Data-Driven Model for Solar Panel Performance and Dust Accumulation

Abstract

Solar panel deployment is vital to generate clean energy and reduce carbon emissions, but sustaining energy output requires regular monitoring and maintenance. This is particularly critical in countries with harsh environmental conditions, such as Qatar, where high dust density reduces solar radiation reaching panels, thereby lowering generating efficiency and increasing maintenance costs. This paper introduces a data-driven model that uses the relationship between generated and consumed energy to track changes in solar panel performance. By applying statistical analysis to real and simulated data, the model identifies when efficiency losses are within the parameters of normal variation (e.g., daily fluctuations) and when they are likely caused by dust accumulation or system ageing. The findings demonstrate that the model provides a reliable and cost-effective way to support timely cleaning and maintenance decisions. It offers decision-makers a practical tool to improve residential solar panel management, reducing unnecessary costs, and ensuring more consistent renewable energy generation.

1. Introduction

Dust accumulation on photovoltaic (PV) modules is a critical operational challenge in arid and semi-arid regions. Numerous studies have shown that soiling reduces solar transmittance, increases module temperature, and accelerates efficiency losses, thereby raising maintenance costs and undermining renewable energy targets. High dust density, frequent sandstorms, and limited rainfall exacerbate this issue in Qatar and across the Middle East and North Africa, making effective monitoring and decision-making strategies essential [1]. Recent research has advanced understanding of this problem from multiple perspectives. Javed et al. [1] conducted one of the first systematic field characterizations of naturally accumulated dust in Doha. Their ten-month study linked dust accumulation rates, particle size distribution, and mineralogy with PV performance degradation, revealing seasonal variations and particle agglomeration processes.

Moving beyond characterization, Shaaban et al. [2] developed a data-driven “dust estimation unit” trained on field measurements in Dubai. By using irradiance, temperature, and PV output as predictors, they demonstrated how regression models could estimate dust loading and trigger condition-based cleaning. More recently, Shukla et al. [3] provided a controlled laboratory benchmark by applying multiple dust types and graded doses under artificial irradiance. Their comparative analysis of support vector regression (SVR) and artificial neural networks (ANNs) highlighted the superior predictive accuracy of the former for PV power under soiling.

While these efforts have enriched scientific and technical understanding, they remain focused on either detailed dust characterization, or advanced machine learning prediction of dust load and power loss. Both approaches often require specialized sensors, experimental facilities, or computational expertise, limiting their direct adoption in small-scale residential or SME contexts. Hence, this paper addresses that gap by introducing a data-driven model (DDM) based on the generated-to-consumed energy ratio (GtoC). Unlike previous studies [1,2,3], the proposed framework offers a cost-effective, scalable, and easily interpretable decision-support tool that distinguishes normal performance variation from degradation caused by dust or ageing.

2. Data-Driven Decision-Making (DDDM)

Data plays a crucial role in decision-making, providing valuable insights and evidence to support informed choices. DDDM is the process of using objective data to inform business decisions, rather than relying on intuition or personal judgment. This approach has become increasingly important in modern business environments, where vast amounts of data are available from various sources such as social media, customer data, and internal and industrial reports [4,5]. Research has shown that DDDM leads to more accurate and effective decision-making outcomes [6].

By collecting and analyzing relevant data, decision-makers gain a deeper understanding of patterns, trends, and relationships within their domain. This empirical evidence enables them to make better-informed decisions grounded in objective information. It also reduces reliance on subjective opinions and intuitions, which are often influenced by bias. As a result, the risk of erroneous judgments and suboptimal outcomes is minimized [7].

As illustrated in Figure 1, the benefits of DDDM include valuable insights, continual growth, improved program outcomes, optimized operations, prediction of future trends, and actionable insights [8]. One of the key advantages of DDDM is its ability to uncover patterns and trends that may not be immediately apparent through other means. Data analysis enables the identification of correlations between different segments, allowing for tailored strategies for different groups [9]. Additionally, DDDM helps organizations mitigate the risk of errors or bias that can arise when decisions are based solely on subjective judgment. By relying on objective data, businesses can minimize the potential for decision-making errors and ensure that they make the best choices possible with the information at hand [10]. Hence, decision-makers across various fields and industries recognize the significance of relying on data-driven approaches to enhance their decision-making processes [6].

