1. Introduction
The installed capacity of photovoltaic (PV) technology has increased rapidly in recent years. In 2021, solar was again the power-generating technology with the highest net installed capacity, with a 56% share of all new installations for the year.
In addition, from the total 940 GW installed at the end of 2021, in the most conservative scenario, it is expected a capacity of at least 1991 GW worldwide by 2026, driven by the lower cost compared with nuclear, fossil, and other renewable sources [
1] (see
Figure 1).
The integration of all this new PV capacity into the grid represents a technical challenge by itself [
2,
3]. In order to reduce the cost of PV technology and increase profits, the solar industry is constantly pursuing techniques to maximize the efficiencies and minimize the losses and the costs associated [
4,
5,
6].
The accumulation of dust, dirt or other contaminants on the photovoltaic panels is known as soiling. Soiling decreases the irradiance that reaches the photovoltaic cells, reducing the generated power. The progressive accumulation of soiling can reach levels of 15–20% in dry locations, causing similar reductions in the energy production and therefore in the income of the power plant. In [
7], a study in Palestine revealed a difference of 13.1% in annual energy production between weekly cleaned panels and uncleaned panels.
Soiling reduced the global solar power production at least 3–4% in 2018, causing revenue losses in the range of 3–5 billion euros, according to a recent analysis [
8].
For these reasons, it is important to monitor and predict soiling profiles and establish an optimal cleaning schedule.
This paper presents the results of the application of a soiling model in large grid-connected PV plants in Spain. In previous work [
9], the idea was presented. In this work, the proposed method is validated by the comparison between the proposed model and soiling sensors in five PV power plants, with a total installed power of 200 MW, over two years.
2. Impact of Cleaning Activity
In most PV plants, especially in dry locations, cleaning activity is one of the biggest operational costs, both in terms of economics and resources. Each time a large grid-connected PV plant is cleaned, it costs between EUR 450 and 900 per MW installed. It is common to use water to perform this activity (
Figure 2). The amount of water depends on the selected method. Between 2.50 to 6.00 m
3 of water are needed to clean an area equivalent to 1 MWp of PV panels.
Table 1 includes the typical resources consumption and cost of conventional cleaning in a large PV power plant, expressed in different units.
Due to the resources consumption, economic impact and time spent in this activity, the O&M solar industry is seeking new and innovative techniques for panel cleaning and soiling mitigation [
10,
11]. New techniques include fully automated robots with dry cleaning (no water consumption). This new fully autonomous cleaning method is only employed in 0.13% of the current global solar capacity [
8], but it is expected to grow in the next years.
3. Methods for Determining Actual Soiling Losses
3.1. Soiling Metrics
One of the most common metrics used to quantify the amount of soiling is the soiling ratio (
SRatio). The standard IEC 61724-1 [
12] defines
SRatio as the ratio of the actual power output of a PV array under given soiling conditions and the power that would be expected if the PV array were clean and free of soiling:
where
Psoil is the actual power output and
Pclean the power output expected in clean conditions (all the other weather conditions must be the same).
The soiling loss, also known as soiling level (
SL) in the IEC standard [
12] is calculated as:
The soiling ratio has a value of 1 in the absence of soiling (0% soiling loss) and decreases while soiling deposits on the PV modules. In this paper, we refer to soiling loss or soiling level as SL, which represents the percentage of losses due to soiling in a certain time period (in this case, one day)
3.2. Methods to Quantify Soiling
In order to monitor soiling in PV power plants, there are currently four different approaches, summarized in
Table 2.
The first method (a) is the most accurate [
27] but requires not only additional equipment, but also to maintain a clean reference cell (generally cleaned manually). Moreover, in large-connected PV power plants, the soiling measured at a point of the solar field (a or b) may not be representative of the soiling in the whole plant.
