The spherical Couette flow problem—the motion of a viscous fluid confined between two concentric, differentially rotating spheres—serves as a canonical model for understanding fundamental instabilities, nonlinear dynamics, and transitions to turbulence in rotating fluids. Its relevance extends to geophysical and astrophysical contexts, including the fluid dynamics of planetary liquid cores and accretion disks. Despite decades of study, the full bifurcation structure and the precise scaling of transitional Reynolds numbers remain areas of active investigation, particularly for wide-gap configurations and moderate-to-high Reynolds numbers where complex three-dimensional flows emerge.
This project proposes a systematic numerical investigation of spherical Couette flow using a newly developed, high-order spectral code written in Fortran. The numerical method employs an expansion in spherical harmonics for the angular directions, combined with Chebyshev polynomials in the radial direction to efficiently resolve thin boundary layers near the spherical shells. This hybrid spectral approach is particularly well-suited for the spherical geometry, offering superior accuracy per degree of freedom compared to finite-difference or finite-element methods. The code leverages established high-performance libraries: ISPACK for optimized spherical harmonic transforms, FFTW for fast Fourier transforms, and LAPACK for efficient linear algebra operations, all of which will be installed and managed using the SPACK package manager to ensure reproducibility and optimal performance on the HPC system.
The project has two primary objectives. The first is to perform a rigorous benchmarking and performance assessment of the new code. This includes scaling tests to evaluate parallel efficiency and identify computational bottlenecks—particularly the spherical harmonic transforms, which dominate the computational cost. The second objective is to conduct a targeted set of first production simulations across a range of Reynolds numbers (Re ≈ 10**2 to 10**4) to validate the code against well-established benchmark results from the literature. These simulations will aim to reproduce known flow regimes, including steady axisymmetric Taylor vortices, non-axisymmetric spiral waves, and the onset of time-dependent and chaotic dynamics.
Achieving these goals demands substantial computational resources. The requested allocation of 10,000 core-hours per month will enable the PI to execute the necessary parameter sweeps and high-resolution runs, explore the code's performance envelope, and collect the preliminary data required to refine future, more extensive investigations. This initial allocation is critical for establishing the credibility and performance of the new computational tool, paving the way for a larger-scale study of turbulence and transport in rotating spherical flows.