Skip to content
BY 4.0 license Open Access Published by De Gruyter Open Access March 9, 2023

Contribution to the building of a weather information service for solar panel cleaning operations at Diass plant (Senegal, Western Sahel)

  • Mohamed Cherif Aidara EMAIL logo , Pape Abdoulaye Fam , Derrick Kwadwo Danso , Eric Mensah Mortey , Amy Mbaye , Mamadou Lamine Ndiaye , Abdou Latif Bonkaney , Rabani Adamou , Sandrine Anquetin and Arona Diedhiou
From the journal Open Geosciences


The accumulation of dust on the surface of solar panels reduces the amount of sunlight reaching the solar cells and results in a decrease in panel performance. To avoid this loss of production and thus, to improve the performance capacity, solar panels must be cleaned frequently. The West African region is well known for its high solar energy potential. However, this potential can be reduced by the high occurrence of dust storms during the year. This article aims to provide a contribution to the construction of a meteorological information service for solar panel cleaning operations at Diass solar plant in Senegal (Western Sahel). It is based on a full year in situ experiment comparing the power loss due to dust between solar panels cleaned at different frequencies and those not cleaned. The model to determine the cleaning frequencies is based on the deposition rate of airborne particles, the concentration of airborne particles, and the density of the dust that has a major impact on the power loss. Cleaning frequencies are presented at seasonal scale because in the study area, dust episodes differ according to the seasons. A cost–benefit analysis is also performed to demonstrate the advantage of using weather information service to support the dust cleaning operations at Diass plant. As results, it is found that cleaning every 3 weeks is required during the dry seasons, December–January–February and March–April–May. During the rainy season, cleaning every 5 weeks is recommended in June–July–August, while in September–October–November cleaning every 4 weeks is sufficient to maintain an optimal performance of the solar panel. The total costs of cleaning operations based on these results are reduced compared to the current costs of cleaning and the benefits are much higher than without cleaning action.

1 Introduction

The African continent represents a huge potential market for the photovoltaic sector. Many international investors show interest and support solar power plant projects in African countries. For example, the following solar projects has been launched in the past few years: “Energy for All in Africa,” “New Deal” for energy in Africa, “Scaling solar,” “Power Africa.” The number of investments and initiatives to improve the continent’s electrification rate is increasing. At the end of 2015, Africa had 2,100 MW of photovoltaic solar installations [1]. Since then, the continent has almost quadrupled the installed capacity of solar plants increasing to more than 8 GW at the end of 2019 [1]. However, this enhancement is still modest compared to the solar energy potential that can be achieved in Africa. Projects are developing at a spectacular rate, and if this dynamic continues, the International Energy Agency predicts that by 2030, solar energy could represent 14% of the installed capacity in Africa [2]. The solar energy revolution is happening in many countries in Africa. For instance, in February 2016, Morocco had inaugurated “Noor” the seventh-largest thermodynamic solar power plant in the world with more than half a million solar panels spread over more than 480 ha [3], which should allow Morocco to meet its energy needs for more than one million households within the next 5 years.

Eight months later, in October 2016, Senegal inaugurated “Senergy 2,” which at that time was the largest solar power plant in West Africa with 75,000 photovoltaic panels and a capacity of 20 MW, covering the needs of 200,000 Senegalese households. Furthermore, other power plants have been built in the country: Malicounda 20 MW in October 2016, Santhiou Mékhé 30 MW in June 2017, Merina Dakhar 30 MW in October 2017, and Kahone 20 MW in July 2018. Touba-Kael and Kahone 60 MW planned for 2019 and Diass 25 MW planned for 2019 [4]. In 2021, the Zagtouli power plant in Burkina Faso, with a maximum production capacity of 33 MW, is the largest solar farm in West Africa. It thus completes the long list of power plants installation in Africa.

However, despite the important solar resource availability in Africa [5], development of solar photovoltaic power plants presents some major challenges associated with cloud cover [6] and dust [79]. In particular, dust presents many drawbacks for solar energy development in the Sahel region of Western Africa [10] due to the presence of the Sahara Desert in the north. The accumulation of dust on solar photovoltaic (PV) panels deteriorates the performance of the panels [11]. Continuous dust accumulation can also reduce the service lifespan of solar power plants [12].

In this general context, the number of studies dealing with the effects of dust on solar systems has increased considerably in recent years. For example, in Saudi Arabia, more than 5% degradation in photovoltaic solar panel production was observed just after 1 month of outdoor exposure [13]. Tests of amorphous silicon panels were conducted in Nigeria under severe weather conditions during the dust season in December, January, and February 2007 and 2008 [14]. The authors found that for 70 days without cleaning, solar absorption decreased by 20%. Tests carried out on several polycrystalline panels in Libya between February and May 2011 showed that the effect of dust could contribute to a 50% reduction in the panel’s output power [15]. Jiang et al. [16] showed that when the deposited dust density increases from 0 to 22 g/m², the reduction in output efficiency increased from 0 to 26%. In the work of Kalogirou et al. [17], a 14, 15, and 1% decrease in output power was observed for monocrystalline, polycrystalline, and amorphous silicon, respectively. According to Hammad et al. [18], the average dust efficiency reductions are 0.768 and 0.607%/day with the multi-variable linear regression models and the artificial neural network model, respectively. A study of the drop in efficiency of a photovoltaic solar panel due to dust was carried out in Baghdad, Iraq [19], in a controlled experimental set-up. It revealed a loss of efficacy of 6.24, 11.8, and 18.74% for 1-day, 1-week and 1-month exposures, respectively. Adinoyi and Said [20] observed a power reduction of up to 50% when PV systems were left uncleaned for more than 6 months, while a single dust storm can reduce power by 20%. In another study conducted in the eastern province of Saudi Arabia [21], it was found that the overall reduction in transmittance under outdoor conditions was about 20% for dust deposits of 5 g/m2 after 45 days of exposure. These various studies show the importance of frequent cleaning of the surface of solar panels in order to reduce losses due to dust accumulation on solar panels.

