Experimental Evaluation of PV Inverter Anti-Islanding with Grid Support Functions in Multi-Inverter Island Scenarios

As PV and other DER systems are connected to the grid at increased penetration levels, island detection may become more challenging for two reasons: 1.) In islands containing many DERs, active inverter-based anti-islanding methods may have more difficulty detecting islands because each individual inverter's efforts to detect the island may be interfered with by the other inverters in the island. 2.) The increasing numbers of DERs are leading to new requirements that DERs ride through grid disturbances and even actively try to regulate grid voltage and frequency back towards nominal operating conditions. These new grid support requirements may directly or indirectly interfere with anti-islanding controls. This report describes a series of tests designed to examine the impacts of both grid support functions and multi-inverter islands on anti-islanding effectiveness. Crucially, the multi-inverter anti-islanding tests described in this report examine scenarios with multiple inverters connected to multiple different points on the grid. While this so-called 'solar subdivision' scenario has been examined to some extent through simulation, this is the first known work to test it using hardware inverters. This was accomplished through the use of power hardware-in-the-loop (PHIL) simulation, which allows the hardware inverters to be connected to a real-time transient simulation of an electric power system that can be easily reconfigured to test various distribution circuit scenarios. The anti-islanding test design was a modified version of the unintentional islanding test in IEEE Standard 1547.1, which creates a balanced, resonant island with the intent of creating a highly challenging condition for island detection. Three common, commercially available single-phase PV inverters from three different manufacturers were tested. The first part of this work examined each inverter individually using a series of pure hardware resistive-inductive-capacitive (RLC) resonant load based anti-islanding tests to determine the worst-case configuration of grid support functions for each inverter. A grid support function is a function an inverter performs to help stabilize the grid or drive the grid back towards its nominal operating point. The four grid support functions examined here were voltage ride-through, frequency ride-through, Volt-VAr control, and frequency-Watt control. The worst-case grid support configuration was defined as the configuration that led to the maximum island duration (or run-on time, ROT) out of 50 tests of each inverter. For each of the three inverters, it was observed that maximum ROT increased when voltage and frequency ride-through were activated. No conclusive evidence was found that Volt-VAr control or frequency-Watt control increased maximum ROT. Over all single-inverter test cases, the maximum ROT was 711 ms, well below the two-second limit currently imposed by IEEE Standard 1547-2003. A subsequent series of 244 experiments tested all three inverters simultaneously in the same island. These tests again used a procedure based on the IEEE 1547.1 unintentional islanding test to create a difficult-to-detect island condition. For these tests, which used the two worst-case grid support function configurations from the single-inverter tests, the inverters were connected to a variety of island circuit topologies designed to represent the variety of multiple-inverter islands that may occur on real distribution circuits. The interconnecting circuits and the resonant island load itself were represented in the real-time PHIL model. PHIL techniques similar to those employed here have been previously used and validated for anti-islanding tests, and the PHIL resonant load model used in this test was successfully validated by comparing single-inverter PHIL tests to conventional tests using an RLC load bank.