betaFeedback

WorksPhysics and Astronomy

Unifying heat transport model for the transition between buoyancy-dominated and Lorentz-force-dominated regimes in quasistatic magnetoconvection

Abstract

In magnetoconvection, the flow of electromagnetically conductive fluid is driven by a combination of buoyancy forces, which create the fluid motion due to thermal expansion and contraction, and Lorentz forces, which distort the convective flow structure in the presence of a magnetic field. The differences in the global flow structures in the buoyancy-dominated and Lorentz-force-dominated regimes lead to different heat transport properties in these regimes, reflected in distinct dimensionless scaling relations of the global heat flux (Nusselt number Nu\textrm{Nu}) versus the strength of buoyancy (Rayleigh number Ra\textrm{Ra}) and electromagnetic forces (Hartmann number Ha\textrm{Ha}). Here, we propose a theoretical model for the transition between these two regimes for the case of a quasistatic vertical magnetic field applied to a convective fluid layer confined between two isothermal, a lower warmer and an upper colder, horizontal surfaces. The model suggests that the scaling exponents γγ in the buoyancy-dominated regime, Nu∼Raγ\textrm{Nu}\sim\textrm{Ra}^γ, and ξξ in the Lorentz-force-dominated regime, Nu∼(Ha−2Ra)ξ\textrm{Nu}\sim(\textrm{Ha}^{-2}\textrm{Ra})^ξ, are related as ξ=γ/(1−2γ)ξ=γ/(1-2γ), and the onset of the transition scales with Ha−1/γRa\textrm{Ha}^{-1/γ}\textrm{Ra}. These theoretical results are supported by our Direct Numerical Simulations for 10≤Ha≤200010\leq \textrm{Ha}\leq2000, Prandtl number Pr=0.025\textrm{Pr}=0.025 and Ra\textrm{Ra} up to 10910^9 and data from the literature.

Introduction

Magnetoconvection (MC) governs most astro- and geophysical systems and is relevant to various engineering applications [25, 7]. The former include, for instance, outer layers of stars and liquid metal planetary cores [12], examples of the latter comprise liquid-metal batteries, induction heating, casting, liquid-metal cooling for nuclear fusion reactors and semiconductor crystal growth [6]. MC occurs in an electrically conducting fluid that is subjected both to a magnetic field and an imposed temperature gradient. The buoyancy forces induce convective fluid motion due to thermal expansion and contraction, while the magnetic field affects this motion and distorts the global flow structure through the Lorentz force, which eventually influences the heat transport in the system. The resulting main two control parameters, the strength of the imposed thermal driving and that of the external magnetic field, are encoded in independent dimensionless groups, the Rayleigh number Ra and Hartmann number Ha, respectively.

One of the key objectives in MC research is to provide scaling relations for the heat transport through the system, represented in dimensionless form by the Nusselt number Nu, as a function of Ra and Ha. However, the heat transport scaling relations also depend on the flow configuration, including the angle between the magnetic field and gravity, the geometry of the container and the boundary conditions (BCs), and on whether the buoyancy forces dominate over the Lorentz forces in the system or vice versa. This inherent complexity results in the need, at least in principle, to derive separate heat transport scaling relations to describe each specific flow regime itself and transitions between distinct regimes. The considerable difficulty of doing so in a coherent manner is exacerbated by non-universal scaling relations even within specific regimes – the scaling relations in the buoyancy-dominated and Lorentz-force-dominated regimes themselves change with the control parameters, and transitions between the different regimes are also non-universal.

The objective of this paper is to offer a unifying heat transport model for the transition between the buoyancy-dominated and Lorentz-force-dominated regimes in quasistatic MC. We focus on Rayleigh–Bénard convection (RBC) [2] with an applied vertical magnetic field and assume that the magnetic field is constant in the entire domain, without being affected by a fluid motion or finite magnetic diffusion. The model uses the theoretical predictions by Grossmann and Lohse [9, 10, 24] for RBC without magnetic field and transfers the approach by Ecke and Shishkina [8 §3.3] for transitions in rotating RBC to the case of RBC with a vertical magnetic field. To verify the proposed model, we compare the theoretical predictions with results for liquid metal MC obtained by direct numerical simulation (DNS) carried out by us and others [16, 27, 3, 26], as well as experiments [5, 13, 28, 26]. In addition, we carried out simulations for a different working fluid at higher Prandtl number Pr to compare our DNS data with that from Ref. [15]. The predictions of the proposed model agree well the experimental and DNS data.