Several studies proposed frequencies for cleaning the surface of solar panels. Martinez-Plaza et al. [22] showed that weekly cleaning is more than sufficient to maintain constant yield for panels in the Qatar region. In Jordan, Hammad et al. [18] estimated the optimal cleaning frequency between 12 and 15 days. Chiteka et al. [23] presented a new approach to optimize cleaning frequencies in Muzarabani, Zimbabwe. The authors showed that cleaning has been necessary every 2 weeks to minimize losses both due to frequent cleaning and losses caused by not cleaning the panels. Abu-Naser [24] studied the frequency of cleaning solar panels for maximum financial benefit. They proposed a formula for the optimal number of days between cleaning cycles of a solar panel by minimizing the cost of cleaning the panel and the loss of revenue due to dirty panels. The results show that it takes 22 days for the cost due to the lack of power generation to reach $250. In ref. [25], it was found that the cost of cleaning photovoltaic solar panels installed in Perth, Western Australia, would be much higher than the losses caused by dust and, therefore, cleaning is not justified. Thus, the system operator can rely on natural cleanings, such as rain and wind. In Europe, it was found [26] in three places (Murcia in Spain, Munich in Germany, and Stockholm in Sweden), that the clean-up is justified in Murcia and to some extent in Munich, but not justified for Stockholm assuming a cleaning cost of 2,500€.

In West Africa, the impact of the dust accumulation on the solar photovoltaic panels surface has been identified by several studies [9,27,28]. However, the development of dust mitigate strategies is poorly developed. Because this area is under the influence of dust throughout the year [29,30], it is necessary to limit their adverse effect on solar energy production.

The objective of this study is thus to determine the cleaning frequencies according to the different seasons in the Sahel region of West Africa, specifically in Senegal. In a region where dust episodes are frequent, the determination of dust cleaning frequencies helps to limit PV production losses due to dust accumulation and to reduce maintenance costs. Section 2 describes the method used with a description of the atmospheric particles deposition rate, the cleaning frequency model used, and the input parameters. The experiment deployed to study the impact of cleaning frequencies on the performance of the panels is also presented. Results are presented in Section 3 and the conclusion, in Section 4, ends this manuscript.

2 Methodology

Airborne dust deposit on the surface of photovoltaic panels causes a progressive degradation of the performance of solar systems. This article aims at identifying optimal cleaning frequencies of panels in order to limit performance losses due to dust deposits. The model used was developed by Jiang et al. [31] and is based on the deposition rate of airborne particles, the dust density accumulated on the surface of solar panels, and on the concentration of atmospheric dust particles.

Sunshine, also called insolation, is a measure of the amount of solar radiation received by a surface over a period of time. For the study of the solar potential, we used the data measured at the Dakar site by a pyranometer SMP10 of the company Kipp & Zonen over the experimental period every 10 s.

2.1 The study area

The study area is the Diass solar plant, located near the city of Dakar, Senegal, West Africa (Figure 1a). The dust at this site is characterized by the presence of laterite, which is different from the type of dust encountered in urban areas. The choice of the Diass site is justified by its proximity to the Dakar site: it is the closest large-scale plant (about 42 km to Dakar). This power plant has a current capacity of 15 MW and an extension of 7 MW is already planned. It has 55,584 polycrystalline solar panels of 270 W each. These panels are divided into 2,340 strings of 24 modules each. A tractor equipped with a cleaning brush is used to clean the plant. During the rainy season, the cleaning is done naturally by the rain and during the dry season, the installation is cleaned once every 4 weeks.

Figure 1 
                  (a) Map of West Africa and the location of the experimental site as well as the solar power plant in Diass, the black square box shows Senegal. Within the Senegal box, the annual cycle of the (b) mean dust optical depth at 10 μm and (c) mean dust layer altitude for the year 2019. The envelop gives the spread of the daily values.
Figure 1

(a) Map of West Africa and the location of the experimental site as well as the solar power plant in Diass, the black square box shows Senegal. Within the Senegal box, the annual cycle of the (b) mean dust optical depth at 10 μm and (c) mean dust layer altitude for the year 2019. The envelop gives the spread of the daily values.

In West Africa, seasonality of rainfall is particularly marked and is determined by the large and apparent seasons (dry season and wet season), which are themselves composed of sub-seasons.

The dry season is characterized by scant rainfall. In general, it starts from the beginning of November to the end of April; its length varies across different areas in West Africa [32]. It is much longer when one progresses from low to mid-latitudes with sometimes a delay of up to 1 month or even more [33]. In the Sahel, this season covers the winter period (i.e., December–January–February [DJF]) and spring period (i.e., March–April–May [MAM]) [34].

The wet season or rainy season is characterized by the occurrence of rainfall events in West Africa. This is a period during which the incoming solar irradiance decreases due to an intensification of cloud cover [35]. This season is also composed of two sub-seasons, summer (i.e., June–July–August [JJA]) and autumn (i.e., September–October–November [SON]) [33].

This study makes use of the dust data, obtained through the AERIS data infrastructure portal.[1] These daily data at 12 km resolution are retrieved from the Infrared Atmospheric Sounder Interferometer measurements (on board Metop-A satellite).

