Вы находитесь на странице: 1из 45

Solar Energy

Manuscript Draft

Manuscript Number: SE-D-19-02017

Title: Modeling and Quantifying Dust Accumulation Impact on PV Module


Performance

Article Type: Research paper

Section/Category: Photovoltaic materials, cells and systems

Keywords: Effect of Soiling on PV Performance; Soiling Loss Index (SLI);


Soiling Monitoring System; PV Cleaning Schedules; Feed-in tariff.

Abstract: Natural dust accumulation or soiling on the surfaces of PV


modules in large-scale PV power plants has a significant effect on the
overall performance of these power plants, especially those located in
arid areas such Jordan. In this manuscript, the soiling induced effects
are quantified and modeled. Experimental setups for three common types of
PV modules are utilized to acquire the data used for the models.
Projection of the performance and economic impacts on large scale PV
plants is carried out to emphasize the necessity to consider the soiling
effects during the planning phase. Optimization of PV cleaning schedules
for the best return-on-investment, in addition to the importance of
determining typical soiling rates for forecasting models in PV power
plants under planning is presented as well.
Highlights (for review)

“Modeling and Quantifying Dust Accumulation Impact on PV Module Performance”

by Mohammad Al-Addous, Zakariya Dalala, Firas Alawneh, and Christina B. Class

Highlights

• Definition of Soiling Loss Index (SLI) and its measurements

• Characterization of dust accumulation induced losses

• Experimental investigation of different PV technologies performance under soiling

• Economic analysis of cleaning against energy gain


*Manuscript
Click here to view linked References

Modeling and Quantifying Dust Accumulation Impact


on PV Module Performance
Mohammad Al-Addousa , Zakariya Dalalaa , Firas Alawneha , Christina B.
Classb,∗
a
Department of Energy Engineering, German Jordanian University, Amman, Jordan
b
School of Basic Sciences, Ernst-Abbe-Hochschule Jena, Germany

Abstract

Natural dust accumulation or soiling on the surfaces of PV modules in


large-scale PV power plants has a significant effect on the overall performance
of these power plants, especially those located in arid areas such Jordan. In
this manuscript, the soiling induced effects are quantified and modeled. Ex-
perimental setups for three common types of PV modules are utilized to
acquire the data used for the models. Projection of the performance and
economic impacts on large scale PV plants is carried out to emphasize the
necessity to consider the soiling effects during the planning phase. Optimiza-
tion of PV cleaning schedules for the best return-on-investment, in addition
to the importance of determining typical soiling rates for forecasting models
in PV power plants under planning is presented as well.
Keywords:
Effect of Soiling on PV Performance, Soiling Loss Index (SLI), Soiling
Monitoring System, PV Cleaning Schedules, Feed-in tariff.


Corresponding author
Email address: cclass@acm.org (Christina B. Class)

Preprint submitted to Solar Energy July 22, 2019


1 1. Introduction

2 Photovoltaic (PV) modules production has been sharply increasing over


3 the past decade, supported by the large number of PV projects commissioned
4 daily. The global interest to change course in the nature of energy supply
5 has been motivating the investments in renewable energy projects, where
6 PV and wind dominate the market share among other potential sources [1].
7 Expansion in scale and penetration is a key target in most national energy
8 sector policies around the world, where the energy market is expected to
9 witness continuously growing deployment of renewable energy production
10 projects [2].
11 On the technical level, operating PV systems in the field requires con-
12 tinuous monitoring of all possible factors that might affect the anticipated
13 performance, and proposing ways to mitigate their impact. In [3], a represen-
14 tation of a European suitability map for the installation of PV systems based
15 on a Geographical Information System Multi-criteria Assessment (GIS-MCA)
16 method using a set of relevant geographical variables was presented to help
17 allocate EU regional solar energy generation funds more efficiently. In [4],
18 the influence of dust, different tilt angles and orientations on the PV system
19 performance was carried out where an average of 2.5% variance in specific
20 yield was obtained. The inclination effect on performance was reported to be
21 in the range between 5.6% to 17.3%. In [5], the influence of temperature in-
22 crease on the energy production capacity of the installed off-grid PV system
23 was assessed and a 2% efficiency enhancement was obtained through con-
24 trolled cooling of the PV modules. Soiling effect has recently been receiving
25 more attention [6, 7], where previous studies showed that power degrada-

2
26 tion can reach 15% due to dust and dirt accumulation [8] Soiling can cause
27 the prevention of effective solar irradiance being absorbed by PV cells and,
28 consequently, significantly reduces the power production capability of PV
29 modules [6]. In arid and semiarid areas, where the solar irradiance potential
30 is usually near maximum, the soiling induced power generation reduction can
31 reach 50% [9]. In [10], the pattern of the dust distribution in different parts
32 of the world was investigated, and it was found that the Middle East and
33 North Africa exhibit the worst dust accumulation zones in the world.
34 Previously, several research efforts focused on quantifying the induced
35 soiling effects on the performance of large scale PV power plants, and ways
36 to consider these effects in the design phase [11–14]. An experimental set
37 up involving 100 glass samples with different tilt (β) and azimuth angles (γ)
38 was developed to investigate the induced dust effects on the performance
39 and the transmittance of the glass. It was evaluated at regular intervals
40 where a reduction in glass normal transmittance has been found strongly
41 depend on the dust deposition density in conjunction with plate tilt angle,
42 as well as on the orientation of the surface with respect to the dominant wind
43 direction [12]. In [13], the effects of pollutant types on the PV performance
44 was experimented. In addition, a comprehensive summary of the related
45 research on the PV pollutant types was presented. In [14], three different
46 models of PV power are used to investigate the effect of urban climate on PV
47 performance. The results show that the dimming of solar radiation in the
48 urban environment is the main reason for the decrease of PV module output.
49 Causes of dust accumulation and affecting factors were investigated as
50 well in [15–21].

3
51 An appraisal on the current status of research in studying the impact of
52 dust on PV system performance and identifying challenges to further perti-
53 nent research and a framework to understand the various factors that govern
54 the settling/assimilation of dust and likely mitigation measures were dis-
55 cussed in [15]. Lab prepared dust was utilized to study the effects of dust
56 accumulation on the surface of photovoltaic cells experimentally, where pre-
57 cise dust density was controlled to relate the actual output degradation [16].
58 It was concluded that dust accumulation considerably deteriorates the per-
59 formance of the photovoltaic cells. However, in carrying out the investigation
60 on the effect of dust and particulate pollution, the physical characteristics of
61 dust must be determined and correlated to the observed effects. Effects of
62 dust, humidity and air velocity that can influence the PV cell’s performance
63 have been simultaneously studied [17]. It is shown that each of these three
64 factors affect the other two and it is concluded that in order to have a pro-
65 found insight of solar cell design, the effect of these factors should be taken
66 into consideration together. In [18], the influence of the properties of PV
67 module itself on dust deposition and efficiency degradation was experimen-
68 tally investigated, such as the cell types and surface materials. The results
69 indicated that dust pollution has a significant impact on PV module output.
70 With dust deposition density increasing from 0 to 22gm−2 , the correspond-
71 ing reduction of PV output efficiency grew from 0 to 26%. The reduction
72 of efficiency has a linear relationship with the dust deposition density, and
73 the difference caused by cell types was not obvious. Moreover, the surface
74 material may influence dust deposition and accumulation considerably. The
75 poly-crystalline silicon module packaged with epoxy degraded faster than

