ICME 2023: App.: AM Microstructure I
Program Organizers: Charles Ward, AFRL/RXM; Heather Murdoch, U.S. Army Research Laboratory
Tuesday 1:20 PM
May 23, 2023
Room: Caribbean VI & VII
Location: Caribe Royale
Session Chair: Adam Kopper, Mercury Marine
1:20 PM Cancelled
Multi-phase-field Simulation of Rapid Solidification in SUS316L Stainless Steel Using Artificial Neural Network-based Thermodynamic Calculation: Akinori Yamanaka1; Masahito Segawa1; Shoichiro Nakamura1; 1Tokyo University of Agriculture and Technology
Additive manufacturing is a powerful method to produce an industrial part with flexible shape and superior mechanical properties. In the additive manufacturing process, a laser heating causes a melting of material, followed by a rapid cooling that results in formation of unique solidification microstructures. The multi-phase-field method is a powerful numerical simulation method to quantitatively predict the microstructure evolution during the rapid solidification in the additive manufacturing process. In this study, the rapid solidification in SUS316L stainless steel during the additive manufacturing is simulated using the non-equilibrium multi-phase-field model that is able to simulate microstructure evolutions under a strong non-equilibrium condition. The thermodynamic calculations for the multi-component SUS316L stainless steel in the multi-phase-field simulation were accelerated using aritificial neural networks. In this presentation, we present a computational framework to train the aritificial neural networks with the thermodynamic database and to implement the trained neural network into the multi-phase-field simulation.
1:40 PM
Microstructure Variability Prediction in Powder Bed Metal Additive Manufacturing: Aashique Rezwan1; Theron Rodgers1; Daniel Moser1; 1Sandia National Laboratories
Laser powder bed fusion (LPBF) for stainless steels can produce components with novel designs and material properties. However, LBPF processes introduce variance in mechanical properties and as built geometries even with identical settings on a single machine. Quantifying the uncertainties introduced by LBPF is essential for component qualification. This work presents probabilistic predictions of microstructure variability in LPBF. A high-fidelity thermal fluid model is used to predict the melt pool behavior (i.e., temperature/phase) and it’s variance due to processing uncertainties. These data are used to predict microstructure size, shape and crystallographic texture using a coupled grain nucleation and kinetic Monte Carlo model for grain growth. Variability due to microstructure model parameters are also considered. This analysis will provide a tool for designers to predict a generalized margin of design for the mechanical properties of additively manufactured parts.SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525
2:00 PM
Identifying Scaling Laws for Discretization Error in Process-Structure Simulations of Laser Powder Bed Fusion: Joshua Pribe1; Brodan Richter2; Patrick Leser2; Edward Glaessgen2; 1National Institute of Aerospace; 2NASA Langley Research Center
Process-structure simulations for metal additive manufacturing involve computing temperature fields, grain growth, and other relevant physics on a discretized domain. The discretization length scale is typically adjusted to achieve a desired convergence level for output quantities without incurring excessive computational cost. However, the same discretization may not produce acceptable error across different material systems or process parameters, requiring ad hoc adjustments for different input parameters. This work seeks to define scaling laws that relate a normalized discretization measure to errors in microstructural and crystallographic texture-based outputs across input parameter space. A case study is conducted using a computational materials framework that couples an analytical thermal model with a microstructural evolution model to simulate the laser powder bed fusion process. Normalizing length scales in the model are identified through dimensional analysis. The key outcome is a more rigorous and consistent approach for selecting a discretization length scale in process-structure simulations.
2:20 PM
Prediction of Prior Austenite Structure as a Function of Processing Parameters in Additively Processed High-strength Steel: Stephen Cluff1; Clara Mock1; Brandon Mcwilliams1; 1DEVCOM Army Research Laboratory
The ability to control microstructure in high-strength steel via additive manufacturing (AM) is of interest for many structural applications, providing a means to homogenize mechanical properties across varied part geometries and implement designed heterogeneities. The localized application of heat during AM, coupled with predictive modeling, enables the optimization of microstructure for steels showing sensitivity to AM parameters. One such steel is known as AF9628. This steel’s final microstructure is mediated by the prior austenite microstructure. The current work presents a model that captures the evolution of the parent austenite grain structure in AF9628 during AM processing using a Potts/Monte Carlo method implemented in the SPPARKS software. This model is used to predict austenite grain size and morphology as a function of processing and can be used as a design tool for the control of austenitic and final microstructures.
2:40 PM Break