LV and MV Cable Testing 5

[EEbyChoice]

I have quite a few on-site backup power generation facilities (located at customer sites, commercial and industrial)ranging in size from 16 MW down to 350 KW. Age ranges are from 3 years to 12 years or maybe a little older.

I am trying to establish an electrical preventative/predictive maintenance program for these facilities that includes cable diagnostic testing. I am interested in determining which types of testing should be used at the various facilities.
I have experience with DC Hi Pot testing and DC insulation resistance testing, but there are many other types of testing now available. We already conduct regular thermographic surveys of each site, but I have some concern about cable insulation system integrity at our older sites.

Most of these sites have never had any type of cable testing done so I don't have much history to work with.

I need to minimize risk to installed assets and also prevent the possibility of testing related customer outages or unavailability of equipment.

Any feedback would be appreicated.

 

[busbar]

 
One reference for low- and medium-voltage cables is §7.3 NETA 2001 Maintenance Testing Specs. (Revision anticipated March 2005.)

http://www.netaworld.org

 

[benlanz]

You are not going to find an effective solution at NETA. The NETA standard has not caught up to IEEE standards yet. NETA is rushing up date their standard to match IEEE 400. I happen to know this because I am a writing work group member of the NETA cable test task force, and I am an officer on the IEEE 400 comittee.

The only proven, effective way to establish an electrical preventative/predictive maintenance program is using Partial Discharge testing. It is the most expensive option, but you will get what you are looking for.


IEEE 400 states:

IEEE 400 favors Partial Discharge Testing
7.5
In summary, if the cable system can be tested in the field to show that its partial discharge level is comparable with that obtained in the factory tests on the cable and accessories, it is the most convincing evidence that the cable system is in excellent condition.


Direct Voltage Testing
5.5 Summary of Advantages and Disadvantages
Some of the advantages and disadvantages of dc testing are listed below:

5.5.1 Advantages
Relatively simple and light test equipment, in comparison to ac, facilitates portability.

Input power supply requirements readily available.

Extensive history of successful testing of laminated dielectric cable systems and well established data base.

Is effective when the failure mechanism is triggered by conduction or by thermal consideration.

Is effective on inter-face problems of joints and terminations and surface problems of terminations.

Purchase cost generally lower than that of non-dc test equipment for comparable kV output.

5.5.2 Disadvantages
Is blind to certain types of defects, such as clean voids and cuts.

May not replicate the stress distribution existing with power frequency ac voltage. The stress distribution is sensitive to temperature and temperature distribution.

May cause undesirable space charge accumulation, especially at accessory to cable insulation interfaces.

What does this mean? DC is a cheap test, but it is not effective on modern cable installation.

May adversely affect future performance of water-tree-affected extruded dielectric cables.

Leakage current readings may have wide variations due to atmospheric conditions and lack of control of charges at termination lugs.


-Cheers

 

[busbar]

benlanz, would you please comment on applicability/merits of VLF cable testing {and IEEE P400.2}?

 

[EEbyChoice]

Thanks for the information on IEEE 400. I have read through this document and there is good information to be found. Below is a specific case of what I am working with:

Currently, we are testing 5KV cables terminated at one end in the generator (4160 VAC) and at the other end in switchgear enclosures. The terminations at the generator are splices, not bus connections, so we leave them connected and run Polarization index tests on the cable/generator windings together. If we get a bad test, we will disconnect the generator and retest individually to see if the issue is with the generator or the cable. Any cautionary information I should have regarding this technique??

I am going to do some research locally to see who is capable of doing PD testing.

Thanks,

 

[alehman]

I agree with benlanz re. PD tests. It requires extended time monitoring to be beneficial. I think the PI test will be of limited value for finding defects in MV cable. Typical PI test voltage might be 2500VDC. Hipot would be the next best choice (VLF AC for older cables).

I would suggest the first time you have to disassemble the splices that you install terminators and a bolted connection to allow easy disassembly in the future.

 

[benlanz]

busbar,

I would be happy to comment on VLF in regard to 400.2. I am not apart of this writing group but, I have sat in on numerous .2 meetings and I am familiar with the issues.

VLF hipot is better than DC hipot (on extruded cables) and better than doing nothing at all. Manufacturers claim they know the growth rate of defects and that simple hipot for a calculated amount of time will fail any potential defect. This would be a great idea if there was only one defect per cable and if all defects had a uniform growth rate. VLF hipot will fail week points in the cable but, it will also grow small defects and thus age the cable. On new cable the VLF hipot may fail defects, but gross workmanship defects will still survive the test to fail years down the road. The bottom line, VLF hipot will tell you the cable did not fail during the hipot and it might not fail as soon as it is switched back into service. Hipot does not measure the response of the cable and does predict future performance.

