In short: CIGRE TB 761 is guidance on transformer condition assessment, not a certifiable standard with one mandatory health-index formula. Prepared by WG A2.49, it discusses purpose-specific Transformer Assessment Indices, alternative scoring methods, failure mechanisms and uncertainty. A defensible implementation explains its choices and evidence; it does not claim universal CIGRE compliance from a particular input list, score or software feature.
Maintenance, refurbishment and replacement decisions benefit from condition evidence alongside age, service history and operational constraints. CIGRE Technical Brochure 761, published in 2019 as Condition assessment of power transformers by WG A2.49, describes how to develop and use Transformer Assessment Indices (TAIs) for those decisions.
What does it mean, in practice, to align an assessment method with TB 761?
A utility or customer may choose to incorporate CIGRE guidance into a specification, but that does not make every recommendation a universal regulatory or insurance requirement. The useful question is whether the chosen method fits its purpose, represents the relevant failure mechanisms, handles uncertainty and supports a reviewable engineering decision.
1. Define the assessment purpose and depth
TB 761 describes assessment methods with different levels of information and uncertainty. It does not mandate the three stages below or a fixed test package for each. The following is one practical way to organise an investigation, with its depth chosen for the asset and the suspected failure mechanisms:
- Fleet screening: Review available routine oil tests, DGA, inspections and operating history. Record which components and failure mechanisms are covered and where information is missing. This is a practical workflow, not a minimum test list prescribed by TB 761.
- Targeted assessment: Where screening or asset importance warrants it, select additional tests that address the suspected mechanism. Depending on the case, these may include furans, dielectric measurements, frequency response analysis or bushing diagnostics. More tests are useful only when their interpretation supports the decision.
- Specialist investigation: Difficult cases may require detailed engineering analysis, internal inspection or material testing. Such work is selected according to the problem, safety constraints and expected value of the information; it is not reserved exclusively for post-failure investigation.
Data confidence: TB 761 Chapter 4 discusses uncertainty and missing information. A system should make their effect visible rather than silently treating an unmeasured variable as healthy. Rescaling, using a worst-case assumption or reporting a range are methodological choices that need justification. Dynamic scaling by three compulsory tiers is not a TB 761 requirement.
2. Assess components and failure mechanisms
TB 761 discusses the failure mechanisms and diagnostic methods associated with transformer sub-components. The user selects the mechanisms relevant to the purpose of the index. A separate numerical score for every component is one possible representation, not a universal requirement. The important point is that an aggregate must not hide evidence of a serious individual defect.
A. Solid insulation and cellulose ageing
Cellulose degradation reduces the mechanical strength of paper insulation. Degree of polymerisation can be measured on a paper sample; estimates derived from dissolved markers are indirect and depend on the paper, oil history and model used. TB 761 discusses these methods without requiring one DP estimate for every assessment.
- Evidence to review: 2-FAL and other cellulose-related indicators can support an ageing assessment. They do not directly measure the condition of the most degraded paper in a transformer. Oil treatment or replacement, paper type and temperature history affect interpretation. A low estimated DP warrants engineering review, but a universal DP threshold cannot by itself prescribe retirement, establish short-circuit strength or certify that an insulation system is pristine.
B. On-Load Tap Changers (OLTC) and Bushings
OLTCs and bushings are important parts of the assessment because they have distinct failure mechanisms and diagnostic needs. Their contribution to observed failures varies with the population and failure definitions; no universal percentage is assumed here.
- OLTC evidence: Consider operating count, inspection and maintenance history, compartment-specific oil results, temperature behaviour and relevant electrical or mechanical tests. Interpret DGA using a method appropriate to the switching design and compartment; main-tank criteria are not automatically transferable.
- Bushing evidence: Capacitance, dissipation factor or power factor, inspections and other measurements can support condition assessment. Online leakage-current monitoring may be useful for selected assets, but TB 761 does not impose it as a universal compliance condition. Interpretation must consider bushing technology, measurement conditions and trend.
3. Aggregation, weighting and visibility of defects
TB 761 Section 2.4 and Table 2-2 compare several scoring approaches, including summation, multiplication, estimated failure probabilities, worst-case and hybrid methods. Their advantages and drawbacks differ. Two issues deserve particular attention:
- Weighting factors: Explain why a mechanism has a particular weight and test whether that choice can hide a problem. Weighting by perceived importance does not correct weak evidence, and a gas concentration alone is not a complete assessment of the mechanism or its severity.
- Preventing masking: A serious individual defect should remain visible when other indicators are satisfactory. TB 761 presents worst-case and hybrid methods as options for doing this; it does not require every algorithm to force the entire index into a universal red zone. Keep urgent defect flags and the underlying evidence accessible alongside the aggregate score.
4. Online monitoring and research directions
New monitoring and analytical methods can improve the evidence available to an engineer. They should be assessed against the relevant failure mechanisms and the decisions they support. Their availability does not create new universal TB 761 compliance requirements:
- Online DGA and edge processing: TB 761 Chapter 5 already discusses online monitors. Higher-frequency measurements may reveal changes between laboratory samples, provided measurement quality, baseline and fluid-specific interpretation are addressed. Online monitoring is a risk-based equipment choice, not a mandatory feature of every index.
- Partial-discharge monitoring: UHF and other PD measurements may contribute useful evidence for suitable equipment and defects. Sensor location, sensitivity, interference and interpretation need to be established for the application. TB 761 does not make a particular sensor installation a general acceptance or compliance requirement.
- Machine learning and remaining life: TB 761 discusses machine learning as one possible scoring approach, with data and validation limitations. It does not establish neural-network prediction of an exact end-of-life year. Separate a condition score, a forecast of a measured variable and a validated estimate of failure probability or remaining life; each is a different claim requiring different evidence.
Conclusion
Applying TB 761 thoughtfully means defining the purpose of an assessment, examining relevant failure mechanisms, choosing a scoring method and exposing uncertainty. The result should help the user identify which assets need attention and then inspect the evidence needed for a practical decision.
For O&M managers and grid engineers, a documented method can make maintenance and investment proposals easier to review. It does not guarantee a longer asset life, regulatory acceptance or an irrefutable risk estimate. The supporting records should show the assessed defects, confidence, alternative actions and consequences relevant to the decision.
Technical references: CIGRE TB 761, Condition assessment of power transformers (2019), WG A2.49: executive summary (p. 5); Chapters 1–2 and Table 2-2 (scoring alternatives, pp. 26–28); Chapter 4 (uncertainty); Chapter 5 (online monitoring); Chapter 7 (probability of failure); Section 8.1 (solid insulation).