Model for the transition between the buoyancy dominated and Lorentz force dominated regimes

We consider a layer of electrically conducting fluid confined between two infinitely wide and long plates, driven by a buoyancy force generated by an imposed vertical temperature difference between the top and bottom plates, and subjected to a uniform vertically orientated magnetic field. When the magnetic field is weak compared to the buoyancy force, we recover classical RBC scaling for the dimensionless convective heat flux Nu−1\mathrm{Nu}-1, that is, the total dimensionless heat flux Nu\mathrm{Nu} less its conductive contribution, Nu−1∼(Ra/Rac,b)γ∼Raγ\mathrm{Nu}-1 \sim(\mathrm{Ra}/\mathrm{Ra}_{c,b})^{\gamma} \sim\mathrm{Ra}^{\gamma}, for an exponent γ\gamma, where where Rac,b\mathrm{Ra}_{c,b} is the critical Ra for RBC bulk onset for a given container geometry. When the Lorentz force is strong compared to the buoyancy force, we expect a similar scaling law Nu−1∼(Ra/Rac,L)ξ\mathrm{Nu}-1 \sim(\mathrm{Ra}/\mathrm{Ra}_{c,L})^{\xi}, for an exponent ξ\xi. Here, the dependence on the critical Rayleigh number Rac,L\mathrm{Ra}_{c,L} in the Lorentz-force-dominated regime is kept due to its dependence on Ha, that can be obtained from the linear stability theory [4] Rac,L∼Ha2\mathrm{Ra}_{c,L} \sim\mathrm{Ha}^{2}, with no dependence on the Prandtl number.

Although these buoyancy-dominated and Lorentz-force-dominated scaling laws appear disconnected, they are intrinsically linked under the assumption that they must overlap at some intermediate region between the two extreme regimes, where neither the influence of the Lorentz force or buoyancy force on the convective heat transport can be ignored. At the transition between these regimes, the two corresponding scaling laws will scale in the same way,

Raγ∼Nu−1∼(Ha−2Ra)ξ,(1)\mathrm{Ra}^{\gamma} \sim\mathrm{Nu}-1 \sim\left(\mathrm{Ha}^{-2}\mathrm{Ra}\right)^{\xi}, \tag*{(1)}

To construct a relationship between the two scaling exponents γ\gamma and ξ\xi, we assume that this cross-over occurs when the thicknesses of the thermal and viscous boundary layers (BLs) scale in the same way, motivated by the observation that the viscous layer is nested within the thermal BL in the Lorentz-force-dominated regime and vice versa in the buoyancy-dominated regime. In a domain of height HH over a semi-infinite plate, the average thermal BL thickness from laminar Prandtl-Blasius boundary layer theory [18, 21] is δθ=H/(2Nu)∼Nu−1\delta_\theta= H/(2\mathrm{Nu}) \sim\mathrm{Nu}^{-1}, while the laminar viscous BL influenced by a vertical magnetic field is the Hartmann layer [11, 7], δHa=cH/Ha∼Ha−1\delta_{\mathrm{Ha}} = cH/\mathrm{Ha} \sim\mathrm{Ha}^{-1}, where cc is a constant. Assuming that δθ∼δHa\delta_\theta\sim\delta_{\mathrm{Ha}} at the transition implies Nu∼Ha\mathrm{Nu} \sim\mathrm{Ha}. Since this transition is seen to typically occur at high Nusselt numbers, meaning Nu≈Nu−1\mathrm{Nu} \approx\mathrm{Nu}-1, we obtain

Ha∼Nu∼Raγ∼RaξHa−2ξ∼Ra−2ξγ+ξ,(2)\mathrm{Ha} \sim\mathrm{Nu} \sim\mathrm{Ra}^{\gamma} \sim\mathrm{Ra}^{\xi}\mathrm{Ha}^{-2\xi} \sim\mathrm{Ra}^{-2\xi\gamma+\xi}, \tag*{(2)}

resulting in the following relationship between the exponents,

ξ=γ1−2γ,orγ=ξ1+2ξ.(3)\xi= \frac{\gamma}{1-2\gamma}, \quad\text{or} \quad\gamma= \frac{\xi}{1+2\xi}. \tag*{(3)}

One can see that a larger (smaller) exponent in one regime requires a larger (smaller) exponent the other regime, and that γ\gamma is always smaller than 1/21/2.

In Figure 1 we present a sketch of the proposed scaling relations for (Nu−1)(\mathrm{Nu}-1) versus Ha\mathrm{Ha} (Figure 1 a) and Ra\mathrm{Ra} (Figure 1 b), according to the relations (3). Once γ\gamma is known for any specific Pr\mathrm{Pr}, the exponent ξ\xi can be calculated from Eq. (3). These scalings can then be used define coordinates (Nu−1)Ra−γ(\mathrm{Nu}-1)\mathrm{Ra}^{-\gamma} and Ha−1/γRa\mathrm{Ha}^{-1/\gamma}\mathrm{Ra} with respect to which the heat transport dependence for different values of Ha\mathrm{Ha} and Ra\mathrm{Ra} collapse onto a master curve, as sketched in Figure 2. The transition then should take place in a Rayleigh-number range that scales as Ha1/γ\mathrm{Ha}^{1/\gamma}.

