Forced Oscillation Source Location

The foundational theory posits that a forced oscillation injects energy into the grid. By calculating the dissipating energy flow from PMU data, the net energy contribution of each generator or bus can be determined. A component generating net positive dissipating energy is a source, while one absorbing it is a sink. This transforms a complex dynamic problem into a tractable energy balance calculation.

The method differentiates between sources on the generation and load sides:

• Generator Source: A malfunctioning turbine governor or faulty power system stabilizer injecting periodic mechanical power oscillations.
• Load Source: A cyclical industrial load, like a large compressor or arc furnace, drawing pulsating power. The DEF calculation is applied identically to both, with the sign and magnitude of the energy flow pinpointing the origin regardless of its electrical nature.

Source location accuracy is directly proportional to PMU observability. The DEF is calculated at each PMU-equipped bus. By mapping the spatial distribution of energy injections, the source is triangulated to the area with the highest positive energy density. A dense, time-synchronized network of PMUs is critical to avoid mislocating the source to a nearby resonant node that is merely amplifying the disturbance.

A key characteristic is the ability to distinguish a forced oscillation from a poorly damped natural mode (electromechanical oscillation).

• Forced Oscillation: Energy flows from a specific, identifiable source. The oscillation persists only while the driving force is active.
• Natural Mode: Energy is exchanged between synchronous machines in a standing wave pattern with no single source. The DEF method yields a near-zero net energy balance for natural modes, preventing false source identification.

Real-time source location is being integrated into System Integrity Protection Schemes as a remedial action trigger. Upon detecting a sustained, dangerous forced oscillation and triangulating its source, a SIPS can automatically:

• Send a trip signal to the specific generating unit or load.
• Issue an operator alert with the exact geographic coordinates. This prevents the oscillation from exciting inter-area modes and causing a wide-area blackout.

The classical DEF method assumes a quasi-steady-state sinusoidal injection. Challenges arise with:

• Non-Sinusoidal Sources: Square-wave-like injections from converter-driven loads require advanced signal decomposition.
• Subsynchronous Oscillations (SSO): Interactions with series-compensated lines or wind farm converters demand a modified energy function that accounts for the network's frequency-dependent impedance. Ongoing research focuses on extending the DEF method to these complex, non-traditional source types.

Raw PMU data must be rigorously conditioned before analysis. This stage involves:

• Detrending: Removing the nominal 60/50 Hz frequency to isolate the oscillatory component.
• Band-pass filtering: Extracting the specific frequency of the forced oscillation while rejecting noise and other electromechanical modes.
• Bad data rejection: Filtering out spikes and dropouts that would corrupt the energy calculation.

Source location is not a single measurement but a network-level inference. By calculating the dissipating energy flow on multiple branches across a wide-area model, operators can observe the energy radiating outward from the source bus. The bus with the largest consistent net energy injection across multiple connected lines is identified as the geographic origin of the disturbance.

Critical distinction for diagnosis:

• Natural (Electromechanical) Oscillations: Characterized by a specific mode shape and damping ratio inherent to the system's inertia. They are a response to a transient event.
• Forced Oscillations: Driven by a persistent, periodic external input (e.g., a malfunctioning turbine governor). Their frequency matches the driving force and is independent of system modes. Source location targets the latter.

The DEF method is deployed as an online application within a Phasor Data Concentrator (PDC) or a WAMPAC platform. It operates on a sliding window of streaming data (e.g., 5-10 seconds) to continuously scan for sustained oscillations. Upon detection, it automatically alerts operators with the geographic coordinates of the suspected source, enabling rapid manual or automated mitigation.

Typical root causes identified by this technique include:

• Turbine governor instability: A feedback loop malfunction in a steam or hydro turbine causing cyclic mechanical power output.
• Cyclic loads: Large industrial motors or compressors with a periodic load pattern.
• Control system interactions: Poorly tuned power system stabilizers (PSS) or HVDC controls that inject oscillations instead of damping them.