Real-time Fault Detection System For Large Scale Grid-Integrated Solar Photovoltaic Power Plants

Abstract

The global solar photovoltaic (PV) installations are increasing rapidly in an effort to slow down the process of global warming and climate change. The clean and green energy from solar PV power plants is being utilized at every level from utility scale to distributed generation applications. The large scale PV plants are normally installed in grid-tied topology because of its simple and easy to install and operate design. As the size of grid-tied PV power plants increases, the probability of error occurrence also increases. When a small PV plant is installed on the rooftop of a house it is much easier to trace a fault and get the system back on track, however, as the sizes of PV plants grow the string sizes increase in PV arrays and it becomes a cumbersome and tiring activity to find the nature and location of the fault. In this study, an algorithm with system for fault detection and identification is presented which is able to sense the abnormality in the DC power output and identify the fault in the DC side of PV power plant. The algorithm proposes mathematical operations which use basic statistical tools of mean and standard deviation to formulate the fault detection mechanism. Additionally, string power ratio is used to detect fault at the maximum power point tracking (MPPT) unit. In order to implement and test the proposed fault detection system for large scale grid-integrated PV plants, a simulation model of PV system is developed in MATLAB Simulink based on 125 kWp grid-tied PV power plant installed at Wah Nobel Group of Companies, Wah Cantt., Pakistan. The modeled PV plant is configured such that it is able to very closely match its DC power output to that of reference PV plant. Different faults have been studied which can cause reduction in energy yield of PV plants. The study discusses and keeps its focus on the faults of faulty PV modules in a string, single or multiple strings failure, MPPT unit failure, partial shading on a PV string and soiling on a PV string. Weather conditions, irradiance and PV module temperature, are obtained from the reference PV plant for several days and simulations are conducted to test the viii performance of proposed fault detection system. The results obtained from tests show that the proposed fault detection and identification system is able to detect an abnormal event in the PV system and accurately identify the nature of fault as well.