betaFeedback

WorksPhysics and Astronomy

Turbulent Rayleigh-Bénard convection with bubbles attached to the plate

Abstract

We numerically investigate turbulent Rayleigh-Bénard convection with gas bubbles attached to the hot plate, mimicking a core feature in electrolysis, catalysis, or boiling. The existence of bubbles on the plate reduces the global heat transfer due to the much lower thermal conductivity of gases as compared to liquids and changes the structure of the boundary layers. The numerical simulations are performed in 3D at Prandtl number Pr=4.38 (water) and Rayleigh number 107≤Ra≤10810^7\le Ra \le10^8. For simplicity, we assume the bubbles to be equally-sized and having pinned contact lines. We vary the total gas-covered area fraction 0.18≤S0≤0.620.18 \le S_0 \le 0.62, the relative bubble height 0.02≤h/H≤0.050.02\le h/H \le0.05 (where HH is the height of the Rayleigh-Bénard cell), the bubble number 40≤n≤14440 \le n \le 144, and their spatial distribution. In all cases, asymmetric temperature profiles are observed, which we quantitatively explain based on the heat flux conservation at each horizontal section. We further propose the idea of using an equivalent single-phase setup to mimic the system with attached bubbles. Based on this equivalence, we can calculate the heat transfer. Without introducing any free parameter, the predictions for the Nusselt number, the upper and lower thermal boundary layer thicknesses, and the mean centre temperature well agree with the numerical results. Finally, our predictions also work for the cases with much larger Pr (e.g. 400400), which indicates that our results can also be applied to predict the mass transfer in water electrolysis with bubbles attached to the electrode surface or in catalysis.

Introduction

In wall-bounded buoyancy-driven turbulence, the boundary conditions play a crucial role in the flow structure and the global transport property of the system. Bubbles attached to the wall affect these boundary conditions. They often occur in various industrial applications. One example is water electrolysis, where bubbles are generated at the electrodes and can significantly reduce the global mass transport of the system by reducing the active electrode area [26, 28, 31, 18], leading to the decrease of the electrolyser efficiency. Another example is catalysis, where bubbles are generated by chemical reactions and can block the catalytic surface, thus also reducing the mass transport [21, 16, 29]. One example from daily life is heating water. When a pot of water is heated from below, many tiny gas bubbles nucleate at the bottom wall, reducing the heat transfer efficiency of the system due to the much lower thermal conductivity of gas as compared to the liquid. In all of these examples, bubbles attached to the wall influence the boundary layer (BL) and thus affect the global heat or mass transfer of the system. Therefore, it is highly desirable to quantitatively understand how much the global transport properties are changed. This is the motivation for our study.

As model system we choose Rayleigh-Bénard (RB) convection which is the paradigm of thermally driven turbulence (see the reviews of [2, 15, 4, 19]), where a fluid between two parallel plates is heated from below and cooled from above. In previous studies, the effect of various plate properties on the flow structure and the heat transport were examined, e.g., staggered conducting and insulating strips on the plate [27, 3], temporally-modulated temperature [10, 30], plate with roughness [34, 9, 33], and plates with different wettabilities [12]. Here we pick RB convection as model system to study the effects of attaching bubbles on the global heat or mass transport in buoyancy-driven turbulence.

We employ direct numerical simulations with an advanced finite difference method combined with the phase field method [14]. The bubbles are put at the lower (hot) plate with pinned contact lines so that they cannot move or detach from the plate. To focus on the effects of the Rayleigh number (dimensionless strength of the thermal driving) and the bubble geometry, we disregard mass exchange between the bubble and the liquid, and keep the bubbles at constant volume. The calculations are performed for various Rayleigh numbers and geometries, expressed through the relative area covered by the bubbles, bubble height, bubble number, and type of bubble distribution. To define the thermal BL thickness in the multiphase system, we extend the traditional temperature profile slope method, based on heat flux conservation. With this, we propose an equivalent single-phase system, for which we apply the Grossmann-Lohse (GL) theory [6, 7, 22] to predict the heat transfer and the temperature profile for RB convection with bubbles attached to the hot plate. Without introducing any new free parameter, these predictions well agree with our numerical results. They work for both moderate and large PrPr, which indicates that they can be applied to predict both the mass transport in water electrolysis with bubbles attached to the electrode surface and to catalytic surfaces on which bubbles have formed.

The organization of this paper is as follows: The numerical method and setup are introduced in § 2. The flow features and heat transfer are shown in § 3. We define the thermal BL thicknesses of this new two-phase system in § 4, and propose an equivalent single-phase system in § 5 to calculate the heat transfer. The paper ends with conclusions and an outlook.

Numerical method and setup

The three-dimensional simulations are performed in a cubic domain of dimensions H3H^3. The numerical method (Liu et al. [14]) combines the phase-field method (Jacqmin [8]; Ding et al. [5]; Liu & Ding [13]) and an advanced finite difference direct numerical simulation solver for the Navier-Stokes equations (Verzicco & Orlandi [25]; van der Poel et al. [24]), called AFiD. Numerical details, validation cases, and convergence tests were already presented in our previous study (Liu et al. [14]).

