Authors: Yu-Hsiang Lan (University of Illinois Urbana-Champaign); Misun Min (Argonne National Laboratory); Paul FIscher (University of Illinois Urbana-Champaign); Elia Merzari (Pennsylvania State University); Tri Nguyen (Pennsylvania State University, Hanoi University of Science and Technology); Haomin Yuan, Patrick Shriwise, and Stefan Kerkemeier (Argonne National Laboratory); Andrew Davis, Aleksander Dubas, and Rupert Eardley-Brunt (United Kingdom Atomic Energy Authority); Maximiliano Dalinger and Luiz Aldeia Machado (Pennsylvania State University); Kristopher Rowe (Argonne National Laboratory); Ananias Tomboulides (Aristotle University of Thessaloniki, Argonne National Laboratory); and Ramesh Balakrishnan (Argonne National Laboratory)
Abstract: We present pioneering multiphysics simulations of lead-lithium and water flows in the ITER WCLL fusion test blanket set using NekRS on exascale platforms, including Frontier and Aurora. Developed at Argonne, NekRS is a high-performance, open-source multiphysics solver that employs rapidly convergent high-order spectral element methods with minimal numerical dissipation and dispersion. Advanced multilevel preconditioners, high-order time-splitting schemes, and adaptive communication strategies are enabled through OCCA-based kernels, ensuring portability and scalability across emerging HPC architectures. This work presents the first complete, high-fidelity, high-resolution multiphysics model of a fusion blanket, including turbulent flow resolution, heat transfer, buoyancy, magnetohydrodynamics, tritium breeding and transport, and water activation. We also present the largest spectral element model in NekRS to date and the largest simulation of fusion systems to date.