Scalings of Nu − 1 versus (a) Ha and (b) Ra, according to the theory.

Figure 1. Scalings of Nu − 1 versus (a) Ha and (b) Ra, according to the theory.

Figure 2. Schematic representation of the normalized convective heat transport Nu−1\mathrm{Nu}-1 displaying the transition from the Lorentz-force-dominated regime, Nu−1∼(Ha−2Ra)ξ\mathrm{Nu}-1 \sim(\mathrm{Ha}^{-2}\mathrm{Ra})^\xi, to the buoyancy-dominated regime, Nu−1∼Raγ\mathrm{Nu}-1 \sim\mathrm{Ra}^\gamma, according to our model. The scaling exponents ξ\xi and γ\gamma follow Eq. (3), while the transition scales with Ha−1/γRa\mathrm{Ha}^{-1/\gamma}\mathrm{Ra}.

To close the model, a theoretical prediction for either γ\gamma or ξ\xi must be made. The former is readily available through Grossmann–Lohse (GL) theory [9, 10, 24] for RBC without magnetic field, applicable here in the buoyancy-dominated regime. For any given Pr\mathrm{Pr} and Ra\mathrm{Ra}-range, the theory provides accurate predictions of the value of γ\gamma, for containers of aspect ratio Γ≳1\Gamma\gtrsim1. For Γ≪1\Gamma\ll1, the data can be rescaled according to the method suggested in [22, 1], which we do not discuss here, as in the present study Γ=1\Gamma= 1.

Status Quo – experimental and numerical data.

To verify the theoretical model we compare its predictions against data obtained from experiments of liquid-metal MC [5, 13, 28, 26] and DNS conducted by us and others [16, 27, 3, 26]. However, before doing so we provide a brief overview of the data collated from the literature and produced by us to demonstrate the considerable challenges that arise when trying to draw firm conclusions on the scaling of the heat transport with HaHa and RaRa.

We simulate an incompressible, viscous buoyancy-driven flow of an electrically conducting fluid in the presence of an imposed magnetic field for very small magnetic Reynolds number Rem≪1Re_m \ll1 and magnetic Prandtl number Prm≪1Pr_m \ll1 by numerically solving the MC equations within the Oberbeck–Boussinesq and quasistatic approximation,

∂u∂t+u⋅∇u+∇p=PrRa[∇2u+Ha2(j×eB)]+Tez,(4)\frac{\partial\boldsymbol{u}}{\partial t}+\boldsymbol{u}\cdot\nabla\boldsymbol{u}+\nabla p=\sqrt{\frac{Pr}{Ra}}\left[\nabla^2\boldsymbol{u}+Ha^2(\boldsymbol{j}\times\boldsymbol{e}_B)\right]+T\boldsymbol{e}_z, \tag*{(4)}
∂T∂t+u⋅∇T=1RaPr∇2T,(5)\frac{\partial T}{\partial t}+\boldsymbol{u}\cdot\nabla T=\frac{1}{\sqrt{RaPr}}\nabla^2T, \tag*{(5)}
∇⋅u=0,(6)\nabla\cdot\boldsymbol{u}=0, \tag*{(6)}
j=−∇ϕ+u×eB,(7)\boldsymbol{j}=-\nabla\phi+\boldsymbol{u}\times\boldsymbol{e}_B, \tag*{(7)}
∇2ϕ=∇⋅(u×eB).(8)\nabla^2\phi=\nabla\cdot(\boldsymbol{u}\times\boldsymbol{e}_B). \tag*{(8)}

where u\boldsymbol{u} is the velocity, TT the temperature, pp the kinematic pressure, j\boldsymbol{j} the electric current density, ϕ\phi the electric potential, and ez\boldsymbol{e}_z and eB\boldsymbol{e}_B are unit vectors that point, respectively, upward (opposite to gravity) and in the direction of the magnetic field B=B0eB\boldsymbol{B}=B_0\boldsymbol{e}_B. The magnetic field is aligned with the buoyancy force, eB=ez\boldsymbol{e}_B=\boldsymbol{e}_z.