Figure 1 presents the evolutions of (Figure 1b) the dust optical depth at 10 μm and (Figure 1c) the altitude of the dust layer, during 2019. Figure 1b, the mean dust optical depth (solid curve) is highest in June and July. This period marks the beginning of the rainy season in parts of the West African Sahel region. This dust, maximum in June/July over Senegal, can be linked to mineral dust aerosols transported from their sources in the Sahara region and other areas in the Sahel region of West Africa [36,37]. Although the dust maximum occurs at the beginning of the rainy season, the mean dust layer altitude is the highest during this period. The implies that their residence time in the atmosphere is higher (i.e., dust in the atmosphere takes a longer time to settle on PV panels). Since it is the rainy season, large fractions of these dust particles are washed from the atmosphere by rain before they can settle on the solar panels. On the other hand, mean dust layer altitude is low in the dry season and the particles settle a lot faster. In addition, the lack of rain in the dry season helps the dust particles to settle on the panels without being washed out of the atmosphere.

Figure 2 shows the average solar potential in the Dakar region during the days of the cleaning frequency tests. The specific red bars remind the cleaning dates as indicated in Table 1. During the experiment period, maximum sunshine is noted for the 63rd day corresponding to 1 February 2019 and minimum sunshine is noted for the 15th and 70th day corresponding to 15 December and 8 February 2019.

Figure 2 
                  Solar potential during the days of the cleaning frequency tests. The red bars refer to the cleaning dates.
Figure 2

Solar potential during the days of the cleaning frequency tests. The red bars refer to the cleaning dates.

Table 1

Cleaning frequency

Cleaning days Ipv_1 Ipv_2 Ipv_3
26 December 2018 Cleaned Cleaned Not cleaned
02 January 2019 Cleaned Not cleaned Cleaned
09 January 2019 Cleaned Cleaned Not cleaned
23 January 2019 Cleaned Cleaned Cleaned
06 February 2019 Cleaned Cleaned Not cleaned
13 February 2019 Cleaned Not cleaned Cleaned

Bold values are just to emphasize the days without cleaning action.

2.2 Seasonal cleaning frequency model

One of the main concerns of the photovoltaic industry is the efficient cleaning of the surface of solar panels. Generally, the dusty surface of solar panels is cleaned manually using water or taking advantage of the “natural method” with rainfall events. However, new cleaning technologies are increasingly emerging, such as waterless cleaning technologies [28] and robotic cleaning [38]. Other anti-dirt technologies are also developed such as electrodynamic dust shield technology, to repel dust by electrostatic force, or passive anti-soiling coatings with optical transparency, self-cleaning, and anti-reflective properties [39]. In the absence of proper planning, all of these cleaning methods may not be effective. It is important to determine fixed cleaning periods according to the seasons and also according to the exposure area of the solar panels.

This study provides cleaning frequencies in the region of interest by applying the model developed by Jiang et al. [31]. This model is obtained by using the particle deposition rate obtained in equation (1).

Thus the cleaning time, T (in second) is given by

(1) T ( s ) = M d × A ÷ ( A × C d × V d ) = M d C d × V d ,

where M d is the particle accumulation density for a specific power loss (g/m²), A is the surface area of the solar panels (m²), V d is the particle deposition rate (m/s), and C d is the mass concentration of particles in ambient air (µg/m3).

According to Jiang et al. [31], a 5% loss of power performance is an indication to clean the surface of solar panels. Based on this threshold and using regression model in Figure 10 of Ndiaye et al. [40] in the same region, we estimate that a dust deposition density of about M d = 0.41 g/m² led to 5% loss of power performance.

2.3 Deposition rate model

The deposition rate (V d in m/s) is the ratio between the particle flow and the atmospheric concentration of the aerosol in the vicinity of the surface. This deposition rate depends on many parameters, including terrain topography, substrate, micro-weather conditions (turbulence), aerosol characteristics (grain size), or external fields (gravity) [41]. Depending on their diameter, particles will be subjected to three main families of phenomena to which elementary mechanisms of aerosol transfer and deposition are linked. They are sedimentation, Brownian diffusion, and impact processes. For soiling study, the largest particles are the main concerns [42]. The model developed by Zhao and Wu [43] is based on three layers as proposed by You et al. [44] and allows to estimate the deposition rate of spherical particles on inclined surfaces. In this study, the airborne dust was assumed to be a spherical particle. Zhao and Wu [43] proposed an enhancement on the three-layer particle deposition model by considering four particle transport mechanisms: Brownian diffusion, turbulent diffusion, gravitational sedimentation, and turbophoresis. According to Lai and Nazaroff [45], this Eulerian model hypothesizes that there is a very fine particle concentration boundary layer in the turbulent boundary layer where the particle flow, J, is constant. The particle flow is described by a modified form of Fick’s law:

(2) J = ( ε p + D ) C d Y i ϑ S C d + V t C d ,

where ε p is the diffusivity of the particles within the near-surface boundary, D is the Brownian diffusivity of the particles, ϑ S is the sedimentation velocity due to gravity, i is the parameter for the orientation of the surface, i.e., for a horizontal surface oriented upward (ground) i =1, for a horizontal surface oriented downward (ceiling) i = −1, for a vertical surface i = 0, V t is the turbophoretic velocity, C d is the particle concentration, and Y is the distance between the particles and the surface. This model can accurately estimate the rate of particle deposition on the surface of a plate.

Depending on the particle diameter, the deposition rate model can be divided into four parts, and they are called “fine zone,” “coarse zone,” “zero zone,” and “transition zone.” The equation for each zone consists of two parts: the equation itself and the corresponding applied range. In the “fine zone,” where the particles have small diameters, the inclined angles have no effect on the deposition rate of the particles. The “coarse zone” is for large diameter particles and for surfaces with an inclination angle of less than 90°. In this area, the deposition rates of the particles are proportional to the cosine function of the angle of inclination of the surface. The “zero zone” is reserved for surfaces with an inclination angle greater than 90° and is therefore not used in this study, as the angle of the installed PV panels is less than 90°. The “transition zone” concerns the remaining data.

