TC05: Uncertainty Quantification in Multiscale Materials Simulation
Infra-red Invisibility Stickers

TC05: Uncertainty Quantification in Multiscale Materials Simulation

Jonathan Guyer, National Institute of Standards and Technology

Benchmarking Problems for Phase Field Codes

Written by Aashutosh Mistry

The phase-field method has proved to be useful in understanding the physics of alloying, grain growth, and other problems pertaining to materials science. Given the presence of multiple research groups and different numerical implementations, there is a need for standardization. Jonathan Guyer and colleagues from the National Institute of Standards and Technology and the Center for Hierarchical Materials Design (CHiMaD) have been spearheading such efforts along with collaboration from universities and national laboratories. In their opinion, one should use standardized problems and solutions to benchmark different codes. The selection of benchmarking problems is nontrivial. Various considerations for choosing these standardized tests were (1) problems should only be solved numerically; (2) problems should have different physics and complexities; (3) problems should not require heavy computing resources; and (4) simulation output should be defined such that it allows for comparison across results from different solvers. With these criteria in mind, the researchers have agreed upon six problems, ranging from spinodal decomposition and Ostwald ripening to stokes flow and electrochemistry. Their initial findings for comparison across different solvers reveal that even the choice of time stepping scheme can have a strong effect on the simulation outcome. Such efforts will hopefully unify different available schools of thoughts and establish guidelines for accurate physically realistic phase-field simulations.

Comments

The comments to this entry are closed.