4
76 other modules with glass surface under the same dust concentration. The
77 effect of soiling on energy production for large-scale ground mounted photo-
78 voltaic plants in the countryside of southern Italy was evaluated in [19]. The
79 results presented in this work show that both the soil type and the washing
80 technique influence the losses due to the pollution. A 6.9% loss for the plant
81 built on a sandy soil and a 1.1% for the one built on a more compact soil
82 have been found.
83 In [20], the soiling affect to PV concentrators in comparison with flat pan-
84 els has been experimented. In general, rshort circuit current ranged between
85 14% to 26%. The tilt angle influences the amount of energy collected by a
86 PV module has been detailed experimentally in [21].
87 The direct relation between the tilt angle and soiling was investigated
88 in several research efforts to indicate the best tilt angle configuration for
89 maximizing energy production [6, 22, 23]. Multiple tilt angle configurations
90 throughout the year is suggested to mitigate the soiling loss, though, it is
91 not practical for most on-grid systems where fixed structure is desired. It
92 was found that soling effect is heavily dependent on the specific location and
93 season. In [24], it was found that the degradation in PV power production
94 reached 50% for 6 months without cleaning in eastern parts of Saudi Arabia.
95 In [25], the average soling index was estimated to be 4.4% for a year with
96 peaks of 20% during drier times. Long term degradation due to natural
97 elements, including dust accumulation was measured to range between 8% -
98 12% in Australia [26].
99 Performance degradation due to soiling induces economic losses, associ-
100 ated with unreliable estimations for the energy yield while sizing a specific

5
101 PV power plant. To mitigate these impacts, several efforts have been made
102 including scheduled cleaning, which in itself, might induce considerable cost
103 in large PV systems especially in drier regions [27, 28]. Thus, cleaning should
104 not be the trivial resort to solve the problem of soiling, unless the production
105 losses due to soiling exceed the cost of cleaning [29]. Approximate formulas
106 to estimate the economic losses due to dust were generated [30], although
107 with limited accuracy.
108 Quantifying the induced soiling losses on the system production is vital,
109 especially with the rising cost of operation and maintenance contracts; where
110 forced cleaning is usually included. In rooftop installations, which is the com-
111 mon case in urban areas, the cleaning task is not necessarily an easy part
112 where the inclusion of customized cleaning robots is extremely expensive, not
113 to mention the water budget needed in such cleaning efforts. Thus, in the
114 process of contracting new PV projects, calculating the soiling losses will im-
115 pact the budgeting scenarios and eventually, alter the planning schemes. As
116 literature is still modest in this regard, this work is directed to investigate the
117 soiling induced losses and to predict the feasible cleaning schedules needed
118 for a specific installation. A complete system setup is designed and installed
119 in the Jordan Valley where continuous data acquisition to all system infor-
120 mation is handled. The outdoor system is operated such that short term and
121 long term predictions are achieved. The season where dust accumulation is
122 highest is summer, and hence is the focus of the current study. The potential
123 loss due to natural soiling is clarified utilizing indicative measurements and
124 figures. The loss of radiation as a result of soiling is extrapolated by engaging
125 the short circuit current measurements in the PV module and three different

6
126 PV modules were installed in the location for comparison purposes. The
127 average soiling loss is calculated and projection on the power production loss
128 is evaluated, where more accurate estimations for the future production for
129 such installed PV systems are made. The manuscript details the experimen-
130 tal system setup along with the data acquisition system information. The
131 methodology, research activities and results are presented coherently.
132 The concept of soiling loss index is introduced in section 2. Section 3
133 presents the experimental setup and measurement results. Based on these
134 results soiling effects are modelled and analyzed in section 4. After presenting
135 the annual energy analysis (section 5), the economic analysis is presented in
136 section 6 before concluding the paper in section 7.

137 2. Definition of Soiling Loss Index

138 The Soiling Loss Index (SLI) is defined as the soiling induced loss in the
139 irradiance reaching the PV cells inside the PV module. This loss represents
140 the major loss due to the loss of the transmittance properties of the front
141 glass of the PV module because of soiling, if all other operating factors are
142 kept the same. The SLI uses the effective irradiance of a clean, or reference,
143 PV module and a dirty, or test, PV module as shown in Fig. 1 to quantify
144 the SLI percentage. The values are based on the ideal mathematical model
145 for solar day as well as the ideal solar cell model, which captures reduction
146 of current due to soiling.

147

148 Accordingly, SLI is defined using the following equation:

7
Figure 1: Effect of PV Module Soiling on Incident Global Irradiance

Gef f,Dirty − Gef f,Clean


SLI = · 100%. (1)
Gef f,Clean
149 Therefore, the output current-voltage characteristics or IV curves for two
150 identical PV modules, one with a clean surface and the other with a soiled
151 surface are different as shown in
152 Fig. 2 displays the effect of soiling on output IV characteristics based on
153 the ideal mathematical model for the solar day and and ideal solar cell model
154 as well as an assumed reduction of the current due to soiling.
155 The Maximum Power Point (Pmp = Vmp · Imp ) is depicted as a small red
156 circle in Fig. 2.

157

158 The effective irradiance information can be extracted from the short-

8
Figure 2: Effect of PV Module Soiling on Output IV Characteristics

159 circuit current of the PV module according to the following equation:


Isc 1
Gef f = · G0 · , (2)
Isc0 [1 + α(T − T0 ]
160 where
Isc measured short-circuit current of the PV module
Isc0 short-circuit current of the PV module at STC
(Standard Test Condition)
T measured backside temperature of the PV module
161

T0 backside temperature of PV module at STC, T0 = 25 ◦ C


G0 Incident global solar irradiance on the surface of PV module at
STC, G0 = 1000 W/m2
α Temperature coefficient of short-circuit current (% ◦ C−1 )
162 Table 3 in appendix 8.1 summarizes all symbols used in this paper.
163 The effective irradiance measurement utilizing the clean reference PV
164 module could be substituted by the clean and calibrated pyranomter of the

9
165 IV curve tracer, which is utilized in the experimental setup in this manuscript.

166 3. Experimental Setup, Results and Analysis

167 The test setup designed was installed in the southern part of the Jordan
168 Valley, near the town of Karma. A simplified schematic diagram of the
169 installed system is shown in Fig. 3. The system setup consists of a PV module
170 under testing, measurement devices including the pyranometer for irradiance
171 measurements, the back side temperature sensor, IV curve tracer and a data
172 acquisition system connected with a computer through USB communication
173 port.