Approach (c), calculating soiling from PV yield, can be representative for large plants and do not require additional sensors, but it is affected by other problems the plant may have. Recently, several models have been presented for modelling the soiling from environmental conditions (d). A good comparison of the different methods is presented in [
28].
For large PV plants, the best option is a combination of different methods when available, combining direct measurements with specific equipment (a, b), and soiling modelling from environmental conditions (d).
3.3. Methodology Used in This Work
In this work, a new model of soiling from environmental conditions (d) is presented. This soiling model was validated thanks to the results measured in 5 PV power plants using reference cells (a) and soiling from performance calculated on sunny days at midday (c).
In order to have a representative value of soiling for the whole power plant, the output from each inverter was statistically treated. In this way, inverters with other problems affecting their performance did not affect the soiling calculation.
4. Soiling Loss Model
In the absence of cleaning events, the soiling level (SL), measured as soiling losses in percentage, increases over time. Previous studies have considered soiling losses models with linear rates, which would be acceptable for low levels of soiling.
However, after the analysis of measurements for almost two years at five plants with a total capacity above 200 MW, it was observed that if the panel was completely clean, the soiling rate increased faster than if the panel was dirty. There was a saturation level where the panel maintained its soiling ratio.
4.1. Soiling Loss Empirical Model
The empirical model presented here approximates the evolution of soiling losses (SL) as a complementary exponential growth, up to a saturation level, according to the following expression:
where
t is the equivalent time expressed in days, measured from a total cleaning event (called from here on
eqDaysFromCleaning),
is the maximum saturation level of soiling at the site (which can be adjusted in a range from 20 to 30%), and
k is the time constant in days that represents the soiling rate. The time constant
k can be obtained from a regression analysis from historical plant soiling data in dry periods.
The model considers that the soiling level is reduced to zero (or to a certain value) when a heavy rainfall takes place, or an intentional cleaning is performed. Thus, the variable
eqDaysFromCleaning is related to day
d and calculated as:
where:
Factor = 1 in case of no event;
Factor = 0 in case of heavy rain (total cleaning event);
0 < factor < 1 in case of partial cleaning;
factor > 1 in case of a dust event.
We should consider two different kinds of cleaning events, both of which reduce the amount of soiling in the panels:
4.2. Natural Cleaning and Characterization of Rainfall
Depending on the location of the power plant and the local environment conditions (wind, dust, and pollution), the effect of rainfall can be quite different, and its characterization and forecast can be a key aspect regarding the optimization of the cleaning schedule.
Previous studies have found that 5 mm/day of rainfall is enough to clean photovoltaic modules for up to 0.5% of soiling [
30]. However, the analysis of the five PV power plants showed that some locations required significantly more rainfall to clean completely the modules.
Figure 3 shows the measured efficiency of a PV power plant in Spain and the local precipitation for one and a half year. The performance (%) on sunny days at midday is represented in magenta. Blue bars represent daily precipitation (mm/day). Manual cleaning events and dust events (Saharan dust) are also indicated in the figure. The daily precipitation data used in this study were accessed via an API from the AEMET (Spanish National Meteorology Agency) OpenData database [
31].
It can be observed that only heavy rain (above 10–12 mm/day) succeeded in cleaning the panels. This fact was corroborated by the increase in the power plant performance.
It can also be observed that the effect of heavy raining was not the same after a long dry period or after a certain amount of precipitation. Thus, the proposed method for characterizing precipitation took into account different thresholds to determine the cleaning of the panels.
For a partial cleaning (not a heavy rain), and according to our experience in Spain, factor used in Equation (4) was set between 0.2 and 0.8, when the cumulative rainfall over the last two days was between 10 and 5 mm, respectively (with a linear relationship), and when the rainfall was above 12 mm/day, it was considered a total cleaning event (factor = 1).
The rainfall characterization was obtained from the historical precipitation data from the closest weather station. Therefore, it is possible to predict the impact of rain on soiling and cleaning scheduling, even for new power plants or if the instrumentation is not fully available.