Thus, the particle deposition velocity, V d (m/s), is estimated by the following equation [44]:

(3) V d = ( 5.15 × 10 8   U 5.63 11 ) d p 1.263 , d p < 0.0512 (   U ) 0.4227 ,   3.7   × 10 5   d p 1.9143 ( cos θ ) , d p > 0.3577 ( cos θ ) 0.41 , cos θ > 0 , 0 , d p > g ( U , cos θ ) ,   cos θ 0 , f ( U , cos θ ,   d p ) , for the others,

where U * is the wind friction speed (m/s), d p is the particle diameter (µm), and θ is the angle of inclination of the panels (°).

The size of the dust particles, d p , is specific to the area and thus varies according to the region. Gac et al. [46] carried out a particle size analysis on samples taken in Dakar, and showed that atmospheric dusts have an average particle size of 10–15 µm; 6.5% of particles have a diameter of less than 2 µm, 91% have a diameter between 2 and 50 µm, and only 2.5% correspond to sand. Drame et al. [29] showed the seasonal cycle of size distribution based on the AERONET station records. The size distribution is a classification of the number of particles by size. Over Dakar, the maximum size distribution is recorded during MAM while the minimum dust distribution is recorded in DJF, with an average radius of about 2 µm.

Concerning the concentration of dust particles in the atmosphere, C d (equation (1)), only PM10 will be considered since Sow et al. [47] showed that, in Dakar, fine particles (PM10 and PM2.5) are the most important pollutants observed in the region and that their rates far exceed the annual thresholds set by the WHO and the national standard (5–6 µg/m3). The concentrations of PM10 and PM2.5 vary, respectively, from 120 to 180 µg/m3 and from 25 to 48 µg/m3 confirming the strong presence of PM10 particles in the atmosphere. Figure 3 shows the average concentration of PM10 particles in the Dakar region from the Monitoring Unit of Air Quality in 2018.

Figure 3 
                  Average PM10 concentrations in Dakar in 2018.
Figure 3

Average PM10 concentrations in Dakar in 2018.

As mentioned in equation (3), the rate of dust particle deposition also depends on the angle of inclination of the solar panels. This angle depends strongly on the installation site and the position of the sun, so it is theoretically equal to the latitude of the site [48,49]. The technology of solar trackers, which is developing more and more, makes it possible to adjust the angle of inclination over time. However, this technology has limitations in the case of large solar power plants. In Senegal (17°10 and 17°32 west longitude and 14°53 and 14°35 north latitude), most solar installations are inclined at an angle of 15°. Only one out of nine PV solar plants (Sakkal power plant) currently have a tracking system.

In this study, we considered an 15° inclination of the PV panels which optimizes the annual production in Dakar.

As a short synthesis, the different steps to determine the dust cleaning frequencies are:

  1. Obtain the inclination θ by considering the angle used to incline the panels to the Senegal (15°) and the size of the dust particles, d p , is determined using data from the literature.

  2. Compute the particle deposition velocity, V d (m/s), equation (3).

  3. Obtain the concentration of particles C d using data from the Monitoring Unit of Air Quality and the dust deposition density M d based on previous study by our research team on site.

  4. Calculate the seasonal cleaning frequency model, T (s), equation (1)

2.4 Cost–benefit model

The estimation of the cost associated with the cleaning of the entire plant was discussed with the site manager and all stakeholders involved. The expenses related to the purchase of fuel and water are included as well as the maintenance of the machine and the payment of all related service providers. The total cost of cleaning in a sub-season is the number of cleaning operations multiplied by the estimated cost of a cleaning.

The cost of energy lost by the Diass solar plant for each sub-season is also calculated assuming that the plant did not undergo any cleaning actions during that period. To estimate the energy lost by the panels over a period of time, the daily energy production of each module is calculated. The difference between the energy produced by a clean module and a dirty module is used to estimate the rate of energy lost by a dusty module relative to a clean module.

The energy produced by a module in Wh, E module, is calculated using the following equation:

(4) E module = t 0 tf U ( t ) * i ( t ) dt ,

where U(t) refers to the voltage delivered by the panel in V, i(t) is the current of the module in A, and t represents the measurement instants.

2.5 Experimental design

The objective of the experiments is to be able to validate the results obtained by using the cleaning frequency model. This validation experiment is performed during the DJF sub-season.

The experiment lasted from 12 December2018 to 14 February 2019. It consisted of recording the short-circuit current of a set of photovoltaic panels implemented at the International Center for Training and Research in Solar Energy (CIFRES; Figure 4), differentiated by their cleaning frequency.

Figure 4 
                  Experimental platform for testing cleaning frequency at the CIFRES; photo taken by Aidara on 15 March 2019 at 19 h.
Figure 4

Experimental platform for testing cleaning frequency at the CIFRES; photo taken by Aidara on 15 March 2019 at 19 h.

At the lower row of the array, three panels (panel 1, panel 2, and panel 3) are arbitrarily chosen to perform cleaning frequencies. The purpose of this experiment is to determine the frequency of cleaning the surface of the panels for the environmental conditions typical of this region. Thus, panel 1 is cleaned every day and serves as a reference panel, panel 2 is cleaned every 2 weeks, while panel 3 is cleaned every 3 weeks. Table 1 shows the periodicity of cleaning applied at the platform level.

Since the systems to be compared are identical and located in the same place and therefore subject to the same conditions, it is quite simple to make the comparison simply by using the maximum power. The power degradation rate, P , is calculated for panels 2 and 3 according to the reference panel (panel 1). It is thus calculated using the following equation:

(5) P ( % ) = P ref P i P ref × 100 ,

where P ref is the power of the reference panel (panel 1) and P i is the power of panels 2 and 3.

