- Email: [email protected]
Windbreak walls reverse the negative effect of crosswind in short natural draft dry cooling towers into a performance enhancement
International Journal of Heat and Mass Transfer 63 (2013) 162–170
Contents lists available at SciVerse ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Windbreak walls reverse the negative effect of crosswind in short natural draft dry cooling towers into a performance enhancement Y. Lu ⇑, Z. Guan, H. Gurgenci, Z. Zou Queensland Geothermal Energy Centre of Excellence, Qld 4072, Australia School of Mechanical and Mining Engineering, The University of Queensland, Qld 4072, Australia
a r t i c l e
i n f o
a b s t r a c t
Article history: Received 20 September 2012 Received in revised form 14 December 2012 Accepted 28 March 2013 Available online 27 April 2013
Crosswind effect on the cooling performance of large natural draft dry cooling towers (NDDCT) has been verified to be unfavourable by many researchers. Small size natural draft cooling towers (height <30 m) proposed for geothermal and other renewable power plants are expected to be more negatively affected. CFD modelling has been carried out to numerically analyse the heat transfer performance of a 15 m-high small NDDCT under different crosswind speeds. Simulations show that, at certain crosswind speeds, the crosswind significantly degrades the cooling performance. However, the negative effect of the crosswind can be turned into positive in small natural draft cooling towers by introducing windbreak walls that guide the air mobilised by crosswind through the heat exchangers. When windbreak walls are used, the results show that the tower performance improves with increased crosswind velocity. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: Crosswind Natural draft cooling tower Windbreak wall CFD modelling Numerical simulation
1. Introduction Natural draft dry cooling towers (NDDCT) are widely used in Rankine cycle power plants in arid areas around the world. Geothermal and solar thermal power plants are more likely located in arid areas and are likely to use forced- or natural-draft air cooling. In a natural draft dry cooling tower, no fans are required. The flow of air through the bundles of heat exchangers is by means of buoyancy effects. Buoyancy occurs due to a difference in air density between the inside and outside of the tower resulting from the temperature difference. The greater the temperature difference and the height of the tower structure, the greater the buoyancy force. In natural draft cooling tower design, both the aerodynamic balance and thermodynamic balance should be satisfied at the same time which can be expressed as follows [1]:
DP ðqao qai ÞgðHt Hhx Þ ¼
X
K resist
q
a
v2
2
Q ¼ ma cpa ðT ao T ai Þ ¼ ml C pl ðT li T lo Þ ¼ hu AF T DT lm
ð1Þ ð2Þ
The first equation means the total pressure drop over various components of the tower must be balanced by the buoyancy force. The Eq. (2) states that the heat transferred into the air is equal to the ⇑ Corresponding author at: Queensland Geothermal Energy Centre of Excellence, Qld 4072, Australia. Tel.: +61 7 336 542 48. E-mail address: [email protected] (Y. Lu). 0017-9310/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2013.03.075
heat extracted from the cooling liquid (water) and this heat is transferred through the heat exchangers. The above equations for NDDCT design and selection do not include the crosswind effects. The effect of the crosswind is common and seen during operations of both wet and dry cooling towers in thermal power plants. Early studies on the crosswind influence on natural draft cooling towers were focused on either experimental methods such as full scale tower measurements [2] or laboratory tests [3–5]. However, numerical analysis (CFD) became the preferred method since the last decade [6,7]. Both the experimental and the numerical studies in the past have discovered that crosswind has a negative influence on the NDDCT cooling performance. For instance, a study showed that the heat transfer rate decreased by more than 30% at crosswind velocities above 10 m/s [7]. Wind break walls, by using either experimental [8] or numerical method [9,10], were found to improve the thermal performance of natural draft cooling towers under windy conditions. However, all the studies above focused on either Heller-type or surface-condenser-type indirect large natural draft cooling towers with heights above 100 m. The Queensland Geothermal Energy Centre of Excellence (QGECE) has been developing small natural draft cooling towers (NDDCT) for geothermal and solar thermal power plants [11]. It is expected that crosswind will have significant negative effect on the performance of small NDDCT. To turn this negative effect into positive, a novel windbreak arrangement has been analysed and the results are reported in this paper.
163
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Nomenclature A area (m2) a constant Aa, Ar air-side area, fin-root area, respectively (m2) Ac surface area of numerical cell (m2) C inertial resistance factor C1, C1e, C2, C3e constants in turbulence equations Cp specific heat (J Kg1 K1) Cl coefficient in turbulent viscosity dr outer diameter of tube (m) F source term for momentum equations FT temperature correction factor H height, elevation (m) h convective heat transfer coefficient (W m2 K1) Gb turbulent kinetic energy source term due to buoyancy Gk turbulent kinetic energy source term due to mean velocity gradients Kresist pressure loss coefficient K, Ke, Kt laminar, effective, turbulent thermal conductivity, respectively (W m1 K1) k turbulent kinetic energy (m2 s2) m mass flow rate (kg/s) nr number of tube rows in a heat exchanger bundle p pressure (Pa) Pr, Prt laminar, turbulent Prandtl number, respectively pt, pd tube transversal pitch, tube diagonal pitch, respectively (m) Q heat transfer rate (W) Qr, Qrw heat transfer rate of radiator without or with windbreak wall (W) q heat flux (W m2) Rec Reynolds number based on minimum free flow area S modulus of the mean rate-of-strain tensor S/ volumetric source term for variable quantity / T temperature (K) DTlm logarithmic mean temperature difference (K)
2. CFD model
U, V, W Vc v x, y, z
velocity components in x-, y-, z-direction (m s1) numerical cell volume (m3) velocity scalar (m s1) Cartesian co-ordinates
Greek letters a permeability (m2) b bulk thermal expansion coefficient (K1) e turbulent kinetic energy dissipation (m2 s3) / scalar quantity (u, v, w, T, k, e, . . .) C/ diffusion coefficient for variable quantity / q, q density, mean density (kg m3) l, le, lt laminar, effective, turbulent viscosity, respectively (kg m1 s1) rk, re turbulent Prandtl number for k and e, respectively s stress tensor Vectors ~ v
velocity
Subscripts a, l air side, liquid (water) side b base of tower cw cross wind e effective hx heat exchanger i, o inside or inlet, outside or outlet r radiator t tower u overall w windbreak wall 0, ref reference value
(4) are proposed by Ganguli et al. [13] and Robinson et al. [14], respectively.
2.1. Tower size scaling The first stage of the study was a theoretical analysis to find the possible smallest size of NDDCT for a particular small renewable power generator under given windless conditions using a onedimensional (1D) mathematical model based on Eqs. (1) and (2) [12]. The geometry of tower is assumed to have cylinder rather than hyperbolic shape. While hyperbolic shape provides better structure strength for reinforced concrete towers and has slightly lower air flow resistance, it increases the building costs especially if the tower is built with steel structure as appropriate for small cooling towers. By contrast, a cylindrically shaped steel tower offers more economic for building and remote area installation though it sacrifices the overall cooling performance a little bit. In fact, the wall profile has very limited influence on the heat transfer and the air flow resistance inside the tower. According to the above two equations Eqs. (1) and (2), contracting and diffusing of the air flow area can cause a difference in the pressure loss, but this change is negligible compared to the total resistance, the largest part of which comes from heat exchangers. The heat transfer and flow characteristics of the heat exchangers in 1D model are based on standard correlations developed for finned tube bundles. This study considers a specific bundle design with three staggered circular-finned tube rows, whose air-side heat transfer correlation Eq. (3) and pressure drop correlation Eq.
