ML20058L549

Engineering Evaluation Rept,Electrical Inverter Operating Experience - 1985 to 1992

ML20058L549
Person / Time
Issue date:	12/31/1993
From:	Ibarra J NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD)
To:
Shared Package
ML20058L544	List: ML20058L549 NUREG-1455, Forwards AEOD/E93-03, Engineering Evaluation Rept, Electrical Inverter Operating Experience - 1985 to 1992. Findings Show That Inverter Failures Continue to Cause ESF Actuations,Reactor Trips & Turbine Runbacks
References
TASK-AE, TASK-E93-03, TASK-E93-3 AEOD-E93-03, AEOD-E93-3, NUDOCS 9312170071
Download: ML20058L549 (49)

Text

{{#Wiki_filter:. 4 AEOD/E93-03 ENGINEERING EVALUATION REPORT EI.ECTRICAL INVERTER OPERATING EXPERIENCE - 1985 TO 1992 DECEMllER 1993 Prepared by: Jose G. Ibarra Office for Analysis and Evaluation of Operational Data U.S. Nuclear Regulatory Commission YDR off I h}f06 PDR

.. ~ P CONTENTS r; 1. SUM 51ARY 1 e 2. DISCUSSION 1 2.1 Introduction 1 2.2 Data Sources........... 2 2.3 Vendor Maintenance Guidelines................. 2 { 3 2..t Nine Mile Point Unit 2 and Other Events... 2.5 Electrical Inverter Generic Communications 4 4 3. ANALYSIS AND FINDINGS..... 7 4. SAFETY SIGNIFICANCE.. 11 l 5. CONCLUSIONS............... 12 j 6. REFERENCES........ h i t TABLES Table 1 Events reported in LERs by category and number of failures..... 5 Table 2 Failed components....................... 6 t FIGURES Figure 1 Invener failures per vear - LERs................................ 8 I 9 Figure 2 Inverter failures per year - NPRDS. Figure 3 Number of plant events 10 t i APPENDICES l APPENDIX A - MAINTENANCE RECOMMENDATIONS APPENDIX B - INVERTER /UNINTERRUPTIBLE POWER SUPPLY FAILURES i APPENDIX C - INVERTER FAILURES BY PLANT (1985-1992) l l I iii I i

_.m g t ABBREVIATIONS [ i 4 AIT Augmented Inspection Team ASP Accident Sequence Precursor i BOP balance of plant i CDPr core damage probability . i 5 EDG emergency diesel generator ENS Emergency Notification System . j ESF engineered safety feature ESFA engineered safety feature actuation i HPCI high-pressure coolant injection HVAC heating, ventilation, and air conditioning j IE Inspection and Enforcement IIT Incident Investigation Team IN information notice i LCO limiting condition for operation LER licensee event report MFP main feedwater pump i MSIV main steam isolation valve i MTBF mean time between failures J. N M P-2 Nine Miles Point Unit 2 i NPRDS Nuclear Plant Reliability Data System l RCIC - reactor core isolation cooling l RPI rod position indicator RPS reactor protection system j SCR silicon-controlled rectifiers SCSS Sequence Coding and Search System l i TS Technical Specification i UPS uninterruptible power supply i P t V h

I L

SUMMARY

The total number of documented electrical inverter failures has decreased in the last 7-% years. Component failure continued to be the dominate root-cause. Human error j was the second significant root-cause. Capacitors were the most failed component. The other failed components were transformers, silican-controlled rectifiers (SCR), and l transfer switches. In 15 percent of the everm., the licensee did not identify the root-cause failure. Other root-causes were incorrect setpoints, lack of maintenance, and inadequate procedures. Electrical inverter failures caused engineered safety feature actuations (ESFAs), reactor trips, and turbine runbacks. The decrease in electrical inverter failures as identified in the licensee event reports (LERs) was due largely to three factors: 1) i better cooling of the units, 2) more preventive maintenance, and 3) more inverter l replacements. l The maiu sources of information were databases of LERs and component failure data, and two inspection reports. The Sequence Coding and Search System (SCSS) database l was used to locate the LERs describing inverter problems attributable to maintenance. The Nuclear Plant Reliability Data System (NPRDS) was searched to find the industry-wide inverter component problems. Three problems, addressed in two 10 CFR 50.72 reports and an NRC daily report, that were not addressed in LERs were reviewed. i An Incident Investigation Team (IIT) report and an Augmented Inspection Team (AIT) i report that investigated the losses of power and annunciators at Nine Mile Point Unit 2 (NMP-2) were also reviewed. l The maintenance recommendations of Exide Electronics, Elgar, and Solid State Controls from many sources were reviewed. These vendors recommended both periodic minor and major preventive maintenance. Minor maintenance is a visual and physical inspection of components. Major maintenance involves more vigorous verification tasks and component replacement. t i 2. DISCUSSION [ 2.1 Introduction Within a period of 8 months, August 1991 to March 1992, there were a total of 11 electrical inverter events reported to the NRC that either involved a loss of control room instrumentation, reactor trips, or ESFAs. Two of these events were serious enough for the licensees to declare a Site Area Emergency and an Alert, and for the NRC to 1 investicate these events in detail with an IIT and an AIT. P Case Study AEOD/C605," Operational Experience Involving Losses of Electrical l Inverters," December 1986, concluded that there was no significant improvement.of the number of inverter failures per year in the years preceding the study, and that high elevated temperatures was a major cause of failures. Due to the IIT findings at NMP-2, l and the 11 inverter failures in 8 months, the findings of the inverter case study once again became a concern. What is the operating experience with inverters 7-% years after l l i

i t i the case study? What preventive maintenance is performed on the electrical inverters? Are the same components failing? The purpose of this engineering evaluation is to i answer these questions using operating reactor data from January 1985 to October 1992. r 2.2 Data Sources AEOD performed a database search using SCSS for uninterruptible power supplies (UPS) and electrical inverters that failed in events or problems attributed to their maintenance in licensees' root-cause determination. A UPS includes both an inverter and a rectifier in one unit; for the purpose of this engineering evaluation, a UPS unit i was called an electrical inverter. AEOD/C605 had covered inverter problems through 1984; this engineering evaluation covers inverter failures after 1985. Thus, an SCSS l search found a total of 106 LERs, involving electrical inverter failures from January 1985 to October 1992, which became the prime information used in this study. l i An NPRDS search found 842 component failures in electrical inverters from January i 1985 to September 1992. This data was used to verify the decreasing trend in inverter failures as indicated by the LER data. The failures were summed per year and plotted. NPRDS has more reported failures since all failures, regardless of significance are reported. The purposes of the NPRDS and NRC SCSS databases are different, and therefore, most of NPRDS failures did not result in events meeting the NRC reporting threshold. The Emergency Notification System (ENS) that contains 10 CFR 50.72 reports was used to capture two events that were not covered by the LERs: Waterford Unit 3 ENS 24719 and Comanche Peak Unit 2 ENS 23451. In addition, the NRC daily report was used to identify Callaway event identified as MR 3-92-0332. 2.3 Vendor Maintenance Guidelines t The main event that prompted this engineering evaluation was the_NMP-2 IIT report, NUREG-1455," Transformer Failure and Common-Mode Imss of Instrument Power at i l Nine Mile Point _ Unit 2 on August 13, 1991."- The IIT concluded that the lack of periodic maintenance on the logic batteries coincident with degraded voltage caused the loss of five UPS units. This event, along with others, raised concern about the maintenance of the electrical inverter units. The 106 events reviewed involved several inverter vendors. However, this report reviews documentation from three of the vendors: Elgar, Exide,- l and Solid State Control. These vendors' technical guidelines were reviewed for recom-l mended maintenance on both safety and nonsafety-related equipment maintenance. L Safety-related electrical inverters have regular maintenance due to their equipment { classification, however, there are no mandatory maintenance requirements for the i nonsafety related inverters. Utilities usually provide maintenance for the nonsafety 'f inverters for economic reasons. These units are used for equipment categorized as important to safety. In addition, utilities try to prevent repetitive electrical inverter l 2 l ~ O

j L failures since such losses could lead to nonconformance with 10 CFR Part 50, j Appendix B. The new maintenance rule,10 CFR 50.65.(b), effective in 1996, includes l safety-related and nonsafety-related components "whose failure could prevent safety-related structures, systems and components from fulfilling their safety-related function or whose failure could cause a reactor scram or actuation of a safety related system," which includes many nonsafety-related electrical inverters. l t 2.4 Nine Mile Point Unit 2 and Other Events The failure of the electricalinverters can have potentially serious consequences. In addition to causing reactor trips, ESFAs, and turbine runbacks, the inverters provide power to equipment and instrumentation that aid in the plant recovery and emergency assessment capability. Three events, two at NMP-2 and one at Susquehanna l demonstrate these points. On August 13,1991, at NMP-2 a transformer fault occurred which resulted in the loss of five UPS units and a Site Area Emergency declaration. Control room annunciators, i balance of plant (BOP) instrumentation, emergency lighting, in-plant communications, and plant computer systems were lost. UPS maintenance had consisted primarily of l replacing air filters, recording meter and alarm indications, removing dust and dirt, and applying heat-sink grease to SCRs. If the UPS logic batteries had been fully functional j during the transformer fault, the UPS units would have maintained their power outputs. i Improved UPS preventive maintenance, which included logic battery replacement, would have precluded the event. The replacement interval for the logic batteries was stated in the vendor guidelines to be 4 years, but the operating environment reduced battery life. NUREG-1455 documented the detailed investigation by the IIT. On March 23,1992, at NMP-2, another loss of annunciators occurred due to dead UPS logic batteries. This event was initiated by a technician error which tripped one of the two lines supplying offsite power to the plant. A second human error caused total loss-of-offsite power. The power loss led to a loss of control room annunciators and the declaration of an alert. The event was further complicated when two running emergency diesels generators tripped. The cause'of the UPS failure was a failed internal logic batteg. These batteries were replaced following the IIT with a maintenance replacement cycle of 15 months but the batteries had failed after only 7 months. i Inspection Report 50-410/92-81 documented the investigation by the AIT. i On July 31,1991, with Susquehanna Steam Electric Units 1 and 2 at full power, loss of an offsite ac source coincident with Unit 1 pre-existing conditions, resulted in a scram i and main steam isolation valve (MSIV) isolation on Unit 1 and a half-scram on Unit 2. An UPS failure occurred on Unit 2. Upon loss of the normal power that provides preferred ac power to the UPS unit, the UPS unit automatically transferred as designed to its self-contained battery bank backup. But the battery bank had three failed cells that rendered the backup power unavailable. This inverter failure resulted in the loss of nurnerous indicators and instrumentation. High ambient temperature inside the UPS unit reduced the battey life from 10 years to 3 years. 3

2.5 Electrical Inverter Generic Communications Information Notice (IN) 87-24, " Operational Experience Involving Losses of Electrical Inverters," June 4,1987, includes the findings of AEOD/C605. The NMP-2, August 13, 1991, event led to issuance of IN 91-64, " Site Area Emergency Resulting from a Loss of Non-Class 1E Uninterruptible Power Supplies," October 9,1992. This information was later augmented by IN 91-64, Supplement 1, which described the UPS logic circuit and the decrease of the battery life when the ambient temperatures are higher than design. In addition to the above three ins, there have been other previous electrical inverter problems that resulted in generic communications, Inspection and Enforcement (IE) Circular 79-02, " Failure of 120 Volt Vital AC Power Supplies," January 11,1979, IE IN 79-29,"1.oss of Nonsafety-related Reactor Coolant System Instrumentation During Operations," November 16,1979, Information Bulletin 79-27, " Loss of Non-Class IE Instrumentation and Control Power System Bus During Operation," November 30,1979, and IE IN 84-80, " Plant Transients Induced by Failure of Non-Nuclear Instrumentation Power," November 8,1984. These generic communications document inverter problems ~ due to improper voltage, failed components, and the impact to the reactor systems of losing control and instrument power from the inverters. 3. ANALYSIS AND FINDINGS ~I Analysis of the 106 applicable LERs identified 128 specific failures which were grouped into 11 classification categories. Overall, the reported events show that the inverter failures caused ESFAs, reactor trips, and turbine runbacks. Table 1 lists the categories and the number of failures for each category. This includes both safety related and i nonsafety-related inverter failures because of similarity in equipment and the effects of the failures on safety related equipment. Component failure was the largest category (72) with human error the second largest i (15) category. The licensees were not able to report the cause of failure in 15 events. i Improper maintenance was identified as the failure cause for six of the events. In six of i the events, the inverter setpoints were found to be wrong or too restrictive for proper' inverter operation. Procedures were listed as the root-cause when no procedures existed to address the tasks or instructions were not clear within the existing procedures. A major concern of the AEOD/C605 report was that the inverter operating } temperatures were higher than the design temperatures which resulted in reduced inverter component life. This concern was communicated to the licensees in IN 87-24. Several LERs addressed this issue; corrective actions have included doing more periodic thermographic monitoring, upgrading the heating, ventilation, and air conditioning. (HVAC), and installing fans and vents. These corrective actions seem to have reduced the number of inverter failures due to heat. P ( I P

3 Table 1 Events reported in LERs by category and number of failures Order Failurt Category Failures 1. Component Failure 72 2. Human Error 15 3. Unknown Causes 15 4 Inadequate Maintenance 6 5. Setpoints 6 6. Procedures 4 7 High Temperature 4 8. Desien 3 9. Documentation 1 10. Training 1 11. Random Failure 1 l Design, documentation, training, and random failure were the remaining categories. Table 2 lists the components that failed in the order of largest number of identified failures. The list was compiled considering both safety and nonsafety-related inverters. 1 The capacitors are the dominating failed component, followed by transformers, SCRs. transfer switches, diodes, and inductors. The rest of the list is insignificant when compared to the total number of component failures. Electrical inverter vendors may have recognized the electrolytic capacitors failure dominance since the maintenance guidelines specifically address these capacitors. Exide Electronics, Elgar, and Solid State Controls maintenance guidelines were reviewed for maintenance recommendations. The inverter manufacturers divided preventive maintenance into a minor and major periodic maintenance schedule. The minor l maintenance is mainly a visual and physical inspection of the UPS units. These inspections are intended to detect degradation prior to component failure. Degradations, j such as capacitor leaks and component discoloration are typical precursors to failure. r i Major maintenance includes guidance on tests to verify the wave shapes, voltage levels, alarms, and equipment settings; and functional tests to perform on power transfer mechanisms, trip functions, and unit load conditions. The vendors also recommend periodically replacing components based on the results of environmental tests to account for premature aging. ] )

I These preventive maintenance recommendations have a direct impact on the number of failures in the failure categories in Table 1 and Table 2 with the exceptions of human error, procedure, documentation, and training which are plant specific. The preventive maintenance recommendations, if followed, would reduce the number of electrical i inverter failures. Several licensees' responded to the inverter failures by doing more preventive maintenance. In other instances, where high temperatures were a concern, i some licensees have started monitoring temperature and taken action to remove the heat - buildup. Related to this, there have been several events since 1985 where old electrical inverters were replaced with newer, more reliable models. These corrective actions appear to have reduced the number of inverter failures. j Table 2 Failed components i Order Component Failing Failures

• 1.

Electrolytic Capacitor 11 2. Transformer 9 3. Silicon ~ controlled Rectifiers 7 f 4 Transfer Switch 6 5. Diode 5 6. Inductors, Connections 4 Voltage Regulator Card 4 7. Power Fuse, de/dc Converter 3 i 8. Instrumentation and Control Board, 2 Oscillator i Current Limiting Card 2 f 9. Relay Coil, Transistor 1 Resistor, Power Supply 1 Insulation Breakdown 1 l Bridge Rectifier, Filters 1 Breakers, Cooling Fan 1 Gating / Sequencing Card 1 Number of failures for each component listed. A summary of each vendor's maintenance guidelines is in Appendix A. The 106 events that were the basis for this evaluation are summarized in Apperidix B. Each summary describes the specific electrical inverter problem. Inverter failures at individual reactor anits are listed in Appendix C. l l 6

e '4 l Inverter failures were the direct initiator of the event 94 percent of the time. Of the 106 reported events, one-half of the inverter failures caused an ESFA and other half led to reactor trips. Turbine runbacks forestalled reactor trips 5 percent of the time. In 69 percent of the events, the reactors were at power. In 6.6 percent of the events, the reactor was in standby. The reactors were in shutdown mode 16 percent of the time i when the failures occurred. In 7.5 percent of the events, the plants were in a refueling mode. In one event, the plant was still under construction. In three events, the equipment was declared inoperable and there was no actual failure of the inverter. i LER data plotted in Figure 1 shows the number of failures per year is decreasing and may have leveled out in the last four years. A simple regression analysis confirmed the downward trend for the 7-% years of data. The AEOD/C605 case study found 50 to 60 failures per year. Despite improved search methods, these new failure rate figures are considerably less and show a positive improvement. To double-check the decrease ofinverter failures, a NPRDS search of inverter failures was done. Figure 2 is a plot of the NPRDS reported failures per year. This includes i incipient failures. No attempt was made to review these failures in detail. NPRDS data has more reported failures since both significant and insignificant failures are reported to the NPRDS database. Figure 2 also shows decreasing failures. A simple regression analysis of the data also confirmed a decrease in failures reported to NPRDS but the decrease wa.s not as significant as that shown LER data shown in Figure 1. j Figure 3 shows the number of inverter failure events and the corresponding number of plants that experienced those number of events. A complete listing of the plant unit is given in Appendix C. Only one inverter failure was experienced in each of 37 plants during the evaluation period. No plant experienced five inverter events, and only one plant experienced six inverter events. Multiple unit sites usually have the same equipment vendor, so the problems experienced on one unit usually will occur sooner or later in its sister units. The compiled data averages one inverter failure per month over the last 4 years. There i were 11 inverter failures, during the 8 month period between August 1991 and March 1992, which is not considered to be statistically significant, especially since three failures were at NMP-2 with one vendor. 4, SAFETY SIGNIFICANCE Inverter losses can lead to failure of 120 V buses and the associated instrumentation and i controls, Between 1985 and 1991, there were six Accident Sequence Precursor Program (ASP) calculations involving inverters and/or instrument bus losses. The conditional core damage probability (CDPr) for five events ranged from E-5 to E-4. The ASP program calculated the CDPr of a reactor trip to be between E-7 and E-6. Of the ~ approximately 200 reactor trips per year,5 could be expected to be due to inverter failures, and thus, would typically not be a significant contributor to core damage. 7 j i

A I i l j. ) IWERTER FAILURES L 1 1 LER Data to October 1992 1 i t e i ^ 25 " t i i i I I i l t i l 1 a ,p a. 20dit!"td! t i 6 :s l b 4 s i ',t;l .,,s l I j sup% 4 i m t 1 l [yl'o @d i i 151 Qg[ g[ih@;;d@W h& 'I'$$1 g+gwg' l ?aA4 W i s .e~, c n sD @pcg\\?;)g i:y gN ? git v Ne t;, 'ss u::a g"n g 1 pn ou b,;w7:;;r ;4 8: g r @4:q q 8 gn. 3; g g [ Q @gj 'M ti y/ l&g} d" n ana 1 0, h~.' a'@ [W;eM"f kP IFjftTl fiQY N'

\\

% ti ygg jpgpf gygg pf,g@d / f a ' - d' g[. ' lk;w$M j Qi i ? d $$b sd$e $g$pp![s $@$l $[4 4MW l es ! a JJ lig e $a, J gg k u t: p d< 4 m: w pl a R M d +gpg

g d r ge 4,s s typg ge s y ;;

j:r:p w; g y r g $g,g' A g.,'s.wid gp p, g;gg, s -J ..u-g f

L GN.wej W.

e ..J,gpsa, i

,ygis,,

, :., o :q..eseg 2;gep w 2 7j l .a. 3 N ji

?

• d 5,

l b:t r mgqq

ve -

w w: n-g gpw: h K I I lk h -I $l[g h lg hf) -l$M f2 h if ,.l le l,j MM1 I r i;k$,$ 4l 'i e%A -MMhb4@..A Hb::;,j --?)7:4Mgp%,9 s 4"m rasq en l-w 3 r Os O i 4 ^ 1985 1986 1987 1988 1989 1990 1991 1992 f f Figure 1 Inverter f ailures per year r LERs e f l 8 5 C

~ . i. I i I INVERTER FAILURES NPRDS Data to September 1992 1 1 i 1 i s t I r 140~' 1, i i i i i i \\ b J I a l k / W md. l 6. 120 tw,a 1 i n~, m~ t*%vj wa = .wsrud tg:= 4@, i i 2 t:, pp:a, t o:,E1 ...-:w q: Mi ( i t i:;gn hq [ g c. g,... . 'ga,. , h, m, .. g n .nw:c ev.; ,a. ['4,2'+f' dam] ]gA{qu gg$el t jg [;qq1j 6 s sw e 00 4 l t mgW,- I t u-e .m

j we nm a

se teg'gp;l,Ags i } s s. uyp '; Mw::

• L.

y

• wp..

t s

sy)g a. wp:. W: gg;~ w:e e, m;b

emw

/ l 2nm'N ( 4:3 c:

$. y r.

w y$&qp h

hy,i; k'

;l

IMg, i

>%[f

G:

c' ? 1 (ddq# 80 J Qh : Q g { g hMi i l Rh W R$ W n i'i

1 $1 1

h h h Y'k h hkh -h j b hMI 'l l h !$%g) 40 - lg_DyPI) I j'f @sb,sm:ti*hgt

p. g~,j(dy, j, ( g,j d

.g gg g g gp l mm m 1 y f 5,l t[EfM.h;'] [ql%g%x:$(' lag:A, tygg .g ; 1 t '.3 " :m S.:,1,RN I, f.4 Y $ cow,gk1 tp:;,ij j C 3.,,.g:\\'. y.# y,i t@ gj;! q., p ,m.g, oef s 4 s t,$,3$c p" ,i J3: en,sa, p, , p$'s. ; a w-w g taw.. , y: 1 N hmy y 6:3 t, .o a n N' w.g.y y,,l g Ai '2' %\\' hv W J If L ')i; wh T, '1

.,0'f

,ed e, e a 20 a hfL MAR ? Ja ( P' s : Says ta j p-i s t h s s i ts " !i i. s j f, '1j o:,4-@y a .' l

y, y

&i c. ke l to e w

1. sy a

av ' 3 8 g .;4y, s e y,g, ' ? i d. f Lp s t s y 4

• j%)

Fg i gh? s cg_,m t-b$AE W /-{, L,H M,-L jj W$m*,$,-; 2WQQ$4 J e l o 1985 1986 1937 1988 1989 1990 1991-1992 i i 1 1 Figure 2 inverter f ailures per year NPRDS I .1

+ b I l l i INVERTER FAILURES Data January 1985 to October 1992 i i l f 50 m I i f 5 I. i f f I i 40 i ! i i I N p; - u 4 U t t I m i s i b k. i! i e e t 30 - E' i i r y l o L i f 1 I i l P b l a I J: l l a 20 I i n 4 i t t y l s [ i - i i j 1 / / + f' 'l h f i 4 1 10 - ii d L ) { i ...W / i t j s m m ,/ -l + 1-i t isagen i 4 5 m 1..,N / .. J. 1/ . / // 0g i i I 1 2 3 4 5 6 j 1 l Number of Events i Figure 3 Number of plant events i i 10 t

~ However, common cause inverter failures can create risk significant events due to the loss of both equipment and control room instrumentation. Loss of both compounds problems for the operators during recovery. The five simultaneous inverter failures at - NMP-2 (LER 50-410/91-017) quantified in ASP as having a CDPr of 3.8E-4, was 1 modeled as a loss of feedwater and failure of a transformer, with two trains of low pressure core injection and one train of residual heat removal unavailable. The NMP-2 ever* also demonstrated the human contribution that is not quantified in present probabilistic risk assessment models; there was initial confusion in the control room j when annunciators, computers, essential lighting, rod position indicators, and communications were lost due to the loss of the inverters. Exide estimates the mean time between failures (MTBF) for its manufactured units is l 150,000 hours if preventive maintenance is performed and the configuration includes a i maintenance bypass. The 150,000 hours MTBF is illustrative of electrical inverters of the i size that are covered in this report. AEOD/C605 case study calculated 56,000 hours MTBF in 1986 based on LERs.- This case study noted the calculation only included reported inverter failures. Using the current data for the 7-% year period, the calculated l MTBF for the LER data is 258,000 hours, and 31,000 hours for the NPRDS data. The 258,000 hours MTBF figure only includes loss of inverter events that meet the LER j threshold criteria. Hours between failures would decrease if all the inverter failures within the period were counted. The 31,000 hours MTBF figure includes all failures, i incipient as well as total failure. Therefore, this MTBF figure is also skewed but closer to a realistic figure. These calculations demonstrate that real operating experience l reliability is not well known. Taking the more realistic MTBF figure, there is still room for improvement before the licensees approach the manufacture's MTBF goal. l r -) 5. CONCLUSIONS j From August 1991 to March 1992, there were 11 events of electrical inverter failure that j caused operational concern. The most serious event was the NMP-2 transformer and i common-mode loss-of-instrument power which was investigated by an IIT. The IIT l i concluded that the preventive maintenance of the UPS units was deficient in replacing logic batteries. This study includes evaluations of 106 events containing 128 electrical inverter problems associated with mamtenance. j Component failures were dominant and accounted for more than 55 percent of the losses i of electrical invec'ers. Capacitors, transformers, SCRs, and transfer switches were the components that faiwd most often. The second most dominant failure identified was l human error. Of equal quantity with human error were the unidentified failures. In j these cases, the licensee investigations have not yielded a root-cause of the electrical l inverter problem. Other root-causes are inadequate maintenance, setpoints, procedures, high temperatures, and design. l i The total number of electrical inverter failures has decreased in the last 7-% years. ~ Electrical inverter failure is causing an ESFA one-half of the time, and a reactor trip the ~; other half of the time. The LERs identified three factors that have contributed to the failure decrease: 1) better cooling,2) more preventive maintenance, and 3) more i 11 J ...-c- ., - ~ ~

i I frequent inverter replacements. There have been numerous generic communications. over the past 13 years that address the electrical inverter failures. The review of recommended preventive maintenance by the three representative vendors found that maintenance is divided into minor and major maintenance. Minor maintenance is a visuai and physical inspection of all the components. Major maintenance is a more vigorous task of verification, measurement, and component replacement. The recommended practices could reduce the number of failures because the degradation of the components could be detected before the components fail. However, current regulatory requirements which impose regular maintenance affect only the safety-related electrical inverters. The nonsafety electrical inverters may be covered by the new maintenance rule that will be in effect in 1996. 6. REFERENCES 1. F. Ashe," Operational Experience InvoMng Losses of Electrical Inverters," AEOD/C605, December 1986. t 2. U.S. Nuclear Regulatory Commission, ' Transformer Failure and Common-Mode Loss of Instrument Power at Nine Mile Point Unit 2 on August 13,1991," U.S. NRC Report NUREG-1455, October 1991. 3. U.S. Nuclear Regulatory Commission, Information Notice 87-24, " Operational Experience Involving Losses of Electrical Inverters," June 4,1987. i 4. U.S. Nuclear Regulatory Commission, Information Notice 91-64, " Site Area [ Emergency Resulting from A Loss of Non-Class 1E Uninterruptible Power Supplies," October 9,1992. 5. U.S. Nuclear Regulatory Commission, Information Notice 91-64, Supplement 1, " Site Area Emergency Resulting from A Loss of Non-Class 1E Uninterruptible i Power Supplies," October 9,1992. j 6. U.S. Nuclear Regulatory Commission, IE Circular 79-02, " Failure of 120 Volt Vital AC Power Supplies," January 11,1979. 7. U.S. Nuclear Regulatory Commission, IE Information Notice 79-29, " Loss of f Nonsafety-related Reactor Coolant System Instrumentation During Operation," i November 16,1979. 8. U.S. Nuclear Regulatory Commission Information Bulletin 79-27c" Loss of Non-Class 1E Instrumentation and Control Power System Bus During Operation," l November 30,1979. 1 12

-l t 9. U.S. Nuclear Regulatory Commission, IE Information Notice 84-80, " Plant Transients Induced by Failure of Non-Nuclear Instrumentation Power," November 8,1984. 10. Inspection Report 50-410/92-81, "NRC Region I Augmented Inspection Team (AIT) Review of the March 23,1992, Nine Mile Point Unit 2 Loss of Offsite Power and Control Room Annunciator," April 10,1992. i t t i i ) r h 'f i i i i f I i .. j 'i ? } l 13 7 e w-et w-- w

.Ls h k a-4 t APPENDIX A - MAINTENANCE RECOMMENDATIONS i a h f I I L f I i i-6 f b r Y e k a P l [ r l I k I t )

F I APPENDIX A Maintenance Recommendations This appendix describes a summag of maintenance guidelines recommended by Exide Electronics, Elgar, and Solid State Control. The maintenance recommendations are specified in the technical guidelines reviewed for this technical review. l EXIDE ELECTRONICS ANNUAL RECOMMENDATIONS a Shut down the unit and inspect the physical condition of all components and connections. Inspect all electrolytic capacitors for leaks. Operate unit with low load or no load, and check the logic functions. l Verify all manual settings to assure no changes which jeopardize UPS reliability. Verify the all wave shapes wherever applicable. Connect a variable UPS load and adjust load to verify corresponding settings response. Check all the trip functions, fuses, and overload conditions. Verify the transfer mechanisms for alternate power are operable. Exide recommends minor maintenance on their UPS at a frequency that depends on the l environmental conditions in which they operate. Customers should check and replace j the air filters when dirty, check for discolored components such as resistors or capacitors and verify that the D-cell batteries function properly. f I A1

i ELGAR MONTHLY Inspect intake air filters, clean or replace as required. Inspect batteries for overheating, tightness of all electrical connections, and leaks. Visually inspect connectors and printed wire assemblies for overheating, tightness of connection, and damage. Inspect fans for proper blade rotation and freedom from obstruction. Measure and verify or adjust specifications: charger output voltage inverter output voltage and frequency inverter synchronizing circuitry transfer switch operation alarms and indicators inverter output over-and under-voltage settings ANNUA.LLY Verify specifications for the torque on the silicone controlled rectifier mounting i bracket bolt and flatness. BIANNUALLY Replace fans. l EVERY 9 YEARS Replace all oil-filled and electrolytic capacitors as noted in manual. In some cases, replacernent interval may be 7 years, refer to manual. To perform most of this work, the maintenance personnel must have a thorough knowledge of the unit and its system. ELGAR recommends contacting them before i performing the task if the knowledge and equipment is not available at the site, I h l A-2 i

i SOLID STATE CONTROL Replace maintenance components at the required intervals to maintain the IE system integrity. Replace in accordance with the manufacturers recommendation's or as determined by the Wyle Laboratory aging analysis. include circuit boards in the component replacement schedule. Replace e components at the designated intervals in lieu of replacing the complete circuit board. Replacement of components should be by a qualified technician, or the boards returned to the manufacture for the replacement work. Inspect fan motor strain relief busing for wire entry on 2 year intervals and replace if deterioration is found. Lubricate fan motor bearings on 2 year intervals if bearing oil cups are present. Clean the system of dust and debris. Pressurized air to 40 psi maximum may be used to clean the system. Do not use cleaning agents containing hydrocarbons on the components or in the system. Components with a life less than the qualified system life are being operated at a margin below which component was analyzed for. Thus, the design life and the replacement period are the same time periods. The fans are not required to maintain the proper inverter operation. s r W I 1 i [ i i A-3

T 4 l 1 APPENDIX B -INVERTER /UNINTERRUPTIBLE POWER f UPPLY FAILURES { b i t t ? I F t f a s

APPENDIX B Inverter /Uninterruptible Power Supply Failures January 1985 to October 1992 DATE PIANT DOCKET /LER NUMBERS 01/01/85 Peach Bottom 3 50-278/85-001 With the unit in mode 1, the inverter supplying power to high-pressure coolant injection (HPCI) instrument failed. Investigation of the Topaz inverter model N250-GWR-125 l 115, found three failed 14K microfarad capacitors. 01/10/85 Limerick 1 50-352/85-007 With the unit less than 5 percent power, a half-scram signal was initiated on one reactor I protection system (RPS) channel at two different times. The events occurred as result of a temporary loss of power to the 1B RPS and UPS 120 V ac distribution panel. In each event, a faulty voltage regulator card in the IB static inverter caused an overvoltage condition resulting in the trip of the electrical supply breakers to the electrical panel. 01/26/85 Cook 2 50-316/85-003 With the unit in mode 1, the reactor tripped due to the failure of a vital instrument inverter. The cause of the failure was not determined. All suspect components were replaced. A design change will replace existing inverter with more reliable inverters. 02/08/85 Limerick 1 50-352/85-024 With the unit in a cold shutdown, a half-scram on the 'B' RPS channel occurred along with the various outboard NSSS subsystems isolations._ The event occurred as a result of temporary loss of power to the 1B RPS and UPS 120 V ac distribution panel. Voltage fluctuations from a voltage regulator board in the static inverter caused an overvoltage condition resulting in the trip of the electrical supply breakers. The voltage fluctuations were caused by high ambient temperatures in the static inverter cabinet. Hoods and fans were installed in the top of the inverter cabinets to circulate the air and reduce the temperature in the cabinet. 02/28/85 Limerick 1 50-352/85-026 With the unit in a cold shutdown, a half-scram on the 'A' RPS channel occurred along with various NSSS subsystems isolations. The event occurred as the result of a temporary loss of power to the 1A RPS and UPS 120 V ac distribution panel. Voltage fluctuations from. a voltage regulator board in the static inverter caused an overvoltage condition resulting in the trip of the electrical supply breakers to the panel. The voltage fluctuations were caused by high ambient temperatures in the static inverter cabinet. B-1

