Not the usual suspects
Not long ago we heard from an electrician who had a customer with a very common problem - an overheated electrical panel. What made the situation somewhat unusual was the fact that the overheating didn’t result from any of the usual causes.
The electrician was called in to troubleshoot an electrical panel with 42 circuit breakers that supplied power to dozens of servers and other nonlinear loads. The IT tech at the customer site noticed that the face of the 225 A main circuit breaker, which was supposed to be white, was noticeably yellow. When he felt the face he found that it was not only discolored but was seriously warm to the touch.
Although there had been no outages, the resources powered through this panel are mission critical, so the customer couldn’t afford to wait until things went bad to address the problem. He decided instead to get to the root of the problem before an outage occurred.
Troubleshooting with the power on
Since even a planned shutdown would have been very difficult to schedule, all troubleshooting was done with the power on. In an initial survey of the environment, the electrician noted that the neutral conductor consisted of two 4/0 conductors, an oversized neutral commonly found in panelboards that serve nonlinear loads. That told him the system was designed properly.
His next task was to bring out his Fluke 1587 Insulation Multimeter to measure the phase-to-phase and phase-to-neutral voltage on the line side of the 225 A main circuit breaker. He was looking for any anomalies that could point to a problem. Those measurements all were well within the normal range - nothing too high or too low - so he moved on.
Next he looked at the current draw on each of the incoming phase conductors with the Fluke 376 Clamp Meter and found that the current was well below 225 A on all three phases - ranging from 108.9 A to 130.3 A. Because the panel powered nonlinear loads, the next logical step was to look for current harmonics. The electrician clamped the 376 on the neutral conductor and found the current to be quite low - just 38.9 A. If harmonics were causing a problem, the current on the neutral would have been equal to or higher than one of the phase currents.
Ruling out harmonics
What about the loads served by the panel? Maybe there were voltage harmonics there? This time the electrician brought out his Fluke 434 Power Quality Analyzer to check the harmonic content of the loads served by the panel in question. The harmonic content was consistent with the type of loads served. The voltage was good and there wasn’t a lot of extra high current or harmonics; once again he hit a dead end.
So it appeared that harmonics weren’t the problem either.
Discovering the location of the problem
Figure 1. An infrared image from the Fluke Ti55 IR thermal imager of the phase conductors of the main circuit breaker indicates that the C phase conductor is significantly warmer than the A and B phases.
But he wasn’t finished. Next he recorded the voltage drop across the line and load side phase conductors of the main circuit breaker. A loose or bad connection or an internal problem with the circuit could produce a large voltage drop - something greater than 100 millivolts (mV). In this case the voltage across the A phase was 51 mV, and across B phase was 41 mV. But then when he got to the C phase the voltage was 137 mV, which made it pretty obvious where the heat was coming from.
Since the problem still didn’t warrant a shutdown, the electrician brought out his Fluke Ti55 IR thermal imager and lo and behold the C phase showed up significantly warmer than the A and B phases (see Figure 1).
At the top of the circuit breaker the wire was very warm and grew cooler as the camera moved farther way from the circuit breaker. This indicates that the problem was most likely at the lug or where the lug connects to the circuit breaker.
Before leaving the site, the electrician attached the Fluke 1735 Three-Phase Power Logger to the conductors feeding the 225 A main circuit breaker and left it for a week to measure current. That didn’t show any anomalies with the current (see Figure 2).
Figure 2. A week-long recording of the current on the conductors feeding the main circuit breaker showed no anomalies indicating that the problem was at the lug, or where the lug connects to the circuit breaker.
Because the troubleshooting isolated the location of the problem but didn’t pinpoint the exact cause, the customer and electrical consultants decided to replace the entire guts of the panel - including all the circuit breakers - and pull new conductors.
“We’re going to that extent because the cost of downtime and criticality of the system are so significant that the customer wants it fixed. They don’t want to have to arrange another outage anytime soon,” the electrician explains.