A performance enhancement of a natural draft dry cooling tower in crosswind via inlet flow field reconstruction

Energy & Buildings 164 (2018) 121–130

Contents lists available at ScienceDirect

Energy & Buildings journal homepage: www.elsevier.com/locate/enbuild

A performance enhancement of a natural draft dry cooling tower in crosswind via inlet flow field reconstruction Weiliang Wang∗, Junfu Lyu, Hai Zhang, Qing Liu, Guangxi Yue, Weidou Ni Key Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, China

a r t i c l e

i n f o

Article history: Received 11 September 2017 Revised 24 December 2017 Accepted 2 January 2018

Keywords: NDDCT Crosswind Degrading mechanism Flow field reconstruction Labyrinth Performance enhancement

a b s t r a c t With the wide utilization of a natural draft dry cooling tower (NDDCT) in power generation in arid areas, the degradation of its performance under crosswind conditions is increasingly concerned. Based on the influencing mechanisms of the crosswind, the paper reconstructs the destructed inlet flow field with a labyrinth structure. The effect of the labyrinth structure is firstly assessed by means of a verified computational fluid dynamics (CFD) model. Then, on the basis of CFD results, the labyrinth structure is further optimised by adopting a quantification method using a flow loss factor (FLF). Numerical results revealed that the proposed flow field reconstruction approach could increase the ventilation rate of a NDDCT by ∼62% under high speed crosswind condition, correspondently reducing the overall coal consumption by 23,10 0–33,0 0 0 t annually for a 660 MW coal-fired unit. Moreover, the negative effect of the crosswind on the performance of a NDDCT could be reversed to a positive one. The numerical results are well validated by the modelling experiments conducted in a wind tunnel. © 2018 Elsevier B.V. All rights reserved.

1. Introduction A traditional wet cooling power plant is usually a large fresh water consumer, consuming tens of million tons of water per year in waste heat rejection [1]. Building a wet cooling power plant is unpractical in arid countries and regions [2]. Thus, indirect dry cooling technology (IDCT), which adopts surface-air-cooled heat exchangers was proposed to resolve the problem [3]. Nowadays, IDCT is increasingly used not only in coal-fired power plants, but also in concentrating solar thermal power plants [5]. In addition, IDCT equipped with a natural draft dry cooling tower (NDDCT) is prevailing in low cost of operation and maintenance and long service time [4].However, as the main facility of IDCT, the performance of a NDDCT is sensitive to the ambient crosswind [6]. Highspeed crosswind may degrade the ventilation rate of a NDDCT by ∼36%, resulting an increment of ∼7.5 °C in the air temperature inside the tower [7], or an increment of ∼7 °C in cycling water [4], or a decrement of more than 25% in the heat transfer efficiency [8]. Generally, a NDDCT is consisted of a heat exchanger bundle (radiators), a plenum chamber, and an effective plume part [9]. Previous studies found that, the crosswind forms an unfavourable pressure distribution at the tower inlet [7], causing a horizontal air flow, or even a cross ventilation in the tower [10]. The inlet

∗

Corresponding author. E-mail address: [email protected] (W. Wang).

https://doi.org/10.1016/j.enbuild.2018.01.003 0378-7788/© 2018 Elsevier B.V. All rights reserved.

air flow streams from the leading and rear radiators converge and then produce complex vortices [11]. These vortices disturb the hot plume from the cooling tower. In addition, a back flow could be induced by the separation vortex at the leading edge of the tower outlet [12]. Crosswind may also squeeze the plume flow, leading to a smaller cross section of the plume and a higher flow resistance along the path line [13–15]. Along with the continuous pursuing of high efficiency, high reliability and low cost in power industry [16], a few approaches, including windbreaks [17], enclosure [18,19], and newly proposed wind collecting ducts [20], are proposed to overcome the cooling performance degrade of a NDDCT under the crosswind condition. However, there is still no distinguished approach which satisfies both the performance improvement and the construction feasibility. Recently, our study found the crosswind changes the inlet flow field, inducing mainstream vortices inside the tower, and thereby degrades the ventilation. In addition, a high speed crosswind can generate low pressure area at the NDDCT outlet to reduce the ventilation [20]. This paper presents a flow field reconstruction to the inlet flow field when the crosswind present, to overcome the performance degradation for a NDDCT. On the ground of a thorough analysis of the non-uniform flow field around a NDDCT, two labyrinth structures equipped outside the posterolateral radiators are proposed,