Equations (4)–(8) have been non-dimensionalized using the container height HH, the free-fall velocity uff≡(αgHΔ)1/2u_{ff}\equiv(\alpha gH\Delta)^{1/2}, the free-fall time tff≡H/ufft_{ff}\equiv H/u_{ff}, the temperature difference between the bottom and top plates, Δ≡T+−T−\Delta\equiv T_+-T_-, and the external magnetic field strength, B0B_0, as units of length, velocity, time, temperature and magnetic field strength, respectively. The dimensionless control parameters are the Rayleigh number RaRa, the Prandtl number PrPr, and the Hartmann number HaHa,

Ra≡αgΔH3κν,Pr≡νκ,Ha≡B0Hσρν,(9)Ra\equiv\frac{\alpha g\Delta H^3}{\kappa\nu},\qquad Pr\equiv\frac{\nu}{\kappa},\qquad Ha\equiv B_0H\sqrt{\frac{\sigma}{\rho\nu}}, \tag*{(9)}

where σ\sigma is the electrical conductivity, ρ\rho the mass density, α\alpha the thermal expansion coefficient, gg the acceleration due to gravity, ν\nu the kinematic viscosity, and κ\kappa the thermal diffusivity. We apply no-slip BCs for the velocity at all boundaries, u=0\boldsymbol{u}=0, constant temperatures at the end-faces, i.e., T=T+T=T_+ at the bottom plate at z=0z=0 and T=T−T=T_- at the top plate at z=Hz=H, and adiabatic BC at the side walls, ∂T/∂n=0\partial T/\partial n=0, where n\boldsymbol{n} is the vector orthogonal to the surface. All solid boundaries are considered electrically insulated; Neumann BCs for the electric potential are ∂ϕ/∂n=0\partial\phi/\partial n=0. The simulation domain is cubic of height HH, width WW, length LL, H=W=LH=W=L, i.e. has aspect ratio Γ≡L/H=1\Gamma\equiv L/H=1. Most simulations are carried out for liquid metals such as GaInSnGaInSn at Pr=0.025Pr=0.025, for RaRa up to 10910^9 and HaHa up to 2000. Some DNS are conducted also for Pr=8Pr=8, to extend the parameter range studied in Ref. [15]. In the bulk, spatial flow fluctuations are resolved down to 2–5, occasionally 10 Kolmogorov microscales, near the rigid walls we resolve the thermal and Hartmann BLs [23]. Our DNSs have been carried out with an MC extension of GOLDFISH [14, 20, 19]. Further details of the simulations are provided in appendix A.

In Figure 3 we present Nu−1Nu-1 as a function of RaRa (Figure 3 a-Figure 3b) and HaHa (Figure 3 c-Figure 3d), respectively, from our DNS, experimental [5, 13, 28, 26] and DNS data [16, 3, 26] for liquid metals, 0.025≤Pr≤0.0290.025\le Pr\le0.029 (Figure 3 b, Figure 3d) and a fluid with Pr=8Pr=8 (Figure 3 a, Figure 3c). In addition, in Figure 3 (b, Figure 3d), we plot for comparison the DNS data [27] for free-slip BCs. In Figure 3 (a-Figure 3b) one can see that NuNu generally increases with growing RaRa, but with different slopes for different HaHa, which are steeper for larger HaHa. In the double logarithmic plots of Figure 3 (a-Figure 3b), the curves of the (Nu−1)(Nu-1)-vs.-RaRa dependences for different HaHa approach each other when RaRa increases. In Figure 3 (c-Figure 3d) one can see that NuNu remains to be almost unaffected by the magnetic field for relatively small HaHa, but for a strong Lorentz force (large HaHa) they gradually decrease with growing HaHa. Here, again, the decreasing slopes are different for different RaRa, and the transition to the regime, where the heat transport is affected by the magnetic field, depends on RaRa. In summary, the data in Figure 3 look rather different in different experiments and DNS. In what follows, we show that our model results in a collapse of all data points on a single master curve.

