You are on page 1of 16

Energy Conversion and Management 185 (2019) 800–815

Contents lists available at ScienceDirect

Energy Conversion and Management


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

Experimental investigations on PV cleaning of large-scale solar power plants T


in desert climates: Comparison of cleaning techniques for drone retrofitting
Mohammed Al-Housani, Yusuf Bicer , Muammer Koç

Division of Sustainable Development (DSD), College of Science and Engineering (CSE), Hamad Bin Khalifa University (HBKU), Qatar Foundation (QF), Education City,
Doha, Qatar

ARTICLE INFO ABSTRACT

Keywords: This study experimentally investigates the effectiveness of various PV cleaning techniques for potential retro-
Solar energy fitting into unmanned aerial vehicles, drones, for large-scale solar power plant cleaning operations in desert
Photovoltaics climates. The ultimate objective of this study is to improve the efficiency, reduce the cost, time and environ-
Thin film mental impact of cleaning techniques by integrating them with autonomous drone cleaning to ensure cost-
Cleaning
effectiveness and competitiveness of Solar PV installations. This study considers the following factors: (i)
Soiling
Drone
cleaning effectiveness (power output change in W and %), (ii) frequency of cleaning (daily, weekly, and
monthly), (iii) overall cost of cleaning (USD/m2), (iv) adaptability to drone, (v) energy consumption, and (vi)
weight. The cleaning techniques tested for their effectiveness and overall cost are (1) cloth wipers, (2) brushes,
(3) vacuum cleaners, and (4) some combinations. Cleaning frequency is varied from 1-day to 1-month intervals
to find the optimal cleaning period. The results of this study show that there is a significant reduction in the PV
efficiency, and hence an increase in the cost of produced electricity, for monthly cleaning periods regardless of
cleaning technique in desert climate conditions. The microfiber based-cloth wiper is found to be the optimum
method from both cost and performance point of view. Microfiber based-cloth wiper achieved on average 7.7%
and 3.1% performance improvement (compared to control panel) for weekly cleaning frequency in winter and
summer seasons, respectively. The assessment of drone retrofitting was also conducted, which resulted that the
brush and microfiber based-cloth wiper with their low weight, small size, and ease of use are best-suited options
for drone-based solar panel cleaning. The evaluation factors for microfiber based-cloth wiper are found to be (i)
3.1% for cleaning effectiveness in summer, (ii) 1 week for frequency of cleaning, (iii) 0.41 USD/m2 for overall
cost of cleaning, (iv) 9.5 (out of 10) for adaptability to drone, (v) 10 (out of 10) for energy consumption and (vi)
9 (out of 10) for weight.

1. Introduction The dust problems mainly occur in desert areas, whereas soiling can
occur in many other locations since it also includes snow and other
Production of electricity from clean energy sources is a critical mi- particles. Dust accumulation on PV panel surfaces disturbs the trans-
tigation strategy to overcome the global warming challenge. The mission of light, which adversely affects the efficiency of the solar pa-
countries located in the sunbelt region and the Middle East have higher nels and the overall performance of the system [1]. Soiling of PV panels
solar energy potentials than many other regions in the world, however, is known to cause significant performance and revenue losses in solar
with a significant disadvantage of soiling, which necessitates frequent power installations. Furthermore, dust compounded with humidity can
and costly cleaning of solar panels. The main challenge of implementing form the micro surface structures on PV panels, which are difficult to
solar energy systems in these hot dry climates is dust and soiling on the remove using basic cleaning techniques particularly if water usage
surface of the panels and collectors in addition to high-temperature needs to be avoided. It was found that dust accumulation zones are
levels. Both reduce the efficiency of solar PVpanels significantly and highly populated in the Middle East and North Africa regions [2].
hence increase the cost of electricity produced, making solar energy Therefore, there has been a significant increase in the number of recent
systems not competitive enough with abundant fossil fuel resources publications regarding PV soiling and mitigation strategies [3].
such as natural gas. Hence, this study attempts to address how to deal with the PV


Corresponding author.
E-mail addresses: malhousani@hbku.edu.qa (M. Al-Housani), ybicer@hbku.edu.qa (Y. Bicer), mkoc@hbku.edu.qa (M. Koç).

https://doi.org/10.1016/j.enconman.2019.01.058
Received 13 September 2018; Accepted 21 January 2019
0196-8904/ © 2019 Elsevier Ltd. All rights reserved.
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Nomenclature GPOA Global plane of array


MPPT Maximum power point tracking
I Current (A) NASA National Aeronautics and Space Administration
Ir Irradiance (W/m2) PV Photovoltaic
P Power (W) QSTP Qatar Science and Technology Park
T Temperature (C) STF Solar Test Facility
V Voltage (V)
v Wind velocity (m/s) Subscripts

Acronyms C Cell
Max Maximum
CA Contact angle Mpp Maximum power point
CdS Cadmium Sulfide o Ambient
CdTe Cadmium telluride oc Open circuit
DHI Direct horizontal irradiance POA Plane of array
DNI Direct normal irradiance sc Short circuit

soiling issues and remove the dust on the surface of the PV panels by this study. Syafiq et al. [7] also compared the number of recent pub-
bringing an alternative autonomous and low-cost solution for large- lications about various type of cleaning techniques. They have observed
scale solar power plants to ensure reliable and low-cost electricity that cleaning-robots presented the highest number of publications
generation. Several PV cleaning techniques are first tested in a solar among other techniques (electrodynamic dust-shield, self-cleaning su-
facility through long-term experimentation, and then, they are assessed perhydrophobic glass, and self-cleaning superhydrophilic glass) be-
in terms of various aspects for retrofitting into unmanned aerial ve- tween 2009 and 2017 with a total number of 9423 original research
hicles. articles. Consequently, the existing trend shows that researchers are
In the following sections, we present the importance of PV soiling more confident about the mechanical-cleaning method over others. In
and numerous solutions conducted by many researchers around the this study, we improve the mechanical cleaning process by proposing
world. drones and terminating water use, which can further enhance the
practicality of the proposed PV cleaning technique.
Jing et al. [8] studied the effect of wind on the PV cleaning process.
1.1. The significance of PV soiling
They included the particle resuspension theory to estimate the wind
cleaning process of the dust particles deposited on the flat solar pho-
According to the study Guo et al. [1] performed in Doha, Qatar,
tovoltaic (PV) module surfaces. They found that cleaning process for
solar panels lose around 0.4–0.8% of their power output per day. This small particles via natural wind was not effective, because mainly the
means that a PV panel can lose about 12–24% of its power output if it is
large shear velocity was needed for detachment of the particles from the
not cleaned for a month. Al-Shehri et al. [4] evaluated the solar PV surface. Therefore, the cleaning of the panels naturally by the wind
cleaning efficiency for different types of brush materials of nylon, cloth,
cannot be a feasible solution for climates such as Qatar where humidity
and silicon rubber foam. They concluded that while some materials had also plays an adverse role, causing sticking of dust particles (in some
a notable impact on the solar panels, no permanent or significant ne-
kind of micro mud bumps) on the panels.
gative impact was found to affect the efficiency of solar panels. How- Dust is not a problem only for PV modules but also for solar col-
ever, the cleaning efficiency of the nylon brushes was not as high as
lectors. Li et al. [9] experimentally investigated the impact of dust on
cleaning using water and delicate wipers. In addition, they concluded in the absorptivity and emissivity of absorber plates of solar collectors.
their further study [5] that selection of the optimal brush is essential to
They tested the dust deposition and simulated the dust deposition
reach the required quality of cleaning while preventing damage on the density over absorptivity reduction. The maximum efficiency reduction
surface of the solar panels. Deb and Brahmbhatt [6] proposed an au-
in the solar collector was obtained as 19.23% on the 110th day of dust
tonomous brush-based PV cleaning machine working without water. exposure, which is an average dust deposition density value of surface
The device is placed on the PV surface and covers the whole length. It
7.07 g/m2. Elminir et al. [10] tested 100 glass samples for dusting on a
moves parallel to panels and sweeps the row. By using frames, they transparent cover of solar collectors. They evaluated the transmittance
eliminate the need for rails. This is an important advantage. However,
of the glass over a period of about seven months and found out that the
at the end of the row, the device needs to be shifted to another row. The dust deposition density ranges from 15.84 g/m2 (tilt angle of 0) to
device works with a battery powered microcontroller. The main dif-
4.48 g/m2 (tilt angle of 90 and 135 deviation from the north). It was
ference between their study and this study is the employment of au- approximately 52.54–12.38% reduction in transmittance. Furthermore,
tonomous flying vehicle, drone, which eliminates (i) the construction of
Xu et al. [11] proposed a model to estimate the optimal tilt angle for
cleaning frame, (ii) placement on the PV surface and provides addi- minimum soiling effects. They also validated the model with experi-
tional benefits of (i) highly flexible mobility, (ii) cleaning on-demand
mental results. The results of their study showed that the output voltage
basis and (iii) installation flexibility of cleaning apparatus. declines by over 60% when the dust deposition density is about 4.71 g/
Syafiq et al. [7] prepared a review paper focusing on self-cleaning
m2.
surfaces for PV cleaning applications. The main idea of self-cleaning
systems is to eliminate external cleaning. However, the cost, as well as
preparation of self-cleaning surface, remain as a challenge on the im- 1.2. PV cleaning solutions
plementation of the method. As also concluded by Syafiq et al. [7], PV
glass coatings still require significant improvement to be applied at There are several cleaning alternatives mentioned in the literature
outdoors in terms of performance reliability. Many aspects and mod- ranging from manual to robotic-based autonomous types. Some of them
ification still need attention. Hence, adapting an already commercially are not suitable for hot desert climates as they require intensive water
available method (brush/microfiber cloth) and converting into water- use, which is a scarce resource. Freshwater is commonly produced from
free and human-free cleaning is a promising approach as reported in water desalination, which needs high thermal or electrical energy

