Strong coupling between height of gaps and thickness of thermal boundary layer in partitioned convection system
Lin Ze-Peng, Bao Yun
Department of Mechanics, Sun Yat-Sen University, Guangzhou 510275, China

 

† Corresponding author. E-mail: stsby@mail.sysu.edu.cn

Abstract

A direct numerical simulation (DNS) method is used to calculate the partitioned convection system with Ra number ranging from 107 to 2 ×109. Using the boundary layer thickness to normalize the height of gaps d, we find a strong consistency between the variation of the TD number (the average value of the temperature in each heat transfer channel is averaged after taking the absolute values) with the change of the height of gaps and the variation of the TD number with the change of Ra number in partitioned convection. For a given thickness of partition, heights of gaps are approximately equal to 0.5 or 1 time of the thermal boundary layer thickness at different Ra numbers. TD number representing temperature characteristics is almost the constant value, which means that TD number is a function of only. Analysis of local temperature field of area in gaps shows that the temperature distribution in the gaps are basically the same when is certain. The heat transfer Nu number of the system at is larger than that of , both of them have the same scaling law with Ra number and .

PACS: 47.27.te
1. Introduction

Thermal convection is widespread in nature and daily life, which has drawn great focus among researchers.[1,2] Rayleigh-Bénard (RB) convection is a classical physical model in the field of natural convection research,[3] and heat transfer enhancement is one of the hot spots in such field.[4] In recent years, a rising amount of studies have focused on the non-standard RB convection. Compared with the standard RB convection system, it was found by experiments and numerical simulations that some non-standard RB thermal convection systems can enhance heat transfer efficiency, such as RB convection with rough plates,[57] corilois force,[810] with phase transition,[11] with polymer addition,[12,13] partitions.[1418] There are both passive and active ways to improve the heat transfer efficiency. Due to the fact that the passive method does not introduce external powers, it has been widely concerned and deeply studied for a long time. However, the heat transfer enhancement is generally less than one time.[1921]

As a passive method, the partitioned convection system has achieved an extraordinary heat transfer enhancement. Bao et al.[14] added vertical partitions at equal distances in the RB convection devices with aspect ratio of 5, and left gaps at both ends of the partitions away from the top and bottom plates. It was found that the heat transfer efficiency increased significantly while the number of partitions increases. Through numerical simulation, Bao et al.[15] found that when the number of partitions increased to 28, adjusting the height of gaps can further improve the heat transfer efficiency, which was excited up to 3.1 times that of the non-partition device. Lin et al.[16] further studied the influence of different geometrical parameters on the flow and heat transfer characteristics of the system, and found that the height of gaps is the key geometric parameter in the partitioned convection system. Lin and Bao[17] focused on the TD number, representing the strength of heat separation in this system, and found that TD reflects the magnitude of buoyancy, which is balanced by the pressure difference between the top and bottom of the heat transfer channels. In addition, there is a scale relation with exponent −1.64 between TD and the height of gaps in a certain range,[17] and a scale relation with exponent 0.4 between TD and Ra.[18] Therefore, in this paper, size of TD number will be concerned when the temperature characteristics of systems are studied.

In the partitioned convection system, the height of gaps is usually chosen on the same order of magnitude as the thickness of thermal boundary layer. The change of the thickness of the thermal boundary layer of the RB convection system under different Ra numbers is based on a certain law. In this paper, the height of gaps of the partitioned convection system is related to the thickness of the thermal boundary layer of RB convection system, and the temperature characteristics and heat transfer characteristics of the partitioned convection system are discussed when the ratio of height of gaps to thickness of boundary layer takes some specific values.

2. Numerical methods

RB convection is an enclosed system with fluid heated from the bottom and cooled from the top. The convective cavity has height H, width D, and thickness W. We consider an RB convection system with vertical partitions equally distributed in the cell. As illustrated in Fig. 1(a), the partitions are connected with the front and back sidewalls. Both ends are kept away from the cooling and heating plates in a distance d. Figure 1(b) is a zoom-in view of the convective cell. Here l represents the thickness of the partitions and b the width of the vertical channels.