3 Results and discussion

3.1 Cleaning frequency of solar PV panels

Figure 5 presents the different seasonal frequencies for 1 year found after model simulation. The cleaning criteria chosen in this case is a loss of 5% of the maximum power of the panels. The JJA sub-season has the longest cleaning time of up to every 6 weeks, mainly explained by the increased rainfall events during this period. The SON sub-season follows with a frequency of every 4 weeks. This sub-season records fewer rainfall events; however, the concentration of airborne particles is not very high (Figure 3). Therefore, solar panels do not require much cleaning during the SON sub-seasons.

Figure 5 
                  Cleaning frequencies for 1 year.
Figure 5

Cleaning frequencies for 1 year.

On the other hand, DJF and MAM have the shortest cleaning times of one every 3 weeks. Indeed, during these periods the absence of rain [50,51] and high particle concentrations (Figure 3) contribute to a significant accumulation of dust on the surface of the panels. The study of Younis and Onsa [52] on cleaning operations and their effects on photovoltaic performance in Africa and the Middle East support our results. Indeed, their study conducted in Zimbabwe (Southern Africa) concluded that the minimum interval between each cleaning operation is 15 days. Although the conditions are different from our study area, the frequencies found are close. Still in Younis and Onsa [52], another study carried out in West Africa (Senegal, Burkina Faso) approved a weekly cleaning program for crystalline silicon modules and once every 3 weeks for amorphous silicon modules, which guarantees an annual energy recovery of 5%. This accumulated dust must be removed from the surface of the panels regularly to allow them to function properly, hence these short cleaning frequencies noted. These cleaning frequencies are first reported in the Sahel region and can help in the proper maintenance of solar installations.

3.2 Impact of cleaning frequencies on panel performance

Figure 6 shows the evolution of the short-circuit current I sc of the three panels (Ipv_1, Ipv_2, and Ipv_3). The dots represent the cleaning days for both modules (orange for Ipv_2 and black for Ipv_3). In most cases, after cleaning the current increases; however, there are periods when after cleaning the current drops considerably due to low irradiation. In these 3 months experimentation, the short-circuit current of panel 1, reference panel cleaned every day, is practically constant except for some fluctuation certainly due to the variation in sunlight from one day to the other. Similarly, it is always superior to the other two currents because of its dust-free surface. For the first day (12 December 2018), all the panels are cleaned and almost the same values of the short-circuit current are noted for the three modules. For 26 December 2018, only module 3 is not cleaned, so that after this day, the short-circuit currents of the other two modules register greater increases. However, it should be noted that the increase in I sc can also be due to strong irradiance (about 500 W/m²; Figure 2). On 09 January 2019 only module 3 is not cleaned and there is a general decrease of the I sc due to the low irradiance (about 300 W/m²; Figure 2). However, the largest decrease is observed for module 3 which is not cleaned. For 23 January 2019, all three panels are cleaned and an increase of their I sc is observed after this day. They all have practically the same I sc.

Figure 6 
                  Evolution of the short-circuit current of the three panels. The dots represent the cleaning days, orange for Ipv_2 and black for Ipv_3.
Figure 6

Evolution of the short-circuit current of the three panels. The dots represent the cleaning days, orange for Ipv_2 and black for Ipv_3.

Over a certain period of time, the short-circuit current of panel 2 is practically the same as that of panel 3, certainly due to their level of dust accumulation. The general observation is that just after a cleaning (modules 2 and 3), the current of these modules increases to reach its initial value.

In Figure 7, it can be seen that each time a module is cleaned, its power degradation rate decreases considerably. The relatively high rate noted at the beginning of the experiment can be explained by losses due to other parameters and long exposure before the start of the experimentation. After 2 weeks without cleaning, we notice a power loss of approximately 12%. For cleaning every 3 weeks, it causes a loss of about 15%. In comparison to these two cleaning frequencies, it is clear that it is better to clean modules every 2 weeks instead of 3 weeks because it generates less power loss.

Figure 7 
                  Influence of cleaning on the power degradation of panel performance. The dots represent the cleaning days, blue for Ipv_2 and orange for Ipv_3.
Figure 7

Influence of cleaning on the power degradation of panel performance. The dots represent the cleaning days, blue for Ipv_2 and orange for Ipv_3.

During the DJF sub-season, the model estimates a cleaning frequency of 3 weeks while the cleaning frequency found experimentally is 2 weeks. Also, during this period, the model power losses are less than 5%, while the experimental cleaning gives power losses of 12%. This difference can be explained by the variation of the concentration of dust particles and climatic parameters (wind speed, relative humidity, and ambient temperature) that were not taken into account in the model. This can also be explained by the fact that the panels used may experience some loss due to aging over time (installed since 2013).

3.3 Cost–benefit analysis of an optimal cleaning frequency

The total cost of cleaning the entire Diass plant is estimated at €400 according to the manager and stakeholder interviews. If the plant is to be cleaned according to the frequency found in this study, the cleaning operations from December to February will be performed four times (every 3 weeks according to the model). The total cost of these cleanings during this period of dust appearance will be €1,615.

If no cleaning operations are performed, the energy lost by the Diass plant during the DJF period is estimated at 1,122,240 kW h. Knowing that the selling cost is €0.10/kW h, the total cost of the lost energy will be €112,224.

The cost of the energy lost by the Diass plant if no cleaning action is performed during the DJF season is much higher than the cost of cleaning during this same period. It is therefore justified to find the optimal cleaning frequency to reduce the effect of dust on the panels and to save water, especially in dry regions where it is a precious resource.

