An arc flash began in a switchgear cabinet near the control room, but a just-installed arc-flash relay shut off the power before major damage could occur and potentially saved the company millions in downtime and equipment replacement costs
The damage caused by an arc flash increases rapidly with time
A plant that makes fiberglass insulating blankets and glass insulating bricks had been having major issues with utility power. It was experiencing voltage dips that caused product quality problems, and the utility claimed the problem wasn’t on its end. Much of the electrical equipment at the plant dates to the 1970s, so the owner of the plant, a U.S. refractory company, hired an engineering firm to conduct an arc-flash analysis in an effort to reduce the arc-flash hazard categories of equipment throughout the plant.
The engineering firm’s analysis uncovered a dangerous situation. A 480V cabinet, which was fed from a 3,500kVA transformer, exceeded arc-flash hazard Category 4, meaning there was no level of personal protective equipment (PPE) that would allow a worker to approach the cabinet if it were open. Therefore, the door to the cabinet was kept closed, and everything related to it was done remotely. But the cabinet, along with the heavy transformer associated with it, is located in a room next to the control room within feet of the operators.
The refractory company then called on Evans Enterprises, Oklahoma City, to install power quality metering that would detect any anomalies in the incoming power. Evans installed shunt trips on all the circuit breakers. The plant owner also asked Evans to try to reduce the arc-flash hazard category, so the firm installed a PGR-8800 arc-flash relay from Littelfuse, which could easily be retrofitted in the existing panel.
The cabinet in question contains a three-phase, 4,000A main breaker that feeds four 800A MCC feeder breakers, plus one 2,500A breaker that feeds an SCR power controller that controls current to three electrodes in the glass melting furnace. It has three sections: the back section for the cabling, the middle section for the bus, and the front section where the breakers are mounted. Across the top of the cabinet is another section set aside for control equipment, with no voltage greater than 120V present. The Evans technicians installed the arc-flash relay in this section and mounted two of the relay's light sensors in each of the other three power compartments. They connected the trip output of the relay to the remote trip input of the main breaker. In total, the installation of the relay was a simple job that took about three hours, including placing the sensors.
After the relay was installed, an engineering firm rerated the cabinet's arc-flash hazard rating from a Hazard Category 4 to a Hazard Category 2, a level approachable using 8 cal/cm2 PPE.
The timing of this installation was fortunate. The cabinet sits over a channel in the concrete floor with a dirt bottom. Cables from the load sides of the breakers in the cabinet emerge from the bottom and go down to enter buried conduits that end at the level of the pit bottom. While this would not be permitted in a new installation, this one had been grandfathered in.
Over the years sand from the glass manufacturing process and dirt and debris had accumulated in the channel, nearly filling it, and had found its way into the open ends of the conduits. The cables coming up through the pit were quite old, and the insulation had become brittle. About a week after the arc-flash relay was installed, an arc started on one of the cables a few inches below the surface of the dirt. The arc tracked up the cable to the main bus inside the cabinet, where it became a phase-to-ground arc flash that threatened to destroy the cabinet and potentially the plant — not to mention the effect on any people who might be nearby.
As the flash developed, the relay responded by sending a trip signal to the main breaker in less than 1 msec. The breaker then mechanically cleared the fault with a total clearing time of 51 msec. Without the arc-flash relay, the breaker would have relied on over-current relays, for which a typical ‘trip signal’ time is 16 msec to 32 msec. As a result of the shortened reaction time, damage was minimized. A thumb-size piece of copper was burnt off one of the bus bars and the cabinet and the switchgear inside it suffered only minor damage.
Had the relay not been there, the refractory company could have had costs estimated between $800,000 to $1 million. A switchgear cabinet with a 4,000A main breaker and one 2,500A and four 800A load breakers, plus the necessary bus bars, insulators and cables would have cost around $100,000 to replace. Just getting it into place would have required cutting a hole in the side wall of the plant.
But that would have been only a small fraction of the cost when adding in downtime. The lead time on electrical equipment of that size would be anywhere from 10 to 24 weeks — and that's just for delivery, not installation. That 24 weeks of downtime represents 46% of a year’s production.
As it turned out, the repairs to the switchgear consisted of replacing a few insulators at a cost of about $6,000, and that was because the obsolete equipment could not be purchased from the manufacturer. Fortunately Evans was able to obtain the necessary parts from cabinets they had in stock, and the plant was back up and running within 24 hours, from the initial incident at 6 p.m. to 5 p.m. the next day.