Numerical Study of Mixed Convection in Uniformly Heated Air-Based Tubular Cavity Receiver of Solar Parabolic Trough Concentrator with Different Boundary Conditions

Simultaneous forced and free convection heat transfer in a horizontal heated cylinder has been the subject of much research, due to particular importance in many industrial engineering applications such as PTC's solar absorber, which is the key thermal component of a PTC.

Roldlan et al. [1] used steam as the heat transfer fluid to develop a computational Fluid Dynamics (CFD) model to estimate the pipes thermal behavior of parabolic trough solar collectors PTC and validated the simulation results with experimental data. Song et al. [2] proposed helical screw-tape inserts to improve thermal efficiency by homogenizing the absorber tube temperature distribution.

For laminar flow in horizontal tubes, natural convection plays an important part in flow and heat transfer. Barozzi et al. [3] performed experiments to investigate the effect of the Reynolds and Rayleigh number variations on heat transfer results of mixed laminar convection for both horizontal and inclined pipe with 0˚to 60˚ slope; and compared these results with theoretical predictions. Mori et al. [4] carried out experiments to study the effect of buoyancy force on forced convection for fully developed air flow under constant wall heat flux. They showed the effect of secondary flow on the thermal and dynamic behaviour in the cross section, and that the natural convection increases the value of the Nusselt number.

Bergles and Simonds [5] made an experimental study on the effect of natural convection on a laminar flow of water in a glass tube heated by a reasonable and constant heat flux and compared the experimental results with correlations. They find that glass tubes are advantageous for heat transfer.

By the use of the integral method of the boundary layer Morri and Futagami [6] did an analytical study on the effect of the thermogravitational force on the laminar flow of a fluid at a number of Pr close to unity heated by a wall heat flux constant and compared the theoretical results with the experimental data.

In an entrance zone of a horizontal semicircular duct with constant wall heat flux condition, Lei and Trupp [7] examined experimentally combined convection for laminar water flow; for ranges of 400 to 1600 mean Reynolds numbers and high heat flux “Rayleigh number reaches 4.6 x 10⁸”; and they correlate flow and heat transfer results.

Shome and Jensen [8] have numerically studied the effect of viscosity on mixed convective flow in an isothermal pipe. According to the results, the friction factor is more significantly impacted by the change in viscosity than the Nusselt values.

Cheng et al. [9] have numerically studied the effect of LVG lateral vortex generators on the performance of the PTC receiver with different conditions "Reynolds number, incident solar radiation, LVG geometric parameters". They are found that the insertion of LVGs at the level of the internal surface of the solar absorber increases the performance of the energy system and decreases the thermal losses.

By the use of dimensionless numbers; Huang et al. [10] made a qualitative study of the effect of natural convection on the thermal and dynamic behavior of turbulent flow of super-heated steam Pr=1.5 in a PTC absorber. And compared the development of the number of Nusselt Nu and of friction factor for different conditions of NUHF wall condition and UHF conventional condition. In the same context, Li et al. [11] carried out the same study, but with the laminar regime.

Huang et al. [12] have made a numerical study to illustrate the effect of dimples on the thermal and dynamic behavior of turbulent flow in a solar absorber of a PTC subjected to a phenomenon of mixed convection. the deeper dimples are found to be more energy efficient and to increase the performance of a PTC solar absorber most significantly when subjected to non-uniform flow.

Lu et al. [13] have made a theoretical study on the performance and heat transfer of the receiver of a cylindrical parabolic concentrator by the use of the energy balance between the heat transfer fluid and the various components of the solar absorber and the external environment, and the results are validated by the experimental. it is found that the thermal performance of a solar absorber subjected to a non-uniform heat flux is higher than that of a uniform flux on the circumference of the wall.

Myoz and Abánades [14] have numerically analysed the influence of internal fins in a solar absorber of PTC on thermal and pressure losses. They study several configurations of geometry, number of fins, and helix angle. They find that the internal fins can increase the efficiency of the solar plant. Hachicha et al. [15] have developed an optical numerical model coupled with another thermal model based on the finite volume method FVM to evaluate the inlet solar flux towards the absorber of a PTC and the energy balances of this solar receiver. Its optical model is validated by analytical results and the thermal model validated by experimental measurements obtained by Sandia National Laboratories. Polyakov [16] solved the flow and temperature field distribution of turbulent mixed convection in horizontal tubes analytically. However, these solutions are only suitable for thermal gravitation that is rather modest. Experiments were used by Petukhov et al. [17] to study turbulent mixed convection in horizontal tubes. Abdelmeguid and Spalding [18] use a finite difference procedure to show the effect of secondary flow induced by the buoyancy force on the thermal and dynamic behavior of a turbulent flow of a fluid subjected to a constant heat flux on the wall with different angles of inclination from 0 degrees (horizontal tube) to 90 degrees (vertical tube). Grassi and Testi [19] used perfluorohexane as the working fluid to experimentally study mixed convection in weakly turbulent flow. They developed correlations of the Nusselt number in the lower and upper part of the tube showing agreement with the experimental data in the band of deviation of 10%.