Fig. 1. (a) The model diagram of partition convection device. (b) An enlarged portion of the partition convection device.

Previous studies have shown that when the number of partitions is large enough, the flows in the channels will become laminar. In such case, the heat transfer efficiency will be the same for both two- and three-dimensional partitioned RBC systems.[15] Since the focus of this paper is on the laminar regime, we use two-dimensional direct numerical simulation (DNS) for all cases in this paper, which is good enough to reveal the physics of this system. Under the Boussinesq approximation, the non-dimensional governing equations of two-dimensional RB thermal convection is given by

where is velocity, P pressure, θ temperature, and the unit vector of vertical direction. Rayleigh number and Prandtl number are the control parameters of this system. All boundaries are non-slip and impermeable. The top and bottom plates are isothermal with fixed temperature θ=0.5 and θ=−0.5 respectively, while the sidewalls are adiabatic. Staggered grids are used in our simulations. The temporal and spatial discretization schemes have first- and second-order precision respectively. In this paper, the horizontal grids are equidistant, and the vertical grids are refined near boundaries and gaps. The number of grids in the gaps is more than 10 to ensure that the horizontal flows in the gaps are resolved. We choose Pr=5.3, , and the width of partition l=0.12. For multi-partition systems, when the number of partitions is large, the influence of the aspect ratio of the convective cell on the flow and heat transfer efficiency is negligible.[16] In order to reduce the CPU time, the aspect ratio is selected as 2 in this paper. For systems with higher Ra number, the more meshes are generally required, so the grid size of the sample with is tested. By calculating the systems with grid scales of 1000×450 and 2000×900 respectively, it is found that their Nu numbers differ by less than one percent. Therefore, all the samples in this paper are uniformly calculated using the grid scale of 1000 × 450.

3. Systems with series of d and Ra

In the laminar partitioned RB system, the fluid in each channel moves vertically in alternate directions. Due to the continuity, horizontal flow is formed in the gaps, shearing the thermal boundary layer and transporting the hot/cold fluid to the upward/downward channels. Such process makes the upward/downward channels have higher/lower temperature than the average of the cell, as shown in Fig. 2. This separation between hot upward flow and the cold downward one enhances the global heat transfer effectively. Thus, this “heat-separation” phenomenon is essential to this system. In this section we will focus on the influence of gaps’ height d and Ra number over this phenomenon.

Fig. 2. Temperature field and streamline of partitioned convection system.

We average the temperature in the heat transfer channel in the horizontal direction and then integrate it in the longitudinal direction to obtain the TD number[15] in each heat transfer channels, which represents how strong the heat is separated. The absolute values of TD number in the channels are averaged to get the total TD number of the system, which is able to reflect the magnitude of temperature drift in the heat transfer channels of the system. The formula for calculating the TD number is expressed as follows:

where means that the physical quantity in the i-th heat transfer channel is averaged laterally, and n is the number of partitions. Through a series of calculations, the relationships between TD number and height of gaps and Ra number are studied, respectively.

Figure 3(a) shows the TD number as a function d at Ra = 108 with channel width b = 0.092, 0.116, and 0.145. Here d ranges from 0.01 to 0.025. We observed a scaling law . The dependence of TD on Ra is shown in Fig. 33(b) with d=0.02 and b=0.074. There are two states in the TD–Ra relation. When Ra is small, TD basically maintains between 0.3 and 0.5, increases gradually and then begins to decline. As Ra increases, it enters the second state, in which TD decreases while Ra increases, and there is a scaling relation between TD and Ra ( ). Thus, the TD number varies with d and Ra number, and there are scaling laws with a negative exponent in a certain range of d and Ra.

Fig. 3. (a) TD number under various geometric parameters of partitions. (b) TD number with the change of Ra, and the limit value of temperature is 0.5.

