Characterization of dust accumulated on photovoltaic panels in Doha, Qatar

Solar Energy 142 (2017) 123–135

Contents lists available at ScienceDirect

Solar Energy journal homepage: www.elsevier.com/locate/solener

Characterization of dust accumulated on photovoltaic panels in Doha, Qatar Wasim Javed a, Yiming Wubulikasimu a, Benjamin Figgis b,c, Bing Guo a,⇑ a

Texas A&M University at Qatar, PO Box 23874, Doha, Qatar Qatar Environment & Energy Research Institute, HBKU, PO Box 5825, Doha, Qatar c ICube Laboratory, University of Strasbourg–CNRS, Strasbourg, France b

a r t i c l e

i n f o

Article history: Received 7 June 2016 Received in revised form 31 October 2016 Accepted 29 November 2016 Available online 23 December 2016 Keywords: Dust accumulation rate Exposure time Particle size Chemical composition XRD XRF

a b s t r a c t In this study, samples of dust naturally accumulated for various exposure times on photovoltaic (PV) panels were collected and characterized over a period of ten months in a solar test facility located in Doha, Qatar. The dust accumulation rate (DAR) over the exposure time was determined gravimetrically. The dust samples were characterized using particle size analysis, X-ray fluorescence (XRF), X-ray diffraction (XRD), and scanning electron microscopy (SEM). The cleanness index change rate (CICR), a measure of how fast the PV power output degrades due to soiling, was found to have strong negative correlation with DAR, but the CICR/DAR ratio was found to differ between winter and summer. The DAR and the mean particle size of the accumulated dust both decreased with increasing exposure time, reaching relatively steady values for longer exposure times. Calcium was found to be the most abundant element in the accumulated dust, followed by silicon, iron, magnesium and aluminum. Calcite, dolomite, and quartz were the dominant minerals in the accumulated dust, with gypsum being a minor component. Dust collected after dust-storm events had higher proportions of halite and quartz contents than non-dust-storm days, depending on the direction of the wind. Also, dust particles accumulated on PV panels appeared to agglomerate as the exposure time increased. The data provided in this paper will be useful for quantitatively determine the degree of soiling and its effect on PV performance in Qatar and regions with similar environmental conditions. The data will also be useful for the selection of soiling mitigation technologies. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Environmental concerns and growing energy demand have increased interest in photovoltaic (PV) solar power worldwide in recent years, as a promising renewable energy source. Qatar, as well as other countries in the Middle East and North Africa (MENA) region, has a tremendous potential for development and deployment of solar power generation, due to the high solar irradiation levels and the availability of land. However, the extreme climatic conditions and desert environment in the region pose significant challenges to the successful deployment of solar power generation (Sarver et al., 2013). Numerous studies have shown that dust accumulation on solar surfaces can cause significant degradation of their solar conversion Abbreviations: DAR, dust accumulation rate; CICR, cleanness index change rate; PM10, particulate matter concentration with aerodynamic particle size 610 lm; WS, wind speed; WD, wind direction; RH, relative humidity. ⇑ Corresponding author. E-mail address: [email protected] (B. Guo). http://dx.doi.org/10.1016/j.solener.2016.11.053 0038-092X/Ó 2016 Elsevier Ltd. All rights reserved.

efficiency (Darwish et al., 2015; Said and Walwil, 2014; Sayyah et al., 2014; Tanesab et al., 2015). The effect of soiling is a function of dust loading on the PV module. Dust accumulation rate on module surfaces mostly depends on airborne particle concentration, distribution of aerodynamic particle size, and weather conditions, which are all site specific factors (Said and Walwil, 2014). The dust accumulation rate may also vary with the outdoor exposure time (Mastekbayeva and Kumar, 2000). For example, in the Greek capital, Athens, dust deposition masses of 0.1–1 g m
2 were recorded for the outdoor exposure periods of 2–8 weeks (Kaldellis and Kokala, 2010). A dust loading of 6.184 g m
2 was reported for an exposure period of ten months (February to December) in Dhahran, Saudi Arabia (Adinoyi and Said, 2013). In another work, a dust accumulation rate of 132 mg m
2 day
1 was found in Mesa, Arizona (Boppana, 2015). Similarly, mass accumulation rates of 1– 50 mg m
2 day
1 in Colorado (Boyle et al., 2015) and 150– 300 mg m
2 day
1 in the Minia region, Egypt (Hegazy, 2001), were observed depending on the tilt angle and location. More recently, dust accumulation rates of 10–80 and 5–20 mg m
2 day
1 on 40°

124

W. Javed et al. / Solar Energy 142 (2017) 123–135

inclined glass plates have been reported at Commerce City and the Erie site, respectively, in Colorado, US (Boyle et al., 2016). The characteristics of deposited dust have a significant impact on soiling-related PV performance degradation. Various laboratory studies have shown that the impact of dust soiling on PV energy yield is dependent on the dust physicochemical properties (Appels et al., 2013; Jiang et al., 2011; John et al., 2016; Kaldellis and Fragos, 2011; Kaldellis et al., 2011; Kaldellis and Kapsali, 2011; Khatib et al., 2013; Sulaiman et al., 2011). These laboratory studies have identified 15 types of dust pollutants, of which six pollutant types (i.e. ash, limestone, red soil, calcium carbonate, silica, and sand) are believed to have a greater effect than others on PV performance degradation at a given surface loading (Darwish et al., 2015). However, most of these studies have used artificial dust, and there has been little research regarding the physicochemical characteristics of naturally deposited dust in real-world operating conditions. Soiling studies with adequate natural dust chemical composition information would be useful for understanding the relation between soiling-induced effect and chemical composition. Understanding physicochemical characteristics of accumulated dust is also essential for the development of soiling mitigation technologies. Chemical composition, along with particle size distribution, determines how dust particles interact with the PV surface, and hence affects the applicability and effectiveness of PV soiling mitigation technologies (Horenstein et al., 2013; Johnson et al., 2005; Kazmerski et al., 2016; Mazumder et al., 2015). Various theories suggest that, by rendering the PV module surface hydrophilic or hydrophobic, it is possible to mitigate soiling. However, it has been shown that the affinity of particles to PV surfaces is not only dependent on the PV surface properties, but also the particle properties. These aspects highlight the importance of understanding the characteristics of deposited dust (Kim et al., 2016). There are also active methods of PV soiling mitigation, through the mechanical or electrostatic removal of dust particles from the surface. Information on the particle properties will be useful for assessing the adhesion force of the particles, or for determining their motion in an electric field (Mazumder et al., 2015; Quesnel et al., 2015). Soiling-related PV performance degradation is also dependent on particle size distribution of dust deposited on the surface (ElShobokshy and Hussein, 1993; Pulipaka et al., 2016; Said and Walwil, 2014). Particle size plays a significant role in reflectance, scattering, and absorption of light incident on solar cells and in turn, leads to PV performance degradation. The greater tendency of resuspension for larger particles, even at moderate wind speed, promotes accumulation of smaller-size dust particles (Weber et al., 2014). Finer particles may cause more significant performance degradation of PV modules than larger particles for the same mass of dust. Smaller particles, having the greater specific surface area, may be distributed more uniformly as compared to the coarser dust particles, thus reducing the voids between the particles through which light can pass (Tanesab et al., 2015). Weber et al. (2014) reported that particle sizes deposited on the PV surface mostly lie within a range of 1–50 lm. However, overall there has been limited research on particle size distribution of dust accumulated on PV panels in desert environments. Characterization of accumulated natural dust and its impact on PV system efficiency are limited, given the fact that dust accumulation is a complex phenomenon and varies with site-specific environmental and weather conditions (Kazmerski et al., 2016; Mani and Pillai, 2010). However, through systematically study dust accumulation in various locations of interest to PV power generation, it is hoped that the PV soiling problem may be better understood and addressed. This study was carried out in Doha, Qatar. It was intended to characterize dust accumulation on PV surfaces through a systematic approach. Specifically, this study aimed to