The phase field method is widely used in simulations of multiphase turbulent flows (Soligo et al. [20]), where the liquid-gas interface is represented by contours of the volume fraction CC of the liquid. The corresponding volume fraction of gas is 1−C1-C. The evolution of CC is governed by the Cahn-Hilliard equation,

∂C∂t+∇⋅(uC)=1Pe∇2ψ,(1)\frac{\partial C}{\partial t}+\nabla\cdot(\mathbf{u}C)=\frac{1}{Pe}\nabla^2\psi, \tag*{(1)}

where u\mathbf{u} is the flow velocity, and ψ=C3−1.5C2+0.5C−Cn2∇2C\psi=C^3-1.5C^2+0.5C-Cn^2\nabla^2C the chemical potential. We set the Péclet number Pe=0.9/CnPe=0.9/Cn and the Cahn number Cn=0.75Δx/HCn=0.75\Delta x/H with Δx\Delta x being the mesh size and HH being the height of the RB cell. The parameters PePe and CnCn are taken according to the sharp-interface approach proposed in Ding et al. [5]; Yue et al. [32]; Liu & Ding [13].

The flow is governed by the Navier-Stokes equation, the heat transfer equation, and the incompressibility condition,

ρ~(∂u∂t+u⋅∇u)=−∇P+PrRa∇⋅[μ~(∇u+∇uT)]+Fst+G,(2)\tilde{\rho}\left(\frac{\partial\mathbf{u}}{\partial t}+\mathbf{u}\cdot\nabla\mathbf{u}\right)=-\nabla P+\sqrt{\frac{Pr}{Ra}}\nabla\cdot\left[\tilde{\mu}(\nabla\mathbf{u}+\nabla\mathbf{u}^{T})\right]+\mathbf{F}_{st}+\mathbf{G}, \tag*{(2)}
ρ~c~p(∂θ∂t+u⋅∇θ)=1PrRa∇⋅(k~∇θ),(3)\tilde{\rho}\tilde{c}_{p}\left(\frac{\partial\theta}{\partial t}+\mathbf{u}\cdot\nabla\theta\right)=\sqrt{\frac{1}{PrRa}}\nabla\cdot(\tilde{k}\nabla\theta), \tag*{(3)}
∇⋅u=0,(4)\nabla\cdot\mathbf{u}=0, \tag*{(4)}

where θ\theta is the dimensionless temperature, Fst=62ψ∇C/(Cn We)\mathbf{F}_{st}=6\sqrt{2}\psi\nabla C/(Cn\,We) the non-dimensionalized surface force, where We=ρlU2H/σWe=\rho_lU^2H/\sigma is the Weber number, with the surface tension σ\sigma. The vector G={[C+ΛβΛρ(1−C)]θ−ρ~/Fr}z\mathbf{G}=\{[C+\Lambda_{\beta}\Lambda_{\rho}(1-C)]\theta-\tilde{\rho}/Fr\}\mathbf{z} represents the dimensionless gravity. All dimensionless material properties (indicated by a tilde, q~\tilde{q}) are defined in a uniform way, q~=C+Λq(1−C)\tilde{q}=C+\Lambda_q(1-C), where Λq=qg/ql\Lambda_q=q_g/q_l is the ratio of the material properties of gas and liquid, marked by the subscripts gg and ll, respectively. The global dimensionless parameters controlling the flow are listed in Table 1. The most important response parameter of the system is the heat transfer, which is quantified by the Nusselt number Nu=Q/(klΔ/H)Nu=Q/(k_l\Delta/H), with QQ being the dimensional heat flux.

Material and geometrical parameters(40°C)WaterAir
Densityρ\rhokg/m3kg/m^3992.21.127
Kinematic viscosityν\num2/sm^2/s0.6591×10−60.6591 \times10^{-6}16.92×10−616.92 \times10^{-6}
Thermal conductivitykkW/(m K)W/(m\,K)0.62860.02735
Thermal diffusivityκ=k/(cpρ)\kappa= k/(c_p\rho)m2/sm^2/s1.52×10−71.52 \times10^{-7}241.0×10−7241.0 \times10^{-7}
Specific heat capacitycpc_pJ/(kg K)J/(kg\,K)41801007
Thermal expansion coefficientβ\betaK−1K^{-1}3.84×10−43.84 \times10^{-4}32.1×10−432.1 \times10^{-4}
Surface tensionσ\sigmaN/mN/m0.0696
Gravity accelerationggm/s2m/s^29.8
Temperature differenceΔ\DeltaKK0.1 – 10
Domain heightHHmm0.1 – 10
Dimensionless parameterRealistic valuePresent value
Rayleigh numberRa=βlgH3Δ/(νlκl)Ra = \beta_l g H^3 \Delta/(\nu_l\kappa_l)10610^6 – 101410^{14}10710^7 – 10810^8
Prandtl numberPr=νl/κlPr = \nu_l/\kappa_l4.354.38
Weber numberWe=ρlU2H/σWe = \rho_l U^2 H/\sigma10−110^{-1} – 10510^5100
Froude numberFr=U2/(gH)Fr = U^2/(gH)10−410^{-4} – 10−210^{-2}1
Density ratioΛρ=ρg/ρl\Lambda_\rho= \rho_g/\rho_l0.0010.01
Kinematic viscosity ratioΛν=νg/νl\Lambda_\nu= \nu_g/\nu_l25.725.7
Thermal conductivity ratioΛk=kg/kl\Lambda_k = k_g/k_l0.04350.042
Thermal diffusivity ratioΛκ=κg/κl\Lambda_\kappa= \kappa_g/\kappa_l158158
Thermal expansion coefficient ratioΛβ=βg/βl\Lambda_\beta= \beta_g/\beta_l8.368.36

