Introduction

Ensuring reliable protection is one of the primary challenges for distribution systems with increasing distributed energy resources (DERs). Paired with computational intelligence and advanced system management capabilities, these systems are also referred to as Active Distribution Network (ADNs). ADNs have remote infeed from DERs resulting in bidirectional power flow and, variable fault current levels and voltage profiles, which may cause blinding and/or sympathetic tripping for overcurrent relays (OCRs). The utilities operate with an anti-islanding requirement to ensure that line protection is unaffected by DERs. However, this will change with the revised IEEE 1547 (2018), which mandates minimum DER ride-through to support the faulted system unless tripped by the protection. As a result, the OCRs need to detect faults with DERs connected.

Fault pickup for OCRs is set between (usually twice) the nominal line load and (half of) the minimum fault current (I f min) in their protection zones/. This makes the pickup sensitive to line loading, short circuit capacity, the power flow direction, and DER remote infeed. For grid-connected systems, the difference in I f min between primary and backup protection may be sufficient for relay coordination. However, with nested microgrids, this may not hold good. Traditional coordination can severely delay the response to primary zone faults. Adaptive overcurrent relaying (AOCR) schemes address this by adjusting the pickup based on the operating conditions (DER penetration, grid forming source location and others) to help maintain the selectivity and sensitivity.

Most approaches in related research formulate AOCR as an optimization problem. Given the computational burden, optimization based approaches cannot be solved within the overcurrent relays. Consequently, there is an inherent bias for centralized architectures using external controllers and a dependence on communicating volumes of analog data to maintain adaptivity. This can be a concern for a system with a large number of nodes/relays. To work around this challenge, limits the cases based on significant changes in topology, achieves the same by using pre-defined adaptive settings in multiple relay groups, selected given the system equivalent impedance. Decentralized approaches using agents have also been proposed with AOCR formulated as an optimization problem. Expectedly, as notes, the process is communication intensive and time-consuming (100 s of milliseconds) which needs to be addressed to avoid misoperation in time-critical cases.

Other approaches incorporate the impact of line load using the previous day load data from smart meters to determine AOCR pickup. This approach is susceptible to DER output, and can negatively affect the relay’s sensitivity. But, impact of line load on the relay’s pickup needs to be considered as distribution systems move towards integration of nested microgrids. Finally, all existing AOCR schemes have a common assumption – they reprogram the new relay settings using communication or as setting groups. With setting groups, protection engineer also needs to ensure that the new group is applicable in entirety (not just the OCR element) to the new system state. As relay manufacturers also note, when reprogramming the static relay settings or changing the setting group, the relay disables itself (all its functionalities) for a short period. This is a limitation as the new settings need to be reflashed into the microprocessor for immediate access. Existing AOCRs are tested for the first fault detection and usually specific to either the grid or microgrid modes, of operation. This creates a new use-case for potential misoperation in ADNs with high DER penetration which is elaborated (Section III), addressed and validated (Section VI-B) in this paper.

In view of the concerns associated with existing AOCR schemes, this paper proposes a novel dynamic AOCR strategy. The scheme has been designed for ADNs with high DER penetration, capable of operating in both grid-connected and microgrid modes. A key advantage is being implementable within the overcurrent relays (no external controllers needed) allowing them to change the settings dynamically. To achieve these objectives, the paper re-formulates the AOCR problem to simplify the computational burden and moves away from the optimization-based approach. The scheme is tested in Matlab/Simulink with transient domain simulations for high PV penetrations with ride-through capabilities for grid and microgrid modes of system operation. Section IV describes the proposed dynamic AOCR scheme to address these concerns. The test system and corresponding DER/protection relay models are discussed in Section V. Section VI presents the results. Finally, Section VII concludes the paper.

Challenges for existing adaptive approaches

The key challenges with existing AOCR schemes emerge with the upcoming DER interconnection requirements and transitioning/operating as microgrids. Addressing these will be formative for the development of more versatile AOCR schemes supporting grid/microgrid modes of operation.

A. Underreaching OCRs With Multiple DER Generation Levels

Considering DERs as constant current sources, existing AOCRs often downscale the DER fault current contribution for different levels of output w.r.t. rated power. However, DERs can feed maximum fault current even when producing low unfaulted output power because they behave like a terminal voltage controlled current source. depends on highly variable parameters like distance from the fault, fault impedance, and fault type. As a result, existing AOCRs may overcompensate DER output for lower output power and under-reach.

B. Delayed Trip Time for a Fault in the Primary Protection Zone

Microgrids can have significantly lower fault currents compared to grid-connected systems. This negatively affects the OCR sensitivity in primary/secondary zones. In the microgrid mode, traditional time-based coordination can cause a significant delay in a relay’s trip time for primary zone faults. Existing AOCR approaches do not discuss this problem.

C. Current Direction Reversal Due to the Mode Switching

Current direction reversal is possible for i) the grid-connected system given the DER output states, or ii) microgrids depending on the location of the grid forming DER. Existing AOCR approaches seldom discuss the directional aspects of protection. However, adapting directional supervision when available is important to ensure selectivity towards internal vs external faults. Without accounting for directional supervision, it is difficult to create a robust scheme for the coexistence of grid-connected and microgrid modes.

These observations show that the existing AOCRs lack applicability to ADNs with both grid and microgrid modes, are vulnerable to under-reaching and changing fault current, and can be too complex to implement in traditional distribution relays. The next section elaborates a new emerging vulnerability difficult to address with current practices to programming the relay.