As mentioned above, the height of gaps is on the same order of magnitude as the thickness of thermal boundary layer. Therefore, the relationship between the height of gaps and the thickness of thermal boundary layer is worth discussing. Zhou et al.[22] gave the definition of thickness of thermal boundary layer in turbulent thermal convection. Thickness of thermal boundary layer is defined as the height where the linear fitting line of the temperature near the bottom intersects the horizontally averaged temperature of the bulk region. Two-dimensional RBC system with aspect ratio of 2 has been calculated and Ra number ranges from 105 to 1010. Our calculation results show that thickness of thermal boundary layer decreases as Ra increases under a scaling law .

We next normalize d in Fig. 3 using the thickness of the thermal boundary layer , to reveal the importance of the gap height in the heat separation. We plot the TD number in Fig. 3(a), in which Ra is fixed, together with those in Fig. 3(b), in which d is fixed, as a function of . It is a surprise to observe that these two sets of results collapse well, as shown in Fig. 4. These results indicate that within a certain parameter range, TD number in the heat transfer channel depends only on the ratio between the gap height d and thermal boundary layer thickness .

Fig. 4. TD number under different heights of gaps d which are indicated by black solid rectangle and different Ra number which are indicated by red solid circles with the change of .

Therefore, it is necessary to further explore the effect of on the temperature and heat transfer characteristics of the partitioned RBC.

4. Effect of on physical properties of partitioned convective system

We then fixed the thickness of partitions l=0.12, numbers of partitions n=8, the width of channels b=0.116, and Ra changes from 107 to 2 ×109. The thickness of thermal boundary layer varies with Ra. By changing the height of gaps d, we focus on those cases with and , and study their temperature distribution and heat transfer.

4.1. Influence of on TD

First, we focus on the change of TD with Ra number under different .

Figure 5 shows TD as a function of Ra with fixed . It can be seen that whether or . In such cases, TD seems to be independent of Ra. TD equals 0.37 when and 0.13 when . As a quantity reflecting the global temperature distribution property of the system, TD is independent of the aspect ratio of the system and the width of the heat transfer channel. We find that for a given thickness of partitions, TD number depends only on the ratio , i.e., , regardless of Ra. This means that the magnitude of determines the strength of heat separation in this system.

Fig. 5. TD number with the change of Ra number when or 1. The 1d results are indicated by the red solid circles and the 0.5d results are indicated by blue solid triangles.
4.2. Temperature distribution near and within the gaps

The gaps and their adjacent area of the partitioned convection system are very close to the thermal boundary layer. Due to the injection of partitions, the thermal boundary layer is suppressed, which is related to the global heat transfer enhancement. In this section we focus on the temperature fields in the gaps and the vicinity at Ra equals 107, 108, and 109 and or . Figure 6 shows the temperature field in the area of and near the lower gap. Figures 6(a), 6(c), and 6(e) correspond to Ra= 107, 108, and 109 at , and figures 6(b), 6(d), and 6(f) correspond to Ra= 107, 108, and 109 at , respectively. The flows in the gap are all from the cold channels to the hot ones. It is evident in the figure that although the thickness of boundary layer decreases while Ra increases, cases with the same are with similar temperature distribution.

Fig. 6. Temperature field of the gap’s area below the fourth partition of the system and the bottom area of the channels adjacent to it. Panels (a), (c), and (e) correspond to Ra= 107, 108, and 109 when . Panels (b), (d), and (f) correspond to Ra= 107, 108, and 109 when .

From the temperature distribution shown in Figs. 6(a), 6(c), and 6(e), one can see that when the cold fluid passes through the lower gap from left to right, temperature gradually increases due to heating of the bottom plate, and the direction of the gap flow is horizontal. The finest thermal boundary layer is observed at the entrance of the gaps, where the vertical temperature gradient is the largest. The temperature at the outlet is between 0.3 and 0.4. The temperature of fluid at the bottom of the right channel is also between the two isotherms whose values are equal to 0.3 and 0.4. Since is fixed, the height of the gap is different between Ra numbers. However, their temperatures of the fluid entering the right channels are close, and the heating processes of the fluid in the gaps by the bottom plates are also similar. In Figs. 6(b)6(d), the flows are similar to the former series, except that the fluid has not been heated completely when they flow through the gaps, which may be caused by larger heights of the gaps. Also, there is still cold fluid in the upper part of the gaps. Thus, the temperature of the fluids exiting the gaps ranges from 0.1 to 0.2, which is lower than that of the cases.

