Sei sulla pagina 1di 11

Applied Energy 211 (2018) 1297–1307

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Demonstration project of a cooling system for existing PV power plants in T


Portugal

André F.A. Castanheiraa, João F.P. Fernandesb, , P.J. Costa Brancob
a
Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal
b
IDMEC, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal

H I G H L I G H T S

• ATheprototype of the cooling system was installed in 20 kW PV power plant in Lisbon.


• The ON/OFF cooling cycles allow the cooling of several PV strings in parallel.
• The optimal cooling cycle of 20%/80% was experimentally obtained for Lisbon.
• The start-up temperature of the cooling system was 30 °C for Lisbon and Beja.
• financial return of the system is estimated to be 2 years for Lisbon and Beja.

A R T I C L E I N F O A B S T R A C T

Keywords: Water cooling continues to be an attractive solution for mitigating thermal effects in photovoltaic (PV) panels.
Photovoltaic systems However, a technically viable and economically effective design of a cooling system for utility-scale solar power
Temperature plants already operating is still in its early developments. In this context, this work not only presents the design
Efficiency and characterization of a water cooling kit system specifically conducted for already existing and operating PV
Water-cooling system prototype
plants, but also takes into account various aspects of its potential as a commercial-scale product. One considers
not only water and energy losses in the cooling system, optimizing it for a real industrial application, but also
economic aspects as the investment cost, durability and maintenance have been taken into account. In this
specific context, the cooling kit was designed to minimize its cost and also the amount of water used. An ana-
lytical model is also presented and validated to predict the PV temperature of a cooled and non-cooled panel, all
based on meteorological data at the desired installation place. Prototyping and testing were done on a 20 kW PV
plant functioning in a roof of an industrial building in Lisbon (Portugal). Experiments showed how cooling/
heating thermal time constants of the PV panels and water ON/OFF time cycles affect PVs’ efficiency. A large
amount of test data was produced and analyzed, showing how the cooling kit can increase annual energy
production up to +12%.

1. Introduction was mounted in 25 PV panels, reaching a peak power of 5 kW. The main
contributions are the analysis of the cooling system performance (water
Different techniques or even new materials to cool and clean PV and energy losses and an increase of power), already under an in-
panels are reaching its maturity. From different cooling techniques, the dustrial ambient and for large-scale applications, its design (auxiliary
water active cooling ones are those that present more increase of output equipment) and operating condition optimization (water cycles), and
electrical power [1]. However, a knowledge gap still exists when those the economic analysis, where the technology and economic aspects
solutions must be installed and operate in already existing utility-scale must be balanced.
solar power plants. With all the investments already done in photovoltaic technologies,
In this context, a water cooling kit system was designed to already several PV plants were installed worldwide with a fixed maximum ca-
existing and operating PV plants. The results of the performance testing pacity, which typically cannot be increased without a change in the
of the cooling system are presented and discussed when installing in a contract or an enlarging of the power plant. In addition to this limita-
20 kW PV park in an industrial building. In this park, the cooling system tion, the efficiency of PV cells is strongly dependent on its own


Corresponding author.
E-mail address: joao.f.p.fernandes@tecnico.ulisboa.pt (J.F.P. Fernandes).

https://doi.org/10.1016/j.apenergy.2017.11.086
Received 18 August 2017; Received in revised form 19 November 2017; Accepted 20 November 2017
0306-2619/ © 2017 Elsevier Ltd. All rights reserved.
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

