Online DGA is a monitoring approach in which a permanently installed analyzer measures dissolved gases at regular intervals without taking the transformer offline.
Short answer: offline (laboratory) DGA is still the accuracy reference and the route to standards compliance, but it gives you one data point per sampling interval. Online monitoring turns that snapshot into a continuous trend and normally sees a developing fault weeks earlier. The two are complements rather than substitutes — most operators run online monitors on their critical units and keep annual lab DGA across the rest of the fleet.
- Offline wins on accuracy, standards recognition, and cost per transformer across a large fleet.
- Online wins on detection speed, rate-of-change alarms, and remote or unmanned sites.
- Rule of thumb: monitor online where the cost of a failure exceeds the cost of the instrument; sample the remaining fleet on a schedule.
The Fundamental Choice
Every transformer asset manager faces the same decision: rely on periodic laboratory DGA testing, invest in continuous online monitoring, or implement a hybrid strategy. The right answer depends on transformer criticality, accessibility, failure consequences, and budget.
Offline (Laboratory) DGA
How it works: An oil sample is drawn from the transformer (typically annually), transported to a laboratory, and analyzed using gas chromatography per IEC 60567. Results are available in days to weeks.
| Advantage | Disadvantage |
|---|---|
| Gold-standard accuracy (GC reference method) | Snapshot — one data point per year |
| IEC/IEEE/ASTM regulatory compliance | Weeks of delay from sampling to result |
| Lower capital cost per transformer | High cumulative cost: $1,200+ per test × 30–50 years |
| Covers entire fleet affordably | Misses rapidly developing faults |
| Independent third-party data | Sampling quality variability |
| No installation required | No rate-of-change trending |
Annual lab DGA is appropriate for:
- Distribution transformers (<10 MVA)
- Transformers with no fault history and stable DGA trends
- Fleet screening — identifying which transformers need closer attention
Online (Continuous) DGA Monitoring
How it works: A monitor is permanently installed on the transformer, extracting and analyzing oil automatically at configurable intervals (typically every 1–4 hours). Results are transmitted to SCADA or asset management systems in real time.
| Advantage | Disadvantage |
|---|---|
| Continuous trending — hourly data | Higher capital cost per unit |
| Rate-of-change alarms — catches rapid faults | Requires installation (valve access) |
| Zero sampling/transport delay | Requires communication infrastructure |
| Detects faults 30–90 days before lab sampling would | Requires periodic sensor verification |
| Enables condition-based maintenance | |
| Lower total cost over 30–50 year lifecycle |
Online monitoring is recommended for:
- GSU transformers (failure takes entire plant offline)
- Transmission transformers ≥220 kV
- HVDC converter transformers
- Critical industrial transformers (steel, chemical — $M/day downtime)
- Remote/unmanned substations
- Offshore wind farm transformers
- Any transformer where failure consequential cost exceeds $1M
Cost Comparison: 30-Year Lifecycle
| Cost Element | Annual Lab DGA | Online PAS DGA |
|---|---|---|
| Capital equipment | $0 | $15,000–$40,000 |
| Installation | $0 | $2,000–$5,000 |
| Annual lab testing (1 sample/year) | $1,200 | $0 (confirmation only) |
| Consumables (carrier/calibration gas) | $0 | $0 (PAS) |
| Maintenance | $0 | $500/year (inspection) |
| 30-year total | $36,000 | $32,000–$60,000 |
The crossover: Online PAS DGA becomes cost-competitive with annual lab testing at approximately 15–20 years of operation — while providing dramatically better protection.
But the real financial case is not about DGA costs — it’s about avoided failure costs:
| Failure Scenario | Cost |
|---|---|
| Distribution transformer (10 MVA) | $100K–$500K (equipment + installation) |
| Transmission transformer (100 MVA) | $1M–$3M + outage costs |
| GSU transformer (500 MVA) | $3M–$10M + $500K/day generation loss |
| Nuclear GSU | $10M+ + regulatory impact |
A single avoided catastrophic failure pays for online DGA monitoring on an entire fleet of transformers.
The Hybrid Strategy (Industry Best Practice)
| Tier | Method | Transformers | Purpose |
|---|---|---|---|
| Online (continuous) | PAS multi-gas monitor | Critical/high-value (top 20%) | Early warning, trending, rate-of-change alarms |
| Offline (annual) | Lab GC | All transformers (100%) | Fleet screening, regulatory compliance, online data verification |
| Offline (event-driven) | Lab GC | Any transformer after alarm | Confirmation of online monitor readings |
Benefits of the hybrid approach:
- Online monitors provide 24/7 protection on the assets that matter most
- Annual lab GC covers the entire fleet cost-effectively
- Lab results validate online monitor accuracy (cross-check)
- Online trending data informs sampling strategy (increase frequency when trends change)
Data Comparison: Online vs. Lab
A 2016 IEEE/PES study compared online DGA monitors against laboratory GC across multiple technologies:
| Metric | Lab GC | Online PAS |
|---|---|---|
| C₂H₂ detection | ≤0.5 ppm | 0.1–0.5 ppm |
| Repeatability | ±3% | ±3% |
| Data points per year | 1 | 8,760 (hourly) |
| Fault detection latency | Weeks to months | 30–60 minutes |
| Rate-of-change data | No | Yes |
| Standards recognition | Full IEC/IEEE/ASTM | Growing (compliant via diagnostic methods) |
The rate-of-change advantage: A transformer developing an arcing fault might go from normal to catastrophic in 2–4 weeks. With annual sampling, there’s a 96% chance the fault is completely missed. With hourly online monitoring, the rising acetylene trend is detected within hours, providing weeks of lead time for a controlled shutdown.
