Surge Arrestor Failures 3

[CuriousElectron]

Greetings,
Does one expect for a surge arrestor end-of-life event to result in a ground fault? If a surge arrestor fails, we want to make sure it doesn't explode.
The pressure-relief rating of the SA should be higher than maximum SLG fault calculated for the system.

Thoughts?
Regards,
EE

 

[davidbeach]

Like many such things, the answer is “it depends”. For distribution class arrestors connected here, there, and everywhere we use a ground disconnect that is blown open by the current through the failed arrestor. Works well for that application.

On the other hand, if the arrestor on a substation transformer fails I want to know it now rather than waiting for a line patrolman to find the open disconnect. The arrestor is a sufficiently important part of the transformer protection that tripping for a failed arrestor is fine.

Don’t know of a failure mode for MOV arrestors that doesn’t involve either a temporary or permanent ground fault.

 

[bacon4life]

Once in a while we catch an arrestor at the end-of-useful-life using routine thermal imaging. For substation class arrestors, routine leakage current testing may detect end-of-useful-life. Some folks use surge counters to trigger testing of substation arrestors and others just test substation arrestors on a time based system. David nailed it for catastrophic failures of the MOV.

For distribution class arrestors, there may be failure modes where the explosive ground disconnect triggers prematurely, resulting in an open failure.

Unrelated to typical MOV failure, any mechanical connection can fail open circuit due to poor installation technique, material defects or corrosion in the arrestor body or the connected wires.

Rather than just considering single line to ground faults, consider the maximum fault duty to which the arrestor is exposed:
-Surges often affect more than one phase, and such a surge could result in multiple simultaneous failed arrestors and a multiphase fault.
-On delta systems and impedance grounded systems, using L-G fault current would clearly be unreasonable.

 

[davidbeach]

And on delta and impedance grounded systems don’t use arrestors with the same MCOV that you’d use on a solidly grounded system.