2.1. DDDM in the Renewable Energy Industry

DDDM in the renewable energy industry has been widely recognized for its significant role in improving renewable energy planning, optimization, and management [11]. By utilizing data and advanced analytics, decision-makers can gain valuable insights to support informed choices and drive the transition towards a sustainable and clean energy future. Renewable energy industry stakeholders (among others) have been implementing Data-Driven Decision Support Systems (DDDSS) to enhance decision-making processes, by harnessing large amounts of data and advanced analytical tools. In the context of the renewable energy industry, the following key elements are vital for the effectiveness of DDDSS [4]:

• Data collection: Relevant data must be gathered from various sources specific to the industry, including internal databases, external sources such as renewable energy generation data, environmental factors, market trends, and customer data. Ensuring the quality of collected data is crucial to obtain accurate and reliable insights.
• Data storage: Once collected, industrial data needs to be stored in a manner that is easily accessible and facilitates quick analysis. This may involve storing data in specialized databases or cloud-based storage solutions designed for renewable energy applications.
• Data analysis: Advanced analytical tools and techniques are employed to analyze the data specific to the industry. This includes using machine learning algorithms, statistical analysis, and energy modeling techniques to uncover patterns, trends, and insights relevant to renewable energy generation, optimization, and forecasting.
• Data visualization: Results of data analysis are presented in a visually intuitive manner to facilitate easy understanding and interpretation. This includes interactive dashboards, charts, maps, and other visualizations that help stakeholders grasp the complex relationships and make informed decisions regarding renewable energy projects and investments.

One key aspect where DDDM is valuable in the renewable energy sector is resource assessment and site selection. Through the analysis of historical weather data, topographical information, and other relevant data sources, decision-makers can identify optimal locations for renewable energy projects such as wind farms or solar installations [12]. By considering factors such as wind speed, solar radiation, and land suitability, DDMs can help maximize the energy generation potential and improve the overall efficiency of renewable energy systems.

Furthermore, data-driven approaches can aid in the prediction and forecasting of renewable energy generation. By utilizing real-time monitoring data, weather forecasts, and predictive modeling techniques, decision-makers can estimate the expected output from renewable energy sources with greater accuracy [13]. These forecasts enable effective grid integration, better resource allocation, and improved energy management, ultimately optimizing the utilization of renewable energy resources.

DDDM also plays a critical role in monitoring and maintenance of renewable energy infrastructure [10]. By collecting data from sensors, IoT devices, and remote monitoring systems, decision-makers can continuously monitor the performance and health of renewable energy systems [13,14,15]. This proactive approach allows for early detection of issues, timely maintenance, and improved operational efficiency, ensuring the optimal functioning of renewable energy assets [16].

Therefore, DDDM is instrumental in the renewable energy sector, facilitating resource assessment, forecasting, and monitoring of renewable energy systems. By leveraging data and advanced analytics, decision-makers can optimize the deployment and management of renewable energy resources, contributing to a more sustainable and environmentally friendly energy landscape [12].

2.2. Data Driven Model (DDM) for Solar Panel Systems

Regular monitoring, maintenance, and upgrades are crucial for the long-term sustainability and efficiency of solar PV systems [17]. Advanced monitoring technologies, such as IoT and Intelligent Wireless Sensor Networks, are utilized in large solar panel plants for performance evaluation [18]. However, these technologies are too costly for smaller and localized private applications. Therefore, there is a need for a practical and affordable approach like GtoC for homes and businesses with a limited number of solar panels [19]. Additionally, having enough data is essential for making the right decisions and establishing strategies. For instance, GtoC can be used as an indicator to support decision-making when the government desires to establish national renewable energy policies, to incentivize teams and consumers to install solar panel systems or support solar energy generation.

The implementation of GtoC within a DDDSS can assist various stakeholders in developing effective strategies and making informed decisions [20]. Moreover, GtoC can be used as an indicator to compare the current reading with the previous reading, when the system was at its peak performance, to support planning, cleaning, maintenance, or replacement of parts or whole systems [19]. This is crucial as dust, dirt, and ageing can reduce the performance of PV systems, particularly in Qatar, where high dust density occurs during specific months of the year [21]. For example, if the GtoC changes, it could indicate a decline in panel performance, which can be restored by performing cleaning to restore the normal GtoC. Figure 2 illustrates how dust can impact GtoC data.

3. Methodology

3.1. Methods

As the objective is to introduce a data-driven system as a new mechanism to improve decision-making for solar roof panels, the Plan–Do–Study–Act (PDSA) methodology was adopted to establish the DDM to serve the purpose of this research project. PDSA has been widely used within healthcare and other industries to improve processes and achieve desired outcomes (Figure 3) [22,23]. Adopting PDSA as a methodology for establishing a DDDSS for solar panel roofs can be particularly effective because it allows for continuous improvement of the system over time. The PDSA cycle is a four-step iterative process that involves planning, executing, evaluating, and refining a change [23]:

• In the initial Plan phase, the problem is identified (e.g., target data collection sites in Qatar), and a plan is developed to collect and analyze the data.
• In the Do phase, the plan is implemented, and data is collected (on GtoC).
• The Study phase involves analyzing the data collected in the previous step, to determine if the plan was successful in addressing the identified problem by identifying patterns or trends.
• Finally, in the Act phase, the findings from the Study phase are used to refine and improve the plan, which is then implemented again in the next PDSA cycle.

As can be seen in Figure 4, the process of creating a DDM involves several key stages [24], as attested by previous research [8]. It begins by defining the mission or problem at hand and then proceeds to identify the relevant data sources. Once the data is collected, it undergoes a thorough cleaning and organizing process, to ensure its quality and suitability for analysis [25]. The next stage involves performing statistical analyses on the cleaned and organized data, utilizing appropriate methods and techniques [26,27]. This step aims to extract meaningful insights, identify patterns, and derive conclusions based on the information derived from the data and statistical analysis [27].

3.2. Proposed DDM

Countries like Qatar are prone to sandstorms, which can lead to the accumulation of sand on solar panels and degrade the amount of energy generated [28]. The proposed DDM, as depicted in Figure 5, aims to contribute to the monitoring of changes in the performance of solar panel systems. By comparing the amount of GtoC under ideal working conditions with the amount of GtoC after dust accumulation, statistical analysis can determine if the impact of dust significantly degrades the amount of energy generated. This analysis can be conducted for a single home, a specific region with multiple homes, homes distributed across a city, or the total number of homes in a country. Based on these findings, informed decisions can be made to address the impact of dust accumulation on solar panel performance.

3.3. Identifying Mission/Problem

Solar panels may experience degradation in the amount of generated energy due to aging, lack of maintenance, and other reasons [29]. It is important to monitor the extent of this degradation in order to determine the necessary course of action. With multiple solar panels installed on residential roofs, a reactive approach is often taken, waiting until the system completely stops or experiences a significant drop in energy generation before conducting checks. However, a proactive approach is more advantageous for sustaining clean energy generation [30]. Therefore, the objective of this study is to establish a mechanism that can effectively identify degradation in the amount of energy generated by home solar panels, providing stakeholders with valuable information to make informed decisions.

3.4. Identifying Data Sources

Identifying the right data sources is a critical factor in enabling the proposed approach to function as intended. In the related research conducted, it was found that the GtoC indicator can serve as a simple and useful data source [8]. This indicator can be extracted from a single home, a specific region, or even the entire country, depending on the analyses needs to be made. By utilizing this data, valuable insights into energy generation and consumption patterns, which can inform decision-making processes by different stakeholders and support the development of sustainable energy strategies. As previously established through an analytical study, GtoC data provides a wealth of information that can be used to supply DDMs with the necessary data to make informed decisions. GtoC data can be generated based on the following equations derived from previous research [19].

3.4.1. Estimated Daily Solar Panels Energy Generation Formula

$$G={\displaystyle \frac{\left(p_{s}1000p_n{p}_e{s}_{PD}\right)}{1000}},$$

(1)

where $G$ is the total generated electrical energy from solar panels (kWh), $p_s$ is the solar panel area (m²), $p_n$ is the number of solar panels, $p_e$ is solar efficiency, and $s_{PD}$ is sun hours per day.

3.4.2. Generated to Consumed Electrical Energy Ratio (GtoC) Formula

$$G_{to}C=\left({\displaystyle \frac{G}{C}}\right)\times 100\%,$$

(2)

where $G$ is the total generated electrical energy from solar panels in KWh, and $C$ is the consumed electrical energy by the house in KWh.

3.4.3. Average $G_{to}C$ for Number of Homes in City or District Formula

$${\left(G_{to}C\right)}_A={\displaystyle \frac{\left(\sum _{i=1}^n{\left(G_{to}C\right)}_1+{\left(G_{to}C\right)}_2+\cdots +{\left(G_{to}C\right)}_n\right)}{n}},$$

(3)

where n is the total number of houses in the city or district.

3.4.4. National $G_{to}C$ Formula

$${\left(G_{to}C\right)}_N={\displaystyle \frac{\left(\sum _{i=1}^n{\left(G_{to}C\right)}_{A1}+{\left(G_{to}C\right)}_{A2}+\cdots +{\left(G_{to}C\right)}_{An}\right)}{n}},$$

(4)

where n is the total number of cities or districts in country.

