English

Non-Oberbeck-Boussinesq effects in coldwater

Fluid Dynamics 2026-05-05 v2 Materials Science Geophysics

Abstract

Water exhibits an anomalous nonlinear temperature-density (ρ\rho-TT) relation as it approaches freezing, along with an increase in viscosity, and a decrease in thermal conductivity. These departures from the standard Oberbeck--Boussinesq approximation, which assumes constant material properties and a linear ρ\rho-TT relation, can modify convection in ice-bounded aquatic systems, yet their effects remain unexplored. Here, we examine these effects via the canonical Rayleigh--B\'enard convection framework using direct numerical simulations. We show that non-Oberbeck--Boussinesq effects lower the mean fluid temperature relative to the standard case and break the classical symmetry of the mean temperature profile. The magnitude of this symmetry breaking depends on both the Rayleigh number RaRa and the temperature-dependent material properties retained in the governing equations. We further identify a small but measurable shift in the critical Rayleigh number, RacRa_c. After accounting for this shift, the nondimensional heat transfer rate, NuNu, follows the classical scaling with supercriticality, while ReRe remains consistent with the Grossmann--Lohse unifying theory, Re(RaRac)1/2Re\propto (Ra-Ra_c)^{1/2} for low-RaRa convection (regime Iu\mathrm{I}_u) and Re(RaRac)4/7Re\propto (Ra-Ra_c)^{4/7} at high-RaRa (regime IIIu\mathrm{III}_u). Unlike the classical expectation that the latter scaling arises at high Prandtl number, here it is obtained at an intermediate Prandtl number, Pr10Pr\sim 10. Our results establish how near-freezing material anomalies affect both local and global properties of convection, with implications for heat distribution and mixing in cryospheric liquid waters.

Keywords

Cite

@article{arxiv.2604.24979,
  title  = {Non-Oberbeck-Boussinesq effects in coldwater},
  author = {Gustavo Estay and Daisuke Noto and Hugo N. Ulloa},
  journal= {arXiv preprint arXiv:2604.24979},
  year   = {2026}
}

Comments

Update: Fix momentum equation