4.3. Characterization of Manual Cleaning
In small power plants, manual cleaning can be well characterized, considering that the soiling level (SL) is 0% after the cleaning is performed. This process is similar to a heavy rain, as shown in
Figure 3. However, in large PV plants, the cleaning activity can last for several weeks. Some panels are cleaned first and others later, resulting in a different degree of soiling in each area of the plant. For the estimation of the degree of soiling in the whole plant (for instance for soiling losses calculation), a gradual decrease from day to day should be considered. This can be observed in
Figure 4.
4.4. Dust Events
Dust events are important to consider in locations with sandy winds. For example, in Spain, where winds from the Sahara and dirty rains (depositing mud on the panels) can occur.
In Spain, in March 2022, there was an extraordinary dust event, causing a high dust deposition in almost the entire Iberian Peninsula and the south of Europe. The dust concentration in the same locations went up to 1000 µg/m
3, values extremely high, which had not been observed in the current century [
32].
Figure 5 shows the intrusion in Europe through the aerosol optical depth (550 nm) indicator.
The impact of this kind of dust events on PV panels is very high. However, it is not easy to obtain a good value of soiling from DC performance, as it requires clear sky (Method c in
Table 2). In plants where soiling sensors are installed (Method b in
Table 2), it was found that the result was not always reliable, as it required a manual cleaning of the sensor’s clean cell (not performed every day) and also required a certain level of irradiance.
The dust events impact on the model was adjusted based on the PM10 concentration, detecting these “events” when the PM10 concentration was higher than a threshold (i.e., 50 µg/m
3). According to our experience in Spain,
factor used in Equation (4) could be obtained as a function of the PM10 concentration (PM10):
Other models have been proposed in the literature to convert PM data into soiling losses, as summarized in [
33].
The PM10 concentration data used in this study were accessed via an API from the European Environment Agency Air Quality Download Service [
34].
Figure 6 represents the model and soiling sensor during the first half of 2022. When comparing to the soiling sensors, it can be observed that sometimes the dust events were not even registered by the sensors, or registered with a delay, probably because the sensors had not been cleaned properly.
5. Soiling Model Validation
A validation of the proposed model was performed to corroborate its accuracy. The data of five large size PV power plants were used for this purpose. Moreover, in these power plants, two different soiling sensors were installed, whose data were used as well.
It should be taken into account that the soiling sensor data were not accurate, as stated before, because sensors required manual cleaning by power plant operators (usually performed weekly).
In each power plant, two soiling sensors were used. In
Appendix A (
Figure A1,
Figure A2,
Figure A3,
Figure A4 and
Figure A5), the difference between the sensors and between the sensor and the proposed model is represented. The comparisons were accomplished by plotting the mean absolute error and the quartiles of the values.
Table 3 summarizes the main values of the comparison. An important fact is that the difference between the two sensors at the same power plant was not negligible.
The variance of the data of different sensors installed at the same locations (column Sensor 1 vs. Sensor 2 of
Table 3) also showed the errors in the sensors themselves (due to calibration, manual cleaning, or the need for clear sky conditions). These errors in some cases were higher than the difference between the model and one of the sensors. On average, the mean absolute error between the model and sensors was 0.71%, lower than the difference between the two sensors installed at the same power plant, which was 0.98%.
6. Economic Aspect of Soiling Loss Forecasting
Many studies in different regions and countries have been carried out to quantify the reduction of performance in PV modules due to soiling. Most of them do not consider the effect of rain [
35,
36] or perform an economic analysis [
37]. Only a few include economic aspects, which help to determine the optimum cleaning schedule [
38,
39].
In order to make a decision regarding the manual cleaning of the panels, it is important to know the economic impact of the soiling losses. Afterward, it should be determined when the gain due to the cleaning is economically higher than the cost of cleaning [
40].