To show the similarity of these cases, we again normalized the heights of the gaps by the thickness of the boundary layer, and the temperature distributions under different Ra in the gaps are discussed in details. Three temperature contours θ=0.1, 0.2, and 0.3 with are extracted from the results shown in Figs. 6(a), 6(c), and 6(e), and presented in Fig. 7(a). The gray rectangular in the figure is the region where the partition is located. Figure 7(b) is the result of , in which three temperature contours of −0.1, 0, and 0.1 are selected.

Fig. 7. When , three isotherms of 0.1, 0.2, and 0.3 are selected (a) and , three isotherms of −0.1, 0, and 0.1 are selected (b) at Ra=107 (dots), Ra=108 (dash Dot), and Ra=109 (solid). The gray area in the figure is the area where the partition is located.

Figure 7 convinces that the temperature distribution in the gap is similar when is the same, also the fluid is under similar heating process when passing through the gap. As shown in Fig. 7(a), when , the isotherms are divided into two regions. Before entering the gaps, the isotherms basically horizontal. Upon approaching the gap, the heights of isotherms gradually decrease, reaching the lowest position at the entrance. After entering the gap, three isotherms gradually rise and end at the bottom of the partition, and the thickness of the thermal boundary layer also increases. In Fig. 7(b), when , the isotherms are divided into three regions. Before entering the gap, the isotherms of different Ra vary greatly, but they all reach the lowest point at the entrance of the gap. After entering the gap, the isotherms of the same temperature value for different Ra numbers almost collapse, and the height rises gradually. The isotherms of −0.1 end up at the bottom of the partition, and the isotherms of 0.0 and 0.1 can extent out of the gap.

In a word, the thickness of the thermal boundary layer of different Ra numbers varies greatly, but as long as are the same, the distributions of temperature in the gaps are basically the same.

4.3. Effect of on Nu number

As shown in the previous section, in the partitioned convection, we find that for the same , the same TD will be produced. In the last part of this section, we will again focus on the effect on heat transfer efficiency, which is the widespread interest in this field.

Figure 8 shows change of Nu with Ra for the two series of simulations, together with Nu in the non-partitioned convection system as a comparison. As shown in Fig. 8, for and , Nu is much higher than the case without partitions, which is an example of the heat transfer enhancement in the partitioned convection system. Within the given range of Ra, the maximum heat transfer efficiency enhancement happens at Ra= 107, whose Nu is 2.2 times of the non-partitioned one for , and 2.6 times for . With the increase of Ra, the enhancement of heat transfer efficiency decreases gradually. We also observed a scaling relation between Nu and Ra for both and 1, which yields . Even though, the enhancement effect in the partitioned convection is still huge. The smallest heat transfer efficiency at for is nearly twice as large as that without partitions.

Fig. 8. Semi-log plot of Nu/Ra 0.29 as functions of Ra. 0.5d ( ) and 1d ( ) are indicated by blue solid triangle and red solid circles, respectively. nogb (Nu/Ra 0.29 of system without partitions) are indicated by black solid rectangle. The dash lines represent the best power-law fits of corresponding cases.
5. Conclusion

A series of studies have been carried out on the physical characteristics of the partitioned convection system. Because the height of the gaps is usually selected within the thickness of thermal boundary layer of RB convection, the characteristics of temperature drift characteristics in the partitioned convection system are studied in detail for a given thickness of partitions l=0.12, and the following conclusions are obtained.

(i) By using the thickness of thermal boundary layer to normalize the height of gaps, TD numbers of different gaps’ heights or different Ra numbers are analyzed, and it is found that the curves of these two changes with coincide basically.

(ii) The ratio of gaps’ heights to thickness of thermal boundary layer is selected as constant value, and it is found that TD numbers have nothing to do with Ra numbers, and when , when .

