Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-17T02:31:40.697Z Has data issue: false hasContentIssue false

Experimental study of the convection in a thin cylindrical gas layer with imposed bottom and top fluxes and imposed side temperature

Published online by Cambridge University Press:  11 March 2025

F. Rein*
Affiliation:
Aix Marseille University, CNRS, Centrale Marseille, IRPHE, 13013 Marseille, FranceIRSN, 13108 St Paul lez Durance, France
L. Carénini
Affiliation:
IRSN, 13108 St Paul lez Durance, France
F. Fichot
Affiliation:
IRSN, 13108 St Paul lez Durance, France
B. Favier
Affiliation:
Aix Marseille University, CNRS, Centrale Marseille, IRPHE, 13013 Marseille, France
M. Le Bars
Affiliation:
Aix Marseille University, CNRS, Centrale Marseille, IRPHE, 13013 Marseille, France
*
Corresponding author: F. Rein, florian.rein@protonmail.com

Abstract

We investigate convection in a thin cylindrical gas layer with an imposed flux at the bottom and a fixed temperature along the side, using a combination of direct numerical simulations and laboratory experiments. The experimental approach allows us to extend by two orders of magnitude the explored range in terms of flux Rayleigh number. We identify a scaling law governing the root-mean-square horizontal velocity and explain it through a dimensional analysis based on heat transport in the turbulent regime. Using particle image velocimetry, we experimentally confirm, for the most turbulent regimes, the presence of a drifting persistent pattern consisting of radial branches, as identified by Rein et al. (2023, J. Fluid Mech. 977, A26). We characterise the angular drift frequency and azimuthal wavenumber of this pattern as functions of the Rayleigh number. The system exhibits a wide distribution of heat flux across various time scales, with the longest fluctuations attributed to the branch pattern and the shortest to turbulent fluctuations. Consequently, the branch pattern must be considered to better forecast important wall heat flux fluctuations, a result of great relevance in the context of nuclear safety, the initial motivation for our study.

JFM classification

Information

Type
JFM Papers
Copyright
© The Author(s), 2025. Published by Cambridge University Press
Figure 0

Figure 1.$(a)$ Photo of the experimental set-up with the laser sheet inclined for better visualisation (it is horizontal during acquisition). The dotted yellow rectangular region shows the PIV camera field of view. $(b)$ Sketch of a vertical cross-section of the experimental set-up with magnified views of the bottom plate configuration below the working tank and below the side boundary, respectively.

Figure 1

Figure 2.Sketch of the modelled fluid layer in a vertical plane through the cylinder. The top boundary condition is no slip for the gas experiment simulations, representing the lid used in the set-up, while a free-slip condition is applied to model the metal layer in the nuclear application.

Figure 2

Figure 3.$(a)$ Non-dimensional temperature averaged in time and over the six thermocouples in the bulk of the experiment, as a function of $Ra_{\phi }$. The dash–dot line shows the $Ra_{\phi }^{-1/5}$ asymptotic scaling from Rein et al. (2023a) (the inset shows the compensated plot). $(b)$ Non-dimensional horizontal r.m.s. velocity in the $(x,y)$ plane at height $z= {6}\,\text {cm}$ as a function of $Ra_{\phi }$. The dash–dot line and continuous line show a $1/3$ and $2/5$ best fit power law based on experimental data. For both plots, square ($\square$), diamond ($\Diamond$) and circle ($\circ$) symbols represent respectively data from DNS simulations at $R_F=0.5$ and $R_F=0.6$, and the experiments. Empty circles indicate when ${\rm SF}_6$ gas is used. For all DNS simulations, $\Gamma =4$ and $Pr=0.7$.

Figure 3

Figure 4.$(a)$ Compensated horizontal advective flux following (4.4) as a function of $Ra_{\phi }$. $(b)$ Compensated radial r.m.s. velocity following (4.5) as a function of $Ra_{\phi }$. For both plots, different symbols correspond to different aspect ratios $\Gamma$ and colour variation from red to green involves $R_F$ variation from 0.1 to 0.9. Data are from DNS with $Pr=0.1$ and a free-slip upper boundary, for which our data set is more complete in terms of $\Gamma$ and $R_F$ systematics, but the same behaviour is expected with the experimental configuration. Empty/full symbols indicate respectively filtered/DNS simulations. The shadings represent three times the standard deviation of the horizontal advective flux time series at the statistically stationary state.