4 Conclusion

This article aims at proposing numerical and experimental methodologies to contribute to the development of a weather information service for the maintenance of solar plants faced to dust accumulation. The interest of this study is to prevent the maintenance of the solar panels in order to avoid considerable loss of performance due to sprouting as the Sahel area is under the frequent influence of dust episodes. In large solar power plants in the Sahel, there is a real problem with cleaning the surface of the solar photovoltaic panels. One of the reasons for this is the difficulty of cleaning and, above all, the high cost of water. The cleaning of large power plants requires the regular use of a large amount of water. This is why it is important to minimize the cleaning periods in order to use as little water as possible and at the same time to keep the efficiency of the panels.

Optimal seasonal cleaning frequencies are estimated from a case study on the Diass solar plant in Dakar, Senegal.

The modeling approach reveals that with a seasonal breakdown of the area, the frequencies vary with the seasons. Thus, due to the natural cleaning associated with rainy events, the cleaning time is longer for the JJA season followed by the SON season. During the other two dryer seasons, DJF and MAM the cleaning frequencies are higher due to dust load in the atmosphere and the scarcity of the rainfall events. The model identifies a 3- week frequency for both seasons. This is confirmed by the experimental approach aiming to assess the effect of several cleaning frequencies on the power loss.

This work contributes to the preventive maintenance of solar PV installations by providing efficient cleaning frequencies. However, these values are only valid in the zone considered but the model can be used in other locations using the specific input parameters of the region. Moreover, the choice of the Diass site is justified by its proximity to the Dakar site but the two sites may have some different characteristics such as the type of dust deposited on the surface of the panels. This may lead to differences in the seasonal cleaning frequency model between the two sites.

In the future, it is crucial to improve the model in integrating other influential climatic parameters such as wind speed, relative humidity, and ambient temperature. There is also a need to continue the experiments with the aim to provide measurements of dust deposition during time as well as measurements of dust accumulation on the panels versus time to better understand the evolution of power loss over time and assess the optimal cleaning frequency with seasons. In a context of climate change, it is essential to support the development of the photovoltaic industry in implementing research to better understand the barriers such as the presence of dust and to develop weather services to support panel cleaning operations.

  1. Funding information: The research leading to this publication is co-funded by International Center for Training and Research in Solar Energy (CIFRES; University Cheikh Anta Diop, Dakar, Senegal) and by Institut de Recherche pour le Développement, France (IRD), grant number UMR IGE Imputation 252RA5. In the frame of this study, the lead author (Mohamed Cherif Aidara) was hosted as visiting scientist during 3 months at the International Joint Laboratory LMI NEXUS in Côte d’Ivoire at the Centre National de Calcul de Côte d’Ivoire (CNCCI, National High Performance Computing Centre of Côte d’Ivoire).

  2. Author contributions: Mohamed Cherif Aidara, Pape Abdoulaye Fam, and Arona Diedhiou performed the design experiment and the conceptualization of the article. Mohamed Cherif Aidara, Derrick Kwadwo Dans, and Eric Mensah Mortey processed the data. All the authors analyzed and discussed the results and contributed to the drafting of the manuscript and the revised version.

  3. Conflict of interest: The authors declare no conflict of interest.


[1] Jäger-Waldau A. Snapshot of photovoltaics—February 2020. Energies. 2020;13(4):930. 10.3390/en13040930.Search in Google Scholar

[2] Monforti F, Belward A, Bisselink B, Bódis K, Brink A, Dallemand JF, et al. Renewable energies in Africa: Current knowledge. European Commission, Joint Research Centre (JRC), Ispra; 2011.Search in Google Scholar

[3] Feukeng L. Solar Show Africa, African operators in solar energy to meet in Johannesburg, Afrik21. 2018.Search in Google Scholar

[4] Aidara MC, Ndiaye ML, Nkounga WM. Correlation between dirt on the photovoltaic module surface and climatic parameters in the Dakar region, Senegal. Seventh International Conference on Renewable Energy Resources Applications; 2018. p. 517–21.10.1109/ICRERA.2018.8566807Search in Google Scholar

[5] Plain N, Hingray B, Mathy S. Accounting for low solar resource days to size 100% solar microgrids power systems in Africa. Renew Energy. 2019;131:448–58. 10.1016/j.renene.2018.07.036.Search in Google Scholar

[6] Danso DK, Anquetin S, Diedhiou A, Adamou R. Cloudiness information services for solar energy management in West Africa. Atmosphere. 2020;11:857. 10.3390/atmos11080857.Search in Google Scholar

[7] Dajuma A, Yahaya S, Touré S, Diedhiou A, Adamou R, Konaré A, et al. Sensitivity of solar photovoltaic panel efficiency to weather and dust over West Africa: comparative experimental study between Niamey (Niger) and Abidjan (Côte d’Ivoire). Comput Water Energy Environ Eng. 2016;5:123–47. 10.4236/cweee.2016.54012.Search in Google Scholar

[8] Bonkaney A, Madougou S, Adamou R. Impacts of cloud cover and dust on the performance of photovoltaic module in Niamey. J Renew Energy. 2017;2017:8 pages. 10.1155/2017/9107502.Search in Google Scholar

[9] Bonkaney AL, Madougou S, Adamou R. Impact of climatic parameters on the performance of solar photovoltaic (PV) module in Niamey. Smart Grid Renew Energy. 2017;8(12):379.10.4236/sgre.2017.812025Search in Google Scholar

[10] Diop D, Drame MS, Diallo M, Malec D, Mary D, Guillot P. Modelling of photovoltaic modules optical losses due to Saharan dust deposition in Dakar, Senegal, West Africa. Smart Grid Renew Energy. 2020;11:89–102. 10.4236/sgre.2020.117007.Search in Google Scholar

[11] Karmouch R, Hor HE. Solar cells performance reduction under the effect of dust in Jazan Region. J Fundam Renew Energy Appl. 2017;7(2). 10.4172/2090-4541.1000228.Search in Google Scholar

