About this Abstract |
Meeting |
TMS Specialty Congress 2025
|
Symposium
|
8th World Congress on Integrated Computational Materials Engineering (ICME 2025)
|
Presentation Title |
Benchmarking Massively Parallel Phase-Field Codes for Alloy Solidification |
Author(s) |
Jiefu Tian, David Montiel, Kaihua Ji, Trevor Lyons, Jason Landini, Katsuyo Thornton, Alain Karma |
On-Site Speaker (Planned) |
Jiefu Tian |
Abstract Scope |
Phase-field (PF) modeling is a powerful method for simulating solidification microstructures in processes ranging from casting to additive manufacturing. Massively parallel PF simulations must be conducted on high-performance computing platforms to access experimentally relevant length and time scales. However, direct comparisons of high-performance PF codes are rare, except for those based on simplified benchmarking problems. This study benchmarks two state-of-the-art PF codes through direct quantitative comparison of their predictions for 3D convection-free directional solidification of a binary alloy for which microgravity experimental benchmark data is available: GPU-PF, a CUDA-based finite difference code utilizing GPUs, and PRISMS-PF, an MPI-based open-source finite element code with adaptive meshing. The comparison shows that both codes produce essentially identical microstructures but differ in their performance and convergence as a function of mesh size. This comparison highlights each tool’s unique strengths and reinforces their value in advancing PF modeling for alloy design within an ICME framework. |
Proceedings Inclusion? |
Undecided |