(1) quantify the rate of dust accumulation on PV panels for various exposure periods, and (2) determine the physical and chemical properties of accumulated dust, and their correlation with environmental variables, time of exposure and soiling-induced PV performance loss. 2. Methodology 2.1. Sampling of accumulated dust Sampling of dust accumulated on PV panels was carried out at the Solar Test Facility (STF) (at latitude 25°190 32.6100 N and longitude 51°250 59.8300 E) located at Qatar Foundation, Doha, Qatar. Samples of accumulated dust were collected from the surface of twelve panels of a PV array (CdTe thin film frameless, Pmax 90 W, tilted at 22° and facing due South). Each panel is 1.2 m in width, 0.6 m in height and 6.8 mm in thickness. A photo of the test PV array is shown in Fig. 1. Of the twelve panels, four panels were used to collect daily samples, i.e., accumulated dust was collected every 24 h from these four panels. The 24-h dust samples were collected from 11 January 2015 to 20 February 2015 (winter period), and from 1 June 2015 to 15 July 2015 (summer period). The other eight panels were used to collect two-week, one-month, twomonth and six-month accumulated dust samples; two panels for each sampling frequency. The matrix of dust sampling from the PV panels is shown in Fig. 2. The environmental variables, shown in Fig. 2, have no strong trend that might be constructed to have a role in causing the dependence on exposure time. Dust accumulated on the surface of the PV panels was scraped off with a rubber spatula and carefully collected in a polystyrene Petri dish. The spatula used for scraping dust was made from a polyvinyl chloride acetate (PVCA) card that measures 90 mm by 45 mm by 1 mm. A long side of the spatula was gently pressed against the PV panel surface and moved down to scrape the dust down the lower edge of the panel. The falling dust was collected in the Petri dish. Multiple scrapes were used for each area of the PV panel until the area was clean; i.e. there was no visible dust present. This procedure was carried out for the entire PV panel so as the ensure all dust on the panel was collected 2.2. Analyses of the dust samples All collected dust samples were subjected to the analyses described below. Daily collected dust samples during a month were subjected to gravimetric analysis individually and then

Fig. 1. Photo of a test PV array used in this study. The 12 panels on the outside (perimeter) were used for dust sample collection.

W. Javed et al. / Solar Energy 142 (2017) 123–135

125

Fig. 2. PV Dust Accumulation Sampling Matrix and the environmental variables (Each strip represent an exposure period. Arrows indicate two dust-storm days of February 9 and June 23, 2015).

combined into one-month agglomerated samples for the other analyses. However, daily samples collected on dust-storm days (February 9 and June 23, 2015) were submitted for all analyses separately, without being combined with other daily samples. For gravimetric analysis, the Petri dish was weighed before and after collecting the dust sample to determine the dust sample mass, using a Discovery DV215CD semi-microbalance (Ohaus Corp., Pine Brook, NJ, USA) with a sensitivity of 0.01 mg. The dust samples collected in Petri dishes were acclimatized to the temperature and humidity conditions in the laboratory (20–22 °C; 45 ± 5% RH) for at least 48 h before weighing. Particle size distribution of the collected dust samples was measured using an LS13-321 MW laser diffraction particle size analyzer (Beckman Coulter Inc., Brea, CA) in the dry powder mode. Mineralogy characterization was performed using an Ultima IV X-ray powder diffractometer (Rigaku Corp., Tokyo, Japan) and integrated PDXL2 powder diffraction software. While a ZSX-Primus II X-ray fluorescence (XRF) spectrometer (Rigaku Corp., Tokyo, Japan) was used for elemental analysis. A KratosAxisUltra, X-ray photoelectron spectrometer, XPS (Kratos Analytical, Manchester, UK) was utilized for surface compositional characteristics. Particle morphology analysis was carried out on a Quanta 400 scanning electron microscope (FEI, Hillsboro, OR, USA) equipped with an Apollo XP EDS system (EDAX Inc., Mahwah, NJ, USA). The XRD, XRF, EDS, and XPS analysis were used to established and intercompare the compositional characteristics. The EDS and XPS give the surface compositional information and were used to monitor the consistency in the results and complement the bulk information from XRF.

2.3. PV performance and environmental variable measurements Concurrent PV performance data were collected at the STF testing site. From the PV performance data, the Cleanness Index

Change Rate (CICR), a measure of how fast the PV power output degrades due to soiling, was determined to quantify the PV soiling level. The value of the Cleanness Index ranges from 0 to 1, with 1 for a perfectly clean PV module. Details of the PV data collection and Cleanness Index calculation have been described elsewhere (Guo et al., 2015). In addition, ambient dust concentration (PM10), wind speed (WS), wind direction (WD), and relative humidity (RH) were also measured at the site. PM10 concentration was measured using a TSI 8533EP DustTrak DRX Aerosol Monitor (TSI Inc., Shoreview, MN, USA). WS was measured by a wind speed transmitter (Thies Clima, Gottingen, Germany), WD by a wind direction transmitter (Thies Clima, Gottingen, Germany), and air temperature and RH by a hygro-thermo transmitter-compact (Thies Clima, Gottingen, Germany). The 24-h arithmetic mean values were calculated for PM10, WS, and RH. The 24-h mean of WD was computed by treating all angular measurements as points on the unit circle and computing the resultant vector of the unit vectors determined by data points (Guo et al., 2015).

3. Results and discussion 3.1. Dust accumulation rate 3.1.1. Seasonal variation of DAR The daily average dust accumulation rate (DAR) along with the daily average of four environmental variables, i.e., PM10, WS, RH, and WD, are shown in Fig. 3. It can be seen that daily DAR follows the same trend as the concentration of PM10, while on days when WS was higher (except dust-storm events) DAR was reduced. Mean values of DAR and environmental variables in the winter (11 Jan to 20 Feb 2015) and summer (01 Jun to 15 Jul 2015) periods are given in Table 1. The DAR for the bi-monthly exposure was higher (142 mg m
2 d
1) in the winter period than in the summer period (98 ± 15 mg m
2 d
1). The CICR for the bi-monthly exposure

126

W. Javed et al. / Solar Energy 142 (2017) 123–135

Fig. 3. Daily (24-h) dust accumulation rate (DAR) and corresponding environmental variables during the winter (11-Jan to 20-Feb) and summer (1-June to 15-July) sampling periods.