Figure 3: Schematic Diagram of PV Module


Soiling Experimental Setup

Figure 4: Site Photos for Experimental Se-


tups of 130 WP , 310 WP and 80 WP PV
Modules

174 Three different types of PV modules with 130 WP , 310 WP and 80 WP

10
175 were utilized for the experimental work. The technical details are specified
176 in Table 4 in the Appendix.
177 Polycrystalline are most deployment technology in large scale PV projects
178 in semiarid areas while thin film-based PV modules still represent a modest
179 penetration. Clear comparative performance studies are inevitable to estab-
180 lish general selection criteria and guidelines between both technologies that
181 best suit specified regions, with specific environmental conditions.
182 Current-voltage (IV) measurements, global incident solar irradiance mea-
183 surements and module-backside temperature measurements were made avail-
184 able by utilizing a high precision IV curve tracer (Solmetric PV Analyzer
185 1000S). The data was acquired in real time. Three modules were installed
186 and natural dust accumulation was allowed during the summer season, which
187 represents the typical season for dust accumulation in Jordan. The poten-
188 tial degrading of PV module performance is the highest during this season.
189 The three experimental procedures took place at different time frames and
190 occasionally, some of them were cleaned to establish reference reading points
191 for comparison as will be shown later in this section. The IV-curve measure-
192 ments were acquired at solar noon time where irradiance is maximum and
193 above 800 W/m2 according to the IEC 60904. This avoids any differences
194 in soiling loss due to zenith angle of sun, module current dependence on ir-
195 radiance level or spectral differences. At the beginning of our research, we
196 noticed that the calculated SLI from measured IV curves which were cap-
197 tured during mornings and evenings (G < 800 W/m2 ) were not stable, so
198 they were filtered out from our SLI analysis to generate stable and uniform
199 data. A photo of the installed setup is shown in Fig.4.

11
200 3.1. 130 Wp PV Module SLI Experiment
201 Fig. 5 shows the results for the dust accumulation experiment which was
202 performed on the PV module rated at 130 WP during the period from July,
203 27th to November, 7th of 2017. The PV module was cleaned twice, one on
204 30/07/2017 and another one on 11/10/2017 as it is apparent from the figure.
205 The daily SLI is plotted with linear curve fitting (linear regression) between
206 the two cleanings as shown in Fig. 6. The daily SLI was estimated as -0.16%,
207 which is reflected directly on the output power, and daily energy produced
208 by the PV module.

Figure 5: SLI Measurements for 130 WP PV Module during Test Period: 27/07 -
07/11/2017 (16 weeks)

209 3.2. 310 Wp PV Module SLI Experiment


210 Same procedure is applied to the PV module rated at 310 WP . Fig. 7
211 shows the results for the daily irradiance loss at noon time during the period
212 extending from May, 22nd to November, 7th of 2017. This figure shows that
213 the PV module was cleaned twice, one on August, 21st and another one on
214 November, 10th , 2017. For the time in between the cleaning times, where
215 the dust is naturally accumulating, the daily irradiance loss at noon time

12
Figure 6: Determination of Daily SLI for 130 WP PV Module (Daily SLI = -0.16%) for
Test Period: 30/07 – 11/10/2017 (11 weeks)

216 was plotted with linear curve fitting as shown in Fig. 8. The daily SLI was
217 estimated at -0.13%, which is reflected directly on the output power, and
218 daily energy produced by the PV module.

Figure 7: SLI Measurements for 310 WP PV Module during Test Period: 22/05 -
07/11/2017 (25 weeks)

219 3.3. 80 Wp PV Module SLI Experiment

220 Fig. 9 shows the results for the daily irradiance loss at noon time during
221 the period extending from September, 9th to October, 31st of 2018. This
222 figure shows that the PV module was cleaned twice, one on September, 2nd

13
Figure 8: Determination of Daily SLI for 310 WP PV Module (Daily SLI = -0.13%) for
Test Period: 22/05 – 20/08/2017 (14 weeks)

223 and another one on October, 24th of 2018. The daily irradiance loss at noon
224 time was plotted with linear curve fitting as shown in Fig. 10. The daily SLI
225 was estimated at -0.45% , which is reflected directly on the output power, and
226 daily energy produced by the PV module. This thin-film module experiences
227 the highest SLI compared to the used polycrystalline ones.

Figure 9: SLI Measurements for 80 WP PV Module during Test Period: 02/09 -


31/10/2018 (9 weeks)

14
Figure 10: Determination of Daily SLI for 80 WP PV Module (Daily SLI = -0.45%) for
Test Period: 02/09 – 24/10/2018 (8 weeks)

228 3.4. Experimental SLI Values

229 In real world tests in 2017 SLI values were determined experimentally
230 for three different types of PV modules using linear regression. Table 1
231 summarizes the experimental results.

Table 1: Summary of experimental SLI Values

Module Test duration SLI Value R2 see

130 WP Module 11 weeks -0.16% 0.911 Fig. 6


310 WP Module 14 weeks -0.13% 0.9839 Fig. 8
80 WP Module 8 weeks -0.45% 0.0.9759 Fig. 10

232 These results are used for the modeling and impact analysis as described
233 in section 6.1.

15
234 4. Modeling and Impact Analysis of Soiling Effects

235 In order to analyze the induced effects of soiling on the annual energy
236 output of the PV module, modeling of the maximum power output of the
237 PV module under clean and dirty conditions is necessary. For the clean PV
238 module, the maximum power output will be estimated based on the incident
239 global solar irradiance and PV temperature according to eq.( 3).

G
P = P0 · · [1 + γ(T − T0 )], (3)
G0

240 where:

P is the maximum power output of the PV module (W)


Po is the maximum power output of the PV module (W)
at STC conditions
G is the incident global solar irradiance (W/m2 ), tilt angle = 30◦ .
Go is the incident global solar irradiance at STC
(typical 1000 W/m2 )
γ is the power temperature coefficient (% ◦ C−1 )
T is the operating PV temperature (◦ C) (see eq. (4))
T0 is the operating PV temperature at STC (typical 25 ◦ C)

241 The operating PV temperature is calculated using the following equation:

N OCT − 20
T = Ta + · G, (4)
800

242 where:

T is the operating PV temperature (◦ C)


Ta is the ambient temperature (◦ C)
16
N OCT is the Nominal Operating Cell Temperature obtained from the
PV module datasheet
G is the incident global solar irradiance (W/m2 ) either for clean or
dirty PV module

243 For the dirty PV module, the maximum power output will be estimated
244 based on the same variables including the soiling loss index that varies over
245 time due to soiling rate, natural cleaning by rain and forced water cleaning.
246 The maximum power output of the soiled PV module is determined according
247 to the following equation:

G · (1 + SLI)
P = P0 · · [1 + γ(T − T0 )], (5)
G0

248 where:

P is the maximum power output of the PV module (W)


P0 is the maximum power output of the PV module (W)
at STC conditions
G is the incident global solar irradiance (W/m2 ), tilt angle = 30◦ .
SLI is the Soiling Loss Index (%)
G0 is the incident global solar irradiance at STC
(typical 1000 W/m2 )
γ is the power temperature coefficient (% ◦ C−1 )
T is the operating PV temperature (◦ C)
T0 is the operating PV temperature at STC (typical 25 ◦ C)

