Getting the Most From TOA: Database Management and Fleet Analysis

Transformer Oil Analyst™ (TOA) is designed to reduce the effort required for DGA fleet management while improving the quality of the risk picture it produces. But like any analytical tool, its outputs are constrained by its inputs. The CSEV (Cumulative Severity) and HF (Hazard Factor) metrics at the core of R-DGA methodology [1] are history-integrating: they improve in reliability and diagnostic power as the historical record behind them grows.

This means that the practices governing how data enters and is maintained in TOA directly affect the quality of the fleet risk picture it generates. This guide covers the practical steps that maximise analytical value.

Setting Up the Transformer Database

A well-structured transformer database is the foundation of useful TOA operation. Each transformer record should include the fields that drive accurate DGA interpretation and enable meaningful fleet comparison.

Core identification fields. Asset ID, name or designation, location (substation, voltage class, bay), manufacturer, serial number, in-service date, and rated MVA. These fields link TOA records to the asset register and enable results to be reported in operational terms.

Interpretation-critical fields. Two fields significantly affect how DGA results are interpreted and compared:

• Oil type: mineral oil, natural ester, synthetic ester, or silicone. The thresholds and fault classification methods appropriate for each fluid differ [2]. Applying mineral oil interpretation to ester-fluid DGA results produces incorrect assessments.
• Oil preservation system: conservator (open to atmosphere), sealed tank, or nitrogen blanket. The preservation system affects how CO and CO₂ equilibrate in the oil, influencing carbon gas interpretation. Conservator-type transformers lose CO₂ to the atmosphere and typically show lower CO₂/CO ratios than sealed units.

Additional fields that improve analysis include transformer design type (core vs. shell), cooling class (ONAN, ONAF, OFAF), oil volume, and winding configuration.

Populating these fields before analysis begins prevents systematic interpretation errors. A transformer flagged as mineral oil in TOA that is actually filled with natural ester will receive mineral oil threshold comparisons that overcount false positives on CO and CO₂.

Importing Historical DGA Data

The single highest-value data management step for most new TOA implementations is importing historical DGA records from previous systems, including spreadsheets, legacy software, paper records, or laboratory archive files.

The reason this matters: CSEV is a history-integrating metric [1]. For a transformer with only two recent samples, CSEV reflects only the fault severity visible in those two data points. For a transformer with ten years of quarterly samples, CSEV reflects the cumulative fault history visible across 40 data points, a materially more reliable and informative severity estimate. The longer the history, the more the CSEV smooths out individual measurement variability [3] and the more accurately it characterises the transformer's cumulative risk profile.

A transformer with a 20-year DGA history in TOA produces a population-calibrated risk assessment grounded in two decades of condition data. The same transformer with two recent samples produces an estimate with substantially wider uncertainty. Importing historical data is not a cosmetic improvement. It directly improves the accuracy of the risk assessment that informs maintenance decisions.

Delta-X Research provides data import support for common formats, including laboratory report imports and legacy software exports. Contact us if you have historical data that needs to be converted.

Sample Entry Discipline

The ongoing data quality that sustains reliable fleet analysis depends on consistent sample entry practice.

Enter results promptly. Laboratory results should be entered into TOA within a few days of receipt, not accumulated for batch entry monthly or quarterly. The fleet ranking view is only current as long as the underlying data is current. An HF value computed from a sample six months ago may no longer reflect the transformer's actual condition.

Use collection date, not analysis date. DGA laboratories typically process samples several days after collection; some have backlogs of one to two weeks. Recording the laboratory analysis date as the sample date introduces a systematic error into trend analysis. The dissolved gas concentrations reflect the transformer's condition at the time of oil collection, which is what matters for trend calculation.

Record the laboratory and analytical method. ASTM D3612 [4] defines multiple extraction methods that produce quantitatively different results for the same oil sample. Interlaboratory variability is documented and can be sufficient to shift results across condition boundaries [3]. Recording the laboratory source alongside each sample result enables informed interpretation of apparent trend changes that coincide with laboratory changes.

Note unusual sampling circumstances. If a sample was collected immediately after a fault event, after recent oil processing, after a laboratory error required resampling, or from a non-standard sampling point, note it. These circumstances affect interpretation and should be visible when reviewing the trend record. A spurious high reading that was actually a sampling error should be flagged, not silently included in the trend analysis.

The Fleet Ranking Review Workflow

Once the database is populated and current, the fleet ranking view in TOA provides the most operationally useful output: a ranked list of every transformer in the database by R-DGA severity metrics, updated each time new results are entered.

A practical monthly or post-batch-entry review workflow:

Sort the fleet by HF. Transformers in the upper quartile by HF are the current-activity priority: they are generating fault gases at rates that are elevated relative to their cumulative history and the population baseline [1]. Review the individual trend chart for each to confirm the context of the elevated HF: is this a new development or a sustained pattern?

Review the CSEV/HF scatter. Transformers with both high CSEV (long-term severity history) and elevated HF (current activity rate) represent the most concerning segment of the fleet. These are the units where both the cumulative record and the current trajectory indicate above-population risk.

Flag significant changes from the previous review. A transformer that was in the lower half of the fleet ranking last month and is now in the upper quartile has experienced a significant change in its condition trajectory. Regardless of whether any threshold has been crossed, a trajectory change of this magnitude warrants investigation.

Schedule follow-up for flagged units. The output of the fleet review is a prioritised list: units for accelerated resampling, units for on-site inspection, and units for escalation to detailed diagnostic testing. This list should flow directly into the maintenance planning process.

This workflow can be completed in less than an hour for a fleet of several hundred transformers, substantially less time than reviewing individual transformer reports sequentially, and it produces a more consistent and defensible prioritisation output.

For implementation support, data import assistance, or training on TOA fleet analysis workflows, contact us. For a full overview of TOA's capabilities, visit the TOA page.