(PV modules cleaned every two months) was also higher (
1.1% d
1) during the winter period as compared to the summer period (
0.4% d
1). Apparently, this was due to the lower WS and higher RH in the winter period, the favorable conditions for dust accumulation on PV panels, even though PM10 concentration was lower in the winter period. As reported in our previous study (Guo et al., 2015), the daily PV soiling rate (referred to as daily DCI) was positively correlated with WS and negatively correlated with RH. DAR results for various exposure periods from this study are compared with other reports, as shown in Table 2. The daily DAR in this study is in the same range as found in Bankok, Thailand (Mastekbayeva and Kumar, 2000), Minia region, Egypt (Hegazy, 2001), Rumah, Saudi Arabia (Jones et al., 2016) and Baghdad, Iraq (Saidan et al., 2016). The DAR of our study is much higher than Athens, Greece (Kaldellis and Kokala, 2010), Mexico City (Weber et al., 2014), Mesa, Arizona (Boppana, 2015) and Commerce City and Erie, CO, US (Boyle et al., 2016). The various dust accumulation results were recorded with widely varying exposure time, ambient dust concentration, and weather conditions. Hence, there is wide variability in the dust accumulation rates, showing its dependence on the exposure period, time of the year, deposition surfaces, geographical location and prevailing climate conditions (Said and Walwil, 2014). This variation highlights the need for measurement of ambient dust concentration and weather conditions in PV soiling studies at a particular location, so that the soiling results may be interpreted with the relevant input variables (environmental variables), and generalizable soiling relations may be quantitatively derived.

3.1.2. DAR dependence on exposure time The DAR of PV modules also depends on the time of exposure to outdoor environmental conditions. In this study, when examining the relation between DAR and exposure time, we excluded periods that contained rain events. Therefore, what is presented here is primarily the effect of ‘‘dry” exposure time on DAR. However, in some sampling periods, water condensation apparently occurred. We observed signs of water streams flowing down the PV panel. Most such observations were in the winter, when RH was high. The DAR results are shown in Fig. 4. It was observed that recently-cleaned panel surfaces retain more dust than dirty ones. DAR decreased sharply during the first two weeks of exposure, and then became relatively steady as exposure time increased. DAR decreased with increasing time of exposure. For example, the average DAR for the six-month exposure is 80 mg m
2 d
1 (14,547 mg m
2 over 180 days), which is lower than the averages for two-month (120 mg m
2 d
1), one-month (140 mg m
2 d
1), two-week (155 mg m
2 d
1) and 24-h (260 mg m
2 d
1) exposure periods. It can be seen that DAR decrease sharply during the first two weeks of deposition and after that became relatively steady. In other words, daily DAR was higher in the early stages of dust accumulation and steadied as exposure time approached sixmonths. The reason for this trend may be that as dust accumulates on the PV panels, the resuspension rate increases, and hence, the net dust accumulation rate decreases over time. Dust accumulation as a function of exposure time for a short period (one month) has been reported by Mastekbayeva and Kumar (2000). They also observed a sharp decrease in DAR during the initial two weeks of deposition, which dropped sharply from 440 mg m
2 d
1 during the first week to about 120 mg m
2 d
1 after two weeks, and after that, it became relatively steady up to one month of the study period. The accumulated dust mass increased with exposure time as shown in Fig. 5. On average, an accumulated dust mass of about 5 g m
2 was recorded after a month-long accumulation, which then increased to 15 g m
2 after six months of field exposure. The cumulative dust mass over a particular period can be predicted by fitting a power curve, as given in Fig. 5, where ‘‘D” is the exposure time measured in days. For example, one would be able to predict the total amount of accumulated dust after a year from these equations (excluding the cleaning effect of rain). In this study, the amount of dust collected after a year would be 32.87 g m
2 (17.41 and 15.46 g m
2 respectively for Winter and Summer six-month periods), assuming that the weather conditions and all other factors are constant. It must be noted that the trends may vary for different climatic conditions and geographical locations. 3.1.3. Correlation between DAR and other variables The relationship between the DAR and airborne PM concentration, as well as other environmental variables, is not commonly reported in the literature. Also, there is only scarce information about the relationship between accumulated dust characteristics and PV performance reduction. Net dust accumulation on a PV panel is the result of simultaneous deposition of airborne dust and resuspension of deposited dust; the environmental variables

Table 1 Mean values of DAR and other variables. Period

Winter Summer

Mean values (±STD) Bi-monthly DAR (mg m
2 d
1)

Bi-monthly CICR (% d
1)

CICR/DAR ratio (% per 100 mg m
2)

PM10 (lg m
3)

WS (m s
1)

RH (%)

WD (°)

142 98 (±15)


1.1 (±0.9)
0.4 (±0.8)


0.7
0.4

140 (±50) 153 (±60)

1.94 (±0.90) 2.48 (±1.10)

62 (±5) 36 (±10)

270 (±80) 324 (±60)

127

W. Javed et al. / Solar Energy 142 (2017) 123–135 Table 2 Comparison of measured daily DAR of this study with other studies from various cities around the world. Location

Sampling period

Deposition surface

Exposure period

Tilt angle, orientation

DAR (mg m
2 d
1)

Reference

Doha, Qatar

Jan-Oct 2015

PV modules (1.2  0.6 m2)

Daily Biweekly Monthly Bimonthly Six-month

22°, South

215–306 155 140 120 80

This study

Bankok, Thailand

Apr-May 1998

Plastic sheets (30  30 cm2)

6 days- one month

15°, South

125–440

Mastekbayeva and Kumar (2000)

Minia region, Egypt

1999

Glass plates (7  7 cm2)

Weekly

0°, Horizontal 20°, South

230–330 150–200

Hegazy (2001)

Rumah, Saudi Arabia

Jan-Dec 2015

Glass plates (5  5 cm2)

Weekly for 12 months

0°, Horizontal 15°, South

25–190 35–165

Jones et al. (2016)

Baghdad, Iraq

Sept 2014

PV module

Daily Weekly

30°, South-west

210 57

Saidan et al. (2016)

Athens, Greece

Aug-Sept 2009

PV modules (988  448 mm2)

2–8 weeks

30°, South

15–20

Kaldellis and Kokala (2010)

Mexico City

Dec 2012-May 2013

Glass plates (60 cm2)

Weekly

0°, Horizontal

65 (24–102)

Weber et al. (2014)

Colorado, US

Aug 2011-Jun 2014

Glass plates (10  10 cm2)

1–5 weeks

0°, Horizontal 40°, South

20–50 12–35

Boyle et al. (2015)

Mesa, AZ, US

20–22 Oct 2014

PV module

2 days

33°, South-west

132

Boppana (2015)

