International Journal of Thermal Sciences 136 (2019) 70–85

Contents lists available at ScienceDirect

International Journal of Thermal Sciences

journal homepage: www.elsevier.com/locate/ijts

Utilization of partial through-flow tower shell to cope with the excess T

cooling capacity of dry cooling tower in extremely cold days with crosswind
Huan Maa, Fengqi Sia,∗, Kangping Zhub, Junshan Wangb
a
Key Laboratory of Energy Thermal Conversion and Control of Ministry of Education, School of Energy and Environment, Southeast University, Nanjing, 210096, PR
China
b
CPI Shentou Power Generation Co.,Ltd, Shuozhou, 036800, PR China

ARTICLE INFO ABSTRACT

Keywords: During low ambient temperature periods, the huge heat transfer temperature difference would result in an excess
Natural draft dry cooling tower cooling capacity of the natural draft dry cooling tower. So low cooling water temperature and back pressure can
Excess cooling capacity be easily achieved in cold days due to the unavoidable air leakage in practical operation, and the excess cooling
Crosswind capacity is a great threat to the safe operation of the indirect dry cooling system. To cope with that problem, a
Numerical simulation
novel partial through-flow tower shell is proposed. By numerical simulation of the tower, aerodynamic fields in
Thermal performance
and around the tower were presented, and cooling water temperature drop was exported as well. It was found
that thermal performance of the tower would get enhanced by crosswind under the air leakage condition, but
that can be counteracted by the proposed partial through-flow tower shell. The through-flow tower shell is
effective to reduce air inflow rate and cooling capacity of the tower, and increase back pressure and cooling
water temperature after the coupling of the tower and condenser. Compared with the sideward or leeward
through-flow tower shell, the windward through-flow tower shell would provide a much larger feasible region
for safe operation of the system, and achieve the maximum anti-freezing result at the moderate wind speed. At
the fairly high wind speed, the sideward through-flow tower shell would bring the adverse effect, while the other
two forms are still active, although their anti-freezing effects degrade a bit in this condition.

1. Introduction temperature may appear, especially for the windward radiators under
crosswind [16], which would increase the freezing risks of the radia-
The indirect dry cooling system with natural draft dry cooling tower tors. In addition, the low outlet water temperature of the tower would
(NDDCT) has been widely used in power plants in the regions with rich reduce the condensation temperature of exhaust steam in condenser,
coal and lack of water [1,2], due to the significant advantage of saving and lead to a low back pressure. However, excessive low back pressure
water. Circulating cooling water absorbs heat from exhaust steam in the would not only have no beneficial effect on the economic operation but
condenser, and then releases the waste heat to ambient atmosphere pose a threat to safe operation of the turbine. In recent years, some
through heat exchanger bundles of the tower. Thus, ambient conditions attentions have been gradually drawn on anti-freezing work for the air-
have great impacts on the running of the tower, and abundant re- cooled heat exchanger of the NDDCT under low ambient temperatures.
searches have been carried out in the past decades to reveal the adverse As a regular approach, the louvers at the inlet of the delta-type air-
influence mechanisms of crosswind [3,4] and high ambient tempera- cooled heat exchanger are used to control the inflow air flow rate [17],
ture [5,6] on thermal performance of the NDDCT. To regain the lost and no air would flow into the tower with the louvers entirely closed in
cooling capacity, a lot of measures have been proposed for the tower theory. Nevertheless, the unavoidable air leakage in practical operation
including inlet air spray cooling [7,8], cooling water heterogeneous could not eliminate the freezing risk of heat exchanger under extremely
distribution [9,10], windbreak walls [11–15] and so on. low ambient temperatures.
The excess cooling capacity of the NDDCT under low ambient Cooling water flow rates regulation is proposed as an anti-freezing
temperatures is another concern that must be reckoned with. In ex- approach [18,19]. However, cooling water redistribution is determined
tremely cold days, the huge heat transfer temperature difference would by the structure of circulating water pipelines, and can be realized by
enlarge cooling capacity of the tower greatly, and low outlet water the adjustments of additional valves at the inlet of water pipeline of

∗
Corresponding author. School of Energy & Environment, SouthEast University, No.2 Sipailou Road, Nanjing, 210096, China.
E-mail address: [email protected] (F. Si).

https://doi.org/10.1016/j.ijthermalsci.2018.10.012
Received 12 May 2018; Received in revised form 26 September 2018; Accepted 8 October 2018
Available online 16 October 2018
1290-0729/ © 2018 Elsevier Masson SAS. All rights reserved.
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