0:333 ha ¼ 0:38Re0:6 ðAa =Ar Þ0:15 K=dr c Pr
K resist ¼ 18:93nr Re0:316 c
ð3Þ
0:927 0:515 pt pt dr pd
ð4Þ
It was found that an NDDCT with an internal horizontal heat exchanger placement could be as small as 15 m in height and 12 m Table 1 Proposed design conditions.
*
Design point
Value
Tower aspect ratio* Total thermal efficiency, % Water inlet temperature, °C Ambient air dry bulb temperature, °C Water mass flow rate, kg/s Heat exchanger tube diameter, mm Tube length, m Fin diameter, mm Transversal tube pitch, mm Total heat exchanger front area, m2 Heat exchanger tubes Bundle arrangement
1.25 15 40 20 16 21 8 51 60.6 73.7 3 Rows, 3 passes Horizontal inside the tower
Defined as the ratio of total height to base diameter.
164
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
in base diameter for a plant with net power generation of 100 kW under the proposed design conditions in Table 1. At the design point as defined in Table 1, the heat exchangers can reject around 578 kW heat at no crosswind condition. However, with crosswind, the heat rejected by the heat exchangers will be different. CFD models of the NDDCT of Table 1 have been built in the commercial CFD software ANSYS FLUENT to study its heat rejection performance at different crosswind speeds, with and without a windbreak. 2.2. Model geometry The geometry of tower in the CFD model, including tower support, is a cylinder with the size given above to reflect the practicality of steel construction for small towers. In the model with windbreak, three solid, zero-thickness walls are placed in the tower bottom with equal separating angles of 120°. The heights of the walls are equal to the inlet height of tower, similar to the wall proposed by Du Preez and Kröger [15]. The computational domain (to simulate outside ambient air) is also of cylindrical shape with 90 m in height and 72 m in radius. Past CFD studies show that the distances from tower to domain boundaries affect the numerical results to a certain extent. So this CFD model uses 12 times the tower diameter for the domain diameter and 6 times for the domain height.
without thickness, whose heat transfer rate is calculated by Eq. (6) [18].
q ¼ hðT r T ao Þ
where the heat transfer coefficient, h, is a function of the heat exchanger characteristic parameters and the air inlet velocity, vai, normal to radiator. Tr is the radiator temperature and Tao is the air temperature downstream of the radiator [17]. For air flow pressure drop, the radiator model can simulate resistance to air flow in the direction normal to radiator face. However, it does not provide resistance in other two directions, i.e. velocity components parallel to radiator face. This will cause overestimation of the possibility of vertices occurring near the radiator, since real structure of fin tube heat exchanger bundles can prevent horizontal air flow, allowing air flow through heat exchanger only vertically. Therefore a porous media model is added to represent the pressure loss within the heat exchanger, leaving the radiator model to represent heat transfer only. Porous media model offers an ideal approximation for modelling the heat exchangers in this context where the internal structure detail is not of concern but a distributed resistance is important. A porous media zone introduces an additional source term in momentum equation of each i-axis [18]:
le 1 v þ C qv 2i 2 a i
Fi ¼ 2.3. Boundary conditions As shown in Fig. 1, the wall boundaries are set as slip walls because the boundary layers are sufficiently thin so cannot influence flow separations at tower inlet and outlet [16]. The velocity inlet boundary is defined at the windward side of the domain. The velocity profile is applied in this boundary defined by Eq. (5), where a is recommended as 0.2 [17].
v cw ¼ v ref
y yref
!a ð5Þ
The pressure outlet boundary condition is applied in leeward side as well as on the top of domain. The temperatures on both the inlet and outlet boundaries are set equal to the ambient temperature, i.e. the air inlet temperature of heat exchanger. Several ways of modelling heat exchangers in CFD can be found in open literature. The radiator model in FLUENT is used in this study to represent the heat exchanger bundles as a lumped face
ð6Þ
ð7Þ
where a and C are determined by the friction factor of heat exchangers in the 1D model. By this modelling strategy, the vertical air flow can be guaranteed by setting the resistances in other two directions much larger than that in vertical direction (y axis). The tower support is set to the porous jump boundary which is simplified as a cylinder face with same pressure resistance coefficients corresponding to those of supporting structures in real towers. 2.4. Governing equations Since the air velocity in this study is far below 0.3 Mach, the incompressible air model with constant density is assumed. A buoyancy generating term is introduced in vertical component of the momentum equation using the Boussinesq’s approximation to reflect the buoyancy effect caused by the density difference.
Table 2 Summary of governing equations. Equation
/
C/
S/
Continuity x Momentum
1 U
0
0
V
le le
@ ~ @p @x þ r ðle @x v Þ þ F x
y Momentum z Momentum
W
le
Energy
T
Ke Cp
@ ~ @p @z þ r ðle @z v Þ þ F z
k
k
e**
e
l þ rlkt l þ rlet
where
le = l + lt, lt ¼ qC l ke , K e ¼ K þ K t ; K t ¼ CPrp lt t 2
l 1 Gk = ltS2, Gb ¼ bg Prtt @T @y , b ¼ T 0 P
0 h i Ske C 1 ¼ max 0:43; Skþ5 , C1e = 1.44, C3e = tanh(V/U), C2 = 1.92, rk = 1.0, re = 1.44,
e
Pr = 0.74, Prt = 0.85, T0 = 293.15 **
The realizable k–e turbulence model is used in this modelling.
@ ~ @p @y þ r ðle @y v Þ q0 bðT T 0 Þg þ F y
1 Cp
qAc Vc
Gk + Gb qe 2
e qC 1 Se qC 2 kþepffiffiffiffi v e þ C 1e C 3e Gb k
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
72
165
Unit: m
Pressure outlet
Velocity inlet Pressure outlet
90
Radiator 12
Porous media zone Wall
15 3 y Porous jump
Wall
z x
Fig. 1. Geometry of 3D models with or without walls.
The model is simulated by solving a series of conservation equations of physical quantities, whose general term is expressed as:
div ðq~ v /Þ ¼ div ðC/ grad/Þ þ S/
ð8Þ
The expressions of /, C/ and S/ for the above equation are shown in Table 2. The source terms Fx, Fy, Fz in each momentum equation refer to c porous media resistance defined as Eq. (7), while qA in energy Vc source represents heat transfer of heat exchangers with q calculated by Eq. (6).
Fig. 2. Structured meshes in tower body and ground.
The implicit partial differential governing equations are discretised with the second order of upwind discretization scheme and are decoupled using pressure-based segregated algorithms: SIMPLE [18]. The convergence criterion is that the scaled residuals for all variables (except energy) drop to the order of 105 and the monitored variables remain stable when iterating. The calculation process is iterated for more than 15,000 steps and converged results can be obtained.
Fig. 3. 3D streamlines inside and under cooling tower when crosswind speed is (a) 0 m/s, (b) 0.5 m/s, (c) 2 m/s, (d) 4 m/s, (e) 6 m/s and (f) 8 m/s.