3.5. Cleaning and Organizing Data

Cleaning and organizing the data in DDMs is crucial to ensure accurate and reliable results. It is important to note that only 20% of the data will be used for analysis in order to make informed decisions; additionally, organizing the flow of the data is essential to ensure a meaningful approach that captures relevant information [31]. In this research, the implementation of the tree-topology approach (as shown in Figure 6) is considered the most appropriate, as it enables a better flow and collection of data for subsequent statistical analysis. The tree-topology approach offers several advantages, including improved data handling, efficient data processing, and enhanced data integrity [31]. By utilizing this approach, the collected data can be managed and analyzed, effectively, leading to more accurate and reliable research outcomes.

Figure 6 illustrates the tree topology, which facilitates the collection of GtoC data across three tiers: home, city, and national levels. This hierarchical structure allows for statistical analysis to be conducted at different levels, enabling informed decision-making based on the data obtained from each tier. In order to establish a DDM to support decision-making regarding the impact of dust accumulation on solar panel performance and energy generation, an assumption was made based on a study of 100 homes. The GtoC data was calculated using information from a previous study that analytically assessed the feasibility of deploying solar panels in Qatari homes, representing the ideal operational scenario [19]. Subsequently, the data was degraded at different levels, including −5%, −10%, and −20%, as shown in

Table 1. Utilized GtoC data.

Month	Ideal Situation	Degradation by [−5%]	Degradation by [−10%]	Degradation by [−15%]	Degradation by [−20%]
Jan	39.8	37.8	35.8	33.8	31.9
Feb	42.6	40.5	38.3	36.2	34.1
Mar	44.2	42.0	39.8	37.6	35.4
Apr	47.7	45.3	42.9	40.5	38.2
May	50.6	48.0	45.5	43.0	40.4
Jun	51.2	48.7	46.1	43.5	41.0
Jul	50.9	48.4	45.9	43.3	40.78
Aug	48.2	45.8	43.4	41.0	38.6
Sep	46.5	44.1	41.8	39.5	37.12
Oct	43.2	41.0	38.9	36.7	34.6
Nov	39.9	37.9	35.9	33.9	31.9
Dec	39.0	37.1	35.1	33.2	31.2
Av. GtoC	45.3	43.0	40.8	38.5	36.3

. This degraded data was then utilized to feed the model and perform the subsequent stage of statistical analysis, transforming the data into meaningful information that can be understood by various stakeholders.

4. Results

4.1. Descriptive Analysis

Descriptive analysis allows for the representation of data and provides a general overview of its behavior. This analysis, such as calculating means, is extracted from the data presented in Table 1. The radar plot in Figure 7 illustrates the different degradation scenarios, enabling a comparison of each degradation level with the ideal scenario. Additionally, comparing the GtoC average for each degradation case with the average of the ideal scenario, as depicted in Figure 8, offers another method of assessing the different levels of degradation. However, it is important to note that descriptive analysis alone does not provide a meaningful comparison to determine if the changes are significantly different from the ideal scenario. To contextualize these findings and gain further insights, a comparative analysis should be incorporated. By combining descriptive and comparative analyses, the situational implications of the data can be better understood [32].

4.2. Comparative Analysis

Comparative analysis is a valuable approach for making meaningful comparisons between the degradation situation and the ideal situation. Since the comparison involves two distinct groups, conducting statistical tests becomes essential. One well-known test for such comparisons is the t-test, which examines the means of each group and determines if there is a significant difference. In cases where the level of degradation is high, the statistical test will reveal a significant difference [32].

Table 2. T-test statistical analysis summary.

Degradation	Calculated t-Value	Calculated p Value	Result Significant [Yes/No]
−5%	1.27	0.109	No
−10%	2.60	0.008	Yes
−15%	4.00	<0.001	Yes
−20%	4.17	<0.001	Yes

displays the results of various tests conducted to compare the different levels of degradation with the ideal situation. These tests involve calculating the t-value and p-value to establish the level of significance. The t-value reflects the extent of degradation, with higher values indicating greater degradation. The p-value is crucial for determining whether the observed degradation is statistically significant or not. Table 2 shows that the higher the t-value, the greater the level of degradation, while the p-value indicates the significance of the observed degradation.

4.3. Conclusion for Action

Figure 9 illustrates the relationship between the t-value and degradation provides decision-makers with a visual representation that facilitates the understanding of the statistical findings presented earlier. The graph includes a horizontal line marked at 0%, representing the ideal situation. When there is minimal degradation, as seen in the case of a 5% degradation, the line closely aligns with the horizontal line. However, as the value of t increases with greater degradation, the line deviates towards the vertical axis. By observing these changes, decision-makers can determine the appropriate course of action. For example, when the degradation is at 5%, it can be concluded that the situation falls within the normal range. In cases where the degradation reaches 10%, it is recommended to implement a plan for solar panel cleaning. When the degradation exceeds 15%, it becomes crucial to conduct cleaning to address the significant degradation. The graphical representation in Figure 9 of the t-value against degradation offers decision-makers an intuitive way to interpret the data and make informed decisions based on the observed changes in the line’s position.

4.4. Integrating Panel Efficiency Degradation over Time

The model integration of panels’ efficiency degradation due to age is a critical consideration in comprehensive modeling for solar energy systems. PV panels experience a normative decrease in efficiency over time, which directly undermines the overall energy output of the system. To accurately depict the long-term performance of such systems, it is imperative to incorporate this degradation factor into the modeling process. In this context, it is observed that the panel efficiency diminishes progressively over the years. The established model incorporates a degradation rate of 0.2 for the initial year, followed by a subsequent decrease of 0.7 percent in each subsequent year [33]. This degradation pattern significantly influences the amount of energy generated by the system. By integrating this dynamic efficiency degradation into the model, a more holistic understanding of the system’s behavior emerges, allowing for informed decision-making processes, particularly in matters concerning system upgrades or maintenance [34].

To illustrate the application of this degradation model, Figure 10 depicts a flowchart exemplifying the programming function used for the calculation. Furthermore, Figure 11 offers a 3D visual representation based on the data sourced from Table 1. The provided flowchart serves as a comprehensive guide, detailing the step-by-step process for calculating the efficiency for each year, thereby enhancing the clarity surrounding the evolution of efficiency over time. While the degradation pattern may not strictly adhere to a linear trajectory, Figure 11 encapsulates the fundamental concept of efficiency reduction in panels for each successive year. Additionally, it affords valuable insights into the efficiency fluctuations throughout each month, extrapolated from the initial energy generation data established during the system’s setup phase. It is noteworthy that this dataset signifies the system’s peak performance during its initial installation [35].

5. Discussion

The continuous improvement of decision-making processes is essential for sustaining solar panel performance and ensuring reliable clean energy generation. Without such approaches, energy losses can accumulate, maintenance costs increase, and national sustainability targets risk being undermined. The DDM proposed in this study demonstrates its effectiveness in supporting proactive monitoring of solar PV systems, particularly in dust-prone regions such as Qatar. By leveraging GtoC data, the model enables stakeholders to distinguish between normal performance fluctuations and degradation caused by dust accumulation or system ageing.

The integration of descriptive and comparative analyses, including t-tests, provides decision-makers with clear statistical evidence to guide timely interventions. For example, while minor deviations may be within acceptable variation, more significant performance drops indicate the need for cleaning or corrective maintenance. This structured approach moves beyond reactive practices and supports condition-based strategies that are both efficient and cost-effective.

Another strength of the model lies in its ability to incorporate long-term efficiency degradation due to panel ageing. By accounting for the gradual decline in PV output over time, the model offers a more comprehensive view of system performance and supports forward planning for upgrades or replacements. The visual outputs further enhance interpretability, allowing stakeholders to grasp performance trends and degradation patterns with greater clarity.

Importantly, the proposed approach remains accessible and scalable. Unlike advanced machine learning techniques or sensor-intensive systems, the DDM relies on routinely available energy data, making it applicable across residential, community, and national contexts. It can also be expanded to provide daily analyses, enabling faster responses to sudden efficiency drops and strengthening overall system resilience.

In summary, this study highlights how a cost-effective, data-driven framework can equip decision-makers with practical insights for sustaining solar panel efficiency. By addressing dust accumulation and ageing in an integrated manner, the DDM contributes to more informed, timely, and sustainable energy management strategies.

6. Conclusions

This paper introduces a DDM for decision-making within the context of solar panel systems. The model leverages GtoC data to perform statistical analysis and translate it into actionable information for stakeholders. The proposed DDM offers a cost-effective approach with clear and understandable statistical methodologies, making it accessible to a wide range of stakeholders.

The DDM utilizes descriptive and comparative analysis, including the t-test, to provide insights into the performance of solar panels. It is worth noting that additional tests and comparisons can be incorporated into the model to further enhance its effectiveness. Furthermore, the model can be extended to enable daily comparisons instead of monthly assessments, particularly if prompt action is required to address degradation issues in a timely manner.

Overall, the DDM presented in this paper serves as a valuable tool for decision-makers in the solar panel industry. By utilizing descriptive and comparative analyses, stakeholders can make informed decisions to optimize the performance and maintenance of solar panel systems.

For future work, the model needs to be tested in practice to validate its applicability in real-life scenarios. It may also be useful to benchmark its results against other similar models to evaluate relative performance and effectiveness.