temperature, being lower at higher temperatures [2,3]. This creates a not only the panel temperature decrease but also the configuration
limiting phenomenon in the energy PV production, where in days of of the hydraulic circuit to be implemented. Elements as the electric
high solar radiation the PV energy production stagnates due to the pump and also the recently proposed water spreading systems must
higher PV cells’ temperature. This effect decreases the PV plant energy be studied in a more representative way. For example, nozzles for
production and creates a negative economic impact. water spreading show critical obstructions problems when tested in
Recent developments in the PV cooling techniques over the last four real conditions. Dust and/or dirt deposition in the panel may cause
years show that the topic is still a strong area of interest. The different blockage of the nozzles, augmenting the hydraulic circuit pressure
types of cooling systems can be divided into passive and active ones. and thus the power consumed by the pump.
In the area of the passive cooling system, different techniques have • Research about water cooling systems to be commercialized and
been developed [4–9]. The research work in [4] presents a radiative installed in existing PV power plants is a gap in this topic. In this
cooling technique that uses a colorless and transparent silicon coat on context, an assessment study of the interplay between technical and
the front surface of the PV to generate thermal radiation and lower the economic risks is needed since these will dictate if the proposed
PV temperature. This layer is transparent for the range of solar radia- system becomes a viable and competitive option.
tion wavelengths but has a high emissivity which results in significant
thermal radiation. From simulation results it is possible to estimate a It is in this context that this paper builds on the water cooling kit
decrease in the PV operating temperature by 18.3 °C, however, these system for existing PV plants, namely those installed in Portugal, con-
still lacks experimental validation. In [5] is presented a passive cooling sidering various aspects of its potential as a commercial-scale product.
technique using natural vapor in the back of the PV, as heat transfer The project was based on a prototype installed in 5 kW (25 PV panels)
fluid. The authors suggest as a real application the installation of PV of a 20 kW PV plant already operating in the industrial facility rooftop
panels in the top of rivers or canals to use the evaporation of the water own by our commercial partner in Lisbon, Portugal.
as the coolant. Using a laboratory set-up, it was verified that the PV The novelty of this work concerns the installation of a water cooling
temperature could be reduced from around 60 °C to 40 °C, resulting in system for an already installed PV power plant, its adequate design for
an increase of 7.3% of electric power. the PV plant location and climate conditions, the optimization of the
Other recently developed techniques are the use of phase change ON/OFF water cycles (number of strings per water pump and start-up
materials (PCM) as passive coolants [6]. In [7] a simulation study was temperature) to maximize the profit from the water cooling system and,
done by introducing PCM spheres to a water storage tank that is con- finally, a detailed economic study, based on the experimental results,
nected, by pipes, to the back surface of a PV panel. With an optimized considering the water losses, the auxiliary system energy consumption
system, simulation results showed to be possible to obtain an increase and other aspects related to the water cooling system operation.
of 13.7% of electric power, when compared with the non-PCM case. The cooling system uses a front water flow with a semi-closed
Different types of active cooling systems have been studied, using looped system. I.e. uses a closed-loop water circuit where some water
fluid flow in the front [10–12] or back [13–15] of PV cells. Recently, an losses can occur, and its ON/OFF water cycles and start-up temperature
approach using a water cooling system in both PV sides showed to be are optimized for the minimization of costs and maximization of energy
possible to highly increase the electric power [16]. A 16.3% of increase production when installing in an already existing PV plant. To com-
of peak power was obtained using one PV panel cooled by sprinklers on plement the cooling system analysis, an analytical model was done to
both sides and tested under real ambient conditions in a Mediterranean provide the possibility of predicting a PV plant energy production based
climate. This active water cooling technique presents a high potential to on weather conditions available online, with and without the cooling
be implemented, however, the optimization of the water cycles and the system.
economic analysis in a real application were still not addressed. The work is structured in five main sections: the Section 2 where the
Different from [16], the research presented in [17] studied the concept of the system is described, Section 3 with the system im-
convective effect on the backside surface of a PV panel and respective plementation, Section 4 where the experimental tests to the prototype
electrical efficiency. Authors built an experimental setup using two installed in a 20 kW PV plant are presented, Section 5 with the devel-
panels to simulate typical Mediterranean climatic conditions. After opment and validation of the analytical model, and finally in Section 6
several tests, it was concluded that a redesign of the backside surface in an economic analysis and system optimization done for two regions of
conventional PV panels could improve its thermal dissipation to in- Portugal, based on the analytical model and the weather condition
crease the panel average efficiency. However, none design was sug- available from past years.
gested by the authors, neither an economic study for the needed
changes. 2. Concept of the water cooling system
Similar to [17], a backside surface cooling system using water flow
was studied in [15] for climate conditions at Jeddah city in Saudi The cooling system consists of a water cooling system to lower the
Arabia. A multiphysics model integrating the electrical, thermal and PV surface temperature, thus increasing its efficiency and using a set of
structure physics of a PV panel was used to study the PV panel per- valves to optimize the water flow. The main purpose is to use a uniform
formance. Different from similar studies, which never approach the water layer to lower its surface temperature and to create a refractive
evolution of mechanical stresses in the panel structure when cooled, layer to decrease the solar radiation reflected by the glass (refractive
authors showed that the cooled panel can present one order of mag- index: 1.3 to water and 1.5 to glass [10]).
nitude lower stress levels than an uncooled panel. However, none la- The water is stored in a tank which can be at ambient temperature
boratory scale experiment was even settled to validate the multiphysics or buried, to reduce the temperature of the water coming from the PV
model. panels. The water is pumped into to the main channel and then, using
Based on those latest research about the active cooling systems, one different controlled valves, is spread to different strings of PV strings.
can clearly point out key challenges that have to be addressed: Using sprinklers, the water is spread along the PVs at a low pressure and
speed to create a uniform layer of water. Finally, the water is collected
• There is still a gap between the laboratory prototypes and the real at the bottom of the PV strings and is carried to the water tank, where it
industrial applications. Laboratory tests confirm that it is possible to is processed and cleaned. To avoid the accumulation of biofilm soiling
increase the electrical output of a PV panel, but the costs related to and debris, the system requires filters. The proposed solutions for small
the operating system are not considered. This is an important factor plants are chlorine tablets placed in the tank, which can be acquired for
for the future implementation of real applications. 44 USD for 10 years of operation and anti-sediment filter (48 USD). In
• Optimization of the water cooling cycle must be made concerning larger plants, UV filters may also be used for larger tanks (374 USD plus