166
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Fig. 4. The temperature contour at the mid-xy plane when wind speed is (a) 0 m/s, (b) 0.5 m/s, (c) 2 m/s, (d) 4 m/s, (e) 6 m/s and (f) 8 m/s.
cell layer near walls is 0.08 m while the maximum cell size inside the tower is 0.15 m. And a finer mesh is used in the regions of heat exchanger and tower outlet where cell size is less than 0.1 m. Fig. 2 shows the final mesh used in the CFD model. 3. Results and discussions
Fig. 5. Velocity vectors at mid-xy plane when crosswind speed = 6 m/s.
Grid-independence has been tested by analysing the no-crosswind cases at different mesh sizes. When the cell quantity is over 3 million, the conserved variables monitored change by less than 0.5% compared with results in mathematics model. Final model uses about 3,750,000 structured prism cells. The thickness of first
The small tower performance was first simulated under nowind conditions. This was followed by the cases with and without windbreak walls under various crosswind speeds. The numerical results in no-wind case matches the calculation results of 1D model quite well, which itself has been validated against the industrial cooling tower data in [1]. The differences between the 1D and 3D models in outlet air mass flow rate and mean temperature through the heat exchangers are about 0.8% and 1.5%, respectively. This 1Dverified CFD model is then used for modelling at various crosswinds. When crosswinds are introduced, the crosswind speed was varied from zero (no wind) to maximum of 18 m/s at the reference elevation of 10m as defined in Eq. (5). 3.1. Crosswind influence on the tower performance without windbreak wall With horizontally-blowing wind, the airflow inside the cooling tower is not only driven by the buoyancy force, but also is
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Fig. 6. Pressure contours at the heat exchanger inlet face when crosswind speed is (left) 4 m/s and (right) 8 m/s showing the negative pressure zone.
Fig. 7. Velocity vectors at mid-xy plane for case study.
Fig. 8. Comparisons of different crosswinds effects between models with (maw, Qrw, Qw) and without walls (ma, Qr, Q).
167
168
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Fig. 9. Air streamlines passing through the heat exchanger in direction of (left) backward and (right) forward for wind-wall case at crosswind speed of 2 m/s.
Fig. 10. Pressure contours at heat exchanger inlet face with the windbreak walls installed when crosswind speed is (left) 4 m/s and (right) 8 m/s.
subjected to the outside crosswind. Fig. 3 shows the airflow 3D streamlines inside the tower as well as at the bottom of the cooling tower while Fig. 4 is the air temperature contours at the central vertical cross section of the tower at various crosswind speeds. At low wind speeds such as 0.5 m/s, the air flow inside the tower is near uniform, and the ambient air enters into the tower bottom only through the windward side. As the speed of crosswind increases, two vertices form due to the different mechanisms applying in each region and penetration or ‘‘downwash’’ [19] of hot air in the downstream of the tower outlet is observed. The upper vortex is caused by the crosswind forming a high speed zone acting like a lid above tower outlet. The warm air exiting from the tower at a much slower speed (around 0.4 m/s) cannot break through the wind lid. In fact, the warm air is quickly cooled down near the tower exit by the cross wind and some cooler air sinks back into the cooling tower (Fig. 5). This phenomenon was referred to as the cold inflow in large industrial towers, which can be usually assessed with the value of the Froude number [20]. The result is the reduction of the effective draft height of the cooling tower. While at the tower bottom, hot air inside the tower at the windward side is sucked down because of the negative pressure underneath the heat exchangers (Fig. 6) caused by the
crosswind speed. This part of air re-enters into the heat exchanger bundles at leeward side, forming another hot air circulation. The lower vortex largely decreases effective transfer area of heat exchanger bundles and makes heat transfer in this region rather complicated. Although the emergence of these two air vertices inside tower is attributed to different mechanisms, the suction effect under the heat exchangers is believed to play a dominant role. This has been proved by a complementary case study. In this case study, the space outside the tower is divided into two parts by virtual horizontal faces. And in each simulation, crosswind flow is applied in either upper or lower part of computation domain to study their effect separately. The velocity vectors at the cross section of the mid-xy plan (Fig. 7(a)) show that in the case when crosswind is applied on the tower outlet (upper part) only, the inside tower air flow field does not change much compared with no-crosswind case. However, when the crosswind is applied only at the tower inlet (lower part), air flow reverses its direction (Fig. 7(b)), for the zone below the heat exchangers now has lower pressure. The effect of crosswind on the NDDCT cooling performance is better understood by separately examining the two cooling
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
components: the heat taken away by the air updraft leaving through the tower outlet at the top Q, and the heat taken away, Q0 , by the air that leaves through the bottom part of the tower after circulating through the heat exchangers driven by the lower vortex explained earlier. The sum of these two components is equal to the total heat taken away from the heat exchangers (radiator), Qr, which is independently computed by FLUENT as the radiator heat transfer, namely Qr = Q + Q0 . The first component, i.e. the heat Q dissipated through cooling tower outlet is calculated using the second term in Eq. (2). Here the air mass flow rate ma is the net value of the air flowing through the radiator. It is extracted from the numerical results as the upward mass flow rate minus the downward one in the case when inverse air flow occurs at heat exchangers. The heat exchanger exit temperature Tao is calculated by averaging the numerically calculated air temperatures at the face of tower outlet. The heat exchanger inlet temperature Tai is assumed to be equal to the ambient air temperature. The second cooling component, Q0 , is calculated by subtracting Q from Qr. When the air flow is unidirectional inside the tower, Q0 is zero and thus Qr = Q. Solid lines in Fig. 8 indicate the change of ma, Q and Qr under different crosswind conditions. The heat dissipated through the tower outlet Q keeps declining along with the rise of crosswind speed while the mass flow rate ma decreases and remains nearly constant after 10 m/s. However, in this small tower, the crosswind does not always exert a negative effect on the cooling tower performance in terms of total heat transferred by radiator Qr. Qr reaches its lowest point at a crosswind speed around 5 m/s and then increases with increase of crosswind speed. This means under high-speed (>5 m/s) crosswind conditions, Q0 becomes a dominant component in the overall heat transfer rate. This phenomenon is seldom seen in large NDDCT installations since a tall tower provides a relatively large draft force for hot air and normal crosswind speeds are not high enough to cause inverse flow at heat exchangers against this relatively large updraft. With the existence of crosswind, the total transferred heat Qr could decrease by 37% compared with no-crosswind condition, which leads to a significant drop in net power generation at certain cross wind speeds. 3.2. Crosswind influence on the tower with windbreak walls Three simple but very effective windbreak walls have been introduced aiming to improve the performance of small NDDCTs under crosswind. The function of these walls is to divert crosswind flow through the heat exchangers rather than flowing across the bottom of the tower. The walls are installed in the bottom of the tower with 120 degrees separation. When there is no crosswind, the cooling air enters into the tower freely without any obstruction from the walls. If crosswind exists, the walls stop the crosswind flowing across the bottom, change the direction of the crosswind, and force it flow through heat exchanger. Since more air flows through the heat exchanger, it improves the performance of the tower. The preliminary simulations found that the total cooling performance is substantially enhanced regardless of the wind direction but the level of the improvement is dependent on the wind directions. In this paper, only one simple case is reported that the coming crosswind is facing right against sector A which is symmetric about x axis as seen in Fig. 9. The results of simulation with windbreak walls are plotted using dashed lines in Fig. 8. The subscript ‘‘w’’ is used to refer to the results obtained with windbreak walls. There is a significant increase in total radiator heat transfer, Qrw, as well as in air mass flow rate maw with increasing crosswind speed. This is directly related to the change in air flow field in the tower bottom. In sector A, crosswind is stopped by the walls so that nearly all the air flows upward through the
169
heat exchangers, while in sectors B and C, the draft effect of cooling tower remains sufficient to drive outside air flows into it through the heat exchangers continually although vortices occur in the wake area of walls. The increase of upward mass flow rate inside the tower can be attributed to the air pressure under the heat exchangers being higher than that without the windbreak walls at the same wind speed (Figs. 6 and 10). And among the three sectors, the pressure in sector A is remarkably higher. Since there is no downward air flow through the heat exchanger, the heat dissipated through tower outlet Qw equals to Qrw as expected. The addition of a wind break wall significantly improves the total thermal performance of a cooling tower. For the case study considered, a 30% increase in total heat transfer is estimated when the crosswind speed is around 5 m/s, a speed that would produce a substantial reduction of tower performance without a windbreak. This clearly shows the benefit of using wind break walls.