122

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

Nomenclature d ITD NDDCT P q T U v z

difference initial temperature difference natural draft dry cooling tower pressure (kPa) mass flow rate (kg/s) temperature (K) potential flow (kg/(s m2 )) average velocity (m/s) the vertical height m

Greek letters  differential error ρ air density (kg/m) ξ local resistance coefficient  flow resistance (1/m2 ) Superscripts ∗ total value Subscripts 0 the baseline value bottom the area inside the radiator chimney the area right inside the tower chamber f flow inlet the area prior to the inlet of the NDDCT m mass outlet the area above the outlet of the NDDCT r reference value radiator the area between the radiator fins t wind tunnel total the overall streamline field

Fig. 1. Schematic diagram of a typical NDDCT.

to reconstruct the flow field of the NDDCT and facilitate the overall ventilation. To comprehensively assess the effect of labyrinth structures, numerical and experimental studies are conducted. By using a recently proposed flow loss factor (FLF) to quantifies the effect local flow field change, the labyrinth dimensions are optimised via a developed computational fluid dynamics (CFD) model. A hot state test rig simultaneously meeting the scaling laws of Froude and Euler numbers is also adopted to verify the simulation results [20]. 2. Methods 2.1. Problem descriptions A schematic diagram of a typical NDDCT is shown in Fig. 1, where thousands of radiators are equipped evenly around the tower inlet under the expansion platform. The investigated NDDCT in this study is installed in a 660 MW coal-fired power plant in China. The basic dimensions are listed in Table 1. Under wind free condition, drafted by the pressure difference between the inside

Table 1 Dimensions of the investigated NDDCT. Items

Values (m)

Total height Thickness of the expansion platform Height of the radiators Height of the radiator support Diameter of the outlet Diameter of the throat Diameter of the radiator bundle base

170 1.5 24 2 84.47 82 146.17

Fig. 2. A CFD model of NDDCT with labyrinth structure.

and outside, the cooling air enters the tower uniformly through the vertically arranged radiators. Then the air flow is heated by the cycling water inside the radiators, maintaining a natural draft status. As the crosswind stagnates at the windward sections, it accelerates at the side sections, and converges at rearward sections [19]. A labyrinth is proposed to reconstruct the flow field outside the NDDCT inlet as shown in Fig. 2. It includes a windbreak, a labyrinth top with a certain circumferential width set above the windbreak, a labyrinth fence set vertically along the outer edge of the top, a horizontal baffle and a vertical baffle set along the back side edge of labyrinth top and labyrinth fence. Thereafter, a labyrinth cavity with an inlet and an outlet is constructed, where the inlet is among the outer edges of the windbreak, the leading edges of the

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

123

Table 2 Parameters of the investigated labyrinth structures.

Top width Vertical Baffle width

Case 1

Case 2

Case 3

Case 4

Case 5

Case 6

15° 10 m

30° 10 m

45° 10 m

15° 20 m

30° 20 m

45° 20 m

Fig. 5. The variation trend of ventilation loss along with different FLF’s.

Fig. 3. Experimental model of the wind collecting scheme.

45°, respectively, and the width of the vertical baffle is 10 m and 20 m, respectively. It is known that the heat rejection in a NDDCT equals to the exhaust heat released from the exhaust steam in the condenser. For a constant power load, the crosswind may deteriorate the heat transfer in the radiators, hence increase the ITD (initial temperature), and consequently increase the back pressure of the steam turbine. However, the NDDCT performance does not greatly affect the heat rejection. Because as the back pressure increases in a certain range, the steam rate per unit power generation increases slightly, and meanwhile the exhaust heat released in the condenser decreases slightly [21–23]. According to a case study, a 5 kPa increment of back pressure leads to only an ∼2% increment of the exhaust heat, which is negligible compared to an ∼30% variation of the NDDCT performance caused by the crosswind. Based on the analysis, the overall heat rejection inside a NDDCT is regarded as constant under different crosswind conditions. 2.2. CFD modelling

