Estimation of the amount of retained austenite in austempered ductile irons using neural networks

Many of the properties of austempered ductile iron depend on the austenite which is retained following the bainite reaction. A neural network model within a Bayesian framework has been created using published data to model the retained austenite content. The model allows the quantity of retained austenite to be estimated as a function of the chemical composition and heat treatment parameters. The computer programs associated with the work have been made freely available (http //www.msm.cam.ac.uk/map/mapmain.html)