249 In order to annually simulate the maximum output power for both clean

17
250 and soiled PV modules using equations (3) and (5), 10-minutes records for
251 one year are needed for both incident global solar irradiance and ambient
252 temperature. These records were obtained from a local weather station for
253 the year 2014.
254 The weather data will be made available under a creative commons licence
255 for non commercial use at http://www.renewables-and-water.org/ghor/
256 Ghor.html.
257 For the clean PV module, 10-minutes maximum output power (P) is
258 calculated for the corresponding 10-minutes incident global solar irradiance
259 (G) and PV temperature (T) according to equations (3) and (4).
260 For the soiled PV module, 10-minutes maximum output power (P) is
261 calculated according to equations (5) and (4) for the corresponding measured
262 10-minutes incident global solar irradiance (G) and PV temperature (T). As
263 SLI the values summarized in Table 1 determined by linear regression / curve
264 fitting as presented in sections 3.1 to 3.3 were used. SLI was set to 0 when
265 cleaning was modelled. For winter months from November to February, the
266 PV modules are assumed to be naturally cleaned three times per month due
267 to natural rainfall at equidistant times. It is assumed that the rain occurs in
268 regular intervals.
269 Local investigation of the PV industry in Jordan showed that typical
270 cleaning scenarios are monthly and biweekly. Cleaning is mainly performed
271 by external contractors based on a fixed schedule, and can thus not be per-
272 formed on short notice based on an actual needs.
273 Figures 11a to 11c shown below give the annual SLI profile for 130 WP
274 (Fig. 11a), 310 WP (Fig. 11b) and 80 WP (Fig. 11c) PV modules for the

18
275 monthly cleaning scenario during summer months.

Figure 11a: Annual SLI for 130 WP PV Module with Monthly Cleaning Scenario (Daily
SLI = -0.16%)

276 For the 130 WP wafer-based PV module, the daily SLI is experimentally
277 determined to be -0.16%, with the maximum monthly SLI (occuring a day
278 before washing) reaching around -5.0% by modeling the annual SLI profile
279 as suggested.
280 For the 310 WP wafer-based module, the daily SLI = -0.13% reaching the
281 maximum monthly SLI of around -4.0 %.
282 For the 80 WP thin-film PV module, the daily SLI was modelled to be
283 -0.45%, and accordingly the maximum monthly SLI reached around -14.0 %.
284 Figures 12a to 12c below show the simulated maximum output power
285 of two sample days. The simulation includes a cloudy day based on the
286 data of March 30th 2014 as well as a sunny day (June 29th 2014). The
287 simulation computed the maximum output power for the three experimented
288 PV modules based on the described models. A day before the cleaning day
289 of the corresponding PV module and at the end of the month, were both
290 selected.

19
Figure 11b: Annual SLI for 310 WP PV Module with Monthly Cleaning Scenario (Daily
SLI = -0.13%)

Figure 11c: Annual SLI for 80 WP PV Module with Monthly Cleaning Scenario (Daily
SLI = -0.45%)

20
Figure 12a: Simulated Maximum Output Power for 130 WP PV Module (Daily SLI =
-0.16%) for a Cloudy Day on 30 March (Left) and a Sunny Day on 29 June (Right) with
Monthly Cleaning Scenario

291 5. Annual Energy Analysis (Clean vs Dirty)

292 Energy analysis for both clean and dirty cases for the three PV module
293 samples have been performed to obtain the monthly energy output for both
294 cases. For energy (kWh) analysis, the power (kW) over time is integrated.
295 The intervals of 10 minutes = 1/6 h were added and kW converted to kWh.
296 For the 130 WP , 310 WP and 80 WP PV modules, the energy analysis
297 for clean and dirty cases are summarized in Tables 5 to 7 in the Appendix
298 in section 8.
299 For 130 WP , 310 WP and 80 WP PV modules, the annual relative energy
300 losses due to soiling with one cleaning scenario per month were obtained
301 1.37%, 1.65% and 10.32%, respectively as shown in Table 2 listed below.
Table 2: Annual Relative Solar Radiation and Energy Losses for the Three PV Modules
(Monthly Cleaning Scenario)

130 WP PV Module 310 WP PV Module 80 WP PV Module


Relative Relative Relative Relative Relative Relative
Month Solar Rad. Energy Solar Rad. Energy Solar Rad. Energy

21
Loss due Loss due Loss due Loss due Loss due Loss due
to Soiling to Soiling to Soiling to Soiling to Soiling to Soiling
Jan -0.83% -0.72% -0.67% -0.59% -2.33% -2.17%
Feb -0.76% -0.65% -0.62% -0.54% -2.14% -1.99%
Mar -2.69% -2.32% -2.19% -1.92% -7.57% -7.08%
Apr -2.37% -2.06% -1.93% -1.71% -6.67% -6.26%
May -2.53% -2.22% -2.05% -1.84% -7.10% -6.71%
Jun -2.40% -2.11% -1.95% -1.74% -6.76% -6.37%
Jul -2.50% -2.20% -2.03% -1.82% -7.03% -6.64%
Aug -2.50% -2.16% -2.03% -1.79% -7.04% -6.60%
Sep -2.36% -2.03% -1.92% -1.69% -6.64% -6.22%
Oct -2.21% -1.91% -1.80% -1.59% -6.23% -5.83%
Nov -0.80% -0.70% -0.65% -0.58% -2.26% -2.12%
Dec -0.84% -0.73% -0.68% -0.61% -2.35% -2.21%
Year -1.90% -1.65% -1.54% -1.37% -5.34% -5.02%

302 In order to compare the effect of cleaning times per month in summer
303 months on energy output and consequently on the economic performance
304 of the whole PV system, especially for Mega size projects, another scenario
305 for two cleanings per month (biweekly cleaning scenario) was considered for
306 comparison purposes. This, a comparison between monthly cleaning scenario
307 and biweekly cleaning was performed.
308 The annual SLI profile for the biweekly cleaning scenario was modeled
309 for 130 WP , 310 WP and 80 WP PV modules as shown in the Figures 13a to
310 13c below. The simulation model for the biweekly cleaning scenario was per-
311 formed for the three experienced modules using the same modeling approach
312 as for the monthly cleaning scenario analysis. Both simulation results for
313 monthly and biweekly cleaning scenarios will be compared in the next sec-
314 tion.