Fig. 4. System diagram of the hot state NDDCT test rig. 1. The chimney of the model, 2. radiator resistance pieces, 3. electrical heating rods, 4. variable voltage power, 5. crosswind, 6. pressure sampling ports, 7. thermocouples, 8. pitot probe, 9. movable rocker, 10. primary signal cable, 11. transmitter, 12. DC Power, 13. standard signal cable, 14. A/D converter, 15. digital signal cable and 16. computer.

top and the leading edge of the fence, and the outlet is between the radiators and the inner edges of the horizontal/vertical baffles. The windbreak has a size of 26 m (in height) × 25 m (in width). The labyrinth top is 12 m above the windbreak. The fence is 10 m apart from the right edge of the windbreak. The horizontal baffle is 10 m in width. The labyrinth top has a sector shaped configuration; while the fence is a cambered surface, whose upper edge is shared by the outer edge of the top. Considering the influence of the width of the labyrinth and the width of the vertical baffle, we set six different cases to investigate as shown in Table 2, where the width of the labyrinth (angle of the sector) is set as 15°, 30° and

Because the flow field around a NDDCT is symmetric against the crosswind, only half of the NDDCT is considered in developing the CFD model, as shown in Fig. 2. Correspondingly, the calculation domain is limited in a half-cylinder configuration. The size of the calculation domain is ten times large as the model size in each direction, with a diameter of 1200 m and a height of 1700 m. Consequently, a reasonable velocity profile of crosswind around the NDDCT could be estimated as a constant value is set at the domain inlet. The outlet boundary is set as an outflow condition. Adiabatic wall with no slip condition is set to the surfaces like the ground, the inside/outside cooling tower shells and the support and joint faces between adjacent radiators. The pressure-based solver in FLUNET with pressure-velocity coupling SIMPLEC method is adopted. And the second-order upwind differencing scheme is used to discretise the governing equations of the momentum, energy, turbulent kinetic energy and dissipation rate. As the heat rejection inside the tower is regarded as a constant, the radiator bundle is simplified as a constant heat source. The air flow is assumed to be in fully developed turbulent regime. The variation of air density is negligible. Boussinesq approximation is adopted in the momentum equation in consideration of the buoyancy force. The governing equations of the steady state,

124

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

(a) Radiator inlet pressure at Vt of 1.5 m/s

(b) Radiator outlet pressure at Vt of 1 m/s

Fig. 6. Validation of the pressure distribution for the CFD model.

(a) Baseline case

(b) Labyrinth-15°-10 m

(c) Labyrinth-30°-10 m

(d) Labyrinth-45°-10 m

Fig. 7. The reconstruction effect on inlet flow fields at 20 m/s.

buoyant, and turbulent flow with heat transfer includes continuity, momentum, energy, and turbulence modelling equations. The standard k − ε model is used to describe the turbulent flow [24]. Grid checking is done with grid numbers of 8.3 M, 12.2 M, 13.3 M and 15.7 M at the crosswind speed of 0.5 m/s and 4 m/s, respectively. It was found that the ventilation rate at the same boundary

condition varies slightly with the increase of grid number, and is only ∼0.02% when the grid number changes from 13.3 M to 15.7 M in both crosswind conditions. Then the hexahedral meshing model with a grid number of 15.7 M was developed. Detailed descriptions about the CFD model were given in the previous study [4].

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

(a) Baseline case

(b) Labyrinth-15°-10 m

(c) Labyrinth-30°-10 m

(d) Labyrinth-45°-10 m

125

Fig. 8. The reconstruction effect on inlet pressure fields.

2.3. Experiment modelling

2.4. Quantification modelling

The windward view of the experimental model of the NDDCT with labyrinth structures is exhibited in Fig. 3. The model is scaled down by 200 times from the designed dimensions. The NDDCT model, mainly consisting of a chimney, resistance bundle (cooling deltas), heating rods, is built up according to the scaling law and placed in a wind tunnel. For the benefit of monitoring and processing, most of the structures are made of acrylic glass. The resistance bundle is composed of a series of parallel arranged zigzag iron pieces to mimic the resistance characteristic of the radiators. Three tuneable electrical heating rods are circumferentially mounted onto the inner side of the resistance pieces to mimic the heat rejection from the radiators. The schematic diagram of the experimental system is shown in Fig. 4, where a scale of 1/200 NDDCT model, a wind tunnel and a measurement unit are included. The wind tunnel can supply precisely controlled crosswind to mimic the specific environmental crosswind.

