Algorithm Development in Materials Science and Engineering: Computational Simulations and Algorithms for Study Structure-Processing Relations
Sponsored by: TMS Materials Processing and Manufacturing Division, TMS: Computational Materials Science and Engineering Committee, TMS: Integrated Computational Materials Engineering Committee, TMS: Phase Transformations Committee, TMS: Solidification Committee
Program Organizers: Mohsen Asle Zaeem, Colorado School of Mines; Mikhail Mendelev, NASA ARC; Bryan Wong, University of California, Riverside; Ebrahim Asadi, University of Memphis; Garritt Tucker, Colorado School of Mines; Charudatta Phatak, Argonne National Laboratory; Bryce Meredig, Travertine Labs LLC

Wednesday 2:00 PM
March 17, 2021
Room: RM 36
Location: TMS2021 Virtual

Session Chair: Ebrahim Asadi, University of Memphis

2:00 PM
Real Time Boundary Condition Acquisition and Integration of Heats of Fusion and Phase Transformation Using an Implicit Finite Element Newton Raphson Based Approach for Thermal Behavior Prediction in Additively Manufactured Parts: Deepankar Pal¹; Madhu Keshavamurty¹; Grama Bhashyam¹; ¹Ansys
     The transient thermal behavior during additive processing as a function of space and is a key contributor towards an Integrated approach to Computational Additive Manufacturing suited for part qualification. Simulation outputs namely the thermal gradients and cooling rates are further used for microstructure and Crystal plasticity predictions. Additionally, the continuum residual stresses could be accurately predicted. Therefore, an accurate prediction of transient thermal response warrants for accurate downstream predictions.To address the above-mentioned challenges and opportunities, a new approach was established where the top surface thermal boundary conditions were captured using an Infrared thermal camera and further emissivity corrected using static thermal simulation for application in transient thermal analysis during Additive Processing. In addition, heats of fusion and phase transformations were incorporated to arrive at an accurate simulation response. This response was validated against experimentally available melt pool metrics and further contrasted against Goldak and other heat source assumptions.

2:20 PM
Global Local Modeling of Melt Pool Dynamics and Bead Formation in Laser Bed Powder Fusion Process Using a Comprehensive Multi-Physics Simulation: Faiyaz Ahsan¹; Jafar Razmi¹; Leila Ladani¹; ¹Arizona State University
    Understanding the physical mechanism of laser powder bed fusion (LPBF) additive manufacturing process can benefit significantly through computational modeling. LPBF is a state-of-the-art manufacturing process which involves rapid heating of the powder bed by the laser heat source and subsequent flow of fluid and solidification around melt pool. This work aims to assess the impact of various process parameters like laser power, scan speed on the shape and geometry of the solidified bead from melt pool, which dictate the final build property and establish a relation between them to optimize the process. The computational model was be validated by comparing the result with experiment. This work also attempts to calculate the convective heat transfer coefficient around melt pool region during LPBF process. Marangoni convection, recoil pressure and buoyancy force are included to properly simulate the melt pool dynamics along with non-gaussian beam to model the laser-powder interaction.

2:40 PM
Multi-scale Modeling of Hierarchical Microstructure in Ceramic Composites: Matthew Guziewski¹; David Montes de Oca Zapiain²; Jennifer Synowczynski-Dunn¹; Remi Dingreville²; Shawn Coleman¹; ¹Army Research Laboratory; ²Sandia National Laboratory
    Microstructural features are known to influence the mechanical response of multiphase ceramic composites. To elucidate their contributions, a multi-scale modeling effort utilizing atomistic and mesoscale simulations has been developed. Using observed orientations within experimental samples, high-throughput atomistic models investigate the distribution of grain boundary structures and energetics through a Monte Carlo approach to grain boundary optimization, considering both the ground and metastable states of these interfaces. Then, utilizing machine learning techniques in conjunction with strategic sampling, models are produced to capture the variance of relevant properties, such as strength, fracture energy, and mobility, within the microstructure. These models are used to parameterize phase field models, allowing for the study of the influence of various granular distributions on the microstructural evolution and macroscale response of the material. This multi-scale approach has the potential to shorten the material design loop and improve performance through the prediction of favorable microstructures and processing methods.

3:00 PM
Analysis of Dendrite Growth and Microstructure Evolution during Solidification of Al 6061 via 2D and 3D Phase Field Models: Neil Bailey¹; Yung Shin¹; ¹Purdue University
    A 3-dimensional (3D) phase field model is developed and used to predict dendrite growth and microstructure development during the laser welding processes of Al 6061 alloy. A 2D version of the model is used to validate the model using isothermal solidification of an Al-4wt.%Cu alloy to compare, with good agreement between dendrite growth velocity and overall morphology. The 3D model is then used to simulate dendrite growth and microstructure development in a laser welding process using Al 6061. By using a validated, comprehensive laser welding model to calculate the complex and strongly transient temperature field, the predicted 2D cross-sections of the 3D simulated microstructure are compared with micrographs from an experimental workpiece as well as 2D simulated results. The 2D cross-sections taken from the 3D simulation results match well with the experimental micrographs and they are a better prediction of dendrite growth and solidification microstructure than the 2D simulation results.

3:20 PM
A Machine Learning Approach for Predicting Melt Pool Size in Wire-feed DED Process: Amit Verma¹; Zhening Yang¹; Ali Gruzel¹; Anthony Rollett¹; ¹Carnegie Mellon University
    Wire-feed direct energy deposition (WFDED) is an up-and-coming AM method because of its high deposition rates and flexible application. However, the melt pool size in WFDED is hard to predict because the operating parameters are not independent. To better dimension control and predict the microstructure of as-built part, it is crucial to find a way to predict the melt pool dimensions in WFDED. In this work, we analyzed Ti-6Al-4V single-bead samples made with various power levels and speeds. Random forest (RF) method is used to develop a model to predict the melt pool size in WFDED and two other analytical methods are used to compare with our RF method in this research. An exhaustive dataset was used to train our RF model and the operating parameters affecting melt pool size most significantly are found. The analysis provides insights into the scope of data analytics methods for quantifying process uncertainty.