1298
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

Fig. 1. Cooling system’s overall scheme.

lump (77 USD) and coal cartridge (20 USD)). 3. Implementation of the cooling system
The main characteristics of this water cooling system concept are
based on the following: A prototype of the water cooling system was installed on a string of
5 kW (25 panels) of a 20 kW PV solar power plant located in Bobadela,
• The reduction of the water pump fatigue: Lisbon, Portugal. The solar power plant consists of four 5 kW PV strings:
three strings of monocrystalline silicon PVs and one of poly-crystalline
To reduce the water pump fatigue, the pump operates at only one PVs, as shown in Fig. 3. For the experimental tests, two strings of mono-
speed (corresponding to the minimum water flow) and with the crystalline PVs were chosen, one to be the reference (string 1) and the
minimum required start/stop operations. To achieve this, a set of other to be cool with the developed prototype of the cooling system
controlled valves (v1 to v5 in Fig. 1) is installed and only one operates at (string 2). The mono-crystalline panels present a rated power of 190 W
one time. With this, the water pump flow remains constant. and the characteristics are on Table 1. Note that the maximum power of
the panel decreases with the temperature, −0.45%/°C.
• The optimization of the water flow: The prototype cooling system schematic is shown in Fig. 4. The PV
panels are installed on a rooftop with an inclination of 33° along the
From experimental tests, it was verified that the PV cooling thermal strings array, Fig. 5.
constant is much higher than the heating one. Using the difference Experimental tests were done for only one string (String 2 in Fig. 3)
between thermal constants, it is advantageous to use a cyclic cooling with twenty-five panels, where the cooling time during which the
process which cools only one string at one time, as seen in Fig. 2. With sprinklers are working is controlled by an electric valve. Different water
this process, the water pump can be dimensioned by only one string, cycle ratios ON/OFF are tested to analyze the impact of the efficiency
reducing the investment and operating costs of the system. increase and the water and electric water pump consumption. One ON/
OFF cycle corresponds to the time the panel is being cooled (ON) or not
(OFF).

Fig. 2. An example of pump operation with 3 strings.

1299
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

String 2
(mono)
String 1
(mono)

String 3
(mono)
String 4
(poly) Fig. 5. Water cooling system overall.

Fig. 3. PV solar power plant installed on the rooftop of the company building.

Table 1
Temperature characteristics.

Characteristic

Nominal Operating Cell Temperature (NOTC) 45 ± 2 °C


Temperature Coefficient of Pmax −0.45%/°C
Temperature Coefficient of Voc −0.34%/°C
Temperature Coefficient of Isc 0.050%/°C

The water is stored in a 150L water tank (Fig. 6) and it is pumped Fig. 6. A pump placed in the water tank.
through a manual valve and a controlled valve into the two arrays of
PVs of string 2. The escape valve simulates the presence of other PV The major issue found in this system is related to the water losses in
strings being cooled, redirecting the water flow when the controlled form of water evaporation. In peak hours and with excessive wind
valve is closed and maintaining the water pump flow. Using sprinklers, speeds, the water evaporation losses may be significant.
the water is spread along the PVs panels with a low pressure to assure a
uniform distribution along them and to reduce the water leakage,
4. Experimental tests
Fig. 7. Finally, in Fig. 8, the water collecting system can be seen, made
of PVC gutters connected to the water tank. Due to the rooftop in-
A series of experimental tests were carried, with different ON/OFF
clination, the water flows to the tank (installed on the roof) by gravity.
cycles, to study the optimization of the cooling system and associated

Fig. 4. Schematic of the prototype water cooling system.

1300
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

Fig. 7. Sprinklers located on top of the panel.

Fig. 9. Placement of the temperature sensors in the PV panels.

Cell A
Cell B

Fig. 8. PVC gutters to collect the water and return it to the tank.

costs. The tests were performed between 9 am and 5 pm of months of


June to November. During the tests, an ambient temperature variation
between 21 and 37 °C and an irradiance between 500 and 1100 W/m2
Fig. 10. The temperature of the two cells between day 30 of July and 1 of August.
was measured.
Those tests were divided into a set of three:

1. Measurement and analysis of the temperature and power profiles


(thermal constants and reduction of power).
2. Impact analysis of the cooling system in the PV efficiency.
3. Measurement of water losses during the operating time.
IJOFF
These tests allow us to model the electrical and thermal behavior of
the PV panels, with and without the cooling system.

