Neutral overheating on a harmonic-rich 400 kVA transformer with power-factor correction
A 400 kVA, 11 kV/400 V Dyn11 transformer feeds a mixed office and small data-centre building. The load is dominated by switch-mode power supplies, LED drivers, electronic ballasts and inverter-driven chillers; the building also has a 100 kvar automatic capacitor bank for power-factor correction. During a thermal survey at about 80 percent load, the three phase currents read 372 A, 365 A and 358 A, and the neutral current reads 340 A. The neutral bar and neutral tails in the main board reach 96 degrees C by infrared, while the phase conductors sit at 62 degrees C. Phase current THD is 31 percent and neutral current THD is 71 percent, with a dominant third harmonic. The transformer neutral is rated at 100 percent of phase current. The client reports occasional unexplained reboots of IT equipment and a capacitor-bank fuse that has blown twice this year.
1. Explain why the neutral carries almost as much current as a phase conductor despite nearly balanced loading, and why the neutral runs hotter than the phases.
Read the worked answer
The load is balanced in fundamental current, so the fundamental components of the three phase currents sum to zero in the neutral, as expected. The problem is the harmonic content.
Triplen harmonics - the 3rd, 9th, 15th and so on - are zero-sequence in a four-wire system: in all three phases they are displaced by 3 x 120 = 360 degrees, that is, they are in phase with each other. Instead of cancelling in the neutral, they add arithmetically. With a 31 percent phase THD dominated by the third harmonic, a neutral current approaching the phase current is entirely predictable: In approximately 3 x I3.
Why hotter than the phases? Three reasons:
- The neutral is sized at 100 percent of phase, so it has the same copper but carries a current rich in 150 Hz and higher components, where skin and proximity effects raise the AC resistance.
- The neutral bar is a single, often narrower, solid path with fewer parallel ways and poorer ventilation than the phase bars, and its joints are less often torqued.
- Neutral current is continuous in these loads even when the fundamental load is balanced, so there is no cool-down period.
The 96 degrees C reading is above the safe continuous rating of PVC-insulated tails and of many bar joints; the reboots are consistent with a rising neutral-to-earth potential as joints heat and oxidise.
2. Assess whether the transformer is overloaded and state what you would recommend, with numbers.
Read the worked answer
The fundamental loading looks acceptable: 400 kVA at 400 V corresponds to 577 A per phase, and the measured 372 A is 64 percent of that. That is misleading.
Harmonic currents increase winding eddy-current losses in proportion to the square of the frequency, so the transformer's effective capacity falls as THD rises. With a phase THD of 31 percent and a third-harmonic-dominated spectrum, a conventional transformer of this type should be derated to roughly 70-80 percent of nameplate for continuous duty, i.e. an effective capacity around 280-320 kVA. The thermal survey and the 31 percent THD indicate the unit is operating near that limit, not at 64 percent of it.
Recommendations:
- Carry out a proper harmonic study (a week-long power-quality log with K-factor and neutral current recording) before any purchase decision.
- If the spectrum is confirmed, replace or supplement with a K-rated transformer (K-13 or K-20 as the study indicates), or accept a documented derating.
- Double-size the neutral from the transformer to the main board and from the board to the distribution points, and re-torque the neutral joints to the manufacturer's figure with a calibrated torque wrench.
- Consider a zig-zag or active harmonic filter at the main board to strip the triplen components before they reach the transformer.
- Measure neutral-to-earth voltage at the IT racks; more than a few volts explains the reboots.
3. The capacitor-bank fuse has blown twice. Explain the mechanism and state what must not be done.
Read the worked answer
The capacitor bank plus the source impedance (transformer leakage reactance plus supply inductance) forms a parallel resonant circuit. Its resonant frequency is:
fr = f1 x sqrt(Ssc / Qc)
With a 100 kvar bank on a 400 kVA transformer of, say, 5 percent impedance (Ssc = 8 MVA), fr is in the region of 50 x sqrt(80) = 447 Hz, close to the 7th harmonic at 350 Hz and the higher-order spectrum of switch-mode loads. When a harmonic current is injected near resonance, the impedance seen by that harmonic is high and the harmonic voltage and circulating current are amplified several times. The capacitor carries a current proportional to frequency, so at the 7th harmonic its current is seven times the fundamental value for the same voltage, and the fuse ruptures.
What must not be done:
- Do not simply fit a bigger fuse or a bigger capacitor bank. A larger bank lowers fr further, possibly towards the 5th harmonic at 250 Hz, making the amplification worse and putting the transformer and cables at risk.
- Do not remove the bank to "cure" the harmonics and leave the power factor penalty unmanaged; the harmonic distortion will remain, and the neutral overheating will not improve.
The correct measure is a detuned (reactor-protected) capacitor bank with a tuning factor of about 7 percent, which places the resonant frequency below the 5th harmonic, or an active filter. Verify the capacitor rating for the harmonic-rich voltage and confirm the bank's contactors and fuses are rated for capacitor duty.