Commerce City, US Erie, CO, US

Aug 2011-Jun 2014 Nov 2012-May 2014

Glass plates (10  10 cm2)

1–5 weeks (mostly 2 weeks)

0°, Horizontal 40°, South 0°, Horizontal 40°, South

10–120 10–80 5–60 5–20

Boyle et al. (2016)

Fig. 4. The DAR decreases with exposure time in a ‘‘power law” manner. One twomonth (February-March) dust sample was included in the winter, and two samples (April-May and June -July) were included in the summer period. One six-month sample (from March to Aug) was included in the summer period.

mostly influence have been reported to be WS and RH (Figgis et al., 2016). The correlation between daily DAR and environmental variables for 24-h exposure time was also examined in this study. PM10 at the site varied between 40 and 340 lg m
3 with an average of 146 ± 50 lg m
3 during the sampling period. Relations between daily DAR and environmental variables are complex as the WS and RH both have a dual role in overall dust accumulation on the PV surfaces. In general, the average daily DAR was higher at elevated PM10 concentration, higher RH, and lower WS. (Fig. 3). PM10 has a less strong correlation (r = 0.41) with DAR that is due to the fact that particles larger than 10 lm in aerodynamic diameter are deposited but are not counted in the PM10 measurement.

Fig. 5. Cumulative dust loading increases with the exposure time in a power law manner.

The particle size analysis also shows that 35% of particles in the daily-deposited dust are smaller than 10 lm. The rate of resuspension is a function of RH and WS (Figgis et al., 2016; Kim et al., 2016). Higher RH promotes high DAR by reducing the resuspension rate of dust particles. Overall, daily DAR was greater on days with the higher RH levels. Higher RH promotes adhesion of dust particles to PV surfaces and inhibits their resuspension by the wind (Kim et al., 2016). This aspect is consistent with the fact that higher RH increases dust accumulation onto PV modules and hence promotes the adhesion of dust, as the water content of the fallen particles forms a bonding force between the particles and the PV surface. Such an effect of RH on PV soiling has been confirming in our previous study (Guo et al., 2015). The RH affects the threshold WS for resuspension of deposited dust particles. Ibrahim et al. (2004) demonstrated that the threshold WS to detach 50% of stainless steel microspheres (64–76 lm) from

128

W. Javed et al. / Solar Energy 142 (2017) 123–135

glass surfaces in a wind tunnel increased from 3.6 to 13.4 m s
1 as RH increased from 18 to 67%. This result suggests that the dust resuspension effect of high-speed winds is enhanced when the deposited dust particles contain the little moisture, which makes them less sticky and more likely to be carried away by the wind. When humidity is sufficiently high, condensation may occur on the accumulated dust (Figgis et al., 2016). There are two possible outcomes related to condensation. First, this may lead to cementation, in which a crust forms between the particle and surface over successive cycles of condensation and drying (Sarver et al., 2013; Sayyah et al., 2014). Secondly, the condensation may lead to dew formation and results in a partial cleaning effect (Caron and Littmann, 2013). However, there is little information available about this cleaning effect. In contrast, high-speed winds cause higher resuspension of already deposited dust on PV panels and hence decrease DAR. In general, low-speed wind increases dust settlement, while highspeed wind tends to remove dust from PV surfaces and results in cleaning (Figgis et al., 2016; Mani and Pillai, 2010). However, on some high wind-speed days, DAR was high. Those were duststorm events typically, when PM10 was very high. Nonetheless, PV panel position (e.g., tilt angle, orientation) in relation to the wind also affects dust deposition. The correlation between DAR and WD is the weakest of correlations examined in this study (Table 3). This might be explained by the fact that the NW and NWW prevailing winds covered the entire range of wind speeds, but winds from other directions typically only occurred in the low WS range (Guo et al., 2015). Because DAR relatively strongly depended on WS, the peculiar WS, and WD relation led to the weak correlation between DAR and WD (Guo et al., 2015). However, in locations where there is no such biased WS distribution, the dust accumulation rate can exhibit a significant correlation with WD. A study carried out in Minia, Egypt (Elminir et al., 2006) quantified dust deposition on the glass plates installed at different orientations and tilt angles, simultaneously. The results indicate that at lower tilt angles, the dust deposition was higher for glass samples facing the south and southwest, opposite to the prevailing wind direction (from the north and northeast). For north and northeast facing surfaces, the dust accumulation was lower. There is a relatively high correlation between daily DAR and CICR, as shown in Table 3. Of interest is the ratio of CICR to DAR, which quantifies the PV performance degradation as a function of dust loading on the PV surface. As can be seen in Table 1, the CICR/DAR ratio of the winter period is higher than that of the summer period. The values of CICR/DAR ratio found in this study are similar to the ratio of PV power output loss to dust loading as reported by other researchers in a laboratory study (ElShobokshy and Hussein, 1993). However, PV performance loss due to dust loading is affected by several factors, including particle size, chemical composition and the angle of solar incidence (ElShobokshy and Hussein, 1993; Zorrilla-Casanova et al., 2013). From summer to winter, all these factors may vary and contribute to the

difference in the CICR/DAR ratio. Further investigation is needed to provide a satisfactory explanation for this difference. Daily DAR has a stronger correlation with environmental variables than does CICR, according to the correlation coefficients in Table 3. DAR is a quantity that depends on PM10, WS, and RH, whereas the CICR depends directly on DAR (and hence indirectly on PM10, WS, and RH) and apparently other variables that are yet to be determined. As reported in our previous study (Guo et al., 2015), CICR was also negatively correlated with ambient PM concentration, positively correlated with WS, and negatively correlated with RH. The data given in supplementary material also showed a strong correlation between monthly cumulative dust loading and monthly CICR. Overall, daily PV performance was highly reduced on days with higher DAR levels. The data revealed that the phenomenon of dust accumulation is extremely complex and challenging to practically understand in the outdoor environment, given all the factors that influence dust accumulation. 3.2. Dust characteristics 3.2.1. Particle size distribution The size distribution of accumulated dust particles depends on the exposure time of PV modules to the outdoor environmental conditions. Mean particle size of the accumulated dust was found to decrease with increasing exposure time. The 90th percentile (by volume) particle diameter of accumulated dust was 44, 36, 32 and 27 lm for 24-h, one-month, two-month and six-month long exposure times, respectively, as shown in Fig. 6. Deposited dust for 24-h exposure has a higher proportion of larger particles (25–50 lm) and a median particle diameter of 17 lm, compared to dust accumulated for one-month exposure having median particle size of 14 lm. For the six-month exposure, 90% (by volume) of accumulated dust consists of particles smaller than 27 lm and has median, mean, and mode particle diameters of 9, 13, and 16 lm, respectively. Month-by-month accumulated dust samples also show significant variation in particle size distribution (Fig. 7). It was found that mean particle size of accumulated dust was smaller in April and May. This might be due to the higher wind speeds and lowest relative humidity during these two months (data given in the Supplementary material). Mean particle size was larger in June and July, which had more dust storms. Typically, 90% particles (by volume) of dust are smaller than 36 lm on average for all month-long dust samples. The volume-weighted mean and median particle sizes of one-month accumulated samples vary within a broad range of 12– 25 and 9–16 lm, respectively. It was also observed that about 35% of daily-deposited dust and 40% of monthly-deposited dust comprised particles smaller than 10 lm, which depicts the strong correlation between DAR and airborne PM10 concentration (Table 3). Overall, the size analysis indicates that the longer the exposure time (i.e., the older the dust accumulation) of the PV panel surfaces, the higher proportion of smaller size particles as seen in the case of six-month long dust sample. Mean and median particle

Table 3 Pearson correlation coefficient matrix for daily DAR and CICR with the different environmental variables.