As investigated previously [20], energy conservation is expressed by Bernoulli equation [25] as Eq. (1), where P∗ represents the total pressure including the dynamic pressure and altitude pressure, a flow resistance f and a potential flow Uf defined by Eqs. (2) and (3). Consequently, an inverse proportional relationship between the flow resistance and mass flow rate is obtained as expressed by Eq. (4). Using a differential method, a FLF is deducted as Eq. (5) to depict the effect of the pressure loss on the overall dimensionless mass flow rate. The subscripts – r and – t denote the reference and the total conditions, respectively.

P1∗ = P2∗ + ξ ·

 f = Uf =



1 ρ2 v22 2

ξ

S 2 ρ P ∗

Uf = qm · f

(1)

(2) (3) (4)

126

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

(a) Baseline case

(b) Labyrinth-15°-10 m

(c) Labyrinth-30°-10 m

(d) Labyrinth-45°-10 m

Fig. 9. The reconstruction effect on streamlines.

FLF =

f f−r−total

= FLF0 +

qm−r−total qm

Consequently, FLF is adopted to quantitatively assess the effect of the flow change in each region on the overall ventilation rate.





d ( qm ) P ∗ d ( P ∗ ) − Pr∗−total 2P∗ qm

3. Results and discussion

 (5)

By choosing a far field surface above the ground as the inlet pressure surface of the controlled volume, and another far field surface above the outlet as the outlet pressure surface, the total FLF is calculated according to Eq. (5). The baseline NDDCT performance is calculated at conditions of crosswind range from 0 to 30 m/s. The calculated total FLF and ventilation loss of the baseline NDDCT under different crosswind conditions are exhibited in Fig. 5, where the ventilation loss refers to the reduced mass flow rate compared to crosswind free condition. The variation trend of the FLF-total shows a good agreement with that of the ventilation loss, i.e. FLF is proportional to the correspondent ventilation loss.

3.1. Validation of the CFD model Fig. 6 shows a comparison of the pressure distributions at the inlet and outlet of the radiators between the experiment and simulation. From Fig. 6(a), we can infer that the crosswind stagnates at the inlet centre of the windward radiator bundle (0°/360°), forming a high pressure zone. Then it accelerates at the side sections and induces flow separation at the leeward side, resulting in correspondent low pressure zones. These results are consistent to the finding in previous studies [26]. Fig. 6(b) shows that the circumferential pressure distribution at radiator outlet is relatively even, only a slight increment at the leeward section. This can be explained as the result of asymmetric air intake between the windward and the leeward.