By a numerical study, Forooghi and Hooman [20] focused on the effect of the buoyancy force on the deterioration of the heat transfer and the thermal uniformity of a turbulent flow of supercritical fluid in a cylindrical pipe with different angles of inclination.

Faheem et al. [21] have made a comparative numerical study between several turbulence models used to simulate turbulent mixed convection in a horizontal smooth pipe with a Renolds number ‎Re=23000 and Richardson number Ri=1.4 and they compare their results with expressions empirical. They find that the K-e model is the most favorable model for showing the thermal and dynamic behavior of the flow in a fully developed regime.

Mohammed and Salman [22] observed that the presence of laminar mixed convection in a horizontal cylindrical tube is unfavorable to heat exchange at low Reynolds number while it is favorable for higher Reynolds numbers. Choi and Choi [23] demonstrated the increase in heat exchange as well as the appearance of natural convection rolls, linked to the establishment of mixed convection in a horizontal tube that is not uniformly heated around its circumference.

By using the Monte Carlo ray-tracing method, coupled with a finite volume solver; Wirz et al. [24] formulated a 3D optical-thermal numerical model to evaluate the non-uniform incident solar flux distribution around the PTC absorber, and the different heat loss and energy efficiency of the latter. They used the spectral optical data for the absorber tube with selective coating and for the glass envelope for more precision.

In this paper, a three-dimensional numerical simulation on laminar mixed convection of air fluid Pr=0.72 in the vacuum receiver tube of the parabolic trough solar collector is carried out under different boundary conditions. We change the mass flow rate; heat flux; Tinlet and we notice the effect of the Reynolds number Re; Grashof number Gr; and Richardson number Ri respectively. The effect of solar elevation angle is also investigated by the change of the location of heat flux. And finally, we present the effect of wall thickness and material.

The flow is three-dimensional, fully developed, laminar. All the thermo-physical properties of the fluid keep constant at reference temperature T₀ evaluated by the fluid bulk temperature at inlet. The density varies with temperature with Boussinesq approximation (The gravity acting along the direction y). So, the governing equations can express as:

$\frac{\partial}{\partial \mathrm{x}_{\mathrm{i}}} \rho \mathrm{u}_{\mathrm{j}}=0$                  (1)

$\frac{\partial}{\partial x_j} \rho u_i u_j=-\frac{\partial \mathrm{p}}{\partial \mathrm{x}_{\mathrm{i}}}+\mu \frac{\partial^2 \mathrm{u}_{\mathrm{i}}}{\partial \mathrm{x}_{\mathrm{j}}^2}-\beta \rho\left(\mathrm{T}-\mathrm{T}_0\right) g_{\mathrm{i}}$                  (2)

$\frac{\partial}{\partial \mathrm{x}_{\mathrm{j}}} \rho \mathrm{u}_{\mathrm{j}} \mathrm{T}=\frac{\partial}{\partial \mathrm{x}_{\mathrm{j}}}\left[\frac{\mu}{\operatorname{Pr}} \frac{\partial \mathrm{T}}{\partial \mathrm{x}_{\mathrm{j}}}\right]$                  (3)

In this work, steady state simulations are run. The commercial computational fluid dynamics (CFD) solver ANSYS FLUENT 14, which is based on the finite-volume method, is used to conduct all of the numerical simulations. The first and second upwind schemes, respectively, discretize the convective terms in momentum and energy equations. The pressure is discretized by using the standard scheme. The pressure-velocity coupling is dealt with the SIMPLE algorithm.

Table 1. Grid independence checking “Nusselt number”

3.png

Figure 3. Grid in cross section

The receiver tube was modelled and meshed in three dimensions using Gambit modelling and meshing software. The grids in cross section are shown in Figure 3. Because the velocity and temperature gradients are greater near the inner wall than in any other areas in cross section, we can see that the meshing is severe in the radial direction close to the wall. The averaged Nusselt number at Re=2200 and heat flux “full concentration condition” =110.9358w/m² from these grid systems are listed in Table 1. The grid system with 200000 cells is chosen to strike a compromise between the computational load and the numerical accuracy since it has relative deviations of averaged Nusselt at 200000 cells of 0.134%.