Table 1. The upper part of the table shows the material and geometrical parameters controlling the flow and their typical values. The lower part of the table shows all resulting dimensionless parameters. Here U=βlgHΔU = \sqrt{\beta_l g H \Delta} is the free-fall velocity. The index gg stands for gas, and ll for liquid.

The values of the control parameters are chosen mainly based on the properties of air and water (see Table 1), though for better numerical efficiency we take the density ratio Λρ=0.01\Lambda_{\rho}=0.01, about 10 times larger than in reality. Note that the exact value of this parameter hardly affects our results. Since the bubbles are pinned and the buoyancy and surface tension forces applied on bubbles are always balanced, we take Fr=1Fr=1 and We=100We=100, also for numerical convenience. The geometrical parameters of the bubbles are the relative covered area 0.18≤S0≤0.620.18\le S_0\le0.62, the non-dimensionalized height 0.02≤h/H≤0.050.02\le h/H\le0.05, their number 40≤n≤14440\le n\le144, and the spatial distribution (uniform, random and half-covered). Note that from S0S_{0}, h~=h/H\tilde{h} = h/H and nn, we can also calculate the bubble volume (=S0h/2+nπh~3/6)(= S_{0}h/2+n\pi\tilde{h}^{3}/6), the bubble contact radius (=S0/(nπ))(= \sqrt{S_{0}/(n\pi)}), the bubble contact angle (=arccos⁡{[S0/(2nπh~)−h~/2]/[S0/(2nπh~)+h~/2]})(= \arccos\{[S_{0}/(2n\pi\tilde{h})-\tilde{h}/2]/[S_{0}/(2n\pi\tilde{h})+\tilde{h}/2]\}). Since the local Weber number of the bubble is relatively low due to the small bubble size, the bubbles maintain their spherical-cap shape although the bubbles could deform and the flows inside and outside the bubbles are both solved. The bubble shape is closer to spherical with larger h~\tilde{h} and smaller S0/nS_{0}/n.

To ensure that the contact lines are pinned on the plate, we set the value of CC on the hot bottom plate equal to the initial value as the boundary condition of the phase field. The other boundary conditions are no-slip velocities on the top and bottom plates, fixed temperature θcold=0\theta_{cold} = 0 (top) and θhot=1\theta_{hot} = 1 (bottom), and periodic conditions in the horizontal directions. Stretched grids with 2883288^{3} gridpoints are used for the velocity and temperature fields, and uniform grids with 5763576^{3} gridpoints for the phase field [14], which corresponds to at least 12 gridpoints used for the bubble height. The mesh is sufficiently fine and is comparable to corresponding single-phase studies [23, 17].

Flow features and heat transfer

A typical thermal structure in RB convection with bubbles attached to the hot plate is shown in Figure 1. Although there is continuous emission of thermal plumes, the bubbles almost maintain the shape of spherical cap due to the pinned contact lines and sufficiently strong surface tension. This is indeed the relevant situation during most of the time for the bubbles in electrolysis or catalysis. In Figure 1(a) and (b), we observe the plumes rising up from the gaps between the bubbles. Inside the bubbles (see the inset in Figure 1(c)), pure thermal conduction takes place since the local Rayleigh number inside the bubble Rag≈βggh3(Δ/2)/(νgκg)Ra_{g} \approx\beta_{g}gh^{3}(\Delta/2)/(\nu_{g}\kappa_{g}) is small enough to remain under the onset of convection Rac≈1708Ra_{c} \approx1708 (namely Rag<100Ra_{g} < 100) for all cases. Furthermore, considering that the thermal conductivity of gas is much lower than that of liquid (i.e. for their ratio $\Lambda_{k} = 0.042 \ll 1), the heat transfer through the gas phase is negligible. Therefore the overall heat-conducting ability of the fluid near the hot bottom plate is lower than that near the cold top plate. Consequently, since the total heat flux is the same across each horizontal plane, the temperature drop across the hot bottom BL becomes larger than that across the cold top BL, leading to the mean centre temperature smaller than 0.5 (e.g. 0.461 in Figure 1c), closer to the temperature of the cold top plate to that of the hot bottom plate.