801
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

input. The most basic form of solar panel cleaning is mechanical in Chile. They found out that monthly cleaning and random rainfall
cleaning, which includes all sorts of cleaning robots and manual labor. showed positive impacts as main solutions. Additionally, based on the
Generally, for large solar facilities, a custom-designed robotic system is various energy prices, they defined a critical cleaning period of 45 days
proposed. This type of systems focuses on cleaning the largest area with for a real case, independent on cleaning cost and energy prices.
the least amount of time and usually require few operators [12]. There In brief, reduced electricity production efficiencies of PV panels due
are smaller robots containing brush with less water usage. Although to dust accumulation has directed various researchers into the in-
these robots offer autonomous cleaning and are usually very robust, vestigation of effective and affordable PV cleaning methods. Existing
they require (i) customization, (ii) a significant amount of costly in- PV cleaning methods can get quite expensive, and not effective since
stallation such as special rails, and (iii) complex maintenance under hot most of them require water and fixed installations. The fixed cleaning
climate conditions. There is another type, which is the sprinkler-based installations increase the capital investment cost, at the same time,
robot, using more water to clean the panels. Although they work ef- some of them use fossil fuels to run the cleaning. This is a contradictory
fectively in dry sandy areas, the main disadvantage is large amounts of issue, where a clean energy source is utilized but at the same time, fossil
water requirement. Thus, currently, manual labor is commonly em- fuel is consumed. Water is scarce in hot dry climatic regions such as
ployed for cleaning solar panels using water [12,13]. Qatar. Hence, consuming valuable water resource for PV cleaning ap-
There is also electrical curtain method for cleaning PV panels, in plications contradicts with the ultimate goals of sustainability from
which an electrical wave is applied creating a traveling wave that economic, environmental and social perspectives. Therefore, this study
would prevent any particle to settle on the PV surface. In order to create attempts addressing these issues based on a real need for cost-effective,
this traveling wave, some parallel electrodes would need to be em- efficient, flexible, automated PV cleaning method and technology.
bedded in a dielectric surface of the panel [14], making this solution
costly in addition to some safety concerns during rainy days.
Coating of the PV surfaces with anti-dust materials is another way of 1.3. Drone-based PV cleaning
dust protection. There are two types of self-cleaning nano-films; super-
hydrophilicity and super-hydrophobic films. The popular type of In this research, the use of flexible systems, such as drones, is con-
coating film is TiO2 [15]. The solution requires water, which is not well sidered as an alternative to fixed robotic installations and/or manual
suited for desert climates [16]. Arabatzis et al. [17] monitored the cleaning methods. In this study, there are still brush/vacuum/micro-
outdoor performance of coated and uncoated PV panels and arrays for fiber cloth, which are commercially available and practical cleaning
several months at different climate conditions, where they used a self- apparatuses. The motivation for utilizing drone combines the strength
cleaning, photocatalytic, anti-reflective glass coating. In their study, the of autonomous and human-free cleaning without using water. This
coated surface showed better dust removal ability due to its super- considerably affects the overall cost of cleaning when large-scale PV
hydrophilicity, in which one step application could on average achieve power plants are considered. It is important to understand that the
5–6% power gain for long-term operation. Zhong et al. [18] also studied drone is the platform, which carries different loads such as cleaning
a novel super-hydrophilic coating with high stability and corrosion tools. The power source is one of the most important components of a
resistance to be applied in PV modules. The coating material, TiO2/ drone. Having the right power source can determine or limit all other
silane coupling agent, can be deposited on the surface of PV panels by aspects and characteristics of the drone. A high and lightweight power
spraying for large-scale applications. According to Piliougine et al. [19], source means the drone can fly higher, faster, longer distances, while
the modules with coating film have an average daily soiling loss of 2.5% carrying heavier payloads.
whereas for the uncoated modules, this loss is about 3.3%. Soiling of Drones emerge as a suitable tool for the solar industry because of
bifacial PV panels has also recently been investigated by Luque et al many advantages such as a wide range of surveillance capability, long
[20]. It was observed that the soiling rate in the monofacial mini- range inspection, and efficient data logging capability [24]. The recent
module is about 0.301%/day, whereas, a rate of 0.236%/day was found studies proposed using drones for PV surveillance applications to re-
for the bifacial module [20]. duce the human workforce [24], as well as aerial thermographic in-
For some regions having regular rains, a 5 mm of rainfall can be spection of photovoltaic plants [25]. Similarly, drones have been ap-
evaluated as sufficient to clean solar panels and restore its production plied in construction and buildings industry [26,27]. There are even
capacity [21]. Nevertheless, in desert climates like Middle East, rainfall some studies to find the optimal location by using IR imaging ther-
is scarce, especially in the April-November period. This means that the mographic cameras on the drone [28]. In addition, monitoring of PV
plant/facility owner needs to resort to human intervention to clean the modules was carried out for identifying irregularities of PV panels in a
panels. Because, leaving them to natural conditions will cause dete- solar power plant via an infrared measurement aerial system [29].
rioration on panel performance. Moharram et al. [22] studied removing Drones have already been proposed for maintenance of wind turbine
the dust particles on the PV panels with a non-pressurized water system blades and it is considered as one of the future research areas in wind
in Cairo, Egypt. They found that the efficiency of the PV panels de- energy sector [30] although there are some legal requirements to be
creased by significantly after 45 days of cleaning using non-pressurized satisfied with using drones in special industries such as military and
water, while the efficiency remained constant when anionic and ca- mapping [31].
tionic surfactants were used for cleaning. They proposed that a com- Using drones in PV cleaning applications brings many potential
bination of surfactants diminish the volume of water desired for advantages to the industry along with some disadvantages. Table 1
cleaning as well as the energy for spraying the water [22]. On the other below shows some potential advantages and disadvantages of using
hand, Urrejola et al. [23] prepared a comprehensive study considering drones for PV cleaning applications.
the effects of soiling on thin-film and crystalline type PV panels located With the rapid growth of the drone/UAV industry, many new ap-
plications for the drones are emerging. This rapid expansion of

Table 1
Potential advantages and disadvantages of drones in PV cleaning applications.
Advantages Disadvantages

• Automation of cleaning procedures • Capital cost and short flight time


• The high mobility of system, which allows its use on multiple solar farms or installations • New field of applications requiring research
• Cleaning can be water free • Requires electrical or fuel recharging stations
802
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Fig. 1. (a) Mechanical brush [35], (b) Microfiber based cloth wiper [36], (c) Vacuum cleaner [37].