Figure 3. The dimensionless convective heat transport, i.e., Nu−1\mathrm{Nu}-1, as functions of (a, b) Ra\mathrm{Ra} and (c-d) Ha\mathrm{Ha}, for (a, c) Pr=8\mathrm{Pr}=8 and (b, d) 0.025≤Pr≤0.0290.025 \leq\mathrm{Pr} \leq0.029. The color scales are according to (a-b) Ha\mathrm{Ha} and (c-d) Ra\mathrm{Ra}. Legend: Γ=1\Gamma=1 box, Pr=8\mathrm{Pr}=8, DNS; Γ=1\Gamma=1 box, Pr=0.025\mathrm{Pr}=0.025, DNS; Γ=1\Gamma=1 box, Pr=8\mathrm{Pr}=8, DNS Lim et al. 2019; Γ=1\Gamma=1 cyl., Pr=0.025\mathrm{Pr}=0.025, DNS Akhmedagaev et al. 2020; Γ=4\Gamma=4 box, Pr=0.025\mathrm{Pr}=0.025, DNS Liu et al. 2018; Γ=1\Gamma=1 cyl., Pr=0.025\mathrm{Pr}=0.025, DNS Liu et al. 2019; Γ=3.30\Gamma=3.30 box, Pr=0.025\mathrm{Pr}=0.025, DNS Yan et al. 2019; Γ=1.655\Gamma=1.655 box, Pr=0.025\mathrm{Pr}=0.025, DNS Yan et al. 2019; Γ=2\Gamma=2 cyl., Pr=0.025\mathrm{Pr}=0.025, DNS Xu et al. 2023; Γ=1\Gamma=1 cyl., Pr=0.027\mathrm{Pr}=0.027, Exp. Xu et al. 2023; Γ=2\Gamma=2 cyl., Pr=0.027\mathrm{Pr}=0.027, Exp. Xu et al. 2023; Γ=1\Gamma=1 cyl., Pr=0.025\mathrm{Pr}=0.025, Exp. King and Aurnou 2015; Γ=2\Gamma=2 cyl., Pr=0.026\mathrm{Pr}=0.026, Exp. Grannan et al. 2022; Γ=1\Gamma=1 cyl., Pr=0.025\mathrm{Pr}=0.025, Exp. Cioni et al. 2000; Γ=1\Gamma=1 cyl., Pr=0.029\mathrm{Pr}=0.029, Exp. Zürner et al. 2020.

Model validation

We now validate the model using the data presented in Figure 3. To calculate the scaling exponent γ\gamma in the buoyancy-dominated regime, for the considered Pr\mathrm{Pr} and Ra\mathrm{Ra}-ranges, we use GL theory, which gives γ\gamma about 0.30 for Pr=8\mathrm{Pr}=8 and about 0.31 for Pr=0.025\mathrm{Pr}=0.025. These values agree very well with fits to data for the Ra\mathrm{Ra}-ranges in the buoyancy-dominated regime for both values of Pr\mathrm{Pr}, see appendix A for further details. Using Eq. (3) with γ=0.30\gamma=0.30 for Pr=8\mathrm{Pr}=8 and γ=0.31\gamma=0.31 for Pr=0.025\mathrm{Pr}=0.025, we calculate the exponent ξ\xi in the Lorentz-force-dominated regime, which equals ξ=0.75\xi=0.75 for Pr=8\mathrm{Pr}=8 and ξ≈0.82\xi\approx0.82 for Pr=0.025\mathrm{Pr}=0.025. In Figure 4 we plot all data presented previously Figure 3 using the coordinates suggested by our model and visualised in Figure 2. This results in a clear collapse of the data onto master curves for Pr=8\mathrm{Pr}=8 (Figure 4a) and 0.025≤Pr≤0.0290.025 \leq\mathrm{Pr} \leq0.029 (Figure 4b). Some deviation of the data from the master curve in Figure 4b is observed when Ra\mathrm{Ra} is relatively small and the flow is in the wall-mode regime, that is, before the onset of bulk convection where heat transport is confined to the near-wall region. Since the influence of no-slip side-walls is outside of the scope of our model, the observed deviations are expected in this regime.

Figure 4. All data from Figure 3 follow master scaling curves if plotted as Figure 2 suggests, for (a) Pr=8\mathrm{Pr}=8 and (b) 0.025≤Pr≤0.0290.025 \le\mathrm{Pr} \le0.029. The values of γ\gamma are calculated from GL theory, and the values of ξ\xi are calculated from Eq. (3). Pink and blue lines show the predictions of the slopes in the buoyancy and Lorentz-force-dominated regimes, respectively. The symbols have the same meaning as in Figure 3.

Conclusion

In conclusion, we have proposed a heat transport model for the transition between the buoyancy- and Lorentz-force-dominated regimes of vertical MC. We validated the model using our DNS and data available in the literature. We wish to emphasize that the proposed model is parameter-free. For a given Pr\mathrm{Pr} and Ra-range, one can calculate the scaling exponent in the buoyancy-dominated regime, using GL theory. Then, using Eq. (3), one can calculate the scaling exponent ξ\xi in the Lorentz-force-dominated regime and collapse the data on a master curve by rescaling the coordinate axes as in Figure 2. The model can in principle be extended to include the effect of a fluctuating magnetic field, this merely results in an adjustment of the prefactors.

