CALPHAD Uncertainty Quantification and TDBX
- PDF / 5,536,019 Bytes
- 10 Pages / 593.972 x 792 pts Page_size
- 54 Downloads / 278 Views
https://doi.org/10.1007/s11837-020-04405-z Ó 2020 The Minerals, Metals & Materials Society
AUGMENTING PHYSICS-BASED MODELS IN ICME WITH MACHINE LEARNING AND UNCERTAINTY QUANTIFICATION
CALPHAD Uncertainty Quantification and TDBX ´ N FREY,1,2 SAM SORKIN,1 YU LIN,1 ABHINAV SABOO,1 RAMO JIADONG GONG,1 GREGORY B. OLSON,1 MENG LI,3 and CHANGNING NIU 1,4 1.—QuesTek Innovations LLC, Evanston, IL 60201, USA. 2.—Present address: ETH Zurich, Zurich, Switzerland. 3.—Department of Statistics, Rice University, Houston, TX 77005, USA. 4.—e-mail: [email protected]
CALPHAD uncertainty quantification (UQ) is the foundation of materials design with quantified confidence. We report a framework and software packages to enable CALPHAD UQ assessment and calculation using commercial CALPHAD software (Thermo-Calc). This Bayesian inference framework is coupled with a Markov chain Monte Carlo algorithm to establish uncertainty traces with a given thermodynamic database file (TDB) and corresponding experimental data points. This general framework is demonstrated with the Ni–Cr binary system. The algorithm is firstly validated on synthetic data with known ground truth. Then it is applied to real experimental data to generate posterior traces. We develop a file format named TDBX, which provides a single source of truth by combining the original TDB content and the traces for each assessed Gibbs energy parameter. CALPHAD UQ calculations are performed based on the TDBX file, from which uncertainties for phase boundaries, enthalpy curves, and solidification range are collected as examples of basic design parameters. This TDBX file with corresponding scripts are made open-source. The combination of CALPHAD UQ assessments and calculations connected by TDBX supports uncertainty-assisted modeling, enabling the integrated application of modern design with uncertainty methodologies to computational materials design.
INTRODUCTION Chemical thermodynamics forms the cornerstone of materials science and engineering, and plays a key role throughout the materials design process. Designing new materials using an integrated computational materials engineering (ICME) approach begins with the use of chemical thermodynamics in the form of phase diagrams to identify material concepts for further exploration and ends with the use of chemical thermodynamics in the form of numerical process–structure models that enable optimization and sensitivity analysis of a material’s properties with respect to process conditions.1 As the ICME design approach is so dependent on chemical thermodynamics, a design is only as This work is supported by DOE SBIR Award DE-SC0017234. (Received July 3, 2020; accepted September 22, 2020)
accurate as the thermodynamic data and models upon which the design is based. Understanding the uncertainties from within the CALPHAD databases is the foundation of a design with quantified confidence. CALPHAD UQ studies date back to at least three decades ago. Frameworks in the early studies include frequentist approaches such as the ‘‘spread and Bayesi
Data Loading...
