Designing a Periodic Table for Alloy Design: Harnessing Machine Learning to Navigate a Multiscale Information Space
- PDF / 2,783,494 Bytes
- 10 Pages / 593.972 x 792 pts Page_size
- 84 Downloads / 235 Views
https://doi.org/10.1007/s11837-020-04388-x Ó 2020 The Minerals, Metals & Materials Society
AUGMENTING PHYSICS-BASED MODELS IN ICME WITH MACHINE LEARNING AND UNCERTAINTY QUANTIFICATION
Designing a Periodic Table for Alloy Design: Harnessing Machine Learning to Navigate a Multiscale Information Space SCOTT R. BRODERICK1 and KRISHNA RAJAN
1,2
1.—Department of Materials Design and Innovation, Buffalo, Buffalo, NY 14260, USA. 2.—e-mail: [email protected]
University
at
We provide an overview of how to apply statistical learning methods to directly track the role of alloying additions in the multiscale properties of alloys. This leads to a mapping process analogous to the Periodic Table where the resulting visualization scheme exhibits the grouping and proximity of elements based on their impact on the properties of alloys. Unlike the conventional Periodic Table of elements, the distance between neighboring elements in our Alloy Periodic Table uncovers relationships in a complex high-dimensional information space that would not be easily seen otherwise. We embed this machine learning approach with an epistemic uncertainty assessment between data. We provide examples of how this data-driven exploratory platform appears to capture the alloy chemistry of known engineering alloys as well as to provide potential new directions for tuning chemistry for enhanced performance, consistent with accepted mechanistic paradigms governing alloy mechanical properties.
INTRODUCTION The complexity of the role of elemental chemistry in the properties, processing, and performance of superalloys has been and continues to be an area of intense study. Of course, this is due to the wellknown multidimensional/multiscale impact of alloying additions on the phase stability, microstructural evolution, dislocation dynamics, and chemical and thermal resistance, to mention just a few of the metrics that need to be met. At present, there are sophisticated models and experimental techniques that address specific segments of engineering design such as: developing new materials (e.g., first-principles calculations), refining legacy materials (e.g., through processing and microstructural modification), and engineering design and manufacturing. Linking together information across multiple scales requires accounting for the interaction of the myriad parameters that govern materials development and the complexity of the engineering performance.1–4 Current approaches that utilize a datadriven approach do so in conjunction with physicalbased and/or heuristically driven models. The (Received July 6, 2020; accepted September 14, 2020)
computational design of alloys has largely focused on the issue of phase stability,5–8 and this in itself is a massive combinatorial problem in selecting which combination of elements need to be added to the base alloy chemistry. The search for elemental substitutions and/or additions needed to refine metal alloy compositions and enhance their properties is a classical problem in metallurgical alloy design. Finding appropr
Data Loading...