4. Conclusions Crosswind would stop small natural draft cooling tower functioning if no windbreak walls are used at high crosswind areas. This paper demonstrates that the negative effect of the crosswind can be reversed into a performance enhancement in short natural draft dry cooling towers. CFD modelling has been done to quantify the crosswind effects on cooling performance of small size NDDCT with and without windbreak walls. A new approach has been introduced to simulate the heat transfer and pressure drop of the heat exchanger in the cooling tower model: a combination of a FLUENT ‘‘radiator’’ element to represent the convective heat transfer term and a porous media zone to represent the heat exchanger pressure drop. Simulations under different crosswind speeds indicate that the tower heat transfer significantly suffers from crosswind when no wind break wall is installed. On the other hand, with wind break walls, the negative effect of the crosswind is turned into an advantage and both the total heat transfer and mass flow rate are enhanced. The results are internally consistent and the numerical predictions under no crosswind conditions are in agreement with the correlations developed using industrial data. Experiments are being planned to test the fundamental assumptions of the representation used in this model so as to produce a more rigorous experimental validation of the numerical method. The following observations apply based on the present results: 1. Without wind break walls, the air flow field inside the tower is disturbed by the horizontally-flowing crosswind forming two major vortices leading to inverse flow through the heat exchangers. The main reason is the suction effect of the wind passing underneath the heat exchangers. 2. With such inverse air flow, the total heat transfer Qr between the heat exchangers and the air is no longer unidirectional and can be dissipated through both the tower top (Q) and the tower bottom (Q0 ) at the same time. And at certain wind speed, larger part of heat is dissipated underneath the heat exchangers, which unexpectedly increases the total cooling performance of heat exchangers. 3. When windbreak walls are employed, the cooling performance is consistently improved with increasing crosswind speeds in the considered wind direction. The results show that the use of natural draft dry cooling towers is feasible with small renewable power plants if wind break walls are employed. This is a significant conclusion and will assist the commercial viability of remote area renewable power generators.
170
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Appendix A. Fig. 1–10. References [1] D.G. Kröger, Air-Cooled Heat Exchangers and Cooling Towers, II, Pennwell Corp, Tulsa, 2004. pp. 38–91. [2] A.F. Du Preez, D.G. Kröger, Effect of wind on performance of a dry-cooling tower, Heat Recovery Syst. CHP 13 (2) (1993) 139–146. [3] M. Gao, F.Z.H. Sun, Research on the Effect of Cross-wind to Temperature Difference and Efficiency of Natural Draft Counter flow Wet Cooling Tower, Springer Berlin Heidelberg, Berlin, Heidelberg, 2007. pp. 513–517. [4] Q.D. Wei, B.Y. Zhang, K.Q. Liu, X.D. Du, X.Z. Meng, A study of the unfavorable effects of wind on the cooling efficiency of dry cooling-towers, J. Wind Eng. Ind. Aerodyn. 54 (1995) 633–643. [5] D.D. Derksen, T.J. Bender, D.J. Bergstrom, K.S. Rezkallah, A study on the effects of wind on the air intake flow rate of a cooling tower: Part 1. Wind tunnel study, J. Wind Eng. Ind. Aerodyn. 64 (1) (1996) 47–59. [6] G. Ge, F. Xiao, S. Wang, L. Pu, Effects of discharge recirculation in cooling towers on energy efficiency and visible plume potential of chilling plants, Appl. Therm. Eng. 39 (Journal Article) (2012) 37–44. [7] R. Al-Waked, M. Behnia, The performance of natural draft dry cooling towers under crosswind: CFD study, Int. J. Energy Res. 28 (2) (2004) 147–161. [8] Y. Chen, F. Sun, H. Wang, N. Mu, M. Gao, Experimental research of the cross walls effect on the thermal performance of wet cooling towers under crosswind conditions, Appl. Therm. Eng. 31 (17) (2011) 4007–4013. [9] R. Al-Waked, M. Behnia, The effect of windbreak walls on the thermal performance of natural draft dry cooling towers, Heat Transfer Eng. 26 (8) (2005) 50–62.
[10] Z. Zhai, S. Fu, Improving cooling efficiency of dry-cooling towers under crosswind conditions by using wind-break methods, Appl. Therm. Eng. 26 (10) (2005) 1008–1017. [11] H. Gurgenci, Z. Guan, Dirigible natural draft cooling tower in geothermal and solar power plant applications, in: Proceedings of 14th IAHR International Conference on Cooling Tower and Air-Cooled Heat Exchangers, Stellenbosch, 2009. [12] Y. Lu, Z. Guan, H. Gurgenci, Development of small natural draft dry cooling towers for geothermal power plants, in: The 15th IAHR Cooling Tower and Aircooled Heat Exchanger Conference, Beijing, 2011, pp. 308–316. [13] A. Ganguli, S.S. Tung, J. Taborek, Parametric study of air-cooled heat exchanger finned tube geometry, Am. Inst. Chem. Eng. Symp. Series C 81 (245) (1985) 122–128. [14] K.K. Robinsion, D.E. Briggs, Pressure drop of air flowing across triangular pitch banks of finned tubes, Chem. Eng. Prog. Symp. Series 62 (64) (1966) 177–184. [15] A.F. Du Preez, D.G. Kröger, The effect of the heat exchanger arrangement and wind-break walls on the performance of natural draft dry-cooling towers subjected to cross-winds, J. Wind Eng. Ind. Aerodyn. 58 (3) (1995) 293–303. [16] H.C.R. Reuter, D.G. Kröger, Computational Fluid Dynamics Analysis of Cooling Tower Inlets, American Society of Mechanical Engineers, New York, 2011, pp. 81104. [17] L.J. Yang, X.Z. Du, Y.P. Yang, Space characteristics of the thermal performance for air-cooled condensers at ambient winds, Int. J. Heat Mass Transfer 54 (15) (2011) 3109–3119. [18] ANSYS Inc., FLUENT User’s Guide, ANSYS, Inc., USA, 2011. [19] P. Arora, A.K. Saha, Three-dimensional numerical study of flow and species transport in an elevated jet in crossflow, Int. J. Heat Mass Transfer 54 (1–3) (2011) 92–105. [20] F.K. Moore, Cold inflow and its implications for dry tower design, in: Second Conference on Waste Heat Management and Utilization, Miami, 1978.
