Capacitor exploding in a paper plant

Question

Two years ago I was in an internship in a paper plant in the Superintendence of Electrocontrol, in the area of maintenance of electrical equipment. Well, they had a problem regarding some capacitor banks that were installed to mantain the power factor at a certain level. The system was simple, it had multiple capacitor banks that were switched to change the reactive power and control the power factor, and they did have many switchings over a day. The problem they had is that, capacitors would not last long, they kind of exploded on a month basis, and this was expensive due to the maintenance and the need of more capacitors to replace the useless ones. I am kind of puzzled, because it was a 3-phase system, and the capacitors nominal voltage was higher than the voltage of the source they were connected (380 line voltage, and capacitors were approximately 600 V)

My question is, why did capacitors "explode" so quickly? They should have lasted at least 3 - 6 months (1 year expected). I was thinking of over-voltage due to the transients when switching and producing like a "chain-reaction", less capacitors => higher voltage transient. However there is long time since I have not solved a circuit problem, could someome explain it in more mathematical terms? Am I correct in my statement?

Answer 1

I would suspect harmonic distortion of the system voltage due to nonlinear loads. The capacitors are sized to carry current due to the complex impedance of the capacitors at the line frequency. If the actual voltage waveform has significant content of voltages at harmonic frequencies, that increases the capacitor current because the impedance is lower at those frequencies. In addition, harmonic distortion can precipitate a harmonic resonance situation further increasing the capacitor current.

The nonlinear loads can be electronic motor speed controls or other power electronic equipment on the three-phase distribution system or even electronic equipment on single-phase distribution systems that are fed from the three-phase systems.

Many large motors in paper plants are powered through electronic power conversion speed controls.

Answer 2

The most likely cause of failure in this case is that the capacitor got too hot. Like anything, capacitors get hot when current flows through them. The amount of current flow depends on the voltage and the complex impedance of the cap at the line frequency. In the simplest possible analysis, RMS current will be \$V_{line} \times 2 \pi f C\$. Typically, capacitors used for this sort of application will have an AC voltage spec, which takes all that into account.

Of course, you may have multiple capacitors in parallel which are not sharing current properly. In that case the one with the lower impedance (higher capacitance) dominates.

Or there might be other effects contributing to capacitor currents besides the line voltage. I've heard of cases where resonant effects between different capacitors in a plant expose them to higher currents than they would otherwise be expected to see. I've even heard of cases where the failure of one capacitor changed the resonance, making the problem go away.

Until someone replaced the blown capacitor, of course...

Answer 3

Capacitors can fail early if their ripple current rating and ambient operating temperature rating is being violated.

If the ripple current rating for these 600V rated caps was 20A, but they were seeing peak AC current ripples of 30-40A (unit magnitudes are examples only), then the capacitors will wear down much faster than what they are rated for due to internal temperature rise thanks to ESR (equivalent series resistance).

Ambient temperatures being high will make this even worse.

The reality of contact resistance and parasitic inductance means no capacitor will share current exactly the same. Some may bear just above their ripple current rating, while others down the chain (if you model parallel caps as a chain connected by parasitic resistances in the contacts and wires between them) have far less. The ones closer to the input source will charge and discharge current more than the others, and will fail faster due to the above reasons.

I also found a useful document on PF correction capacitor banks and their failure modes here.