Table 2 Table 4
Ratings of thin-film solar panels used in the experiments at standard testing Uncertainty values of PV measurement instruments.
conditions.
PV Parameter Uncertainty
Ratings at Standard Conditions Value Unit
DC Voltage 0.5%
Nominal power (+/−5%) 85 W DC Current
Voltage at Pmax 48.5 V DC Power 0.5%
Current at Pmax 1.76 A AC Voltage 0.5%
Open circuit voltage 61 V AC Current 0.5%
Short circuit current 1.98 A AC Power 0.5%
Maximum system voltage 1000 V Module back surface temperature 0.4%
Temperature coefficient of Pmpp −0.25 %/°C
Temperature coefficient of Voc, high temp (> 25 °C) −0.27 %/°C
Temperature coefficient of Voc, low temp (-40 °C to + 25 °C) −0.20 %/°C Table 5
Temperature coefficient of Isc +0.04 %/°C
Uncertainty values of meteorological measurement instruments.
Limiting reverse current 3.5 A
Meteo Parameter Uncertainty

applications has inspired this research and experimentation to de- Irradiance – Global Horizontal (GHI) < 2%
Irradiance – Plane of Array (22° tilt) (POA) < 2%
termine the current usability of drones in the solar PV cleaning appli- Irradiance – Total Normal (TNI) < 2%
cations. The solar cleaning application is chosen due to the presence of Irradiance – Direct Normal (DNI) 1%
a need to overcome those challenges, the fast expansion of the whole Wind speed, 2 m, and 5 m above ground Greater of 2.5% or 0.4 m/s
solar industry, and the theoretical suitability of the drones to be used in Air temperature 0.2 °C
cleaning applications.
Drone technologies have been developing rapidly in recent years,
The main disadvantages of drone are the short flight time and the
which direct researchers to pursue novel studies in this field. Drones
required recharging periods. These disadvantages can be overcome my
have the potential to revolutionize the way solar panels are being
customizing the power source of the drone and by optimizing the
cleaned, similar to the way it transformed other fields in the near past.
cleaning schedule. The most important advantage of using a drone for
A couple of commercial products are available by some companies to
PV cleaning is that it is an autonomous, mobile, hence flexible solution,
clean the solar panels. One of them is named as Cleandrone [32], which
and it can be applied anywhere on-need basis, thus eliminating costly
uses an onboard glass cleaning device and detachable cleaning fluid
fixed installations as in robotic cleaning methods. This can significantly
container. In this one, they use a liquid for cleaning. They also propose
reduce the cost of cleaning by eliminating the robot arms, robot trails,
to use it for concentrated solar energy applications to clean the mirrors.
and water use and by providing flexible operation whenever required.
In addition, another company proposed so-called Solarbrush [33] by
Drones can also offer water-free cleaning based on the cleaning tools
integrating a brush as a tail of a drone. It is observed that the com-
used. These advantages are the main motivations to investigate the
mercial companies are adapting the drones by integrating a flexible
potential use of a drone in PV cleaning applications.
robotic arm, in which there is a cleaning apparatus at the end. This is
However, first of all, the most suitable cleaning techniques, tools,
the most common way. However, depending on the cleaning apparatus,
and frequencies need to be determined for specific weather and climate
the mounting structure may vary.

Table 3
Employed cleaning plan for thin-film PV panels in summer.

803
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Table 6 for cleaning technology, time, frequency, cost, and effectiveness. This
The total calculated uncertainty of the PV performance and meteorological research initially tests the effects of various cleaning techniques on the
variables in PV cleaning. performance of thin-film photovoltaic panels in Doha, Qatar under both
PV Performance Parameter Uncertainty (%) moderate winter and hot summer conditions. The cleaning techniques
in this study are selected based on the fact that they will be retrofitted
DC Voltage 0,50 on a drone/UAV (unmanned aerial vehicle) to clean the panels in a
DC Current 0,50
mobile, autonomous, and efficient manner. All of the selected cleaning
DC Power 0,50
Module back surface temperature 0,40 techniques require zero water. There have been some studies of PV
Total Uncertainty 0,95 cleaning for the silicon type panels, but the focus of this study is thin-
Meteorological Parameters film type solar panels. In addition, the cleaning apparatuses used in this
Irradiance 2,0 study are commercially available and cost-effective combinations,
Wind speed 2,5 which were not previously tested.
Air temperature 1,0 The specific objectives of this study can be summarized as follows:
Total Uncertainty 3,35

• Determine the most efficient cleaning frequency for hot desert cli-
conditions. This research aims to experimentally determine the most mates in summer and winter conditions
suitable cleaning techniques, tools and frequencies to be integrated into • Measure the soiling/dust accumulation on the panels in hot desert
drone-based PV cleaning. climates
Although there are few scientific papers related to the use of drones • Assess the selected PV cleaning methods from a cost and cleaning
for PV inspection and surveillance applications [24,34], there is a lack performance point of view
of scientific studies to assess the utilization of drones for PV cleaning • Evaluate the most suitable cleaning method to be retrofitted to a
applications. In this study, there is a long-term experimental quantifi- drone
cation of cleaning performance of the selected PV cleaning techniques. • Evaluate the PV cleaning methods’ integrality and adaptability to
In addition, this study is unique to assess the adaptability of these drones for drone-based large-scale solar power plant cleaning.
cleaning solutions into a drone by considering diverse parameters
ranging from cleaning effectiveness, payload to size and cost. The structure of this paper is as follows: Introduction section pre-
There is a necessity to define optimal cleaning procedure accounting sents an overview of PV cleaning and emerging methods. A literature
review is conducted to expand the understanding of latest available

Fig. 2. Comparison of daily power outputs of CdTe thin-film PV panels cleaned by various PV cleaning techniques as well as control panel (a) summer, (b) winter.

804
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

technologies in both photovoltaic (PV) industry and drone industry, • Solar panel 3 (PV26): Mechanical brush and Vacuum cleaner
and help integrate them in this research. The second section describes • Solar panel 4 (PV25): Microfiber based-cloth wiper and Vacuum
the experimental methodology to select the cleaning techniques to be cleaner
evaluated. The cleaning techniques are selected based on the under- • Solar panel 5 (PV30): Control panel (no cleaning)
standing established from the literature. After that, the most suitable
location for the experiments is selected, which was Solar Test Facility Five thin-film solar modules are installed on a platform tilted 20-
(STF) in Qatar, having targeted thin-film PV panels. The experiments degrees to south direction, where the specifications are tabulated in
are conducted to study the effect of different cleaning techniques, dif- Table 2. The outputs of these solar panels are compared before and after
ferent frequencies, the effect of temperature, and the effect of soiling on the cleaning in order to measure the change in performance. The panels
the panels. The third section presents and discusses the findings fol- are cleaned at the same time of the day. Visual changes are also re-
lowed with the conclusions and recommended future work. corded using a camera, and the amount of dust removed is measured on
glass coupons located on a separate PV panel (This PV panel is not used
2. Experimental methodology and analysis in the experiments).
The evaluation process is briefly described below:
In the experiments, the most suitable cleaning apparatuses are de-
termined based on energy usage, cleaning efficiency and cleaning time • Determination of the different cleaning technologies that can be
for different types of dust accumulation. Defining the appropriate used for the drones (Microfiber based-cloth wiper, brush, vacuum
cleaning frequency is also necessary, which may vary depending on cleaner, and combinations).
different apparatuses. Some of the cleaning techniques are combined to • Testing of each cleaning tool separately, in order to find out the
observe whether there will be a reduction in individual deficiencies. most suitable technology with the highest efficiency and applic-
The selected cleaning apparatuses, as well as their associated solar ability for drones.
panel numbers, are listed below and shown in Fig. 1: • Determination of cleaning capabilities for different dust levels.
• Calculation of energy needs and efficiencies for each cleaning
• Solar panel 1 (PV28): Mechanical brush technique
• Solar panel 2 (PV27): Microfiber based-cloth wiper • Investigation of equipping a drone with the cleaning apparatuses

Fig. 3. Comparison of normalized daily power outputs from CdTe thin-film PV panels cleaned by various PV cleaning techniques (a) summer, (b) winter.