DAR CICR PM10 RH WS WD *

DAR

CICR

PM10

RH

WS

WD

1
0.60* 0.41* 0.33*
0.35*
0.09

1
0.15
0.26 0.22 0.08

1
0.17 0.19
0.07

1
0.39*
0.37*

1 0.48*

1

Significant at p < 0.05 using 2-tailed t-test. The correlation analysis is based on the daily dust samples for 24-h exposure time and daily average values of other variables.

W. Javed et al. / Solar Energy 142 (2017) 123–135

Fig. 6. Box plot showing particle size distribution of accumulated dust as a function of exposure time.

Fig. 7. Box plot showing particle size distribution of accumulated dust by monthlong exposure time.

size in the accumulated dust was found to decrease with increasing exposure time, with the 90th percentile particle size falling from 44 lm for 24-h exposure to 27 lm for six-month exposure. The observed smaller grain-size particles suggest that the dust is originated mainly from local active construction or vehicular sites with minimal contribution of the desert sand, which has larger grains. Particle size plays a significant role in both dust settling and resuspension of dust on a PV surface. The high tendency of resuspension of larger particles, even at moderate WS, results in a greater fraction of smaller-sized particles in the accumulated dust at longer exposure time. Boor et al. (2013) reported that multilayer deposits show significantly higher resuspension than single layer deposits. Furthermore, the bigger particles moving along the surface after detachment strike stationary particles on the surface, inducing further detachment (Kim et al., 2016). Fine dust particles readily stick to the PV surfaces when there is no rain. This condition will be aggravated by high humidity to create a cementing effect on the surfaces (Sayyah et al., 2014). Typically, we found that 95% (by volume) of accumulated dust particles lie in the 1–50 lm diameter range. The higher accumulation of smaller-size particles on tilted PV panels is due to the greater resuspension of larger particles from dry surfaces at higher WS. Other reports of outdoor experiments also show that particle sizes of deposited dust on the PV surface mostly lie in the range of 1–50 lm (Weber et al., 2014). Particle size is also an important factor in the accumulated dust’s ability to scatter and attenuate solar irradiation. Therefore, for the same dust loading (mass) on the PV surface, PV perfor-

129

mance degradation may be dependent on the particle size accumulated. Finer particles may cause stronger degradation in PV performance than larger particles, for the same mass loading (Qasem et al., 2014). It is attributed to the fact that finer particles have more specific surface area and distributed more uniformly as compared to the coarser dust particles, thus reducing the voids between the particles through which light can pass (ElShobokshy and Hussein, 1993; Tanesab et al., 2015). Smaller dust particles scatter shorter wavelengths more than the bigger particles, result in more light attenuation, especially at shorter wavelengths, and this trend increases with accumulated dust mass (Qasem et al., 2014). We anticipate that for particles of mixed size distributions, smaller particles can fill the voids between large particles. Hence, the distribution becomes more packed, and further reduces the available PV area for light capture. In addition, the fundamental energy requirement for cleaning is also dependent on particle size distribution of the deposited dust, either mechanically or electrostatically. For example, Kawamoto and Shibata (2015) reported a consistent drop in the dust removal efficiency by an electrodynamic shield for smaller-size (<25 lm) particles while larger (50–300 lm) particles are much easier to remove. Other methods of cleaning are likely to encounter a similar decrease in cleaning efficiency as particle size distribution decreases. For example, Appels et al. (2013) observed that bigger dust particles (60 lm) seem to be more easily removed by rain than smaller particles (2–10 lm). These impacts of particle sizes highlight the importance of investigating the size characteristics of deposited dust. 3.2.2. Dust chemical composition The chemical composition of deposited dust has a significant impact on particle adhesion with the glass surface and is consequently essential for cleaning method designs, either mechanical, electrodynamic, or the design of coating system (Horenstein et al., 2013; Johnson et al., 2005; Kazmerski et al., 2016). Information on the chemical composition of dust will be useful for assessing optical properties of the particles, or for determining the motion of the particles in an electric field (John et al., 2016; Mazumder et al., 2015). Knowledge of the chemical composition of dust may also help to identify its origin. XRD analysis of all 24-h and one-month accumulated dust samples show no significant variation in major mineral composition. Typically, dust accumulated on modules for various exposure times have a similar mineral composition, as illustrated in Fig. 8. It was determined that calcite (CaCO3), dolomite (CaMg[CO3]2) and quartz (SiO2) are the dominant minerals of all collected dust samples. Both calcite and dolomite compounds accounted for more than 70% of the dust particle contents in all one-month accumulated samples (Fig. 9). The next most prevalent minerals are quartz and gypsum, with the remainder other primary silicate minerals (i.e., palygorskite, albite, cristobalite and kaolinite). Of these dominant minerals, six-month accumulated dust has a high proportion (15%) of gypsum (CaSO42H2O) compared to the 24-h, one-month and two-month accumulated samples (Fig. 8). High contents of calcite and dolomite in the dust mainly originate from local soils. Engelbrecht et al. (2009) reported that Qatar and UAE soils mostly contain significant amounts of calcite (33– 48%) and dolomite minerals compared to other Gulf countries such as Iraq and Kuwait, where soil has substantial amounts of quartz and feldspar minerals. They also suggested that both Qatar and the UAE receive wind-blown dust from the Arabian Peninsula, Iraq, and Kuwait. Similarly, Kazmerski et al. (2016) analyzed the composition of deposited dust gathered from module surfaces in desert regions of Middle East, i.e., Libya, Dhahran, Riyadh, and Baghdad and found that the deposited dust is dominated by quartz, followed by calcite and primary silicates (i.e., Feldspar and Muscovite). Aissa

130

W. Javed et al. / Solar Energy 142 (2017) 123–135

Fig. 8. Mineralogical composition by XRD analysis of accumulated dust as a function of exposure time.

et al. (2016) and Yilbas et al. (2015) have also reported the high content of calcite along with varying components of other silicate minerals in dust particles collected from various locations in the Middle East. So, it can be anticipated that dust mineralogy varies substantially for various locations and climate zones in the region.

