Variable speed drives (VFDs) turn processing machines and pumps at a plant that makes “mud” (joint compound) for finishing interior drywall. The plant is almost fully automated, and depends on the smooth operation of the electrical system.
The plant had experienced a power issue: when the power went off at one power panel some power outlets went off but others stayed on. The lighting supplied from the same board showed a similar pattern. Because the power issue didn’t cause immediate disruption, the plant operators were uncertain how long the power was off. They realized there was a problem when they noticed that the displays were no longer updating. So, although the PLCs continued to work there was no display information to help determine what was happening. The plant operators discovered that the server running the SCADA system and the fiber optic switch connecting the PLCs had lost power; both these devices are “protected” by a 20 minute UPS.
Power to the circuit had been restored before the electrical consultants arrived. Preliminary investigations revealed that an electrical contractor in the adjacent building had by agreement fed a ground cable to the ground bar in the affected building. Initial measurements showed that the cable was carrying 2A of ground current. The electrical specialists also discovered that the water pipe, which had many cables bonded to it, had eroded due to electrolytic action and was leaking.
The detailed investigation
The first part of the detailed investigation was to measure the ground current on the system. The electrical specialists confirmed, by using an earth ground tester, that there was 2A on the neighbor’s cable and 1 or 2A on some of the company’s own ground cables. The interconnecting bonds at the ground bar had obviously been installed at different times over many years, resulting in a “rat’s nest” of cables.
The investigators then tried to make measurements of ground current at the buildings ground rods. As is usual in most such cases, understanding a buried ground system is very difficult unless you are prepared to dig and investigate. After a short time they gave up and decided to investigate the ground system at the main breaker outside of the building. They had to suit up in PPE (personal protective equipment) so they could make the measurement. At the same time, while the covers were off the switch gear, they carried out an infrared survey.
The infrared survey showed that the main switchgear and cable connections were running at a completely normal temperature, showing no hot spots. The investigators took the opportunity to check the voltage and current harmonics present using a three-phase power quality analyzer. The readings showed very low levels of distortion and the supply looked very clean.
The electrical consultants took the investigation inside the building, to the service panel where the power had been lost. They confirmed more details of the ground currents using a current clamp meter. They then decided to disconnect the neighbor’s ground connection. Once this was done the ground currents were measured once again, which showed that the disconnected ground cable was not having any affect on the circuits within the problem building.
They then decide to carry out an infrared survey on the small distribution transfer in the service closet where the ground bar was located. When they removed the front panel of the transformer, the thermal imager showed hot spots on the transformer connection of phase C. This was the circuit that failed in the previous week. The loose connection was so bad that during the survey the connection sparked due to a small disconnection. In response they arranged to power the local power panel down so that repairs could be made. This caused a disruption to the office facility and the production area, because the local displays could no longer connect with the SCADA system.
All the connections in the transformer were tightened-there were many lose connections. The panel feeding the transformer was also checked and major issues were discovered. More than fifty percent of the connections in the panel were loose, the neutral was connected to ground at the correct point (according to NEC code), the panel was not properly grounded due to lose connection on conduit. There were many supplementary grounds on local circuits, again not in line with code. The main breaker in the panel was cracked and required immediate replacement. It was agreed that the electrical consultants would replace two panels and bring everything up to code in the next few weeks.
Summary
It took the use of all the test equipment available to find and correct the problems. In each case they were used to detect different aspects of the situation. The earth ground clamp meter and the current clamp meter both showed high levels of ground current. Most likely these results were due to the VFDs in both plants being connected to the same transformer. The power quality analyzer showed that the voltage and current harmonics at the service entrance contained very little distortion. The thermal imager showed hot spots in an internal distribution transformer on one phase, which was the phase exhibiting the problem earlier. Remedial work will be carried out based on the investigation.
The crisp, clear images on the Fluke Ti32 thermal imager helped narrow down the possibilities fast.
Suiting up with PPE protected the investigators from the possibility of arc flash.
Troubleshooting with the Fluke 435 power quality analyzer went quickly with on-screen display of trends and captured events, even while background recording continued.
by Frank Healy, Fluke Marketing Manager, Power Quality Specialist
Frank has advised and consulted on many aspects of instrumentation for electrical engineering, including installation testing, earth / ground testing, and power, for more than 20 years. He started his career as a hospital engineer in the UK. frank.healy@fluke.com
Comments (2)
I would think that the equipment nearest to the problem would be the FIRST place to look.
1) Posted 9:04 am, 16 November 2011 by Paul Kail
2) Posted 8:06 am, 23 November 2011 by Dana Payton