4.1. Temperature and power results


Fig. 11. An example of experimental and theoretical PV temperature for an ON/OFF
The temperature of two consecutive PV strings, one with the water cooling cycle.
cooling system (string 2, Fig. 3) and other without it (string 1, Fig. 3),
were measured for different days and months. Temperatures were required so that the panel is completely cooled (temperature of the PV
measured using probes in one panel of each string placed in front of the panel reaches the water temperature).
panels’ glass. For each panel, two temperature sensors were placed as The second set of experiments is related to the influence of the
shown in Fig. 9. In the point A, at the top of the PV, after the water temperature in the output power of the PV strings. Experimental results
stream is uniform; and point B in the middle of the panel. Temperature show that the reduction of the temperature is significant when using the
differences between the two points were lower than 2 °C, Fig. 10. Their water cooling system and a maximum power increase of 17% was
average 5-min value was considered as the PV temperature. measured during the whole experience.
The first set of experiments were related to the thermal constants of As an example, for the 11th of August of 2016, the temperature
the PV with the cooling system. Fig. 11 shows the PV temperature of profiles for strings 1 and 2 (without and with the cooling system, re-
one ON/OFF water cycle for one day, where it is possible to calculate spectively) are shown in Fig. 12, with different ON/OFF cycles for
the ON (with the water running) and OFF thermal constants. From the different hours. In this day, the temperature difference between the
all experimental data obtained, the average ON and OFF thermal con- cooled and non-cooled panels reach up to 22 °C. The effect of the de-
stants were 0.6 and 11 min, respectively. Therefore, being the cooling crease of temperature on the cooled string resulted in an increase of
transient faster than the heating one, one can take advantage of it to power.
optimize the water flow by using a cyclic cooling process, i.e., using an In Fig. 13 is shown the power output from the cooled and non-
intermitted water flow for different strings. The experimental tests cooled string, showing an increase up to 15%. Note that the largest
showed that a minimum of 2 min with the system switched ON is

1301
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

Fig. 12. Experimental results for 11th August 2016.


Comparison between the cooling panel (red) and the
reference panel (black) a) temperature.

Fig. 14. Correlation between the increase of electrical efficiency and the ON/OFF cycles,
where “On Cycle” = ON/(ON + OFF) times.

time of the cooling system. For example, with this, it is possible to


expect more income for two strings with a 50% ON cycle (one string
Fig. 13. Power output comparison between the cooled string (dashed line) and the re-
being cooled at each time) than just one string with 100% of ON cycle.
ference one (continuous line), 11-August-2016. Additionally, it was verified that the influence of the wind in the
semi-closed water circuit could not be neglected. A small fraction of the
water that falls on the panel eventually leaves the circuit via evapora-
difference occurs between 11:30 h and 15:30 h when the ambient
tion due to the wind influence. After several tests with different ON/
temperature and solar irradiance are higher. Aside from the increase of
OFF cycles, it was verified that these losses can vary between 10 and
power related to the decrease of temperature, it was also verified that
25 l/h for a range of wind speed up to 6 m/s. This highlights the im-
because of the water flowing in front of the PV panels, those were also
portance of increasing the number of string fed in parallel by the same
cleaned from dust and dirt. This cleaning process keeps the panel
water pump.
cleaned for 2–3 days, even without the cooling system operating, thus
increasing the power output of the string. Even with the cooling system
4.3. Experimental results discussion
not running, due to the water cleaning from the previous day, the pa-
nels already presented an increase of 5% of power. When the cooling
From the experimental results the following conclusions were
system is ON, the power increases up to 15%.
reached:
For the day shown in this figure, the experimental increase of en-
ergy, in this 5 kW string, corresponded to an increase of 3.6 kWh
(1.26 USD) over the 24 kWh (7.02 USD) produced without the cooling • The range of water flow and cycles were enough to keep a uniform
temperature along the PV panel.
system. The cost of water pump energy and water losses were 0.29 USD
and 0.12 USD (72 l), respectively. Therefore, a gross gain of 0.85 USD • heating and cooling of the PV panels can be approximated to
The
exponential transients with the cooling and heating thermal time
was measured, corresponding to an increase of 12.1% of income.
constants of 0.6 and 11 min, respectively.

4.2. ON/OFF cycles efficiency increase • When changing the ON/OFF cycles, the experimental efficiency
enhancement changed between 7% and 17% for the cycles of 20%/
80% and 100%/0% (ON/OFF cycles). With this conclusion, it is
The optimization of the water flow is an important aspect of this
possible to explore the possibility of using only one water pump to
water cooling system. The sizing and energy consumption of the water
feed different PV strings. One example is to use an ON/OFF cycle of
pump and the resulting water losses are critical for the economic via-
20%/80%, to allow the use of one water pump for 5 PV strings (each
bility of the system. The ON/OFF cycles allow us to have several PV
one with 5 kW), where each string is cooled at 20% of the time
strings fed by the same water pump and, using an intermittent cycle as
cycle.
described in Fig. 2, decrease the water and electricity consumption of
the PV cooling. • Additionally, on 11th of August of 2016, using the developed pro-
totype for only one 5 kW PV string, it was measured an energy in-
In Fig. 14 are presented the experimental results for the average
crease of 3.6 kWh (1.26 USD) over the 24 kWh (7.02 USD) of normal
daily increase of efficiency with different ON/OFF cycles, measured
production and a total water cooling system variable costs of
during the 5 months of experiments.
0.41 USD. With this, a gross gain of 0.85 USD was measured,
As it can be seen, the increase of efficiency is not linear with the ON

1302
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

corresponding to an increase of 12.1%.