system.
Switching off certain cooling sectors like windward sector off-
working [20] under extremely low ambient temperatures, is beneficial
to anti-freezing of the air-cooled heat exchanger, but it's the last choice
during practical operation of the tower only when louvers adjustment
doesn't work. On the one hand, during the water-discharge process of
switching off the windward sector, freezing risks of delta-type radiators
in this sector would get a rapid increase owing to the flow velocity
decrease of cooling water inside the finned tube bundles. On the other
hand, at the beginning of the water-filling process of rerunning the
sector which has been switched off, the low flow velocity of inside
cooling water would increase freezing risks of radiators in this sector as
well. In addition, other matching systems, like nitrogen-charge pro-
tection system, have to participate during the water-discharge/filling
processes. As a result, cooling sector should not easily be switched off in
practical operation of the tower.
Though anti-freezing of air-cooled heat exchanger of the NDDCT is
significant, back pressure of the system is also a great concern which
shouldn't be ignored under low ambient temperatures. The essential
work is to cope with the excess cooling capacity of the NDDCT in cold
days. Therefore, a novel partial through-flow tower shell is proposed.
Fig. 1(a) shows the current NDDCT with radiators vertically arranged
around tower base. Tower shell and the seal plate above cooling deltas
are supported by the X-shape supporters. If the taller X-shape suppor-
ters are adopted, the tower shell would be lifted, and the generated
extra space marked green in Fig. 1(b) is freed up. Ambient air can flow
through the generated space, and the inflow air rate can be controlled
with additional facilities like louvers if needed. Hence, the generated
space can be seen as the through-flow part of the entire tower shell, as
shown in Fig. 1(c). By numerical simulations, effects of the partial
through-flow tower shell on thermal performances of the tower and the
whole system were researched and analyzed. By monitoring the lowest
water temperature and back pressure, appropriate ranges of ambient
temperature and wind speed were then determined for the safe opera-
tion of the system in this paper.

2. Modeling

In a typical indirect dry cooling system shown in Fig. 2, the con-

denser has cooling area of 35,000 m2, and the material of the inside
cooling water pipes is stainless steel TP304. The outer diameter and
wall thickness of these pipes are 24 mm and 0.5 mm respectively, and
the design cooling water velocity in these pipes is 2.3 m/s. The NDDCT
of hyperbolic shape has the air inlet height of 28 m and total height of
173 m. The base, throat and top diameters of the tower are 155 m, 88 m
and 91 m respectively.

2.1. Numerical model of the NDDCT

Due to the symmetric geometry along crosswind direction, only half

tower is modeled and simulated in this paper. The delta-type radiators
(i.e. cooling deltas) around half tower are divided into 3 sectors named
as windward, sideward and leeward sectors from θD = 0° to θD = 180°,
where θD is the circumferential angle of cooling delta, as shown in
Fig. 3. Every two intersecting cooling columns (named as columns A
Fig. 1. Schematic of the NDDCT with vertical delta-type radiators, (a) the and B) with finned tube bundles inside form a cooling delta with the
current tower and its components, (b) the proposed tower, (c) through-flow louver arranged at the inlet, and according to the specified layout in
tower shell arrangement along crosswind direction. Fig. 3, circumferential angle of column A would be a little smaller than
that of column B for each cooling delta.
each cooling sector. Furthermore, cooling water redistribution achieved The through-flow part of tower shell is arranged above the seal plate
by the reported method failed to be further verified by the results ex- with the height of 1/4 heat exchanger height, as seen in Fig. 1(c). The
ported by pipeline network hydraulic calculation of circulating water through-flow tower shell is also divided into 3 parts along tower semi-
circumference, named as windward, sideward and leeward through-

71
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 2. Schematic of the indirect dry cooling system mainly composed of the NDDCT and condenser.

Fig. 3. Cooling delta distribution and cooling sector division along crosswind direction.

flow tower shell according to crosswind direction. calculated based on the state equation of ideal gas. The steady gov-
The computational domain shown in Fig. 4(a) is large enough to erning equations describing air flow and heat transfer can be expressed
eliminate the unrealistic impacts of the domain boundaries on near- as the following form:
tower air flowing. The multi-block hybrid approach is utilized for
meshing owing to the complex configurations in the computational ui
= +S +S (i = 1,2,3)
domain. Hexahedral grids with mesh interval sizes of 0.3 m and 1.8 m xi xi xi (1)
are employed for the cooling deltas and tower body respectively, as
shown in Fig. 4(b). The part between cooling deltas and tower body is where ρ is air density, ui is the velocity in xi direction, φ, Γφ are variable
meshed with successive ratio increasing grids of triangular prism, and and its diffusion coefficient. For mass, momentum, energy conservation
the outer space takes successive ratio increasing grids of hexahedral and turbulent equations, φ stands for 1, air velocity component ux (or
structure. Adjusting grid interval sizes, three grid systems with mesh uy, uz), air temperature ta, turbulent kinetic energy k and turbulent
cells of 5,304,672, 6,726,799 and 7,911,976 were generated. Mesh dissipation rate ε respectively. Sφ is the internal source term of air
independence test was performed at ambient temperature of 22.7 °C governing equation, and Sφ′ is user defined source term only inserted to
and wind speed of 1.4 m/s, with the results summarized in Table .1. the momentum and energy equations of computational cells in the
Comparing the two highest grid density solutions, the differences of cooling columns, describing momentum loss and heat absorption of
outlet water temperature and air inflow rate of the tower are only inflow air when flowing through heat exchanger.
0.02 °C and 14 kg/s, and relative errors are less than 0.04%. As a result, Due to the great advantage of low computing cost, k-ε viscous
grid number of 6,726,799 was finally adopted, weighing computing models have been widely adopted in the research of NDDCT as long as
cost against accuracy. their calculation precisions are satisfactory [22]. Among various k-ε
Air is assumed as an incompressible ideal gas, because air static models, the standard k-ε model [3,7–9,11,21,23] is mostly adopted, but
pressure variation in the computational domain is reported to be just there are also many researchers using the realizable k-ε model
1–2% of the local atmospheric pressure [21]. Then, air density is [1,10,13,16–20]. So model comparison was performed and the results

72
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 4. Schematic of computational domain, (a) the half computational domain with boundary conditions specified, (b) the generated multi-block meshes.