(iii) Even if the thermal boundary layer will change with different Ra number, the temperature distributions of the gap and its adjacent area are basically the same when the are fixed under different Ra numbers.

(iv) Under the same Ra number, the heat transfer efficiency of systems when is higher than that of the systems when . With a given magnitude of , there is a scale relationship between Nu number and Ra number, .

Reference
1 Lappa M 2005 Cryst. Res. Technol. 40 531 https://dx.doi.org/10.1002/(ISSN)1521-4079
2 Ahlers G Grossmann S Lohse D 2009 Rev. Mod. Phys. 81 503 https://dx.doi.org/10.1103/RevModPhys.81.503
3 Bodenschatz E Pesch W Ahlers G 2000 Annu. Rev. Fluid. Mech. 32 709 https://dx.doi.org/10.1146/annurev.fluid.32.1.709
4 Xia K Q 2013 Theor. Appl. Mech. Lett. 3 052001 https://dx.doi.org/10.1063/2.1305201
5 Du Y B Tong P 1998 Phys. Rev. Lett. 81 987 https://dx.doi.org/10.1103/PhysRevLett.81.987
6 Wei P Chan T S Ni R Zhao X Z Xia K Q 2014 J. Fluid Mech. 740 28 https://dx.doi.org/10.1017/jfm.2013.638
7 Joshi P Rajaei H Kunnen R P J Clercx H J 2017 J. Fluid Mech. 830 R3 https://doi.org/10.1017/jfm.2017.660
8 Kunnen R P Clercx H J Geurts B J 2006 Phys. Rev. E 74 056306 https://dx.doi.org/10.1103/PhysRevE.74.056306
9 Kooij G L Botchev M A Geurts B J 2015 Int. J. Heat & Fluid Flow 55 https://dx.doi.org/10.1016/j.theriogenology.2019.08.003
10 Weiss S Wei P Ahlers G 2016 Phys. Rev. E 93 043102 https://dx.doi.org/10.1103/PhysRevE.93.043102
11 Guzman D N Xie Y Chen S Rivas D F Sun C Lohse D Ahlers G 2016 J. Fluid Mech. 787 331 https://dx.doi.org/10.1017/jfm.2015.701
12 Cheng J P Zhang H N Cai W H Li S N Li F C 2017 Phys. Rev. E 96 013111 https://dx.doi.org/10.1103/PhysRevE.96.013111
13 Cheng J P Cai W H Zhang H N Li F C Shen L Qian S Z 2019 Phys. Fluids 31 023105 https://dx.doi.org/10.1063/1.5083195
14 Bao Y Chen J Liu B F She Z S Zhang J Zhou Q 2015 J. Fluid Mech. 784 R5 https://doi.org/10.1017/jfm.2015.610
15 Bao Y Lin Z P Ding G Y 2017 Computer Aided Engineering 26 57 (in Chinese) https://doi.org/10.13340/j.cae.2017.05.011
16 Lin Z P Bao Y 2018 Sci. Sin. Phys. Mech. & Astron. 48 054702 https://doi.org/10.1360/SSPMA2017-00330
17 Lin Z P Bao Y 2018 Sci. Sin. Phys. Mech. & Astron. 48 104702 (in Chinese) https://doi.org/10.1360/SSPMA2018-00183
18 Bao Y Lin Z P 2019 Sci. Sin. Phys. Mech. & Astron. 49 044701 (in Chinese) https://doi.org/10.1360/SSPMA2018-00252
19 Hajmohammadi M R 2018 Int. J. Refrigeration 88 16 https://dx.doi.org/10.1016/j.ijrefrig.2017.11.032
20 Fugmann H Di Lauro P Sawant A Schnabel L 2018 Energies 11 1322 https://dx.doi.org/10.3390/en11051322
21 Xue Y Ge Z Du X Yang L 2018 Energies 11 1398 https://dx.doi.org/10.3390/en11061398
22 Zhou Q Stevens R J A M Sugiyama K Grossmann S Lohse D Xia K Q 2010 J. Fluid Mech. 664 297 https://dx.doi.org/10.1017/S0022112010003824