Another important issue to consider is that the benefit of a manual cleaning does not only depend on the current soiling ratio, but also on the rainfall forecast (number of days expected until the following rainfall) and on the expected soiling rate.
An economic model for the soiling losses (expressed in EUR or USD) can be obtained from the soiling loss model, presented in the previous section, the electricity production, and the electricity price forecast. This greatly facilitate the decision-making regarding manual cleaning and permits its optimal scheduling.
7. Conclusions
The soiling losses in a PV plant depend on numerous factors related to ambient conditions such as rain, dust, humidity, wind, irradiation, and temperature, among others. Soiling sensors are a great tool to measure soiling, but it has been observed that sometimes the sensor output is not accurate, due to several reasons, such as calibration, manual cleaning, or the need for clear sky conditions. Moreover, the measurement of the soiling at one location of a large power plant is not representative.
This paper presented an empirical soiling model based on environmental conditions. The proposed model took into consideration the natural cleaning produced by rain with different impacts on the power plant efficiency depending on the daily rainfall. Moreover, the dust events were also modelled, as well as soiling loss forecasting.
The proposed soiling model based on environmental conditions has several advantages; one of them it is the possibility to make predictions and to optimize the cleaning schedule, reducing the total cost and increasing the energy production.
The model was validated by the data of large PV power plants for almost two years of operation. These five power plants were located in Spain with a total peak power of 200 MW. The soiling model can be adapted to any PV power plant regardless of its location and rated power.
The soiling model presented in this work was compared at the five locations with the soiling measured by sensors. On average, the mean absolute error between the model and the sensors was 0.71%. This difference was lower than the difference between two sensors installed at the same location, which was 0.98%. It was observed that dust events were not always registered by the sensors, or they were registered with a delay, because they required manual cleaning.
Future work should consider parameters other than rainfall and PM, to reduce the modeling error. As a conclusion, it is recommended to combine different techniques (model from environmental conditions and sensors if available) in order to improve the cleaning schedule decision-making process. Other future work should be the use of artificial intelligent to optimize the cleaning process.
8. Patents
The method presented in this paper has been submitted for a patent (submission number P202230237).
Author Contributions
Conceptualization, M.R.; methodology, M.R. and C.A.P.; software, M.R. and F.R.; validation, V.D. and A.M.; formal analysis, M.R.; investigation, M.R.; resources, A.M.; data curation, M.R., F.R. and V.D.; writing—original draft preparation, M.R.; writing—review and editing, M.R., C.A.P., F.R., V.D. and A.M.; visualization, M.R. and V.D.; supervision, V.D. and A.M.; project administration, M.R.; funding acquisition, A.M. All authors have read and agreed to the published version of the manuscript.
Funding
This research received no external funding.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
Not applicable.
Acknowledgments
The authors would like to thank Enel Green Power and Endesa Generación SA for sharing the data used in this analysis, and to all O&M Solar team, for sharing their knowledge and insightful discussions.
Conflicts of Interest
The authors declare no conflict of interest.
Appendix A
Left plot: soiling (%) model (continuous line), sensor 1 (blue dots), and sensor 2 (red dots).
Right plot: mean absolute errors and the quartiles of the values, comparing the model vs. S1 (sensor 1), vs. S2 (sensor 2), and S1 vs. S2 (sensor 1 vs. sensor 2)
Each sensor was a commercial kit composed of two panels, and the soiling ratio output was obtained by a comparison of the output of one panel with that of the clean reference panel (cleaned manually).
Figure A1.
Soiling data from the power plant at location 1.
Figure A1.
Soiling data from the power plant at location 1.
Figure A2.
Soiling data from the power plant at location 2.
Figure A2.
Soiling data from the power plant at location 2.
Figure A3.
Soiling data from the power plant at location 3.
Figure A3.
Soiling data from the power plant at location 3.
Figure A4.
Soiling data from the power plant at location 4.
Figure A4.
Soiling data from the power plant at location 4.
Figure A5.
