Life expectancy of microprocessor-based protective relays (IED) 4

[oldfieldguy]

The question has arisen at my workplace as to what we expect out of these things. I have devices installed in excess of twenty years ago.

What do I tell our managers about their life expectancy? We're starting to see sporadic failures on devices.

That is not to mention that software, firmware and hardware revisions have passed up my installed devices long ago. My employer replaces laptop computers at five years. What am I supposed to believe about IED's that protect the plant power system that have been in service twenty years?

What's the industry consensus?

old field guy

 

[JG2828]

Good question. I would like to add that most manufacturers stop supporting the relay communication software when they stop supporting the relays. That is a real problem if the software uses proprietary protocols to communicate with the relay and the software only runs on obsolete versions of windows (NT).

 

[ScottyUK]

After 20 years I would be getting concerned about support and be thinking seriously about replacement. IC's become obsolete, and continued support simply isn't possible in all instances. Based on our failure rates, not many Alstom / Areva / GE Grid MiCOM relays will see their 20th birthdays without intervention. The best I can hope for is that a newer, more reliable product is developed which is physically interchangeable with the current generation.

The software is a very good point: at my previous employer I was running MS-DOS as one boot option on my laptop so I could talk to the ABB REG216 relays in our HV substation. The REOR100 event recorder software ran under Win 3.11. That was at 17 years or so into their life, and in software and operating system terms that is forever.

 

[itsmoked]

Of course there's the support software as mentioned. That can render good hardware useless for a company's use if adjustment and service tasks become troublesome or undoable.

Microcontrollers that are not re-programmable will probably run happily for about 20~25 years.
Microcontrollers that are programmable.. Twenty years is pushing it especially if it is a safety device and not, say, a pinball machine where one might lose a game and 25⊄ in a failure. You can expect about 8~10 years before soft errors can start coming out of the firmware due to the re-programmable memory starting to 'forget' due to multiple physical mechanisms.

This can be mitigated by reprogramming the processors via reading them out while they're still viable and simply reprogramming them. Some companies bright enough to understand this offer re-programming service where they'll send out a new set of chips and you return yours so they can reprogram them and send them to the next guy. That has your 'machine' down only long enough to swap the firmware chips. Other companies have to keep updating their firmware and this gets done automatically with no one catching on until they stop updating the firmware because they finally (think) they've reached nirvana.

The above usually works well only for systems with plug-in EPROMs and in some cases plug-in EPROM based microcontrollers. For other systems the only alternative would be a new set of control boards with 'freshly programmed firmware'.

OTPROMs or One-Time-Programmable-Read-Only-Memory can last indefinitely if it is the type that blows fuses during programming as then there are no memory cells to forget with. However other companies have OTPROMs that are merely windowless and therefore un-eraseable but still have memory cells that will forget. Often they can still be re-programmed to freshen them up but, of course, you can only exactly reprogram them as no values can actually be changed.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[davidbeach]

We're using relays that can be interacted with using a terminal emulator as well as somewhat friendlier setting software. We're seeing equipment failures at 22-24 years of age that are requiring the relays to go back to the factory for refurbishment. The one's we're having the most issue with wire up very differently from what we'd replace them with, so an update is a full project. Grabbing a refurbished replacement is quicker.

 

[LionelHutz]

I believe both Multilin and SEL offer 10 year warranties so those relays should be serviceable for at least 10 years. Relays developed today using any decently thought out and planned architecture will not have software that becomes outdated until long after you cease to care. The software will be modular and basically include a definition that will continue to carry forward into the core software which can continually be updated to match the current version of Windows.

 

[davidbeach]

SEL's 10 year warranty is effectively much longer; they're still covering relays that over 20 years old, no questions asked. We're preemptively sending in spares for refurbishment, including a new main board, so that we can replace pairs when one finally dies. That's much easier than sending in one failed relay for repair and then turning around in a few months having to do the same with its mate. The 25 year old designs have the two relays so interconnected that it's as much work to replace one as it is to replace both.

 

[cranky108]

I would not say no questions asked, as we had some that were water damaged, and we had to pay to fix them, but other than that SEL has replaced all the failed ones we had.

But I agree that 20 to 25 is looking like the end of the bathtub.

The microprocessors are not like the EMs we took out of service at 75 years.

 

[Mbrooke]

So its safe to say EMs are outlasting the microprocessors?

 

[davidbeach]

Sad but true. But since event analysis is a core part of my job I have no interest in going back to the EM days. The relay feature set has increased so much that I’m ready to replace everything more than 7 years old or so, but funding for that doesn’t exist.

 

[Mbrooke]

Sad indeed. But then again the reduced need for testing makes up for it. Seen POCOs go as far as eliminating test switches.


