Protective Relays Testing Intervals. What standard states times?

[AJ2002]

I am looking for the testing intervals for protective relays? Every two years seems to be a rule of thumb but many standards reference the manufacuture's manuals for recommended testing time frame. The manufactures in many cases say testing is not required after the initial setup, calibration and functional tests. I referenced NETA testing spec's for relays that told what to test but not how often. I referenced C37.90 there was no interval stated. I referenced IEEE Yellow it defers to C37. IEEE Buff states period testing required, but does not state when.

Does anyone have a standard to reference? Thank you for your responses.

 

[busbar]

Ypu might try

http://www.netaworld.org/files/ItemFileA42.pdf,

although keep in mind NETA has a vested interest in the testing business.

 

[oldfieldguy]

The NETA specification is a good starting point, as are manufacturer instructions.

Most manufacturers in the area of the US Gulf Coast seem to do 2-5 years, three years being perhaps the most popular. The whopping big electrical utility I used to work for did five years on electromechanicals. They were talking about doing away with full testing on microprocessor based relays after initial commissioning, relying on the relay's self-test features and metering checks on the relay, and circuit function checks for the equipment being protected.

If you've got relays in adverse conditions such as elevated humidity, dirt or temperature, then annual checks might be a good idea. I've seen relays shaken apart by mechanical vibration, and premature demise due to high temperatures in metal-clad substations in the summertime.

old field guy

 

[Zogzog]

I agree with ofg, electro mechanicals need to be tested every 3 years or so, because they do fail (The dirt factor). But besides start up testing I have never failed a micro processor based relay, I dont see the point in testing them at all after commisioning.

 

[slavag]

Hi.
Please see attached for your information.
thread238-211068

Best Regards.
Slava

 

[etronics]

NPCC has issued new standards for testing intervals of EM relays, solid state and microprocessor based relay, I would look there first.

 

[stevenal]

The selinc website has papers that question the need for any routine testing of microproccessor relays after commissioning. They test themselves continuously with none of the potential missteps that can occur when a relay is removed from and placed back into service.

 

[dpc]

My only concerns with elimination of maintenance testing of digital relays:

Firmware upgrades

Physical wiring and output relays

 

[davidbeach]

Which firmware upgrades? Firmware upgrades that solve a known problem that apply to the particular relay as used at a particular location should be made as soon as practicable and not wait for scheduled routine maintenance. Firmware upgrades that only add features not needed for the particular installation or change unused functions of the relays should, in general, be avoided as every change has some risk of introducing new errors.

The regional reliability organizations (WECC, etc.) are moving toward a requirement for testing of "protection systems", of which the relay is only one part. As far as I'm concerned*, a numeric relay can be tested by reviewing the alarm log looking at the metering. The testing of the rest of the protection system will require the relay to do various things, such as tripping the breaker and closing the breaker. That is quite different from conventional relay testing.

* My opinion does not necessarily align 100% with the practices of my employer and I am not speaking for them.

 

[dpc]

I'm just saying that a relay may need to be re-tested after a firmware upgrade. The issue of whether or not to upgrade is another interesting question. Unfortunately not all manufacturers identify specific issues fixed by a particular firmware upgrade.

 

[davidbeach]

Oh, absolutely. In many ways testing after a firmware upgrade is more important than initial testing as that particular relay was never through the factory testing with that firmware.

 

[cranky108]

I read what the NPCC said in the paper above, and I agree that the micro-processor relays need to be tested. But not in the same way as the Electromechinical relays do.
It also aligns with one of the SEL papers on testing.

 

[slavag]

Hi.
Small additional Q's.
What is a billing meter testing interval?
In case of fault, are you show to your customer SEL,GE, Siemens,Ziv, ABB, Areva, etc.. papers?
Are you think that self supervision is all, are you have papers what is tested and what is interval?
What is "free maintanance", what, SEL is responsible for my site and pay for the damages?
Best Regards.
Slava
Sorry for sarkazm.

 

[cranky108]

Customers usually don't ask about testing, unless they have a problem with one of there relays. It's good customer service to help. (Although they can access the same info. on the net).
NERC is pushing testing and rational behimd testing interval in the US.
SEL hasen't ever charged us for replacing or fixing one of there relays, even if the problem was our fault (Bad wiring).
The self testing only applies to the microprocessor, the inputs, and outputs, including the analog parts are not self testing, And therefor still require testing.
But NERC is going beyond just relay testing. They want to include documentation of battery testing, circuit checking, PT's, and CT's. It's the setting up of the paper trail that's consuming more time.

 

[slavag]

Cranky, great!!!!!
"The self testing only applies to the microprocessor, the inputs, and outputs, including the analog parts are not self testing, And therefor still require testing."
It's short and full answer.
Best Regards.
Slava

 

[521AB]

Inputs and outputs can be supervised, actually. It depends how much you want to go deep with the supervision.
The problem is if the self-supervision should be supervised, which is possible, but then we enter in MTBF/complexity issues and supervising the self-supervision brings the system to be so complex that it will have a higher probability to fail than a non-supervised supervision.

Anyway, I think hat you should consider that there is not only the relay (numerical relay) to test.. the relay is connected to CTs, VTs, to the trip coil or trip unit, to the battery..... Are all these connections STILL ok? Are they maybe a little loose?
You should think that around the numerical relay there is still a large part of electromechanical stuff, and it has also to be tested. Do you test it every 5 years?

 

[stevenal]

I hope no one is ignoring the battery for five years at a time. That interval is far too long.

 

[DTR2011]

What really needs to be addressed is how numerical relays are tested. It kills me to see how many around the world are still doing static testing, pick up / drop out testing, etc on a digital device. SEL is correct in many of their thoughts, however I have seen too many utilities do away with testing altogether, which hides other hidden failures of the IED's.

This is especially prevalent when IED's were newer and the folks that commissioned the relays were not familiar with best practices for getting all of the settings back into the relay after testing. I am sure there are 'horror stories' about this. Older relay books for electro mechanical relays called out for blocking certain contacts open and closed while testing discrete elements. A thoughtful relay tech will use a $20 bill to do the blocking as this will not be forgotten at the end of the day.

I believe I read somewhere that the biggest failure mode for relays is not a relay failure itself, but a bad setting or blocking due to testing.

Most IED SW has a way to compare the settings that should be in the relay with the settings that are in the relay.

I am a big fan of either fault record playback (transient testing) or some kind of accurate state simulation. This is often done on the transmission level, but can also be easily applied to the MV protection schemes.

The problem is that most relay techs don't have access to either a goof fault record (which can also be made in ATP/EMTP or good relay testing network simulation SW) or a state simulation (from Aspen, etc). If better communication between the testing departments and the system protection departments is established, cost effective and very comprehensive testing can be performed in about 5 minutes.

The other problem is getting the buy in for both parties that this type of testing is not taking away someones work, or creating work for others.

If the COMTRADE or state simulation data is made readily available from engineering, it takes longer to get an outage and hook up the test leads than it does to run the test. Careful examination of the event reports (if properly configured) will reveal the performance of all elements in the associated protection scheme.

 

[oldfieldguy]

stevenal--

That 'battery' thing also applies to relays sitting on the shelf as spares, as do thoughts of electrolytic capacitors de-forming during long periods without charge.

old field guy

 

[stevenal]

Oldguy,

True. The NERC standard, however, does not concern itself with spare parts on the shelf.

My point is, that everyone speaks of the interval as if only one interval exists. The covered equipment will require more than one.