ML20059M753
| ML20059M753 | |
| Person / Time | |
|---|---|
| Site: | Diablo Canyon  |
| Issue date: | 08/17/1993 |
| From: | Kong A AFFILIATION NOT ASSIGNED |
| To: | |
| References | |
| OLA-2-I-MFP-018, OLA-2-I-MFP-18, NUDOCS 9311190287 | |
| Download: ML20059M753 (15) | |
Text
1
, 60- 2 13 3 2 3 - Oc 8 N?&WN i
c 3._ y p p _. j g 8[nl93 W a FEle66
. +.~,,.: .fh ,
'93 r7 28 F 6 n 5 Results of AnalyticalInvestigations to Determine The Root Causes of Medium Voltage Cable -
j Failures at Diablo Canyon Power Plant 2nd DRA FT. S/2?M.1 PreparedigI NUCLEAR POWER GENERATION BUSINESS l' NIT Diablo Canyon Power Plant integrated Problem Response Team Tracking AR # 0294742 l
Prenared by CU3TOMER ENERGY SERVICES BUSINESS UNIT
~
Technical and Construction Services ,
Gas and Electric Distribution Department San T:an:isco, CA Albert Kong ,
and CENERAL SERVICES BUSINESS UNIT l
Technical and EcologicalServices i Emironmental Engineering and Electrical and Mechanical Engineering Sections San Ramon, CA Richard A. McCurdy
.w u an ; , w.u co?.' cAn Verne L Wyman
> ~* u Q(iA._ on. .,; C b :o lP C 0-N.$
1 ~ &:
- u bum.1:ALandLi eMLLCC.-
/
" , , " ,_ . _. __ __ .._ _ 'r .:
- . v
-._.-._-.__.__. -i .._.__ _..-
t- .-
C' $ra YyI Db5 := {:ll:'l#> ~
! --, . i.c m 9311190287 930817 0" '-T
! 1 i PDR. ADOCK 05000275 ,pon -
, RentterJ. khI te h> icet o
! )
F DRAFT 4
Table of Contents .
Introduction i
SkV Cable Failure Investinations October 1989 failure - ASW pump 2-2 May 1992 failure - Bus 14D feeder cable October 1992 failure - ASW pump 12 15kV Cable Failure Investinations ' '
Chemical Analysis February 1993, CWP 1 1,15 kV Cable Failure .
March 1993. CWP 12,15kV Cable Failure, Follow-up Testing on Additional CWP 1 1 Cable Failure Water Sample and Severs' 4kV, Auxiliary Saltwater Pump -
Related Samples Diseussion Conclusions Recommendations pe . ,
J..
2
- . . - ~ . - . . - ~= - - . . - - .
4 'l DRAFT T-Appendices Appendix A, Okonite report No. 463, Evaluation ofSkVShielded Okoguard-Okoprene Power Cable Returned From PG&E Diablo Canyon - Unit 11, February 5,1990 and TES summary of j this report, dated June 1,1990.
Appendix B. TES, Laboratory Report No. 420DC-92.856, Diablo Canyon Power Plant 4kVLoad Center Feeder Cable Failure (Ref ARn A0266115).
Appendix C, Okonite report No. 483, Examination ofCables Returned From PG&E - Diablo i
Canyon. December 22,1992.
Appendix D, CTL, Inc. report No.93-029, Assessment of the Condanon ofa 20 year old SkV EPR Cable Agedin Service, Progress Report No.1, February 17,1993 Appendix E, CTL, Inc report No.93-055, Assessment bfthe Condition ofa 20 year old SkVEPR .,
Cable Agedin Service, Progress Report No. 2, April 9,1993 i
Appendix F - Results of Hydrocarbon Analysis Perfonned on CWP l-1, Failed Cabl- Jacketing Appendix G - Results of Thermal Analysis Performed on CWP l-1, Failed Cable Jacketing ,
Appendix H - Results of Analysis of White Deposit Found Inside CWP l-1, Failed Cable Jacketing Appendix 1 - Results of Analysis of" Yellow Liquid", Copper Shield Corrosion Products, Spare 4 Inch Conduit Pull Box Liquid and Miscellaneous Other Liquid and Solid Samples Submitted During the Week of March 23,1993, CWP l-2 Cable Failure (Also includes several ASW 1 .
and 1-2 samples)
Apper. dix J- Results From the Analysis of Samples Obtained by Altran Engineering on March 13, 1993, CWP l 215kV Cable Failure :
.v o.-
l l
J 1
3 ..
l
l
- 1 DRAFT !
't ,
Introduction The Diablo Canyon Power Plant facility has experienced five medium voltage cable failures, the first occurring in October 1989 and the last in March 1993. The first 3 failures were on cable with a SkV insulation rating and installed on 34,4160 volt circuits. Rese cables l prosided power to auxihary salt water pumps and bus feeder circuits. The last 2 failures I have been on cable with a 15kV insulation rating and installed on 34,12kV circuits. These
)
circuits power large condenser cooling water pump motors. The origin of all cables begin at j the switch-gear, located in the turbine building, and terminate at the cooling water intake - l structure, a distance of approximately 2200 ft.
)
He cables were all rnanufactured in 1972 by The Okonite Co. at their plant in Santa Maria, CA. They were installed in 1974 and have been in sersice since 1984. The SkV and 15kV -,
rated cables have the same design and are constructed of the same materials. The only i differences are their conductor size and EPR insulation thickness. The SkV rated cable has' l 115 mits ofinsulation and the 15kV cable has 220 mits ofinsulation. !
'l Beginning with the first SkV cable failure, the TES Electrical Test and Analysis Unit was l requested by the Plant Electncal Maintenance Group to determine the root cause for the failures Okonite offered their cable testing facilities, and expertise, to aid in determining the cause of the failure After the third failure of 5kV cable in October 1992, TES and plant personnel, decided to contract an outside independent cable testing laboratory (Cable Technology Laboratories, In1) to analy ze one of the three cables removed. The other two cables (one faulted and one good) were sent to the Okonite cable test facility for testing.
Following the first 15kV cable failure in February 1993, an integratedprob/cm response team was formed to focus a maximum effort in determining the causes for the failures. The team was compnsed of personnel from the plant staff, nuclear engineering, TES , an outside cable expert (ALTRAN), and a Company expert on underground medium voltage shielded cables. Okonite representatives, also volunteered their testing facilities, in New Jersey, for continued support of these investigations.
It was agreed by all team members, and Okonite, that the SkV and 15kV cable failure mechanisms were substantially different. He 15kV cables had undergone a chemical attack, evidenced by the deterioration of the neoprene jacket material, along with the underlying filler tapes and tinned copper shield tape. To the contrary, the SkV cables showed no immediate signs of chemical attack, and in all cases, the insulation had failed at a single pomt 4
DRAFT T*
Summary of SkV Cable Failure investigations October 1989 failure ASW oumo 2-2 Extensive testing at the Okonite cable test facility on sections of the failed cable, did not reveal a root cause for the failure. Electrically and physically the cable insulation tested in as new condition. He neoprene jacket material showed only normal signs of aging. Inspection of the cable insulation near the fault point, did not disclose any major voids that may have contributed to the insulation puncturing.
Okonite and TES concluded the railure os an isolated incident possibly caused by stresses placed on the cable during ir;tallation. No fbrth:r action was recommended, other than, consideration be given to replacing this c tble known to be in wet areas with eable that uses
~
"today's productions methods"(Okonite). Refer to Appendix A, for detailed Okonite report No 463, Evaluation ofSkVShielded Ohoguard-Okoprene Power Cable Returned From PG&E Diablo Canyon - Unit 11. February 3, )990 and TES summary of this teport, dated June 1,1990.
Mav 1942 failure - Bus la feeder cable No electrical tests were performed on the 35 foot section of cable containing the fault. During the process used to locate the fault point, most of the jacket material was removed from this cable.Ris was the only section of cable replaced after this failure and no other cable was available for testing.
Physical tests and inspection of the insulation and jacket materials were performed by TES and Okonite, but did not reveal a clear indication as to the root cause for the failure.
The failure was again attributed to an unknown or isolated incident. No further action was recommended, other than, suggesting a plan be developed to replace cable in a minimum of !
\
high priority circuits. .
Okorute did not issue a report on their evaluvion of the failed section of cable. Their verbal comments were included in the TES report o. uis cable failure. Refer to Appendix B, for TES, Laboratory Report No. 420DC 92.856, Diablo Canyon Power nant 4kVLoad Center Feeder Cable Failure (Ref ARM A0266115). l October 1992 failure - ASW oumo 1-2 Following this failure, TES contracted with Cable Technology Laboratories, Inc. to evaluate one good cable of the three 430 foot length cables removed The other two cables, one good ,
and the other containing the fault, were sent to the Okonite.
Okonite again found, that the physical properties of the insulation were normal and displayed )
mmimal signs of aging. The electrical properties were also normal for cable this age and vintage. They concluded there was no definitive reason for the failure. Refer to Appendix C,
]
for detailed Okonite report No. 483, Examination ofCables Returned From PG&E - Diablo Canyon. December 22,1992. ,
5 9
DRAFT i
Results of the CTL tests also showed that the cable was in good condition and the manufacturing quality was supenor for this vintage cable. It met all of the requirements specified for this type cable, with one exception; following the AC breakdown test, a microscopic examination of the EPR insulating near the point of breakdown, revealed several voids in the insulation. Dey ranged in size from 2 to 10 mils. The largest void measured was 10 mils. De original PG&E cable specifications (1971) allowed for no voids greater than 5 mils. Refer to Appendix D, for detailed CTL, Inc. report No.93-029, Assessment ofthe Condmon ofa 20 year old SkVEPR Cable Agedin Service, Progress Report No.1, February 17,1993 TES requested CTL to conduct additional breakdown tests on the remaining sections of good cable in order to substantiate the prs ce of voic's. Ir. addition, an accelerated water ;
absorption test was performed in order to determine if the dielectric constant and stability factor of the EPR insulation material had changed with age. The results showed these properties remained within current industry standards.
Results of electrical breakdown tests on three sections of cable following the water absorption test, showed a slight decrease in breakdown voltage, when compared to earlier tests with dry cable. However, breakdown levels were still relatively good for a 20 year old cable.
Following these additional breakdown tests, the dissections of each cable near their points of breakdown did reveal more voids.Re largest void in the EPR insulation measured 7.9 mils (see Figures I and 2). Refer to Appendix E, for detailed CTL,Inc. report No.93-055, Assessment of the Condstron ofa 20 year old SkVEPR Cable Aged in Service, Progress Report No. 2, April 9,1993 15kV Cable Failure Investiaations. February and March 1993 failures - CWP 1-1 and 1-2 5e results of these investigatioris are summarized together because the failures occurred less than six weeks apart and it was apparent that they had been chemically attacked by a common agent.
i Some electncal AC withstand tests were conducted at TES on three short len;. (~10ft) of l good sections of cable removed from the conduit leadin; to the switch gear (CWP l-1) in the j turbine building. All three cable were tested beginning at rated AC insulation withstand voltage of 35kV, and continued with increasing 9kV voltage steps, holding 5 minutes, while waiting for a breakdown to occur.
All three cables withstood a voltage level of 62kV without an insulation failure. The test samples could not be tested to breakdown because the 10ft sections were too short, and the terminating stress cones could not be made sufficiently long enough to avoid breakdown at the terminations.
No electncal tests were perfonned by TES on the cable removed following the March failure (CWP l 2). He plant staff decided further tests were unnecessary at this time.
6
DRAFT F, j Chemical Analysis During the course ofinspections immediately following the CWP l-1 and 1-2 cable failures, l
many samples were collected and submitted to TES for exanunation and analysis. The samples consisted of cable samples, insulation, and solid and liquid samples obtained from '
conduit and pull boxes. In all, more than 17 samples were submitted for testing. Samples were obtained, both by DCPP plant personnel and by an outside consultant (ALTRAN Engineering) he selection of tests performed on individual samples was largely based on looking for those chemicals which had the potential of causing degradation of the cable neoprene jacketing material along with limited information given on the identity of the sample. This was somewhat clouded by the fact that some of the samples may have been contaminated with substances used to aid in the removal of cable following the CWP l-1 cabic failure.
Other physical tests were also performed which assessed the degree of degradation of the cable jacketing.
What follows is a description of the testing and corresponding summary of results. Detailed results are presented in Appendices F through J, which consists of various TES C' : min!
Analysis Senices Unit laboratory test reports, specialized contract lab test reports, and other supporting data. This information was generated roughly during the time period in which the samples were received, however some were tested later, pending results from preliminary tests.
Also included with these test results are data from several Unit I and 2 Auxiliary Saltwater Pump (ASW) 5 kV conduit and pull box water sample analyses. Samples were submitted in March along with a set of the Unit 2 CWP samples.
CWP b1.15 kV Cable Failure Sections of failed cable were received at TES on 2/12/93. This included a portion nearest the point where the fault actually occurred.
Analysis of Cahic Material for Diesel Fuel and Other Hydroca.c.ns Portions of the cable and jacketing in close proximity to the fault were sealed in a heated l container (105'C). A sample of the head space gas was then analyzed for hydrocarbons, including diesel fuel, using gas chromatography. No detectable concentrations of any light hydrocarbons or diesel were found. Copies of the resulting chromatograms are shown in l Appendix F. There was an unknown peak that could not be identified, however, this was l attnbuted to some component present in the rubber formulation, i.e., an oil, plasticizer, etc. l Preliminary Examination of Failed Copper Shielding for CorrespondingTesting Personnel from the TES Corrosion Unit examined the copper shielding for conosion. ;
Preliminary X-ray fluorescence (XRF) scans were made on the bluish green deposits present.
Samples of the copper were also put under long teim bacteriological tests for sulfur reducing bacteria.
7 1
i
i
. l DRAFT ,
I 1,
Results of the XRF scans showed sigmficant amounts of chlonne, sulfur and oxygen present, strong indicators of chlondes and sulfates. Naturally, large concentrations of copper were also present as elemental copper, as mixed chlorides, and as the oxide.
Results from the bacteria tests were negative. It appears that no corrosion took place as a result of bactenal action. 4 1
l Thermal Analysis of the Neoprene Jacketing I l
Portions of neoprene jacketing were obtained from several areas of the cable: A so-called " good" section which visually appeared to have been unaffected by any physical stress or chemical agents (good tear resistance and the presence of exterior wax like ;
coating noted) a sample obtained within I foot from the fault, and fmally a sample taken approximately 4 feet from the fault.
The samples were submitted to the Rose Consulting Laboratory for analysis by Dynamic Mechanical analysis (DMA) and Thermo-Gravimetric analysis (TGA). All results are attached in Appendix G. Note that throughout all of the appendices there is reference made to the cables as both 12kV or 15kV. The results confirm that chemical changes have taken place in the failed portions of the neoprenejacket material which have led to the degradation. Rose, based on available inform.+ tion and earlier analyses, postulated that these chemical changes occurred from long term (a period of years) exposure to water while under mechanical stress.
At the time these tests were conducted, Rose was not given any additional infonnation concerning the cable operating conditions, temperature or other chemical data. This -
information suggests that other outside agents may have been involved in the degradation of the neoprene and that water was indeed not the sole cause for failure.