Image not extracted

Figure 1. Rayleigh-Bénard convection with bubbles on the hot plate for Ra=108Ra = 10^{8}, Pr=4.38Pr = 4.38, S0=0.18S_{0} = 0.18, h~=0.02\tilde{h} = 0.02 and n=144n = 144. (a) Volume rendering of a snapshot of the thermal structures, (b) horizontal slice at the height of the lower (hot) boundary layer thickness, and (c) mean (over time and space) temperature profile with an inset showing thermal structures inside one bubble. The temperature field is color-coded and the interface between liquid and gas is marked in green. As seen from the mean temperature profile in (c), due to the bubbles the mean centre temperature θˉ=0.461\bar{\theta} = 0.461 is lower than the average temperature 1/21/2 of the top and bottom plates (dashed black line).

In Figure 2, we plot the heat transfer NuNu (normalized by NusNu_s of single-phase system) as function of the geometrical parameters. As shown in Figure 2(a), unsurprisingly, Nu/NusNu/Nu_s decreases with increasing the relative bubble-covered area S0S_0 due to the decreasing conducting area. We also observe that Nu/NusNu/Nu_s decreases with increasing bubble height hh (normalized by the thermal BL thickness λ\lambda) in Figure 2(b), since for larger bubbles there is less liquid (the major conducting fluid) in the thermal BL. With increasing h/λh/\lambda, the simulation results gradually deviate from the predictions, since the larger h/λh/\lambda, the more parts of the bubbles enter the bulk, which gradually deviates from our assumption that bubbles only affect the thermal BL structure.

Image not extracted

Figure 2. Nusselt number normalized by that in the single-phase system. (a) Nu/NusNu/Nu_s as function of the relative bubble-covered area S0S_0. Here the (relative) bubble height h~=0.02\tilde{h}=0.02 and the bubble number n=40n=40. (b) Nu/NusNu/Nu_s as function of h/λh/\lambda for S0=0.32S_0=0.32 and n=40n=40, with λ\lambda being the thermal BL thickness (values taken from the hot bottom BL). (c) Nu/NusNu/Nu_s as function of nn for S0=0.18S_0=0.18 and h~=0.02\tilde{h}=0.02. Symbols denote the numerical results and lines the predictions of eqs. (5) and (6)–(8), where colors are for Ra=107Ra=10^7 (red), Ra=3.2×107Ra=3.2\times10^7 (blue), and Ra=108Ra=10^8 (black). The orange symbol + and the green symbol × in (c) denote the cases with random bubbles distribution as shown in (d) and the half-covered distribution as shown in (e). All the other symbols in (a)–(c) are for the uniform distribution. All error bars and deviations between simulations and predictions are within 5%.

In contrast to the significant effects of S0S_0 and h~\tilde{h} on the heat transfer, Nu/NusNu/Nu_s is insensitive to the bubble number nn (see Figure 2c), since nn only contributes to the second term in the bubble volume (=S0h~/2+nπh~3/2)(=S_0\tilde{h}/2+n\pi\tilde{h}^3/2), which is of order O(h~3)O(\tilde{h}^3) and thus negligible compared to the first term (O(h~))(O(\tilde{h})). We further show the effects of the spatial bubble distribution, including uniform, random, and the half bubble-covered distributions in Figure 2(c). With all three types of distribution, the values of Nu/NusNu/Nu_s are almost the same. This indicates that the overall heat flux is insensitive to the spatial bubble distribution, at least in our model system. This differs from the previous studies [11, 9], where the large-scale flow is changed by the solid elements on the plate. The rigid solid surfaces imply no-slip velocity boundary conditions, which affect the flow structure much more than the boundary conditions of continuous velocity and shear stress on the deformable bubble interfaces.

Analogous trends are also found in the relationship between the mean centre temperature θˉ\bar{\theta} and the geometrical parameters, as shown in Figure 3. In all cases, the temperature profile is asymmetric due to the bubbles, namely θˉ\bar{\theta} is lower than the average temperature 1/21/2 between top and bottom plates, as explained above. The temperature profile can be quantitatively described by θˉ\bar{\theta} and the top and bottom thermal BL thicknesses, the values of which are calculated in § 5.

Image not extracted

Figure 3. (a) Mean centre temperature θˉ\bar{\theta} as function of S0S_0. (b) θˉ\bar{\theta} as function of h/λh/\lambda. (c) θˉ\bar{\theta} as function of nn. All the cases are the same as in Figure 2. The dashed line denote the average temperature 1/21/2 of the top and bottom plates. All error bars and deviations between simulations and predictions are within 5%.

Such an asymmetric temperature profile has already been studied in the context of single-phase RB convection under non-Oberbeck-Boussinesq (NOB) conditions [1], where in water the viscosity and thermal diffusivity are temperature dependent and smaller at the hot bottom plate than at the cold top plate. Also this leads to an asymmetric temperature profile. Ahlers et al. [1] employed an extended Prandtl-Blasius BL theory for the NOB conditions in the two BLs, and coupled them by imposing heat flux conservation. With this they could calculate the centre temperature and the thermal BL thicknesses, which well agree with the experimental results. Note that that analytical calculation must be adapted to be applicable to the situation here, since here the BL flow is not parallel to the plates due to bubbles. In addition, whereas in the NOB case one has no-slip and no-penetration boundary conditions throughout, here on the spherical-cap shaped bubble interface we have the emerging condition of the velocity and shear stresses continuity. However, as in Ahlers et al. [1], we use the concept of heat flux conservation to define the thermal BL thickness (see § 4), and then propose the idea of an equivalent single-phase system to mimic the system with attached bubbles (see § 5).