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

127

The measured pressure is in the range of 0.1 Pa–1 Pa. Such a low pressure is hard to be detected by an ordinary pressure sensor, resulting few experimental results reported in this issue. This paper presents a method adopting a high precision micro pressure transmitter with an accuracy of ± 0.0625 Pa. From Fig. 6(a) and (b), it can be seen that the maximum errors of experimental and numerical results are up to 69% and 30%, with the average of ∼30% and ∼5%, respectively. However, the results are rather good considering the measurement accuracy and some unavoidable environment disturbance in such a low pressure measurement system. Nevertheless, the trends of calculated pressure distribution profiles are well validated by the experimental data. 3.2. Effect of the flow field reconstruction The flow fields around the NDDCT inlet at crosswind of 20 m/s of baseline and three reconstruction cases are shown in Fig. 7, where the crosswind blows from the right hand side. It can be seen from Fig. 7(a) that, after adopting the labyrinth structure, the accelerated flow on the side surface is broken up, and the tangential air flow is mostly directed to be in radial direction. That means, the uneven flow field outside the radiator is greatly improved. Hence, the air intake both in front of and inside the structure is obviously enhanced. Meanwhile, the symmetric mainstream vortices induced by the uneven air intake are greatly reduced and moved to the rear side. Fig. 7(c) and (d) shows that, as the increment of the labyrinth width, the air intake is getting circumferentially more and more uniform, and the swirling intensity of the mainstream vortices is getting weaker and weaker. The pressure fields around the NDDCT inlet at crosswind of 20 m/s for baseline and three reconstruction cases are shown in Fig. 8. For baseline case, as Fig. 8(a) shows, the pressures outside the side radiators are even inferior to the adjacent inside ones, indicating backward flows happen in this areas. However, after the flow field reconstruction, as shown if Fig. 8(b), the backward flows are eliminated immediately. Besides, the uniformity of the pressure distribution inside the radiators is enhanced obviously. From Fig. 8(c) and (d) we can also find, as the labyrinth width increases, the high pressure area outside the radiators enlarges correspondently, and the pressure distribution inside the radiators becomes more uniform. Besides, the pressure gradient inside the mainstream vortices is apparently smaller as descending of the swirling intensity. As the mainstream vortex from the inlet is found to be the main degrading factor for the crosswind [20], the streamline inside the NDDCT is also discussed. Along with the changing of velocity and pressure fields, the streamline inside the NDDCT changes greatly after the adoption of labyrinth structure. Compared to the baseline case as shown in Fig. 9(a), the mainstream vortices move to the rear side, and the controlled volumes of the vortices reduce greatly in all the other cases as shown in Fig. 9(b–d). When the width of the labyrinth structure increases to 45°, the swirling intensity of the mainstream vortex becomes very weak, and the upward mainstream is more even. 3.3. Quantification of the flow fields To discuss the flow characteristics along the streamline, six pressure surfaces are selected as shown in Fig. 10, including far field inlet surface, radiator inlet surface, radiator outlet surface, chimney inlet surface, tower outlet surface, and far field outlet surface. The FLF’s between two neighbour pressure surfaces are calculated as FLF-inlet, FLF-bottom, FLF-chimney and FLF-outlet respectively, and FLF-total represents the FLF between the far field inlet surface and the far field outlet surface. The FLF’s of each flow section are calculated and shown in Fig. 11, where Fig. 11(a)–(f) exhibit the FLF’s of Cases 1, 4, 2, 5,

Fig. 10. The locations of selected pressure surfaces.

3 and 6, respectively. In comparison to Fig. 11(g), most of FLF’s are reduced greatly. That means the labyrinth structure is effective to improve the flow field and enhancing the NDDCT ventilation. Combining Fig. 11(a), (c) and (e), we can see that at a highspeed crosswind condition, along with the increment of the labyrinth width, FLF-total decreases gradually. And the decrement of FLF-bottom and FLF-chimney contribute the most, which is consistent with the previous analysis on the flow field. When crosswind is ∼10 m/s, a V-type trend of FLF-total is found, where FLFtotal first decreases and then increases. The increment of FLF-radiator surpasses the decrement of FLFchimney in Fig. 11(e). On consideration of the overall crosswind range, Case 2 seems to be the best among these three as shown in Fig. 11(c). Combining Fig. 11(b), (d) and (f), we find that a V-type trend of FLF-total exists under high-speed crosswind conditions. And the increment of FLF-radiator caused by the FLF-inlet surpasses the decrement of FLF-chimney. Generally speaking, the cases with 20 m width vertical baffle are superior to those with 10 m width vertical baffle cases in FLF comparison. 3.4. Effect and validation of the reconstruction The ventilation rates of the NDDCT are calculated and shown in Fig. 12. In accordance with the discussion in FLF section, the ventilation rate with labyrinth increases greatly when a crosswind is present. In middle-speed crosswind conditions (∼10 m/s), Cases 2, 4 and 5 are superior to the other. While in high-speed crosswind conditions (∼20 m/s), Cases 5 and 6 have better performance in ventilation. On consideration of the overall performance and construction cost simultaneously, Case 5 with a labyrinth width of 30° and a vertical baffle width of 20 m is suggested. Compared to the baseline case, the ventilation rate increases in the investigated crosswind range by adopting the labyrinth structure. An increasing rate of ∼62% is achieved at a high-speed crosswind. Besides, the ventilation rate increases when a crosswind is present compared to the wind free condition. For example, an increasing rate of ∼10% is achieved ∼20 m/s. That means the crosswind pay a positive effect on the performance of a NDDCT.

128

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

(a) Labyrinth - 15° - 10 m

(b) Labyrinth - 15° - 20 m

(c) Labyrinth - 30° - 10 m

(d) Labyrinth - 30° - 20 m

(e) Labyrinth - 45° - 10 m

(f) Labyrinth - 45° - 20 m

(g) Baseline case Fig. 11. The FLF’s on investigated cases.