The significant point is that the thermal analysis results confirmed defmite degradation of the neoprene polymer and the cablejacket formulation.
Analysis of white deposits found inside failed neoprene jacketing Small quantities of white deposits, resembling white crystals, were observed oa the inside :f a portion of the neoprene jacketing which was distant from the section of cable near the fault point. These deposits were observed by scanning electron microscopy and analyzed by XRF.
Results showed the material similar in composition to a component found in the l LUBADUCT pulling compound. XRF results confirmed the criemical constituents were I also similar and that the material was partially composed of a type of glass or silica bead.
Results are shown in Appendix H.
l CWP 1-2.15kV Cable Failure. Follow-up Testina on Additional CWP 11 Cable Failure
{
Water Sample and several 4kV. Auxiliary Saltwater Pump Related Samples Sections of cable from the CWP l-2 failure were submitted to TES, beginning the week of March 8,1993. Additionally, a sample of yellow colored fluid taken directly
)
i from the failed cable interior was also submitted alcng with samples obtained from the )
pull-boxes. Finally, sections of stripped neoprene jacketing were also submitted.
8
DRAFT t-Analpis of Yellow Liquid (water)
A metals scan was performed using Inductively Coupled Plasma Spectroscopy (ICP).
Results showed significant concentration of sodium (2500 ppm), with a variety of other elements at lower concentrations, also present.
He sample was then tested for anions using ion chromatography, Test results show high concentrations of chlorides (3800 ppm) present.
The pH of the solution was found to be 7.5 with a conductivity of 12 milli Siemens.
Finally, the soluthn was sent to a contract labor:. tory for analysis by gas l
chromatography-mass spectrometry (GCMS). This technique is specific for many types of volatile and semi-volatile organic materials. Results showed no detectable l
concentrations (less than 20 to 100 parts per billion, ppb) of any materials except for
! 660 ppb of benzoic acid and 210 ppb of Bis (2-Chloroethyl) ether. Rese materials may be contaminants, impurities in the neoprene,acketing, semi-conducting tape or EPR rubber insulation which were extracted by the water.
i The DCPP Chemistry Dept. also performed analyses on a replicate of this sample.
I Similar tests performed by their lab showed concentrations of sodium, potassium, magnesium, calcium, chlorides, and sulfates in the same order of magnitude found by the lab at TES. The conductivity and pH were also similar and DCPP concluded that the water was similar in composition to diluted sea water.
Results for all of these tests are shown in Appendix !.
Analysis of Green Deposits Found on the Copper Shielding be deposits were analyzed quahtatively by XRF. Results showed sigruficant concentrations of copper, chlorir.:, sulfur a .d exygen to be present.
Subsequent analysis of the material by ICP showed large amounts of copper (40%)
sodium (2000 ppm) calcium (1.2%) aluminum (5600 ppm) iron (2000 ppm) tin (410 -
ppm) and zine (8900 ppm). IAwer concentrations of other metals were also detected.
lon chromatographic analysis of the dissolved deposits showed concentration of 10%
chlorides and .3% of sulfares.
The material appears (based on solubility observations and quantitative analysis) to be primarily a mixture ofimpure copper I and 11 chlorides and copper sulfate.
Results from all of these samples are shown in Appendix 1. in Lab Test Report #
420DC 93.308.
I 9
1 l
l DRAFT Analysis of Water Samp!cs from the Spare 4 Inch Conduit CWP 1-2 and CWP 1 1, Circuit u a ater These two samples were tested for metals (ICP). Results ahowed lower concentrations of sodium (600 ppm) than did the yellow liquid. Much lower concentrations of other metals were also detected.
The Circuit B water did contain high concentrations of aluminum (1800 ppm) and silicon (4800 ppm) which was not present in the Spare 4 inch conduit. Remember that this sample was collected following the CWP l-1 failure.
This same circuit B water also had a slight hydrocarbon odor. It was therefore analyzed for total petroleum hydrocC ns, benzene , toluene, and xylene. A small amount of hydrocarbon,250 ppm was detected in the crude oil range. This could have originated from a heavy weight gear oil or other similar origin. No other hydrocarbons were detected.
Anion analysis of these two samples showed lower concentrations of chlorides and sulfates (140 to 830 ppm) than the yellow liquid.
The pH of these two samples was essentially the same (J.6 - 6.8) as was the conductivity (3.0 - 3.5 Milli Siemens).
All other results for this series of tests are presented in Lab Test Report # 420DC.
93.308 which can be found in Appendix 1.
Analysis of Additional Samples submitted during the week of March 23,1993.
During this time period, a container with various water (including samples from the SkV ASW conduits and pull boxes) and sediment samples was submitted for analysis.
The samples were hquid and collected from:
- Conduit K0696, CMP l-2
- Pull box BPO25, Unit 2
- Pull box BPO33C from ASW 2-1
- Spare conduit K0671, ASW 1-1 (between pull boxes BPOl9 and BPO39B)
Sediment samples included material from:
- Pull box BPOl9, ASW 2-1
- Conduit K0670, ASW 1-1 between BPOl9 and BPO39B
- ASW 1 1, between BPOIS and 39B All liquid samples were analyzed as received for metals, chlorides, fluorides, sulfates, pH and conductivity. During the course of testing , it was noted that the CWP l-2 liquid sample had a unique odor. For this reason, it was tested for volatile and semi-volatile organics using GC/MS.
i 10
DRAFT d.
All results are also contained in Appendix 1. b 2eneral, the constituents found in the l-I and 1-2 ASW samples, were lower in concentration then those detected in the 1-2 CWP samples. Concentrations of sea water-related constituents (sodmm, magnesium, potassium, chlorides, sulfates, etc.) were significantly lower than those found in the CWP samples. Conductivity was also lower, while pH was somewhat higher, but still near the neutral range for most waters. The sample from BP025, Unit 2 was higher than all of the other samples, with a pH of 8.9.
The results from the GC/MS analysis of the CWP l-2 liquid sample did not show significant quantities of any organic materials other than trace (ppb) amounts of chloroform, Bis (2 chloroethyl) ether, dimethlyl phthalate, DI-N Butyl Phthalate and BIS (2-ethylhexyl) Phthalate.
The sedimr nmples wer: extracted with de ionized water and also analyzed for these same cons. ms. A portion of each original sample was also analyzed for total solids.
Additionally, quahtative XRF scans were run on each of the sediment samples.
i Results for these samples show a similar elemental pattern with lower concentrations than those found in the liquid samples. This was however expected since these san fes were simply extracted with water and then tested. The XRF scans showed similar results with some additional elements also present, namely zine and lead. They can probably be traced to black particles found in some of the sediment samples and are believed to be small pieces of cable jacketing.
No other major contaminants or unusual findings were noted during the course of this series of tests.
Samples Obtained by ALTRAN Engineering After the CWP 1-2 Failure .
Additional solid samples of various materials were obtained by ALTRAN Engineerinr3 on March 13,1993 shortly after the CWP l 212 kV cable failure and sent to TES. They consisted of the following:
. A portion of cable jacket from the faulted phase . We duct edgfof pull box BPO39B
= A portion of cable jacket from the faulted phase in the duct and nearer pull box BPO39B
= A wipe sample taken from the bottom of pull box BPO39B
. A mud sample taken from the sump of BPOl7A
. A flake of scale found in the duct of KO697 prior to cable removal XRF scans were taken from a wax like residue found on both samples of cable jacket.
Additionally, infrared scans were made on the residue using fourier transform infrared spectroscopy (FTIR). Results were compared to identical scarts made otta portion of 15 kV cable that appeared new and in good condition. Results showed that these residues were 11
DMR ;
identical and were similar in composition to a matenal called docosanol which is a waxy solid used in some lubricants. It is believed that this material may have been applied to the cable jacketing during the time of manufacture as a pulling aid.
The wipe sample from pull t .ix No. BPO39B was also analyzed by XRF and FTIR. Results show this material resembles simply grit or dirt with a small amount of organic matter.
Similarly, the mud sample is composed of the same type ofinorganic material as the wipe sample (XRF scan) No FTIR spectra were made on the mud.
Finally, an XRF scan of the flake obtained from the KO697 duct showed that it's composition was again similar to the wipe and mud san' ele inorganically, although sisually this material seemed to resemble a dried coating of some kind, perhaps dried cable pulling soap.
Discussion of Cable Failures ,
l SkV Cable The results of all the testing completed by TES, Okonite, and CTL did not reveal an absolute root cause for the three cable failures. The strongest evidence, suggesting a cause, is the voids found by CTL in the EPR insulation material following several AC breakdown tests. ;
I Afte :ompleting the additional AC breakdown tests, CTL found a total of 65 voids in the ;
I insulation material near the points of breakdown. Sixty-five percent of the voids detected were classified as being in the 2 to 5 mils size group and thirty-five percent in the 5 to 10 !
mits size group. The largest void measured was 7.9 mils (refer to CTL progress report No. !
2). It is possible that these voids were created by the test procedure used to electrically -
breakdown the cable insulation.
Voids in these size groups are not likely to cause partial discharge activity to develop at normal operating voltage. For example, a SkV rated cable with nomin9 i =:lation thi-kness of 115 mils and installed on a 4160 volt circuit, is stressed at only 29.5 volts peak (rms x V2) per mil when referenced to ground. Studies have shown that in order to develop partial discharge actisity in a 5kV cable, the voltage stress per mil would have to exceed A volts peak per mil.
This same study showed that the energy dissipated, in micro-joules per discharge, varies with the size of the void and increases as an exponential function of the void diameter. This being the case and assuming the resulting insulation erosion is proportional to the energy ;
dissipated, it was evident, that the larger voids not only discharge at a lower electrical stress, but also cause greater insulation damage with each discharge.
Other studies have shown that multiple discharges must take place within a void before they are sensed by a partial discharge detector. For this reason, a single void, larger than 10 mils and very symmetncal, may go undetected by the factory tests using standard commercial partial discharge detectors. One researcher stated ...Although many failures that occur are unexplained, they very probably could be caused by one or a few undetectable voids that are large enough to result in eventual cable failure due to partial discharge."
12
DRAFT a
Because of the nature of the SkV cable failures experienced at the Diablo Canyon Plant site, and the current research efforts mentioned, a few large voids or contaminants in the insulation may have possibly gone undetected before leaving the factory. Also important to note, is that all three failures occurred a few weeks or days after there had been some precipitation in the region. Standing water was observed by maintenance persons in the ,
pullboxes following the failures.
It is possible the higher humidity or wet environment conditions could have increased partial discharge activity already developing in a undetected larger void, thus accelerating detenoration of the insulation, and leading to its ultimate failure. However, no evidence was found to suppon this hypothesis.
As mentioned in the first TES summary report on a SkV cable failure (Appendix A), the other possibilities for causing the failures, but still without substantial indicators are:
. Electncal surges caused by switching operations.
. Isolated insulation defect that disappeared during damage caused by fault.
. A dormat failure due to improper field proof testing.
. Incorrect installation practices.
- Cable aging.
The investigatis e studies for all threc SkV cable failures have provided sufficient information to ruled out the followini; possible causes:
~
. Poor cable design and construction.
. Ses ere physical damage to cable, hke a cut in jacket that penetrates insulation.
. Overheating of cable due to self heating or high ambient temperatures.
. Chemical attack.