i Hoods and fans were installed in the top of the inverter cabinets to circulate the air and reduce the temperature in the cabinet. M/07/85 Shoreham 50-322/85-013 l t While in mode 4. the unit experienced failure of the HPCI due to an inverter loss. The i cause of the inverter failure was a cooling fan which is used to circulate air through the l cabinet to cool the inverter and its circuitry. That heat built up was sufficient enough to i cause a capacitor to fail and cause failure of the inverter. Components replaced were the fan, capacitor, and resistor. N/25/85 Oconee 1 50-269/85-007 With the unit in mode 1, the reactor tripped when static transfer switch on inverter failed. 05/23/85 Quad Cities 1 50-254/85-004 i With the unit in mode 1, a half-scram was received when an inverter failed twice. The cause of inverter loss was two failed transistors in the logic power supply. 05/25/85 Byron 1 50-454/85-053 With the unit in mode 1, th; eactor tripped when an instrument bus inverter failed. The inverter failure was due to filings shorting the winding in a transformer. [ 06/06/85 Turkey Point 4 50-251/85-013- -r With the unit at full power, the reactor tripped when an inverter supplying power to a instrument bus failed. Investigation found a ground in the input filters on the supply side of the power fuses in the NIS drawer. 07/16/85 Turkey Point 3 50-250/85-018 While in hot standby condition, a vital inverter supplying an instrument bus tripped. Investigations could not determine the root-cause. The inverter was inspected and checked per maintenance instructions. A fuse was found blown but was not considered the root-cause of the trip. The on-going corrective action is to replace the inverters with ' a different manufacturer. 07/17/85 Turkey Point 4 50-251/85-019 With the unit at full power, the reactor tripped due the loss.of a vitalinstrument power from an inverter. Root-cause failure was a current limiting card. B2

i i i I i 08/01/85 Turkey Point 3 50-250/85-023 l With the unit in mode 1, the reactor tripped due to icyrter failure supplying a vital l instrument panel. The failure root-cause was the oscillator and logic power supply. The j voltage regulator and synchronizer, dwell angle control and oscillator and logic power supply modules were replaced. A fuse was found blown and replaced. l 1 08/02/85 Duane Arnold 50-331/85-031 r i With the unit at power, the reactor core isolation cooling (RCIC) system was declared inoperable due to the failure of an inverter. The static inverter was seen emitting smoke - after control annunciation alarms. The smoke within the static inverter was oil which had leaked from a capacitor. The inverter is a Topaz inverter model 125-GW-125 60-115. A preventive maintenance procedure is being developed and will be scheduled every refueling outage. i 08/02/85 Farley 2 50-364/85-01.2 With the unit in mode 1, the reactor tripped due to failure of vital inverter. One inverter of the same channel was already in a trip condition for maintenance. Initiating i failure of inverter was a ferroresonant transformer. The voltage output was reduced to 60 Vs instead of the nominal 120 Vs. l 09/25/85 Grand Gulf 1 50-416/85-038 } i With the unit at full power, a reduced inverter output voltage resulted in actuation of the standby gas treatment and primary and secondary containment isolation. The inverter i output voltage reduction was due to a failed transformer. Root-cause was insufficient loading of the inverter. The vendor recommends operating inverter over 12.5 amp equivalence but the failed inverter was normally loaded to less than 10 amps. Corrective action is a design change to add greater loads to the 120 V ac engineered safety feature (ESF) inverters. 10/07/85 Turkey Point 3 50-250/85-031 With the unit at full power a turbine run back occurred. This was initiated by. personnel error due to inadequate procedure that caused the cycling of a breaker. The cycling of t the breaker generated the failure of an rod position indicator (RPI) inverter. The root-cause of the inverter was a electrical spike from a relay coil. Fuses were blown. A procedure change will be made to caution cycling of breaker. 1 E 10/12/85 Cook 1 50-315/85-052 ? While in a cold shutdown, a failure of the critical control room power inverter cause ESFAs. Investigations found shorted diodes and two blown fuses which were replaced. { In addition, other diodes, SCRs, and capacitors that may have suffered degradation as a 1 result of the incident were also replaced. 3 B-3 -j a

e s 10/26/85 Cgstal River 3 50-302/85-023 With the unit in mode 1, a turbine / reactor trip occurred following a 120 V vital inverter failure. The inverter had been in an undervoltage condition prior to the failure causing i erroneous indication and leading the operators to trip the main turbine. Though taken i prematurely, the operator actions were conservative for the indications. The post trip investigation determined that a defective oscillator board caused a low voltage failure. 11/13/85 WNP-2 50-397/85-059 With the unit at fifty percent power, a reactor scram occurred due to the failure of power to the feedwater control system. The feedwater control system failure was caused l by loss of critical instrument power inverter IN-3. This resulted in the failure to supply power to one of two critical instrument buses. While troubleshooting inverter IN-3, the inverter was inadvertently shutdown when the normal operating procedure was used to transfer inverter power supplies. While continuing to troubleshoot inverter IN-3, a test lead was accidently shorted to ground which caused a fuse in the power supply to a critical instrument bus to blow. This caused a second automatic start of the ESF system and a partial containment isolation. 12/12/85 Kewaunee 50-305/85-023 With the unit at full power, a turbine / reactor trip occurred due the failure of a vital bus. l

1. ass of an inverter generated ESFAs. The root-cause was the failure of a constant j

voltage transformer. l 1 01/03/86 Zion 2 50-304/86-001 While in a refueling outage, a mnmentary voltage fluctuation on the output of an inverter power supply caused a power-operated relief valve lifting. The cause of the output voltage fluctuation was not determined. F 03/11/86 Duane Arnold 50-331/86-005 i With the unit in mode 1, the RCIC was declared inoperable due to inverter failure. investigations found the output setpoint low for the trip setting. In addition, the de battery input to inverter was high due to battesy equalization. The root-cause was inadequate maintenance instractions. } 04/11/86 Turkey Point 3 50-250/86-017 { Turbine runback to 93 percent from 100 percent when RPI failed inverter swapped + power supply from normal to alternate supply. Investigation determined a random failure of the output transformer. The RPI power supply is a Westinghouse 5 kV A inverter model No. ESELIV. A failed power diode is a Westinghouse model l 2057A59H01. r B-4 l l