FWIW, can you still buy new EM relays like distance and time over current?

 

[davidbeach]

Eliminating test switches is extraordinarily shortsighted as far as I can tell. I'll accept opening and closing of test switches as an acceptable difference between the as-tested and as-operated conditions. Not sure how anybody can equate as-tested and as-operated conditions without test switches. Perhaps a lack of imagination on my part, or something else on the part of those will operate in a condition that was never tested. A wire reterminated after testing does not leave the protection system in an as-tested condition. At least in NERC/FERC land we've move way past relay testing into the realm of protection system testing and I'm in the camp that doesn't see testing in a non-operational condition as relevant.

Last I heard you could still buy new EM replacements for IAC, CO, and KD relays, if not others. A brand new KD might well cost as much as a brand new SEL-411L, but as far as I know they can be had. Other than having to remove the mothballs from the factory, a new KD is nearly 100% profit at what ever cost; anybody that actually wants one really, really, really wants one and will pay any price so why not the top price? The ones we have are sunk costs, fully amortized. Can't possibly imagine going there again; they don't do so well with two breaker (ring, BAAH, DBDB, etc.) applications when compared with the modern relays. That's why the two breaker configurations didn't really take off until newer relays could provide two electrically independent trip contacts.

oh, but the EM relays don't give a sh!t about EMP and other esoteric threats to the power system. My gut feeling is that if an EMP event takes out our relays there probably is a whole heck of a lot of load left serve in the immediate aftermath either.

 

[Mbrooke]

David, have a look at this:

https://cdn.selinc.com/assets/Liter...witches_MT_20070907_Web.pdf?v=20150812-084434

It is being entertained and there are POCOs that are more even more optimistic then the writers.


In terms of single breaker designs, I think PGE got it right initially- they are still being built all over the world. BAAH is over-rated in my book


EMPs- not sure how credible the concern is- but I would imagine you would have mach larger issues if that truly was to happen. FWIW, there is a rumor that the 1965 blackout was cuased by EMPs from UFOs.

 

[krisys]

It is like the human beings of new generation with fast food culture and the older generation of traditional (ethnic food) food culture. even though the life expectancy has improved now, the modern generation is much more sick as compared to the older generation.

I find some similarities between the relays and the humans beings!

 

[Mbrooke]

I do as well- I agree with your analogy.

 

[DTR2011]

A few years ago I was asked to set up and test a replacement ABB/West HU (EM-87T) Relay to replace a unit deemed failed by others. I was told the cost was $9k for the relay alone. This was at a nuke plant and I believe this was cheaper than redesigning a new scheme, with all of the approvals required.

 

[marks1080]

If I had to set some kind of industry policy for lifecycling IEDs - which I'm surprised NERC and other regulators haven't done yet - I'd go with a cycle between 5 to 10 years. I would like to see the manufacturers make the IEDs more 'rackable' - being on tracks or whatever, with a 'plug in module' for ALL connections to the relay. This way you could swap out an IED that has already been bench tested and commissioned with settings in about 2 minutes.

I think this discussion naturally leads into discussions regarding maintenance cycles for IEDs as well. I think the industry will start moving towards ditching maintenance on IEDs and just swapping them out for fresh models at some regular interval.

 

[oldfieldguy]

For sake of discussion, my problem is that I have an installed base of GE IED's. they range in capability from SR-369 (Motor protection) up through the UR series for feeders, transformers, motors,and one sad old generator.

Age on these range from late 1990's to 2005.

The present software offered by GE for current relays does not communicate with 1990'vintage SR series installed in a facility I 'inherited'. I had to get an older DOS-based program and run it under an emulator to deal with these.

Some UR series installed in the early 2000's are a different hardware and firmware revision than the newer versions we installed in 2005. The UR platform is modular but modules do not interchange from the older to the newer versions, or vice versa.

Since I'm now dealing with a large company with a wide installed base of devices, we're looking at codifying a maintenance and replacement policy.

old field guy

 

[cranky108]

From above, at least one relay manufacturer makes their relays that are change able with plugs. We don't use them because they seemed flimsley and not to our standards.
So we use test switches.

As far as problems with GE software, and other also, That's exactly one of the reasons we don't use GE relays.
I can still plug a dumb terminal into any of my standard microprocessor relays from the 80's through a new one.
That's not to say the versions, or microprocessor are the same, but I can still communicate with all of them.

And here is where I say, price is not everything. And you need to make your management understand that.

Besides, old dos requirements are a new securty feature from young hackers.

 

[edison123]

Though I am motorhead/generatorhead, I found this thread quite interesting and informative.

Thank you all.

Muthu

http://www.edison.co.in