Transformer Health Indices: How Condition Scoring Supports Fleet Decisions

Transformer condition assessment draws on many data sources: DGA results, oil quality parameters (moisture, acid number, IFT), electrical test results (power factor, capacitance, winding resistance), physical inspection observations, loading history, and nameplate age. Each provides a partial picture of the transformer's condition and remaining service life.

A transformer health index is an attempt to synthesise these inputs into a single summary score for fleet comparison and maintenance prioritisation. Health indices have become widely used in utility asset management and they serve real purposes. They also have structural limitations that are important to understand, particularly in how the DGA component is constructed.

What Health Indices Are Designed to Do

The primary value of a transformer health index is portfolio visibility. A utility operating hundreds or thousands of transformers cannot present detailed condition profiles for every unit to senior management or regulators. A health index score such as "17% of the transmission fleet is in Condition C or D" communicates portfolio risk in terms that are accessible to non-specialist decision-makers and useful for capital budget justification.

CIGRE TB 227 [1] and TB 445 [2] both address health index frameworks as tools for life management and maintenance strategy, noting that the primary application is fleet-level prioritisation and capital planning rather than individual asset diagnosis.

Health indices also support regulatory reporting in jurisdictions where grid owners are required to report on the condition of critical infrastructure assets. A standardised scoring framework provides the consistent, repeatable assessment that regulatory reporting requires.

How Health Indices Are Constructed

A typical transformer health index assigns numerical scores to individual condition indicators and combines them into a total score using a weighting scheme.

Condition indicator scores. Each input, including DGA condition level, oil quality rating, electrical test results, inspection findings, and age-based assessment, is mapped to a score on a common scale (for example, 0–100 or A–F). The mapping specifies what DGA result earns what score, what moisture content earns what score, and so on.

Weighting. The scores are combined using weights that reflect the relative importance of each condition indicator for the overall transformer risk assessment. DGA is typically the most heavily weighted input [2], reflecting its sensitivity to active internal faults. Physical age may receive lower weight relative to condition indicators.

Aggregation. The weighted combination produces a total health index score. Common approaches include linear weighted averages, lowest-score dominance (the overall score is driven primarily by the worst individual indicator), and tiered systems where the overall category is determined by the worst category of any individual indicator.

The DGA Component: Why It Matters and What It Should Include

The DGA input to a health index is the most consequential component for detecting active fault risk. How it is constructed determines how much of the available DGA information the health index actually uses.

Threshold-based scoring. The most common approach assigns DGA scores based on whether gas concentrations are within, above, or significantly above the limits in IEEE C57.104-2019 [3]. A transformer with all gases in Condition 1 scores full points; one in Condition 3 scores poorly.

This approach is straightforward to implement and auditable, but it inherits all the limitations of threshold-based DGA interpretation [4]. Two transformers with the same concentrations receive the same score regardless of whether one has been stable for ten years and the other has tripled in the past six months. The trajectory information, often the most important signal for near-term failure risk, is invisible.

R-DGA-based scoring. A more accurate DGA component uses CSEV and HF from R-DGA methodology [4] as the input to the health index scoring function. A transformer with HF in the upper population quartile receives a poor DGA sub-score; one with HF in the lower quartile receives a good sub-score, regardless of whether any absolute threshold has been crossed.

This approach provides a fundamentally more accurate risk signal in the DGA component because it captures both the cumulative severity history and the current fault rate, the two dimensions that determine whether a DGA result represents elevated risk. Dukarm et al. [4] demonstrated that threshold-based assessment produces both false alarms (penalising transformers with elevated but stable, normal-for-population concentrations) and missed detections (not flagging transformers whose trajectory indicates risk accumulation below threshold). Both distortions degrade the accuracy of the health index as a fleet risk signal.

Combining DGA sub-score with other indicators. When R-DGA metrics drive the DGA component, the health index captures active fault risk accurately. The oil quality indicators (moisture, acid number, IFT) and the insulation life indicators (furans, degree of polymerisation) provide the long-term condition dimensions that complement the fault-sensitivity of DGA. Electrical test results (power factor, capacitance ratio, frequency response analysis) provide diagnostic depth for mechanical and dielectric condition. CIGRE TB 812 [5] identifies the failure mode distribution for the transformer population; health index weighting should reflect the probability that each failure mode would be detected by each monitoring method.

Structural Limitations of Health Indices

Even a well-constructed health index has limitations that users should keep in mind.

Weighting subjectivity. There is no universally validated weighting scheme for transformer health indices. The relative weights assigned to different condition indicators reflect engineering judgment; different organisations applying different weighting schemes will produce different scores for the same transformer. When comparing health index scores between organisations or benchmarking against industry targets, the weighting methodology must be considered.

Aggregation discards information. Combining multiple indicators into a single number necessarily loses some of the information in each input. Two transformers with the same overall health index score may have very different maintenance implications if one scores poorly on DGA (active fault risk) and the other scores poorly on age (long-term life concern but no current fault activity). The health index treats both identically; detailed analysis distinguishes them.

They are screening tools, not diagnostic tools. A health index identifies the segment of the fleet warranting closer attention. It does not diagnose the nature of the concern or recommend a specific maintenance action. When a transformer is flagged by its health index, the appropriate next step is detailed condition review, including full DGA trend analysis using TOA, review of other condition indicators, and potentially a site visit or targeted electrical testing, before any maintenance action is decided.

CIGRE TB 445 [2] is explicit that condition scoring is a fleet management and prioritisation tool; individual asset decisions require individual asset analysis.

Practical Integration With R-DGA Fleet Screening

Health indices and R-DGA fleet screening are complementary rather than competing tools. R-DGA fleet screening (CSEV/HF ranking in TOA) provides the most sensitive and specific DGA-based risk signal, updating automatically each time new DGA results are entered, with the full trajectory sensitivity of the population-based methodology. The health index integrates this DGA signal with other condition data sources into a portfolio-level view.

A practical workflow: use TOA's R-DGA fleet ranking as the DGA input to the health index, producing a combined score that accurately reflects both active fault risk (from DGA) and long-term condition (from other indicators). Use the combined health index for capital planning and portfolio communication. Use TOA's detailed trend and fault classification views for individual asset diagnostic decisions.

For technical background on how R-DGA metrics support both fleet screening and health index construction, visit the Science page. For product information, visit the TOA page or contact us.