Power Factor Correction Capacitors Can Make Your Harmonics Problem Worse

The client installed a 200 kVAr capacitor bank to fix a 0.72 power factor. Three months later they had nuisance tripping on six distribution boards, overheating on the capacitor bank contactors, and a transformer that hummed louder than it had before the "fix."

The power factor had improved — to 0.95. The utility penalty charges had disappeared. By every metric on the electricity bill, the project was a success. But the installation was now in resonance at the 5th harmonic, and the capacitor bank was amplifying harmonic currents that the system had previously tolerated.

This is not a rare or exotic failure mode. It happens routinely in commercial and industrial installations where variable speed drives, LED lighting, UPS systems, and other non-linear loads generate harmonic currents. Installing capacitors without considering harmonics is like adding a megaphone to a room where someone is already shouting.

Every electrical system has a natural resonant frequency determined by the interaction between the system's inductive impedance (primarily the supply transformer) and any capacitance connected to the network. The resonant frequency is:

f_r = f₁ × √(S_sc / Q_c)

Where f₁ is the fundamental frequency (50 or 60 Hz), S_sc is the short-circuit power at the point of connection (in kVA), and Q_c is the capacitor bank rating (in kVAr).

If this resonant frequency lands on or near a harmonic frequency present in the system, the impedance at that frequency drops dramatically. Harmonic currents that were previously limited by the system impedance are now amplified. The capacitor bank acts as a low-impedance path for harmonic currents, absorbing far more current than its fundamental-frequency rating would suggest.

The result: capacitor overheating, fuse blowing, contactor welding, and — in severe cases — capacitor failure with audible bangs and the distinct smell of dielectric fluid.

In most industrial installations, the 5th harmonic (250 Hz at 50 Hz systems, 300 Hz at 60 Hz) is the dominant harmonic component. It is produced by all six-pulse rectifier loads, which includes the vast majority of variable speed drives, UPS systems, and DC power supplies.

Here is why the 5th harmonic is particularly dangerous in the resonance context: consider a 1000 kVA transformer with a 5% impedance, connected to a 200 kVAr capacitor bank on a 50 Hz system.

S_sc = 1000 / 0.05 = 20,000 kVA

f_r = 50 × √(20,000 / 200) = 50 × 10 = 500 Hz

500 Hz is the 10th harmonic. No immediate problem — 10th harmonic content is usually very low. But what if the utility short-circuit level changes (as it does seasonally), or if the capacitor bank is only partially energised? With only 50 kVAr of the bank connected:

f_r = 50 × √(20,000 / 50) = 50 × 20 = 1000 Hz

Still safe. But with the full 200 kVAr and a weaker supply (say, one transformer out of service, S_sc drops to 5,000 kVA):

f_r = 50 × √(5,000 / 200) = 50 × 5 = 250 Hz

That is exactly the 5th harmonic. The system is now in resonance with the strongest harmonic component in the network. This is when things go wrong.

Harmonic resonance does not always announce itself with a dramatic failure. Often it builds gradually, and the symptoms are attributed to other causes:

• Capacitor fuse blowing: If the capacitor bank fuses blow repeatedly without an obvious overcurrent event, harmonic currents are the likely cause. The RMS current through the capacitors exceeds their rating due to harmonic absorption.
• Transformer overheating: Harmonic currents cause additional losses in the transformer windings (I²R losses increase with harmonic frequency squared for eddy current losses). A transformer that ran cool before the capacitor installation now runs hot.
• Audible noise: The transformer and capacitor bank produce audible tones at the resonant frequency. A 250 Hz hum is distinctly different from the normal 100 Hz (2nd harmonic of 50 Hz) transformer hum.
• Nuisance tripping: Circuit breakers and RCDs trip without apparent overload. The harmonic currents flowing through the system create voltage distortion that affects sensitive protective devices.
• Neutral conductor overheating: Triplen harmonics (3rd, 9th, 15th) sum in the neutral conductor rather than cancelling. If the capacitor bank shifts the resonant frequency near a triplen harmonic, neutral currents can exceed phase currents.

The standard solution to harmonic resonance in capacitor banks is the detuned reactor. A series reactor is connected in series with each capacitor step, deliberately shifting the resonant frequency of the capacitor-reactor combination below the lowest significant harmonic.

The detuning factor (p) is expressed as a percentage, representing the ratio of the reactor's impedance to the capacitor's impedance at the fundamental frequency. Common detuning factors are:

• 7% detuning: Tunes the LC combination to approximately 189 Hz (3.78th harmonic at 50 Hz). This places the resonant frequency safely below the 5th harmonic. Suitable for most installations with moderate harmonic content.
• 14% detuning: Tunes to approximately 134 Hz (2.67th harmonic). Used in installations with high 3rd harmonic content (single-phase loads, LED drivers) where even the 3rd harmonic resonance must be avoided.
• 5.67% detuning: Tunes to approximately 210 Hz (4.2nd harmonic). A compromise that provides less detuning but reduces the reactor losses and cost. Suitable only when the harmonic content is well-characterised and the 5th harmonic is the only concern.

The detuned reactor does not eliminate harmonics. It prevents the capacitor bank from amplifying them. For harmonic mitigation, active filters or passive tuned filters are needed — but that is a different problem with a different budget.

Harmonic currents affect cable sizing in two ways that are often overlooked:

• Increased RMS current: A cable carrying fundamental current plus significant harmonic content has a higher true RMS current than the fundamental alone. A power analyser measuring true RMS will show this; a simple clamp meter measuring only the fundamental will not.
• Additional losses: Harmonic currents cause additional I²R losses in the cable (skin effect increases with frequency) and additional eddy current losses in any surrounding metallic containment. Both effects increase cable temperature.

AS/NZS 3008 Table 27 and IEC 60364-5-52 provide harmonic derating factors for cables carrying significant harmonic content. These factors are often forgotten when the cable is sized for fundamental current only. A cable that passes the current-carrying capacity check at 50 Hz may be undersized when the true harmonic-loaded RMS current is considered.

When sizing cables feeding capacitor banks — especially detuned reactor banks — the cable must be rated for the total current including harmonic components, not just the fundamental reactive current.