5. Analytical model

To complement the cooling system analysis, a thermoelectric ana-


lytical model was done to provide the possibility of predicting a PV
plant energy production based on weather conditions available online,
with and without the cooling system.
The analytical model consists of the estimation of the PV efficiency
based on (a) the PV temperatures without the cooling system for the
specific weather conditions and (b) the PV temperature with the cooling
system. For the cooled PV, the model was built using the experimental
thermal constants obtained.
Fig. 15. Comparison between experimental data (dashed line) and theoretical data
5.1. PV panel temperature (continuous line), for 9th August of 2016.

Usually, PV panel temperature can be estimated by a simplified


Table 3
model, based on the ambient temperature, Tamb and the irradiance, G .
Initial costs for different strings.
This model is described in (1) [18], where NOCT (Nominal Operation
Cell Temperature) is given by the panel manufacturer in its datasheet. Material 1 2 3 4 5
This variable represents the temperature reached by the cell in normal String Strings Strings Strings Strings
(USD) (USD) (USD) (USD) (USD)
operating conditions, defined by Tamb = 20 °C and G = 800 W/m2.
Water pump 50 50 50 50 50
TPV = Tamb + G (NOCT −20)/800 (1) Water tank + water + filters 300 300 300 300 300
Controller kit 50 50 50 50 50
Controlled vales 25 50 75 100 125
5.2. Cooled PV panel temperature Pipe system + Gutters 175 310 445 580 715
Sprinklers 40 80 120 160 200

To estimate the PV temperature with the water cooling system and Total Cost 640 840 1040 1240 1440
Cost per string 640 420 347 310 288
to obtain the optimal cycle of the prototype, the following solution is
proposed: by analysis of the experimental thermal constants of the
cooled panel temperature (Fig. 11), the cooling and heating of the PV
panel in the ON/OFF cycle could be described by exponential transient
curves.
Eqs. (2) and (3) were deduced, where each parameter is shown in
Table 2. Eq. (2) translates the cooling curve of the PV temperature, and
(3) it's heating.
t
Tcooling = (TPVncool−TPVcool ) e− τON + TPVcool (2)
t
( )
Theating = (TPVncool−TPVcool ) 1−e− τOFF + TPVcool
(3)
To validate the analytical model, the comparison between the ex-
perimental and estimated temperature was done. In Fig. 15 is presented
Fig. 16. An optimal number of strings for one or two pumps (a case study of Lisbon,
the experimental and analytical results for the PV temperature, with
Portugal, 2015).
different ON/OFF cycles in each hour, on 9th of August of 2016. The
analytical model is capable of predicting the behavior of the PV tem-
1200 120
perature with enough accuracy for different cycles of ON/OFF, for Income Increase Annual Water and Pump Costs [USD]
different ambient temperatures and solar radiations. Using the relative 1150 Revenue
Annual Income Increase

100
standard deviation formula, the deviation between the theoretical and Water Pump Energy Costs
and Revenue [USD]

1100
experimental results are about 1.6%. Water Costs
80
1050
6. Economic analysis and system optimization
1000 60
To fully understand the potentiality of this cooling system, an 950
40
900
Table 2
Model parameters meaning. 850 20
25 26 27 28 29 30 31 32 33 34 35
Parameters Start-up Temperature [ºC]

TPVncool Theoretical panel temperature Fig. 17. Annual income increase, revenue and water and energy pump costs with dif-
ferent start-up temperatures (Lisbon, Portugal, 2015).
TPVcool Sum of Tamb with a ΔT value. ΔT < 4 °C (was fitted to experimental
data)
τON Cooling thermal time constant of the panel, equal to 0.6 min (set economic analysis must be done, considering its energy increase, along
experimentally)
with its costs (water pump energy consumption and water losses). To
τOFF Heating thermal time constant of the panel, equal to 11 min (set
experimentally) have significant conclusions, this analysis must be done for at least a

1303
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

Fig. 18. Hourly average temperature of the PV


panels without (grey line) and with the cooling
system operating (black line) for the whole year
2015 (case study of Lisbon, Portugal).

Fig. 19. Daily number of hours with the cooling


system operating, for a start-up temperature of 30 °C
(case study of Lisbon, Portugal, 2015).

Table 4
Annual revenue for Lisbon, 2015. With a tank mounted at ambient temperature.

