SOFT COMPUTING TECHNIQUES TO DETERMINE THE TENSILE PROPERTIES OF GAMMA BASED TITANIUM ALUMINIDES

Abstract

Titanium aluminides based on TiAI and Ti3A1 are emerging as a revolutionary high temperature material due to some attractive properties such as low density (~ 3.8 g/cm3), high specific strength a/p, high specific stiffness E/p, high temperature strength retention, and resistance against `titanium fire'. However one of the major obstacles to these titanium alumindes is their ambient temperature brittleness, making it difficult to fabricate them into structural. components by conventional processing methods. Several experimental studies have been conducted to overcome the problem of brittleness at room temperature. But there is a need of theoretical model for optimization of process parameters. To solve this problem, the new emerging soft computing techniques are used. Artificial Neural Network (ANN) and Adaptive neuro-fuzzy inference system (ANFIS) have been used to model these complex non-linear models. The tensile properties, percent elongation and 0.2 % yield strength have been optimized with respect. to processing parameters i.e. alloying composition, heat treatment conditions, micro-structure types and working temperature. The models yielded very good results and will prove substantial in predicting the tensile properties of titanium aluminides. The very good agreement between the predicted values from the ANFIS and experimental data is comparable with the difference between experimental data published in different sources. This gives a confidence in the prediction from the model.