Contents lists available at SciVerse ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Windbreak walls reverse the negative effect of crosswind in short natural draft dry cooling towers into a performance enhancement Y. Lu ⇑, Z. Guan, H. Gurgenci, Z. Zou Queensland Geothermal Energy Centre of Excellence, Qld 4072, Australia School of Mechanical and Mining Engineering, The University of Queensland, Qld 4072, Australia
a r t i c l e
i n f o
a b s t r a c t
Article history: Received 20 September 2012 Received in revised form 14 December 2012 Accepted 28 March 2013 Available online 27 April 2013
Crosswind effect on the cooling performance of large natural draft dry cooling towers (NDDCT) has been verified to be unfavourable by many researchers. Small size natural draft cooling towers (height <30 m) proposed for geothermal and other renewable power plants are expected to be more negatively affected. CFD modelling has been carried out to numerically analyse the heat transfer performance of a 15 m-high small NDDCT under different crosswind speeds. Simulations show that, at certain crosswind speeds, the crosswind significantly degrades the cooling performance. However, the negative effect of the crosswind can be turned into positive in small natural draft cooling towers by introducing windbreak walls that guide the air mobilised by crosswind through the heat exchangers. When windbreak walls are used, the results show that the tower performance improves with increased crosswind velocity. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: Crosswind Natural draft cooling tower Windbreak wall CFD modelling Numerical simulation
1. Introduction Natural draft dry cooling towers (NDDCT) are widely used in Rankine cycle power plants in arid areas around the world. Geothermal and solar thermal power plants are more likely located in arid areas and are likely to use forced- or natural-draft air cooling. In a natural draft dry cooling tower, no fans are required. The flow of air through the bundles of heat exchangers is by means of buoyancy effects. Buoyancy occurs due to a difference in air density between the inside and outside of the tower resulting from the temperature difference. The greater the temperature difference and the height of the tower structure, the greater the buoyancy force. In natural draft cooling tower design, both the aerodynamic balance and thermodynamic balance should be satisfied at the same time which can be expressed as follows [1]:
DP ðqao qai ÞgðHt Hhx Þ ¼
X
K resist
q
a
v2
2
Q ¼ ma cpa ðT ao T ai Þ ¼ ml C pl ðT li T lo Þ ¼ hu AF T DT lm
ð1Þ ð2Þ
The first equation means the total pressure drop over various components of the tower must be balanced by the buoyancy force. The Eq. (2) states that the heat transferred into the air is equal to the ⇑ Corresponding author at: Queensland Geothermal Energy Centre of Excellence, Qld 4072, Australia. Tel.: +61 7 336 542 48. E-mail address: [email protected] (Y. Lu). 0017-9310/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2013.03.075
heat extracted from the cooling liquid (water) and this heat is transferred through the heat exchangers. The above equations for NDDCT design and selection do not include the crosswind effects. The effect of the crosswind is common and seen during operations of both wet and dry cooling towers in thermal power plants. Early studies on the crosswind influence on natural draft cooling towers were focused on either experimental methods such as full scale tower measurements [2] or laboratory tests [3–5]. However, numerical analysis (CFD) became the preferred method since the last decade [6,7]. Both the experimental and the numerical studies in the past have discovered that crosswind has a negative influence on the NDDCT cooling performance. For instance, a study showed that the heat transfer rate decreased by more than 30% at crosswind velocities above 10 m/s [7]. Wind break walls, by using either experimental [8] or numerical method [9,10], were found to improve the thermal performance of natural draft cooling towers under windy conditions. However, all the studies above focused on either Heller-type or surface-condenser-type indirect large natural draft cooling towers with heights above 100 m. The Queensland Geothermal Energy Centre of Excellence (QGECE) has been developing small natural draft cooling towers (NDDCT) for geothermal and solar thermal power plants [11]. It is expected that crosswind will have significant negative effect on the performance of small NDDCT. To turn this negative effect into positive, a novel windbreak arrangement has been analysed and the results are reported in this paper.
163
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Nomenclature A area (m2) a constant Aa, Ar air-side area, fin-root area, respectively (m2) Ac surface area of numerical cell (m2) C inertial resistance factor C1, C1e, C2, C3e constants in turbulence equations Cp specific heat (J Kg1 K1) Cl coefficient in turbulent viscosity dr outer diameter of tube (m) F source term for momentum equations FT temperature correction factor H height, elevation (m) h convective heat transfer coefficient (W m2 K1) Gb turbulent kinetic energy source term due to buoyancy Gk turbulent kinetic energy source term due to mean velocity gradients Kresist pressure loss coefficient K, Ke, Kt laminar, effective, turbulent thermal conductivity, respectively (W m1 K1) k turbulent kinetic energy (m2 s2) m mass flow rate (kg/s) nr number of tube rows in a heat exchanger bundle p pressure (Pa) Pr, Prt laminar, turbulent Prandtl number, respectively pt, pd tube transversal pitch, tube diagonal pitch, respectively (m) Q heat transfer rate (W) Qr, Qrw heat transfer rate of radiator without or with windbreak wall (W) q heat flux (W m2) Rec Reynolds number based on minimum free flow area S modulus of the mean rate-of-strain tensor S/ volumetric source term for variable quantity / T temperature (K) DTlm logarithmic mean temperature difference (K)
2. CFD model
U, V, W Vc v x, y, z
velocity components in x-, y-, z-direction (m s1) numerical cell volume (m3) velocity scalar (m s1) Cartesian co-ordinates
Greek letters a permeability (m2) b bulk thermal expansion coefficient (K1) e turbulent kinetic energy dissipation (m2 s3) / scalar quantity (u, v, w, T, k, e, . . .) C/ diffusion coefficient for variable quantity / q, q density, mean density (kg m3) l, le, lt laminar, effective, turbulent viscosity, respectively (kg m1 s1) rk, re turbulent Prandtl number for k and e, respectively s stress tensor Vectors ~ v
velocity
Subscripts a, l air side, liquid (water) side b base of tower cw cross wind e effective hx heat exchanger i, o inside or inlet, outside or outlet r radiator t tower u overall w windbreak wall 0, ref reference value
(4) are proposed by Ganguli et al. [13] and Robinson et al. [14], respectively.
2.1. Tower size scaling The first stage of the study was a theoretical analysis to find the possible smallest size of NDDCT for a particular small renewable power generator under given windless conditions using a onedimensional (1D) mathematical model based on Eqs. (1) and (2) [12]. The geometry of tower is assumed to have cylinder rather than hyperbolic shape. While hyperbolic shape provides better structure strength for reinforced concrete towers and has slightly lower air flow resistance, it increases the building costs especially if the tower is built with steel structure as appropriate for small cooling towers. By contrast, a cylindrically shaped steel tower offers more economic for building and remote area installation though it sacrifices the overall cooling performance a little bit. In fact, the wall profile has very limited influence on the heat transfer and the air flow resistance inside the tower. According to the above two equations Eqs. (1) and (2), contracting and diffusing of the air flow area can cause a difference in the pressure loss, but this change is negligible compared to the total resistance, the largest part of which comes from heat exchangers. The heat transfer and flow characteristics of the heat exchangers in 1D model are based on standard correlations developed for finned tube bundles. This study considers a specific bundle design with three staggered circular-finned tube rows, whose air-side heat transfer correlation Eq. (3) and pressure drop correlation Eq.