The dominant minerals, i.e., calcite and dolomite, are attributed to construction activities in the local urban environment, as these are associated with limestone, marble, Portland cement, and concrete. Considering that Qatari soils are ‘‘Calcisols” and dominated by calcite and dolomite minerals (Engelbrecht et al., 2009), wind erosion from land and construction sites is likely to be the primary source of dust deposited on PV module surfaces. The dust particles elemental analysis by XRF for all collected dust samples for various exposure times indicated that calcium from calcite was the most abundant element, followed by silicon from quartz and minor elements including iron, magnesium, aluminum, sodium and potassium (Fig. 10). All detected constituent elements were matched to the corresponding phase identified by XRD. The elements (Ca, Si, Fe, Mg and Al) are present in all dust samples and mostly associated with natural soil crust, while enriched in the coarse dust particles. The 24-h dust samples have a relatively high proportion of Na and Cl elements originating mainly from sea spray (Yilbas et al., 2015), compared to accumulated dust samples associated with longer exposure times, suggesting these elements are associated with large size particles. We observed that 24-h accumulated dust has a high fraction of larger-size particles. It is reported that mixing of dust particles with sea salt systematically leads to the growth of particles and dust particles become larger as more sea salt is mixed with them (Zhang and Iwasaka, 2004). The sampling site is close to (approximately 10 km) to the Arabian Gulf so dust particles are likely affected by sea salt. It was further evident that the 9-Feb dust storm coming from the sea had very high fractions of these seaassociated elements. The six-month-long accumulated dust has relatively high content of sulfur (S), corresponding to gypsum, compared to dust samples of shorter exposure times (Fig. 10a). The S concentration can be associated with Ca in the dust as gypsum. Moreover, it has been reported that the absorption of SO2 via heterogeneous reaction on the mineral dust, particularly CaCO3, is an important sink for SO2 and contributes to sulfate particle formation (Zhao et al., 2013).

Fig. 9. Mineralogical composition by XRD analysis of monthly accumulated dust.

W. Javed et al. / Solar Energy 142 (2017) 123–135

131

Fig. 10. The comparison of the bulk elemental composition by XRF analysis of accumulated dust (a) by exposure time (b) by month.

So, it can be anticipated that deposited dust particles can age during the long periods of exposure to SO2 in the atmosphere. In addition to natural minerals, traces of contaminating elements (Zn, Cu, Ti, Ni, Cr) were also observed. The sampling site is located in an urban area and surrounded by light vehicular traffic so may be affected by contamination sources such as the resuspension of road dust particles, vehicle fleet exhaust, brake and tire wear. All collected dust samples were also analyzed by EDS and XPS to determine the surface chemistry of particles. EDS and XPS results (not shown) allowed us to compare and confirm the elemental composition of the examined dust particles by XRF. The surface analyses (EDS, XPS) did not show a substantial difference in elemental composition of a dust sample when compared to the XRF results. However, some differences were observed in S, K, Na, and Cl concentrations, showing differences in bulk and surface concentrations. The surface analysis revealed slightly higher concentrations of S, K, Na, and Cl, reflecting the association of these elements with particle surfaces. This suggests that these hygroscopic species, present on the dust particle surface, render the particle are more likely to adhere to the glass and bind together to agglomerate or form cement like films. In such process, soluble materials on the surface of the particles are dissolved when

exposed to high humidity, morning dew, or light rain and then upon drying formed a cementitious layer (Kazmerski et al., 2016; Sarver et al., 2013). 3.2.3. Dust particle morphology The SEM analysis indicated that dust particle morphology changes with increasing exposure time. The dust particles took irregular shapes in shorter exposure time samples. In longer exposure time samples, particles tended to be agglomerated and covered by small features. Long-exposed dust samples had more fine-size particles. The change in particle shape and size may result in a change in chemical composition of dust particles (Yilbas et al., 2015). Such particle morphology change with exposure time can be seen in Fig. 11, which shows the images for 24-h, one-month, and six-month accumulated dust samples. The daily (24-h) dust-accumulated sample has bigger particles and individual particles have rougher and clearer edges. Particles are uniformly distributed on the copper tape, covering the whole area without any identified particles cluster (Fig. 11a). The onemonth dust sample has more small size dust particles with a few bigger ones with the size of >20 lm, compared to 24-h deposited samples during the same month of accumulation. In the case of

132

W. Javed et al. / Solar Energy 142 (2017) 123–135

Fig. 11. The SEM images of dust samples deposited for (a) 24-h, (b) one-month and (c) six-month exposure periods, at 200 (1st column), 800 (2nd column) and 1600 (3rd column) magnification.

six-month accumulated dust sample, there is a predominance of finer (<10 lm) particles which have a tendency to agglomerate. Most smaller particles are stuck together, forming numerous large clusters. The few bigger particles (with the size of 17–22 lm) are mostly covered with and surrounded by many smaller (<5 lm) particles, forming large agglomerates (Fig. 11c). The high proportion of smaller particles over an extended period of accumulation was due to the natural processes of resuspension by which larger size particles were removed by wind or light rain. These results are in accordance with the earlier results of particle size distribution. It was observed that more particles appeared to be agglomerated and covered by small features with longer accumulation time. It was reported that soluble material on the particles’ surfaces (like Na, Cl, K, S, C) reacts with

moisture/humidity to be dissolved, which results in agglomeration of grains into clusters (Kazmerski et al., 2016; KlugmannRadziemska, 2015). The dissolution of these alkali and alkaline compounds results in the formation of a liquid solution (mud) between particles, and upon drying it adheres the particles by bridging the gaps/cavities in-between the dust particles (Yilbas et al., 2015). This results in the cementation of dust particles with the PV module surfaces especially in hot-humid conditions like Doha. It is observed in the field that the six-month long exposed module has a thick cement-like dust layer that is firmly bound to the glass surface of the module. This cement film formed due to the binding of the dust particles together as well as adhesion to the glass surface, which is tough to remove, and required extensive scraping.

W. Javed et al. / Solar Energy 142 (2017) 123–135

133

Fig. 12. The comparison of particle size distribution of accumulated dust during the normal days and dust-storm events.

3.3. Dust-storm episodes Dust storms occur in this arid region of the Middle East. Dust storms severely cause a reduction of visibility, decline of solar radiation and, in consequence, reduce the output of solar devices. Eight dust-storms (having daily mean PM10 concentration of >200 lg m
3) were recorded during the study period (January– October 2015). Of these, two are characterized and described in this section, one occurring in the winter period on 9-Feb and another in the summer time on 23-Jun. The first dust-storm event had daily mean dust (PM10) concentration of 207 lg m
3 and the second one 220 lg m
3. Dust mass accumulated during these dust episode days was 762 and 266 mg m
2, respectively. As expected, the dust mass accumulated on PV surfaces during the dust-storm days was significantly higher than on normal days. It was also observed that the 9-Feb dust storm resulted in more dust accumulation on PV panels than the other one of 23-Jun, as it was accompanied by relatively the lower WS and higher RH - favorable environmental conditions for the dust accumulation (data given in the supplementary material). The corresponding cleanness index change rate (CICR, a measure of PV performance due to soiling) was found as
2.41% and
0.64% per day during these dust event days, respectively.

