Two architectures for online DGA
When a site has multiple oil-filled transformers to monitor, the deployment question is structural: give every transformer its own analyzer (single-channel, 1:1), or let several tanks share one analyzer (one-to-many, multi-tank shared). Both are engineered solutions; they answer different economics. The whitepaper compares them as an explicit engineering trade-off and details the anti-cross-contamination controls that make the shared architecture trustworthy.
Single-channel (1:1) vs one-to-many
| Comparison | Single-channel (1:1) | One-to-many (shared) |
|---|---|---|
| Measurement independence | Each unit measured independently and continuously | Shared unit; sequential switched measurement |
| Accuracy and reliability | Optimal; no switching link | Affected by switching valves, flushing sequence, cross-contamination |
| Average measurement cycle | Short per-unit cycle | Approximately single-unit cycle × number of units |
| Sudden-fault response | Fast; independent alarm per unit | Delayed for a unit not yet polled |
| Per-asset cost | High | Low; amortized across units |
| Maintenance | Points grow with unit count | Fewer points, but a single point of failure affects all connected units |
| Best fit | Critical assets — UHV, core main transformers | Many units, moderate per-unit value — e.g., four main transformers at a hydropower plant |
The measurement-cycle arithmetic
The most direct cost of sharing is time. A single analyzer can measure one tank at a time, so with N units the average cycle for each channel is approximately the single-unit cycle multiplied by N. A unit that would be seen hourly on its own analyzer is seen every few hours in a shared configuration — and a sudden fault on a unit that was not the last polled waits for its turn.
That arithmetic drives the engineering recommendation: for critical assets with high value, high failure cost, and a continuous-monitoring requirement — UHV substations, converter-station main transformers — adopt single-channel 1:1 to preserve independence and response speed per unit. For many units of moderate per-unit value, one-to-many significantly reduces per-asset cost.
The cross-contamination problem
The core technical risk of one-to-many is inter-channel cross-contamination: residual oil or gas sample from the previous tank remains in the piping, valves, degassing chamber, and detection cell, and bleeds into the next tank’s measurement. The result is false elevations and distorted trends — data that looks plausible but is wrong.
Five combined countermeasures are applied in engineering practice:
- Switch-and-flush — after channel switching, the new tank’s oil flow displaces the piping and degassing chamber; flush duration is set longer than the time to displace several chamber volumes.
- Gas-path purge — the detection gas chamber is evacuated or purged between channels to reduce gas-path memory.
- Low-adsorption materials — piping and valve bodies use materials with low adsorption of characteristic gases, avoiding “adsorption-release” tailing.
- Data timestamping — every value is stamped with channel and switching timestamps; the host computer discards switching-transition data and adopts only the stable segment.
- Channel priority — measurement frequency increases for deteriorating or higher-voltage units and decreases for stable units, allocating the limited measurement resource to risk.
These controls are not optional extras; they are the difference between a shared analyzer that reports useful trends and one that reports artifacts. CIGRE TB 296 makes the general point that interpretation quality is directly related to the controllability of the sampling stage.
Specifying one-to-many in procurement
If a shared architecture is chosen, the engineering rules should be written into the technical specification rather than left to the vendor. The whitepaper’s procurement guidance says to specify the switching and flushing sequence, the cross-contamination control metrics, and the maximum number of connected units — and to evaluate the resulting polling cycle against the fault-detection requirements. Regardless of the scheme, oil-gas separation efficiency and response time (T63) should be part of acceptance.
PAS DGA for shared and dedicated monitoring
PAS DGA supports both architectures. For critical main transformers, a dedicated DGA-900 9-gas plus moisture monitor delivers the continuous, independent data a UHV or converter-station unit deserves. For unit groups where cost rules — multiple generator step-up transformers at a hydro plant, step-up substations across a renewable fleet — a shared configuration with proper flushing and purge sequences is the economical path.
The choice is an engineering trade-off, not a technology one. The full deployment economics are covered in our 10-year TCO comparison, and the selection framework in how to select an online DGA monitor walks through the seven dimensions. Contact PAS DGA to model both configurations against your fleet.