0:333 ha ¼ 0:38Re0:6 ðAa =Ar Þ0:15 K=dr c Pr
K resist ¼ 18:93nr Re0:316 c
ð3Þ
0:927 0:515 pt pt dr pd
ð4Þ
It was found that an NDDCT with an internal horizontal heat exchanger placement could be as small as 15 m in height and 12 m Table 1 Proposed design conditions.
*
Design point
Value
Tower aspect ratio* Total thermal efficiency, % Water inlet temperature, °C Ambient air dry bulb temperature, °C Water mass flow rate, kg/s Heat exchanger tube diameter, mm Tube length, m Fin diameter, mm Transversal tube pitch, mm Total heat exchanger front area, m2 Heat exchanger tubes Bundle arrangement
1.25 15 40 20 16 21 8 51 60.6 73.7 3 Rows, 3 passes Horizontal inside the tower
Defined as the ratio of total height to base diameter.
164
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
in base diameter for a plant with net power generation of 100 kW under the proposed design conditions in Table 1. At the design point as defined in Table 1, the heat exchangers can reject around 578 kW heat at no crosswind condition. However, with crosswind, the heat rejected by the heat exchangers will be different. CFD models of the NDDCT of Table 1 have been built in the commercial CFD software ANSYS FLUENT to study its heat rejection performance at different crosswind speeds, with and without a windbreak. 2.2. Model geometry The geometry of tower in the CFD model, including tower support, is a cylinder with the size given above to reflect the practicality of steel construction for small towers. In the model with windbreak, three solid, zero-thickness walls are placed in the tower bottom with equal separating angles of 120°. The heights of the walls are equal to the inlet height of tower, similar to the wall proposed by Du Preez and Kröger [15]. The computational domain (to simulate outside ambient air) is also of cylindrical shape with 90 m in height and 72 m in radius. Past CFD studies show that the distances from tower to domain boundaries affect the numerical results to a certain extent. So this CFD model uses 12 times the tower diameter for the domain diameter and 6 times for the domain height.
without thickness, whose heat transfer rate is calculated by Eq. (6) [18].
q ¼ hðT r T ao Þ
where the heat transfer coefficient, h, is a function of the heat exchanger characteristic parameters and the air inlet velocity, vai, normal to radiator. Tr is the radiator temperature and Tao is the air temperature downstream of the radiator [17]. For air flow pressure drop, the radiator model can simulate resistance to air flow in the direction normal to radiator face. However, it does not provide resistance in other two directions, i.e. velocity components parallel to radiator face. This will cause overestimation of the possibility of vertices occurring near the radiator, since real structure of fin tube heat exchanger bundles can prevent horizontal air flow, allowing air flow through heat exchanger only vertically. Therefore a porous media model is added to represent the pressure loss within the heat exchanger, leaving the radiator model to represent heat transfer only. Porous media model offers an ideal approximation for modelling the heat exchangers in this context where the internal structure detail is not of concern but a distributed resistance is important. A porous media zone introduces an additional source term in momentum equation of each i-axis [18]:
le 1 v þ C qv 2i 2 a i
Fi ¼ 2.3. Boundary conditions As shown in Fig. 1, the wall boundaries are set as slip walls because the boundary layers are sufficiently thin so cannot influence flow separations at tower inlet and outlet [16]. The velocity inlet boundary is defined at the windward side of the domain. The velocity profile is applied in this boundary defined by Eq. (5), where a is recommended as 0.2 [17].
v cw ¼ v ref
y yref
!a ð5Þ
The pressure outlet boundary condition is applied in leeward side as well as on the top of domain. The temperatures on both the inlet and outlet boundaries are set equal to the ambient temperature, i.e. the air inlet temperature of heat exchanger. Several ways of modelling heat exchangers in CFD can be found in open literature. The radiator model in FLUENT is used in this study to represent the heat exchanger bundles as a lumped face
ð6Þ
ð7Þ
where a and C are determined by the friction factor of heat exchangers in the 1D model. By this modelling strategy, the vertical air flow can be guaranteed by setting the resistances in other two directions much larger than that in vertical direction (y axis). The tower support is set to the porous jump boundary which is simplified as a cylinder face with same pressure resistance coefficients corresponding to those of supporting structures in real towers. 2.4. Governing equations Since the air velocity in this study is far below 0.3 Mach, the incompressible air model with constant density is assumed. A buoyancy generating term is introduced in vertical component of the momentum equation using the Boussinesq’s approximation to reflect the buoyancy effect caused by the density difference.
Table 2 Summary of governing equations. Equation
/
C/
S/
Continuity x Momentum
1 U
0
0
V
le le
@ ~ @p @x þ r ðle @x v Þ þ F x
y Momentum z Momentum
W
le
Energy
T
Ke Cp
@ ~ @p @z þ r ðle @z v Þ þ F z
k
k
e**
e
l þ rlkt l þ rlet
where
le = l + lt, lt ¼ qC l ke , K e ¼ K þ K t ; K t ¼ CPrp lt t 2
l 1 Gk = ltS2, Gb ¼ bg Prtt @T @y , b ¼ T 0 P
0 h i Ske C 1 ¼ max 0:43; Skþ5 , C1e = 1.44, C3e = tanh(V/U), C2 = 1.92, rk = 1.0, re = 1.44,
e
Pr = 0.74, Prt = 0.85, T0 = 293.15 **
The realizable k–e turbulence model is used in this modelling.
@ ~ @p @y þ r ðle @y v Þ q0 bðT T 0 Þg þ F y
1 Cp
qAc Vc
Gk + Gb qe 2
e qC 1 Se qC 2 kþepffiffiffiffi v e þ C 1e C 3e Gb k
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
72
165
Unit: m
Pressure outlet
Velocity inlet Pressure outlet
90
Radiator 12
Porous media zone Wall
15 3 y Porous jump
Wall
z x
Fig. 1. Geometry of 3D models with or without walls.
The model is simulated by solving a series of conservation equations of physical quantities, whose general term is expressed as:
div ðq~ v /Þ ¼ div ðC/ grad/Þ þ S/
ð8Þ
The expressions of /, C/ and S/ for the above equation are shown in Table 2. The source terms Fx, Fy, Fz in each momentum equation refer to c porous media resistance defined as Eq. (7), while qA in energy Vc source represents heat transfer of heat exchangers with q calculated by Eq. (6).
Fig. 2. Structured meshes in tower body and ground.
The implicit partial differential governing equations are discretised with the second order of upwind discretization scheme and are decoupled using pressure-based segregated algorithms: SIMPLE [18]. The convergence criterion is that the scaled residuals for all variables (except energy) drop to the order of 105 and the monitored variables remain stable when iterating. The calculation process is iterated for more than 15,000 steps and converged results can be obtained.
Fig. 3. 3D streamlines inside and under cooling tower when crosswind speed is (a) 0 m/s, (b) 0.5 m/s, (c) 2 m/s, (d) 4 m/s, (e) 6 m/s and (f) 8 m/s.
