Reliability Modeling of PV Farm Using Markov Model

Utilization of photovoltaic units in power networks and their participation in the power supply has increased in recent years. Total world capacity of PV units has grown exponentially from 1.5 GW in 2000 to about 300 GW in 2016. This paper presents an analytical method for evaluating the reliability of large photovoltaic farms with regard to the changes in input power and reliability indices of unit components. The proposed method is not only capable of estimating the annual energy production of the photovoltaic units, but also able to calculate the system reliability indices. With Markov approach, Frequency and Duration method is utilized in order to model a photovoltaic farm similar to the multistate conventional units. Probability, frequency and transition rate of each state is obtained using the statistical data of solar radiation as well as operational characteristics of a photovoltaic unit. Due to the large number of solar radiation and PV modules operational states, k-means clustering algorithm is used for data classification. This analytical method is applied to the RBTS to demonstrate the effectiveness of developed approach. Capacity credit using LOLF index, is more than what obtained in the case of LOLE index. This observation can be explained through the variable nature of solar radiation in comparison with conventional units. Hence there is a need to obtain a model which reflects the frequency based reliability indices of PV systems.