9.1 Thermal behaviour according to the axial direction

Figure 6 shows the development of temperature with the z axis at Re=2200 and heat flux q= 189.6353 w/m² “full concentration of heat flux at the wall”. We note that the temperature field stabilizes at a position far enough from the tube inlet. The fully developed region is at z=6m and z=6.5m.

6.png

Figure 6. The temperature distribution for each cross section along the absorber

Table 2. Nu vs z

7.png

Figure 7. Nu distribution with z axis

Table 2 and Figure 7 show the distribution of Nu with z axis. The Nu number established at the fully developed region. For the mixed convection, the fully developed Nu is much higher than for the theoretical Nu of the forced laminar convection in circular tube under UHF Nu₀ = 4,36.

The velocity and temperature distributions were simulated at z = 6.5 m where the axial temperature gradient was kept completely constant “fully developed region”.

9.2 Comparison between forced convection and mixed convection

Figure 8 and Figure 9 showed the difference between mixed convection and pure forced convection. For mixed convection at Gr=3x10⁶ and Re=2200; the effect of buoyancy is clear and it forms a pair of transverse hydrodynamic secondary flow rolls which are symmetrical in the vertical diametric plane and are opposite in direction of rotation. These two vortices caused a drastic change in the velocity and temperature profile.

In Figure 8 The absorber is subjected to a heat flux over the entire circumference of the tube, which causes "under the effect of gravity" and by the mechanism of natural convection, a transport of mass and heat in the cross section in which fluid motion is generated by density differences in the fluid due to temperature gradients. Therefore, the mass is concentrated downwards and pulls the velocity profile lower, which distorts the Poiseuille profile. In addition, the layers of higher temperatures rise higher.

In Figure 9, the heat flux is concentrated on the wall of the lower half cylinder of the tube. We note that the temperature and the axial velocity fields have a bilateral concave shape in the down region of the cross section.

A/ Case one: Total uniform heat flux UHF

8a.png

8b.png

Figure 8. Velocity and temperature distributions at Re = 2200. Left: temperature; middle: secondary flow; right: streamwise velocity

B/ Case two: half down UHF

9a.png

9b.png

Figure 9. Velocity and temperature distributions at Re = 2200. Left: temperature; middle: secondary flow; right: streamwise velocity

10.png

Figure 10. Non-dimensional axial velocity and gradient temperature distributions along vertical diameter at Re=2200 and Gr=3x10⁶

By comparing Figure 8 and 9 we notice the rise of two counter-rotating transverse rollers of a hydrodynamic secondary flow towards the upper half of the cross-section as the heat flux is concentrated towards the wall of the lower half of the tube.

Figure 10(a) shows the adimensional axial velocity distribution W+ in the vertical diametral plane. Knowing that the origins of the x- and y-axes will be the centre of the tube cross-section; and on the y-axis, the upward direction will be positive in all the rest of this study.

From the diagram we notice that the maximum velocity W⁺_max increases and shifts downwards compared to the Poiseuille flow. Due to the increase in the concentration of the heat flux down the tube, W⁺_max is larger and more considerable for the UHF half down case than for the full UHF case, but its position on the diametrical plane is the “same”.

We also notice in Figure 10(b) that the maximum value of the adimensional temperature gradient T⁺_max increases and shifts downwards compared to the case of pure forced convection.

*Nusselt numbers and friction factors “Total uniform heat flux case”

Figure 11 Shows the ratio of Nu/Nu₀ in the fully developed region, relative to Gr under UHF at different Re numbers, where Nu₀ is the Nusselt number for the case of purely forced convection without secondary flow. We notice that, Under the influence of thermogravitational forces, Nu becomes considerably larger than Nu₀ and it reaches values more than twice its value without buoyancy force under UHF, and the effect of Re is neglected.

Figure 12 Presents the performance of heat transfer by mixed convection compared to that by forced convection, for different Gr and Re numbers. Knowing that the ratio f/f₀ represents the friction at mixed convection f compared to that of purely forced convection f₀ and it is an energetically negative term. We note that (Nu/Nu₀)/(f/f₀) is greater than 1 and increases with the decrease of Re number. We can say that natural convection is a positive phenomenon in terms of energy and increases the performance of the absorber.

11.png

Figure 11. Nu/Nu₀ vs Gr

9.3 Comparison mixed convection for different Gr number for the case of half down UHF

In Figure 13, the velocity and temperature distributions at Re=2200 with a half concentration UHF (ϕ=90˚) and varying Gr are depicted. It can be seen that the intensity of the secondary flow is high near the upper half cylinder of the absorber wall, and increases in its scale as the Gr number increases. With the increase of the gravitational thermal effect by the increase of the heat flux the two tourbillons are deformed and rise gradually towards the top of the tube.