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

129

tion rate of the overall coal consumption by 23,10 0–33,0 0 0 t annually. Moreover, a positive crosswind effect on the performance of the NDDCT can be achieved over the investigated crosswind range. Acknowledgement This research is supported by State Key Lab of Power Systems (SKLD13KZ05), Special Funds of the National Natural Science Foundation of China (No. L1522032), and the Consulting Project of Chinese Academy of Engineering (No. 2015-ZCQ-06). Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.enbuild.2018.01.003. Fig. 12. The ventilation rates on investigated cases.

The experiment on the ventilation rate of the scaled model of Case 1 are conducted in wind tunnel test rig as exhibited in Fig. 4. According to the characteristic dimension and scaling law of Froude number, the characteristic velocity of the NDDCT model is fixed, where the model ventilation rate of ∼0.09 kg/s is correspondent to ∼37,500 kg/s. Then the radiator resistance and heating load are designed according to the scaling law of Euler number. Consequently, the crosswind range of 0–20 m/s is modelled to a tunnel wind range of 0–2.4 m/s based on momentum similarity. The ventilation rates of the different experimental models are measured as shown in Fig. 12 in dots. Compared to the baseline model, the ventilation rate of scaled Case 1 increases greatly in all investigated tunnel crosswind range. On high-speed tunnel crosswind condition, a ventilation enhancement is also achieved compared to the wind free condition. These experimental results confirm the numerical prediction in the variation trends and the overall ventilation enhancement. By adopting the calculation method introduced in our previous publication [4], ITD (initial temperature difference) can be obtained. The results show that ITD could be reduced to 29 °C in Case 5 at crosswind of 20 m/s, which is 10 °C less than that of the baseline case. By consulting the operating parameters, the 10 °C reduction of ITD could reduce the back pressure of the steam turbine by 4–5 kPa, correspondingly reducing the net coal consumption rate of the unit by 7–10 g/kW h. By assuming an annual operation time of 50 0 0 h in the investigated 660 MW unit, the suggested reconstruction approach could reduce the overall coal consumption by 23,10 0–33,0 0 0 t annually. 4. Conclusions Because of the deformation of the inlet flow field, the performance of a NDDCT degrades greatly under the crosswind condition. This paper presents a flow field reconstruction with labyrinth structure placed around the NDDCT inlet. A validated CFD model and a scaling experimental model are both developed. By means of the CFD model, the configuration optimisation of the labyrinth structure is conducted. In addition, on the ground of the scaling experimental model, the overall ventilation enhancement by the labyrinth structure is validated. Based on the analysis of flow field reconstruction and flow loss factor (FLF) quantification, a labyrinth structure with a width of 30° in circumference and a vertical baffle width of 20 m is suggested. According to the numerical simulation and experimental validation, the installation of such a structure could increase the ventilation rate of a NDDCT by ∼62% under high-speed crosswind condition, correspondingly reducing the net coal consump-