The Uncertainty Caveat
Laboratory DGA is widely trusted, but it carries real measurement uncertainty — IEC 60567 estimates total laboratory analysis uncertainty at approximately ±15%. Two samples drawn from the same transformer in the same week can differ by more than the apparent gas rise, which is why a single snapshot is a risky basis for a decision.
Online monitors face the opposite challenge: excellent repeatability and trend resolution, but absolute accuracy depends on calibration and degassing efficiency. Running the two side by side resolves both weaknesses — the online monitor supplies the clean trend, and the laboratory supplies an independent absolute reference.
When Online DGA Pays for Itself: Real Cases
Case 1: 1100 MVA GSU — Arcing Fault Detected
A Serveron TM8 online monitor detected acetylene rising from 0 to 19 ppm over 48 hours on a nuclear plant GSU transformer. The transformer was taken offline in a controlled 62-hour outage for repair. Without online monitoring, the fault would likely have progressed to catastrophic failure — estimated cost avoidance: $3.2 million.
Case 2: Steel Mill Arc Furnace Transformer
A PAS-DGA monitor on a 120 MVA arc furnace transformer detected a rapid C₂H₄ increase indicating a developing T3 thermal fault. The transformer was switched to a spare during a scheduled maintenance window. Without monitoring, the fault would have caused an unplanned outage during production: estimated cost avoidance: $2.1 million (4 days lost production).
Case 3: Offshore Wind Farm — 220 kV Export Transformer
Remote monitoring via PAS-DGA detected rising H₂ and CH₄ trends on an offshore platform transformer. A service visit was scheduled during a weather window, and a loose core ground connection was repaired. Without monitoring, the fault would have required an emergency helicopter mobilization: estimated cost avoidance: $500,000+.
Decision Framework
| Question | Yes → | No → |
|---|---|---|
| Is transformer failure consequential cost >$1M? | Online DGA recommended | Consider offline |
| Is the transformer >100 MVA or >220 kV? | Online DGA recommended | Consider offline |
| Is the site unmanned or remote? | Online DGA recommended | Consider offline |
| Are there >5 years of stable DGA history? | Offline may be sufficient | Online provides baseline |
| Is the transformer >20 years old? | Online DGA recommended (aging risk) | Consider offline |
| Is there existing SCADA/comms infrastructure? | Online integration easier | Factor comms cost into decision |
When to Upgrade from Lab-Only to Online
Three triggers justify adding online monitoring to a laboratory-only program:
- An aging or high-value asset where the interval between laboratory samples is too long for the risk — for example, a transformer approaching end of life with rising baseline gases.
- An identified problem — a unit with a known gassing trend needs continuous watching, not another quarterly data point.
- An unmanned or remote site where scheduled sampling is logistically expensive and response times are long.
The full economics — consumables, staffing, and the payback point — are covered in our guide on upgrading from laboratory to online DGA.
Next Steps
- Product Selector — Find the right PAS DGA monitor for your transformers
- DGA Technology Comparison — Which technology fits your application?
- ROI Calculator — Estimate your cost avoidance with online DGA monitoring
How to combine online and offline DGA in one monitoring strategy
- Step 1: Establish a laboratory baseline — run an offline DGA (oil sample) at commissioning and at major maintenance events.
- Step 2: Install the online monitor on critical transformers — set its measurement interval from 1 to 24 hours so it trends between lab samples.
- Step 3: Correlate the two sources — compare the online reading with the next lab sample to confirm agreement.
- Step 4: Let online drive daily decisions — use the continuous trend for early alarms and rate-of-change analysis.
- Step 5: Keep offline sampling for confirmatory and compliance checks — the lab remains the reference, while online fills the gaps.
FAQ
Q: What is the difference between online and offline DGA?
A: Offline (laboratory) DGA involves manually collecting an oil sample and sending it to a laboratory for analysis, typically once per year. Online DGA uses a permanently installed monitor that automatically extracts and analyzes gases at programmable intervals (hourly to daily).
Q: What is the difference between hydrogen-only and multi-gas monitoring?
A: Hydrogen-only monitors (DGA-200/300/500) detect H₂ as a universal early-warning gas — they catch ~90% of developing faults at lower cost. Multi-gas monitors (DGA-900) measure all 7+ fault gases, enabling full fault-type identification via Duval Triangle, IEC ratios, and Rogers Ratio — they tell you not just that there is a fault, but what kind and where.