I wish I could say that VLF hipot was the solution but, it is only a simple tool with limitations.

-cheers

 

[benlanz]

EEbyChoice,

The losses in the motor are going be greater than order of magnitude larger than any serious losses in your cable. The only way to measure the losses, or polarization index, accurately is to measure the cable by itself. The polarization index worked great for paper cables, but it is relatively useless with extruded materials.

I recomend offline partial discharge testing since it is the only PD diagnostic that is backed by IEEE standards. For a testing companies just do a Yahoo search on 'power cable diagnostics'.

-Cheers

 

[benlanz]

Technical justification for the previous comments on VLF

VLF is rather new outside of Europe. The Europeans have been using this voltage source for 25yrs. The earliest developments were at the Technical University of Berlin High Voltage Institute.

Reference:
Research and Development, Annual Report 1992, TU Berlin High Voltage Institute
Prof. Dr.- Ing. Kalkner, Dipl. Ing. R. Bach

Kalkner and Bach show that VLF is exellent in growing electrical trees in XPLE insulation. 50Hz grows e-trees at 2mm per hour and .1Hz (VLF sin) grows e-trees at 12mm per hour(**). The VLF power source is excellent in quickly breaking down insulation. VLF is superior to DC because it can break down 'high impedance,' such as a knife cut. Since the source alternates polarity, space charge is not built up in a DC hipot. Aged cable will fail soon after application of DC hipot, e.g. >2Uo for >15 minutes. VLF voltage source is more expensive than a DC set but, far less than PD diagnostic system.

In short the VLF source is the best HIPOT available in the industry. However, in 25yrs of use the Europeans (especially the Germans who developed this voltage source) have yet to present a large scale study that shows that using VLF will improve the reliability, or predict future performance of a cable. IEEE 400 raises the following questions at the conclusion of the VLF section. (8.5)

1. At what voltage level can a defect be detected?

2. Is it possible to miss a defect which will fault when the cable is returned to service?

3. Does the test aggravate what was a negligible defect so that it will fault when the cable is returned to service?

These questions have never been adequately answered. Consider the following case study.

Background:
A utility performed a PD test on a 1000’ft long 25kV, 1000kcmil XLPE feeder cable. According to the PD test, the cable was condemned. The cable was condemned on the basis that there were too many defects to repair, and it would be more economical to replace the cable. The utility asked a manufacturer of a VLF source to perform a VLF hipot on the cable. If any of the condemned phases survived, the PD test would be repeated.

Test set up:
According to the manufacturer of a VLF source a 25kV class cable should be tested with 0.1Hz VLF at 3xUo for 30minutes. (3xUo =3x13.2Vrms=39.6kVrms). During this experiment the manufacturer explained that their equipment only produces a maximum of 23kVrms, but claims that only the peak voltage needs to reach ~40kV to have a good VLF test. The manufacturer claimed that if the cable fails under test that the cable is ‘bad’ otherwise it is ‘good’. They claim that only 3 to 4% of cables that test ‘good’ will fail within the next three years. The manufacturer advocates a repeat of the 30minute VLF Hipot after a failed cable is repaired to ensure reliability.

Test results:
Each phase was tested independently. A phase passed the test, B phase failed after 20 minutes and C phase failed after a prolonged dwell time of 37 minutes. A PD test was conducted on A phase to measure its condtion.



Observations
A phase was known to have 3 cable PD (defect) sites from the previous test. In the second test these three sites were made worse by the VLF hipot. In addition to these 3 sites 7 new sites showed up. In this case the VLF hipot made the cable much worse, without failing it. This cable would have been placed back in service in worse condition than before the VLF hipot. Another questionable point is that C phase failed after 37minutes. If the manufactures recommendation was followed and the hipot terminated at 30 minutes, this cable would have been put back in service with an imminent failure.

**Comments
The defect growth rate studies cannot be generalized to defects in cables. The experiment uses a needle pushed into cable insulation. This does not represent the electric stress distribution of can electrical tree that is in the presence of a water tree, the most common type of defect in aged extruded cable. This means that defects in cable do not grow at a uniform rate and the VLF hipot is destructive to aged cables. This has been demonstrated in a recent 3 year study including 151, 3-phase feeder cables.

 

[busbar]

Excellent information, benlanz. Thank you very much.

 

[jghrist]

Thank you Benlanz for an excellent post. You get another star from me.

I hadn't heard before of the problems with VLF testing. It seems to me that at least for aged cables, if you can't afford PD tests, you would be better off not testing at all. Would you agree?

