Islanding Detection in PV Inverters: A Case Study

Introduction

When the utility grid is disconnected, grid-tied photovoltaic inverters may continue energizing local loads, creating an unintentional island. This condition, known as islanding, poses serious safety threats to utility workers and can damage customer equipment. Anti-islanding protection is therefore mandatory in modern inverters to detect grid loss within the required time frame.

This case study examines real-world islanding scenarios, compares passive and active detection strategies, and discusses how to shrink the non-detection zone (NDZ) when multiple inverters operate in parallel.

Hazards of Unintentional Islanding

In a typical rooftop PV installation, an inverter monitors grid voltage and frequency. If a fault opens the upstream breaker, the inverter may keep powering nearby loads, forming an island. The main danger is that utility workers assume the line is de-energized, yet the island remains live, creating an electrocution risk.

Beyond safety, islanding can cause voltage and frequency fluctuations that damage sensitive electronic equipment. Induction motors may trip, and reclosing the grid onto an unsynchronized island can generate severe transients. These hazards make rapid detection and shutdown essential for both grid stability and worker safety.

Passive Detection: Voltage and Frequency Monitoring

Passive methods rely on measuring local grid parameters. Under-frequency/over-frequency and under-voltage/over-voltage relays are common. When the grid disconnects, active power imbalance usually drives frequency away from nominal, and reactive power imbalance shifts voltage. However, if local load closely matches generated power, these deviations may be too small to trigger protection, defining the NDZ.

More advanced passive techniques include voltage phase jump detection and harmonic monitoring. Phase jump identifies sudden changes in inverter terminal voltage angle, while harmonic analysis looks for increased lower-order harmonics due to the island's higher impedance. These methods add sensitivity but still fail under certain load conditions.

Active Detection: Frequency Shift and Reactive Power Disturbance

Active methods inject deliberate disturbances into the system to force measurable changes. The Sandia frequency shift (SFS) method, also called frequency bias, applies positive feedback between frequency deviation and inverter current phase. In grid-connected mode, the stiff grid holds frequency constant, so the feedback has no effect. After islanding, any small frequency drift is amplified until a threshold trips the inverter.

Reactive power disturbance, or reactive power variation, periodically injects a small oscillating reactive current. Because the island's load consumes reactive power, the voltage amplitude or frequency fluctuates beyond limits, violating the inverter's ride-through tolerance. Active methods generally shrink NDZ dramatically, but they introduce power quality degradation and can cause unintended interactions among multiple inverters.

Another technique is the slip-mode frequency shift, which changes phase angle based on frequency error. In a parallel system, if each inverter applies a different disturbance frequency, they may cancel each other's effects. Coordinated active methods that use a common perturbation frequency or rotating sequence can avoid these interactions.

Eliminating NDZ in Multi-Inverter Parallel Systems

When several inverters are connected to one distribution feeder, active disturbances from different units can interfere. For example, two inverters using opposing frequency-shift gains might neutralize the signal. To address this, utilities often require inverters to implement active methods with a guaranteed response, such as the Sandia voltage shift combined with frequency shift.

Modern inverters also use impedance-measurement techniques. They periodically inject a high-frequency signal and measure the resulting impedance. A low-impedance grid connection damps that signal, while an island shows higher impedance at that frequency. This method is robust against multi-inverter cancellation because each inverter can detect the impedance change independently, provided the signal frequencies are distinct.

Another practical approach is multimode detection. A single inverter combines passive algorithms with active disturbances, and local controllers communicate via open-source protocols like IEEE 2030.5 to share islanding status. In one field case, a 2 MW solar farm reduced detection time from 2.1 seconds to under 0.5 seconds by switching from passive-only to a coordinated SFS with a rotating disturbance order.

Conclusion

Islanding detection remains a critical safety function for photovoltaic inverters. Passive methods are simple but suffer from NDZ, while active methods significantly reduce NDZ at the cost of potential interference among parallel inverters. No single technique is perfect; the best engineering practice uses a hierarchical approach with passive monitoring, active disturbance, and communication-based coordination.

As solar penetration grows, detection methods must evolve to handle faster grid dynamics and multi-inverter communication. The case study demonstrates that a well-designed, coordinated anti-islanding scheme can eliminate NDZ in practice while maintaining high power quality, making distributed PV safer for workers and equipment.