Fault Current Withstand Capability of Station Class Arresters

Station class arresters since 1980 have been based around various types of metal oxide varistors. Despite having been designed to withstand most power system surges, it is still possible for these varistors to become overloaded or fail such that fault current will flow through the arrester’s core. But since this is an expected event over its life cycle, the arrester is designed to both conduct and withstand fault currents at significant levels and not fracture violently risking collateral damage to nearby equipment. In this regard, it is important to assess if arresters are appropriately rated for the system to which they are being applied. Below are some basic concepts to note:

1. Fault (or short circuit) current is not the same as lightning current but is rather the current that flows from the power system when there is a short circuit. If the short circuit is within the arrester, this is understood to be an arrester-conducting fault current.
2. Violent rupture of an arrester is not caused by lightning current but rather by gases created when fault current heats up and vaporizes materials within the housing. Lightning current can lead to fault current but, without fault current flowing, there would not be a rupture.
3. Arresters are designed to conduct lightning current multiple times over their service life and these operations do not generally result in failure or If an arrester operation does lead to failure, it will likely become a very low impedance path to ground and result in fault current flowing.
4. Arresters are designed to conduct fault current one time at their end of life. Applying fault current to an arrester a second time, such as from reclosing a circuit, can result in violent rupture.
5. When selecting an arrester for a new or old application, it is important to check that the fault current available at the arrester’s location is compatible with the unit’s fault current withstand rating.

station class arrester Fault Current Withstand Capability of Station Class Arresters Design A Types 1 and 2 with porcelain and composite hollow core housings
Above: Design A (Types 1 and 2) with porcelain and composite hollow core housings. Below: Design B (Type 3), with polymeric housing.
CLICK TO ENLARGE

As with all critical arrester parameters, fault current withstand capability is verified by testing at high current laboratories. The relevant test procedure is found in IEEE C62.11 as well as IEC 60099-4, and is one of the arrester tests that are identical in both. To determine fault current withstand capability, the fundamental test procedure is to apply the specified fault current to the arrester, which is expected to take this current for the full duration without interruption. The arrester is permitted to fracture but parts cannot be expelled beyond a limit equal to a circular area with a diameter twice the unit’s height. Per both standards, all arresters must have a fault current or short circuit rating and must be tested at two or four current levels depending on their design.

Arrester design A (Type 1 and Type 2) has substantial internal air volume while design B (Type 3) has no measurable internal air volume. The test currents most often used for design A station class arresters are: 63 kA, 25 kA and 12 kArms for 12 cycles and at 600 amps +/- 200 ampsrms for 1 second (peak levels are plotted in the graphic.) Design B would typically be tested only at 63 kA and 600 amps.

Station Class Arrester station class arrester Fault Current Withstand Capability of Station Class Arresters Fault current withstand curve
Fault current withstand curve.
CLICK TO ENLARGE

To help understand how to apply fault current test results, an Arrester Fault Current Withstand Curve shows the currents at which arresters are tested in the standards. The chart is divided into three zones – Elevated Risk of Violent Rupture, Low Risk of Violent Rupture and Unknown – that describe what might happen if an arrester is subjected to these fault current amplitudes and durations. End of life performance with respect to fault current can then be assessed by plotting expected fault current at the arrester’s location and analyzing the zones this passes through. The plot shows the fault current experienced by two arresters – both certified to 63 kA short circuit capability – from two different circuits and compares how the arresters react to fault current based on location. The green curve represents the fault current experienced by a switchyard arrester with 17,000 peak amps that is interrupted in .0028 seconds; the blue line represents the fault current an arrester at a generator substation might conduct during an end of life event. The current is not interrupted in this case but decays as the generator slows down. For the switchyard arrester, it seems that the arrester should perform well and not rupture from the event. By contrast, the performance of the arrester at the generator station would be questionable and perhaps the supplier should be consulted since it passes into the zone of high rupture risk. As seen from these examples, what may seem a simple ‘yes’ or ‘no’ proposition can sometimes prove more complex.

Jonathan Woodworth