 

[alehman]

Benlanz provided some great information. Clearly he is an expert. The following article has some other viewpoints on testing for discussion (keeping in mind the author's employer makes VLF test equipment). It would be interesting to hear some thoughts.

http://www.netaworld.org/files/ItemFileA504.pdf

 

[benlanz]

jghrist,

According to IEEE 400.2 VLF hipot is not advised on cables with multiple defects (section 6 table 2). This mean anything over 20 years old is suspect!

(6.3.4)
"When testing cables with extensive water tree damage or partial discharges in the insulation, VLF withstand testing may not be conclusive."

VLF hipot is better than DC on aged cable but, I would rather do nothing than VLF hipot aged cable. I question the true economic value of doing nothing vs. PD testing. If you really do the numbers you will see you really can't afford not to do offline PD testing. I am biased toward PD but, that is because it is the most effective!

 

[jghrist]

benlanz,

What are your views on the link provided by alehman

http://www.netaworld.org/files/ItemFileA504.pdf

that indicates PD testing will not find water trees and suggests using VLF at nominal or slightly above nominal voltage as a non-destructive test to measure tan delta?

 

[benlanz]

jghrist,

The statement you have quoted is true but, it is carefully worded to state a marketing position.

It is true that PD will not detect water trees directly. However, large scale studies show that PD can indirectly detect large water trees by locating PD coming from the electrical trees the create. Large water tree are the only ones which will cause electrical trees which lead to insulation failure.

After 20 years of use in Europe there are no large scale (statistically significant) studies that substantiate that claim of the effectiveness of Tan Delta in the field. According to the largest study of it kind (EPRI 1001892 -3 year, 109 field aged cable samples), Tangent Delta (Baur system), is effective at determining the 17% of cables that are 'critically aged' and the 4% that are 'like new'. The 79% which are 'moderately aged' present a scatter of data. Some fail right away, and others, with the same measurement, fail 30 years down the road.

Baur is a very good Austrian company. They are known world wide for their robust fault location technology. When the VLF source was developed, they were one of the first to market one commercially. They then developed a tangent delta system to use with the VLF. When the University of Connecticut Electric Insulation Research Lab demonstrated that PD could be detected in noisy field environment back in the late 80's, Baur went to a small company in Germany and purchased a system to attach to their VLF voltage source. Unfortunately the PD system has not been very effective when it comes to extruded cable. Thus they position themselves by saying PD cannot find water trees but, VLF can detect them.


Disclaimer
I work for a company that sells cable diagnostic technology and testing services. You will probably think that I am biased but, I hope my efforts to be as objective as possilbe will benifit you and your colleages.

-Ben

 

[benlanz]

IEEE 400 states

(9.5.2)
When a cable passes the dissipation factor test [tangent delta is the same thing], it is not possible to declare the cable insulation sound since a localized defect in a long cable may not be detected.

[my commment]

Compare this statement with IEEE 400 conclusion about PD -see my September 13th comment.

-Cheers

 

[alehman]

ben,
You are to be congratulated for providing this excellent and candid information.

 

[EEbyChoice]

Thank you to all for this information. There are several good links to other information contained in these replies and it is definitely worth checking them out.

Bottom line for me is that there is still some debate regarding the type of testing that is most effective for Medium Voltage cables (Although it seems like IEEE 400 has gotten a little ahead of both the NFPA 70B and NETA 2001 standards in their recommended testing).

Things that must be considered:

1) Type of cable being tested
2) Age of cable
3) Type of defect you are looking for
4) Risk tolerance level

I have not found a local contractor who does PD testing on MV cables, or has even begun considering it. I have spoken with two contractors, both still recommend a DC Hi Pot test (15 minutes at ultimate test voltage)combined with a Megger test and a shield continuity test. If anyone knows a reputable contractor in the Midwest who has experience with this technology, please provide contact information.

Thank you all again.

EEbyChoice

 

[alehman]

The company formerly known as Electro-Test is strongly promoting PD testing in the midwest U.S.

http://www.electro-test.com/

 

[benlanz]

Alehman:

Online tests will only see 3% or less of defects in extruded cable insulation and has been shown to be ineffective on paper insulated cables. Sometimes, online testing can find defective terminations and joints. Online tests are not calibrated and therefore cannot rightfully claim any rate of effectiveness, comparison to factory test, or IEEE standards. On the other hand they are very convenient and vigorously marketed.

I would recommend GE. They offer off line PD and dielectric spectroscopy, the combination of which has been shown to be >90% effective.

In the Midwest:
Michael J. Kasten
GE Energy
(262) 797-4909
Michael.Kasten@IndSys.ge.com

US wide:
GE Energy Services World Wide
Kerry Evans
Manager of Global Services
(678) 844-5654
kerry.evans@ps.ge.com