Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exascale Simulations of Fusion and Fission Systems

Misun Min, Yu-Hsiang Lan, Paul Fischer, Elia Merzari, Tri Nguyen, Haomin Yuan, Patrick Shriwise, Stefan Kerkemeier, Andrew Davis, Aleksandr Dubas, Rupert Eardly, Rob Akers, Thilina Rathnayake, Tim Warburton

TL;DR

The paper demonstrates exascale heat and fluid-flow simulations for fusion and fission systems using the GPU-accelerated spectral-element solver NekRS on Frontier and Aurora. It combines high-order spectral-element discretization with overset grids and solid-fluid coupling (CHIMERA) to tackle complex geometries and unprecedented problem sizes, including simulations exceeding $10^{12}$ grid points. A key contribution is the scalable, portable implementation built on OCCA/libParanumal with automated kernel tuning and multilevel preconditioning, enabling efficient performance across heterogeneous HPC architectures. The results establish a new capability for engineering-scale design studies in nuclear energy systems, bridging physics fidelity with practical application, and outlining the path toward accelerated, exascale-informed design of fusion blankets and fission cores.

Abstract

We discuss pioneering heat and fluid flow simulations of fusion and fission energy systems with NekRS on exascale computing facilities, including Frontier and Aurora. The Argonne-based code, NekRS, is a highly-performant open-source code for the simulation of incompressible and low-Mach fluid flow, heat transfer, and combustion with a particular focus on turbulent flows in complex domains. It is based on rapidly convergent high-order spectral element discretizations that feature minimal numerical dissipation and dispersion. State-of-the-art multilevel preconditioners, efficient high-order time-splitting methods, and runtime-adaptive communication strategies are built on a fast OCCA-based kernel library, libParanumal, to provide scalability and portability across the spectrum of current and future high-performance computing platforms. On Frontier, Nek5000/RS has achieved an unprecedented milestone in breaching over 1 trillion degrees of freedom with the spectral element methods for the simulation of the CHIMERA fusion technology testing platform. We also demonstrate for the first time the use of high-order overset grids at scale.

Exascale Simulations of Fusion and Fission Systems

TL;DR

The paper demonstrates exascale heat and fluid-flow simulations for fusion and fission systems using the GPU-accelerated spectral-element solver NekRS on Frontier and Aurora. It combines high-order spectral-element discretization with overset grids and solid-fluid coupling (CHIMERA) to tackle complex geometries and unprecedented problem sizes, including simulations exceeding grid points. A key contribution is the scalable, portable implementation built on OCCA/libParanumal with automated kernel tuning and multilevel preconditioning, enabling efficient performance across heterogeneous HPC architectures. The results establish a new capability for engineering-scale design studies in nuclear energy systems, bridging physics fidelity with practical application, and outlining the path toward accelerated, exascale-informed design of fusion blankets and fission cores.

Abstract

We discuss pioneering heat and fluid flow simulations of fusion and fission energy systems with NekRS on exascale computing facilities, including Frontier and Aurora. The Argonne-based code, NekRS, is a highly-performant open-source code for the simulation of incompressible and low-Mach fluid flow, heat transfer, and combustion with a particular focus on turbulent flows in complex domains. It is based on rapidly convergent high-order spectral element discretizations that feature minimal numerical dissipation and dispersion. State-of-the-art multilevel preconditioners, efficient high-order time-splitting methods, and runtime-adaptive communication strategies are built on a fast OCCA-based kernel library, libParanumal, to provide scalability and portability across the spectrum of current and future high-performance computing platforms. On Frontier, Nek5000/RS has achieved an unprecedented milestone in breaching over 1 trillion degrees of freedom with the spectral element methods for the simulation of the CHIMERA fusion technology testing platform. We also demonstrate for the first time the use of high-order overset grids at scale.
Paper Structure (6 sections, 2 equations, 3 figures, 3 tables)

This paper contains 6 sections, 2 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Formulations
  3. Advanced Algorithms
  4. Performance Measurements
  5. Performance Results
  6. Implications

Figures (3)

  • Figure 1: Rendering of the CHIMERA facility with NekRS simulation results for velocity and temperature.
  • Figure 2: Aurora, Frontier, and Polaris strong-scalings for Navier-Stokes simulation of $17\times 17$ rod bundle using the total number of grid points, $n=1.6B$. The average (wall) time per step, tstep, in seconds is measured over steps 1001--2000. makef is a routine responsible for setting up the right-hand-sides for the momentum equations, involving evaluation of the dealiased advection operator for the velocity.
  • Figure 3: Overset grid results for the CHIMERA geometry (inset). Left: strong-scaling time-per-step for the fluid-thermal (CHT) part of the geometry with and without overset coupling. Right: parallel efficiency for the same cases; also shown is the efficiency for the comparable ZEFR code ($N=4$) running on the V100s of Summit witherden2020.