Number of strings

1 2 3 4 5

Energy production of non-cooled PVs 7.96 MWh 15.05 MWh 22.65 MWh 30.16 MWh 37.74 MWh
Increase of energy 12.31% 11.83% 11.31% 10.84% 10.40%
Annual income increase 245 USD 473 USD 680 USD 868 USD 1042 USD
Pump energy consumption (% of energy increase) 0.48 MWh (49%) 0.48 MWh (27%) 0.48 MWh (19%) 0.48 MWh (15%) 0.48 MWh (12%)
Annual Water costs 43 USD 43 USD 43 USD 43 USD 43 USD
Annual energy pump costs 82 USD 82 USD 82 USD 82 USD 82 USD

Annual revenue 120 USD 348 USD 555 USD 743 USD 917 USD

whole year. 6.1. Initial and operating costs


The economic analysis of the system was done based on the initial
investment and operating costs and the income obtained by the effi- The initial costs of the water cooling system were considered to be
ciency increase. Two case studies were analyzed for a PV power plant the same for both case studies (Lisbon and Beja). Using the same ma-
with 5 strings, each one with 25 panels, similar to the studied PV plant, terials costs, as the ones used in the prototype installed in the RESUL PV
located in (a) Lisbon, Portugal and (b) Beja, Portugal. Lisbon is located plant, one can calculate the total initial costs for each case-study,
near the sea, while Beja is an inner city where in the summer the Table 3.
temperatures are higher. The material costs depend on the number of strings to be cooled.
The system is optimized for each case, regarding the number of Costs related to the water pump, water tank (filters and water) and to
strings per water pump and the start-up temperature of the cooling the controller kit do not change the number of strings, while the con-
system. trolled valves, pipe system, and sprinklers are proportional to the

1304
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

chosen water cycle ON/OFF was 2/8 min.


Additionally, it was studied the impact of the start-up temperature
in the cooling system, i.e., the PV temperature from which the cooling
system starts. Fig. 17 show the annual income increase and revenues
and the water and energy pump costs for different start-up tempera-
tures. The behavior of the water and energy pump costs is expected; by
decreasing the start-up temperature the cooling system will be ON more
hours and, thus, more water and energy pump costs are expected.
The optimal operation points are between 28 °C and 30 °C, for the
case of Lisbon. With these, an annual revenue of 907 USD is estimated
with a gross income of 1034 USD and water and energy pump costs of
43 USD and 82 USD.
Figs. 18 and 19 present the evolution of the PV temperature, with
and without the cooling system, and the number of daily hours with the
cooling system ON, for a start-up temperature of 30 °C. Without the
Fig. 20. Financial return after 20 years for a case study of Lisbon, Portugal, 2015. cooling system, the hourly average PV temperature would be higher
than 50 °C in the months of summer, reaching around 60 °C in the
month of June and July. If we compare these results with the ones
obtained experimentally in Fig. 12, we can verify that, in fact, it is
possible to reach 60 °C in the months of summer, without the cooling
number of strings. As result, the initial estimated costs decrease per system.
string, as more strings are considered. Similarly, in these months, the number of working hours per day of
The operating costs were estimated to be 2–3% of the initial cost per the cooling system increase, reaching 13 h in June, corresponding al-
year (replacement of filters and sprinklers and other costs). most to the number of sunlight hours to this day. The gap shown be-
tween the 5th and 10th day of June corresponds to a “blackout” period
6.2. Case study: Portugal, Lisbon data - 2015 without temperature recordings.