Thermal BL thicknesses

A sketch of the temperature profiles near the plates is displayed in Figure 4, where we also show how the thermal BL thicknesses are defined. Near the cold top plate (see Figure 4a), the thermal BL thickness λcold\lambda_{\mathrm{cold}} is defined through the usual convenient definition via the slope of the temperature profile at the plate from assuming a pure conductive thermal BL and a well-mixed bulk [1]. As λcold\lambda_{\mathrm{cold}} we take that distance from the plate, where the tangent to the temperature profile at the plate reaches the mean centre temperature θˉ\bar{\theta}. However, this definition cannot be directly applied near the hot bottom plate, due to the bubbles. As the bubbles reduce the area occupied by the liquid, the conducting area varies with zz. We therefore correct the solution for pure thermal conduction in the hot bottom BL based on heat flux QQ conservation,

Image not extracted

Figure 4. Sketch of the temperature profile near the (a) cold top and (b) hot bottom plate. The red lines denote the temperature profiles, the black lines the mean centre temperature θˉ\bar{\theta}, the blue lines the solution for pure thermal conduction in the BL, i.e. eq. (5), and the green dashed line the tangent of eq. (5) at the intersection. θ∗\theta^* is the intercept of the tangent. We define the dimensional BL thickness near the hot and cold plates as λhot\lambda_{\mathrm{hot}} and λcold\lambda_{\mathrm{cold}}, respectively, i.e., the distance from the black dashed lines to the corresponding plate.

Q=kl∂θ∂z[1−S(z,S0,h~,n)]=klθˉHλcold.(5)Q = k_l \frac{\partial\theta}{\partial z}[1-S(z,S_0,\tilde{h},n)] = k_l \frac{\bar{\theta}H}{\lambda_{cold}}. \tag*{(5)}

Here, the term on the right is the heat flux on the cold top plate. The gas-covered area S(z,S0,h~,n)=max⁡[(S0+nπzh~)(1−z/h~),0]S(z,S_0,\tilde{h},n) = \max[(S_0+n\pi z\tilde{h})(1-z/\tilde{h}),0] is the insulating area, which is defined with the assumption of spherical-cap shaped bubbles.

The thermal BL thickness near the hot bottom plate λhot\lambda_{hot} equals the distance from the plate to the intersection between the lines of (5) and of θ=θˉ\theta=\bar{\theta}, as shown in Figure 4(b). With the definitions above, both values of λhot\lambda_{hot} and λcold\lambda_{cold} are close to each other in all cases (within 10%), as shown in Figure 5. Since λhot\lambda_{hot} and λcold\lambda_{cold} are almost the same and θˉ<(θhot+θcold)/2\bar{\theta}<(\theta_{hot}+\theta_{cold})/2, the temperature variations across the hot bottom BL and the cold top BL are different, which is reflected in that the temperature profile is asymmetric.

Image not extracted

Figure 5. (a) Thermal BL thicknesses λhot\lambda_{hot} (empty symbols) and λcold\lambda_{cold} (filled symbols) of the hot bottom and cold top BLs, respectively, as functions (a) of S0S_0, (b) of h/λh/\lambda, and (c) of nn. All the cases are the same as in Figure 2. All error bars and deviations between simulations and predictions are within 5%.

Predictions for Nu and the centre temperature using equivalent single-phase RB system

To calculate the heat transfer NuNu and the mean centre temperature θˉ\bar{\theta}, we propose the idea of using an equivalent single-phase setup to mimic the system with attached bubbles. To find the equivalent flow, we first obtain the effective temperature at the hot bottom plate. This is done by plotting the tangent of (5) at the position of λhot\lambda_{hot}, and the obtaining θ\theta-intercept of the tangent as effective temperature θ∗\theta^* (see Figure 4b).

Next, we compare the effective temperature θ∗\theta^* and the mean centre temperature θ‾\overline{\theta} in Figure 6a, which shows that θ∗\theta^* approximately equals to 2θ‾2\overline{\theta} (within ±5%\pm5\%). Thus we assume a nearly equivalent single-phase RB counterpart with θhot=2θ‾\theta_{hot} = 2\overline{\theta} and θcold=0\theta_{cold} = 0, such that θ‾\overline{\theta} is the mean value of temperatures on the two plates, as shown in Figure 6b. Again, we note that λhot\lambda_{hot} is close to λcold\lambda_{cold}, based on our definition of the thermal BL thickness in § 3. Thus, we have the following relations:

Image not extracted

Figure 6. (a) Comparison between the mean centre temperature θˉsim\bar{\theta}_{sim} and the effective temperature θsim∗\theta^*_{sim}, both taken from simulations. The symbols are the same as in Figure 2. The shadow is the zone within ±5%\pm5\%. (b) Sketch of the bubble-system under consideration and the equivalent single-phase RB system. θˉsim\bar{\theta}_{sim} and θsim∗\theta^*_{sim} are plotted this way to support the existence of the equivalent single-phase system.

