Chapter Six - Enhancing relay resiliency in an active distribution network using latest data-driven protection schemes

Abstract

Today’s distribution grids are becoming active due to installation of distributed energy resources (DERs), which are mostly renewable energy sources such as photovoltaic (PV) and wind energy. These DERs have provided an efficient solution for uninterrupted and good power quality supply to the consumers even under substation outage. However, with the inculcation of DERs, various issues arise such as bidirectional power flow, intermittent power output from DERs, and vast variation in short circuit level while operating in standalone mode. These issues lead to the failure of conventional protection schemes. Therefore, a robust protection method is the need of the hour to address these issues. This chapter discusses different protection schemes in an active distribution network and micro grid (MG). Initially, major issues related to micro grid protection has been conveyed and further various possible solutions are discussed in a comprehensive manner. A preliminary protection method, which includes finding separate relay settings for MG operation in grid connected (GC) and islanded mode (IM), is discussed. This method requires offline calculation of settings and regular update whenever operating mode of MG is changed. To overcome this recurrent effort, an adaptive relaying concept is further discussed, where online relay settings update is possible. This scheme is possible in a smart MG which includes intelligent electronics devices (IEDs), smart digital numerical relays (SDNRs), phasor measurement units (PMU), and well-equipped communication channels. The concept of central protection computer (CPC) has made it possible to access real-time electrical data and store them securely. These data can be used extensively to acquire useful insights of an MG or active network to design a decision-making model. These models can provide more reliable protection of MG, even when MGs operating dynamics are changing. Recently, machine learning (ML) has evolved rapidly in this field. Therefore, this chapter also discusses some of the latest ML-based schemes to perform protection tasks such as fault presence, type and phase classification, and fault location prediction. Initially, the steps for designing a ML model are presented for new learners. Using the best ML model, it can be applied to detect and isolate fault in a MG without the need of updating the relay settings. Application of data-driven and ML-based methodology are proving to be efficient in maintaining the resiliency of an active grid.