Acknowledgements

The authors thank R. E. Ecke, D. Lohse and G. Vasil for fruitful discussions and R. Akhmedagaev for providing data and acknowledge the financial support from the Deutsche Forschungsgemeinschaft (SPP1881 ”Turbulent Superstructures” and grants Sh405/7, Sh405/16 and Li3694/1). This work used the ARCHER2 UK National Supercomputing Service (https://www.archer2.ac.uk). Computing resources have been provided by the UK Turbulence Consortium (EPSRC grants EP/R029326/1 and EP/X035484/1).

Appendix

Numerical simulations

All direct numerical simulations have been carried using the direct numerical solver GOLDFISH [14], which has been widely used in previous studies of different convective flows. The new version of the code that applies a fourth-order finite-volume discretisation on staggered grids and a third-order Runge–Kutta time marching scheme [20, 19] has been extended to simulate magnetoconvective flows, where a consistent and conservative scheme [17] is utilised to calculate the current density and the Lorentz force. The three-dimensional DNS are performed using staggered grids that provide fine resolution in the core part of the domain and near the rigid walls [23], to resolve the thermal and Hartmann boundary layers. The main parameters and key observables of the simulations are summarised in Table 1.

Pr\mathrm{Pr}Ha\mathrm{Ha}Ra\mathrm{Ra}Nu\mathrm{Nu}σNu\sigma_{\mathrm{Nu}}NxN_xNyN_yNzN_zTrunT_{\mathrm{run}}NθN_\thetaNHaN_{\mathrm{Ha}}hDNS/hKh_{\mathrm{DNS}}/h_K
85002.0×1062.0 \times10^62.420.011001003003509080.40
5005.0×1065.0 \times10^63.860.021001003005507080.60
0.025101.0×1051.0 \times10^53.080.041501502004049361.06
105.0×1055.0 \times10^54.760.211501502004537361.76
101.0×1061.0 \times10^65.700.321501502006533362.20
105.0×1065.0 \times10^68.770.6715015020010025363.64
101.0×1071.0 \times10^710.680.7815015020016022364.49*
101.0×1081.0 \times10^820.421.982502503505029675.42
105.0×1085.0 \times10^833.181.472502503509021678.76
101.0×1091.0 \times10^940.472.323503504504025878.64*
1005.0×1055.0 \times10^53.290.062002002502060101.07
1001.0×1061.0 \times10^64.390.142002002503050101.34
1005.0×1065.0 \times10^67.740.562002002504035102.40
1001.0×1071.0 \times10^79.690.442002002506031103.09*
1001.0×1081.0 \times10^819.721.362202203004524135.81
1005.0×1085.0 \times10^833.101.762502503506521168.45*
2001.0×1061.0 \times10^62.620.03220220350120101151.13
2002.0×1062.0 \times10^63.690.0122022035022584151.46
2005.0×1065.0 \times10^65.780.2222022035022566152.06
2001.0×1071.0 \times10^77.870.3922022035016056152.65*
2001.0×1081.0 \times10^818.061.8225025040010042175.07
2005.0×1085.0 \times10^830.562.6325025040010032178.66*
5005.0×1065.0 \times10^63.510.062202203506508691.82
5002.0×1072.0 \times10^76.700.342202203501906193.02
5005.0×1075.0 \times10^710.730.642202203503004894.27*
5001.0×1081.0 \times10^813.880.7625025040050048114.74
5005.0×1085.0 \times10^826.751.7725025040015034118.36
5001.0×1091.0 \times10^934.442.59250250400250301110.64*
10001.0×1071.0 \times10^72.410.262202203006009051.92*
10002.0×1072.0 \times10^73.020.0822022030013008052.30
10005.0×1075.0 \times10^75.840.372202203005005653.02
10001.0×1081.0 \times10^89.390.5122022030011004453.77
10005.0×1085.0 \times10^822.781.35220220300601856.44*
20005.0×1075.0 \times10^72.530.092502504009011652.20
20001.0×1081.0 \times10^84.500.112502504001608652.68
20005.0×1085.0 \times10^813.600.632502504002204954.47*

Table 1. Details on the conducted DNS, including the standard deviation σNu\sigma_{\mathrm{Nu}} of the Nusselt number Nu\mathrm{Nu}, the number of nodes NxN_x, NyN_y, NzN_z in the directions xx, yy and zz, respectively; the number of simulation free-fall times used for averaging TrunT_{\mathrm{run}}; the number of the nodes within the thermal boundary layer, and within the Hartmann boundary layer NHaN_{\mathrm{Ha}}; the Kolmogorov microscale, hKh_K, and the relative mean grid stepping, hDNS/hKh_{\mathrm{DNS}}/h_K. Simulations marked by * are those for which grid refinement studies have been completed.