Soiling data from the power plant at location 5.
Figure A5.
Soiling data from the power plant at location 5.
References
- Solar Power Europe. Global Market Outlook for Solar Power:2022–2026; Solar Power Europe: Brussels, Belgium, 2022; ISBN 9789464518610. [Google Scholar]
- Wu, Y.K.; Ye, G.T.; Shaaban, M. Analysis of Impact of Integration of Large PV Generation Capacity and Optimization of PV Capacity: Case Studies in Taiwan. IEEE Trans. Ind. Appl. 2016, 52, 4535–4548. [Google Scholar] [CrossRef]
- Mahdavi, S.; Panamtash, H.; Dimitrovski, A.; Zhou, Q. Predictive Coordinated and Cooperative Voltage Control for Systems With High Penetration of PV. IEEE Trans. Ind. Appl. 2021, 57, 2212–2222. [Google Scholar] [CrossRef]
- Kumar, N.; Hussain, I.; Singh, B.; Panigrahi, B.K. Maximum Power Peak Detection of Partially Shaded PV Panel by Using Intelligent Monkey King Evolution Algorithm. IEEE Trans. Ind. Appl. 2017, 53, 5734–5743. [Google Scholar] [CrossRef]
- Goud, J.S.; Kalpana, R.; Singh, B.A. Hybrid Global Maximum Power Point Tracking Technique With Fast Convergence Speed for Partial-Shaded PV Systems. IEEE Trans. Ind. Appl. 2018, 54, 5367–5376. [Google Scholar] [CrossRef]
- Bosman, L.B.; Leon-Salas, W.D.; Hutzel, W.; Soto, E.A. PV System Predictive Maintenance: Challenges, Current Approaches, and Opportunities. Energies 2020, 13, 1398. [Google Scholar] [CrossRef] [Green Version]
- Abdallah, R.; Juaidi, A.; Abdel-Fattah, S.; Qadi, M.; Shadid, M.; Albatayneh, A.; Çamur, H.; García-Cruz, A.; Manzano-Agugliaro, F. The Effects of Soiling and Frequency of Optimal Cleaning of PV Panels in Palestine. Energies 2022, 15, 4232. [Google Scholar] [CrossRef]
- Ilse, K.; Micheli, L.; Figgis, B.W.; Lange, K.; Daßler, D.; Hanifi, H.; Wolfertstetter, F.; Naumann, V.; Hagendorf, C.; Gottschalg, R.; et al. Techno-Economic Assessment of Soiling Losses and Mitigation Strategies for Solar Power Generation. Joule 2019, 3, 2303–2321. [Google Scholar] [CrossRef] [Green Version]
- Redondo, M.; Platero, C.A.; Moset, A.; Rodríguez, F.; Donate, V. Soiling Forecasting in Large Grid-Connected PV Plants and experience in Spain. In Proceedings of the 2021 IEEE International Conference on Environment and Electrical Engineering and 2021 IEEE Industrial and Commercial Power Systems Europe (EEEIC/I&CPS Europe), Bari, Italy, 7–10 September 2021. [Google Scholar] [CrossRef]
- Einhorn, A.; Micheli, L.; Miller, D.C.; Simpson, L.J.; Moutinho, H.R.; To, B.; Lanaghan, C.L.; Muller, M.T.; Toth, S.; John, J.J.; et al. Evaluation of Soiling and Potential Mitigation Approaches on Photovoltaic Glass. IEEE J. Photovolt. 2019, 9, 233–239. [Google Scholar] [CrossRef]
- Alghamdi, A.S.; Bahaj, A.S.; Blunden, L.S.; Wu, Y. Dust removal from solar PV modules by automated cleaning systems. Energies 2019, 12, 2923. [Google Scholar] [CrossRef]
- IEC 61724-1; Photovoltaic System Performance—Part 1: Monitoring. IEC International Standard: Geneva, Switzerland, 2021.
