Welcome to the first installment of our new power quality column, “Ask the Experts,” moderated by our PQ experts, Dr. Mike Lowenstein and Mark McGranaghan. If you find yourself perplexed by a power quality problem, just e-mail your question to Editorial Director John DeDad at firstname.lastname@example.org. He will get your question in front of our power quality experts, who will answer as many questions as they can each month in this column.
I'm responsible for the reliable operation of a data center and have observed that the transformer serving this center is very hot. We've also been plagued with random circuit breaker tripping. My utility has taken current measurements at the meter and found no evidence of unusual harmonic currents. As far as they're concerned, I don't have a problem with harmonics. What do you think?
Lowenstein's answer: You definitely have a harmonics problem. It's likely that your computers are connected phase-to-neutral in a 3-phase, 4-wire wye system, as shown in the Figure. This method of power distribution is attractive due to the way that return currents combine on the common neutral. If the 120V, 60-Hz currents are balanced (the loads are equal on all three phases), the return currents in the neutral will cancel and the neutral current will be zero. If loads on all three phases aren't equal, the neutral does carry return currents, but only those due to phase unbalance. For 60 Hz, the neutral current will always be lower than any of the phase currents.
However, in addition to the 60-Hz current that provides power to the load, the computer loads are all drawing high amounts of harmonic current. A computer power supply generates harmonic currents, primarily 3rd, 9th, 15th, and higher odd harmonics, with the 3rd harmonic being dominant. These harmonics are independent of the computer's make or model. The amount of harmonic current drawn by a computer can be more than 100% of the fundamental 60 Hz.
These “triplen” harmonics, which are multiples of three, are additive in the common neutral and do not cancel. For example, with 100A of 60-Hz current on each of three phases, the 60-Hz neutral current would be zero. However, with 100A of 3rd harmonic current on each phase, the neutral current would be 300A. When you have many computers connected to a wye distribution system, the neutral current due to the 3rd harmonic can easily exceed any of the phase currents.
Triplen harmonics return on the neutral conductor to the 3-phase transformer, pass through the wye secondary, and are coupled into the delta primary. Transformer theory shows that balanced triplen harmonic currents can't pass out of a delta winding. Instead, they're circulated within the winding and dissipated as heat. In other words, the primary of the transformer is now carrying not only the phase current needed to the supply secondary loads, but also the circulating triplen harmonic currents. Thus, a transformer that should, according to the loads being served, be lightly loaded, can actually be reaching overload. These high currents are responsible for your transformer heating and breaker tripping problems. Even more insidiously, the breaker serving the transformer primary never sees these currents, so the primary winding is essentially unprotected against overloading. It's this “blocking” action in the transformer delta primary that keeps the harmonic currents from being seen at the utility meter.
When you take measurements at the utility meter, you're observing only those currents that will flow through the meter. So you're not measuring the triplen harmonic currents produced by operation of phase-neutral computer loads, even though they're flowing in the secondary and neutral wires of the downstream transformer. The only harmonic currents that will be flowing at the metering point are the harmonic current pairs — 5th and 7th, 11th and 13th, and so on. Three-phase loads, such as adjustable-speed drives and large UPS systems, cause these harmonics. Thus, your utility can, in good faith, tell you there's no harmonic problem, since the problem doesn't show up where they take their measurements.
To evaluate harmonic problems caused by multiple phase-to-neutral connected computers, you must take your measurements on the secondary and neutral of the most downstream transformer serving those loads. Use a true rms ammeter with a clamp-on current probe to avoid errors. The presence of neutral currents almost equal to or exceeding the phase currents is a true indication of harmonic problems.
We have a chiller that trips every time there is a blip in our supply voltage. We have to restart the chiller quickly or other processes will shut down automatically. Also, restarting the chiller can cause its own voltage variations. Are there ways to prevent the chiller from tripping every time we experience one of these events?
McGranaghan's answer: It's not uncommon for motor loads, like chillers, to trip during voltage sags. However, the tripping is usually caused by the sensitivity of a contactor, relay, or control. For instance, a temperature controller might be sensitive to a voltage sag and be causing the entire chiller load to trip unnecessarily. A common method to prevent unnecessary tripping of motor loads is to protect the controls with a constant voltage transformer (CVT). The Figure shows the response of a particular temperature controller with and without a CVT protecting it. With the CVT, it's able to ride through almost any voltage sag (but not an interruption) and will prevent tripping of the motor for these events. Note that it's almost always better to let a motor ride through a short duration voltage sag than to trip it and then restart it. The duty associated with restarting is more severe than the inrush duty that occurs when the voltage sag ends.