22
Figure 12b: Simulated Maximum Output Power for 310 WP PV Module (Daily SLI =
-0.13%) for a Cloudy Day on 30 March (Left) and a Sunny Day on 29 June (Right) with
Monthly Cleaning Scenario

Figure 12c: Simulated Maximum Output Power for 80 WP PV Module (Daily SLI = -
0.45%) for a Cloudy Day on 30 March (Left) and a Sunny Day on 29 June (Right) with
Monthly Cleaning Scenario

23
Figure 13a: Annual SLI for 130 WP PV Module with Biweekly Cleaning Scenario (Daily
SLI = -0.16%)

Figure 13b: Annual SLI for 310 WP PV Module with Biweekly Cleaning Scenario (Daily
SLI = -0.13%)

315 6. Economic Analysis - Comparison between monthly and biweekly


316 cleaning scenarios

317 6.1. Methodology

318 In Jordan and perhaps other countries it is existing common practice for
319 PV systems’ planners, designers and contractors to predefine cleaning sched-
320 ules in the tendering phase. Usually, the soiling induced effects are taken into
321 consideration very roughly, without utilizing localized testing procedures to

24
Figure 13c: Annual SLI for 80 WP PV Module with Biweekly Cleaning Scenario (Daily
SLI = -0.45%)

322 better predict the effective SLI for the anticipated plant. The PV module
323 type is not usually considered as a factor in the suggested SLI profile. The
324 presented study fills the gap in this perspective, where an educated estimates
325 based on experimental setups are proven to be more accurate. The related
326 cost of cleaning could be a major factor in the overall estimated cost of the
327 PV plant. The authors initiative in this manuscript is to highlight the im-
328 portance of utilizing the generated models of SLI profiles rather than the
329 rough and common used values in the market. Such an approach has not
330 been addressed for similar systems in the region.
331 To achieve this goal, the output power loss and cleaning costs for three
332 different 1 MWP plants using the discussed PV modules types, their exper-
333 imentally determined SLI profiles, two suggested cleaning schedules as well
334 as current local cleaning costs were modeled. To determine the total num-
335 ber of modules required for the construction of the different 1 MWP plants,
336 the data of the single modules were used. PV systems are modular, which
337 means that the output energy (in kWh) is approximately proportional to

25
338 the nominal power of PV generator (in kWP ) for typical PV grid connected
339 systems. This means that small size samples can be trusted or scaled for
340 larger systems. The objective of this study is to offer reliable information
341 for designers to take into consideration in the design and tendering phase for
342 large scale designs. String level data were used in literature as well to reflect
343 on the performance of the large scale plants [19].
344 The designed and installed experimental setup was designed to accurately
345 capture the deteriorations induced by dust accumulation and the resultant
346 power degradation behavior was used to predict the total energy loss in larger
347 size plants. Moreover, the experimented season is chosen as the worst case
348 scenario in the region (summer) to better reflect the captured information.

349 6.2. Analysis

350 In order to study the economic benefit from both cleaning scenarios
351 (monthly and biweekly) in large-scale PV power plants, annual energy pro-
352 duction for a one Mega Watt Peak (1 MWP ) PV power plant is estimated
353 based on the simulated annual energy production results shown in Fig. 14
354 for the three PV modules that were experimented in this paper.
355 For a grid-connected PV system, one or more inverters shall be used to
356 convert DC power produced by PV modules into AC power that is injected
357 into the utility grid. This conversion process dissipates power by the in-
358 verter(s), DC and AC cabling, so the output AC power will be lower than
359 the DC power produced by the PV generator. To account for the dissipated
360 power, the Performance Ratio (P R) of the PV power plant is considered
361 which is defined as the percentage between the actual (simulated in our
362 case) and the theoretical energy outputs of the PV power plant according to

26
363 equation (6):

simulated AC energy (kWh) output


PR = . (6)
theoretical DC energy (kWh) output
364 The theoretical energy output is defined using equation (7):

theoretical DC energy (kWh) output = H · APV · η0 , (7)

365 where,

H effective solar radiation received by PV cells (kWh/m2 )


APV total area of PV generator surface (m2 )
η0 standard, nominal or theoretical PV generator efficiency

366 The theoretical PV generator efficiency (η0 ) is defined as the same datasheet
367 or nameplate PV module efficiency, which is calculated according to equation(8):

P0
η0 = , (8)
G0 · APV
368 where,

P0 nominal power of the PV generator (WP )


APV total area of PV generator surface (m2 )
G0 Global solar irradiance at STC conditions (typical 1000 W/m2 )

369 Accordingly, the simulated AC energy output (Eac ) is calculated using


370 equation (9):

 
H
Eac = P R · P0 · . (9)
G0

27
371 The ratio ( GH0 ) represents the Peak Sun Hours (P SH) that is affected by
372 dust accumulation or soiling on PV surfaces by decreasing H, so the previous
373 equation can be expressed as equation (10):

Eac = P R · P0 · P SH. (10)

374 To study the variations of the performance ratio between large and small-
375 scale PV systems and to reach a reasonable estimate of the performance ratio
376 for the analysis, a PV system design software, PVSyst1 was utilized to com-
377 pare the performance between two grid-connected PV systems with different
378 sizes both using the same PV module and inverter types as well as same
379 location (Amman city in Jordan) and module inclination and orientation.
380 The first system is rated at around 1.2 kWP using four PV modules (310
381 Wp PV module from Suntech company, one of the tested PV modules in
382 this paper) and a single inverter rated at 1.2 kWac from ABB company. The
383 second system is rated at around 1000 kWP using 3228 PV modules (same
384 module type like the first system) and 807 inverters (same inverter type like
385 the first system). Both simulation reports gave exactly the same yield and
386 performance ratio results of 1811 kWh/kWP /year and 79.35%, respectively.
387 As the design of the second system comprises many inverters, which is
388 probably not preferable by PV system designers, a single inverter rated at
389 1000 kWac is selected to design and simulate a third system. The simulation
390 report gave better performance results represented by the yield, which is 1905
391 kWh/kWP /year, and the performance ratio, which is 83.46%.

1
https://www.pvsyst.com/

28
392 For the analysis in this paper we use thus 80% as estimate for the perfor-
393 mance ratio of a grid-connected PV system in Jordan.
394 Substituting P R with 0.8 and P0 with 1000 kWp or 1 MWp in the last
395 equation and applying the equation to the three experimented PV modules
396 with the three scenarios, namely, always clean, soiled with monthly cleaning
397 and soiled with biweekly cleaning, the AC energy output results are obtained
398 as shown below in Fig. 15.
399 In order to build a 1,000,000 WP PV generator d1 MWP /130 WP /modulee,
400 i.e. 7,693 PV modules with 130 WP are required. Using the same calculation
401 we obtain 3,226 PV Modules at 310 WP and 12,500 modules at 80 WP ,
402 respectively.

