Tackling Flow Stress of Zirconium Alloys

Abstract

Flow stress during hot deformation is essentially controlled by the chemistry of material, initial microstructure/texture, strain, strain rate, strain path, stress triaxility and the temperature of deformation. A comprehensive literature survey has been performed to realize this fact completely. In the present research, a neural network model under Bayesian framework has been created to correlate the complex relationship between flow stress with its influencing parameters in various grades of zirconium alloys at different deformation conditions. The network has been trained with published experimental database obtained from the different hot deformation experiments of zirconium alloys. Performance of the model has been evaluated; and excellent agreements between experimentally measured and model calculated data are obtained. The analysis permits the estimation of error bars whose magnitude strongly depends on their position in the input space. The model has been employed to different grades of zirconium alloys to confirm that the predictions are reasonably accurate in the context of basic metallurgical/solid mechanics theories and principles. The work has clearly identified the regions of the input space where further experiments should be encouraged and necessary. This model will be useful to design and manufacture the new generation zirconium alloys in future for the nuclear power plant components according to the needs of nuclear engineers/scientists by controlling the alloying elements and other possible conditions. The result shows that neural computation is a very effective tool to model the complex $$\textit{non-linear}$$ behaviour of flow stress of different zirconium alloys under any deformation conditions.