θ∗=2θ‾,λcold=λhot=λs.(6)\theta^* = 2\overline{\theta}, \quad\lambda_{cold} = \lambda_{hot} = \lambda_s. \tag*{(6)}

Here λs\lambda_s is the thermal BL thickness in the single-phase system, which can be calculated as

λs=θ‾HNue,(7)\lambda_s = \frac{\overline{\theta}H}{Nu_e}, \tag*{(7)}

where NueNu_e is the heat transfer estimated from the GL theory ([6, 7]) for the equivalent single-phase system with Ra∗=2θ‾RaRa^* = 2\overline{\theta}Ra, θhot=2θ‾\theta_{hot} = 2\overline{\theta} and θcold=0\theta_{cold} = 0. In the two-phase system and the equivalent single-phase system, the dimensional heat transfer QQ (=Nu k(θhot−θcold)/H= Nu\,k(\theta_{hot} - \theta_{cold})/H) should be the same, which yields

Nu(Ra,Pr,S0,h~,n)=2θ‾Nue(Ra∗,Pr),(8)Nu(Ra, Pr, S_0, \tilde{h}, n) = 2\overline{\theta}Nu_e(Ra^*, Pr), \tag*{(8)}

where Nu(Ra,Pr,S0,h~,n)=(θ‾H)/λcoldNu(Ra, Pr, S_0, \tilde{h}, n) = (\overline{\theta}H)/\lambda_{cold} is the heat transfer for the two-phase system. Combining the relations (6)–(8) for the equivalent single-phase RB system with the relationship (5) between θ‾\overline{\theta} and λcold\lambda_{cold} in § 4, we can now calculate the heat transfer in the system with attached bubbles.

We emphasize that with this approach, using the equations above without introducing any free parameter, for given RaRa, PrPr and bubble geometries, we can now calculate NuNu, θ‾\overline{\theta}, λhot\lambda_{hot} and λcold\lambda_{cold}. The good agreements between the simulations and the predictions for θ‾\overline{\theta}, λhot\lambda_{hot}, and λcold\lambda_{cold} are shown in Figure 2, Figure 3, and Figure 5, respectively, where all deviations between simulations and predictions are within ±5%\pm5\%.

We further check whether our approach is also applicable for large PrPr, which, as explained above, has relevance to transport phenomena in water electrolysis and catalysis. Water electrolysis and catalysis can both lead to natural convection driven by buoyancy, which originates from the density difference of the solute with different concentrations of the electrolysis or catalysis product. Here, the mass transfer is also characterized by NuNu, i.e. the mass transfer normalized by that with pure diffusion (in this context normally called Sherwood number ShSh). The control parameters are the Grashof number (dimensionless strength of the solute driving) GrGr and the Schmidt number (the ratio of viscous diffusion and mass diffusion rates) ScSc, corresponding to Ra/PrRa/Pr and PrPr in RB convection, respectively. The value of ScSc in water electrolysis is always large, e.g., Sc=400Sc = 400 [18].

We performed two simulations (for two different bubble heights) at Pr=400Pr = 400 and Ra=107Ra = 10^{7} (tabulated in Table 2) with a sufficiently fine mesh as explained in § 2. Note the much higher computational costs at this large PrPr, due to the required long time (∼Pr1/2\sim Pr^{1/2}) for the system to enter the statistical steady state. The bubble geometries are characterized by h~=0.03\tilde{h} = 0.03 and 0.050.05, S0=0.32S_{0} = 0.32, and n=40n = 40. The good agreements between our parameter free predictions and the results from the simulations are shown in Table 2. This supports that our predictions can be directly applied to water electrolysis and catalysis.

h~\tilde{h}Nu/NusNu/Nu_sθ‾\overline{\theta}λbot/H\lambda_{bot}/Hλtop/H\lambda_{top}/H
Simulations0.030.900±0.0200.900 \pm0.0200.462±0.0050.462 \pm0.0050.0303±0.00230.0303 \pm0.00230.0294±0.00080.0294 \pm0.0008
Predictions0.030.8800.4520.03400.0340
Simulations0.050.821±0.0220.821 \pm0.0220.438±0.0030.438 \pm0.0030.0315±0.01800.0315 \pm0.01800.0299±0.00060.0299 \pm0.0006
Predictions0.050.8010.4200.03470.0347

Table 2. Comparisons between simulations and predictions at large Pr=400Pr = 400. The other parameters are Ra=107Ra = 10^7, S0=0.32S_0 = 0.32, and n=40n = 40.

