Interpreting Transformer Frequency Response Analyzer Results: A Guide to FRA Curve Analysis

Collecting data with a transformer frequency response analyzer is only the first step in transformer condition assessment. The true challenge—and the true value—lies in interpreting the frequency response curves to identify mechanical defects. For maintenance engineers and asset managers, understanding what the traces reveal about internal winding geometry is essential for making informed decisions. This guide provides a systematic approach to analyzing SFRA results, from visual inspection to numerical evaluation .

The Three Frequency Regions of Analysis

Frequency response curves generated by a frequency response analyzer are typically divided into three distinct regions, each corresponding to different physical components of the transformer. Understanding these regions is fundamental to accurate fault identification.

Low Frequency Region (Below 2 kHz)

The low frequency region is dominated by the core characteristics—specifically the magnetizing inductance and the core losses. In this band, the response is primarily influenced by the core's magnetic properties and its grounding configuration. Deviations in this region typically indicate core-related issues such as residual magnetism, core lamination shorts, or problems with the core ground connection. If your transformer frequency response analyzer shows significant divergence below 2 kHz between phases or compared to the fingerprint, core damage or improper grounding should be suspected .

Medium Frequency Region (2 kHz to 200 kHz)

The medium frequency range is the most critical for detecting winding deformation. In this region, the response is governed by the interaction between the winding inductances and the series capacitances between turns and discs. This is where the mechanical integrity of the winding structure is most visible. Axial displacement (where windings shift vertically) and radial buckling (where windings distort outward) create characteristic resonant peak shifts in this band. When analyzing medium frequency traces from your winding deformation tester, look for:

• Peak Shifts: Movement of resonance peaks to the left (lower frequency) often indicates increased inductance due to winding expansion, while right shifts suggest decreased inductance from compression.

• Peak Amplitude Changes: Damping of resonances (lower amplitude) can indicate increased losses from mechanical damage or shorted turns.

• New Peaks or Missing Peaks: The appearance or disappearance of resonances suggests significant geometric reorganization.

High Frequency Region (Above 200 kHz)

The high frequency region reflects the internal structure of the individual windings and the leads. Here, the response is controlled by the very small capacitances between adjacent turns and the inductance of the connecting leads. Deviations in this region typically point to problems with the tap changer contacts, lead displacement, or movement of internal shielding. While less common than medium frequency faults, high-frequency anomalies can indicate deterioration of the tap changer selector or damage to electrostatic shields .

Common Fault Signatures and Their Patterns

Experienced users of transformer frequency response analyzer equipment recognize that different mechanical faults produce distinct patterns across the frequency spectrum.

Axial Displacement

When windings shift axially (vertically), the capacitive coupling between high and low voltage windings changes significantly. This creates a characteristic pattern: major resonance shifts in the medium frequency region, typically with good correlation between phases at low frequencies but progressive divergence as frequency increases. The affected phase will show a systematic shift of multiple resonant peaks compared to healthy phases .

Radial Buckling

Radial deformation, often caused by high through-fault currents, appears differently. The physical distortion changes both the inductance and the capacitance of the affected section. On the frequency response analyzer display, radial buckling typically creates localized changes—specific peaks may split into double peaks or show dramatic amplitude reduction while neighboring peaks remain relatively stable. This localized nature helps distinguish buckling from generalized displacement .

Shorted Turns

Shorted turns represent a critical failure mode. When a turn-to-turn short occurs, it creates a secondary loop coupled to the main winding. This appears on the frequency response as a significant damping of resonances across a broad frequency range. The affected phase will show lower Q-factors (broader, shallower peaks) compared to healthy phases, and the correlation coefficient will be markedly lower .

Numerical Analysis Methods

While visual inspection by an expert remains valuable, modern diagnostics increasingly rely on numerical indices to quantify the differences between traces. These indices transform subjective observations into objective data that can be tracked over time.

Correlation Coefficient

The Correlation Coefficient (CC) is the most widely used numerical indicator. It measures the similarity between two curves on a scale where 1.0 indicates perfect correlation. IEEE standards suggest that CC values above 0.98 indicate excellent agreement, while values below 0.95 warrant investigation. When using a transformer frequency response analyzer with automated software, pay attention to how the CC varies across frequency bands—a good overall CC but poor medium-frequency correlation may still indicate winding problems .

Sum of Squared Errors and Absolute Difference

The Sum of Squared Errors (SSE) and Absolute Difference (AD) provide complementary information. These metrics are more sensitive to amplitude differences than the correlation coefficient. When comparing traces from your winding deformation tester, increasing SSE values over time suggest progressive deterioration, even before the deformation becomes severe enough to trigger correlation-based alarms .

Comparative Analysis Methods

There are three standard comparison methods used in SFRA interpretation:

• Phase-to-Phase Comparison: Comparing the three phases of the same transformer. This is the most common method for field testing, assuming that at least one phase remains mechanically sound.

• Time-Based Comparison: Comparing current results with historical fingerprints from the same transformer. This is the most sensitive method for detecting gradual deterioration.

• Sister Unit Comparison: Comparing with identical transformers from the same manufacturer and design. Useful when no baseline exists for the test unit.

Common Interpretation Pitfalls

Even with advanced transformer frequency response analyzer technology, interpretation errors can occur. Tap changer position variations are a frequent source of confusion—always verify that the tap position matches the reference measurement. Temperature effects can also shift resonances slightly; significant temperature differences between tests (greater than 10°C) should be noted and considered during analysis. Additionally, be cautious when interpreting tests performed with different lead configurations or cable lengths, as these introduce systematic variations that mimic real faults .

Conclusion

Mastering the interpretation of transformer frequency response analyzer results transforms raw data into actionable intelligence. By understanding the three frequency regions, recognizing common fault signatures, and applying numerical analysis methods, engineers can accurately assess transformer mechanical integrity. This diagnostic capability enables condition-based maintenance, extends asset life, and prevents catastrophic failures. As artificial intelligence continues to advance, the combination of expert knowledge and automated pattern recognition will further enhance the value of SFRA as a cornerstone of transformer asset management .