Within SerendiPV project, an online round-table and open discussion on soiling was organized to introduce the innovations developed during the project, get feedback and encourage collaboration for data sharing. As part of the event, the innovations on soling determination within SerendiPV were introduced.
Check some videos and the participants list on:
During the workshop, valuable feedback from the expertes present en the round tables was gathered. Here you can find a summary:
Importance of soiling and measurement
Experts unanimously recognize soiling as a significant issue in photovoltaic (PV) plants, particularly in regions where cleaning is challenging. Accurate measurement of soiling losses is crucial due to the non-homogeneous nature of soiling. It’s essential to have sensors that are not only precise but also cost-effective, easy to maintain, and require infrequent calibration. The introduced sensors are beneficial as one measures real soiling loss at the Maximum Power Point (MPP), while the other is a multi-zone soiling sensor, capturing the non-uniform distribution of soiling across different zones.
Sensor deployment and cost
Balancing the number of sensors installed in a PV plant with the associated costs is critical. Sensors or models need to be accurate across different locations and conditions. For instance, floating PV modules may experience fewer soiling issues compared to Agri-PVs, which are more susceptible to soiling, including the impact of chemicals. It’s also important that sensors or models measure soiling at the string level, not just the module level, to provide a more comprehensive understanding of soiling impacts.
Data quality and algorithm performance
The effectiveness of soiling measurement approaches depends significantly on the quality of the data. Meta-algorithms can be useful for assessing soiling, but their performance varies with the quality of the available data. High-quality data allows these algorithms to function well, while weak and noisy data presents challenges. In such cases, physical models should be considered to complement the data, ensuring that the model parameters align with intuitive expectations.
Data availability and forecasting
A major challenge highlighted by experts is the lack of publicly available data for developing and validating soiling models. Access to diverse data from sites in different locations and under varying conditions is crucial for model development and validation. Forecasting soiling is also important, as it allows for proactive maintenance and management strategies.
The expert feedback underscores the critical importance of accurate soiling measurement, the need for cost-effective and maintainable sensors, the challenge of non-homogeneous soiling, and the vital role of high-quality data in developing and validating effective soiling models.
Also the general audience, provided interesting views on the topic through a survey, here you can find them summarized:
Estimates of soiling losses
Respondents provided varied estimates for losses caused by soiling in their PV plants, reflecting the complexity of the issue. Loss estimates ranged from 2% to over 6%, with some reports indicating even higher losses in particularly dusty environments, such as deserts. A general estimate suggests soiling could cause 5-15% annual energy production losses. While many respondents provided specific figures, such as 2%, 3%, and 6.5%, others used standard estimates or performance ratio (PR) analysis.
Some participants did not have specific estimates, relying instead on general default values. This variability highlights the influence of location, environmental conditions, and measurement methods on soiling loss estimates.
Nature of soiling problems
When asked to describe the nature of soiling problems briefly, dust was the most frequently mentioned issue, underscoring its significance. Bird droppings also emerged as a common concern, along with natural elements like wind and snow that contribute to the accumulation of particulates on PV surfaces. Additional factors noted included aerosols, industrial pollutants, sand, chemicals, soil degradation, humidity, and site layout. These responses reflect a broad awareness of various contributors to soiling in PV plants.
Challenges with soiling sensors
Participants identified several key issues with soiling sensors, including accuracy, calibration, and cost. Concerns also included maintenance, reliability, and the need for regular cleaning of sensors. Many respondents noted a lack of correlation between sensor data and PV module performance, with specific issues such as bird droppings affecting sensor accuracy. Design selection, environmental impacts, data communication, and sensor representativeness were also highlighted. Participants expressed a desire for improved maintenance practices and better sensor performance to minimize losses.
Reactions to innovations
The majority of participants responded positively to the innovations presented, describing them as interesting, promising, good, or worth exploring further. However, some respondents were cautious, noting that the solutions might not be universally applicable or fully effective for all operational issues. There was a call for more data and external measurements to evaluate the innovations’ effectiveness fully. Concerns about area-relative comparisons and non-uniform soiling were mentioned, indicating that a one-size-fits-all solution might be unrealistic. Some felt that the innovations were more suited to laboratory settings than real-world applications.
Preferences for soiling management
Survey responses indicated a preference for using both sensors and modeling in managing soiling issues. Many participants valued the combined strengths of both methods, recognizing that while models offer valuable insights, they need to be validated by actual measurements. Some respondents preferred sensors alone for their direct, real-time data, while a few favoured modeling or used a project-specific approach. This combined approach is seen as crucial for accurate, data-driven decision-making.
Frequency of soiling monitoring requests
The survey revealed a strong demand for soil monitoring and prediction services. Most respondents indicated that these services are frequently requested, reflecting their importance in maintaining PV plant efficiency. However, some noted that the need for these services depends on specific conditions or is limited to certain types of PV plants. One respondent mentioned that soiling monitoring is not frequently requested, suggesting variability based on location or technology.
Reactions from the developers
The wide feedback received from the stakeholders and interested parties was described as very useful by the innovators. On the one hand, it stressed the importance of the issue, and the pertinence of the developments made. On the other hand, it allowed us to understand what the main reluctances are from clients, defining the points to focus on to achieve a successful market introduction. For example, the ones related to the maintenance of the equipment, that would not be a significant issue if incorporated in the general sensor maintenance plan. Or the no need for calibration and very high data availability.
In addition, the experiences shared allow identifying new improvement lines for the soiling devices. For example, the heterogeneity of the dust accumulation patterns in dusty climates makes the soiling impact different from one module to another, even within the same PV string. Therefore, an additional mismatch loss effect is expected to arise, possibly making the effective soiling losses at generator level larger than the ones measured at module level.
In the months after the session, QPV made a first test to try to quantify this effect, and found very promising results. In a desertic environment, the difference between the losses at module and string level were almost negligible for soiling losses in the range of 1-3%. However, as the soiling losses increased, the difference started to grow. For a soiling loss at module level of 6-7%, the soiling loss at string level reached almost 11%. These findings entail a clear improvement line for the soiling sensors, that will need to be able to measure full strings to correctly address the full soiling impact. In the case of the E-Dust, for example, a new version of the equipment should be able to measure a clean module together with a dusted PV string. Besides, to make the solution applicable, the dusted string must be one constituting the PV generator, so the equipment will need to be able to disconnect it from the rest of the generator (through an adequate switch) and reconnect it again after the measurement.
Conclusion
Overall, the survey results show a strong interest in soiling monitoring and prediction services, with a general trend towards frequent requests.
There is a positive outlook on the innovations presented, although further exploration and development are needed.
The survey also highlights a significant discrepancy between actual soiling losses and those modelled or predicted, underscoring the need for more accurate and practical solutions.