The DNS dataset comprises 37 simulations to cover the necessary ranges in Ha and Ra. To obtain a dataset this large within the available resources, a compromise had to be made in terms of the bulk resolution. Hence a number of simulations do not resolve scales close to the Kolmogorov microscale. To ensure accurate predictions of mean Nu, grid refinement studies have been done for key simulations (marked by * in the hDNS/hKh_{\mathrm{DNS}}/h_{\mathrm{K}} column). For these cases, changing the grid resolution by a factor of at least 2.5 resulted in changes of Nu by less than 1% over a long-time average.

Grossmann-Lohse scaling

Figure 5 shows a comparison between the predictions of Grossmann-Lohse theory [9, 10, 24] for the dependence of the dimensionless heat transport Nu−1\mathrm{Nu}-1 on the Rayleigh number Ra\mathrm{Ra}, Nu−1=ARaγ\mathrm{Nu}-1=A\mathrm{Ra}^{\gamma} for Pr-dependent prefactor AA. Fits to data have been carried out for 3×106≤1093 \times10^{6} \le10^{9} for Pr = 0.025 and 3×106≤1093 \times10^{6} \le10^{9} for Pr = 8. As can been seen from the data presented in Figure 5, the theoretical predictions agree very well with the data.

Predictions of Grossmann–Lohse theory for Nu − 1 versus Ra scaling relations

Figure 5. Predictions of GL theory [9, 10, 24] (thin lines) for Nu − 1 versus Ra scaling relations in RBC without magnetic field, for Pr = 8 (blue) and Pr = 0.025 (pink). The power fits for the considered Ra-ranges are highlighted with thick lines of the corresponding colors.

Flow and temperature field visualisations

Figures 6 and 7 present visualisations of the velocity magnitude and temperature, respectively, at an instant in time during statistically steady evolution in the magnetically dominated and the buoyancy-dominated regimes. In both figures, the respective panels (a) correspond to Ra⁡=107\operatorname{Ra} = 10^{7}, Ha⁡=1000\operatorname{Ha} = 1000, panels (b) to Ra⁡=109\operatorname{Ra} = 10^{9}, Ha⁡=10\operatorname{Ha} = 10. As can be seen from a comparison of the flow and temperature field between both regimes, both velocity and temperature fluctuate on much smaller scales in the buoyancy-dominated regime. The magnetically dominated regime is shows strong vertical flows near the walls, and we note the absence of plumes in the temperature field.

Instantaneous velocity magnitude on the y mid-plane

Figure 6. Instantaneous velocity magnitude UU shown on the yy mid-plane for (a) Ra⁡=107\operatorname{Ra} = 10^{7}, Ha⁡=1000\operatorname{Ha} = 1000 and (b) Ra⁡=109\operatorname{Ra} = 10^{9}, Ha⁡=10\operatorname{Ha} = 10.

Instantaneous temperature field on the y mid-plane

Figure 7. Instantaneous temperature field TT shown on the yy mid-plane for (a) Ra⁡=107\operatorname{Ra} = 10^{7}, Ha⁡=1000\operatorname{Ha} = 1000 and (b) Ra⁡=109\operatorname{Ra} = 10^{9}, Ha⁡=10\operatorname{Ha} = 10.

