This report focuses on efficient data gathering, storage, distribution, and validation, covering data management topics- from sensor selection to permanent data storage. The report is mainly targeted at system designers and plant operators, aiming to provide checklists and recommendations on these topics.

Fault detection is essential to ensure the proper operation of solar-thermal plants. Hence, monitoring personnel frequently analyze the data to detect unusual behavior. While visualization approaches may considerably support the monitoring personnel during their work, no existing application can yet deal with the multivariate and time-dependent sensor data, or does not fully support the users' workflow. Thus, this work introduces the visual framework SunScreen. It allows users to explore the sensor data, automatically detected anomalies, and system events (e.g., already detected faults and services). The feedback from the users shows that they appreciate the tool and especially its annotation functionality. However, the SUS results indicate that it does not meet all requirements yet.

Fault-Detection (FD) is essential to ensure the performance of solar thermal systems. However, manually analyzing the system can be time-consuming, error-prone, and requires extensive domain knowledge. On the other hand, existing FD algorithms are often too complicated to set up, limited to specific system layouts, or have only limited fault coverage. Hence, a new FD algorithm called Fault-Detective is presented in this paper, which is purely data-driven and can be applied to a wide range of system layouts with minimal configuration effort. It automatically identifies correlated sensors and models their behavior using Random-Forest-Regression. Faults are then detected by comparing predicted and measured values. The algorithm is tested using data from three large-scale solar thermal systems to evaluate its applicability and performance. The results are compared to manual fault detection performed by a domain expert. The evaluation shows that Fault-Detective can successfully identify correlated sensors and model their behavior well, resulting in coefficient-of-determination scores between R²=0.91 and R²=1.00. In addition, all faults detected by the domain experts were correctly spotted by Fault-Detective. The algorithm even identified some faults that the experts missed. However, the use of Fault-Detective is limited by the low precision score of 30% when monitoring temperature sensors. The reason for this is a high number of false alarms raised due to anomalies (e.g., consecutive days with bad weather) instead of faults. Nevertheless, the algorithm shows promising results for monitoring the thermal power of the systems, with an average precision score of 91%.

First results of an algorithmen for detection faults in large-scale solar thermal systems.

The number of large-scale solar thermal installations has increased rapidly in Europe in recent years, with 70% of these systems operating with flat-plate solar collectors. Since these systems cannot be easily switched on and off but directly depend on the solar radiation, they have to be combined with other technologies or integrated in large energy systems. In order to most efficiently integrate and operate solar systems, it is of great importance to consider their expected energy yield to better schedule heat production, storage and distribution. To do so the availability of accurate forecasting methods for the future solar energy yield are essential. Currently available forecasting methods do not meet three important practical requirements: simple implementation, automatic adaption to seasonal changes and wide applicability. For these reasons, a simple and adaptive forecasting method is presented in this paper, which allows to accurately forecast the solar heat production of flat-plate collector systems considering weather forecasts. The method is based on a modified collector efficiency model where the parameters are continuously redetermined to specifically consider the influence of the time of the day. In order to show the wide applicability the method is extensively tested with measurement data of various flat-plate collector systems covering different applications (below 200°Celsius), sizes and orientations. The results show that the method can forecast the solar yield very accurately with a Mean Absolute Range Normalized Error (MARNE) of about 5% using real weather forecasts as inputs and outperforms common forecasting methods by being nearly twice as accurate.