l 04/13/86 Wolf Creek 1 50-482/86-022 With the unit in cold shutdown, a control room ventilation isolation signal occurred due to a loss of an instrument bus. An inverter had been de-energized to facilitate battery testing and upon re-energizing a fuse blew because of a faulty transistor. I 04/17/86 Catawba 2 50-414/86-006 With the unit in mode 4 (hot shutdown), a vital instrument bus lost power due to an inverter failure. The root-cause of the inverter failure was an inductor had burned out. 05/10/86 1.2 Crosse 50-409/86-016 While in hot shutdown, the 1A static inverter failed and transferred power to its alternate source. The 1A inverter was replaced with a spare. Reactor tripped three days later when same inverter transferred power again to alternate source and inverter 1C blew a fuse. Inverter 1C caused loss of emergency phone circuits. Failure of inverter 1A l was determined to be an overheated resistor. The interaction between the two inverters l was the phone circuit. 06/03/86 Point Beach 1 50-266/86-003 With the unit at full power, the reactor tripped when power was lost to white vital instrument bus. The inverter breaker tripped when after maintenance the swing inverter was incorrectly placed into operation. The root-cause is personnel error. 06/03/86 Point Beach 2 50-301/86-003 While at 100 percent power, the unit experienced a turbine runback due to the loss of the white vital instrument bus. The white inverter output breaker tripped due to personnel error. The breaker trip was caused when the swing inverter was incorrectly restored to operation on the de bus. 06/09/86 Limerick 1 50-352/86-029 With the unit in mode 4, an ESFA occurred when transferring the power supply via the static inverter transfer switch. The power was interrupted and the RPS/UPS static inverter de-energized. This event was caused by an incomplete connection between a logic card and its mating connector. The card design does not allow for ease of-determining proper connection. The lack of proper connection resulted in de-energizing of the static inverter. _ Procedure revisions are being made to provide a step which will test the integrity of the logic card pin connections prior to the transfer of power. B-5

t i ~ 06/19/86 San Onofre 2 50-361/86-015 il With the unit in mode 1, the reactor tripped when 120 V ac vital bus was de-energized due to failure of inverter. The inverter failure was caused by a shorted capacitor. Failure believed to be an isolated case. i 07/02/86 Fort Calhoun 1 50-285/86-001 With the unit at full power, the reactor tripped due to the failure of a safety related bus. Vital inverter failed but the root-cause was not discussed. 08/07/86 Catawba 2 50-414/86-036 With the unit in hot shutdown, a reactor trip signal was received due to personnel error. During preventive maintenance, technicians were adjusting the output frequency on a j vital inverter when they inadvertently caused the inverter to trip. ~ 08/10/86 Maine Yankee 50-309/86-005 l Reactor trip from full power due to trouble in the feedwater system and failure of an inverter. The root-cause of the inverter failure was a transfer switch. 10/02/86 Point Beach 2 50-301/86-006 With the unit in a refueling shutdown, three reactor trip signals were actuated due to work activities. A reactor trip was generated when a fuse blew while removing power. from a de motor operated valve. In meggar-testing of a new neutron detector cable, a i reactor trip signal was generated when the test induced an erroneous voltage on a nearby signal cable resulting in a source range reactor trip. An instrument bus voltage fluctuation caused the logic circuitry to unblock resulting in a reactor trip. In each of these events, the cause was attributed either outage-related work or erratic operation of an instrument bus inverter. Procedures have been changed and discussions will be held with personnel to emphasize the correct interpretation of reporting requirements and the importance of anticipa* ion actuation signals by work activities. 10/08/86 Lacrosse 50-409/86-031 1 While unit was in cold shutdown, several actuations including the emergency diesel generator (EDG) start occurred. The reason was a momentary disruption in the IB i noninterruptible bus. The IB static inverter has transferred to its alternate source when a fuse had blown in the inverter. ' No root-cause was determined for the inverter failure. 11/17/86 Point Beach 1 50-266/86-005 L With the unit at full power, the reactor tripped when power was lost to the red vital instrument bus. The root-cause of the failure was a control circuit diode in the bus inverter failed. B-6 5

l f 01/22/87 Farley 1 50-348/87-004 With the unit at full power, an erroneous high flux rate tripped the reactor due to failure of an inverter. B!own fuses cause failure of the inverter. The root-cause of the blown fuse was not determined. 02/11/87 Indian Point 3 50-286/87-002 With the unit at full power, the reactor tripped due to a static inverter trip A short circuit on a solenoid valve caused an overcurrent condition on the inverter. The inverter designed was modified to allow isolation of a single branch circuit if a short circuit develops. 02/16/87 Monticello 50-263/87-006 With the unit at full power, the reactor scrammed when inverter failed in one electrical division. Root-cause of the event was a failed feedback control transformer which blew a fuse. Corrective actions include changing power sources to the reactor vessel water level to prevent reactor scram upon losing one electrical division. + 02/23/87 Vogtle ' 50-424/87-005 s i With the unit at full power, several control room ventilation isolations and containment ~ isolation occurred. It was determined that voltage transients were generated on the 120 V ac vital power whenever the safety features sequencer system was re-energized. The apparent cause reason is a conservative setpoint in an inverter circuit which shuts down 1 power to radiation monitors whenever a large inrush of current is experienced. l Corrective actions are to increase the circuits setpoint. 02/23/87 Susquehanna 1 50-387/87-008 [ ^ With the unit at full power, an inverter failure and HPCI out of service alarms were received in the control room. A Technical Specification (TS) limiting condition for operation (LCO) was entered. The event was caused by a ruptured Topaz inverter capacitor which leaked insulating oil into the HPCI panel. 03/28/87 Palo Verde 3 50-530/87-001 Preparing to enter mode 6 for refueling, the control room received control room essential filtration actuation signa' ud containment purge isolation actuation signal f actuations. When the B inverter part of the BOP ESFAs transferred from normal power i to its alternate power source, the actuations occurred. The inverter failure root-cause was attributed to the slow static transfer switch operation and a faulty de to de converter l board. Corrective actions were to recalibrate the static transfer switch and replace the de to de converter board. i B7 l