References

  1. [1]G. Ahlers, E. Bodenschatz, R. Hartmann, X. He, D. Lohse, P. Reiter, R.J.A.M. Stevens, R. Verzicco, M. Wedi, S. Weiss, X. Zhang, L. Zwirner, and O. Shishkina. Aspect ratio dependence of heat transfer in a cylindrical Rayleigh–Bénard cell. Phys. Rev. Lett., 128:084501, 2022.DOI
  2. [2]G. Ahlers, S. Grossmann, and D. Lohse. Heat transfer and large scale dynamics in turbulent Rayleigh–Bénard convection. Rev. Mod. Phys., 81:503–537, 2009.DOI
  3. [3]R. Akhmedagaev, O. Zikanov, D. Krasnov, and J. Schumacher. Turbulent Rayleigh–Bénard convection in a strong vertical magnetic field. J. Fluid Mech., 895:R4, 2020.DOI
  4. [4]S. Chandrasekhar. Hydrodynamic and Hydromagnetic Stability. Clarendon Press, 1961.
  5. [5]S. Cioni, S. Chaumat, and J. Sommeria. Effect of a vertical magnetic field on turbulent Rayleigh–Bénard convection. Phys. Rev. E, 62:R4520–R4523, 2000.pubmed.ncbi.nlm.nih.gov/11089073
  6. [6]P. A. Davidson. Magnetohydrodynamics in materials processing. Annu. Rev. Fluid Mech., 31:273–300, 1999.DOI
  7. [7]P. A. Davidson. Introduction to Magnetohydrodynamics. Cambridge University Press, 2016.DOI
  8. [8]R. E. Ecke and O. Shishkina. Turbulent rotating Rayleigh–Bénard convection. Annu. Rev. Fluid Mech., 55:603–638, 2023.DOI
  9. [9]S. Grossmann and D. Lohse. Scaling in thermal convection: A unifying theory. J. Fluid Mech., 407:27–56, 2000.
  10. [10]S. Grossmann and D. Lohse. Thermal convection for large Prandtl numbers. Phys. Rev. Lett., 86:3316–3319, 2001.pubmed.ncbi.nlm.nih.gov/11327959
  11. [11]J. Hartmann and F. Lazarus. Hg-dynamics ii, experimental investigations on the flow of mercury in a homogeneous magnetic field. Det Kgl Danske Videnskabernes Selskab Math-fys Medd, 15(7):1–45, 1937.
  12. [12]C. A. Jones. Planetary magnetic fields and fluid dynamos. Annu. Rev. Fluid Mech., 43:583–614, 2011.DOI
  13. [13]E. M. King and J. M. Aurnou. Magnetostrophic balance as the optimal state for turbulent magnetoconvection. Proc. Natl. Acad. Sci., 112:990–994, 2015.DOI
  14. [14]G. L. Kooij, M. A. Botchev, E. M.A. Frederix, B. J. Geurts, S. Horn, D. Lohse, E. P. van der Poel, O. Shishkina, R. J. A. M. Stevens, and R. Verzicco. Comparison of computational codes for direct numerical simulations of turbulent Rayleigh–Bénard convection. Comp. Fluids, 166:1–8, 2018.DOI
  15. [15]Z. L. Lim, K. L. Chong, G.-Y. Ding, and K.-Q. Xia. Quasistatic magnetoconvection: heat transport enhancement and boundary layer crossing. J. Fluid Mech., 870:519–542, 2019.arxiv.org/abs/1902.08754
  16. [16]W. Liu, D. Krasnov, and J. Schumacher. Wall modes in magnetoconvection at high Hartmann numbers. J. Fluid Mech., 849:R2, 2018.DOI
  17. [17]M.-J. Ni and J.-F. Li. A consistent and conservative scheme for incompressible MHD flows at a low magnetic Reynolds number. Part III: On a staggered mesh. J. Comput. Phys., 231:281–298, 2012.DOI
  18. [18]L. Prandtl. Über Flüssigkeitsbewegung bei sehr kleiner Reibung. In Verhandlungen des III. Int. Math. Kongr., Heidelberg, 1904, pages 484–491. Teubner, 1905.
  19. [19]P. Reiter, O. Shishkina, D. Lohse, and D. Krug. Crossover of the relative heat transport contributions of plume ejecting and impacting zones in turbulent Rayleigh–Bénard convection. Europhys. Lett., 134:34002, 2021.
  20. [20]P. Reiter, X. Zhang, and O. Shishkina. Flow states and heat transport in Rayleigh–Bénard convection with different sidewall boundary conditions. J. Fluid Mech., 936:A32, 2022.arxiv.org/abs/2111.00971
  21. [21]H. Schlichting. Boundary layer theory. McGraw-Hill, 1979.
  22. [22]O. Shishkina. Rayleigh–Bénard convection: The container shape matters. Phys. Rev. Fluids, 6:090502, 2021.DOI
  23. [23]O. Shishkina, R. J. A. M. Stevens, S. Grossmann, and D. Lohse. Boundary layer structure in turbulent thermal convection and its consequences for the required numerical resolution. New J. Phys., 12:075022, 2010.DOI
  24. [24]R. J. A. M. Stevens, E. P. van der Poel, S. Grossmann, and D. Lohse. The unifying theory of scaling in thermal convection: The updated prefactors. J. Fluid Mech., 730:295–308, 2013.DOI
  25. [25]N. O. Weiss and M. R. E. Proctor. Magnetoconvection. Cambridge University Press, 2014.
  26. [26]Y. Xu, S. Horn, and J. M. Aurnou. The transition from wall modes to multimodality in liquid gallium magnetoconvection. https://arxiv.org/pdf/2303.08966.pdf, 2023.
  27. [27]Yan, M. A. Calkins, S. Maffei, K. Julien, S. Tobias, and P. Marti. Heat transfer and flow regimes in quasi-static magnetoconvection with a vertical magnetic field. J. Fluid Mech., 877:1186–1206, 2019.
  28. [28]Zürner, F. Schindler, T. Vogt, S. Eckert, and J. Schumacher. Flow regimes of rayleigh–bénard convection in a vertical magnetic field. J. Fluid Mech., 894:A21, 2020.DOI

Paper details

vlnc:W_n2khewe6
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents