Predicting Tensile Strength in Laser Powder Bed Fusion (LPBF) of IN718 Using Neural Networks: the Influence of Heat Treatments and Process Parameters

Abstract Additive manufacturing offers significant advantages, especially for fabricating complex components. Inconel 718 (IN718), a nickel-based superalloy known for its high strength and resistance to heat and corrosion, is frequently produced using Laser Powder Bed Fusion (LPBF) technology. Due to the varying thermal gradients and solidification rates during the LPBF process, the resulting microstructure can differ significantly, influencing mechanical performance. This study focuses on predicting the tensile strength of LPBF-manufactured IN718 components using neural networks, taking into account key process parameters such as laser power, scanning strategy, and heat treatment conditions. Various post-processing heat treatments were applied to further enhance the mechanical properties, including homogenizing the microstructure and reducing anisotropy. In this study, 159 experimental data points were used to train and validate the neural network model. The model demonstrated its ability to capture complex relationships between process variables and tensile strength, with a strong R2 value of 0.85. The study highlights the potential of machine learning in optimizing additive manufacturing processes by providing accurate predictions of mechanical properties, such as tensile strength. By incorporating Machine Learning (ML) and empirical equations, other mechanical properties such as the fatigue life of LPBF components can be obtained and this can further reduce Additive Manufacturing (AM) production costs. The use of neural networks in this context opens the door to more efficient manufacturing practices, minimizing the need for physical experiments and improving the performance of complex materials like IN718.