Figure 14: Simulated Annual Energy Production of 130 WP , 310 WP and 80 WP PV


Modules for Clean and Dirty Scenarios

403 In order to economically compare the monthly and biweekly cleaning


404 scenarios for the three different 1 MWP grid connected PV systems, the
405 electricity production revenues are estimated based on the total annual value
406 of electricity fed into the grid and the applicable feed-in tariff. In addition

29
Figure 15: Simulated Annual Electricity Production for 1 MWP PV Power Plant Based
on 130 WP , 310 WP and 80 WP PV Modules for Clean and Dirty Scenarios

407 the total cost of annual cleaning times per year and the typical cost for one
408 cleaning process for 1 MWP in the Jordanian PV market, which is 700.00
409 USD were taken into consideration. This number reflects the average cleaning
410 costs obtained form different contractors and PV operators in Jordan.
411 The plants have different surface areas, based on the type of modules used
412 which range from 6258.44 m2 in case of 310 WP modules to 9000 m2 in case
413 of 80 WP modules. The average local cleaning costs per cleaning cycle are
414 small thus the cost differences due to the different surfaces will be of minor
415 importance.
416 Accordingly, the monthly cleaning scenario which results in 8 cleaning
417 times per year, the total annual cleaning cost is calcuated at 5,600.00 US$,
418 while for the biweekly cleaning scenario, the total annual cleaning cost is
419 calcuated at 11,200.00 US$. Figures 16a to 16c below show the revenues for
420 the three different 1 MWP projects for varying feed-in tariffs and cleaning
421 scenarios.

30
422 For the Mega-scale PV project based on 130 WP and 310 WP crystalline
423 modules, biweekly cleaning is not feasible compared with monthly clean-
424 ing for a wide range of feed-in tariffs. The biweekly cleaning is justified at
425 high rates of feed-in tariffs of more than 0.5 USD which is hardly realistic
426 in the current installations for similar projects. While for the Mega-scale
427 PV project based on 80 WP thin-film PV module, the biweekly cleaning is
428 feasible for feed-in tariffs above 0.15 USD per kWh, which are reachable for
429 some contracts under the Net-Metering schemes ,in which the rates can reach
430 around 0.35 USD per kWh. It is apparent from the results of this study that
431 the type of project and its tariff greatly influence the economics of cleaning
432 schedules. The PV modules technology to be adapted has direct impact on
433 the generation capacity of PV projects depending on the environmental con-
434 ditions, and thus informed design scenarios must be adapted to increase the
435 potential of the installed PV projects.

436 7. Conclusion

437 In this manuscript, the induced effects of soiling on the performance of


438 PV plants are quantified and modeled. Power analysis technique is used to
439 define the soiling rates and to project the impacts on the energy yield, and
440 thus the economic value. Experimental setup was designed and installed,
441 where the most three common PV modules’ types were tested against dust
442 accumulation. Summer season was chosen to run the experiments to gener-
443 ate experimental SLI for each module. Power degradation model was then
444 extrapolated for each type. This information is very important to large scale
445 PV systems’ designers as it serves as the input to most simulation softwares

31
Figure 16a: Annual Revenues Comparison between Monthly and Biweekly Cleaning Sce-
narios for 1 MWP PV Power Plant based on 130 WP PV Module

446 in the design phase, where an estimated SLI profile should be used. The
447 presented study highlights the importance of including the soiling rate and
448 the forecasted needed cleaning schedules in the planning phase of large scale
449 projects. Energy loss could hit 10% easily in semi-arid areas if cleaning sched-
450 ules are not followed. In this manuscript two different cleaning schedules were
451 examined following the common practice for contractors in the region. It has
452 been shown that different PV technologies exhibit different performance un-
453 der increased soiling rates. Thus, surveying different locations for anticipated
454 soiling rate is necessary prior to designing and deploying large scale plants.
455 Depending on the governing tariff of installed PV projects, the number and
456 distribution of cleaning schedules may vary to achieve economic feasibility as
457 found out in this study. For large scale plants, the cleaning cost per installed
458 kWP is less than small scale systems, thus, intensive cleaning schedules are

32
Figure 16b: Annual Revenues Comparison between Monthly and Biweekly Cleaning Sce-
narios for 1 MWP PV Power Plant based on 310 WP PV Module

Figure 16c: Annual Revenues Comparison between Monthly and Biweekly Cleaning Sce-
narios for 1 MWP PV Power Plant based on 80 WP PV Module

33
459 justified against the energy gain.
460 Acknowledgements
461 This work has been supported and funded by the Scientific Research
462 Support Fund of The Ministry of Higher Education and Scientific Research
463 in Jordan under grant number (ENE/1/10/2015). The authors would like
464 to thank the project office of the Scientific Research Support Fund for their
465 continuous support throughout the lifetime of the project.

466 8. Appendix

467 8.1. Nomenclature

Table 3: Nomenclature

PV Photovoltaic
SLI Soiling Loss Index (%)
G Global solar irradiance incident of PV module (W/m2 )
Gef f Dirty Effective global solar irradiance reaching the cells of the dirty
or soiled PV module.
Gef f Clean Effective global solar irradiance reaching the cells of the clean
PV module
ST C Standard Test Conditions for a PV module at global solar
irradiance of 1000 W/m2 , cell temperature of 25 ◦ C and air
mass or solar spectrum of 1.5
N OCT Nominal Operating Cell Temperature at global solar of
800 W/m2 , ambient temperature of 25 ◦ C,
wind speed of 1 m/s and air mass or solar spectrum of 1.5.

34
G0 Global solar irradiance at STC conditions
(typical 1000 W/m2 )
T0 Operating PV temperature at STC conditions (typical 25 ◦ C)
T Operating PV temperature (◦ C)
Ta Ambient temperature (◦ C)
P Operating PV power at ambient conditions (W)
P0 Nominal PV power at STC conditions (WP )
γ Power temperature coefficient for PV module (% ◦ C−1 )
PR Performance Ratio (%)
H effective solar radiation received by PV cells (kWh/m2 )
APV total area of PV generator surface (m2 )
η0 standard, nominal or theoretical PV generator efficiency
P SH Peak Sun Hours

468 8.2. Specifications of PV Modules

469 The following table provides the technical details of the three PV modules
470 used in the experiments.

Table 4: Technical Specifications for PV Modules under Test

PV Module Nominal
130 WP 310 WP 80 WP
Power Rating
Manufacturer Centrosolar, Suntech, Calyxo,
Germany China Germany
Model. No. SM520S STP310-24/Vem CX3pro 80/2
Cell Technology Wafer-based Wafer-based CdTe
poly-Si poly-Si Thin-film

35
(2 Busbars) (4 Busbars)
Cell Dimensions 156x156 mm 156x156 mm narrow strip
Number of Cells 36 72 116
Module Front Glass Tempered 4 mm Temper- 3.2 mm
glass ed glass glass
Module Back Sheet Blue Tedlar White Tedlar 3.2 mm glass
Module Frame Aluminum Aluminum Frameless
Module Dimensions 1500x680x40 1956x992x40 1200x600x6.9
mm mm mm
1200x600x21.4
mm incl.
junction box
Module Weight 12.1 kg 25.8 kg 12 kg
Max. Module Efficiency 12.7% 16% 8.3 %
@ STC
Open Circuit Voltage 21.9 V 44.9 V 56.7 V
@STC
Short Circuit Current 8.20 A 8.96 A 2.17 A
@ STC
Max. Power Voltage 17.4 V 44.9 V 43.5 V
@ STC
Max. Power Current 7.5 A 8.5 A 1.87 A
@ STC
Nominal Power +/- 5% 0/+ 5 W +10% / -5%
Tolerance @ STC (+/-6.5 W) (+8/-4 W)
Current Temp. +0.028% +0.067% +0.02%
−1 −1
Coefficient @ STC ◦
C ◦
C ◦
C−1
Voltage Temp. -0.36 % -0.33 % -0.24 %
−1 −1
Coefficient @ STC ◦
C ◦
C ◦
C−1
Power Temp. -0.45 % -0.41 % -0.25 %

