AI/Data informatics: Design of Structural Materials: Poster Session
Sponsored by: TMS Materials Processing and Manufacturing Division, TMS Structural Materials Division, TMS: Mechanical Behavior of Materials Committee, TMS: Computational Materials Science and Engineering Committee, TMS: Integrated Computational Materials Engineering Committee
Program Organizers: Jennifer Carter, Case Western Reserve University; Amit Verma, Lawrence Livermore National Laboratory; Natasha Vermaak, Lehigh University; Jonathan Zimmerman, Sandia National Laboratories; Darren Pagan, Pennsylvania State University; Chris Haines, Ccdc Army Research Laboratory; Judith Brown, Sandia National Laboratories
Wednesday 5:30 PM
March 17, 2021
Room: RM 32
Location: TMS2021 Virtual
Discovery of Optimized ω-phase Free Ti-based Alloys Using CALPHAD and Artificial Intelligence Approach: George Dulikravich1; Rajesh Jha; 1Florida International University
Ti-Ta-Nb-Sn-Mo-Zr alloy system was divided into 26 sub-groups of 6, 5, 4 and 3 elements at a time. Each group was studied for various phase transformations using CALPHAD approach. Omega (ω) phase is detrimental as it severely impacts mechanical properties and causes embrittlement in shape memory alloys based on titanium. It is a metastable phase and it is formed during heat treatment. All 26 sub-groups were studied for determining phase stability of Omega (ω) phase from -242 ˚C to 1231 ˚C, and ω-phase was observed up to 876 ˚C. Chemistries of several sets of Omega (ω) phase free Ti-based alloys were found. For CALPHAD approach, we used Thermocalc software. We also used supervised machine learning and developed models using Deep Learning libraries (TensorFlow/Keras) in Python. We analyzed our data using unsupervised machine learning where we used Self-Organizing Maps (SOM) and Hierarchical Clustering Analysis (HCA) for discovering patterns within the dataset.
Evaluating Uncertainty in Clustering of Nanoindentation Mapping Data: Bernard Becker1; Eric Hintsala1; Benjamin Stadnick1; Douglas Stauffer1; Ude Hangen1; 1Bruker Nano Surfaces Division
High throughput nanoindentation mapping by nanoindentation (XPM) has potential as a technique for assisting in development of structural materials. Compared to other mechanical tests, nanoindentation is a highly localized measurement that requires minimal sample preparation and has the ability to map properties of various phases over macro length scales. Datasets of many thousands of indents can be gathered in hours, requiring a similar throughput increase in analysis. Clustering is a basic machine learning technique that has been successfully demonstrated for this task, but it’s not clear which algorithms to use, how many clusters to sort data into, and more. To begin addressing this, a quantitative comparison of clustering techniques in terms of bias and variance has been pursued by using bootstrapping of simulated datasets based upon a modelling the probability distribution of the original dataset. This technique is adaptable to evaluating other types data or other datasets as well.
Fast and High-throughput Synthesis of Film and Bulk High-entropy Alloys: Yu Zou1; 1University of Toronto
The synthesis of multicomponent alloys, such as medium- and high-entropy alloys (MEAs and HEAs), out of a vast composition space is a daunting task. Here, we demonstrate a radio frequency inductively coupled plasma (RF-ICP) method for preparing MEAs and HEAs in a fast and high-throughput fashion. Typical bulk MEAs and HEAs are successfully synthesized within less than 40 s per alloy, thereby greatly improving the fabrication efficiency and throughput. This method provides a new opportunity for the accelerated discovery of new multi-principal element alloys. We also use the magnetron co-sputtering method to produce thin film HEAs.
High-throughput Calculation to Predict the Eutectic Point in Quaternary System: Jun Lu1; Yu Zhong1; 1Worcester Polytechnic Institute
Eutectic point plays a significant role in the application of many alloys. However, limited by the visualization technology and heavy computation volume, finding the eutectic point in a quaternary or higher system is challenging and time-consuming. The high-throughput calculation, as a novel approach that combines CALPHAD (Computer Coupling of Phase Diagrams and Thermochemistry) methodology with intelligent data construction and data screening, shows great potential in predicting key properties of metal alloys. By exploiting powerful computing ability, thousands of calculations can be executed by sets of code without supervision instead of manually performing the same calculation procedures. Here, we introduce a way to compute T-Liquidus and T-Solidus of large amount of compositions in complex alloy and screen the eutectic point based on the temperature difference using high-throughput calculation.