Refurbished cable testing questions 1

[TurbineGen]

I have a few questions for the MV cable experts out there regarding compliance testing of shielded MV cables. We have a mobile substation in the U.S. with 9 - 500MCM - 15 kv shielded cables, 3 per phase. These are fed from the mobile substation by 6 bare copper (2 per phase) 500MCM cables. Cables are EPR with stress cones on either end. After less than 2 months in the field, one of the stress cones on C-phase had failed on the far side and another stress cone also on C-phase at the mobile sub end was very near failure. We saw a significant amount of adhesive (mastic?) coming from several of the other stress cones. The manufacturer has agreed to replace all of the stress cones on the cables.

Here's my question: What type of compliance testing should we be doing to ensure that these cables will not fail again?

The vendor says that they do an AC HiPot test to 31.5kv for 5 minutes and a DC 31.5kv HiPot withstand test for 30 minutes. I did the AC HiPot on the cable that was near failure and it passed, thus I feel that this test is inadequate. Looking at several references they suggest only a HiPot and a PD test. I'm not too certain we have the capabilities to perform the PD test.

Thank you in advance.

------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.

 

[Zogzog]

Read this then ask questions if you have any

http://www.eng-tips.com/viewthread.cfm?qid=248906

 

[TurbineGen]

I had read through that thread and I'm still lost as to how the stress cones should be tested. I know the cable insulation is good. What I get from that is the same conclusion I had earlier: A HiPot test will not detect the failure I am seeing.

I am having issues with stress cones overheating and then failing. The cable insulation is ok. The only thing that I can see is to do an off line PD test. Is this my only other option?

The vendor has mentioned using a thermal camera during the AC HiPot tersting and looking for a hot spot. Do you think this would be sufficient? I am highly uncertain of this test.

Unfortunately we do not manufacture cables here at this facility and have no experience doing PD testing on cables. We also need to get this unit back in service in 2 weeks.

Thanks again.

------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.

 

[dpmac]

IMO the cables should be PD tested at 60Hz. This will tell you if the terms were made up properly. You mentioned the terms failing due to overheating, can you elaborate? Have you sent the failed terms out for forensic evaluation? That may answer your quality concerns very quickly. If you have purchased a cable assy. It should be tested just as they would the cable and that would include PD at power frequency. Not sure where in US you are but try these firms they should able to help you.

http://Www.imcorp.com

http://Www.advtest.com

 

[dpmac]

Sorry about that. It should have read "

http://www.imcorptech.com"

 

[Zogzog]

I have a hard time seeing how an IR camera would see anything, maybe a corona camera. There are many companies that can assist you with this, depending on your location. Here is another one

http://nooutageelectricaltesting.com/

 

[TurbineGen]

I will have to look into the PD testing companies. I am in central Virginia. From what I am reading it appears an offline PD test will be the best to determine if there is a problem with the stress cones.

We sent the stress cones out for evaluation and I have not heard back yet. Attached are the photos taken from the job site. I'd like to hear your comments.

Thanks again



------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.

 

[ppedUK]

This looks very similar to a number of failures of new 11kV single core cables that I have come across. The failures in our case were caused by local overheating of the earth braid connections to the copper wire screen, which were effected using constant force springs. The overheating was caused by the circulating currents (the cables were intended to be bonded at ends) circulating via the screen earth braids, similar to your installation. As the connection gets hotter, the insulation at that point gets softer and the springs bite into the insulation and eventually.....magic smoke. I suspect that you're getting melting of the moisture block / mastic at that point and that is what you can see oozing out.

In our case, no amount of testing would have found the problem when the cable was put into service, as the faults developed only after several weeks of running.

In our case there was nothing wrong with the termination kit, but we concluded that constant force springs don't like circulating currents and we changed the method of termination on later installations.

 

[rasevskii]

This would appear to be a problem due circulating currents as said by ppedUK in the braids. There should not (therotically) be any significant current in the braids/shields. There may be ground loops or harmonic currents. If codes allow, lift the ground from the braid at the load end of the cable. If you can, try to measure the current in the grounding braids beforehand.

The panel frames at the load end have to be grounded according to code obvoiusly at some point, but the cable shield/braids should not be that ground path. A separate grounding cable may be an option instead from one end to the other.

We had a lot of trouble with an REF protection once on a 6.6Kv system, only resolved by lifting sheath gronds at the load end of cables (about 50m distance).

rasevskii

 

[TurbineGen]

I think ppedUK got this one. I forgot to check to see if the other end of the cable was grounded. I talked to one of the installers who confirmed that both ends of the cables were grounded. That explains the issue since the mobile substation had a different ground than the load.

So now the next question: Do we ground the shields at just one end and keep the other ends all tied together (but not grounded) or do we only tie the three shields on the same phase together on the other end?

------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.

 

[rasevskii]

You said "three shields on the same phase" that must be wrong it has to be "each shield of the three single core cables" (if that is what you have...) In any case the shields on the load end must be tied back, insulated, and left ungrounded to eliminate any circulating currents, as I see it that is.

I suppose this is a mining application where special codes rule.

rasevskii

 

[VLFit]

You will read replies from people telling you to PD test the cables. Although PD testing is of course a valuable tool, it an expensive and complicated overkill in this case. You have only several short cables. An AC withstand test is all that is needed, whether 50/60 Hz or VLF. If VLF is used, the proper test voltage and time should be observed. To perform an abbreviated time test or too low in voltage a test and then say the method is no good due to a later failure, is not giving the equipment and the physics behind the method the chance to work. Withstand, TD, and PD testign are all valuable for cables, however, depending on the situation, some are better than others and some should be avoided all together. Don't give in to the notion that the only way to test in all cases is pd, especially when the pd guys can not provide you with precise levels of accepatable pd to allow in your terminations in your environment. If a cable passes a proper AC withstand test, who cares what the pd is, it will last for years.

 

[dpmac]

VLFit,

I believe the OP stated in his original post that the cables were tested with 31.5 kvac for 5 min. and the cables still failed in service.

Are you saying they should be tested at a higher voltage or a longer time or both?

 

[ppedUK]

TurbineGen,

In terms of circulating currents, it doesn't matter whether your screens are grounded at one end or both. If the screens of all phases are tied together at both ends, as in your installation, you will get circulating currents whether grounded or not. To eliminate circulating currents, you will have to isolate the screens from each other and from ground at one end. In our installation we couldn't ground at one end only due to the long runs and we accepted that we would have circulating currents, but the constant force springs were not up to the job of carrying circulating currents of that magnitude (around 100A in our case).

The decision of whether to ground at one end or both is determined by calculation of the potential rise of the isolated end of the screen under phase fault conditions. The potential rise limits are usually dictated by local regulations or may be determined by the additional voltage stresses on the insulation.

Incidentally, text books will tell you that if single core cable screens / armours are tied together and grounded at both ends, then installing the cables in close trefoil formation will virtually eliminate circulating currents in the screens / armour. In the real world, you still get significant circulating currents even when installed in trefoil formation throughout their entire length.

 

[TurbineGen]

Looks like I have some calculations to do. I think I can get away with grounding at one end and isolating the other, but I will need to check.

Thank you so much for the help. This was a mystery to me up until now.

------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.

 

[ScottyUK]

VLFit,

Your post made sense until the silly and erroneous comment "If a cable passes a proper AC withstand test, who cares what the pd is, it will last for years.".

I guess from your handle you are in the VLF market - if you have any expertise to contribute on this technique I'm sure people would be interested - it doesn't get the discussion and recognition it should.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[Skogsgurra]

A few pictures from a similar case.

Current induced in cable screens. Opening gnd connection in one end cured it. One solid gnd connection installed.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[TurbineGen]

Thanks, Gunnar.

We have started our inspection of the cables as we are refurbishing the whole lot. A few of the cables have heating right where the semiconducting layer meets the insulator. It looks like the stress barrier wasn't applied properly. In other cables the hetin iriginated from the band where the shield is connected.

So it seems we had two problems: Poor cable termination and circulating currents. It's no wonder they lasted only a couple of weeks.

Thanks for all the explanations.

------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.