References [1] W. Li, J. Chai, J. Zheng, Investigation of natural draft cooling tower in China, Heat Transf. Eng. 38 (11–12) (2017) 1101–1107. [2] E. Al-Bassam, G.P. Maheshwari, A new scheme for cooling tower water conservation in arid-zone countries, Energy 36 (7) (2011) 3985–3991. [3] Y. Sun, Z. Guan, K. Hooman, A review on the performance evaluation of natural draft dry cooling towers and possible improvements via inlet air spray cooling, Renew. Sustain. Energy Rev. 79 (2017) 618–637. [4] W. Wang, H. Zhang, P. Liu, Z. Li, J. Lv, W. Ni, The cooling performance of a natural draft dry cooling tower under crosswind and an enclosure approach to cooling efficiency enhancement, Appl. Energy 186 (2017) 336–346. [5] Y. Sun, Z. Guan, H. Gurgenci, K. Hooman, X. Li, Investigation on the influence of injection direction on the spray cooling performance in natural draft dry cooling tower, Int. J. Heat Mass Transf. 110 (2017) 113–131. [6] Y. Kong, W. Wang, L. Yang, X. Du, Y. Yang, A novel natural draft dry cooling system with bilaterally arranged air-cooled heat exchanger, Int. J. Therm. Sci. 112 (2017) 318–334. [7] Q. Wei, B. Zhang, K. Liu, X. Du, X. Meng, A study of the unfavorable effects of wind on the cooling efficiency of dry cooling towers, J. Wind Eng. Ind. Aerodyn. 54 (94) (1995) 633–643. [8] M. Goodarzi, R. Keimanesh, Heat rejection enhancement in natural draft cooling tower using radiator-type windbreakers, Energy Convers. Manag. 71 (2013) 120–125. [9] C.M. Chu, M.M. Rahman, S. Kumaresan, Effect of cold inflow on chimney height of natural draft cooling towers, Nucl. Eng. Des. 249 (2012) 125–131. [10] G. Tang, M. Su, S. Fu, Numerical study on flow field of a dry-cooling tower in a cross wind, Acta Aerodyn. Sin. 15 (03) (1997) 328–336. [11] Z. Zhai, G. Tang, S. Fu, K. Zhu, Experimental and numerical study of the influence of cross wind on the performance of natural draft dry-cooling tower, Thermal Power Gener. 01 (03) (1997) 3–7. [12] T.S. Fisher, K.E. Torrance, Experiments on chimney-enhanced free convection, J. Heat Transf. 121 (1999) 603–609. [13] X. Zhang, Y. Zheng, Q. Wang, Numerical analysis of the inner and outer flow field of an air-cooling tower, J. Eng. Thermal Energy Power 15 (01) (20 0 0) 52–54. [14] H. Ma, F. Si, Y. Kong, K. Zhu, W. Yan, Wind-break walls with optimized setting angles for natural draft dry cooling tower with vertical radiators, Appl. Therm. Eng. 112 (2017) 326–339. [15] M. Goodarzi, A proposed stack configuration for dry cooling tower to improve cooling efficiency under crosswind, J. Wind Eng. Ind. Aerodyn. 98 (12) (2010) 858–863. [16] Z. Zou, H. Gong, X. Lie, X. Li, Y. Yang, Numerical investigation of the crosswind effects on the performance of a hybrid cooling-tower-solar-chimney system, Appl. Therm. Eng. 126 (2017) 661–669. [17] R. Al-Waked, M. Behnia, The effect of windbreak walls on the thermal performance of natural draft dry cooling towers, Heat Transf. Eng. 26 (8) (2005) 50–62. [18] Z. Zhao, J. Shi, Q. Wei, B. Zhang, The engineering improvement for weakening the bad effect of natural wind on the dry cooling towers, J. Appl. Sci. 16 (01) (1998) 115–123. [19] W. Wang, H. Zhang, Z. Li, J. Lv, W. Ni, Y. Li, Adoption of enclosure and windbreaks to prevent the degradation of the cooling performance for a natural draft dry cooling tower under crosswind conditions, Energy 116 (2016) 1360–1369. [20] W. Wang, The Reconfiguration of the Flow Field of a Natural Draft Dry Cooling Tower on Crosswind Condition for Cooling Enhancement [Doctor], Tsinghua University, Beijing, 2017. [21] Y. Zhao, F. Sun, Y. Li, G. Long, Z. Yang, Numerical study on the cooling performance of natural draft dry cooling tower with vertical delta radiators under constant heat load, Appl. Energy 149 (6) (2015) 225–237. [22] G. Barigozzi, A. Perdichizzi, S. Ravelli, Performance prediction and optimization of a waste-to-energy cogeneration plant with combined wet and dry cooling system, Appl. Energy 115 (2014) 65–74.

130

W. Wang et al. / Energy & Buildings 164 (2018) 121–130

[23] G. Barigozzi, A. Perdichizzi, S. Ravelli, Wet and dry cooling systems optimization applied to a modern waste-to-energy cogeneration heat and power plant, Appl. Energy 88 (4) (2011) 1366–1376. [24] M. Goodarzi, R. Ramezanpour, Alternative geometry for cylindrical natural draft cooling tower with higher cooling efficiency under crosswind condition, Energy Convers. Manag. 77 (1) (2014) 243–249.

[25] Q. Zheng, Z. Lu, Fluid Mechanics, China Machine Press, Beijing, 1979. [26] C. Huang, S. Zhao, Research on resistance performance of indirect-dry cooling tower, J. China Inst. Water Resour. Hydropower Res. 09 (03) (2011) 195–199.