805
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Fig. 4. Average daily power output changes of CdTe thin-film PV panels (average of all cleaning techniques with respect to control panel) (a) summer, (b) winter.

taking into account the most suitable cleaning system nearby, which may increase the dust levels. This is, in fact, a better
condition for the experimentation since the purpose of the study is to
The experimental flow is described as follows: remove the dust from the PV panel surface in an efficient way.

• Four PV panels were cleaned using different cleaning techniques at 2.1. Experimental uncertainty
the same time of the day around 11:00 a.m.
• The cleaning process is conducted without any water or liquid. The uncertainties of the measurement equipment are significant for
• The cleaning apparatuses are applied on the PV surface downwards determining the accuracy of the results. In many cases, the measured
direction. variables have a random variability, which is referred to its uncertainty.
• The fifth PV panel was not cleaned and kept as a control panel. Assuming the individual measurements are uncorrelated and random,
• The cleaned PV panels were compared to the control panel to de- the uncertainty in the calculated quantity can be determined as
termine the improvement rate.
• The experiments were conducted on daily, weekly and monthly Uy =
y 2
Ux2
cycles. x (1)
• The experiments were conducted both in winter and summer sea-
i

sons of the desert climate. where U represents the uncertainty of the variable.
In this section, the uncertainty levels of each measuring device in-
The frequency of cleaning is selected to be (i) daily for the first cluding PV performance parameters and metrological parameters are
week, (ii) weekly for four weeks, and (iii) monthly for two months as reported. Table 4 shows the uncertainty in the power measurement of
shown in Table 3. The measurements are taken from January to July in the solar panels. Table 5 lists the uncertainty values of the utilized
2018 to consider both moderate winter and harsh summer conditions in meteorological parameters.
Qatar. The changes in performance are recorded 24/7 h, using a data The meteorological parameters used in the analysis are total irra-
logger. diance, wind speed, and air temperature. Therefore, these selected
The experiments are conducted in the Solar Test Facility (STF) in parameters are accounted for the total uncertainty calculations as listed
Qatar Science and Technology Park (QSTP). The facility is a small solar in Table 6. The total uncertainty for the meteorological parameters is
farm with different types of solar panels being tested on a gravel/sandy calculated to be 3.35%, whereas it is 0.95% for the PV performance
area in the middle of big buildings. There are a few construction sites measurement parameters as reported in Table 6.

806
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Fig. 5. Comparison of weekly power outputs of CdTe thin-film PV panels cleaned by various PV cleaning techniques as well as control panel (a) summer, (b) winter.

2.2. Cost considerations • The PV panel area is 0.72 m . 2

• The weekly power losses in percentage are calculated from the ex-
The overall cost of cleaning plays a significant role in determining perimental data in the winter season to be 3.42%, 2.95%, 3.63%,
the feasibility of PV cleaning techniques. In order to establish the most 2.28% respectively for Microfiber based-cloth wiper + Vacuum
feasible cleaning technique, a study comparing different operational cleaner, Mechanical brush + Vacuum cleaner, Microfiber based-
costs of the selected PV cleaning techniques are also performed based cloth wiper, Mechanical brush.
on capitals cost, operational cost, and equipment lifetime with rea-
sonable assumptions. Note that drone related costs are not included in 2.3. Drone retrofitting assessment
the cost assessment. The costs are calculated for weekly cleaning
period. The assumptions used in the cost calculations are listed below: In this section, the drones are assessed based on the cleaning tech-
niques suitability. These cleaning methods were selected based on
• The capital costs of the cleaning equipment are normalized for one specific requirements. These requirements are listed as follows:
year considering the lifetime of the equipment.
• The lifetime of brush and microfiber cloth is taken as 1 year to find • Cleaning capabilities for different dust types and levels
the normalized capital cost of cleaning equipment. Therefore, they • Energy needs and efficiency
are used 52 times in a year for a panel. • Weight
• The lifetime of vacuum cleaner is taken as 5 years to find the nor- • Size
malized capital cost of cleaning equipment. Therefore, it is used 260
times in 5 years.
• Ability to be retrofitted to an autonomous cleaning apparatus
• Electricity production and consumption cost are taken to be The logic behind selecting these requirements is a future plan in
0.09 QR/kWh (0.025 USD/kWh). which the most suitable cleaning method will be used over a drone to
• Manpower cost is assumed as 55 QR/hour (15.1 USD/hour). The clean the panels. Thus, these characteristics are important for the se-
cost of manpower per panel is calculated based on the duration of lection of suitable cleaning technique. The importance of each assess-
cleaning. ment parameter is highlighted as follows:
• Vacuum cleaner power rating is 50 W. The cost of electricity is
calculated using the duration of cleaning. 2.3.1. Weight
• Average daily sunny hours is taken as 5.5 h in Qatar in order to find The total weight of the drones is very significant to determine the
the energy cost lost due to one-week soiling on the panel. flight time. Payload capacity of drones is quite diverse ranging from

807
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Fig. 6. Comparison of normalized weekly power outputs from CdTe thin-film PV panels cleaned by various PV cleaning techniques (a) summer, (b) winter.

grams to many kilograms. However, for regular battery-powered 2.3.4. Adaptability


drones, the flight time is very short ranging from several minutes to one Some cleaning apparatuses are difficult to be fixed on the drones
hour. The short flight time is directly related to the total weight of such as vacuum cleaners. On the other hand, a simple cloth wiper is
drones. Furthermore, the cleaning apparatus weight is an extra load, quite easy to be fixed on the drones. This is considered as the adapt-
which will even lower the flight time. Therefore, the weight is selected ability of cleaning apparatus on the drone. Therefore, it is selected as
as one of the evaluation parameters for drone adaptability. one of the evaluation criteria for the overall comparison.
The weighting factors of each evaluation criteria are selected equal
2.3.2. Size since all of them play significant roles. One can also determine varying
Size of a drone and cleaning apparatus determines the flexibility as weighting factors by prioritizing energy consumption, weight or size.
well as flight time. The dimensions of the drone and cleaning apparatus This can be selected based on the specific application and location.
should not be oversized, which will limit the cleaning process on the PV
surface. In case, the drone size is very large, it would be difficult to 3. Results and discussion
approach the PV panel surface. In addition, larger size drones require
more fuel/energy to be consumed during the flight. Therefore, there is In this section, the main findings of the PV cleaning experiments
an optimum size for this type of applications. during both winter and summer seasons in Qatar are presented.
Fig. 2 shows the power output of each panel during the “daily
2.3.3. Energy consumption cleaning frequency”. The figure shows an improvement in all of the
Although both size and weight characteristics mainly define the panels regardless of cleaning method except the for the control panel. In
energy consumption of drones, they are not the sole parameters. Energy the winter period, the variations in the daily panel performance during
consumption of drones is also related to their motors, efficiencies, the period of 28th January to 1st February is due to cloudy days. In the
battery and many other factors. If the selected cleaning techniques also summer, during the three days between 29th June and 2nd July, the
consume energy during operation, this can cause very short flight time. daily cleaning could not be carried out due to logistical issues in Solar
Therefore, energy consumption of the cleaning process is another cri- Testing Facility. Microfiber based-cloth wiper showed higher power
tical parameter for determining the suitability of the cleaning method outputs in the summer season as shown in Fig. 2a, although brush
for drone applications. performed better in the winter season. This fact is due to humidity

808
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Fig. 7. Average weekly power output changes of CdTe thin-film PV panels (average of all cleaning techniques with respect to control panel) (a) summer, (b) winter.