Fig. 13. The comparison of XRD patterns of the mineralogical composition of accumulated dust during the normal days and dust-storm events.

During the dust storms, particle size of the dust accumulated on the PV panels was quite different compared to normal days (Fig. 12). More precisely, large size particles (30–60 lm) were dominant in the case of dust storms, unlike in normal conditions. On the normal days, smaller particles are more prevalent and bigger ones much rarer. Thus, particles deposited during dust storms tended to be larger than those deposited on normal days. One of the most interesting findings of dust characterization is that the chemical composition of dust particles deposited on PV panels during these dust-storm events was entirely different from normal days (Fig. 13 and 14). These XRF and XRD observations were corroborated by the SEM-EDS composition (data not shown). Dust-storm particles have a high proportion of halite (NaCl) and quartz (SiO2) in addition to calcite and dolomite minerals, depending upon the wind direction of the dust-storm event and, in turn, the dust source. The dust storm on 9-Feb came from the Arabian Gulf (140°; the SE direction) and had about 25% halite (NaCl)

Fig. 14. The comparison of the bulk elemental composition by XRF analysis of accumulated dust during the normal days and dust-storm events.

134

W. Javed et al. / Solar Energy 142 (2017) 123–135

(originating from sea spray), and 13% quartz (SiO2). The dust storm on 23-Jun came from the NW (315°) direction and had higher (21%) quartz mineral and 7% halite mineral, as shown in Fig. 13. So, it can be anticipated that dust source during these dust storms is not local. The source of the 23-Jun dust storm is probably the Arabian desert that is mainly comprised of quartz and calcite (Engelbrecht et al., 2009; Kazmerski et al., 2016). The 9-Feb dust storm most likely originated from the desert of UAE and showed mixing of dust particles with sea spray that resulted in the increase of sea-salt elements in the dust (Zhang and Iwasaka, 2004). 4. Conclusion The rate of dust accumulation, DAR, reaches approximately 100 mg m
2 d
1 for a two-month exposure time, but its value can be more than twice as high for short exposure times. The 90th percentile particle size (volume based) of the accumulated dust is about 32 lm for two-month exposure, but this statistic generally decreases with increasing exposure time. Calcium was found to be the most abundant element in the accumulated dust, followed by silicon, iron, magnesium and aluminum. Calcite, dolomite, and quartz were the dominant minerals in all accumulated dust samples. Particle agglomeration occurs as the dust is allowed to accumulate on the PV panel. The dust characterization results will be helpful for understanding PV soiling in Qatar and regions that share a similarity in ambient dust and weather conditions. The data contained in this paper will be useful for prediction of PV performance loss as based on environmental conditions and development of optimum cleaning approach, which is of interest to project developers and operators. Acknowledgments This work was financially supported by the Qatar National Research Fund (QNRF) through a NPRP grant (Project No. 7-9872-372) and a UREP grant (Project No. 15-083-2-030). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.solener.2016.11. 053. References Adinoyi, M.J., Said, S.A.M., 2013. Effect of dust accumulation on the power outputs of solar photovoltaic modules. Renew. Energy 60, 633–636. Aissa, B., Isaifan, R.J., Madhavan, V.E., Abdallah, A.A., 2016. Structural and physical properties of the dust particles in Qatar and their influence on the PV panel performance. Sci. Rep. 6, 31467. Appels, R., Lefevre, B., Herteleer, B., Goverde, H., Beerten, A., Paesen, R., De Medts, K., Driesen, J., Poortmans, J., 2013. Effect of soiling on photovoltaic modules. Sol. Energy 96, 283–291. Boor, B.E., Siegel, J.A., Novoselac, A., 2013. Monolayer and multilayer particle deposits on hard surfaces: literature review and implications for particle resuspension in the indoor environment. Aerosol Sci. Technol. 47 (8), 831–847. Boppana, S., 2015. Outdoor Soiling Loss Characterization and Statistical Risk Analysis of Photovoltaic Power Plants. Arizona State University. Boyle, L., Flinchpaugh, H., Hannigan, M., 2016. Assessment of PM dry deposition on solar energy harvesting systems: measurement–model comparison. Aerosol Sci. Technol. 50 (4), 380–391. Boyle, L., Flinchpaugh, H., Hannigan, M.P., 2015. Natural soiling of photovoltaic cover plates and the impact on transmission. Renew. Energy 77, 166–173. Caron, J.R., Littmann, B., 2013. Direct monitoring of energy lost due to soiling on first solar modules in California. IEEE J. Photovolt. 3 (1), 336–340. Darwish, Z.A., Kazem, H.A., Sopian, K., Al-Goul, M.A., Alawadhi, H., 2015. Effect of dust pollutant type on photovoltaic performance. Renew. Sustain. Energy Rev. 41, 735–744. El-Shobokshy, M.S., Hussein, F.M., 1993. Effect of dust with different physical properties on the performance of photovoltaic cells. Sol. Energy 51 (6), 505– 511.

