Does the Boussinesq approximation correctly describe statistically stationary states of Rayleigh-Bénard convection?
Abstract
It is generally believed that statistically stationary states of Rayleigh-Bénard convection (RBC) can be described by statistically stationary solutions to the Boussinesq approximation. Based on an analysis of the mechanical energy equation, we argue that this belief is not justified, because the expression for mean conversion of internal energy into kinetic energy by compressions and expansions, which is implicitly given by the approximation, does not qualify as an approximation. In consequence, it is questionable whether the exact Boussinesq expression relating the mean kinetic energy dissipation in a stationary state to the Rayleigh and Nusselt numbers is valid. Using assumptions that partly lie outside the Boussinesq approximation, we derive scaling relations for the mean kinetic energy dissipation in a stationary state in three and two dimensions. The relation for the three-dimensional system is similar to the Boussinesq expression, but includes an unknown prefactor, while the relation for the two-dimensional system is completely different. Our arguments suggest that statistically stationary solutions to the Boussinesq approximation in three dimensions, reproduce scaling relations between various statistical quantities quite well, although they should not be regarded as approximations in a strict sense, while solutions in two dimensions completely fail. We suggest that this should be investigated by performing DNS of weakly compressible RBC and compare with DNS of Boussinesq RBC.
Related
- Stationary Structures of the Boussinesq equations
- Statistically calibrated effective projected optical-depth imaging from paired photon-counting returns
- Randomness Certification from Statistically Hiding Commitments
- Statistically Robust Characterization of Polaron and Bipolaron Intrinsic Traps in Amorphous Silica Reveals the Role of Network Flexibility in Electron Localisation
- Transitioning to the ultimate regime of convection in three-dimensional direct numerical simulations
BibTeX
@misc{lindborg2026does,
title = {{Does the Boussinesq approximation correctly describe statistically stationary states of Rayleigh-Bénard convection?}},
author = {Erik Lindborg},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.16634},
eprint = {2609.16634},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.16634},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Does the Boussinesq approximation correctly describe statistically stationary states of Rayleigh-Bénard convection?
AU - Erik Lindborg
PY - 2026
DA - 2026/09/19
PB - arXiv
DO - 10.48550/arxiv.2609.16634
UR - https://arxiv.org/abs/2609.16634
AB - It is generally believed that statistically stationary states of Rayleigh-Bénard convection (RBC) can be described by statistically stationary solutions to the Boussinesq approximation. Based on an analysis of the mechanical energy equation, we argue that this belief is not justified, because the expression for mean conversion of internal energy into kinetic energy by compressions and expansions, which is implicitly given by the approximation, does not qualify as an approximation. In consequence, it is questionable whether the exact Boussinesq expression relating the mean kinetic energy dissipation in a stationary state to the Rayleigh and Nusselt numbers is valid. Using assumptions that partly lie outside the Boussinesq approximation, we derive scaling relations for the mean kinetic energy dissipation in a stationary state in three and two dimensions. The relation for the three-dimensional system is similar to the Boussinesq expression, but includes an unknown prefactor, while the relation for the two-dimensional system is completely different. Our arguments suggest that statistically stationary solutions to the Boussinesq approximation in three dimensions, reproduce scaling relations between various statistical quantities quite well, although they should not be regarded as approximations in a strict sense, while solutions in two dimensions completely fail. We suggest that this should be investigated by performing DNS of weakly compressible RBC and compare with DNS of Boussinesq RBC.
ER -