- GroundWork. PVSOIL: PV Soiling Measurement System. Available online: http://www.grndwork.com/wp-content/uploads/2016/11/GroundWork_PVSOIL.pdf (accessed on 16 October 2022).
- Kintech Engineering. Soiling Measurement Kit. Available online: https://www.kintech-engineering.com/pdf_docs/EN_Soiling_Measurement_Kit_Brochure.pdf (accessed on 16 October 2022).
- NGRSystems. Available online: https://www.nrgsystems.com/assets/resources/NRG-SRA-SMK-Insert-WEB.pdf (accessed on 16 October 2022).
- Gostein, M.; Stueve, W.; Faullin, S. Measuring Soiling and Non-Uniform Soiling of PV Modules without a Clean Reference Device. U.S. Patent 2019/0273466A1, 14 July 2020. [Google Scholar]
- Atonometrics. Mars Optical Soiling Sensor. Available online: http://www.atonometrics.com/mars-optical-soiling-sensor-revolutionary-new-product/ (accessed on 16 October 2022).
- KIPP&ZONEN. DustIQ Soiling Monitoring System. Available online: https://www.kippzonen.com/Product/419/DustIQ-Soiling-Monitoring-System (accessed on 16 October 2022).
- Hoffmann, H.; Carretero, A.J. Method for Detecting the Degree of Soiling of PV Modules. U.S. Patent 2015/0346123A1, 28 March 2017. [Google Scholar]
- Deceglie, M.G.; Micheli, L.; Muller, M. Quantifying Soiling Loss Directly From PV Yield. IEEE J. Photovolt. 2018, 8, 547–551. [Google Scholar] [CrossRef]
- Kimber, A.; Mitchell, L.; Nogradi, S.; Wenger, H. The effect of soiling on large grid-connected photovoltaic systems in california and the southwest region of the United States. In Proceedings of the IEEE 4th World Conference on Photovoltaic Energy Conference, Waikoloa, HI, USA, 7–12 May 2006. [Google Scholar] [CrossRef]
- Bergin, M.H.; Ghoroi, C.; Dixit, D.; Schauer, J.J.; Shindell, D.T. Large Reductions in Solar Energy Production Due to Dust and Particulate Air Pollution. Environ. Sci. Technol. Lett. 2017, 4, 339. [Google Scholar] [CrossRef] [Green Version]
- You, S.; Lim, Y.J.; Dai, Y.; Wang, C.-H. On the temporal modelling of solar photovoltaic soiling: Energy and economic impacts in seven cities. Appl. Energy 2018, 228, 1136. [Google Scholar] [CrossRef] [Green Version]
- Coello, M.; Boyle, L. Simple Model for Predicting Time Series Soiling of Photovoltaic Panels. IEEE J. Photovolt. 2019, 9, 1382. [Google Scholar] [CrossRef]
- Toth, S.; Hannigan, M.; Vance, M.; Deceglie, M. Predicting Photovoltaic Soiling From Air Quality Measurements. IEEE J. Photovolt. 2020, 10, 1142. [Google Scholar] [CrossRef]
- Sengupta, S.; Ghosh, A.; Mallick, T.K.; Chanda, C.K.; Saha, H.; Bose, I.; Jana, J.; Sengupta, S. Model Based Generation Prediction of SPV Power Plant Due to Weather Stressed Soiling. Energies 2021, 14, 5305. [Google Scholar] [CrossRef]
- Micheli, L.; Muller, M.T.; Deceglie, M.G.; Ruth, D. Time Series Analysis of Photovoltaic Soiling Station Data: Version 1.0, August 2017. NREL, 2017. Available online: https://www.nrel.gov/docs/fy17osti/69131.pdf (accessed on 16 October 2022).