36
Coefficient @ STC ◦
C−1 ◦
C−1 ◦
C−1

471 8.3. Annual Energy Analysis

472 The following tables specify the annual energy analysis specifying the
473 monthly values assuming a monthly cleaning from March to October and a
474 natural cleaning by rain three times a month from November to February for
475 all three experimented module types.
Table 5: Annual Energy Analysis for 130 WP PV Module for clean and dirty cases
(Monthly Cleaning Scenario)

Avg. Solar Avg. PV Avg. Solar Avg PV


Am- Rad. PV Energy SLI Rad. PV Energy
bient Sum Temp. Output [%] Sum Temp. Output
Month
Temp. Clean (Clean) (Clean) (Dirty) (Dirty) (Dirty)
[◦ C] [kWh/ [◦ C] [kWh] [kWh/ [◦ C] [KWh]
m2 ] m2 ]
Jan 17.9 163.5 25.01 18.9 -0.78 162.2 24.95 18.7
Feb 18.6 175.1 27.15 19.9 -0.74 173.7 27.09 19.8
Mar 21.6 179.0 29.42 20.4 -2.47 174.1 29.21 19.9
Apr 26.2 211.9 35.78 23.5 -2.40 206.9 35.55 23.1
May 28.4 204.3 37.37 22.8 -2.48 199.2 37.15 22.3
Jun 31.6 199.9 40.69 22.0 -2.40 195.1 40.47 21.6
Jul 33.1 209.1 42.23 22.9 -2.48 203.9 42.00 22.4
Aug 34.2 217.5 43.75 23.4 -2.47 212.1 43.51 22.9
Sep 32.2 216.8 42.02 23.2 -2.40 211.7 41.79 22.7
Oct 28.1 184.1 36.89 20.0 -2.27 180.0 36.70 19.6
Nov 21.4 151.6 28.22 17.2 -0.80 150.3 28.17 17.1
Dec 19.4 155.1 26.17 17.9 -0.83 153.8 26.11 17.7
Year 26.06 2267.8 34.56 252.0 -1.88 2223.0 34.39 247.7

476 For the 310 WP PV module, the energy analysis for clean and energy
477 cases are summarized in Table 6 listed below.

37
Table 6: Annual Energy Analysis for 310 WP PV Module for clean and dirty cases
(Monthly Cleaning Scenario)

Avg. Solar Avg. PV Avg. Solar Avg PV


@ Am- Rad. PV Energy SLI Rad. PV Energy
bient Sum Temp. Output [%] Sum Temp. Output
Month
Temp. Clean (Clean) (Clean) (Dirty) (Dirty) (Dirty)
[◦ C] [kWh/ [◦ C] [kWh] [kWh/ [◦ C] [KWh]
m2 ] m2 ]
Jan 17.9 163.5 24.74 45.7 -0.64 162.4 24.69 45.5
Feb 18.6 175.1 26.82 48.3 -0.60 174.0 26.77 48.0
Mar 21.6 179.0 29.12 49.4 -2.00 175.0 28.95 48.5
Apr 26.2 211.9 35.41 57.3 -1.95 207.8 35.24 56.3
May 28.4 204.3 37.03 55.3 -2.01 200.2 36.85 54.3
Jun 31.6 199.9 40.34 53.6 -1.95 196.0 40.17 52.7
Jul 33.1 209.1 41.88 55.7 -2.01 204.8 41.70 54.7
Aug 34.2 217.5 43.38 57.0 -2.01 213.1 43.20 56.0
Sep 32.2 216.8 41.65 56.7 -1.95 212.6 41.47 55.7
Oct 28.1 184.1 36.55 48.7 -1.84 180.8 36.40 47.9
Nov 21.4 151.6 27.96 41.7 -0.65 150.6 27.92 41.5
Dec 19.4 155.1 25.90 43.2 -0.68 154.1 25.86 43.0
Year 26.06 2267.8 34.23 612.7 -1.52 2231.4 34.10 604.0

478 For the 80 WP PV module, the energy analysis for clean and energy cases
479 are summarized in Table 7 listed below.
Table 7: Annual Energy Analysis for 80 WP PV Module for clean and dirty cases (Monthly
Cleaning Scenario)

Avg. Solar Avg. PV Avg. Solar Avg PV


Am- Rad. PV Energy SLI Rad. PV Energy
bient Sum Temp. Output [%] Sum Temp. Output
Month
Temp. Clean (Clean) (Clean) (Dirty) (Dirty) (Dirty)
[◦ C] [kWh/ [◦ C] [kWh] [kWh/ [◦ C] [KWh]
m2 ] m2 ]
Jan 17.9 163.5 24.74 12.3 -2.20% 159.7 24.58 12.0
Feb 18.6 175.1 26.82 13.1 -2.08% 171.3 26.65 12.8

38
Mar 21.6 179.0 29.12 13.4 -6.94% 165.4 28.54 12.4
Apr 26.2 211.9 35.41 15.6 -6.75% 197.7 34.80 14.6
May 28.4 204.3 37.03 15.1 -6.97% 189.8 36.42 14.1
Jun 31.6 199.9 40.34 14.7 -6.74% 186.4 39.75 13.7
Jul 33.1 209.1 41.88 15.3 -6.97% 194.4 41.26 14.3
Aug 34.2 217.5 43.38 15.8 -6.94% 202.2 42.74 14.7
Sep 32.2 216.8 41.65 15.7 -6.74% 202.4 41.02 14.7
Oct 28.1 184.1 36.55 13.4 -6.38% 172.6 36.03 12.6
Nov 21.4 151.6 27.96 11.3 -2.26% 148.1 27.81 11.1
Dec 19.4 155.1 25.90 11.6 -2.35% 151.5 25.75 11.4
Year 26.06 2267.8 34.23 167.2 -5.28% 2141.6 33.78 158.5

480 References

481 [1] V. Tyagi, N. A. Rahim, N. Rahim, A. Jeyraj, L. Selvaraj, Progress in


482 solar PV technology: Research and achievement, Renewable and sus-
483 tainable energy reviews 20 (2013) 443–461.

484 [2] W. Grossmann, K. W. Steininger, C. Schmid, I. Grossmann, Investment


485 and employment from large-scale photovoltaics up to 2050, Empirica
486 39 (2) (2012) 165–189.

487 [3] C. P. Castillo, F. B. e Silva, C. Lavalle, An assessment of the regional


488 potential for solar power generation in EU-28, Energy Policy 88 (2016)
489 86–99.

490 [4] A.A.Babatunde, S. Abbasoglu, M. Senol, Analysis of the impact of dust,


491 tilt angle and orientation on performance of PV Plants, Renewable and
492 Sustainable Energy Reviews 90 (2018) 1017–1026.