Table 1 between the two models at a high wind speed would be a bit larger than
Mesh independence test for the NDDCT. that under a low wind speed. That is mainly caused by the distortion of
grid number outlet water temperature (°C) air inflow rate (kg/s) aerodynamic field at tower back under a high wind speed. As wind
speed increases to the critical value, flow separation appears at tower
5,304,672 45.112 42,326 back, so turbulence at tower back would get sharply enhanced [14]. To
6,726,799 45.104 42,382 further reveal the effects of turbulence, the distribution of air inflow
7,911,976 45.102 42,396
velocities of cooling deltas was analyzed, and the comparisons are
shown in Fig. 5. The difference of velocity distributions exported by the
are summarized in Table 2. Under the three typical wind speeds, outlet two k-ε models is small and satisfactory, although it increases a bit at
water temperatures exported by the two k-ε models are very close, and the high wind speed. On the whole, the difference between the standard
so does the air inflow rate of the tower. By comparison, the error and realizable k-ε models is negligible. Compared with the standard k-ε

73
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Table 2
The comparison between the standard and realizable k-ε models.
0 m/s 4 m/s 16 m/s

standard k-ε realizable k-ε error standard k-ε realizable k-ε error standard k-ε realizable k-ε error

outlet water temperature (°C) 45.107 45.098 0.009 45.891 45.881 0.010 48.198 48.266 0.068
air inflow rate (kg/s) 42,774 42,699 75 (0.18%) 39,102 39,221 119 (0.3%) 27,624 27,985 361 (1.29%)

Fig. 5. The distributions of air inflow velocities of cooling deltas exported by the standard and realizable k-ε models under various wind speeds.

takes the following formula [5]:

p = 0.5ku ua2, u (2)
For overall heat transfer process, heat flux can be expressed as:
hw
q= (tw twl, i ) (water side)
(3)

q= (twl, i twl, o ) (heat conduction through the wall)

(4)

q = ha (twl, o ta ) (air side) (5)

where hw and ha are water-side and air-side convection heat transfer
coefficients, and λ is thermal conductivity of tube wall. σ is the thick-
ness of tube wall, and β is the ratio of overall heat transfer area to the
internal surface area of tubes. tw, twl,i, twl,o and ta are the temperatures of
inside water, internal wall, external wall and outside air respectively.
Combining above three equations, it can be obtained:
1
q= 1
(tw ta )
hw
+ + ha (6)
As for heat transfer between water and air in the radiator, the
conductive thermal resistance through the tube wall ( / ) and the
convective thermal resistance of water ( / h w ) are small, compared with
the convective thermal resistance of air (1/ ha ). So heat flux is ap-
Fig. 6. Schematic of heat transfer process and temperature difference between proximately expressed as:
water and air.
q = h a ( tw ta ) (7)
model, the realizable k-ε model is an improved version with better Using above assumption, heat flux obtained by Eq. (7) would be a
performance solving flow separation, and so it was adopted in this little larger than that obtained by Eq. (6), but If Eq. (7) is amended as
paper. Eq. (8), the adopted heat transfer temperature difference ( w1 2 w 2 ta )
t +t

Air static pressure drop across each column of the cooling delta would be a little smaller than the real one (tw ta ), as shown in Fig. 6.

74
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

So the combined result is that the error of heat flux exported by Eqs. (6)
and (7) is tiny. In Section. 2.3 below, heat rejections of the tower cal-
culated by the mentioned method are quite close to the experimental
results, with the maximum error less than 4%. Besides, outlet water
temperature distribution of cooling deltas obtained by this method is
quite similar to that reported in Ref. [3] or [21]. So the present method
is credible, and using this method, there is no need to calculate water
temperature along the flow way, which would lighten the computation
burden.
tw1 + tw2
q = ha ta
2 (8)

Therefore, the heat exchanged between air and water for the jth
cooling column is calculated as:
tw1 + tw2, j
Qj = ha, j ta Aj
2 (9)

In this paper, radiator model is adopted on the leeward surface of

cooling delta, but is only used to model pressure loss through cooling
column. Modeling heat transfer through cooling column is achieved by
the user defined functions (UDFs). Before the simulation, a constant
inlet water temperature is given for each cooling column, and its initial
outlet water temperature can be specified equal to inlet water tem-
perature. After every iteration, the outlet water temperature of the
column is updated by the following equation:
tw2, j = tw1 Qj / m w, j cp, w (10)

The pressure drop coefficient and air-side convective heat transfer

coefficient can be expressed as the following equations [13]:

ku = 22.25ua, n0.24 (11)

ha = 36.21ua0.45
,n (12)

Then, the overall heat transfer coefficient is expressed as:

1 1
Kr = 1
= 1 1
+ + +
Fig. 7. Flow chart of the numerical-theoretical iterative procedure of the in- hw ha 0.8
376uw 36.2ua0.45
,n (13)
direct dry cooling system.
where the conductive thermal resistance through tube wall is less than
10−4 here, and is much smaller than convective thermal resistances of

Fig. 8. The coupling of the tower and condenser during iterations and the corresponding variation of outlet water temperature of the tower.

75
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 9. Measuring points arrangement for thermal performance experiments of the NDDCT.

air and water. So its insignificant effect is ignored. specified is adopted for the louvers [17,19]. For the sake of research
In the absence of crosswind, the windward surface in Fig. 4(a) takes simplicity, it's assumed that air leak is uniformly distributed, and the air
boundary type of pressure inlet, while in windy conditions, crosswind leak condition is regarded as the condition, under which opening de-
velocity profile at the windward surface obeys the power law below: gree of louvers is very small. According to the wind-tunnel tests [17],
pressure-jump coefficient of the louver can exceed 14,000 1/m when
ucw, z = ucw (z /10)r (14)
opening degree is smaller than 1%. So simulations under pressure-jump
where z denotes the vertical height above the ground, ucw is referential coefficient of 14,000 1/m were performed, and the results are in good
crosswind speed at the height of 10 m. The exponent r is related to the accord with the field data, which can be seen in Section 2.3.
ground roughness and the atmosphere stability, set as 0.16 [15] in this The scheme of first-order upwind is used for spatial discretization of
paper. the governing equations, and the SIMPLE algorithm is applied in
Boundary type of pressure outlet is employed for the top and lee- pressure-velocity coupling. The numerical calculation is repeated
ward surfaces, and the symmetry boundary is adopted for the symmetry iteratively, and converged results are exported until the monitored
surface, at which the gradients of all the evaluated variables are set as average outlet water temperature of cooling columns remains fairly
zero. The ground and tower body are set as adiabatic and no slip wall stable during consecutive iterations.
boundary. Turbulence level of ambient air on the far field boundaries
including windward, top and leeward surfaces is quite low, and tur- 2.2. Thermal calculating model of the indirect dry cooling system
bulence intensity and viscosity ratio can be roughly set as 0.1% and 0.1
respectively [23]. To describe the existing air leakage under all louvers In condenser, heat transfer equation and energy balance equations
closed, the porous jump model with a large pressure-jump coefficient for both exhaust steam and cooling water are expressed with the

76
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Table 3
Data collection of measuring points in the experiments of the NDDCT.
Measuring points Unit Number Values under Values under
experimental experimental
condition 1 condition 2

o
ambient air C 1 22.85 17.11
temperature 2 22.58 17.19
3 22.81 17.36
4 22.68 17.23
5 22.70 17.27
6 22.68 17.13
7 22.67 17.19
8 22.66 17.25
9 22.67 17.32
10 22.66 17.28
o
internal air C 1 40.43 35.64
temperature 2 40.29 35.41
3 40.81 35.77
4 40.31 35.29
5 40.01 35.08
6 39.96 35.33
7 40.56 35.60
8 40.48 35.82
9 40.17 35.26
10 40.64 35.47
o
inlet water C 1 54.75 47.22
temperature 2 54.80 47.27
o
outlet water C 1 44.90 37.84
temperature 2 44.95 37.94
crosswind speed m/s 1.4 0.9
water mass flow m3/h 70,300 70,300
rate

following forms:
tw1 tw2
Qc = K c A c tc = K c Ac t t
ln tes tw 2 (15)
es w1
Fig. 10. Dimensionless outlet water temperature variation of cooling delta with
Qc = mes (h (tes, xd) h (tes, 0)) (16) circumferential angle, (a) at wind speed of 5 m/s, (b) at wind speed of 7 m/s.

Qc = m w cp, w (tw1 tw 2 ) (17)

Where Qc is heat rejection of the condenser. Kc, Ac and Δtc are the temperatures of the tower respectively ignoring heat loss of circulating
overall heat transfer coefficient, heat transfer surface area and loga- water pipelines.
rithm mean temperature difference for the condenser. mes, tes and xd are With Eqs. (15)–(17), the condensation temperature of the steam is
the mass flow rate, condensation temperature and dryness of the steam. obtained by Eq. (18), and then outlet water temperature of the con-
h(tes,xd) and h(tes,0) are the steam enthalpy and condensate enthalpy. denser is calculated by Eq. (19).
mw and cp,w are the mass flow rate and specific heat of circulating
mes (h (tes, xd) h (tes, 0)) e K c Ac / mw cp, w
cooling water. tw1 and tw2 are the outlet and inlet water temperatures of tes = tw2 +
the condenser, which are equal to the inlet and outlet water
m w cp, w (1 e Kc Ac / mw cp, w ) (18)

Table 4
Comparison of the calculated data with experimental and field data.
experimental data of the tower field data of the cooling system in winter conditions

ambient temperature (°C) 22.70 17.23 −11.88 −9.75

crosswind speed (m/s) 1.4 0.9 0.4 1.3
mass flow rate of cooling water (m3/h) 70,300 70,300 53,700 53,700
internal air temperature (°C) 35.47 40.37 \ \
calculated internal air temperature (°C) 35.04 40.02 \ \
tower inlet water temperature (°C) 54.78 47.25 44.72 46.34
calculated tower inlet water temperature (°C) \ \ 44.65 46.67
tower outlet water temperature (°C) 44.93 37.89 38.90 40.52
calculated tower outlet water temperature (°C) 45.10 38.26 38.72 40.75
back pressure (kPa) \ \ 8.86 9.90
calculated back pressure (kPa) \ \ 8.55 9.77
error of inflow air temperature increment (%) 2.44 1.97 \ \
error of cooling water temperature drop (%) 1.73 3.95 1.84 1.78
error of back pressure (%) \ \ 3.55 1.27

77
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 11. The effect of exhaust steam flow rate on back pressure under the air leakage condition at wind speed of 4 m/s.

mes (h (tes, xd ) h (tes, 0)) experimental data is listed in Table 3, and the processed data is sum-
tw1 = tes +
m w cp, w (1 e K c Ac / mw cp, w ) (19) marized in Table 4. Here, apart from the experimental data, the field
data which comes from the distributed control system (DCS) of the unit
To sum up, flow chart of iterative thermal calculations of the in- is also used as reference. As for the numerical model of the NDDCT or
direct dry cooling system is shown in Fig. 7. The number of iterations the numerical-theoretical iterative model of the indirect dry cooling
when conducting tower simulations is monitored, and is defined with system, the calculated results are close to the experimental or field data,
the symbol of NOI. If NOI ≥ 4000 and MOD (NOI-4,000, 200) = = 0, with the maximum error less than 4%. Moreover, outlet water tem-
the coupling of the tower and condenser occurs. The above first judg- perature distribution of cooling deltas obtained by the present models is
ment is used to ensure that a stable outlet water temperature (tw2) can compared with those reported in Refs. [3] and [21], as shown in
be exported under a specified inlet water temperature (tw1) after the Fig. 10. The two curves in Fig. 10(a) or 10(b) show a very similar trend,
convergence of tower simulations. The MOD (number, divisor) function and the little deviation between them is acceptable. All these indicate
in the second judgment means theoretical thermal calculation of the that the established models in this paper are credible.
condenser would get involved at the interval of 200 steps of tower si-
mulation. So the update of water temperature is automatic, and the 3. Results and discussion
coupling of the tower and condenser repeats, until the error between
the tw1 used in tower simulation and the outputted tw1’ by thermal 3.1. Thermal performance of the NDDCT and thermal parameters of
calculation of the condenser is less than 0.1 °C. Fig. 8 shows the var- indirect dry cooling system without through-flow tower shell
iation of outlet water temperature of the tower during iterations. The
coupling points of the tower and condenser (i.e. the updates of tw1) are There are many factors impacting the operation of the indirect dry
clearly given. Once the condensation temperature tes is exported, back cooling system, and the impacts can be divided into two categories. One
pressure of the turbine, which can be regarded as the condensation is the impact of ambient condition, like ambient temperature and wind
pressure approximately, is calculated by the following empirical for- speed. The other is the impact of operation mode and parameters of the
mula in practical engineering [16]: steam turbine, like exhaust steam flow rate. Fig. 11 shows the effect of
tes + 100 7.46 exhaust steam flow rate on back pressure under the air leakage condi-
pc = 0.00981 tion at wind speed of 4 m/s. Under the same ambient condition, back
57.66 (20)
pressure would increase with exhaust steam flow rate, so low exhaust
steam flow rate is bad for the safe operation of the system in the ex-
2.3. Model validation tremely cold days. As a result, the effects of ambient condition on the
performances of the tower and the whole system at the low exhaust
To evaluate the accuracy of the modeling approach, two thermal steam flow rate were mainly researched and discussed in this paper.
performance experiments of the full-scale NDDCT were carried out, and Heat transfer in the tower under no crosswind is natural convection
the arrangement of measurements is shown in Fig. 9. Wind speed is heat transfer, and flow rate of the inhaled air is quite small under the air
measured at the height of 10 m by the cup anemometer accurate leakage condition with all louvers closed. Due to the large heat transfer
to ± (0.3 + 0.03v) m/s, and its measuring point is placed in open area temperature difference of heat exchanger, the little inhaled cold air gets
far away from the main buildings. Air temperature is measured by strongly heated. Forced by the generated powerful upward buoyancy,
pt100 temperature sensor accurate to ± (0.15 + 0.2%t) °C. Ten the little inhaled airflow turns to rise immediately. Because of the weak
measuring points of ambient temperature are arranged 1 m away from momentum, the inhaled air can hardly reach tower middle after flowing
the radiators at half of the radiator height. So do the measuring points through heat exchanger, and treating tower body as an air passage, the
of internal air temperature. Both the inlet and outlet water temperature inhaled air flows upward along the edge, as shown in Fig. 12(a). Sub-
are measured at the main pipe by pt100 temperature sensor accurate sequently, the inside rising airflow would partly flow back into tower
to ± (0.3 + 0.5%t) °C, and mass flow rate of cooling water is mea- middle when discharged from tower top, which leads to the symmetric
sured by ultrasonic flowmeter with accuracy of ± 1%. All the original recirculation zones in the tower.

78
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 12. Aerodynamic fields at the vertical cross section of the tower without through-flow tower shell at ambient temperature of −11.88 °C and wind speed of, (a)
0 m/s, (b) 4 m/s, (c) 8 m/s, (d) 12 m/s.

Under crosswind, flow fields inside the tower are quite similar. The speed, but would get a slow decrease at high wind speeds. The com-
obvious difference among Fig. 12(b), (c) and (d) is that crosswind bined result is that under this air leakage condition, thermal perfor-
contributes to drive air into the tower from windward cooling deltas. mance of the tower gets improved, as shown in Fig. 13(b). In addition to
Fig. 13(a) shows the dimensionless inhaled air volume flow rates of the overall cooling water temperature drop, the maximum temperature
different cooling sectors. As mentioned, air inflow rate promotion of drop of cooling water, appearing in windward sector shown in
windward sector caused by crosswind is obvious, especially at high Fig. 13(c) and (d), increases with crosswind speed, and at wind speed of
wind speeds. Air inflow rate of leeward sector increases slightly with 16 m/s, the maximum temperature drop of cooling water is nearly twice
wind speed. For the sideward sector, air flow rate increases with wind of the overall temperature drop. Thermal performance improvement of

79
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 13. Thermal performance of the NDDCT under crosswind without through-flow tower shell, (a) the dimensionless inhaled air volume flow rates of different
cooling sectors, (b) the overall and maximum cooling water temperature drops, (c) and (d) are cooling water temperature drop distributions of the two cooling
columns of cooling deltas under various wind speeds.

the tower brought by crosswind would cause back pressure of the in- shown in Fig. 15. On the one hand, cold airflow entering from the
direct dry cooling system to decrease, as shown in Fig. 14(a). Moreover, partial opened tower shell can be considered as the obstruction to the
crosswind would result in the decrease of the minimum outlet water inside rising airflow due to the difference of flow directions, as illu-
temperature of the tower, as shown in Fig. 14(b). If the minimum outlet strated in Fig. 16. On the other hand, the mix of the entered cold airflow
water temperature of the tower is below 0 °C, heat exchanger of the with inside rising airflow would result in the drop of internal air tem-
tower has high risk of freezing. If back pressure drops below the critical perature, and then cause the decrease of buoyancy that drives inside
value, steam expansion at the last stage of the turbine would get badly airflow to rise.
affected, and flow turbulence would appear at the back of the rotor From Fig. 15, it can be clearly seen that the three forms of through-
blade in the last stage, which is a threat to the safe operation of the flow tower shell have the same anti-freezing effect under the windless
turbine. Compared with the minimum outlet water temperature of the condition. Under windy conditions, windward through-flow tower shell
tower, back pressure reaches the critical value more easily, which is in performs best. Its anti-freezing effect increases with wind speed, and
accord with the viewpoint in Ref. [19]. As a result, even though all nearly 40% cooling capacity is reduced at wind speed of 8 m/s. By
louvers of cooling deltas are closed, back pressure can still reach the comparison, the impact of windward through-flow tower shell on the
critical value due to the air leakage, especially in the extremely cold performance of leeward sector is the most, which can be explained by
days with strong wind. Feasible region for the safe operation of the the analysis of flow fields. When windward through-flow tower shell is
system is narrowed by crosswind at low ambient temperatures. opened, the strong entering cold airflow in Fig. 16(b) would drop and
collide with the inhaled airflows of sideward and leeward sectors,
3.2. The effects of through-flow tower shell on thermal performance of the which would seriously affect the rising of these inhaled airflows, as
NDDCT and thermal parameters of indirect dry cooling system shown in Fig. 16(a). At moderate wind speed of 8 m/s, the inhaled
airflow of leeward sector is unable to contend against the strong cold
The proposed through-flow tower shell is of help to reduce cooling airflow entering from the windward through-flow tower shell, and some
capacity of the tower that has been greatly improved by crosswind, as internal air even flows out from the leeward sector, as shown in

80
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

shell can't greatly reduce the air inflow rate of windward sector, as
shown in Fig. 15(d).
By the numerical-theoretical iterative procedure of the indirect dry
cooling system, back pressures of the system and minimum outlet water
temperatures of the tower under various wind speeds are exported in
Fig. 17. The windward through-flow tower shell is the optimal ap-
proach for the safe operation of the system among the three through-
flow tower shells, and its advantage increases with wind speed. At wind
speed of 8 m/s, the critical ambient temperature for the safe operation
of the system is changed from −18.5 °C to −33.8 °C with the support of
the windward through-flow tower shell. That is the feasible region of
the system is effectively expanded by the proposed through-flow tower
shell, especially the windward through-flow tower shell. At the fairly
high wind speed like 16 m/s, except the sideward through-flow tower
shell, the other two forms still take effect, although the positive effect
declines compared with that at a moderate wind speed.

3.3. The comparison of the proposed anti-freezing approach with other

reported approaches for the safe operation of the indirect dry cooling system

Table 5 shows the comparison among several anti-freezing ap-

proaches. Though it's hard to make the comprehensive comparison,
some parameters are helpful for the rough evaluation of anti-freezing
effect. The residual anti-freezing coefficient was introduced in Wang
et al. [18], and the lager the residual anti-freezing coefficient is, the
smaller the freezing risk of heat exchanger becomes. In Wang et al. [20]
and this paper, the decreasing percent of overall cooling capacity of the
tower were used. The larger the decreasing percent of overall cooling
capacity is, the smaller the freezing risk becomes under low ambient
temperatures. As seen in Table 5, the anti-freezing approach presented
in this paper performs quite well under normal wind speeds, although
it's a bit inferior to the other two approaches under the fairly high wind
speed. At the moderate wind speed of 8 m/s, the latter two approaches
in Table 5 have the similar anti-freezing result. Under the windless
condition, the approach of switching off sectors brings the adverse ef-
fect. But the approach proposed in this paper is still efficient to reduce
the cooling capacity of the tower, and is helpful for the anti-freezing of
the air-cooled radiators. On the whole, the anti-freezing approach
Fig. 14. Thermal parameters of the indirect dry cooling system under crosswind proposed in this paper is always effective under the windless and windy
without through-flow tower shell, (a) back pressure of the system, (b) the conations, and the maximum anti-freezing result is achieved at the
minimum outlet water temperature of the tower. moderate wind speed.

Fig. 15(c). At a higher wind speed, more airflow would flow out from 4. Conclusions
the leeward sector, as seen in Fig. 15(d).
The anti-freezing effect of leeward through-flow tower shell is close To avoid the excessive low back pressure and cooling water tem-
to, but a little better than that of sideward through-flow tower shell. perature caused by air leakage in extremely cold days with strong wind,
With the increase of wind speed, both of their anti-freezing effects de- a novel partial through-flow tower shell is proposed to cope with the
grade, comparing Fig. 15 (a), (b) and (c). The main reason is that their excess cooling capacity of the NDDCT.
impacts on the ventilation of windward sector decrease with wind By the analysis of results exported by the established numerical
speed, and at a moderate or higher wind speed, their impacts on the model of the NDDCT, thermal performance of the tower would get
ventilation of windward sector are no longer obvious, as seen in enhanced by crosswind under the air leakage condition, but that can be
Fig. 15(c) and (d). At a fairly high wind speed, the sideward through- counteracted by the proposed partial through-flow tower shell. The
flow would even induce an adverse effect on the anti-freezing of the partial through-flow tower shell would reduce air inflow rate and
tower. When sideward through-flow tower shell is opened at the high cooling capacity of the tower, and then increase back pressure and
wind speed, the flow around the tower, especially outside tower flank cooling water temperature after the coupling of the tower and con-
would generate an extra suction, which is beneficial to the ventilation denser. Compared with the sideward or leeward through-flow tower
of tower. So cooling capacity of the tower improves in this condition. In shell, the windward through-flow tower shell would provide a much
addition to sideward through-flow tower shell, anti-freezing effect of larger feasible region for safe operation of the system, and achieve the
windward or leeward through-flow tower shell degrades at the fairly maximum anti-freezing result at the moderate wind speed. At the fairly
high wind speed as well. That is because thermal performance of the high wind speed, except the sideward through-flow tower shell, the
tower strongly relies on the performance of the windward cooling other two forms still take effect, although the positive effect degrades a
sector at high wind speeds, and the three forms of through-flow tower bit compared with that obtained at the moderate wind speed.

81
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 15. Cooling capacity of the NDDCT and ventilation rates of its sectors with partial through-flow tower shell at ambient temperature of −20 °C and crosswind
speed of, (a) 0 m/s, (b) 4 m/s, (c) 8 m/s, (d) 16 m/s.

Nomenclature Γ diffusion coefficient (kg m−1 s−1)

θ circumference angle (°)
A heat dissipation area (m2) φ scalar variable for airflow
cp specific heat (J kg−1 K−1) Δ the drop of a variable
h convective heat transfer coefficient of air-cooled heat ex- Subscripts
changer (W m−2 K−1)
k turbulence kinetic energy(m2 s−2) a air
ku pressure loss coefficient through the heat exchanger cw crosswind
Kc overall heat transfer coefficient of the condenser (W m−2 c condenser
K−1) D cooling delta
Kr overall heat transfer coefficient of the air-cooled heat ex- DA Column A of the cooling delta
changer (W m−2 K−1) DB Column B of the cooling delta
m mass flow rate (kg s−1) es exhaust steam
p pressure (Pa) j cooling column number along tower circumference
q heat flux (W m−2) n normal direction
Q heat rejection (W) S sector
S source term tw tube wall of heat exchanger
t temperature (K) w circulating cooling water
u airflow velocity (m s−1) 1 inlet
V inhaled air volume flow rate (m3 s−1) 2 outlet
xd dryness of exhaust steam
Greek symbols
Declarations of interest
ρ density (kg m−3)
ε turbulent dissipation rate (m2 s−3) None.

82
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 16. Aerodynamic fields at ambient temperature of −20 °C and crosswind speed of 8 m/s, (a), (c) and (e) are aerodynamic fields at the vertical cross section of the
tower under windward, sideward and leeward through-flow tower shell opened respectively, (b), (d) and (f) are aerodynamic fields at the horizontal cross section of
cooling delta height under windward, sideward and leeward through-flow tower shell opened respectively.

83
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Fig. 17. Thermal parameters of the indirect dry cooling system under the air leakage condition, (a), (c), (e) and (g) are back pressure variations with ambient
temperature at wind speed of 0 m/s, 4 m/s 8 m/s and 16 m/s respectively, (b), (d), (f) and (h) are the variations of minimum outlet water temperature of the tower
with ambient temperature at wind speed of 0 m/s, 4 m/s 8 m/s and 16 m/s respectively.

84
H. Ma et al. International Journal of Thermal Sciences 136 (2019) 70–85

Table 5
The comparison of the proposed anti-freezing approach with other reported approaches.
anti-freezing approaches for the NDDCT in cold days evaluation parameters results under various wind speeds

0 m/s 8 m/s 16 m/s

water flow rates regulation proposed by Wang et al. [18] overall residual anti-freezing coefficient 0.0% 22.6% 31.7%
switching off sectors proposed by Wang et al. [20] decreasing percent of overall cooling capacity of the tower −9.3% 45.4% 67.6%
using windward through-flow tower shell proposed in this paper decreasing percent of overall cooling capacity of the tower 23.2% 40.4% 18.7%

Acknowledgements Tran. 108 (2017) 1924–1940.