increase in the winter time. The humidity causes dust to stick on the frequency. The improvement in the cleaned panels shows once every
surface of the PV. In this case, a larger force is required to be applied for week, whereas the variations in daily outputs are attributed to cloudy/
removing the muddy dust. dusty days.
The recorded total irradiance (Plane of Array) values of each day Fig. 6 reveals the daily power output of the cleaned panels nor-
and time of measurement in Fig. 2a are as follows: 1009 W/m2, 1015 W/ malized to the control panel during each day of the “weekly cleaning
m2, 904 W/m2, 917 W/m2, 912 W/m2, 1016 W/m2, 985 W/m2, and frequency”. As shown in Fig. 6a, there are three clear improvement
950 W/m2, respectively for each day. The irradiance values reported in jumps in performance approximately every week. This improvement is
the figures are the global plane of array (GPOA). due to panel cleaning on that specific day.
Fig. 3 shows the power output of the different panels compared to Fig. 7 shows the average daily power output of the cleaned panels
the uncleaned control panel (PV30). As shown in the figure, the per- with respect to the control panel during the “weekly cleaning fre-
formance of the cleaned panels improves compared to the control quency”. In this case, the weekly cleaning shows approximately 2 W/
panel. This is mainly due to dust accumulation on the control panel, week of improvement compared to the control panel in summer as
while in the case of cleaned panels, the dust is removed every day. depicted in Fig. 7b.
During 1st February, there was a relative decrease of the cleaned panel Fig. 8 presents the daily power output of the last day of the monthly
power output with respect to control panel. This is due to the rain, cleaning cycle. In this figure, only three days (the cleaning day in a
which removed most of the dust from the control panel and therefore month) are shown to better emphasize the cleaning improvements and
led to a relative decrease in cleaned panels’ performance with respect to for better visibility purposes. As shown in Fig. 8a, the cleaning was
the control panel as shown in Fig. 3b. performed on 4th June, which shows the change in the power output
Fig. 4 shows the average power output change of all cleaned panels compared to the control panel in the summer period. Similarly, Fig. 8b
with respect to control panel. It is shown that there is a continuous daily illustrates the changes in the power output compared to the control
improvement of cleaned panels with respect to uncleaned ones. In panel in the winter period when the cleaning was performed on 2nd
Fig. 4a, there is an improvement of about 0.4 W/day on average com- April 2018.
pared to the control panel, including the three days with no cleaning in Fig. 9 shows the average last three days power output of the cleaned
the middle of the week in summer. However, in winter, the improve- panels with respect to the control panel. There is an improvement of
ment is slightly higher corresponding to about 0.5 W/day on average approximately 5 W due to the cleaning in the summer period. However,
compared to the control panel as depicted in Fig. 4b. the improvement was lower in the winter season corresponding to ap-
Fig. 5 shows the daily power output of the “weekly cleaning” proximately 2 W after the cleaning. This is mainly due to the fact that

809
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Fig. 8. Comparison of normalized monthly power outputs from CdTe thin-film PV panels cleaned by various PV cleaning techniques (a) summer, (b) winter.

the monthly cleaning cycle in the winter period occurred during the panel using both the wiper and the vacuum cleaner. As it is visible in
rainy season, which means all the panels were cleaned in each rainy the figure, although this cleaning method can reduce the dust on the
day. This caused the difference between cleaned and uncleaned panels surface, it is not highly effective since the dust particles partially remain
to be relatively small. on the surface. It is important to note that the wiper was used initially
While the daily cleaning provides excellent performance, it is by far and then the panel is vacuumed by the vacuum cleaner.
the most expensive and energy intensive of the three chosen fre- The estimated theoretical improvement is calculated by the differ-
quencies. The weekly cleaning provides slightly lower improvement ence of the output of the control panel on the measurement day and on
compared to daily cleaning while decreasing the cost and energy use. the 1st day of the cleaning cycle. The first day of the cycle is the day
That is a large improvement from an optimization point of view as where all the panels were cleaned thoroughly with water. This measure
shown in Tables 7 and 8. In the monthly cleaning cycle of summer shows the theoretically achievable improvement by cleaning the panel
season, similar to the winter experiments, keeping the panels dusty for a with water. The PV panels were thoroughly cleaned with water, brush,
period of 1-month severely affect the performance of the panels, and on and wiper before starting each cleaning cycle in order to make sure that
the long run, proves to be inefficient from cost and performance point the performance of the selected cleaning apparatuses can be clearly
of view. The improvements are shown for different cleaning frequencies observed. The theoretical results in the summer season are slightly
with a comparison to the control panel. For instance, the monthly better than the winter season. The difference increases as the PV panels
cleaning in summer yielded an improvement of about 7%, meaning that are not cleaned. It reaches about 2% for the monthly cleaning period in
if there are two similar PV panels, one of them was not cleaned for a the summer as shown in Table 8. The weekly data are not very re-
month and the other one was cleaned after 1-month, the cleaned panel presentative here due to the presence of dust storms, which affected all
will have approximately 7% more energy generation than the un- panels and resulted in varying irradiations. This can be commented
cleaned panel. In the winter, weekly cleaning yielded an improvement such that it is not efficient to extend cleaning more than a week.
of about 5.1% in the first week, 7% in the second week, and 11% in the Comparing the winter and the summer cleaning efficacies as shown
third week. This means that if there are two similar PV panels, one of in Table 9, the cleaning process is more effective in the winter than the
them was not cleaned for 3 weeks, and the other one was cleaned every summer, especially for the weekly cleaning frequency. This finding was
week, the difference in their energy output will be approximately 5.1% investigated further by comparing the dust accumulation on the panels
after the first week, about 7% after the second week, and about 11% over the two seasons.
after the third week.
Fig. 10 shows the PV panel before and after the cleaning with the 3.1. Dust accumulation results
combination of microfiber wiper and vacuum cleaner. The PV panel
was cleaned in a vertical direction from the top to the bottom of the Table 10 shows the dust accumulation during the summer season.

810
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

(a)

(b)

Fig. 9. Average monthly power output changes of CdTe thin-film PV panels (average of all cleaning techniques with respect to control panel) (a) summer, (b) winter.

Table 7
Percentage improvements after different cleaning frequencies in the winter season.
Cleaning frequency Time Improvement (with respect to control panel) Theoretical 100% cleaning improvement (with respect to control) Comments

Daily cleaning (30/1/2018) 4.80% 5.1% N/A


1st week
Weekly cleaning (11/2/2018) 5.10% 5.5% N/A
2nd week
Weekly cleaning (19/2/2018) 7% 10% N/A
3rd week
Weekly cleaning (26/2/2018) 11% 18% Rain on 25/2/2018
4th week
Monthly cleaning (2/4/2018) 4.20% 4.4% Rainy season*
8th week

* This point was during the rainy season. Rain cleans all panels equally including the control panel. This caused the relative difference between the cleaned panels
and control panels to be negligible. Therefore, cleaning the panels did not yield a big improvement over the control panel.

Table 8
Percentage improvements after different cleaning frequencies in the summer season.
Cleaning frequency Time Improvement (with respect to control panel) Theoretical 100% cleaning improvement (with respect to control) Comments

Daily cleaning (5/7/2018) 3.7% 4% N/A


1st week
Weekly cleaning (13/6/2018) 2.5% 4.2% Dusty weather
2nd week
Weekly cleaning (20/6/2018) 2.7% 3.3% N/A
3rd week
Weekly cleaning (27/6/2018) 4% 4.5% N/A
4th week
Monthly cleaning (4/6/2018) 7% 9% N/A

811
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Fig. 10. The thin-film PV panel (a) before cleaning and (b) after cleaning with wiper and vacuum cleaner combination.

Table 9
Comparison of summer and winter PV performance improvements compared to control panel.
Cleaning Frequency Improvement in Summer Improvement in Winter

Daily cleaning 3.7% 4.80%


Weekly cleaning (1st week) 2.5% 5.10%
Weekly cleaning (2nd week) 2.7% 7%
Weekly cleaning (3rd week) 4% 11%
Monthly cleaning 7% 4.20%*

* During a rainy period.