By using the dimensionless parameters of velocity W and temperature T, the dynamic and thermal behaviour can be presented qualitatively in Figures 14 and 15. In the horizontal diameter, the temperature distribution in the central area of the diameter plane becomes more uniform and less concave as Gr increases.

In Figure 15(b) for the horizontal diameter, the concavity of the velocity profile becomes more open and less intense with increasing Gr.

12.png

Figure 12. (Nu/Nu₀)/(f/f₀) vs Gr

13a.png

13b.png

13c.png

Figure 13. Velocity and temperature distributions at Re = 2200. Left: temperature; middle: streamwise velocity; right: secondary flow

14.png

Figure 14. ϴ for different Gr number

15.png

Figure 15. W for different Gr number

9.4 Comparison mixed convection for different Ri number “T inlet” for the case of half down UHF

Now the mass flow rate and the heat flux are fixed “m=1.1x10^-3kg/s, q= 665.615 W/m²” and heated air is injected for the simulation at different inlet temperatures. When T_inlet increases the Richardson number decreases and the effect of natural convection becomes less significant.

The velocity and temperature distributions under UHF (ϕ=90˚) with varied Ri are shown in Figure 16. As Ri increases from 0.0477 to 0.4 the velocity and temperature profile become more concave. It is observed that the vortex center moves progressively upwards and slightly towards the tube wall as Ri increases.

Figure 17 and Figure 18 presents the distributions of non-dimensional velocity W and temperature ϴ along horizontal and vertical diameters under half down UHF (ϕ=90˚) with varied Ri number.

In vertical diameter it is seen that the temperature distribution becomes more uniform with the increasing of Ri number (Figure 17(a)) Along the horizontal diameter (Figure 17(b)) it is noticed that the temperature gradient curve is more concave and closer to zero "at the central part of the plane" with the increase of the effect of natural convection.

For flow distributions along the vertical diameter, as Ri increases, the location of the maximum axial velocity moves downward and its value decreased (Figure 18(a)).

In Figure 18(b), with the horizontal diameter we notice that the axial velocity in the center decreased and the curve of axial velocity becomes more concave when Ri number increase.

16a.png

16b.png

666.png

16d.png

Figure 16. Velocity and temperature distributions for different case of Ri number. Left: temperature; middle: streamwise velocity; right: secondary flow

17.png

Figure 17. ϴ for different Ri number “T inlet”

18.png

Figure 18. W for different Ri number “T inlet”

9.4 The effect of solar elevation angle

Figure 19 shows the velocity and temperature distributions at Re = 2200 and Gr = 6x10⁶ under UHF with varied solar elevation angle. With the movement of the heating around the circumference of the wall of the absorber, from the angle of solar elevation ϕ=90˚ to ϕ=60˚ and ϕ=30˚ one of the vortices rises, deforms and lengthens and the other descends downwards clockwise and the symmetry gradually disappeared.

For the case of ϕ=0˚, we notice only one vortex located in the left-lower quadrant of the cross-section.

19a.png

19b.png

19c.png

19d.png

Figure 19. Velocity and temperature distributions at Re = 2200 and Gr = 6x10⁶ under different UHF distributions cases; right: secondary flow; middle: streamwise velocity; Left: temperature

Concerning the axial speed, we note that the bilateral concave dynamic behaviour at ϕ=0˚ deforms and moves towards the heated zone from ϕ=90˚ to ϕ=0˚.

For the case of ϕ=90˚, the thermal behaviour belongs as the most homogeneous. This homogeneity is minimized and gradually degrades until the case of ϕ=0˚ where the profile of T belongs stratifies under isothermal layers oriented slightly towards the heated part "on the left".

9.5 Comparison of fully developed Nu for different cases of Distribution of heat flux “solar elevation angle” about the absorber

The variation of fully developed Nusselt number with Gr are shown in Figure 20 at different solar elevation angles and Re=2200. The distribution and location of heat flux has a great effect on Nu at the same Gr and Re. For a fixed Gr, the Nu number of mixed convection under total UHF and half lateral UHF (ϕ = 0˚) is less than that under UHF (ϕ = 30˚), and this latter is lower than that under UHF (ϕ=60˚ and 90˚). The effect of natural convection is most effective for convective heat exchange when the thermal flux concentrates below the absorber and diffuses from bottom to top with the direction of Archimedes' thrust.

20.png

Figure 20. Fully developed Nu vs Gr at different solar elevation angles Re=2200

9.6 The effect of wall thickness and material

From Figure 21 we notice that the convective exchange is higher and the Nu develops rapidly for the absorber which has the most conductive wall (λ copper = 387.6 w /mk; λ steel = 16.27 w / mk) and which has a lower thickness.

21.png

Figure 21. Nu vs Gr for different wall thickness/material