15kV Cable Throughout the course of testing, the goal was to try and determine if a chemical or other agent could be detected that had the potential of degrading or otherwise, destroying the protective neoprene jacket surrounding the cable.
No evidence or significant quantity of any one substance could be found that could have directly caused this degradation. Evidence of corrosion of the cable shielding was found (copper chlorides). However, once the cable jacketing had been affected, i.e., swelling.
l cracking, etc., then water could have entered the cable interior and proceeded to cause
! gradual corrosion of the shield. Also, most of the water samples thkt were tested had moderate to high conductivity. This caused the corrosive action to develop ever: more rapidly 13 i
I o I DRAFT ,
l
. 3 than if the water had been relatively pure. It should be noted that although the present. J sulfur was detected through XRF analysis, there was no indication of sulfides or formation l
of copper sulfide corrosion products. He presence of sulfur appeared to be in the form of soluble sulfate salts. ;
l i
There is strong evidence that the 15 kV cable has undergone changes as the result of some kind of chemical attack. However if, specific volatile or corrosive chemicals either in a gas or liquid state had been responsible for the attack, no evidence of any such materials could be found on the samples tested. It must be concluded that if they were present while the jacketing was being degraded, they had long since disappeared either by reaction, volatilization, dilution or self-decomposition.
Other theories concerning cable temperature and possible long tenn reactions with standing water, itself being "a chemical", should be considered. Results obtained so far do not preclude the chemical degradation of the neoprene jacketing caused by long tenn exposure to .
a confined quantity of saline water at elevated temperature over an extended period of time.
Further testing and analysis could be performed in the laboratory in order to verify this assumption. He design of such experimentation might include continuous exposure of the neoprenejacketing material to water at elevated temperatures over a period of several months. He water would be made saline with ionic const2uents matching the concentrations of those measured during the testing of the liquid samples obtained from the CWP cable failures. Following the exposure period, thejacket specimens would then be tested for elongation and tensile strength. Hermal analysis would also be performed to determine any chemical changes in the neoprene polymer and in the jacket formulation. Results would be compared to unexposed jacketing and also to portions of in service failed jacket material.
Conclusions SkV Cables Since there were no absolute deciding factors that clearly explained the causes for the failure of the 5kV cables, the establishment of voids and a cable failure relationship has not been proven he presence of voids suggests the dielectric strength of the cable insulation is weakened, however; the studies have shown that small voids are generally not a problem, because it is very unlikely that the size voids detected (2 to 10 mils), would result in partial discharge activity at the low in service voltage stress levels the cables are subjected.
It could be argued, that the weakest points of cable systems have been climinated by the in-service failures and their frequency will decrease with time. This hypothesis is supported by the high ac breakdown voltage levels of all the SkV cable samples tested by the two separate cable testing laboratories.
15kV Cables The results ofintensive chemical analysis investigations did not disclose the specific chemical agents that attacked these cables. It is possible the cables had been attacked 14
DRAFT
/'
scmetime earlier, by a strong chemical that has since disappeared either by reaction, volatilization, dilution or self- decomposition Recommendations SkV Cable In the absence of further studies, the voids can not be shown to have caused the failures. The root cause is still unknown. Based on the good test results, the cable will operate normally and immediate replacement is not necessary. The next cable failure can not be predicted, and it may be prudent to consider an emergency response plan and cable replacement in prionty circuits, in addition, future studies may be considered to understand the atfects of cable life and aging.
15kV Cable To prevent a future chemical attack and to assure the cable systems reliability, the cables and their environments should be accessed and monitored. A cable, ideally energized, might be installed in a spare conduit and removed after five years for examination and testing.
Hermal sensors could also be installed at strategic locations and monitored period cally. In addition, the enclosures or pullboxes emironmental affects could be monitored by simply placing a easily retrievable, unbreakable, clear container with several samples of the jacket matenals in it. Elongation and tensile tests could be perfonned periodically on these samples.
p , f. f F
s 15
_.