Scalable Fully Implicit Runge-Kutta Solvers for Time-Dependent Rayleigh-Bénard Convection

We consider the efficient numerical solution of time-dependent Rayleigh-Bénard convection governed by the incompressible Boussinesq equations. The coupled velocity-pressure-temperature system presents significant computational challenges arising from incompressibility, buoyancy coupling, convection, and the simultaneous resolution of multiple physical time scales. We present a systematic study of fully implicit Runge-Kutta methods for time-dependent Rayleigh-Bénard convection, together with scalable finite element solvers for the resulting stage-coupled systems. We employ high-order, L-stable Radau IIA methods for temporal integration and investigate Taylor-Hood and Alfeld-Sorokina finite elements in two dimensions, along with Taylor-Hood and enriched macroelement discretizations in three dimensions. The macroelement approaches offer strong local inf-sup stability and patch structures well suited to Vanka-type relaxation. In two dimensions, monolithic multigrid achieves nearly stage-independent outer Krylov iteration counts and maintains robust convergence under mesh refinement. On the same mesh with incremental pressure correction, selected Radau configurations reduce elapsed solve time by 18-32% and yield smaller sampled transient differences relative to a numerical reference solution. Three-dimensional experiments demonstrate favorable parallel scalability with Taylor-Hood elements and substantially improved discrete mass conservation with the enriched macroelement discretization. These results identify regimes in which coupled multistage fully implicit integration is computationally competitive when equipped with appropriate multigrid preconditioners.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-02
DOI
https://doi.org/10.48550/arxiv.2611.03361
Primary Topic
Advanced Numerical Methods in Computational Mathematics
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

Scalable Fully Implicit Runge-Kutta Solvers for Time-Dependent Rayleigh-Bénard Convection

Robert C. Kirby, Ahsan Ali
arXiv (Cornell University)
Advanced Numerical Methods in Computational Mathematics
preprint

Scalable Fully Implicit Runge-Kutta Solvers for Time-Dependent Rayleigh-Bénard Convection

Robert C. Kirby, Ahsan Ali
preprint en

Abstract

We consider the efficient numerical solution of time-dependent Rayleigh-Bénard convection governed by the incompressible Boussinesq equations. The coupled velocity-pressure-temperature system presents significant computational challenges arising from incompressibility, buoyancy coupling, convection, and the simultaneous resolution of multiple physical time scales. We present a systematic study of fully implicit Runge-Kutta methods for time-dependent Rayleigh-Bénard convection, together with scalable finite element solvers for the resulting stage-coupled systems. We employ high-order, L-stable Radau IIA methods for temporal integration and investigate Taylor-Hood and Alfeld-Sorokina finite elements in two dimensions, along with Taylor-Hood and enriched macroelement discretizations in three dimensions. The macroelement approaches offer strong local inf-sup stability and patch structures well suited to Vanka-type relaxation. In two dimensions, monolithic multigrid achieves nearly stage-independent outer Krylov iteration counts and maintains robust convergence under mesh refinement. On the same mesh with incremental pressure correction, selected Radau configurations reduce elapsed solve time by 18-32% and yield smaller sampled transient differences relative to a numerical reference solution. Three-dimensional experiments demonstrate favorable parallel scalability with Taylor-Hood elements and substantially improved discrete mass conservation with the enriched macroelement discretization. These results identify regimes in which coupled multistage fully implicit integration is computationally competitive when equipped with appropriate multigrid preconditioners.

arXiv (Cornell University)
Advanced Numerical Methods in Computational Mathematics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon