Applied Thermal Engineering: Kuljeet Singh, Ranjan Das
Applied Thermal Engineering: Kuljeet Singh, Ranjan Das
Applied Thermal Engineering: Kuljeet Singh, Ranjan Das
Research Paper
h i g h l i g h t s
A feedback and control model for the forced draft cooling tower performance.
Model simulation is presented for three different cases under varying heat load.
Useful for solar cogeneration, ORC, diesel engine and HVAC applications.
GA optimized results with feedback control save pump and blower power.
Performance parameters also improve with feedback.
a r t i c l e i n f o a b s t r a c t
Article history: In this work, a feedback model has been proposed in order to control the performance of a mechanical
Received 24 December 2015 draft cooling tower under varying heat load conditions suiting diverse applications such as solar power
Accepted 3 March 2016 generation, HVAC and diesel engine. The model uses specific objective functions to meet the operating
Available online 21 March 2016
conditions related with the process requirement under varying heat load. These objective functions have
been formulated using empirical correlations corresponding to different performance parameters devel-
Keywords: oped from experimental data. Thereafter, an inverse method-based feedback model is made to control
Forced draft
the cooling tower’s performance for different outputs. The simulated model considers different cases clo-
L/G ratio
Inverse optimization
sely relating to practical use of the cooling tower. Further, by monitoring the inlet water temperature
Feedback with time, the controlling variable governed by water to air ratio has been adjusted to optimize the tower
Tower control operation. It is concluded that for different situations, the proposed inverse feedback model is an opti-
mized tool to control the tower performance.
Ó 2016 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.applthermaleng.2016.03.030
1359-4311/Ó 2016 Elsevier Ltd. All rights reserved.
520 K. Singh, R. Das / Applied Thermal Engineering 105 (2016) 519–530
Nomenclature
A approach ð CÞ u uncertainty
a interfacial area m2 =m3 V volume of exchange core m3
cp specific heat at constant pressure ½kJ=ðkg CÞ v independent variable in uncertainty analysis
c coefficient
e effectiveness (%) Greek symbols
F objective function x specific humidity ðkg=kg of dry airÞ
G air flow rate ðkg=sÞ rT standard deviation
H head ðmÞ
h specific enthalpy ðkJ=kgÞ Subscripts
0
h saturated air enthalpy at bulk water temperature a air
ðkJ=kgÞ abs absolute
K mass transfer coefficient kg= m2 s cr. current
L water flow rate ðkg=sÞ ex. exact/required
Me tower characteristic ratio fg fluid–gas mixture
meV water evaporation rate ðkg=sÞ
g guessed value
N number of samples i inlet
n number of independent variables k index for iteration
P power ðkW Þ o outlet
p pressure N=m2 wb wet bulb
R range ð CÞ v vapour
T temperature ð CÞ
cooling tower has been estimated at different levels of heat loads ferent points are recorded using DAQ system. Performance param-
(inlet water temperature) and water to air ratios. Then, empirical eters are explained next.
correlations for performance parameters have been developed
from experiments. These correlations are then used in relevant 3. Performance characteristics and empirical correlation
objective functions, which have been formulated to get a desired
output from the tower under varying heat load conditions. In the In this work, the objective is to establish a feedback system
next section, the experiment setup and procedure have been based on inverse method to achieve a desired performance from
discussed. a forced-draft cooling tower. So, first the important performance
parameters at different levels of T L;i and L=G ratio are estimated
2. Experimental setup and procedure as discussed below.
½1 Range : R ¼ T L;i T L;o ð1Þ
A counter flow forced-draft cooling tower with wire mesh pack-
ing (Fig. 1) is studied having dimensions 0:30 0:30 1:35 m with Depending upon the process, a desired T L;o is needed for differ-
1:2 m fill height. Warm water from the tank is supplied at top of ent chemical processes [12] For example, in ORC, the condenser
the tower by a centrifugal pump through a gate valve and rotame- should be operated at a particular temperature difference between
ter having range 0—11 LPM: Water is distributed over the packing working fluid and cooling water to ensure optimum performance
and droplets of water come into contact with air supplied from the [13]. Moreover, low water temperature is advisable in applications
bottom by means of a centrifugal blower. Air flow rate is measured such as HVAC, but in engines, relatively higher temperatures are
using orifice plate alongwith U-tube manometer containing water required. It is found that low cold water temperature also increases
as working fluid. The cold water is again collected in tank, where the total cost of the tower [14].
water is heated and re-circulated. To maintain a desired tempera- ½2 Approach : A ¼ T L;o T wb;i ð2Þ
ture in tank, a heater of 2 kW is installed in the tank that is pow-
ered through a digital temperature controller. After calibration, The operating cost of the cooling tower is inversely proportional
six K-type thermocouples are used to measure the temperature to approach [14]. So, to minimize the cost of the entire system, a
at different locations i.e. water inlet, water outlet, DBT and WBT balance between range and approach needs to be maintained.
at air inlet and outlet. The data acquisition system (DAQ) of Range T L;i T L;o
National Instruments is used to record the temperature data from ½3 Effectiveness : e ¼ ¼ ð3Þ
Range þ Approach T L;i T wb;i
thermocouples. Experimental procedure is described below.
L=G ratio and T L;i are two major factors which strongly govern During heat transfer process within tower, water droplet is sur-
the cooling tower performance. In this study, five levels of rounded by film of air. The enthalpy difference between film and
T L;i ð36:4—45:3 CÞ have been maintained, which are in line with air provides driving force for cooling. The tower characteristic ratio
different applications such as HVAC, petrochemical industry and is expressed as [7,15],
other plants [12]. At each T L;i , nine levels of L=G ratio ð1—5Þ are Z T L;i
KaV cp;L dT
used, which are widely used in different processes. First, the water ½4 Tower characteristic ratio : Me ¼ ¼ 0 ð4Þ
L T L;o h hG
tank heater is lighted through a digital temperature controller hav-
ing desired setting. After achieving the desired water temperature, Eq. (4) is solved with Chebyshev four point method [16],
pump and air blower are powered. The warm water supplied at top Z
is cooled by air entering from the bottom through sensible and KaV T L;i
cp;L dT T L;i T L;o X4
1
¼ 0 ffi cp;L ð5aÞ
evaporative cooling, which is collected in tank, where it is again L T L;o h hG 4 j¼1
D hj
heated and re-circulated. Under steady state, temperatures at dif-
K. Singh, R. Das / Applied Thermal Engineering 105 (2016) 519–530 521
The enthalpy of the air can be estimated from below [7,17], þ 12:139568 273:152 1:40290431
hG ¼ cp;G ðT G Þ þ x hfg þ cp;v ðT G Þ ð6Þ 102 273:153 ð9Þ
4. Inverse feedback model where heating load is continuously increasing with time and outlet
water temperature is required to be gradually decreased with it.
In actual situation, the heat load imposed on cooling tower by Case-2 can be better understood keeping I.C. engine application in
an industrial system may vary continuously with time. This vary- mind, where at lower engine temperature, the warmer water is
ing load demands a feedback system that controls the tower for needed to be supplied, whereas, with increasing engine tempera-
a desired output. L=G ratio is a parameter that can be easily con- ture, the water temperature needs to be lowered [18]. Therefore,
trolled to get desired performance. In this study, three different in both cases, the desired outlet water temperature is required with
cases are considered to address the cooling tower operation under respect to time and the same has been obtained by continuously
varying heat load. The feedback model is based upon inverse opti- adjusted L=G ratio through inverse feedback system (Eq. (12)). Fur-
mization that uses certain objective functions (as per require- thermore, it is possible that different combinations of L and G sat-
ment), which are formulated using above empirical correlations. isfy a particular L=G ratio: So, depending upon the requirement,
this model also enables the option to select an optimum combina-
4.1. Objective functions tion of L and G for a desired L=G ratio that results in minimum
power consumption. To get multiple combinations of L=G ratio:
Case-1 and 2: First case has been taken for the situation where a the following objective function is formulated,
constant T L;o is required under continuously varying " #2
36:4 C 6 T L;i 6 46:3 C. The below objective function is formulated L L
F2 ¼
to estimate required L=G ratio to get a constant T L;o , G g G ex:
ð13Þ
h i2 h n oi2 Constraints : 0:2183 kg=s 6 L 6 0:109 kg=s
F 1 ¼ ðT L;o Þex: ðT L;o Þg ¼ ðT L;o Þex: T L;i R T L;i ; L=G g 0:0173 kg=s 6 G 6 0:217 kg=s
Constraints : 1 6 L=G 6 5 To minimize F 2 , the Genetic Algorithm (GA) is used because it
ð12Þ intelligently searches the variables satisfying a desired objective
function without violating the above constraints (lower and upper
where R is function of T L;i ¼ T L;i cr: and L=G ratio; whereas, ðT L;o Þex: limits of water and air flow rates). After each run of GA, different
is the required constant outlet water temperature. The constraints combinations of variables ðL and GÞ are obtained by minimizing
are driven by lower and upper limits of T L;i and L=G ratio for this the objective function ðF 2 Þ. Finally, the combination corresponding
setup. T L;i may continuously vary with time (as heat load varies with to minimum power consumption can be selected. Power consump-
process), hence it needs to be updated regularly in the above objec- tion expression is represented below,
tive function. Then objective function is minimized using an opti-
Lðor GÞgH
mization algorithm to obtain required L=G ratio for desired P¼ kW ð14Þ
ðT L;o Þex: . In Case-1 the objective is to obtain constant ðT L;o Þex: over a 1000
period of cooling tower operation with varying heat load for solar
and ORC cogeneration systems [9,13]. Case-2 considers a process
Enable optimum
User input
combination option with case 1 or 2
Yes
Adjust Cooling tower
Required L/G ratio
No L/G ratio
abs. TL, i TL, i 0? TL , i
cr. cr.
For the water-pump, lift height of 1:80 m is considered as the of cooling tower or by decreasing the approach. But, a decreased
required head, whereas for blower, head is taken as the tower approach enhances the energy consumption due to increased fan
height ð1:35 mÞ. power [14]. Therefore, a balance needs to be maintained between
Case-3: As explained earlier, the cooling tower is always used in A and R which is possible at A ¼ R. Consequently, for a given
conjunction with other systems like HVAC, chillers and many 36:4 C 6 T L;i 6 46:3 C appropriate determination of L=G ratio is
more. The power consumption of the system to which cooling governed by,
tower feeds water strongly depends upon the cold water tempera-
F 3 ¼ ½R A2
ture. In applications like condensers of HVAC, low cold water tem- ð15Þ
perature is desirable to save the energy consumption [19]. Low Constraint : 1 6 L=G 6 5
water temperature can be obtained by increasing either the range
Table 1
Uncertainties involved in different measurements and performance parameters.
Measurements
Absolute uG ¼ 1 104 kg=s uL ¼ 0:002 kg=s uT ¼ 0:007 C
uG uL uT
G ¼ 0:01 L ¼ 0:09 ¼ 0:0002
Relative T
Performance parameters
Absolute uR ¼ 0:01 C uA ¼ 0:01 C ue ¼ 0:07% uMe ¼ 0:003 umeV ¼ 4 106 kg=s
uR uA ue uMe umeV
Relative R ¼ 0:003 A ¼ 0:002 e ¼ 0:003 Me ¼ 0:003 ¼ 0:006
meV
TL,i = 36.4°C TL,i = 38.7°C TL,i = 40.3°C TL,i = 43.5°C TL,i = 45.3°C
10
8
6
4
2
16
Approach, A ( C)
14
12
10
8
6
Effectiveness, e ( )
60
50
40
30
Tower characteristic
1.36
ratio, Me
1.02
0.68
0.34
-4
7.70x10
Water evaporation
rate, meV (kg/s )
-4
6.60x10
-4
5.50x10
-4
4.40x10
where R and A are functions of T L;i and L=G ratio; At a given T L;i ,rel-
5:977 107
evant L=G ratio at the intersection point serves as the optimal
9:85 103
5:51 104
4:09 103
1:169 102
because at that point neither the range nor the approach is compro-
mised, which in turn results in overall energy saving. In this case
too, the inverse feedback system can be implemented to operate
c9
the cooling tower at the optimum L=G ratio under varying heat load
or T L;i . Fig. 2 depicts actual implementation of the inverse based
5:324 103
1:316 103
2:286 107
1:46 103
1:894 103
feedback model under continuously varying T L;i . The user can select
the appropriate case depending upon the requirement and conse-
quently the corresponding objective function is chosen. Then, the
c8
values of T L;i and required outlet water temperature ðT L;o Þex: are
updated in the desired objective function, which is further mini-
1:895 103
2:281 102
1:034 103
2:08 106
1:111 102
mized using GA to get the required value of L=G ratio; so that the
cooling tower operation can be controlled by adjusting L=G. More-
over, current inlet water temperature T L;i cr: with respect to time
c7
is continuously monitored and absolute difference
e ¼
T L;i cr: T L;i
is calculated regularly. If e ¼ 0, same L=G ratio is
8:925 102
4:137 105
4:669 106
maintained, whereas, if e > 0, then the feedback loop recalculates
4:552 102
the required L=G ratio to meet the desired objective. Furthermore,
0.3936
if the option of selecting optimum L=G ratio is selected
ðinvolving F 2 Þ, then the desired loop would be followed as shown
c6
in Fig. 2. Next, details of GA have been explained.
7:626 105
7:4 102
4.2. Genetic algorithm (GA)
1.465
0.5299
1.226
c5
GA falls under the category of evolutionary methods that is
based upon biological evolution, which repeatedly modifies the
7:74 102
3:74 105
population using biological rules [20]. It initially creates a random
population within lower and upper limits and satisfying searching
0.2644
0.3077
0.124
constraints that corresponds to number of possible solutions in
c4
binary strings of 10 bits (either 0 or 1). Then binary strings are con-
verted into real number using below relation,
1:146 104
ðUpper Limit Lower LimitÞ
0.6184
1.259
L=G ¼ Lower Limit þ
0.1195
210 1 0.9901
c3
3.322
50.58
(50 in this work) and fitness function is evaluated for this popula-
61.33
known as crossover and mutation are used generate the new pop-
Coefficients used to develop the correlation for different performance parameters.
as represented,
784.5
299.3
696.6
51.44
0:9936
0:9829
0:9914
0:9862
0:9940
ð18Þ
After mutation 11110100 00 10
Range, Rð CÞ
Coefficient
Table 3
Sensitivity of different correlations to uncertainties.
44
TL,i
42
34
33
2.4
L/G ratio
2.2
2.0
54
Effectiveness, e ( )
52
50
48
46
Tower characteristic
0.874
ratio, Me
0.836
0.798
0.760
-4
Water evaporation
5.95x10
rate, meV (kg/s )
-4
5.60x10
-4
5.25x10
-4
4.90x10
2 4 6 8 10 12 14 16 18 20
Time (min)
Fig. 5. Variation of different parameters with time for the cooling tower with and without inverse feedback (Case-1) (dashed and solid lines represent outlet water
temperature, T L;o in the first sub-figure).
required [21]. Let n is the final estimated result based upon n mea- temperature uncertainty (0.007 °C) is in line with available
sured variables, literature, which is 0.01 °C [23]. Next section covers the results
and discussion part.
n ¼ f ðv 1 ; v 2 ; v 3 ; . . . ; v n Þ ð19Þ
6. Results and discussion
Further, for each of the measured variable v, if the correspond-
ing uncertainty is u, then the contribution of each uncertainty in The experimental results for different performance parameters
final result is calculated as follows, are presented in Fig. 3. The range, R decreases with increase in
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi L=G ratio for each T L;i , because, increase in L=G ratio infers that L
2 2 2 2
un 1 @n 1 @n 1 @n 1 @n increases for a given G. Therefore, the amount of air available for
¼ u1 þ u2 þ u3 þ þ un
n n @v 1 n @v 2 n @v 3 n @v n water decreases, which directly decreases heat transfer between
two fluids [24]. Moreover, with the increase in T L;i ; R increases,
ð20Þ
because, as T L;i increases, the heat transfer potential between the
The uncertainties involved in each measurement and different warm water and cold air also increases. Further, if approach, A is
estimated performance parameters have been presented in Table 1. considered, for different values of T L;i , at a low L=G ratio, more air
pffiffiffiffi
The standard deviation of mean rT = N has been used to calculate is available for given amount of water that enables the tower to
the temperature uncertainty [22]. The temperature is recorded supply water at lower outlet temperature T wb;i . Therefore, A
after each 0.1 s, therefore high sampling rate results in low is less with lower L=G, which increases with L=Gratio. Moreover,
relative uncertainty value, whereas, the present value of absolute A increases with T L;i , because if T L;i increases at any value of
40
38 TL,i
34
32
4
L/G ratio
60
Effectiveness, e ( )
50
40
30
Tower characteristic
1.08
ratio, Me
0.81
0.54
-4
5.1x10
Water evaporation
-4
rate, meV (kg/s )
4.8x10
-4
4.5x10
-4
4.2x10
-4
3.9x10
2 4 6 8 10 12 14 16 18 20
Time (min)
Fig. 6. Variation of different parameters with time for the cooling tower with and without inverse feedback (Case-2) (dashed and solid lines represent outlet water
temperature, T L;o in the first sub-figure).
528 K. Singh, R. Das / Applied Thermal Engineering 105 (2016) 519–530
L=G ratio; T L;o also increases. Further, as effectiveness, e is the ratio surements has been revealed in Table 3. Each correlation is more
of R to the difference of T L;i and T wb;i , it also decreases as L=G ratio sensitive to the uncertainty involved in water flow rate ðuL Þ as
increases. Trends of Me reveals that it decreases with increase in compared to other parameters, because the relative uncertainty
L=G ratio, because less air promotes the heat transfer resistance in L is more as compared to other measured quantities. Next, these
and lowers Me [7]. Moreover, Me decreases as T L;i increases, which correlations have been used in the inverse method-based feedback
0 model as discussed earlier.
is due to increased air saturation enthalpy, h corresponding to bulk
water temperature (because of increased inlet water temperature), For demonstrating the inverse feedback model, results are pre-
0
which ultimately rises h hG . Evaporation rate, meV first increases sented for three different cases as discussed in Section 4. Fig. 5 cov-
with increasing L=G ratio and then decreases, because, as L=G ratio ers the detailed analysis for Case-1, where a constant, T L;o is
increases, the water temperature over the tower height increases, required against continuously varying T L;i ðor heat loadÞ. In this
resulting in increased meV . But, at increased L=G, the amount of example, the tower is required to supply water at a constant tem-
air available for a given amount of water becomes comparatively perature ðT L;o Þex: ¼ 33:7 C; where T L;i continuously varies ran-
less, so, the interfacial area between the water droplet and air domly with time. If tower operates without any feedback system
decreases, thereby lowering meV . Using 3rd order polynomial, the ði:e:; at fixed L=G ¼ 2:25Þ, it fails to deliver the water at desired
fitted surfaces for different parameters have been presented in temperature ðT L;o Þex: . The cooling tower equipped with feedback
Fig. 4 with relevant coefficients being mentioned in Table 2. The system continuously senses T L;i and adjusts L=G ratio by minimiz-
goodness of each correlation is also presented in Table 2, which ing the function, F 1 . The update of L=G ratio against time is also
indicate that the R2 value is greater than 0.98 for each correlation. shown, that enables the tower to supply the water at desired tem-
The sensitivity of each correlation to uncertainties in various mea- perature, i:e:; T L;o ¼ ðT L;o Þex: . It is also found that the cooling tower
5 10 15 20 25 30 35 40 45 50 55 60
2.47
Case 1
2.46
10 -3
TL,i = 44.5 C; TL,o = 33.7 C
-5
10 TL,i = 42 C; TL,o = 33.7 C
10 -7
10 -9
5.0 Case 2
4.9
TL,i = 38 C; TL,o = 34.6 C
L/G ratio
2.283
2.280
2.277
10 -3 Case 2
Objective function, F1
10 -5
TL,i = 38 C; TL,o = 34.6 C
10 -7
TL,i = 40 C; TL,o = 32.6 C
10 -9
10 -11
10 -13
5 10 15 20 25 30 35 40 45 50 55 60
Generation
Fig. 7. Variation of estimated parameter ðL=G ratioÞ and objective functions with generations of GA for Case-1 and Case-2.
K. Singh, R. Das / Applied Thermal Engineering 105 (2016) 519–530 529
Temperature ( C)
L
120 G
12
80
70 10
Power consumption, P (kW)
60 8 Optimum L/G
1.0x10 -3 6
8.0x10 -4 4
6.0x10 -4 Pump
Optimum 2
Blower 1 2 3 4 5
4.0x10 -4 combination
Total L/G ratio
2.0x10 -4
2 4 6 8 10 12 14 (b) 5 10 15 20 25 30 35 40 45 50 55 60
2.21
Combination
2.20
2.19
L/G ratio
5 10 15 20 25 30 35 40 45 50 55 60
(b)
Mass flow rate, (kg/h)
-3
10 -5 TL,i = 44.5 C
10
Objective function, F2
10 -7
10 -5
10 -9
10 -7
-11
10
10-9 5 10 15 20 25 30 35 40 45 50 55 60
Generation
-11
10
5 10 15 20 25 30 35 40 45 50 55 60
(c) 2.25
Generation
2.20
Fig. 8. Different combinations of (a) mass flow rate of water and air ðL and GÞ for a
Optimum L/G ratio
constant L=G ratioð2:0846Þ and corresponding power consumption, (b) variation of 2.15
L, G and objective function F 2 for optimum combination during GA optimization.
2.10
having inverse feedback system has better e for most of the time 2.05
period as compared to tower without feedback. A similar trend is
2.00
also observed for Me. Moreover, meV also reduces by implementing
the inverse feedback system in comparison with the tower without 1.95
any feedback, which saves the cost due to less requirement of
make-up water. Therefore, results explained in context of Case-1 1.90
37 38 39 40 41 42 43 44 45
justify the use of inverse based feedback model to get a constant Water inlet temperature, TL,i ( C)
ðT L;o Þex: .
Fig. 6 reveals results for Case-2, where the heat load or T L;i Fig. 9. Optimization study (a) approach and range based optimum L=G ratio (b)
continuously increases with time and the water temperature, T L;o variation of L=G ratio and objective function during GA optimization (c) optimum
L=G ratio with the variation in T L;i (Case 3).
needs to be decreased with this [18]. The cooling tower having
feedback is able to supply the water at the desired temperature,
T L;o ¼ ðT L;o Þex: , whereas, in the absence of feedback, the tower fails, for minimizing F 1 . Further, as the number of generations increases,
as it operates at a fixed L=G ratio: The performance comparison for the objective function corresponding to each case decreases. Next,
both arrangements in terms of e and Me has been also presented. It it is obvious that a given value of L=G ratio can be satisfied by dif-
can be observed that as process (say diesel engine) picks up, the ferent combinations of L and G. So, the present inverse feedback
performance of the tower gets improved by the inverse feedback model also provides the option to select different combinations
arrangement. Further, meV is considerably less as the process picks. of L and G to meet certain performance requirement. For constant
The feedback model continuously updates T L;i and ðT L;o Þex: in the L=G ¼ 2:0846, covering this concept, results (involving 15 different
objective function F 1 and use GA to get required L=G ratio. After combinations) obtaining multiple combinations of L and G have
calculating desired L=G, required adjustments can be made in cool- been presented in Fig. 8a. Moreover, out of these combinations
ing tower by varying either L or G. At two random time levels, the one optimum combination can be selected that results in least
variation of L=G ratio and corresponding objective function, F 1 power consumption. For optimum combination, the iterative vari-
with GA generations for Case-1 and 2 have been presented in ation of two variables ðL and GÞ and the objective function, F 2 are
Fig. 7. GA takes 53–63 generations to determine L=G ratio ratio presented in Fig. 8b.
530 K. Singh, R. Das / Applied Thermal Engineering 105 (2016) 519–530
Fig. 9 presents the case where the cooling tower is required to [2] M.S. Söylemez, On the optimum sizing of cooling towers, Energy Convers.
Manage. 42 (2001) 783–789.
operate at an optimum L=G ratio: As explained previously, the
[3] M.S. Söylemez, On the optimum performance of forced draft counter flow
performance of certain processes (like HVAC) can be improved cooling towers, Energy Convers. Manage. 45 (2004) 2335–2341.
by lowering the cooling tower’s outlet water temperature. But, [4] G.F. Cortinovis, J.L. Paiva, T.W. Song, J.M. Pinto, A systemic approach for
optimal cooling tower operation, Energy Convers. Manage. 50 (2009) 2200–
at the same time high range results in decreased approach, which
2209.
further increases the operating cost. So, the optimum values of [5] E. Rubio-Castro, M. Serna-González, J.M. Ponce-Ortega, Optimal design of
L=G ratio have been proposed for the entire range of cooling effluent-cooling systems using a mathematical programming model, Appl.
tower inlet water temperature, T L;i ð36:4—45:3 CÞ. The intersec- Therm. Eng. 30 (2010) 2116–2126.
[6] E. Rubio-Castro, M. Serna-González, J.M. Ponce-Ortega, M.A. Morales-Cabrera,
tion point of range and approach is considered to be the optimum Optimization of mechanical draft counter flow wet-cooling towers using a
point (Fig. 9a), because if the cooling tower is operated at this rigorous model, Appl. Therm. Eng. 31 (2011) 3615–3628.
value of L=G ratio: neither the performance of tower nor the [7] R.K. Singla, K. Singh, R. Das, Tower characteristics correlation and parameter
retrieval in wet-cooling tower with expanded wire mesh packing, Appl. Therm.
industrial process will be compromised. In Fig. 9a sample results Eng. 96 (2016) 240–249.
are shown for two different values of T L;i ¼ 36:5 and 44:5 C. Sim- [8] D. Manolakos, G. Papadakis, S. Kyritsis, K. Bouzianas, Experimental evaluation
ilarly, this optimum L=G ratio is calculated under different heat of an autonomous low-temperature solar Rankine cycle system for reverse
osmosis desalination, Desalination 203 (2007) 366–374.
loads and the optimum output from whole setup can be obtained [9] B. Twomey, P.A. Jacobs, H. Gurgenci, Dynamic performance estimation of
by minimizing the objective function, F 3 . Variations of L=G ratio small-scale solar cogeneration with an organic Rankine cycle using a scroll
along with objective function, F 3 during optimization process expander, Appl. Therm. Eng. 51 (2013) 1307–1316.
[10] S. Abdelhady, D. Borello, E. Tortora, Design of a small scale stand-alone solar
are presented in Fig. 9b. Furthermore, the inverse feedback model thermal co-generation plant for an isolated region in Egypt, Energy Convers.
can be also implemented to operate the cooling tower at opti- Manage. 88 (2014) 872–882.
mum point under continuously varying heat load, where it has [11] F. Calise, M.D. D’Accadia, M. Vicidomini, M. Scarpellino, Design and simulation
of a prototype of a small-scale solar CHP system based on evacuated flat-plate
to be operated along the curve given in Fig. 9c. In this curve solar collectors and Organic Rankine Cycle, Energy Convers. Manage. 90 (2015)
the optimum value of L=G ratio has been shown for different 347–363.
T L;i , along which cooling tower needs to be operate for optimum [12] J.J. McKetta Jr., Encyclopedia of Chemical Processing and Design, Abrasives to
Acrylonitrile, vol. 1, CRC Press, 1976.
performance. [13] J. Freeman, K. Hellgardt, C.N. Markides, An assessment of solar-powered
organic Rankine cycle systems for combined heating and power in UK
7. Conclusion domestic applications, Appl. Energy 138 (2015) 605–620.
[14] B.G. Lipták, Process Control: Instrument Engineers’ Handbook, Butterworth-
Heinemann, 2013.
In this work, a forced draft cooling tower has been studied [15] F. Merkel, Verdunstungskühlung, VDI-Verlag, 1925.
under varying inlet water temperature and water to air ratio [16] A.K.M. Mohiuddin, K. Kant, Knowledge base for the systematic design of wet
cooling towers. Part I: selection and tower characteristics, Int. J. Refrig. 19
ðL=G ratioÞ. New empirical correlations are proposed to develop (1996) 43–51.
certain objective functions to meet the desired performance [17] J.C. Kloppers, A Critical Evaluation and Refinement of the Performance
depending upon diverse operating conditions. The present inverse Prediction of Wet-cooling Towers, University of Stellenbosch, 2003.
[18] M. Abdulqader, Diesel Generator Auxiliary Systems and Instruments, 2006.
feedback model involving the Genetic Algorithm (GA) predicts the <www.Lulu.com>.
appropriate values of L=G ratio with time under varying heat load [19] J.F. Kreider, Handbook of Heating, Ventilation, and Air Conditioning, CRC Press,
conditions. The suggested feedback model can be useful for differ- 2000.
[20] K. Deb, Optimization for Engineering Design: Algorithms and Examples, PHI
ent cases such as solar cogeneration, HVAC and diesel engine appli- Learning, 2012.
cations. Further, it can be also implemented to operate the cooling [21] R.J. Moffat, Contributions to the theory of single-sample uncertainty analysis, J.
tower at optimum performance reducing the input energy. It can Fluids Eng. 104 (1982) 250.
[22] J.R. Taylor, An Introduction to Error Analysis: The Study of Uncertainties in
be concluded that inverse feedback model in conjunction with
Physical Measurements, second ed., University Science Books, Sausalito,
GA is an optimized, simple and effective tool for adjusting water California, 1983.
to air ratio in order to control the performance of the cooling [23] M. Gao, F.Z. Sun, K. Wang, Y.T. Shi, Y. Bin Zhao, Experimental research of heat
transfer performance on natural draft counter flow wet cooling tower under
tower.
cross-wind conditions, Int. J. Therm. Sci. 47 (2008) 935–941.
[24] K. Singh, R. Das, An experimental and multi-objective optimization study of a
References forced draft cooling tower with different fills, Energy Convers. Manage. 111
(2016) 417–430.
[1] N. Milosavljevic, P. Heikkilä, A comprehensive approach to cooling tower
design, Appl. Therm. Eng. 21 (2001) 899–915.