Figure 4

Figure 5.$(a)$ Hovmöller (space–time) diagram of the radial velocity averaged along the radial direction. The radial velocity is measured at $z=6$ cm (i.e. $z/H=0.54$). $(b)$ Average of the radial velocity time spectra from each PIV measurement location. The vertical dash–dot line shows the frequency ($f_{{max}} =6.7\times 10^{-2}$ Hz) of the peak amplitude. For both, data correspond to experiment $9$ from table 1 where $Ra_{\phi } = 1.26\times 10^8$, using air.

Figure 5

Figure 6.$(a)$ Number of branches as a function of $Ra_{\phi }$. The insets show two snapshots of the radial velocity maps from DNS at $z=0.66$ for $Ra_{\phi }=10^5$ and $10^8$, $R_F=0.6$, $Pr=0.7$ and $\Gamma =4$. $(b)$ Angular drift frequency as a function of $Ra_{\phi }$. The dashed–dotted line represents a best fit based on the experimental data for $Ra_\phi \gt 5 \times 10^7$. In both graphs, square symbols ($\square$), diamond symbols ($\Diamond$) and circle symbols ($\circ$) respectively represent data for DNS at $R_F=0.5$ and $R_F=0.6$, and experimental data. Empty circles indicate the use of ${\rm SF}_6$. All DNS were performed with $Pr=0.7$ and $\Gamma =4$.

Figure 6

Figure 7.Phase-averaged temperature, radial velocity and azimuthal velocity fields, processed during 1 diffusive time. Results are presented for three horizontal slices at $z=0.1$, $0.5$ and $0.9$. Parameters are $\Gamma =4$, $R_F=0.1$, $Ra_{\phi }=10^5$ and $Pr=0.1$. In the present DNS, the pattern drifts anticlockwise.

Figure 7

Figure 8.Wall heat flux PDF from DNS with different values of $Ra_{\phi }$. The fluxes have been scaled by the time-averaged flux $\Phi _{{side}}$. Input parameters are $Pr=0.1$, $R_F=0.1$ and $\Gamma =4$.

Figure 8

Figure 9.(a) Three-dimensional view of the temperature field at the bottom plate (note the upside-down axes system with $z$ pointing downward) with $\Gamma = 4$, $Ra_{\phi }=10^7$, $R_F=0.1$ and $Pr=0.1$. (b) Hovmöller (space–time) diagram from this simulation of the wall heat flux averaged in $z$ and scaled by the mean wall flux $\Phi _{{side}}$.

Figure 9

Figure 10.$(a)$ The PDF of the phase-averaged wall heat flux as a function of $Ra_{\phi }$. The other input parameters are $Pr=0.1$, $R_F=0.1$ and $\Gamma =4$. All depths and azimuths along the side wall have been considered. $(b)$ Same but for the wall heat flux fluctuations related to the phase average. For both plots, the fluxes have been scaled by the time-averaged flux $\Phi _{{side}}$.

Figure 10

Figure 11.Temporal evolution of the wall heat flux measured by our probe spanning a width of 18 mm and a depth $z\in [{4,6}\,\text {cm}, {6,4}\,\text {cm}]$ (orange), and radial velocity measured by PIV at $r\,=\,{34}\,\text {cm}$, $z\,=\,{6}\,\text {cm}$, in the direction aligned with the flux sensor (green). The time is shifted by 2.6 s for the velocity. For each signal, the mean temporal component has been removed to keep only the fluctuating part, which is then normalised by the maximum value of the signal. Data from experiment 3 in table 1.

Figure 11

Figure 12.(a) Wall heat flux PDF for experiment 9$n^o 9$ in table 1 (orange $\circ$) and for a DNS with $Ra_{\phi }=10^8$, $\Gamma = 4$, $Pr=0.7$, $R_F=0.6$ (green $\Diamond$). The continuous blue thick line represents the low-pass filtered DNS results. The thin red line represents a fit by a gamma distribution. The fluxes are all scaled by the time-averaged flux $\Phi _{{side}}$. (b) Same for experiment 9$n^o 9$ and for DNS heat flux fluctuations due to the drifting pattern only (see § 4.4.1).

Supplementary material: File

Rein et al. supplementary material 1

Rein et al. supplementary material
Download Rein et al. supplementary material 1(File)
File 21.1 KB
Supplementary material: File

Rein et al. supplementary material movie 1

Rein et al. supplementary material movie
Download Rein et al. supplementary material movie 1(File)
File 40.8 MB