In the first case study, it was considered a PV power plant located in 6.2.3. Annual revenue
Lisbon, Portugal. To estimate the total energy produced, with and After the optimization of the number of strings per pump and of the
without the cooling system, data was provided by the meteorological start-up temperature, the annual revenue and associated costs are now
institute in Instituto Superior Técnico, University of Lisbon. This data detailed. Table 4 shows the revenue results for the water cooling system
presents the hourly average ambient temperature and solar irradiation installed in Lisbon, Portugal in 2015, with 5 strings per water pump and
for the whole year of 2015. In this year, the Lisbon annual mean tem- with a start-up temperature of 30 °C. All calculations made are based on
perature was 16 °C and the annual solar radiation was around 1.7 kWh/ a panel without a routine cleaning, which can be considered as the
m2. typical operation of panels in Portugal. It was also considered a water
tank installed at ambient temperature.
6.2.1. Variable costs Since the system is to be assembled in already built PV plants, and
For the variable costs, the following considerations were made: their typical lifetime is about 25 years, the calculations were performed
for a period of 20 years. Fig. 20 shows, for this case, the financial return
• The system operates whenever the panel temperature exceeds of around 17000 USD after 20 years.
25–35 °C (depending on the chosen temperature). Additionally, it was also analyzed the case where the water tank is
• Water losses are 15 l/h (average experimental values). buried underground and, therefore, the water temperature is lower and
• The tariff of energy sold to the grid, produced by the panels, is remains almost constant along the year. In this particular case, it was
0.266 USD/kWh (same as in the RESUL PV plant). verified that the increase in energy is almost identical, concluding that
• The energy purchased from the grid to supply the pump is the water temperature is similar to the buried and unburied tank.
0.174 USD/kWh (same as in the RESUL PV plant).
• The water price is 1.614 USD/m (typical value for industrial use in
3
6.3. Case study: Beja, Portugal, 2015
Lisbon).
Considering other locations for the system, it was chosen Beja,
6.2.2. Water cooling system optimization Portugal, due to its different ambient temperature and solar irradiations
To maximize profit, it is necessary to calculate the optimal number range, with annual mean temperature of 22.4 °C and annual solar ra-
of strings that should be cooled by one water pump. Using the devel- diation of 1.8 kWh/m2. This location has a significantly higher tem-
oped analytical model, the PV panel temperature was estimated for two perature in the summer, where the influence of the temperature in the
cases: (1) without the cooling system and (2) with the cooling system PV efficiency is more critical. All considerations, purchase, and sale
starting when the PV temperature reaches 30 °C. For this case, it was tariffs were considered the same as in the previous case.
found that the water pump would be connected 1912 h per year and the
water and energy annually cost spent at the pump are 43 USD and 6.3.1. Water cooling system optimization
82 USD, respectively. These costs were considered independent of the The optimization results for the case of Beja indicate similar results
number of strings because only one string is cooled at a time. for a number of strings per water pump. Furthermore, the same optimal
Fig. 16 shows from which a number of strings it compensates to use start-up temperature points were obtained. These are also between
a second pump, i.e., two independent set of strings with different 28 °C and 30 °C. With these, an annual revenue of 1050 USD is esti-
cooling systems. For this case, one pump refrigerating 5 strings still has mated with a gross income of 1182 USD and water and energy pump
a higher revenue than two pumps cooling 5 strings (one pump cooling 2 costs of 46 USD and 86 USD (2110 h of the cooling system operating
strings and the other 3 strings). To use 5 strings with only one pump the time), (see Fig. 21).

1305
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

6.3.2. Annual revenue


Table 5 shows the revenue results for water cooling system installed
in Beja, Portugal in 2015. The calculations were done considering the
tank installed at ambient temperature. As it can be seen, despite Beja
having higher annual average solar radiation, without the cooling
system, the energy production of the PV plant would be lower than
Lisbon (37.45 MWh in Beja and 37.74 MWh in Lisbon) due to the higher
ambient temperatures. Although, the annual revenue of the cooling
system is higher than in the case of Lisbon.
Fig. 24 shows the financial return after 20 years for the case of Beja.
In this case, the financial return will be higher than in Lisbon (from
17,000 USD to 20,000 USD).
It was also considered that the water tank could be buried, assuming
a water temperature in the tank of 20 °C. This configuration allows the
Fig. 21. Annual income increase, revenue and water and energy pump cost with different water temperature to be lower than in the previous case and, thus,
start-up temperatures (case study of Beja, Portugal, 2015). originating a higher energy increase. The results are presented in
Fig. 25.

Fig. 22. Hourly average temperature of the PV pa-


nels without (grey line) and with the cooling system
operating (black line) for the whole year 2015 (case
study of Beja, Portugal).

Fig. 23. Daily number of hours with the cooling


system operating, for a start-up temperature of 30 °C
(case study of Beja, Portugal, 2015).

As expected, the estimated PV temperatures without the cooling This is the most interesting case for an economic analysis in Beja. It
system are higher than in the Lisbon case. Figs. 22 and 23 present the is observed that there is a financial return after 2 years and during the
evolution of the PV temperature, with and without the cooling system, PV lifetime, it is expected a profit of approximately 22,000 USD, with 5
and the number of daily hours with the cooling system ON, for a start- strings.
up temperature of 30 °C. In Beja, the PV temperatures can exceed 70 °C,
thus enhancing the importance of the cooling system. Setting a start-up 7. Conclusions
temperature of 30 °C, the estimated operation time of the cooling
system is 2110 h. A closed loop water cooling system was developed and optimized

1306
A.F.A. Castanheira et al. Applied Energy 211 (2018) 1297–1307

Table 5
Annual revenue for Beja, Portugal, 2015 with a tank mounted at ambient temperature.

Number of strings

1 2 3 4 5

Energy production of non-cooled plant 7.49 MWh 14.98 MWh 22.47 MWh 29.96 MWh 37.45 MWh
Increase of energy 13.03% 12.53% 11.96% 11.43% 10.96%
Annual income increase 275 USD 527 USD 755 USD 962 USD 1152 USD
Pump energy consumption (% of energy increase) 0.53 MWh (54%) 0.53 MWh (28%) 0.53 MWh (20%) 0.53 MWh (16%) 0.53 MWh (13%)
Annual water costs 46 USD 46 USD 46 USD 46 USD 46 USD
Annual energy pump costs 86 USD 86 USD 86 USD 86 USD 86 USD

Annual revenue 125 USD 377 USD 605 USD 812 USD 1002 USD

Fig. 25. Financial return after 20 years for a case study of Barreiras, Brazil, 2016. Tank
buried and water at 20 °C.