Conclusions and outlook

Turbulent RB convection with gas bubbles attached to the hot plate is numerically investigated for 107≤Ra≤10810^{7} \le Ra \le10^{8} and Pr=4.38Pr = 4.38 and 400400. The bubble geometrical parameters are the relative bubble-covered area S0S_{0}, the relative bubble height h~\tilde{h}, the bubble number nn, and the spatial bubble distribution. Due to the much lower thermal conductivity of gas as compared to liquid, the temperature profile is asymmetric and the heat transfer efficiency of the system is reduced. More specifically, NuNu significantly decreases with increasing S0S_{0} and h~\tilde{h}, but is almost unaffected by nn and the types of bubble distribution.

To predict the heat transfer and the mean centre temperature of the system, we have proposed the idea of using an equivalent single-phase system to mimic the system with attached bubbles. By applying the GL theory for the equivalent system and imposing heat flux conservation in the two thermal BLs, we can predict the heat transfer, the top and bottom thermal BL thicknesses, and the mean centre temperature, without introducing any free parameter. The predictions well agree with the results from the simulations. Briefly, in § 4 we got one relationship between θˉ\bar{\theta} and λ\lambda in eq. (5), and in § 5 we got another relationship between θˉ\bar{\theta} and λ\lambda in eqs. (6) and (8). Then, for only given RaRa, PrPr and bubble geometries, we can well predict the heat transfer and temperature profile in the system with bubbles attached to the bottom plate.

The results of this study can be used not only for the heat transfer in RB convection with bubbles attached to the plates, but also for the mass transfer in electrolysis or catalysis. Our predictions can help to obtain estimates for relevant applications and e.g. optimize the heat or mass transfer and flow features in systems in which bubbles are forming on the plate(s). It would also be interesting to extend our basic idea to other wall-bounded turbulent systems with various plate properties, such as plates with inhomogeneous properties (e.g. wettability or conducting ability).

Acknowledgments

We acknowledge PRACE for awarding us access to MareNostrum in Spain at the Barcelona Computing Center (BSC) under the project 2021250115 and the Netherlands Center for Multiscale Catalytic Energy Conversion (MCEC). K. L. Chong acknowledges Shanghai Science and Technology Program under project no. 19JC1412802.

Declaration of interests

The authors report no conflict of interest.