166
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Fig. 4. The temperature contour at the mid-xy plane when wind speed is (a) 0 m/s, (b) 0.5 m/s, (c) 2 m/s, (d) 4 m/s, (e) 6 m/s and (f) 8 m/s.
cell layer near walls is 0.08 m while the maximum cell size inside the tower is 0.15 m. And a finer mesh is used in the regions of heat exchanger and tower outlet where cell size is less than 0.1 m. Fig. 2 shows the final mesh used in the CFD model. 3. Results and discussions
Fig. 5. Velocity vectors at mid-xy plane when crosswind speed = 6 m/s.
Grid-independence has been tested by analysing the no-crosswind cases at different mesh sizes. When the cell quantity is over 3 million, the conserved variables monitored change by less than 0.5% compared with results in mathematics model. Final model uses about 3,750,000 structured prism cells. The thickness of first
The small tower performance was first simulated under nowind conditions. This was followed by the cases with and without windbreak walls under various crosswind speeds. The numerical results in no-wind case matches the calculation results of 1D model quite well, which itself has been validated against the industrial cooling tower data in [1]. The differences between the 1D and 3D models in outlet air mass flow rate and mean temperature through the heat exchangers are about 0.8% and 1.5%, respectively. This 1Dverified CFD model is then used for modelling at various crosswinds. When crosswinds are introduced, the crosswind speed was varied from zero (no wind) to maximum of 18 m/s at the reference elevation of 10m as defined in Eq. (5). 3.1. Crosswind influence on the tower performance without windbreak wall With horizontally-blowing wind, the airflow inside the cooling tower is not only driven by the buoyancy force, but also is
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Fig. 6. Pressure contours at the heat exchanger inlet face when crosswind speed is (left) 4 m/s and (right) 8 m/s showing the negative pressure zone.
Fig. 7. Velocity vectors at mid-xy plane for case study.
Fig. 8. Comparisons of different crosswinds effects between models with (maw, Qrw, Qw) and without walls (ma, Qr, Q).
167
168
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Fig. 9. Air streamlines passing through the heat exchanger in direction of (left) backward and (right) forward for wind-wall case at crosswind speed of 2 m/s.
Fig. 10. Pressure contours at heat exchanger inlet face with the windbreak walls installed when crosswind speed is (left) 4 m/s and (right) 8 m/s.
subjected to the outside crosswind. Fig. 3 shows the airflow 3D streamlines inside the tower as well as at the bottom of the cooling tower while Fig. 4 is the air temperature contours at the central vertical cross section of the tower at various crosswind speeds. At low wind speeds such as 0.5 m/s, the air flow inside the tower is near uniform, and the ambient air enters into the tower bottom only through the windward side. As the speed of crosswind increases, two vertices form due to the different mechanisms applying in each region and penetration or ‘‘downwash’’ [19] of hot air in the downstream of the tower outlet is observed. The upper vortex is caused by the crosswind forming a high speed zone acting like a lid above tower outlet. The warm air exiting from the tower at a much slower speed (around 0.4 m/s) cannot break through the wind lid. In fact, the warm air is quickly cooled down near the tower exit by the cross wind and some cooler air sinks back into the cooling tower (Fig. 5). This phenomenon was referred to as the cold inflow in large industrial towers, which can be usually assessed with the value of the Froude number [20]. The result is the reduction of the effective draft height of the cooling tower. While at the tower bottom, hot air inside the tower at the windward side is sucked down because of the negative pressure underneath the heat exchangers (Fig. 6) caused by the
crosswind speed. This part of air re-enters into the heat exchanger bundles at leeward side, forming another hot air circulation. The lower vortex largely decreases effective transfer area of heat exchanger bundles and makes heat transfer in this region rather complicated. Although the emergence of these two air vertices inside tower is attributed to different mechanisms, the suction effect under the heat exchangers is believed to play a dominant role. This has been proved by a complementary case study. In this case study, the space outside the tower is divided into two parts by virtual horizontal faces. And in each simulation, crosswind flow is applied in either upper or lower part of computation domain to study their effect separately. The velocity vectors at the cross section of the mid-xy plan (Fig. 7(a)) show that in the case when crosswind is applied on the tower outlet (upper part) only, the inside tower air flow field does not change much compared with no-crosswind case. However, when the crosswind is applied only at the tower inlet (lower part), air flow reverses its direction (Fig. 7(b)), for the zone below the heat exchangers now has lower pressure. The effect of crosswind on the NDDCT cooling performance is better understood by separately examining the two cooling
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
components: the heat taken away by the air updraft leaving through the tower outlet at the top Q, and the heat taken away, Q0 , by the air that leaves through the bottom part of the tower after circulating through the heat exchangers driven by the lower vortex explained earlier. The sum of these two components is equal to the total heat taken away from the heat exchangers (radiator), Qr, which is independently computed by FLUENT as the radiator heat transfer, namely Qr = Q + Q0 . The first component, i.e. the heat Q dissipated through cooling tower outlet is calculated using the second term in Eq. (2). Here the air mass flow rate ma is the net value of the air flowing through the radiator. It is extracted from the numerical results as the upward mass flow rate minus the downward one in the case when inverse air flow occurs at heat exchangers. The heat exchanger exit temperature Tao is calculated by averaging the numerically calculated air temperatures at the face of tower outlet. The heat exchanger inlet temperature Tai is assumed to be equal to the ambient air temperature. The second cooling component, Q0 , is calculated by subtracting Q from Qr. When the air flow is unidirectional inside the tower, Q0 is zero and thus Qr = Q. Solid lines in Fig. 8 indicate the change of ma, Q and Qr under different crosswind conditions. The heat dissipated through the tower outlet Q keeps declining along with the rise of crosswind speed while the mass flow rate ma decreases and remains nearly constant after 10 m/s. However, in this small tower, the crosswind does not always exert a negative effect on the cooling tower performance in terms of total heat transferred by radiator Qr. Qr reaches its lowest point at a crosswind speed around 5 m/s and then increases with increase of crosswind speed. This means under high-speed (>5 m/s) crosswind conditions, Q0 becomes a dominant component in the overall heat transfer rate. This phenomenon is seldom seen in large NDDCT installations since a tall tower provides a relatively large draft force for hot air and normal crosswind speeds are not high enough to cause inverse flow at heat exchangers against this relatively large updraft. With the existence of crosswind, the total transferred heat Qr could decrease by 37% compared with no-crosswind condition, which leads to a significant drop in net power generation at certain cross wind speeds. 3.2. Crosswind influence on the tower with windbreak walls Three simple but very effective windbreak walls have been introduced aiming to improve the performance of small NDDCTs under crosswind. The function of these walls is to divert crosswind flow through the heat exchangers rather than flowing across the bottom of the tower. The walls are installed in the bottom of the tower with 120 degrees separation. When there is no crosswind, the cooling air enters into the tower freely without any obstruction from the walls. If crosswind exists, the walls stop the crosswind flowing across the bottom, change the direction of the crosswind, and force it flow through heat exchanger. Since more air flows through the heat exchanger, it improves the performance of the tower. The preliminary simulations found that the total cooling performance is substantially enhanced regardless of the wind direction but the level of the improvement is dependent on the wind directions. In this paper, only one simple case is reported that the coming crosswind is facing right against sector A which is symmetric about x axis as seen in Fig. 9. The results of simulation with windbreak walls are plotted using dashed lines in Fig. 8. The subscript ‘‘w’’ is used to refer to the results obtained with windbreak walls. There is a significant increase in total radiator heat transfer, Qrw, as well as in air mass flow rate maw with increasing crosswind speed. This is directly related to the change in air flow field in the tower bottom. In sector A, crosswind is stopped by the walls so that nearly all the air flows upward through the
169
heat exchangers, while in sectors B and C, the draft effect of cooling tower remains sufficient to drive outside air flows into it through the heat exchangers continually although vortices occur in the wake area of walls. The increase of upward mass flow rate inside the tower can be attributed to the air pressure under the heat exchangers being higher than that without the windbreak walls at the same wind speed (Figs. 6 and 10). And among the three sectors, the pressure in sector A is remarkably higher. Since there is no downward air flow through the heat exchanger, the heat dissipated through tower outlet Qw equals to Qrw as expected. The addition of a wind break wall significantly improves the total thermal performance of a cooling tower. For the case study considered, a 30% increase in total heat transfer is estimated when the crosswind speed is around 5 m/s, a speed that would produce a substantial reduction of tower performance without a windbreak. This clearly shows the benefit of using wind break walls.