- Bessa, J.G.; Micheli, L.; Montes-Romero, J.; Almonacid, F.; Fernández, E.F. Estimation of Photovoltaic Soiling Using Environmental Parameters: A Comparative Analysis of Existing Models. Adv. Sustain. Syst. 2022, 6, 2100335. [Google Scholar] [CrossRef]
- Khadka, N.; Bista, A.; Adhikari, B.; Shrestha, A.; Bista, D.; Adhikary, B. Current Practices of Solar Photovoltaic Panel Cleaning System and Future Prospects of Machine Learning Implementation. IEEE Access 2020, 8, 135948–135962. [Google Scholar] [CrossRef]
- Hammond, R.; Srinivasan, D.; Harris, A.; Whitfield, K.; Wohlgemuth, J. Effects of soiling on PV module and radiometer performance. In Proceedings of the Conference Record of the Twenty Sixth IEEE Photovoltaic Specialists Conference, Anaheim, CA, USA, 29 September–3 October 1997. [Google Scholar] [CrossRef]
- AEMET OpenData. Sistema para la Difusión y Reutilización de la Información de AEMET. Available online: https://opendata.aemet.es/ (accessed on 16 October 2022).
- AEMET, Informe Acerca de la Intrusión de Polvo de Origen Sahariano Sobre el Territorio Peninsular Español Entre los días 14 y 16 de Marzo de 2022. Available online: https://www.aemet.es/documentos/es/conocermas/recursos_en_linea/publicaciones_y_estudios/estudios/Polvo_Atmosferico/informe-intrusionPolvo-web-AEMET.pdf (accessed on 16 October 2022).
- Micheli, L.; Smestad, G.P.; Bessa, J.G.; Muller, M.; Fernandez, E.F.; Almonacid, F. Tracking Soiling Losses: Assessment, Uncertainty, and Challenges in Mapping. IEEE J. Photovolt. 2022, 12, 114–118. [Google Scholar] [CrossRef]
- European Enviroment Agency. Download Service for E1a and E2a Data. Available online: https://discomap.eea.europa.eu/map/fme/AirQualityExport.htm (accessed on 16 October 2022).
- Jones, R.K.; Baras, A.; Al Saeeri, A.; Al Qahtani, A.; Al Amoudi, A.O.; Al Shaya, Y.; Alodan, M.; Al-Hsaien, S.A. Optimized Cleaning Cost and Schedule Based on Observed Soiling Conditions for Photovoltaic Plants in Central Saudi Arabia. IEEE J. Photovolt. 2016, 6, 730–738. [Google Scholar] [CrossRef]
- Cristaldi, L.; Faifer, M.; Rossi, M.; Catelani, M.; Ciani, L.; Dovere, E.; Jerace, S. Economical evaluation of PV system losses due to the dust and pollution. In Proceedings of the IEEE International Instrumentation and Measurement Technology Conference Proceedings, Graz, Austria, 13–16 May 2012. [Google Scholar] [CrossRef]
- Micheli, L.; Fernandez, E.F.; Muller, M.; Almonacid, F. Extracting and Generating PV Soiling Profiles for Analysis, Forecasting, and Cleaning Optimization. IEEE J. Photovolt. 2020, 10, 197–205. [Google Scholar] [CrossRef]
- Naeem, M.; TamizhMani, G. Cleaning frequency optimization for soiled photovoltaic modules. In Proceedings of the IEEE 42nd Photovoltaic Specialist Conference (PVSC), New Orleans, LA, USA, 14 June 2015. [Google Scholar] [CrossRef]
- Mithhu, M.M.H.; Rima, T.A.; Khan, M.R. Global analysis of optimal cleaning cycle and profit of soiling affected solar panels. Appl. Energy 2021, 285, 116436. [Google Scholar] [CrossRef]
- Alonso-Montesinos, J.; Martínez, F.R.; Polo, J.; Martín-Chivelet, N.; Batlles, F.J. Economic Effect of Dust Particles on Photovoltaic Plant Production. Energies 2020, 13, 6376. [Google Scholar] [CrossRef]
| Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).