39
493 [5] M. Al-Addous, Z. Dalala, C. B. Class, F. Alawneh, H. Al-Taani, Perfor-
494 mance analysis of off-grid PV systems in the Jordan Valley, Renewable
495 Energy 113 (2017) 930–941.

496 [6] R. Conceição, H. G. Silva, L. Fialho, F. M. Lopes, M. Collares-Pereira,


497 PV system design with the effect of soiling on the optimum tilt angle,
498 Renewable Energy 133 (2019) 787 – 796.

499 [7] M. R. Maghami, H. Hizam, C. Gomes, M. A. Radzi, M. I. Rezadad,


500 S. Hajighorbani, Power loss due to soiling on solar panel: A review,
501 Renewable and Sustainable Energy Reviews 59 (2016) 1307–1316.

502 [8] J. Kaldellis, A. Kokala, Quantifying the decrease of the photovoltaic


503 panels’ energy yield due to phenomena of natural air pollution disposal,
504 Energy 35 (12) (2010) 4862–4869.

505 [9] J. Mallineni, K. Yedidi, S. Shrestha, B. Knisely, S. Tatapudi, J. Kuitche,


506 G. TamizhMani, Soiling losses of utility-scale pv systems in hot-dry
507 desert climates: Results from four 4–16 years old power plants, in: 2014
508 IEEE 40th Photovoltaic Specialist Conference (PVSC), IEEE, 2014, pp.
509 3197–3200.

510 [10] S. Ghazi, A. Sayigh, K. Ip, Dust effect on flat surfaces–A review paper,
511 Renewable and Sustainable Energy Reviews 33 (2014) 742–751.

512 [11] H. C. Hottel, Performance of flat-plate solar heat collectors, Trans.


513 ASME 64 91 (1942).

514 [12] H. K. Elminir, A. E. Ghitas, R. Hamid, F. El-Hussainy, M. Beheary,


515 K. M. Abdel-Moneim, Effect of dust on the transparent cover of solar

40
516 collectors, Energy conversion and management 47 (18-19) (2006) 3192–
517 3203.

518 [13] Z. A. Darwish, H. A. Kazem, K. Sopian, M. Al-Goul, H. Alawadhi,


519 Effect of dust pollutant type on photovoltaic performance, Renewable
520 and Sustainable Energy Reviews 41 (2015) 735–744.

521 [14] W. Tian, Y. Wang, J. Ren, L. Zhu, Effect of urban climate on building
522 integrated photovoltaics performance, Energy Conversion and Manage-
523 ment 48 (1) (2007) 1–8.

524 [15] M. Mani, R. Pillai, Impact of dust on solar photovoltaic (PV) perfor-
525 mance: Research status, challenges and recommendations, Renewable
526 and sustainable energy reviews 14 (9) (2010) 3124–3131.

527 [16] M. S. El-Shobokshy, F. M. Hussein, Degradation of photovoltaic cell


528 performance due to dust deposition on to its surface, Renewable Energy
529 3 (6-7) (1993) 585–590.

530 [17] S. Mekhilef, R. Saidur, M. Kamalisarvestani, Effect of dust, humidity


531 and air velocity on efficiency of photovoltaic cells, Renewable and sus-
532 tainable energy reviews 16 (5) (2012) 2920–2925.

533 [18] H. Jiang, L. Lu, K. Sun, Experimental investigation of the impact of


534 airborne dust deposition on the performance of solar photovoltaic (PV)
535 modules, Atmospheric environment 45 (25) (2011) 4299–4304.

536 [19] A. M. Pavan, A. Mellit, D. De Pieri, The effect of soiling on energy


537 production for large-scale photovoltaic plants, Solar energy 85 (5) (2011)
538 1128–1136.

41
539 [20] M. Vivar, R. Herrero, I. Antón, F. Martı́nez-Moreno, R. Moretón,
540 G. Sala, A. W. Blakers, J. Smeltink, Effect of soiling in CPV systems,
541 Solar Energy 84 (7) (2010) 1327–1335.

542 [21] E. Asl-Soleimani, S. Farhangi, M. Zabihi, The effect of tilt angle, air
543 pollution on performance of photovoltaic systems in Tehran, Renewable
544 Energy 24 (3-4) (2001) 459–468.

545 [22] R. Xu, K. Ni, Y. Hu, J. Si, H. Wen, D. Yu, Analysis of the optimum tilt
546 angle for a soiled PV panel, Energy Conversion and Management 148
547 (2017) 100–109.

548 [23] J. Lu, S. Hajimirza, Optimizing sun-tracking angle for higher irradiance
549 collection of PV panels using a particle-based dust accumulation model
550 with gravity effect, Solar Energy 158 (2017) 71–82.

551 [24] M. J. Adinoyi, S. A. Said, Effect of dust accumulation on the power


552 outputs of solar photovoltaic modules, Renewable energy 60 (2013) 633–
553 636.

554 [25] J. Zorrilla-Casanova, M. Philiougine, J. Carretero, P. Bernaola,


555 P. Carpena, L. Mora-López, M. Sidrach-de Cardona, Analysis of dust
556 losses in photovoltaic modules, in: World Renewable Energy Congress-
557 Sweden; 8-13 May; 2011; Linköping; Sweden, no. 057, Linköping Uni-
558 versity Electronic Press, 2011, pp. 2985–2992.

559 [26] J. Tanesab, D. Parlevliet, J. Whale, T. Urmee, T. Pryor, The contribu-


560 tion of dust to performance degradation of PV modules in a temperate
561 climate zone, Solar Energy 120 (2015) 147–157.

42
562 [27] T. Sarver, A. Al-Qaraghuli, L. L. Kazmerski, A comprehensive review of
563 the impact of dust on the use of solar energy: History, investigations, re-
564 sults, literature, and mitigation approaches, Renewable and sustainable
565 energy Reviews 22 (2013) 698–733.

566 [28] A. Sayyah, M. N. Horenstein, M. K. Mazumder, Energy yield loss caused


567 by dust deposition on photovoltaic panels, Solar Energy 107 (2014) 576–
568 604.

569 [29] L. Cristaldi, M. Faifer, M. Rossi, M. Catelani, L. Ciani, E. Dovere,


570 S. Jerace, Economical evaluation of PV system losses due to the dust
571 and pollution, in: 2012 IEEE International Instrumentation and Mea-
572 surement Technology Conference Proceedings, IEEE, 2012, pp. 614–618.

573 [30] M. Faifer, M. Lazzaroni, S. Toscani, Dust effects on the PV plant effi-
574 ciency: A new monitoring strategy, in: 20th IMEKO TC4 Symposium
575 on Measurements of Electrical Quantities: Research on Electrical and
576 Electronic Measurement for the Economic Upturn, Together with 18th
577 TC4 International Workshop on ADC and DCA Modeling and Test-
578 ing, IWADC 2014, IMEKO-International Measurement Federation Sec-
579 retariat, 2014, pp. 580–585.

43

Вам также может понравиться