Fan bank failed, top oil rising, and 35 ppm of acetylene
A 1000 kVA, 11 kV/433 V, Dyn11 oil-immersed distribution transformer serves a traction maintenance depot. The nameplate states ONAN 700 kVA, ONAF 1000 kVA, guaranteed top-oil temperature rise 45 K at rated load, oil-temperature alarm at 90 °C and trip at 105 °C. Site ambient is 44 °C in the afternoon. One of the two ONAF fan banks has been out of service for three days after a control fuse blew and was not replaced. The continuous load log shows 640 kVA. The top-oil thermometer reads 88 °C and the winding temperature indicator 96 °C. A dissolved gas analysis sample shows H2 130 ppm, CH4 55 ppm, C2H4 60 ppm, C2H2 35 ppm, CO 400 ppm and CO2 3500 ppm; the sample taken six months earlier had C2H2 below 1 ppm. The Buchholz relay has collected a small volume of gas. Insulation resistance and winding resistance are normal, and there is no audible discharge. The depot load cannot easily be shed.
1. Assess the thermal condition of the transformer: what rating applies with one fan bank out, how much temperature margin exists at 640 kVA, and does 88 °C by itself justify emergency shutdown?
Read the worked answer
Rating with cooling degraded. ONAF 1000 kVA is only valid with both fan banks in service. With one bank out, the defensible figure for a loading decision is the ONAN rating of 700 kVA; if the banks are identical, half the forced-cooling increment would give roughly 850 kVA, but that is an estimate, not a guaranteed rating. Load = 640 kVA = 91% of the ONAN rating, so the transformer is inside its ONAN capability and the load itself is not the cause of the temperature.
Temperature margin. Rise = 88 − 44 = 44 K, against a guaranteed 45 K at rated load. At 0.91 pu load a healthy unit would be a few kelvin below 45 K, so the unit is running at or above its guaranteed rise - zero margin, consistent with restricted cooling. But 88 °C is 2 K below the 90 °C alarm and 17 K below the 105 °C trip.
Judgement. Thermally this is 'no margin', not 'emergency'. Temperature alone does not justify an immediate shutdown: replace the fan-bank fuse, confirm oil and winding temperatures fall, and hold load at or below the ONAN rating. The decisive evidence for urgent action comes from the oil, not from the thermometer.
2. Interpret the dissolved gas analysis and state whether the transformer should be kept in service. Justify your reasoning using what each gas indicates.
Read the worked answer
Gas-by-gas reading.
- C2H2 (acetylene) 35 ppm, up from under 1 ppm in six months, is the key number. Acetylene is formed only at very high local temperatures - of the order of 700 °C and above - that is, by arcing. It is not a normal ageing product.
- H2 130 ppm accompanies partial discharge or arcing; hydrogen alone is the classic partial-discharge gas.
- CH4 55 / C2H4 60 ppm indicate a thermal component, but at levels far too low to explain the acetylene.
- CO 400 / CO2 3500 ppm show only mild cellulose involvement - no significant paper damage yet.
- The Buchholz gas accumulation confirms that gassing is active, not a one-off artefact of sampling.
Rate of change matters as much as the absolute value. A jump from below 1 ppm to 35 ppm in six months is a fast-developing arcing fault.
Decision: do not leave it in service on the strength of a 'normal' temperature. A unit can show an acceptable top-oil temperature while a local arc destroys insulation, because the fault energy is concentrated and the oil circulation dilutes the heat. Arrange load transfer to the standby or adjacent transformer, take a confirmatory sample at 24 hours, then isolate for ratio analysis, a separate tap-changer oil sample, internal inspection and a repeat DGA before any return to service.
3. The depot engineer argues the acetylene may be coming from the on-load tap changer and wants to keep running. How would you test that hypothesis, and what minimum precautions would apply if the load genuinely cannot be transferred today?
Read the worked answer
Testing the tap-changer hypothesis. Most OLTC diverter chambers have their own oil, isolated from the main tank. Drain and analyse the OLTC oil separately: if C2H2 is concentrated there and the main-tank oil is clean, the arc is in the diverter contacts and the main winding insulation is not implicated - the transformer can return on a locked tap after the OLTC is overhauled. If C2H2 is also present in the main tank, or the design is a shared-oil tap changer, the hypothesis fails and the unit must come out of service. Also check tap-change counters and recent through-fault records; frequent tap changing under load produces diverter arcing as normal wear.
If load truly cannot be transferred today:
- Restore both fan banks before anything else.
- Hold load below the ONAN rating (600 kVA or lower), not the 640 kVA now carried.
- Keep top oil below the 90 °C alarm with margin; log oil and winding temperatures hourly.
- Take DGA samples at 12-hour intervals and watch the acetylene trend; a continuous gas monitor is better if one is available.
- Confirm Buchholz, pressure relief, fire protection and oil bunding are functional, and keep the unit on a documented emergency concession with a fixed deadline of a few days, not weeks.
This is risk acceptance under control, not a repair.