Elminir, H.K., Ghitas, A.E., Hamid, R.H., El-Hussainy, F., Beheary, M.M., AbdelMoneim, K.M., 2006. Effect of dust on the transparent cover of solar collectors. Energy Convers. Manage. 47 (18–19), 3192–3203. Engelbrecht, J.P., McDonald, E.V., Gillies, J.A., Jayanty, R.K., Casuccio, G., Gertler, A. W., 2009. Characterizing mineral dusts and other aerosols from the Middle East – Part 2: grab samples and re-suspensions. Inhal. Toxicol. 21 (4), 327–336. Figgis, B., Ennaoui, A., Guo, B., Javed, W., Chen, E., 2016. Outdoor soiling microscope for measuring particle deposition and resuspension. Sol. Energy 137, 158–164. Guo, B., Javed, W., Figgis, B., Mirza, T., 2015. Effect of dust and weather conditions on photovoltaic performance in Doha, Qatar. In: First Workshop on Smart Grid and Renewable Energy (SGRE). IEEE, pp. 1–6. Hegazy, A.A., 2001. Effect of dust accumulation on solar transmittance through glass covers of plate-type collectors. Renew. Energy 22 (4), 525–540. Horenstein, M.N., Mazumder, M.K., Sumner, R.C., Stark, J., Abuhamed, T., Boxman, R., 2013. Modeling of trajectories in an electrodynamic screen for obtaining maximum particle removal efficiency. IEEE Trans. Ind. Appl. 49 (2), 707–713. Ibrahim, A.H., Dunn, P.F., Brach, R.M., 2004. Microparticle detachment from surfaces exposed to turbulent air flow: effects of flow and particle deposition characteristics. J. Aerosol Sci. 35 (7), 805–821. Jiang, H., Lu, L., Sun, K., 2011. Experimental investigation of the impact of airborne dust deposition on the performance of solar photovoltaic (PV) modules. Atmos. Environ. 45 (25), 4299–4304. John, J.J., Warade, S., Tamizhmani, G., Kottantharayil, A., 2016. Study of soiling loss on photovoltaic modules with artificially deposited dust of different gravimetric densities and compositions collected from different locations in India. IEEE J. Photovolt. 6 (1), 236–243. Johnson, C., Srirama, P., Sharma, R., Pruessner, K., Zhang, J., Mazumder, M., 2005. Effect of particle size distribution on the performance of electrodynamic screens. In: Conference Record of the Industry Applications Conference, 2005. Fourteenth IAS Annual Meeting. IEEE, pp. 341–345. Jones, R.K., Baras, A., Al Saeeri, A., Al Qahtani, A., Al Amoudi, A.O., Al Shaya, Y., Alodan, M., Al-Hsaien, S.A., 2016. Optimized cleaning cost and schedule based on observed soiling conditions for photovoltaic plants in central Saudi Arabia. IEEE J. Photovolt. 6 (3), 730–738. Kaldellis, J.K., Fragos, P., 2011. Ash deposition impact on the energy performance of photovoltaic generators. J. Clean. Prod. 19 (4), 311–317. Kaldellis, J.K., Fragos, P., Kapsali, M., 2011. Systematic experimental study of the pollution deposition impact on the energy yield of photovoltaic installations. Renew. Energy 36 (10), 2717–2724. Kaldellis, J.K., Kapsali, M., 2011. Simulating the dust effect on the energy performance of photovoltaic generators based on experimental measurements. Energy 36 (8), 5154–5161. Kaldellis, J.K., Kokala, A., 2010. Quantifying the decrease of the photovoltaic panels’ energy yield due to phenomena of natural air pollution disposal. Energy 35 (12), 4862–4869. Kawamoto, H., Shibata, T., 2015. Electrostatic cleaning system for removal of sand from solar panels. J. Electrostat. 73, 65–70. Kazmerski, L.L., Diniz, A.S.A., Maia, C.B., Viana, M.M., Costa, S.C., Brito, P.P., Campos, C.D., Neto, L.V.M., de Morais Hanriot, S., de Oliveira Cruz, L.R., 2016. Fundamental studies of the adhesion of dust to PV module chemical and physical relationships at the microscale. IEEE J. Photovolt. 6 (3), 719–729. Khatib, T., Kazem, H., Sopian, K., Buttinger, F., Elmenreich, W., Albusaidi, A.S., 2013. Effect of dust deposition on the performance of multi-crystalline photovoltaic modules based on experimental measurements. Int. J. Renew. Energy Res. 3 (4), 850–853. Kim, Y., Wellum, G., Mello, K., Strawhecker, K.E., Thoms, R., Giaya, A., Wyslouzil, B.E., 2016. Effects of relative humidity and particle and surface properties on particle resuspension rates. Aerosol Sci. Technol. 50 (4), 339–352. Klugmann-Radziemska, E., 2015. Degradation of electrical performance of a crystalline photovoltaic module due to dust deposition in northern Poland. Renew. Energy 78, 418–426. Mani, M., Pillai, R., 2010. Impact of dust on solar photovoltaic (PV) performance: research status, challenges and recommendations. Renew. Sustain. Energy Rev. 14 (9), 3124–3131. Mastekbayeva, G., Kumar, S., 2000. Effect of dust on the transmittance of low density polyethylene glazing in a tropical climate. Sol. Energy 68 (2), 135–141. Mazumder, M.K., Horenstein, M.N., Heiling, C., Stark, J.W., Sayyah, A., Yellowhair, J., Raychowdhury, A., 2015. Environmental degradation of the optical surface of PV modules and solar mirrors by soiling and high RH and mitigation methods for minimizing energy yield losses. In: 42nd Photovoltaic Specialist Conference (PVSC). IEEE, pp. 1–6. Pulipaka, S., Mani, F., Kumar, R., 2016. Modeling of soiled PV module with neural networks and regression using particle size composition. Sol. Energy 123, 116– 126. Qasem, H., Betts, T.R., Müllejans, H., AlBusairi, H., Gottschalg, R., 2014. Dust-induced shading on photovoltaic modules. Progr. Photovolt.: Res. Appl. 22 (2), 218–226. Quesnel, D.J., Rimai, D.S., Schaefer, D.M., Beaudoin, S.P., Harrison, A., Hoss, D., Sweat, M., Thomas, M., 2015. Aspects of particle adhesion and removal. In: Kohli, R., Mittal, K.L. (Eds.), Developments in Surface Contamination and Cleaning, Fundamentals and Applied Aspects, vol. 1. Elsevier Inc., San Diego, CA, USA, pp. 119–145. Said, S.A.M., Walwil, H.M., 2014. Fundamental studies on dust fouling effects on PV module performance. Sol. Energy 107, 328–337. Saidan, M., Albaali, A.G., Alasis, E., Kaldellis, J.K., 2016. Experimental study on the effect of dust deposition on solar photovoltaic panels in desert environment. Renew. Energy 92, 499–505.

W. Javed et al. / Solar Energy 142 (2017) 123–135 Sarver, T., Al-Qaraghuli, A., Kazmerski, L.L., 2013. A comprehensive review of the impact of dust on the use of solar energy: history, investigations, results, literature, and mitigation approaches. Renew. Sustain. Energy Rev. 22, 698– 733. Sayyah, A., Horenstein, M.N., Mazumder, M.K., 2014. Energy yield loss caused by dust deposition on photovoltaic panels. Sol. Energy 107, 576–604. Sulaiman, S.A., Hussain, H.H., Leh, N., Razali, M.S., 2011. Effects of dust on the performance of PV panels. World Acad. Sci. Eng. Technol. 58, 588–593. Tanesab, J., Parlevliet, D., Whale, J., Urmee, T., Pryor, T., 2015. The contribution of dust to performance degradation of PV modules in a temperate climate zone. Sol. Energy 120, 147–157. Weber, B., Quinones, A., Almanza, R., Duran, M.D., 2014. Performance reduction of PV systems by dust deposition. Energy Procedia 57, 99–108.

135

Yilbas, B.S., Ali, H., Khaled, M.M., Al-Aqeeli, N., Abu-Dheir, N., Varanasi, K.K., 2015. Influence of dust and mud on the optical, chemical, and mechanical properties of a PV protective glass. Sci. Rep. 5, 15833. Zhang, D.Z., Iwasaka, Y., 2004. Size change of Asian dust particles caused by sea salt interaction: measurements in southwestern Japan. Geophys. Res. Lett. 31 (15). Zhao, Y., Chen, Z., Shen, X., Huang, D., 2013. Heterogeneous reactions of gaseous hydrogen peroxide on pristine and acidic gas-processed calcium carbonate particles: effects of relative humidity and surface coverage of coating. Atmos. Environ. 67, 63–72. Zorrilla-Casanova, J., Piliougine, M., Carretero, J., Bernaola-Galván, P., Carpena, P., Mora-López, L., Sidrach-de-Cardona, M., 2013. Losses produced by soiling in the incoming radiation to photovoltaic modules. Progr. Photovolt.: Res. Appl. 21, 790–796.