Table 10 calculated for the 1st and 2nd months as, 12.384 g and 8,755 g, re-
Dust accumulation on the different glass coupon samples in the summer. spectively. In the 2nd month, the dust accumulation was less than the
Period Sample # Dusty weight Clean weight Net dust weight 1st month due to the fact that the 2nd month was a rainy period, in
(g) (g) (g) which the rain often cleaned the panels.
The dust accumulation results do not fully explain the difference in
1st Month 1 55.555 55.379 0.176
the cleaning efficiencies between the seasons, as the summer has a
2 54.683 54.636 0.047
3 56.300 56.244 0.056
higher rate of dust accumulation than the winter. Nevertheless, the
Average 55.513 55.420 0.093 obtained results imply that the investigation of different type of dust
2nd month 1 55.438 55.380 0.058
particles/soiling on the panels can be helpful. Because dust accumula-
2 54.691 54.641 0.05 tion in the winter is more dominated by clay-like layer forming on the
3 56.312 56.231 0.081 panels due to the condensation, while the dust accumulation in the
Average 55.480 55.417 0.063 summer is more powdery and easier to remove.

3.2. Cost assessment results


Table 11
Dust accumulation on the different glass coupon samples in the winter. Table 12 shows the main cost parameters in different PV cleaning
Period Sample # Dusty weight Clean weight Net dust weight processes. The table lists a comparison of the total cost for different
(g) (g) (g) cleaning techniques and their associated costs excluding the drone re-
lated costs. This table is prepared for the weekly cleaning schedule since
1st Month 1 55.45 55.407 0.043
it is found as the optimum frequency. The highest cleaning cost is
2 54.681 54.643 0.038
3 56.262 56.231 0.031
calculated for the vacuum cleaner and brush combination due to high
4 56.377 56.306 0.071 manpower cost (more time spent). The lowest cost per the unit PV area
5 56.924 56.892 0.032 is calculated as 21.1 USD/m2/year for the microfiber based-cloth wiper,
Average 55.9388 55.8958 0.043 whereas the highest cost is calculated for the vacuum cleaner and brush
2nd month 1 55.425 55.4 0.025 combination. This implies that complicating the cleaning process brings
2 54.647 54.606 0.041 additional costs, henceforth there should be an optimum point between
3 56.268 56.232 0.036
cost and cleaning effectiveness. Manpower cost forms about 58% of the
4 56.27 56.244 0.026
5 56.911 56.887 0.024 total PV cleaning cost for vacuum and microfiber cloth combination.
Average 55.9042 55.8738 0.0304 These results point out that, in case the cost of manpower can be
minimized by using autonomous cleaning such as drones, the cleaning
cost can be reduced significantly. Consequently, low-cost autonomous
Three glass coupons are used to collect the dust on an unused panel. cleaning techniques are more preferable from a cost point of view. An
Based on the weight change of the coupons, the total estimated dust optimal cleaning technique would include autonomous cleaning tech-
weight on the panel was 26.784 g for the 1st month and 18.144 g for the nology integrated with the least costly and the most effective cleaning
2nd month in the summer season. material found in this study. It can be noted that although there is a
The summer in Qatar has a high frequency of dust storms and scarce need for a drone pilot for drone-based cleaning, the recent technologies
rainfall, while in the winter, Qatar perceives occasional rainfalls, allow fully automated and programmable drones that can eliminate or
whereas the dust storms are rare. The uncertainty level of the mea- minimize the human interaction.
surement balance is 1 mg as per manufacturer datasheet. The cost terms account for the time, money and difficulty of op-
Table 11 shows the results of the dust accumulation during the two eration. The cleaning effectiveness is the other decision parameter,
months in the winter season. Based on the average dust weights on the which is a measure of how much performance improvement can be
coupons, the total estimated dust weight on PV panel surface was achieved in thorough cleaning.

812
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Cleaning using the brush has a low cost, however, the improvement
Total Yearly Cost of Cleaning
after cleaning is relatively small compared to the other methods. For the
microfiber-based wiper, the cost is low, and the cleaning effectiveness is
relatively high. As for the combined methods, even though they pro-
vided high cleaning effectiveness, almost similar to the microfiber-
Per Unit PV Area
(USD/m2/year)

based wiper, their cost is much higher compared to the microfiber


wiper cleaning. The reason behind this fact is that there is a thin layer of
dust that is easily removable on top of the panels. This layer is removed
59.76
54.45

26.42
21.07
completely by all of the methods except the brush. Beneath this layer,
there is another sticky thin layer of dust that is very difficult to remove.
This layer cannot be removed (unless extra forces are applied) by any of
the selected four methods.
Cleaning Per Unit PV Area
Total Cost of One-Time

3.3. Overall comparison

In this section, the overall evaluation of the selected PV cleaning


(USD/m2)

methods with respect to their cleaning effectiveness, cost, and time


1.15
1.05

0.51
0.41

aspects are presented. It is shown that a microfiber-based wiper is the


most effective cleaning technique. It provides one of the best perfor-
mance improvements similar to combined cleaning techniques.
However, it is simpler than the combined methods.
Cleaning Per Panel

Fig. 11 illustrates that the brush and microfiber based-cloth wiper


Cost of One-Time

has the lower cost and the lower maintenance requirement. However,
(USD/panel)

the brush is not much effective in cleaning the panel, primarily because
of the fact that the brush can break the bond of dust particles on the
0.83
0.76

0.37
0.29

panel but cannot fully wipe it off the panel. The microfiber wiper has a
very low cost and requires minimum maintenance. Additionally, the
Manpower Cost

wiper provides the best cleaning effectiveness. Because it has the cap-
(USD/panel)

ability to break of the dust and wipe it off the panel in a single run.
The brush and vacuum cleaner combination, is a much more ef-
0.50
0.44

0.25
0.19

fective cleaning method than brush alone, because of the fact that the
brush breaks off the dust layer while the vacuum cleaner vacuums it off
Electricity Cost Of Cleaning

the panel. However, adding the vacuum cleaner complicates the ap-
paratus and slows down the cleaning, in addition to having a higher
cost. The wiper and vacuum cleaner combination is also an effective
cleaning method, however, there is no significant improvement over
(USD/panel)

wiper only cleaning. This is mainly due to the fact that the wiper cleans
Per Panel

0.000021
0.000021

most of the removable dust, however, it leaves a negligible amount of


dust for the vacuum to remove.
0
0

Fig. 11 illustrates the comparison of different cleaning techniques


across the three main aspects of time, cost, and effectiveness in addition
Normalized Capital

to an overall comparison.
(USD/cleaning)

0.33
0.32

0.12
0.11
Cost
Comparative cost evaluation for different cleaning techniques.
Duration of Cleaning

(minutes/panel)
Per Panel

2.00
1.75

1.00
0.75
Vacuum cleaner and brush combination
Vacuum cleaner and Microfiber cloth

Microfiber based-cloth wiper


wiper combination
Cleaning Method
Table 12

Brush

Fig. 11. Overall comparison of different PV cleaning techniques.