[12] Touati F, Massoud A, Abu Hamad J, Saeed SA. Effects of environmental and climatic conditions on PV efficiency in Qatar. Renew Energy Power Qual J. 2013;262–7. 10.24084/repqj11.275.Search in Google Scholar

[13] Rehman S, El-Amin I. Performance evaluation of an off-grid photovoltaic system in Saudi Arabia. Energy. 2012;46:451–8. 10.1016/ in Google Scholar

[14] Sanusi YK. The performance of amorphous silicon PV system under Harmattan Dust conditions in a tropical area. Pac J Sci Technol. 2012;13(1):9.Search in Google Scholar

[15] Ali K, Khan SA, Jafri MM. Effect of double layer (SiO2/TiO2) anti-reflective coating on silicon solar cells. Int J Electrochem Sci. 2014;9:7865–74.Search in Google Scholar

[16] Jiang H, Lu L, Sun K. Experimental investigation of the impact of airborne dust deposition on the performance of solar photovoltaic (PV) modules. Atmos Environ. 2011;45:4299–304. 10.1016/j.atmosenv.2011.04.084.Search in Google Scholar

[17] Kalogirou SA, Agathokleous R, Panayiotou G. On-site PV characterization and the effect of soiling on their performance. Energy. 2013;51:439–46. 10.1016/ in Google Scholar

[18] Hammad B, Al–Abed M, Al–Ghandoor A, Al–Sardeah A, Al–Bashir A. Modeling and analysis of dust and temperature effects on photovoltaic systems’ performance and optimal cleaning frequency: Jordan case study. Renew Sustain Energy Rev. 2018;82:2218–34. 10.1016/j.rser.2017.08.070.Search in Google Scholar

[19] Saidan M, Albaali AG, Alasis E, Kaldellis JK. Experimental study on the effect of dust deposition on solar photovoltaic panels in desert environment. Renew Energy. 2016;92:499–505. 10.1016/j.renene.2016.02.031.Search in Google Scholar

[20] Adinoyi MJ, Said SAM. Effect of dust accumulation on the power outputs of solar photovoltaic modules. Renew Energy. 2013;60:633–6. 10.1016/j.renene.2013.06.014.Search in Google Scholar

[21] Said SAM, Walwil HM. Fundamental studies on dust fouling effects on PV module performance. Sol Energy. 2014;107:328–37. 10.1016/j.solener.2014.05.048.Search in Google Scholar

[22] Martinez-Plaza D, Abdallah A, Figgis BW, Mirza T. Performance improvement techniques for photovoltaic systems in Qatar: results of first year of outdoor exposure. Energy Procedia. 2015;77:386–96. 10.1016/j.egypro.2015.07.054.Search in Google Scholar

[23] Chiteka K, Arora R, Sridhara SN, Enweremadu CC. A novel approach to solar PV cleaning frequency optimization for soiling mitigation. Sci Afr. 2020;8:e00459. 10.1016/j.sciaf.2020.e00459.Search in Google Scholar

[24] Abu-Naser M. Solar panels cleaning frequency for maximum financial profit. Open J Energy Effic. 2017;06:80–6. 10.4236/ojee.2017.63006.Search in Google Scholar

[25] Tanesab J, Parlevliet D, Whale J, Urmee T. Dust effect and its economic analysis on PV modules deployed in a temperate climate zone. Energy Procedia. 2016;100:65–8. 10.1016/j.egypro.2016.10.154.Search in Google Scholar

[26] Stridh B. Economical benefit of cleaning of soiling and snow evaluated for PV plants at three locations in Europe. 27th European Photovoltaic Solar Energy Conference and Exhibition; 2012. p. 4027–9.10.1109/PVSC.2012.6317869Search in Google Scholar

[27] Ndiaye A, Kébé CMF, Ndiaye PA, Charki A, Kobi A, Sambou V. Impact of dust on the photovoltaic (PV) modules characteristics after an exposition year in Sahelian environment: the case of Senegal. Int J Phys Sci. 2013;8(21):1166–73. 10.5897/IJPS2013.3921.Search in Google Scholar

[28] Mohamed CA, Mamadou LN, Amy M, Mamadou S, Pape AN, Amadou N. Study of the performance of a system for dry cleaning dust deposited on the surface of solar photovoltaic panels. Int J Phys Sci. 2018;13(2):16–23. 10.5897/IJPS2017.4701.Search in Google Scholar

[29] Drame MS, Camara M, Gaye AT. Intra-seasonal variability of aerosols and their radiative impacts on sahel climate during the period 2000–2010 using AERONET data. Int J Geosci. 2013;4:267–73. 10.4236/ijg.2013.41A024.Search in Google Scholar

[30] Marticorena B, Chatenet B, Rajot JL, Bergametti G, Deroubaix A, Vincent J, et al. Mineral dust over west and central Sahel: seasonal patterns of dry and wet deposition fluxes from a pluriannual sampling (2006–2012). J Geophys Res Atmos. 2017;122:1338–64. 10.1002/2016JD025995.Search in Google Scholar

[31] Jiang Y, Lu L, Lu H. A novel model to estimate the cleaning frequency for dirty solar photovoltaic (PV) modules in desert environment. Sol Energy. 2016;140:236–40. 10.1016/j.solener.2016.11.016.Search in Google Scholar

[32] Knippertz P, Fink AH. Dry-season precipitation in tropical West Africa and its relation to forcing from the extratropics. Mon Weather Rev. 2008;136:3579–96. 10.1175/2008MWR2295.1.Search in Google Scholar

