Predicting photovoltaic module degradation using current-voltage
Loading...
Files
Full Text E-thesis
Date
2022-02-01
Authors
Al Mahdi, Hussain
Journal Title
Journal ISSN
Volume Title
Publisher
University College Cork
Published Version
Abstract
With the global increase of photovoltaic (PV) modules deployment in recent years, monitoring techniques to ensure a safe and healthy operation have become crucial. Despite PV modules being considered reliable devices, failures and extreme degradations often occur. They have been frequently classified into two categories: optical and electrical.
Of optical failures, degradation of the ethylene-vinyl acetate (EVA) encapsulant due to prolonged ultra-violet (UV) exposure and other environmental stress factors, such as temperature and humidity, is the most common failure.
Conversely, electrical failures are linked to low shunt resistance (Rsh). Unidentified degradation mechanisms in these categories lead to catastrophic failure. The objective of this dissertation is to find comprehensive techniques to predict and detect these degradations at the initial stages. This can be achieved by continuously monitoring the current-voltage (I-V) curve parameters of the solar cell, from healthy-state (no degradation) to failure-state (critical degradation). The EVA degradation research was accomplished using an electrical circuit simulator (SPICE) to simulate an experimental result. In addition to simulations, experiments were performed to artificially lower the solar cell shunt resistance.
The effect of EVA and shunt resistance degradations on solar cells’ major parameters; maximum power output (Pmax), short-circuit current (ISC), open-circuit voltage (VOC), and fill factor were analysed. As a result, novel linear models were developed and proposed as strong predictors and observers, and also are suitable for implementation in online monitoring systems for operational PV modules. Training a machine learning model to classify degradation mechanisms was also shown effective. Corrections should be made if critical degradation is detected. Aside from the potential risks to the PV system, the thesis showed that replacing a degraded PV module is cost-effective before the critical degradation state.
Description
Keywords
PV failures detection , Machine learning models , PV degradation , PV simulation
Citation
Al Mahdi, H. 2022. Predicting photovoltaic module degradation using current-voltage. PhD Thesis, University College Cork.