813
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

Table 13
Comparison of some literature studies for PV cleaning.
Authors PV Cleaning Method Cleaning Frequency Results

Al-Shehri et al. Brush materials of nylon, cloth, – The enhancement in the maximum power output of solar panels
[4] and silicon rubber foam cleaned with silicone rubber brush was around 1% on average
from the unbrushed initial power output.
Al-Shehri et al. Brush 1 day Selection of the optimal brush is essential to meeting the required
[5] level of cleaning while preventing damage to the surface of the
solar panels.
Arabatzis et al. Self-cleaning, photocatalytic, 16 days and 86 days Under outdoor real conditions, the coated PV panels demonstrated
[17] anti-reflective glass coating an average gain of 5–6% for the monitored period of time.
Piliougine et al. Anti-soiling coated surface Both reference modules manually cleaned with tap Daily energy soiling losses of 2.5% whereas for the uncoated
[19] water and dried every day. The other modules modules this loss is about 3.3%.
were only cleaned by rainfall during a year.
Luque et al [20] Regular cleaning of bifacial For each technology, one module was assigned to Soiling rate of 0.301%/day for the monofacial module, Soiling rate
modules be cleaned once a week, while the other was left to of 0.236%/day for the bifacial module
the soil through the whole experimental period
Urrejola et al. Brushing With water Monthly cleaning (suggested optimum cleaning Lowest decay values in summer 2015: −0.14%, Highest values in
[23] period: 45 days) autumn 2015 seasonal average: −0.56%/day, A monthly decay of
17.36% in the performance ratio.
Pedersen et al. Anti-soiling coating Dust density accumulated each week on normal Over the two-month period, a loss in efficiency of 0.2–0.3%
[38] and anti-soiling glass (estimation based on 1–2% transmission losses)
This study Brush, 1 day The weekly power losses in the winter season are 3.42%, 2.95%,
Microfiber cloth wiper, 1 week 3.63%, 2.28% respectively for microfiber based-cloth
Vacuum cleaner 1 month wiper + vacuum cleaner, mechanical brush + vacuum cleaner,
microfiber based-cloth wiper, mechanical brush.

In addition to the selected cleaning methods in this study, several


other cleaning alternatives are also available from manual to autono-
mous options. However, many of them are not suitable for hot desert
climates because they consume large amounts of water. Freshwater is
not an abundant source in desert climates, although seawater is avail-
able. Nevertheless, the corrosive feature of seawater can damage the PV
panels causing corrosion on the metal frames and structure.
Desalinating the seawater is energy intensive as well.
The main differences of the selected cleaning techniques compared
to other options are; (i) suitable for water-free cleaning, (ii) easier dust
removal compared to water based-cleaning because, there is no water-
sand mixture, which is difficult to remove, (iii) adaptable to autono-
mous cleaning via drones, (iv) requires minimum customization com-
pared to robotic based-cleaning, (v) lower energy requirement.
The comparison of various literature studies on PV cleaning is listed
in Table 13. Most of the studies either considered cleaning with water
or anti-dust coating on the glass. They reported the soiling rates, the
power gain due to cleaning, the soiling losses, the transmission losses,
the performance ratio decay etc. The cleaning frequencies are diverse Fig. 12. Comparative evaluation of various cleaning techniques for drone
ranging from 1 day to even 1 year. adaptability.

3.4. Drone adaptability assessment results 4. Conclusions

Fig. 12 depicts a comparison between the four cleaning methods In this study, various waterless PV cleaning techniques and appa-
with respect to drone retrofit ability. A brush, being the easiest to use, ratuses are evaluated based on their suitability for retrofitting into
lightest, smallest, and requiring no energy to operate has the highest UAVs/Drones to be used in large-scale solar power plant cleaning ap-
overall ranking in terms of retrofit ability to the drone. The microfiber plications. The selected dry-cleaning methods are; brush, microfiber-
wiper is very close to the brush in terms of retrofit ability, being only based cloth wiper, a combination of a brush and a vacuum cleaner, and
slightly bigger and harder to adapt to a drone. It is important to note a combination of microfiber-based cloth wiper and vacuum cleaner.
here that adaptability is meaning how easy it is to mimic the necessary The hot desert climates are the main locations where soiling occurs.
cleaning movement of the tool. The combined cleaning methods rank Therefore, the focus of this study is to determine the optimal thin-film
lower mainly due to the heavy weight, large size, energy requirement, PV cleaning techniques under the winter and summer conditions. The
and low adaptability of the vacuum cleaner. The smallest apparatus cleaning frequency is varied from daily to monthly cycles.
(Brush) is determined to have a size score of 10, while other appara- The critical parameters for drone adaptability of the cleaning ap-
tuses are scored lower based on their size relative to the brush. The paratuses were determined and comparatively evaluated. Based on the
weight and energy consumption parameters are also ranked using the weight, energy consumption, size, and adaptability; selected cleaning
similar logic with the brush having the highest score (equal to micro- techniques are ranked to find out the best performing option.
fiber wiper in energy consumption). Adaptability is a measure of how Furthermore, the cost of cleaning a PV panel is calculated for each
easy to apply this apparatus on a drone. Therefore, the adaptability is method under the assumed conditions. Waiting for more than one week
dependent on other factors such as size and energy consumption. causes larger amounts of dust accumulation on the PV surfaces,

