Research on Diagnostics of Degraded Photovoltaic Devices


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date1 Apr 2022


Photovoltaics are on the rise as the world's fastest-growing energy resource and are expected to supply more than a quarter of total power consumption by humanity within the next 3 decades. This creates pressure to assure the devices entering the market are capable of fulfilling the long-term energy needs. Although these devices undergo certified testing before entering the market, this testing is the bottom line rather than the assurance of a long lifespan.

Therefore, the main focus of the research presented in this thesis is to study the photovoltaic devices throughout their life and describe their performance as they degrade. Degradation of photovoltaic modules is a complex process with hard-to-predict external factors, such as moisture ingress, cyclical thermal expansion, and rare events such as extreme weather conditions during hail storms. The irregularity and unpredictable long-term effects are the major drivers to deliver solutions for continual monitoring while assuring the economics of such an approach are not reflected in the price of generated electricity. Therefore, the monitoring and subsequent diagnostic methods need to be cheap and non-intrusive to the operation of photovoltaic plants.

The thesis delivers one such solution in form of a novel data acquisition unit. It monitors photovoltaic modules and non-intrusively obtains dynamic I-V characteristics and provides its subsequent analysis. The performance of the method is tested on several module technologies and the findings are encouraging enough that the method is protected by a patent.

The diagnostics significantly depends on understanding the complex interaction of factors that influence the performance of photovoltaic modules. Since the degradation is difficult to quantify at the macro-level of a module, there is a lack of simulation tools to efficiently model a degraded module. Many authors either simplify the model of the module to such extent that the inner workings are lost or go to such detail, that it is not possible to simulate the micro-effects on the macro-scale due to the computational cost. In this regard, the thesis presents a novel model of a degraded photovoltaic module, which simulates the sub-cell level detail, while efficiently evaluating the overall I-V characteristic of the module. Furthermore, the proposed model is only a part of a diagnostic method, which further incorporates the I-V characteristic of a module and its electroluminescent image and applies machine learning concepts to deliver a solution that is capable of characterizing individual photovoltaic cells without accessing them directly. This solution has the potential to perform diagnostics on a large scale simply by using the already available data, thus not incurring additional costs.

Despite photovoltaics being both active research and industrial field for the past 3 decades, new technologies are being delivered every year. Regarding the photovoltaic modules, the author is glad to enter the research at this point in time, as the photovoltaics are undergoing a massive shift, overhauling the standard design of photovoltaic cells to raise their efficiency from well established $18 % to $22 % by simply allowing the sunlight to reach the silicon substrate from the rear side. Since this design has no long-term data on degradation, the presented research in this thesis performs a novel way of comparative testing of standard monofacial cells and novel bifacial cells by carefully inducing the same degradation on both. This allows comparing the performance under degradation of both cell types in a fair manner. To ensure the results are not tied to a performance at the cell level, a novel way of investigating the impact of cell-level degradation on the string level performance is proposed in a form of virtual photovoltaic string. A virtual photovoltaic string mimics its real-world counterpart but does not require any hardware, thus the investigation is free from external factors and inexpensive.

The thesis presents the above topics in reversed order, from testing of photovoltaic cells to modeling photovoltaic modules to diagnosing the changes using a data acquisition unit. This is motivated by a didactic perspective to build the reader's understanding of the hurdles in photovoltaics step-by-step since the concepts covered in chronological order go from engineering to fundamental level. Therefore, reversing the order allows the reader to gain an understanding of issues related to photovoltaics first and later on to learn about proposed engineering solutions.

    Research areas

  • Photovoltaics, Renewable Energy, Diagnostics