[11] M. Goodarzi, R. Keimanesh, Heat rejection enhancement in natural draft cooling
tower using radiator-type windbreakers, Energy Convers. Manag. 71 (2013)
The financial support for this research, from China Power 120–125.
International Development Limited (Grant No. ZDST-JH-20160617) [12] H.F. Gu, H.J. Wang, Y.Q. Gu, J.N. Yao, A numerical study on the mechanism and
and the Scientific Research Foundation of Graduate School of Southeast optimization of wind-break structures for indirect air-cooling towers, Energy
Convers. Manag. 108 (2016) 43–49.
University (Grant No.YBPY1701) is gratefully acknowledged. [13] H. Ma, F.Q. Si, Y. Kong, K.P. Zhu, W.S. Yan, Wind-break walls with optimized
setting angles for natural draft dry cooling tower with vertical radiators, Appl.
References Therm. Eng. 112 (2017) 326–339.
[14] H. Ma, F.Q. Si, K.P. Zhu, J.S. Wang, The adoption of windbreak wall partially ro-
tating to improve thermo-flow performance of natural draft dry cooling tower
[1] X.P. Wu, L.J. Yang, X.Z. Du, Y.P. Yang, Flow and heat transfer characteristics of under crosswind, Int. J. Therm. Sci. 134 (2018) 66–88.
indirect dry cooling system with horizontal heat exchanger A-frames at ambient [15] H.G. Zavaragh, M.A. Ceviz, M.T.S. Tabar, Analysis of windbreaker combinations on
winds, Int. J. Therm. Sci. 79 (2014) 161–175. steam power plant natural draft dry cooling towers, Appl. Therm. Eng. 99 (2016)
[2] H. Ma, F.Q. Si, Y. Kong, K.P. Zhu, W.S. Yan, A new theoretical method for pre- 550–559.
dicating the part-load performance of natural draft dry cooling towers, Appl. [16] L.J. Yang, X.P. Wu, X.Z. Du, Y.P. Yang, Dimensional characteristics of wind effects
Therm. Eng. 91 (2015) 1106–1115. on the performance of indirect dry cooling system with vertically arranged heat
[3] M.D. Su, G.F. Tang, S. Fu, Numerical simulation of fluid flow and thermal perfor- exchanger bundles, Int. J. Heat Mass Tran. 67 (2013) 853–866.
mance of a dry-cooling tower under cross wind condition, J. Wind Eng. Ind. Aerod. [17] L. Chen, L.J. Yang, X.Z. Du, Y.P. Yang, Anti-freezing of air-cooled heat exchanger by
79 (3) (1999) 289–306. air flow control of louvers in power plants, Appl. Therm. Eng. 106 (2016) 537–550.
[4] Q.D. Wei, B.Y. Zhang, K.Q. Liu, X.D. Du, X.Z. Meng, A study of the unfavorable [18] W.J. Wang, L.J. Yang, X.Z. Du, Y.P. Yang, Anti-freezing water flow rates of various
effects of wind on the cooling efficiency of dry cooling-towers, J. Wind Eng. Ind. sectors for natural draft dry cooling system under wind conditions, Int. J. Heat Mass
Aerod. 54 (1995) 633–643. Tran. 102 (2016) 186–200.
[5] R. Al-Waked, M. Behnia, The performance of natural draft dry cooling towers under [19] W.J. Wang, X.W. Huang, L.J. Yang, X.Z. Du, Y.P. Yang, Anti-freezing operation
crosswind: CFD study, Int. J. Energy Res. 28 (2) (2004) 147–161. strategies of natural draft dry cooling system, Int. J. Heat Mass Tran. 118 (2018)
[6] H. Ma, F.Q. Si, L. Li, W.S. Yan, K.P. Zhu, Effects of ambient temperature and 165–170.
crosswind on thermo-flow performance of the tower under energy balance of the [20] W.J. Wang, Y. Kong, X.W. Huang, L.J. Yang, X.Z. Du, Y.P. Yang, Anti-freezing of air-
indirect dry cooling system, Appl. Therm. Eng. 78 (2015) 90–100. cooled heat exchanger by switching off sectors, Appl. Therm. Eng. 120 (2017)
[7] Y.B. Sun, Z.Q. Guan, H. Gurgenci, K. Hooman, X.X. Li, L. Xia, Investigation on the 327–339.
influence of injection direction on the spray cooling performance in natural draft [21] Y.B. Zhao, G.Q. Long, F.Z. Sun, Y. Li, C.J. Zhang, Numerical study on the cooling
dry cooling tower, Int. J. Heat Mass Tran. 110 (2017) 113–131. performance of dry cooling tower with vertical two-pass column radiators under
[8] A. Alkhedhair, I. Jahn, H. Gurgenci, Z.Q. Guan, S.Y. He, Y.S. Lu, Parametric study crosswind, Appl. Therm. Eng. 75 (2015) 1106–1117.
on spray cooling system for optimising nozzle design with pre-cooling application in [22] H.T. Liao, L.J. Yang, X.Z. Du, Y.P. Yang, Influences of height to diameter ratios of
natural draft dry cooling towers, Int. J. Therm. Sci. 104 (2016) 448–460. dry-cooling tower upon thermo-flow characteristics of indirect dry cooling system,
[9] M. Goodarzi, H. Amooie, A proposed heterogeneous distribution of water for nat- Int. J. Therm. Sci. 94 (2015) 178–192.
ural draft dry cooling tower to improve cooling efficiency under crosswind, IEEE [23] Y.B. Zhao, G.Q. Long, F.Z. Sun, Y. Li, C.J. Zhang, J.B. Liu, Effect mechanism of air
4th Conference on Thermal Power Plants, CTPP), Tehran, Iran, 2012, pp. 1–6. deflectors on the cooling performance of dry cooling tower with vertical delta ra-
[10] X.B. Wang, L.J. Yang, X.Z. Du, Y.P. Yang, Performance improvement of natural draft diators under crosswind, Energy Convers. Manag. 93 (2015) 321–331.
dry cooling system by water flow distribution under crosswinds, Int. J. Heat Mass

85