814
M. Al-Housani, et al. Energy Conversion and Management 185 (2019) 800–815

therefore, the most efficient cleaning frequency is found to be the [12] Kegeleers M. The Development of a Cleaning Robot for PV Panels. KU Leuven:
weekly cleaning, especially in the summer conditions. Faculty of Engineering Technology; 2015.
[13] Gheitasi A, Almaliky A, Albaqawi N. Development of an automatic cleaning system
Microfiber based-cloth wiper along with the microfiber and vacuum for photovoltaic plants. 2015 IEEE PES Asia-Pacific Power Energy Eng. Conf.IEEE;
cleaner combination are the most effective cleaning methods with ap- 2015. p. 1–4. https://doi.org/10.1109/APPEEC.2015.7380938.
proximately 7.7% performance increase for the weekly period compared to [14] Calle CI, Buhler CR, McFall JL, Snyder SJ. Particle removal by electrostatic and
dielectrophoretic forces for dust control during lunar exploration missions. J
control panel in the winter season. Microfiber based-cloth wiper achieved Electrostat 2009;67:89–92. https://doi.org/10.1016/J.ELSTAT.2009.02.012.
on average 7.7% and 3.1% performance improvement (compared to control [15] Isaifan RJ, Samara A, Suwaileh W, Johnson D, Yiming W, Abdallah AA, et al.
panel) for the weekly cleaning frequency in the winter and summer seasons, Improved self-cleaning properties of an efficient and easy to scale up TiO2 thin films
prepared by adsorptive self-assembly. Sci Rep 2017;7:9466. https://doi.org/10.
respectively. However, combining with a vacuum cleaner increases the 1038/s41598-017-07826-0.
operational costs of cleaning due to electricity consumption. A brush is not [16] He G, Zhou C, Li Z. Review of self-cleaning method for solar cell array. Proc Eng
highly effective in cleaning the panel due to dry cleaning. On the other 2011;16:640–5. https://doi.org/10.1016/J.PROENG.2011.08.1135.
[17] Arabatzis I, Todorova N, Fasaki I, Tsesmeli C, Peppas A, Li WX, et al. Photocatalytic,
hand, to reduce this deficiency, it can be combined with a vacuum cleaner
self-cleaning, antireflective coating for photovoltaic panels: characterization and
showing about 5.4% performance improvement over a week. The largest monitoring in real conditions. Sol Energy 2018;159:251–9. https://doi.org/10.
disadvantage of dry cleaning methods is that a very thin layer of dust re- 1016/J.SOLENER.2017.10.088.
mains in place and is very hard to remove without water. However, if the [18] Zhong H, Hu Y, Wang Y, Yang H. TiO2/silane coupling agent composed of two
layers structure: a super-hydrophilic self-cleaning coating applied in PV panels.
cleaning frequency is carefully selected, then the effects on the performance Appl Energy 2017;204:932–8. https://doi.org/10.1016/J.APENERGY.2017.04.057.
are observed negligible for the short-term. [19] Piliougine M, Cañete C, Moreno R, Carretero J, Hirose J, Ogawa S, et al.
From a drone adaptability point of view, the theoretical analysis Comparative analysis of energy produced by photovoltaic modules with anti-soiling
coated surface in arid climates. Appl Energy 2013;112:626–34. https://doi.org/10.
shows that the microfiber based-cloth wiper and the brush with their 1016/J.APENERGY.2013.01.048.
low weight, small size, and ease of use are the best-suited options for [20] Luque EG, Antonanzas-Torres F, Escobar R. Effect of soiling in bifacial PV modules
drone adaptation/retrofitting. and cleaning schedule optimization. Energy Convers Manage 2018;174:615–25.
[21] Kimber A, Mitchell L, Nogradi S, Wenger H. The effect of soiling on large grid-
For future studies, creating custom-made tools (brushes/wipers) to be connected photovoltaic systems in California and the Southwest Region of the
used in drone retrofitting can significantly improve cleaning performance United States. 2006 IEEE 4th World Conf. Photovolt. Energy Conf. IEEE; 2006. p.
compared to existing tools, which are mainly manufactured for human use. 2391–5. https://doi.org/10.1109/WCPEC.2006.279690.
[22] Moharram KA, Abd-Elhady MS, Kandil HA, El-Sherif H. Influence of cleaning using
In addition, the cleaning time of the day may result in a higher/lower water and surfactants on the performance of photovoltaic panels. Energy Convers
cleaning efficiency, e.g. cleaning in early mornings before sunrise might Manage 2013;68:266–72.
benefit the microfiber-based wiper by taking advantage of dew on the pa- [23] Urrejola E, Antonanzas J, Ayala P, Salgado M, Ramírez-Sagner G, Cortés C, et al.
Effect of soiling and sunlight exposure on the performance ratio of photovoltaic
nels. Therefore, studying the effects of cleaning time selection can be in-
technologies in Santiago, Chile. Energy Convers Manage 2016;114:338–47.
vestigated as well as the effects of different dust deposition mechanisms. [24] Kumar NM, Sudhakar K, Samykano M, Jayaseelan V. On the technologies em-
powering drones for intelligent monitoring of solar photovoltaic power plants. Proc
Acknowledgment Comput Sci 2018;133:585–93. https://doi.org/10.1016/j.procs.2018.07.087.
[25] Gallardo-Saavedra S, Hernández-Callejo L, Duque-Perez O. Technological review of the
instrumentation used in aerial thermographic inspection of photovoltaic plants. Renew
The authors acknowledge the Solar Test Facility (STF) in HBKU-QEERI Sustain Energy Rev 2018;93:566–79. https://doi.org/10.1016/j.rser.2018.05.027.
at QSTP for allowing us to use their facilities during the experiments. The [26] Entrop AG, Vasenev A. Infrared drones in the construction industry: designing a
protocol for building thermography procedures. Energy Proc 2017;132:63–8.
support provided by Qatar Foundation (QF), Qatar and College of Science https://doi.org/10.1016/j.egypro.2017.09.636.
and Engineering (CSE), HBKU, Qatar is gratefully acknowledged. [27] Rakha T, Gorodetsky A. Review of Unmanned Aerial System (UAS) applications in the
built environment: towards automated building inspection procedures using drones.
Autom Constr 2018;93:252–64. https://doi.org/10.1016/j.autcon.2018.05.002.
References [28] Álvarez-Tey G, Jiménez-Castañeda R, Carpio J. Analysis of the configuration and the
location of thermographic equipment for the inspection in photovoltaic systems. Infrared
[1] Guo B, Javed W, Figgis BW, Mirza T. Effect of dust and weather conditions on Phys Technol 2017;87:40–6. https://doi.org/10.1016/j.infrared.2017.09.022.
photovoltaic performance in Doha, Qatar. 2015 2015 1st Work Smart Grid Renew [29] Teubner J, Kruse I, Scheuerpflug H, Buerhop-Lutz C, Hauch J, Camus C, et al.
EnergySGRE; 2015. p. 2015. https://doi.org/10.1109/SGRE.2015.7208718. Comparison of drone-based IR-imaging with module resolved monitoring power data.
[2] Ghazi S, Sayigh A, Ip K. Dust effect on flat surfaces – a review paper. Renew Sustain Energy Proc 2017;124:560–6. https://doi.org/10.1016/j.egypro.2017.09.094.
Energy Rev 2014;33:742–51. [30] Willis DJ, Niezrecki C, Kuchma D, Hines E, Arwade SR, Barthelmie RJ, et al. Wind
[3] Costa SCS, Diniz ASAC, Kazmerski LL. Solar energy dust and soiling R&D progress: energy research: state-of-the-art and future research directions. Renew Energy
literature review update for 2016. Renew Sustain Energy Rev 2018;82:2504–36. 2018;125:133–54. https://doi.org/10.1016/j.renene.2018.02.049.
https://doi.org/10.1016/j.rser.2017.09.015. [31] Vacca A, Onishi H, Cuccu F. Drones: military weapons, surveillance or mapping tools for
[4] Al Shehri A, Parrott B, Carrasco P, Al Saiari H, Taie I. Accelerated testbed for studying the environmental monitoring? Advantages and challenges. A legal framework is required.
wear, optical and electrical characteristics of dry cleaned PV solar panels. Sol Energy Transp. Res. Procedia 2017. p. 51–62. https://doi.org/10.1016/j.trpro.2017.05.209.
2017;146:8–19. https://doi.org/10.1016/J.SOLENER.2017.02.014. [32] Cleandrone – Cleandrone 2018: Barcelona, Spain. http://www.cleandrone.com/
[5] Al Shehri A, Parrott B, Carrasco P, Al Saiari H, Taie I. Impact of dust deposition and cleandrone-2/ (accessed October 28, 2018).
brush-based dry cleaning on glass transmittance for PV modules applications. Sol [33] The world’s first fully automated UAV for cleaning solar panels – Aerial Power
Energy 2016;135:317–24. https://doi.org/10.1016/J.SOLENER.2016.06.005. introduces the worlds first cleaning drone. 2018: London, United Kingdom. https://
[6] Deb D, Brahmbhatt NL. Review of yield increase of solar panels through soiling www.aerialpower.com/solarbrush/ (accessed October 28, 2018).
prevention, and a proposed water-free automated cleaning solution. Renew Sustain [34] Muntwyler U, Schuepbach E, Lanz M. Infrared (IR) drone for quick and cheap PV
Energy Rev 2017. inspection. Proc. 31st Eur. Photovolt. Sol. Energy Conf. Exhib. 2015:1804–6.
[7] Syafiq A, Pandey AK, Adzman NN, Rahim NA. Advances in approaches and methods [35] China Mini Clean Brush from Jinjiang Manufacturer: Jinjiang Jiaxing Company n.d.
for self-cleaning of solar photovoltaic panels. Sol Energy 2018;162:597–619. https://jiaxingbest.manufacturer.globalsources.com/si/6008845131807/pdtl/Car-
[8] Jiang Y, Lu L, Ferro AR, Ahmadi G. Analyzing wind cleaning process on the accu- wash/1159952867/Mini-Clean-Brush.htm (accessed October 3, 2018).
mulated dust on solar photovoltaic (PV) modules on flat surfaces. Sol Energy [36] 3M Commercial Solutions Division. Technical Data Sheet Scotch-Brite™ High
2018;159:1031–6. https://doi.org/10.1016/J.SOLENER.2017.08.083. Performance Cloth. Tech Data Sheet 2017:1. https://multimedia.3m.com/mws/
[9] Li X, Qin H, Zhang Y, Yao W, Li Y, Liu H. Dust effect on the optical-thermal prop- media/1459376O/scotch-brite-high-performance-cloth-2010-technical-data-sheet.
erties of absorber plate in a transpired solar air collector. Energy Convers Manage pdf (accessed October 3, 2018).
2018;169:13–21. [37] Shark® Cordless Pet Perfect® Lithium-Ion Handheld Vacuum (LV801) | Shark® n.d.
[10] Elminir HK, Ghitas AE, Hamid RH, El-Hussainy F, Beheary MM, Abdel-Moneim KM. https://www.sharkclean.com/products/0/all/1991/shark-cordless-pet-perfect-li-
Effect of dust on the transparent cover of solar collectors. Energy Convers Manage thium-ion-handheld-vacuum/#documentation (accessed October 3, 2018).
2006;47:3192–203. [38] Pedersen H, Strauss J, Selj J. Effect of soiling on photovoltaic modules in Norway.
[11] Xu R, Ni K, Hu Y, Si J, Wen H, Yu D. Analysis of the optimum tilt angle for a soiled Energy Proc 2016;92:585–9. https://doi.org/10.1016/J.EGYPRO.2016.07.023.
PV panel. Energy Convers Manage 2017;148:100–9.

815

You might also like