After the power failed at a New England health care facility, a power quality investigation revealed an unlikely culprit.
At about 4:00 one afternoon, I received a call from an anxious electrical contractor at a job site. One of his electricians had caused a power outage at a large health care facility while implementing power quality improvements I had designed. The electrician was in the process of replacing an undersized grounding electrode conductor at the secondary side of a 480V - 208V/120 delta-wye, 75kVA transformer when the building's 2000A, 480/277V, 3-phase, main circuit breaker (CB) tripped. He also caused 225A fuses to blow on the busway riser at the fourth floor. He restored power to most of the building (by resetting the main breaker) before the contractor called me but was reluctant to replace the fuses because he couldn't identify the reason they had blown.
I arrived at the site and went to the fourth-floor electric room, where I met the electrician. His explanation of the sequence of events is as follows:
He replaced an undersized lug on the XO of the transformer, connected a new grounding electrode conductor to this lug, and pulled the conductor through the existing conduit to the vicinity of a ground bus in the corner of the room near the ceiling. The 600V-rated cable was not skinned back at the unconnected end.
He held the cable near the ground bus to measure how much to cut off when he heard a loud noise and the lights went out.
When he went downstairs to get a flashlight, he discovered power to the whole building was out.
He didn't know what he had done wrong, but he assumed it was related to the new ground cable. Returning to the fourth floor, he started to remove the cable at the new lug on the XO of the transformer. As he did so, the ground cable came in contact with one of the adjacent phase lugs, which caused a spark.
He tested the three 225A fuses and found two of them blown.
First, I tried to find the cause of the outage, which would lead us to the likely damage. I examined the bus duct for openings the cable could have entered, but I didn't find any uncovered openings or arc marks. Based on the description of events, I didn't think the cable was long enough to enter the transformer because the electrician held it near the ceiling at the corner of the room.
To help understand the event, I sketched a simplified one-line diagram of the power distribution system. Next, I made the following list of questions:
Why didn't the 125A breaker protecting the transformer open?
Why didn't the 225A fuses in the bus duct switch open before the 2000A main tripped?
Why didn't the 800A breaker in the main switchboard trip before the 2000A main?
What started the problem?
If power to the whole building was off, how did the electrician create a spark when he was removing the ground cable?
I first examined the 125A CB, and found that its adjustable magnetic trip was set at its highest level (1250A). This explained why it did not react quickly.
Next, I inspected the 2000A main CB. Because the service is rated at 480/277V, the main CB has ground fault protection. However, it had been left at the lowest factory setting. This explained why the main CB tripped before the smaller overcurrent devices.
At this point, we acquired better portable lighting so we could more closely examine the equipment on the fourth floor, where the electrician was working. We noticed the H1 and H2 lugs were pitted, and there was a small amount of carbon dust on the mounting bracket adjacent to these two lugs. Originally, I dismissed the significance of the spark, believing it came from the release of a small amount of stored energy in the coils of the transformer. However, the discovery of pitting indicated this transformer was at least partially energized when the arcing occurred. Further examination also revealed pitting on the ends of the strands at the unconnected end of the new grounding electrode conductor.
Based on these discoveries, I formed my theory of what had occurred: The electrician was not holding the ground cable near the corner of the room when the lights went out. He must have lowered the cable, thereby allowing the loose end to enter the front of the transformer and touch H2. This triggered two events:
A short circuit that caused one of the 225A fuses to blow.
A ground fault that tripped the 2000A main circuit breaker.
When the electrician attempted to remove the ground wire, he didn't realize the facilities generator had started, which had re-energized H1 and H3. The spark he described was the short circuit that blew the second fuse.
When an electrical design is in its preparation stage, perform a coordination study to make sure the overcurrent device closest to the fault will trip first. Then, when the equipment is installed, make sure all trip settings are set to the design requirements. This type of work should be done on a de-energized power system. Although the site is a health care facility, it operates on an outpatient basis, and work could have been done at night. In cases where it is necessary to work on a “hot” system, you should follow these basic safety precautions:
Have another qualified person present for emergencies.
Wear goggles and insulated gloves.
When appropriate, use insulating blankets on exposed bus bars and wear a fire-resistant jacket and face shield.
Carry a flashlight for possible emergencies.
Tape the ends of conductors you're installing.
When the lights go out don't assume a total power failure. Some of the power wiring may still be energized, so you should test each phase conductor to be sure.
This case proves a person's memory is not always reliable when recounting a traumatic event, so approach a power quality problem by establishing the facts of the case with your own investigation. Not only will you be assured of gathering the correct information, but you could uncover previously hidden evidence that might lead to the solution.
Gaskell is the president of Gaskell Associates Ltd. Consulting Engineers, Warwick, R.I.