References

  1. [1]AHLERS, G., BROWN, E., FONTENELE ARAUJO, F., FUNFSCHILLING, D., GROSSMANN, S. & LOHSE, D. 2006 Non-Oberbeck-Boussinesq effects in strongly turbulent Rayleigh-Bénard convection. J. Fluid Mech. 569, 409–445.
  2. [2]AHLERS, G., GROSSMANN, S. & LOHSE, D. 2009 Heat transfer and large scale dynamics in turbulent Rayleigh-Bénard convection. Rev. Mod. Phys. 81, 503.DOI
  3. [3]BAKHUIS, D., OSTILLA-MÓNICO, R., VAN DER POEL, E. P., VERZICCO, R. & LOHSE, D. 2018 Mixed insulating and conducting thermal boundary conditions in Rayleigh–Bénard convection. J. Fluid Mech. 835, 491–511.DOI
  4. [4]CHILLA, F. & SCHUMACHER, J. 2012 New perspectives in turbulent Rayleigh-Bénard convection. Eur. Phys. J. E 35, 58.DOI
  5. [5]DING, H., SPELT, P. D. M. & SHU, C. 2007 Diffuse interface model for incompressible two-phase flows with large density ratios. J. Comput. Phys. 226, 2078–2095.DOI
  6. [6]GROSSMANN, S. & LOHSE, D. 2000 Scaling in thermal convection: A unifying view. J. Fluid Mech. 407, 27–56.
  7. [7]GROSSMANN, S. & LOHSE, D. 2001 Thermal convection for large Prandtl number. Phys. Rev. Lett. 86, 3316–3319.DOI
  8. [8]JACQMIN, D. 1999 Calculation of two-phase Navier–Stokes flows using Phase-Field modeling. J. Comput. Phys. 155, 96–127.DOI
  9. [9]JIANG, H., ZHU, X., MATHAI, V., VERZICCO, R., LOHSE, D. & SUN, C. 2018 Controlling heat transport and flow structures in thermal turbulence using ratchet surfaces. Phys. Rev. Lett. 120, 044501.
  10. [10]JIN, X.-L. & XIA, K.-Q. 2008 An experimental study of kicked thermal turbulence. J. Fluid Mech. 606, 133–151.
  11. [11]LIU, B. & ZHANG, J. 2008 Self-induced cyclic reorganization of free bodies through thermal convection. Phys. Rev. Lett. 100, 244501.pubmed.ncbi.nlm.nih.gov/18643586
  12. [12]LIU, H.-R., CHONG, K. L., NG, C. S., VERZICCO, R. & LOHSE, D. 2022 Enhancing heat transport in multiphase Rayleigh–Bénard turbulence by changing the plate–liquid contact angles. J. Fluid Mech. 933, R1.DOI
  13. [13]LIU, H.-R. & DING, H. 2015 A diffuse-interface immersed-boundary method for two-dimensional simulation of flows with moving contact lines on curved substrates. J. Comput. Phys. 294, 484–502.DOI
  14. [14]LIU, H.-R., NG, C. S., CHONG, K. L., VERZICCO, R. & LOHSE, D. 2021 An efficient phase-field method for turbulent multiphase flows. J. Comput. Phys. 446, 110659.DOI
  15. [15]LOHSE, D. & XIA, K.-Q. 2010 Small-scale properties of turbulent Rayleigh-Bénard convection. Annu. Rev. Fluid Mech. 42, 335–364.DOI
  16. [16]OEHMICHEN, T., DATSEVICH, L. & JESS, A. 2010 Influence of bubble evolution on the effective kinetics of heterogeneously catalysed gas/liquid reactions, part I: reactions with gaseous products. Chem. Eng. Technol. 33, 911–920.DOI
  17. [17]VAN DER POEL, E. P., STEVENS, R. J. A. M. & LOHSE, D. 2013 Comparison between two- and three-dimensional Rayleigh-Bénard convection. J. Fluid Mech. 736, 177.
  18. [18]SEPAHI, F., PANDE, N., CHONG, K. L., MUL, G., VERZICCO, R., LOHSE, D., MEI, B. T. & KRUG, D. 2022 The effect of buoyancy driven convection on the growth and dissolution of bubbles on electrodes. Electrochim. Acta 403, 139616.DOI
  19. [19]SHISHKINA, O. 2021 Rayleigh-Bénard convection: The container shape matters. Phys. Rev. Fluids 6, 090502.DOI
  20. [20]SOLIGO, G., ROCCON, A. & SOLDATI, A. 2021 Turbulent flows with drops and bubbles: What numerical simulations can tell us — Freeman Scholar Lecture. J. Fluids Eng. 143, 080801.DOI
  21. [21]SOMORJAI, G. A. & LI, Y. 2010 Introduction to surface chemistry and catalysis. John Wiley and Sons, Inc., Hoboken, New Jersey, second edition.
  22. [22]STEVENS, R. J. A. M., VAN DER POEL, E. P., GROSSMANN, S. & LOHSE, D. 2013 The unifying theory of scaling in thermal convection: the updated prefactors. J. Fluid Mech. 730, 295–308.
  23. [23]STEVENS, R. J. A. M., VERZICCO, R. & LOHSE, D. 2010 Radial boundary layer structure and Nusselt number in Rayleigh-Bénard convection. J. Fluid Mech. 643, 495–507.
  24. [24]VAN DER POEL, E. P., OSTILLA-MÓNICO, R., DONNERS, J. & VERZICCO, R. 2015 A pencil distributed finite difference code for strongly turbulent wall-bounded flows. Comput. Fluids 116, 10–16.arxiv.org/abs/1501.01247
  25. [25]VERZICCO, R. & ORLANDI, P. 1996 A finite-difference scheme for three-dimensional incompressible flow in cylindrical coordinates. J. Comput. Phys. 123, 402–413.DOI
  26. [26]VOGT, H. & BALZER, R. J. 2005 The bubble coverage of gas-evolving electrodes in stagnant electrolytes. Electrochim. Acta 50, 2073–2079.DOI
  27. [27]WANG, F., HUANG, S.-D. & XIA, K.-Q. 2017 Thermal convection with mixed thermal boundary conditions: effects of insulating lids at the top. J. Fluid Mech. 817, R1.DOI
  28. [28]WANG, M., WANG, Z., GONG, X. & GUO, Z. 2014 The intensification technologies to water electrolysis for hydrogen production – A review. Renew. Sust. Energy Rev. 29, 573–588.DOI
  29. [29]XU, W., LU, Z., SUN, X., JIANG, L. & DUNN, X. 2018 Superwetting electrodes for gas-involving electrocatalysis. Acc. Chem. Res. 51, 1590–1598.DOI
  30. [30]YANG, R., CHONG, K. L., WANG, Q., VERZICCO, R., SHISHKINA, O. & LOHSE, D. 2020 Periodically modulated thermal convection. Phys. Rev. Lett. 125, 154502.DOI
  31. [31]YANG, X., BACZYZMAILSKI, D., CIERPKA, C., MUTSCHKE, G. & ECKERT, K. 2018 Marangoni convection at electrogenerated hydrogen bubbles. Phys. Chem. Chem. Phys. 20, 11542.DOI
  32. [32]YUE, P., ZHOU, C. & FENG, J. J. 2010 Sharp-interface limit of the Cahn–Hilliard model for moving contact lines. J. Fluid Mech. 645, 279–294.DOI
  33. [33]ZHU, X., STEVENS, R.J.A.M., SHISHKINA, O., VERZICCO, R. & LOHNSE, D. 2019 Nu ∼ Ra^{1/2} scaling enabled by multiscale wall roughness in Rayleigh–Bénard turbulence. J. Fluid Mech. 869, R4.
  34. [34]ZHU, X., STEVENS, R.J.A.M., VERZICCO, R. & LOHNSE, D. 2017 Roughness-facilitated local 1/2 scaling does not imply the onset of the ultimate regime of thermal convection. Phys. Rev. Lett. 119, 154501.DOI

Paper details

vlnc:W_7p94yjk6
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