Rising Acetylene in a 10 MVA Distribution Transformer: De-rate, Monitor or Withdraw?
A 10 MVA, 33/11 kV, Dyn11 distribution transformer commissioned in 2009 supplies an industrial feeder. Peak demand has grown from 6.5 MW to 8.6 MW at 0.90 power factor, so the unit runs near nameplate for about five hours every evening. Top-oil temperature at peak is 78 °C at 38 °C ambient. Last week's oil sample showed BDV 52 kV and moisture 22 ppm. Three dissolved gas analyses, six months apart, give: H2 90/130/170 ppm, CH4 40/48/55 ppm, C2H4 110/140/175 ppm, C2H6 30/34/38 ppm, C2H2 0/3/9 ppm, CO 320/340/360 ppm, CO2 2400/2500/2650 ppm. The on-load tap changer sits in a separate oil compartment; it was sampled the same day and showed C2H2 15 ppm with no other significant gas. Furan is 0.09 ppm. Because a single-source critical load depends on this transformer, the utility can offer a shutdown only after 45 days. The DGM asks whether the unit can be kept at present loading through the peak season.
1. Interpret the three DGA results, in terms of fault type, severity and rate of change, and state whether you will permit operation at the present 8.6 MW peak for the next 45 days. Justify your answer with the numbers.
Read the worked answer
Gas picture. Total dissolved combustible gas is about 807 µL/L (170 + 55 + 175 + 38 + 9 + 360). The hydrocarbon pattern is ethylene-dominated (C2H4 is roughly 76% of the hydrocarbon sum) and acetylene has appeared and is rising, 0 to 3 to 9 µL/L. That is a high-temperature thermal fault with incipient arcing; on the Duval triangle the point falls in the thermal region near the T2/T3 boundary. The solid insulation still looks healthy: CO2/CO is about 7.4 and stable, furan is 0.09 ppm, moisture 22 ppm and BDV 52 kV are acceptable for a 33 kV unit.
Why present loading is not defensible. 8.6 MW at 0.90 power factor is 9.55 MVA, about 0.96 pu, with top oil already 78 °C at 38 °C ambient. A hot spot that is generating acetylene is being pushed harder for five hours every evening. Forty-five more days at 0.96 pu is an unjustified risk to a single-source load.
Decision. Keep the unit in service but cap it at about 0.80 pu (8.0 MVA, roughly 7.2 MW at 0.90 power factor), with top oil held at or below 85 °C and calculated hot spot at or below 110 °C. Withdraw immediately if acetylene exceeds about 20–25 µL/L, if it rises by more than about 5 µL/L per week, or if ethylene doubles.
2. What operating limits, monitoring and diagnostic steps do you authorise now, and what evidence would separate a tap-changer source from a main-tank hot spot?
Read the worked answer
Operating limits. Cap loading at 0.80 pu; hold top oil at or below 85 °C; verify cooling fans and pumps, oil level and radiator valves; ban overload paralleling and shed non-critical feeders before the evening peak.
Monitoring. Move from routine sampling to weekly DGA with rate-of-change trending, not just absolute limits. Add calibrated top-oil and ambient measurement, infrared scans of bushings, cable boxes and joints at peak, tap-changer operation counter and motor current, core ground-strap current, and a check for oil leakage between the diverter compartment and the main tank.
Separating the two sources. The main-tank sample cannot be explained by the diverter switch if the barrier is tight: 15 µL/L acetylene is unremarkable for a tap changer that arcs under oil, but only the main tank holds the winding. So first prove there is no leakage past the barrier (rising diverter oil level, dye test, gasket inspection). If there is leakage, the whole gas picture is contaminated and the tap changer becomes the prime suspect. If there is none, the acetylene originates in the main tank, and the likely candidates are a poor lead or busbar joint, circulating current from a core ground fault, or blocked oil flow near the winding. Also normalise the diverter gas per tap operation; a steep per-operation acetylene trend points to the reversing switch or contacts.
3. Set out the scope of the 45-day outage and the tests that must be completed before the unit returns to full rating. What contingency do you arrange for the critical load?
Read the worked answer
Outage scope. Drain and remove the tap changer for contact and reversing-switch inspection; check through-leads and barriers; inspect and re-torque all busbar and lead joints to the specified torque; verify the core ground strap and inspect for a second ground; check the internal earth connections and the condition of the oil ducts. If the tap changer is the source, overhaul or replace it; if a joint is the source, remake it and record the resistance.
Tests before return to full rating. Winding resistance with phase imbalance within 2%; ratio on every tap within 0.5% of nameplate; insulation resistance and polarisation index (PI of 2.0 or better) at 5 kV; sweep frequency response against the original fingerprint; oil processing to BDV of 60 kV or better and moisture of 10 ppm or better; a repeat DGA baseline after 24 hours at full load; and confirmation that Buchholz, differential and over-temperature protection settings match the de-rated loading plan.
Contingency. Arrange a mobile transformer or a firm load transfer to the adjacent 11 kV feeder before the shutdown, keep the critical load on a tested DG set with verified changeover, and record the risk acceptance in writing. If the inspection finds no definitive source, treat the unit as permanently de-rated and start replacement planning.