[33] Senghor H. Étude de la variabilité spatio-temporelle et des processus contrôlant la distribution des aérosols désertiques en Afrique de l’Ouest et sur l’Atlantique tropical-est. PhD thesis. Université Cheikh Anta Diop de Dakar (Sénégal); 2017.Search in Google Scholar

[34] Leon JF, Derimian Y, Chiapello I, Tanré D, Podvin T, Chatenet B, et al. Aerosol vertical distribution and optical properties over M’Bour (16.96° W; 14.39° N), Senegal from 2006 to 2008. Atmos Chem Phys. 2009;9249–61.10.5194/acp-9-9249-2009Search in Google Scholar

[35] Danso DK, Anquetin S, Diedhiou A, Lavaysse C, Kobea A, Touré NE. Spatio‐temporal variability of cloud cover types in West Africa with satellite‐based and reanalysis data. Q J R Meteorol Soc. 2019;145:3715–31. 10.1002/qj.3651.Search in Google Scholar

[36] Crouvi O, Schepanski K, Amit R, Gillespie AR, Enzel Y. Multiple dust sources in the Sahara Desert: the importance of sand dunes: Saharan Desert dust sources. Geophys Res Lett. 2012;39. 10.1029/2012GL052145.Search in Google Scholar

[37] Knippertz P, Todd MC. Mineral dust aerosols over the Sahara: meteorological controls on emission and transport and implications for modeling. Rev Geophys. 2012;50:RG1007. 10.1029/2011RG000362.Search in Google Scholar

[38] Parrott B, Carrasco Zanini P, Shehri A, Kotsovos K, Gereige I. Automated, robotic dry-cleaning of solar panels in Thuwal, Saudi Arabia using a silicone rubber brush. Sol Energy. 2018;171:526–33. 10.1016/j.solener.2018.06.104.Search in Google Scholar

[39] Chen EYT, Ma L, Yue Y, Guo B, Liang H. Measurement of dust sweeping force for cleaning solar panels. Sol Energy Mater Sol Cell. 2018;179:247–53. 10.1016/j.solmat.2017.12.009.Search in Google Scholar

[40] Ndiaye A, Kébé CMF, Bilal BO, Charki A, Sambou V, Ndiaye PA Study of the correlation between the dust density accumulated on photovoltaic module’s surface and their performance characteristics degradation. ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2018. In: Kebe CMF, Gueye A, Ndiaye A, editors. InterSol 2017/CNRIA 2017, LNICST. Vol. 204; 2017. p. 31–42. 10.1007/978-3-319-72965-7_3.Search in Google Scholar

[41] Damay P. Détermination expérimentale de la vitesse de dépôt sec des aérosols submicroniques en milieu naturel : Influence de la granulométrie, des paramètres micro-météorologiques et du couvert. PhD Université de Rouen; 2010. p. 217 pages (in French).Search in Google Scholar

[42] Guo B, Javed W, Khan S, Figgis B, Mirza T. Models for prediction of soiling-caused photovoltaic power output degradation based on environmental variables in Doha, Qatar. Presented at the ASME 2016 10th International Conference on Energy Sustainability collocated with the ASME. Charlotte, North Carolina, USA: 2016. p. V001T08A004. 10.1115/ES2016-59390.Search in Google Scholar

[43] Zhao B, Wu J. Modeling particle deposition from fully developed turbulent flow in ventilation duct. Atmos Environ. 2006;40:457–66. 10.1016/j.atmosenv.2005.09.043.Search in Google Scholar

[44] You R, Zhao B, Chen C. Developing an empirical equation for modeling particle deposition velocity onto inclined surfaces in indoor environments. Aerosol Sci Technol. 2012;46:1090–9. 10.1080/02786826.2012.695096.Search in Google Scholar

[45] Lai K, Nazaroff AC. Modeling indoor particle deposition from deposition from turbulent flow onto smooth surfaces. J Aerosol Sci. 2000;31:463–76. 10.1016/S0021-8502(99)00536-4.Search in Google Scholar

[46] Gac JY, Carn M, Diallo MI, Orange D. Le point sur les observations quotidiennes des brumes sèches au Sénégal de 1984 à 1991; 1991.Search in Google Scholar

[47] Sow B, Tchanche B, Fall I, Souaré S, Mbow-Diokhané A. Monitoring of atmospheric pollutant concentrations in the city of Dakar, Senegal. Open J Air Pollut. 2021;10:18–30. 10.4236/ojap.2021.101002.Search in Google Scholar

[48] Abdeen E, Orabi M, Hasaneen ES. Optimum tilt angle for photovoltaic system in desert environment. Sol Energy. 2017;155:267–80. 10.1016/j.solener.2017.06.031.Search in Google Scholar

[49] Lu L, Yang H, Burnett J. Investigation on wind power potential on Hong Kong islands—an analysis of wind power and wind turbine characteristics. Renew Energy. 2002;27:1–12. 10.1016/S0960-1481(01)00164-1.Search in Google Scholar

[50] Nouaceur Z, Murarescu O. Rainfall variability and trend analysis of rainfall in West Africa (Senegal, Mauritania, Burkina Faso). Water. 2020;12:1754. 10.3390/w12061754.Search in Google Scholar

[51] Ahokpossi Y. Analysis of the rainfall variability and change in the Republic of Benin (West Africa). Hydrol Sci J. 2019;63:2097–123. 10.1080/02626667.2018.1554286.Search in Google Scholar

[52] Younis A, Onsa M. A brief summary of cleaning operations and their effect on the photovoltaic performance in Africa and the Middle East. Energy Rep. 2022;8:2334–47. 10.1016/j.egyr.2022.01.155.Search in Google Scholar

Received: 2022-02-09
Revised: 2022-11-23
Accepted: 2022-12-12
Published Online: 2023-03-09

© 2023 the author(s), published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Downloaded on 24.2.2024 from
Scroll to top button