fold: to minimize the thermo-mechanical stresses at the substrate-coating interface arising out of thermal expansion mismatch of the metal substrate and ceramic top layer and secondly to promote coating adhesion. This paper describes the deposition of nickel aluminide coatings on mild steel substrates. A computational technique (ANN analysis) is used to predict the rate of coating deposition under various operational conditions. Prediction of deposition efficiency is significant as it gives an idea about the effectiveness of the spraying technique as well as about the coatability of the material under investigation. This technique involves database training to predict property parameter evolutions in process having large number of interdependent variables such as in the case of plasma spray coating deposition. This paper presents the database construction, implementation protocol and also the set of predicted results related to the deposition efficiency of nickel-aluminide coating. It is seen that the deposition rate depends largely on the plasma arc current, arc voltage etc. The neural network analysis gives an oversight on the parameter inter-dependencies and of the effect of the individual process variable on the efficiency of coating deposition.