The Case of the Backfed Fuse

Sep 1, 2001 12:00 PM, By Mark Lamendola, Technical Editor

A combination of several problems led to a power quality situation that cost thousands in damaged equipment and lost production.

In many cases, the source of a power quality issue is easier to uncover if a proper maintenance program is in place at your facility. When several problems conspire to give you fits, thorough record keeping and adherence to a proper procedure can help you untangle the situation.

A Tennessee appliance factory where I worked as plant electrical engineer had suddenly begun experiencing an increase in motor failures that were interrupting production. When replacing the motors, electricians consistently reported low voltage on one phase, but never the same one. What was going on?

Analog gages on the main panel showed current and voltage on each phase. On the panel meter, phase C carried almost as much current as A and B combined. Although phase C varied from 80VAC to 480VAC, it usually remained near 80VAC. Because my harmonics analyzer showed similar harmonics on each phase, the problem appeared to be unrelated to harmonics.

My next step was to order voltage readings at regular intervals. The company did not approve the purchase of a power monitor, so I paid electricians overtime for information that was far less complete. The Tennessee Valley Authority (TVA) checked the supply to each substation on site — we had the proper voltage on every phase, so we could rule out problems up to the primary side of the factory's substations.

I assigned electricians to walk down the load distribution. Although it didn't explain the imbalance on the panel meters, they found an imbalance of less than 50A on switchgear that held several 2,000A breakers. However, in the process, we discovered why low voltage appeared to move from phase to phase — the phase conductors were mislabeled. Once I recognized the complexity of the problem, I decided to call Mose Ramieh, a consultant regarded as an electrical guru.

Ramieh had a thermographer trace down the 7,620V/480V distribution system but didn't find any bad connections. Ramieh then conducted a harmonics analysis at every substation (the plant had several) but again found nothing significant. He asked me when we had last checked our grounding system, but we had no record it had ever been checked. Although I was eager to get started immediately, Ramieh advised delaying arrangements for ground testing, because of the effort involved in getting approval for the required shutdown. Regardless of when we were able to conduct the tests, we knew it would be necessary to fit them into a small window of time.

Ramieh next inquired about neutral ground bonds on the load side. We hadn't checked for them, but I explained we had noted strange voltage readings. When I explained we had 451VAC phase-to-ground at one cabinet and 473VAC phase-to-ground in the one next to it, Ramieh concluded this was due to unbonded equipment and bonded neutrals, which created an unsafe plant environment.

The next step was to identify the neutral ground bonds. As it turned out, we had one in every cabinet. We immediately began repairs. When equipment shut off for an order change or shift change, we fixed the neutral-bond problem. While conducting repairs, electricians noticed current flowing through the ground during the process — the connectors were hot to the touch.

We began shutdown at midnight, watching the panel meters as we opened the breakers one at a time. When we opened the breaker for the air compressor motors, phase C suddenly dropped to zero. Now we had isolated a problem, and it was somewhere on the load side of the breakers supplying those motors. I left a note asking my boss to call the motor vendor during the day.

Ground testing showed the ground rods were shot, so we drove new ones. The substation, fence, building, and service entrances were at different ground potentials. Electricians made the necessary bonds while we continued testing. We used digital multimeters to do point-to-point ohmic checks and verify bonding between various metal structures and objects. Finally, the facility had adequate grounding.

Ramieh wanted to open the 7,620V/480V transformer to check for phase loss. However, we didn't have the test equipment or safety gear for the work, so we rescheduled for the following week.

The compressor motors used a partial winding scheme instead of a soft-start to reduce sags from across-the-line starting. The vendor conducted a battery of tests, including insulation resistance checks. One motor had a shorted winding due to an apparent insulator failure, and both motors had overheated. The next day the vendor repaired the motors in place — working on one while the other ran. The facility's compressed air system allowed us to valve in air from a bank of six 400-hp compressors that ran off a different substation on the other side of the plant, so we could take one 600-hp compressor out of service for as long as we needed.

Without proper grounding, undesirable currents had no place to go except through the motor windings. Oddly enough, the circulating currents from our grounding errors made it appear as though the phase wasn't missing — just diminished. Now with the grounding problems fixed, we had no voltage on phase C, and the motors wouldn't start.

We looked at the 7,620V/480V transformer with the load disconnected but the primary hot. We measured 7,620V across the phase C fuse — it was open — and verified it with a digital multimeter. We installed a new fuse and then began energizing the switchgear and bringing loads online while watching the panel meters. We saw 480V per phase and balanced current draw. We knew we had solved the problem when the current remained balanced after startup.

This case illustrates how improper maintenance procedures can make “mysterious” problems harder to solve. When you don't have to troubleshoot a web of interactive problems to get to the source of a power quality issue, you can spend more time solving the problem with a thorough approach. When faced with a power quality problem at a poorly maintained facility, however, you must collect the right data quickly and carefully fix the problems in the right order.