Constructing the high temperature efficiency and instability maps of selective laser melted 316L stainless steel through artificial neural network modeling

Abstract

The present work focuses on the high temperature flow behavior of additively manufactured 316 L stainless steel. Toward this end, a series of hot compression tests were conducted in the temperature range of 700–1000 °C under the strain rates of 0.001, 0.01 and 0.1s-1. The high temperature flow curves were accompanied with considerable fraction of strain softening at various thermomechanical conditions. The strain rate sensitivity (m) maps were constructed which comprised the negative and positive m-values from −0.01 to 0.14. Accordingly, the conventional phenomenological models were un-capable of precise predicting the level and trend of the flow stress. In this respect, a single layer Artificial Neural Network (ANN) model was employed for assessment of flow behavior at elevated temperature. The training of the feedforward model was performed through a backpropagation algorithm. The statistical factors were employed to assess the model's reliability. Flow curves were also predicted at thermomechanical conditions which were completely different from those utilized for construction of the model. Interestingly, the developed ANN model not only efficiently tracked the strain hardening portion, but also well predicted the flow softening behavior of the material.

Keywords

• Artificial neural network modelling
• Additive manufacturing
• Mechanical properties
• Stainless steel
• Thermomechanical processing