VIE and Lab DGA: The Case Against Online DGA for Fleet Monitoring
The transformer industry has had continuous gas monitoring technology since the 1990s. In that time, it has been deployed only on a small fraction of the transformers that need it. The economics make fleet-wide deployment nearly impossible: online DGA systems cost $8,000 to $30,000 per unit to install, plus ongoing maintenance. A utility managing 500 transformers could spend up to $15 million equipping the entire fleet before paying a single annual maintenance bill. In practice, utilities deploy online DGA on their highest-risk assets and leave the rest unmonitored.
That is fleet-level blindness, a condition the industry has accepted for decades. VIE exists to close it. The combination of the VIE solution with lab DGA far exceeds the capabilities of online DGA.
The Fleet-Wide Problem Online DGA Cannot Solve
Online DGA is a point solution. It monitors the transformer it is installed on. Every additional transformer requires another installation, another hardware cost, another maintenance contract, another calibration cycle, and more trained eyeballs to analyze the data. The economics do not scale.
VIE's installation is non-invasive and takes one technician under 30 minutes on most transformers, under an hour on the largest units. Sensors bond to the outside of the tank wall (the number scales with transformer size: 3 to 4 sensors on units under 10MVA, 5 to 7 on units between 10 and 100MVA, and 6 to 9 on units above 100MVA). No penetration of the oil system. No outage. No specialized installation team.
Fleet-wide coverage is not a feature. It is the point. A monitoring program that covers your 15 highest-risk transformers and leaves the other 485 invisible has not solved fleet-level blindness. It has documented it.
Online DGA looks backwards. VIE looks forward.
This is the operational distinction that matters most at scale.
Online DGA flags thresholds. Some newer systems can raise an alert automatically when a gas concentration crosses a set level. But that is still a threshold-based response, and it misses the most consequential risk. The greatest danger does not come from the transformers whose gas readings exceed thresholds. It comes from the transformers performing well within those thresholds that fail suddenly when stressed by an extreme weather event or a high load surge. Those are the invisible ones, and online DGA cannot see them.
Installing an online DGA system requires a physical connection to the transformer's oil system (typically through a sampling port or valve on energized equipment). That penetration point must be maintained and resealed over the life of the installation. Calibration and servicing require the same on-site access, on an ongoing basis.
VIE is autonomous. Sensors bond to the outside of the tank wall. There is no penetration of the oil system. There is no new access point to maintain. There is no personnel exposure to energized oil connections. VIE's platform processes vibration, thermal, oil quality, and weather data continuously and surfaces the assets that need attention. Weather conditions have the greatest impact on a transformer's health, and VIE accounts for them at every monitoring interval. For substations with limited access windows, offshore platforms where any personnel deployment is a logistical event, or industrial facilities with classified area restrictions, that autonomy is not a convenience. It is what makes continuous fleet-wide monitoring operationally viable.
What Online DGA Monitors, and What It Misses
Online DGA monitors gases produced by fault conditions. It detects thermal degradation in oil, thermal degradation in paper, partial discharge precursors, and arcing. For high-voltage, heavily loaded transmission assets where gas trending provides lead time, there is diagnostic value in that continuous signal.
But, by the time certain gases appear, the fault is already in progress. Acetylene, for instance, is a consequence of active arcing: it shows up after the event, not before it. Online DGA tells you something has happened. It cannot tell you winding stress is accumulating, core laminations are degrading, or oil quality is declining before those conditions reach the gas-generating threshold.
VIE monitors the failure modes that precede gas formation: winding mechanical integrity through the Radial and Axial Winding Health Metrics, oil quality continuously through V2P and S2P without sampling, thermal behavior through Excess Heat Flux metrics at multiple sensor heights, and partial discharge patterns through vibration anomaly analysis. All five failure mode categories, continuously, on every transformer in the fleet simultaneously.
Online DGA covers one failure mode channel on the transformers it is installed on. VIE covers five failure mode channels on every transformer in the fleet.
Lab DGA Feeds Back Into VIE: This Is the Integration That Matters
Here is the argument that online DGA cannot make, and the one that changes how you should think about lab DGA entirely.
A lab DGA report, read in isolation, tells you the gas concentrations in one transformer's oil at one point in time. A domain expert studying that report can flag threshold crossings and characterize fault types. What the expert cannot do is combine that gas data with months of continuous vibration signatures, thermal profiles, oil quality trends, and local weather history for that specific transformer — and extract the patterns that sit beneath the threshold, in the data between the readings, where the real predictive signal lives.
VIE's AI does exactly that. Every lab DGA result that enters VIE's system is synergistically combined with the continuous data VIE has been collecting to generate insights that no expert could derive from the lab DGA data alone. The AI identifies correlations across multiple data streams simultaneously: rising vibration amplitude and a specific gas ratio trending together, a thermal pattern shifting in the weeks before a lab test flags a change, weather-driven load stress appearing in the vibration signature before it registers in the chemistry. These are not relationships a human analyst can reliably detect by looking at a gas report. They are relationships VIE's AI was built to find.
And the model gets more accurate with every additional data point. Each lab DGA result tightens VIE's understanding of that specific transformer. The accuracy of predictions compounds. The second year of monitoring is more precise than the first because VIE has a year of lab results, field observations, and continuous sensor data to build on. The third year is more precise than the second.
Online DGA produces gas concentration data on the transformers it is installed on. It does not account for weather or load changes, the two most important variables impacting a transformer's health. It does not update its model when new information arrives. It does not learn from MEGGER results, field observations, or inspection findings. VIE integrates all of it, continuously, across the entire fleet.
That is not two monitoring programs running in parallel. That is an intelligence system that outgrows the limitations of any single data source.
The Operational Math
Online DGA: $8,000 to $30,000 per unit installed, $500 to $1,500 per year in maintenance. Covers one transformer. Invasive. Threshold-based. Does not scale. Does not learn.
VIE: Non-invasive. Fleet-wide. Autonomous. All five failure mode categories, continuously. Lab DGA run on VIE's asset-specific recommendation, at $150 to $400 per sample, targets the assets that need it, and every result makes VIE's next prediction sharper.
The question is not whether online DGA is useful in isolation. It is whether an invasive, point-solution, threshold-based program deployed on a fraction of the fleet produces more fleet intelligence than VIE plus targeted lab DGA across the entire fleet.
The answer is not complicated. The assets that fail unexpectedly are rarely the ones that were being watched. And with online DGA, most of the fleet is not being watched at all.