Predictive modeling for low-power direct energy deposition of 316L stainless steel

Low-power direct energy deposition (DED) can be a viable strategy for manufacturing 316L stainless steel parts with low porosity and minimal distortion, enabled by a novel predictive modeling and optimization framework. This study assessed the methodology of a CNC-integrated DED system with four powder nozzles and a custom-made G-Code postprocessor (DEDRA). Over 150 samples were conducted to systematically adjust laser power, traversing speed, powder feed rate, hatch spacing, and interlayer height. Furthermore, ten experimentally derived mathematical models were applied to predict geometry (internal and external) and porosity with an error below 4% across five parameter sets. These models can calculate single-track welds and their superpositions during layer-by-layer deposition, enabling the estimation of thermal distortion and residual stress, which are minimized under the optimized process parameters. Under these optimized conditions, the resulting builds showed porosity between 3 and 4.85%, with a probability of only 0–69 defective parts per million falling outside this porosity range. Dimensional accuracy was preserved without excessive layer growth, while microhardness equaled or surpassed that of forged 316L. Microstructural characterization revealed refined grain morphology, improved homogeneity, and reduced defect formation under optimized conditions. This research addresses the limitations of prior low and high-power, trial-and-error approaches by introducing a predictive framework that enables reliable, repeatable, and energy-efficient additive manufacturing of stainless steel, establishing a clear foundation for future real-time monitoring and process-control advancements.