Fig. 24. Financial return after 20 years for a case study of Beja, Portugal, 2015. Tank
intensity and temperature dependency of photovoltaic cell parameters. Appl Energy
installed at ambient temperature.
2013;111:374–82.
[3] Nicolas B, Raka J, Saïd A, Mohammad A. PV panel single and double diode models:
Optimization of the parameters and temperature dependence. Sol Energy Mater Sol
for already installed photovoltaic (PV) plants, to reduce the negative Cells 2016;148:87–98.
impact of the temperature on their efficiency. Results from a pilot test [4] Zhu L, Raman A, Wang KX, et al. Radiative cooling of solar cells. Optica 1
in a 5 kW PV power plant showed that, with an optimized water flow 2014;1:32–8.
[5] Ebrahimi M, Rahimi M, Rahimi A. An experimental study on using natural vapor-
and ON/OFF cycles, it is possible to increase the PV annual production ization for cooling of a photovoltaic solar cell. Int Commun Heat Mass Transf
by 12%, reaching to 17% of increase peaks (reducing the PV tem- 2015;65:22–30.
perature from around 60 °C to 30 °C) and with water losses between 10 [6] Chandel SS, Agarwal Tanya. Review of cooling techniques using phase change
materials for enhancing efficiency of photovoltaic power systems. Renew Sustain
and 20 l/h.
Energy Rev 2017;73:1342–51.
A theoretical thermal model based on exponential transients to [7] Sweidan A, Ghaddar N, Ghali K. Optimized design and operation of heat-pipe
model the cooling/heating of the PV panel is sufficiently accurate to photovoltaic thermal system with phase change material for thermal storage. J
predict the PV temperature based on meteorological data. Based on this Renew Sustain Energy 2016;8:023501.
[8] Elbreki AM, Alghoul MA, Sopian K, Hussein T. Towards adopting passive heat
model, two case-studies were analyzed: 25 kW system installed in dissipation approaches for temperature regulation of PV module as a sustainable
Lisbon and installed in Beja, Portugal. For these cases, the best eco- solution. Renew Sustain Energy Rev 2017;69:961–1017.
nomic results are obtained with a start-up temperature of 30 °C and [9] Chandrasekar M, Senthilkumar T. Experimental demonstration of enhanced solar
energy utilization in flat PV (photovoltaic) modules cooled by heat spreaders in
with 5 strings of 5 kW per water pump. Economic results show that the conjunction with cotton wick structures. Energy 2015;90:1401–10.
cooling system has a payback time of 2 years and have a 20-years [10] Moharram K, Abd-Elhady M, Kandil H, El-Sherif H. Enhancing the performance of
revenue of 17000 USD, for the case of Lisbon, and around 22000 USD photovoltaic panels by water cooling. Ain Shams Eng J 2013;4(4):869–77.
[11] Fabio S, Alberto B, Anna S, Nicola D. Improving photovoltaics efficiency by water
for the case of Beja. cooling: Modelling and experimental approach. Energy 2017.
[12] Ankita G, Tiwari GN. Performance of a-Si thin film PV modules with and without
Acknowledgments water flow: an experimental validation. Appl Energy 2014;128:184–91.
[13] Zhijun P, Mohammad R, Yiming L. Cooled solar PV panels for output energy effi-
ciency optimization. Energy Convers Manage 2017;150:949–55.
This work was supported by FCT, through IDMEC, under LAETA, [14] Aarti K, Vishal V, Bhim S. Optimization of thermoelectric cooling technology for an
project UID/EMS/50022/2013. The authors also acknowledge RESUL, active cooling of photovoltaic panel. Renew Sustain Energy Rev 2017;75:1295–305.
[15] Siddiqui MU, Arif AFM. Electrical, thermal and structural performance of a cooled
Equipamentos de Energia S.A. for its support on the installation of the
PV module: transient analysis using a multiphysics model. Appl Energy
cooling system in their PV power plant. 2013;112:300–12.
[16] Nižetić S, Čoko D, Yadav A, et al. Water spray cooling technique applied on a
References photovoltaic panel: the performance response. Energy Convers Manage
2016;108:287–96.
[17] Sandro N, et al. Experimental and numerical investigation of a backside convective
[1] Hasanuzzaman M, et al. Global advancement of cooling technologies for PV sys- cooling mechanism on photovoltaic panels. Energy 2016;111:211–25.
tems: a review. Sol Energy 2016;137:25–45. [18] Ross Jr RG. Interface design considerations for terrestrial solar cell modules. In:
[2] Erdem C, Pinar Mert C, Bali, Tulin B. An experimental analysis of illumination 12th Photovoltaic specialists conference; 1976. p. 801–6.

1307

Potrebbero piacerti anche