4. Conclusions Crosswind would stop small natural draft cooling tower functioning if no windbreak walls are used at high crosswind areas. This paper demonstrates that the negative effect of the crosswind can be reversed into a performance enhancement in short natural draft dry cooling towers. CFD modelling has been done to quantify the crosswind effects on cooling performance of small size NDDCT with and without windbreak walls. A new approach has been introduced to simulate the heat transfer and pressure drop of the heat exchanger in the cooling tower model: a combination of a FLUENT ‘‘radiator’’ element to represent the convective heat transfer term and a porous media zone to represent the heat exchanger pressure drop. Simulations under different crosswind speeds indicate that the tower heat transfer significantly suffers from crosswind when no wind break wall is installed. On the other hand, with wind break walls, the negative effect of the crosswind is turned into an advantage and both the total heat transfer and mass flow rate are enhanced. The results are internally consistent and the numerical predictions under no crosswind conditions are in agreement with the correlations developed using industrial data. Experiments are being planned to test the fundamental assumptions of the representation used in this model so as to produce a more rigorous experimental validation of the numerical method. The following observations apply based on the present results: 1. Without wind break walls, the air flow field inside the tower is disturbed by the horizontally-flowing crosswind forming two major vortices leading to inverse flow through the heat exchangers. The main reason is the suction effect of the wind passing underneath the heat exchangers. 2. With such inverse air flow, the total heat transfer Qr between the heat exchangers and the air is no longer unidirectional and can be dissipated through both the tower top (Q) and the tower bottom (Q0 ) at the same time. And at certain wind speed, larger part of heat is dissipated underneath the heat exchangers, which unexpectedly increases the total cooling performance of heat exchangers. 3. When windbreak walls are employed, the cooling performance is consistently improved with increasing crosswind speeds in the considered wind direction. The results show that the use of natural draft dry cooling towers is feasible with small renewable power plants if wind break walls are employed. This is a significant conclusion and will assist the commercial viability of remote area renewable power generators.
170
Y. Lu et al. / International Journal of Heat and Mass Transfer 63 (2013) 162–170
Appendix A. Fig. 1–10. References [1] D.G. Kröger, Air-Cooled Heat Exchangers and Cooling Towers, II, Pennwell Corp, Tulsa, 2004. pp. 38–91. [2] A.F. Du Preez, D.G. Kröger, Effect of wind on performance of a dry-cooling tower, Heat Recovery Syst. CHP 13 (2) (1993) 139–146. [3] M. Gao, F.Z.H. Sun, Research on the Effect of Cross-wind to Temperature Difference and Efficiency of Natural Draft Counter flow Wet Cooling Tower, Springer Berlin Heidelberg, Berlin, Heidelberg, 2007. pp. 513–517. [4] Q.D. Wei, B.Y. Zhang, K.Q. Liu, X.D. Du, X.Z. Meng, A study of the unfavorable effects of wind on the cooling efficiency of dry cooling-towers, J. Wind Eng. Ind. Aerodyn. 54 (1995) 633–643. [5] D.D. Derksen, T.J. Bender, D.J. Bergstrom, K.S. Rezkallah, A study on the effects of wind on the air intake flow rate of a cooling tower: Part 1. Wind tunnel study, J. Wind Eng. Ind. Aerodyn. 64 (1) (1996) 47–59. [6] G. Ge, F. Xiao, S. Wang, L. Pu, Effects of discharge recirculation in cooling towers on energy efficiency and visible plume potential of chilling plants, Appl. Therm. Eng. 39 (Journal Article) (2012) 37–44. [7] R. Al-Waked, M. Behnia, The performance of natural draft dry cooling towers under crosswind: CFD study, Int. J. Energy Res. 28 (2) (2004) 147–161. [8] Y. Chen, F. Sun, H. Wang, N. Mu, M. Gao, Experimental research of the cross walls effect on the thermal performance of wet cooling towers under crosswind conditions, Appl. Therm. Eng. 31 (17) (2011) 4007–4013. [9] R. Al-Waked, M. Behnia, The effect of windbreak walls on the thermal performance of natural draft dry cooling towers, Heat Transfer Eng. 26 (8) (2005) 50–62.
[10] Z. Zhai, S. Fu, Improving cooling efficiency of dry-cooling towers under crosswind conditions by using wind-break methods, Appl. Therm. Eng. 26 (10) (2005) 1008–1017. [11] H. Gurgenci, Z. Guan, Dirigible natural draft cooling tower in geothermal and solar power plant applications, in: Proceedings of 14th IAHR International Conference on Cooling Tower and Air-Cooled Heat Exchangers, Stellenbosch, 2009. [12] Y. Lu, Z. Guan, H. Gurgenci, Development of small natural draft dry cooling towers for geothermal power plants, in: The 15th IAHR Cooling Tower and Aircooled Heat Exchanger Conference, Beijing, 2011, pp. 308–316. [13] A. Ganguli, S.S. Tung, J. Taborek, Parametric study of air-cooled heat exchanger finned tube geometry, Am. Inst. Chem. Eng. Symp. Series C 81 (245) (1985) 122–128. [14] K.K. Robinsion, D.E. Briggs, Pressure drop of air flowing across triangular pitch banks of finned tubes, Chem. Eng. Prog. Symp. Series 62 (64) (1966) 177–184. [15] A.F. Du Preez, D.G. Kröger, The effect of the heat exchanger arrangement and wind-break walls on the performance of natural draft dry-cooling towers subjected to cross-winds, J. Wind Eng. Ind. Aerodyn. 58 (3) (1995) 293–303. [16] H.C.R. Reuter, D.G. Kröger, Computational Fluid Dynamics Analysis of Cooling Tower Inlets, American Society of Mechanical Engineers, New York, 2011, pp. 81104. [17] L.J. Yang, X.Z. Du, Y.P. Yang, Space characteristics of the thermal performance for air-cooled condensers at ambient winds, Int. J. Heat Mass Transfer 54 (15) (2011) 3109–3119. [18] ANSYS Inc., FLUENT User’s Guide, ANSYS, Inc., USA, 2011. [19] P. Arora, A.K. Saha, Three-dimensional numerical study of flow and species transport in an elevated jet in crossflow, Int. J. Heat Mass Transfer 54 (1–3) (2011) 92–105. [20] F.K. Moore, Cold inflow and its implications for dry tower design, in: Second Conference on Waste Heat Management and Utilization, Miami, 1978.