4 k 05/CM/87 Byron 2 50455/87-007 i With the unit in mode 1 and a quarterly surveillance on the power range calibration in progress, the reactor was tripped when one other instrument bus was lost. The cause of the instrument bus trip is due to a failure of an SCR and a capacitor in the inverter. The capacitor failure caused the SCR failure. The failure mechanism of the capacitor was attributed to normal wearout. 05/10/87 Palo Verde 2 50-529/87-013 With the unit at 20 percent power, a TS shutdown was commenced due to an inverter fuse failure. Inverter powers plant protection systems, ESF, and other instrumentation. During troubleshooting the inverter fuse blew again, and its associated static transfer switch did not switch to its alternate power supply. The root-cause of the inoperable inverter was determined to be a loose SCR. 06/15/87 Palo Verde 3 50-530/87-002 While in mode 5 (cold shutdown), the control room received control room essential filtration actuation signal and containment purge isolation actuation signal actuations on both channels of BOP ESFAs. Immediately prior to the actuations the B inverter transferred from normal power to its alternate power source. The root-cause of this event was attributed to the static transfer switch calibration at an unnecessarily high value and a faulty de to de converter board. 06/18/87 River Bend 1 50-458/87-012 m With the unit at 70 percent power, there was a reactor trip with the loss of control power l to panel IVBN-PNL0181. This occurred inadvertently during the troubleshooting of the i battery inverter. Procedural inadequacy caused the inverter loss. Procedure revisions have been completed that will paclude recurrence by requiring the placement of the battery inverter in manual bypass mode prior to troubleshooting. 07/02/87 Crystal River 3 50-302/87-009 With the unit in mode 1, while performing RPS surveillance testing, a vital inverter failed tripping the reactor. Event was initiated when a control rod drive mechanism circuit breaker would not close. The root-cause of the inverter failure was a lower undervoltage setpoint. 07/23/87 Hatch 1 50-321/87-011 With the unit at full power, the reactor tripped due to inadequate ventilation design of the vital inverter. Corrective actions include upgrading HVAC and developing preventive maintenance procedures. High temperatures in the inverter initiated the failure. B-8 j i

-I l l 07/26/87 Hatch 2 50-366/87-006 With the unit in mode 1, the reactor tripped due to loss of a vital bus. Vital inverter failed in a high room temperature. Root-cause is attributed to inadequate HVAC design to ensure vital equipment is properly cooled. Corrective actions include adding HVAC, } developing preventive maintenance procedures, and installing new vital ac inverters. 12/05/87 Crystal River 3 50-302/87-028 i While in a refueling mode two of the vital inverters were bypassed and on their alternate l power sources, an ESFA occurred when one of the vital inverter was lineup with the l energized power supplies. The root-cause of the event was inadequate operator training on the function of the inverter static transfer switches and manual bypass switches. Personnel have reviewed their actions and enhanced training will be gieen. Additionally, a schematic diagram of the vital bus power supplies and switching arrangements will be provided as operator aids in each of the inverter rooms. i 12/21/87 St Lucie 1 50-335/87-017 L While operating at 100 percent, the unit tripped due to the loss of a vital instrument bus. The failure of the instrument inverter caused the subsequent loss of the 120 V bus. The root-cause of the evert was a cognitive personnel error of not adequately following the procedure. A procedme for infrequent operations or manipulations is being drafted. This is to assure a detailed review and briefing by the shift supervisor. A human performance evaluation is being conducted to identify any areas that may be of concern. 01/19/88 Millstone 1 50-336/88-002 Unit was shutdown when loss of crmal power caused ESFAs. One vital instrument inverter was being seniced when another vital inverter failed. Failed de input capacitor caused loss of inverter. Cause of capacitor failure was not known but believed to be random failure. l 02/09/88 Diablo Canyon 1 50-275/88-006 f With the unit in mode 1, fuel handling building ventilation system and control room ventilation isolation signal was initiated when the voltage on an inverter degraded. The j inverter was transferred to alternate power source. The root-cause was a transformer failure. Vendor will preform analysis on failed transformer. Spare transformers will be tested. Maintenance procedure of Westinghouse 7.5 kV A inverters will be revised to l require a high potential test. 02/20/88 Braidwood 2 50-457/88-008 With the unit in startup testing there was a loss of power to an instrument bus. This resulted in a reactor trip. Root-cause seems to be heat damage to the fast-on-connectors I i B-9

~ i I in the inverter. Corrective actions are for inspections of same connectors on all the other inverters on both units. 02/22/88 Oconee 2 270/88-001 i With the unit shutdown, a brace to secure a printed circuit board was found missing during preventive maintenance. Vital inverter declared inoperable. There was a TS i violation and the inverter had operated outside its design basis. Root-cause was lack of a procedure to remove and restore brace. Contributing cause was the lack of documentation on the brace. l 02/24/88 South Texas 1 50-498/88-021 l With the unit in mode 3 prior to initial criticality, a number of control annunciators alarmed along with ESFAs. The actuations were traced to the failure of inverter IV-001. i A failed de to de converter assembly from the inverter was returned to the vendor for i failure analysis. The analysis determined the failure was random and was not the resu'.t i of design or manufacturing defects. 03/27/88 Surry 2 50-281/88-004 i With the unit at full power, a turbine runback initiated due to the loss of one sital l thus causing a current surge which blew the fuse and tripped the output breaker. inverter. Evaluation of the inverter found an internal inductor had failed due to age, Corrective actions are to replace one of two station vital inverters per unit. The remaining two vital bus inverters will be replaced with UPSs during the next refueling e outage. l 05/20/88 Susquehanna 1 50-387/88-009 4 With the unit at full power, an inverter failure alarm was received in the control room. It was determined that the HPCI inverter had tripped and the plant entered a TS LCO. The cause of the trip is attributed to the high voltage trip setpoint of the inverter drifting low. At the time, the battery charger supplying power to the inverter was operating in equalize. The input voltage to the inverter was 144 V dc. The inverter high voltage setpoint is normally 147 V dc. The equalize voltage was lowered to 142.8 V dc and returned the charger to float. These actions lowered the voltage input to inverter between -130 an 134 V dc which allowed the inverter to reset. 06/16/88 Hope Creek 1 50-354/88-016 l With the unit at full power, power was lost momentary from a vital inverter. The unit entered a TS LCO. Preventive maintenance was in progress on the inverter. The apparent cause was a malfunction of the inverter switch. Corrective actions are to trouble shoot the switching function. I B-10 ?

0622/88 Salem 2 50-311/88-014 With the unit at full power, the reactor tripped due to the failure of vital instrument bus. j Root-cause of vital inverter failure was not mentioned. Licensee will do a design change on the 1 out of 4 RCP breakers trip logic and will replace the inverters with state of the art inverters. I i 07/30/88 Salem 2 50-311/88-016 i With the unit in mode 1, the reactor tripped due to a failure of vital instrument bus. Root-cause of vital inverter failure was not mentioned. The licensee will do a design change to the reactor trip logic which occurs when 1 out of 4 RCP breakers open. { 08/06/88 San Onofre 2 50-361/88-027 With the unit at full power, the licensee found out vendor error in settings for inverter shutdown voltage. Setting was at 115 Vs rather than the proper 105 Vs. 11/13/88 Vogtle 1 50-424/88-035 During the a refueling outage, the technicians were conductin;; TS surveillance. A i momentary large current inrush occurred when they reset the sequencer. The current inrush activated a " ZIP" circuit that shutdown the inverter. Corrective actions includes i raising the setpnint which activates the ZIP circuit. l t 11/16/88 Brunswick 2 50-324/88-018 i With the unit at full power, a turbine trip was initiated by a turbine control valve fast 1 closure. During this event, there was a failure of the high reactor water level controls that was caused by a Topaz inverter having a high input voltage. 01/06/89 San Onofre 3 50-362/89-001 j With the unit in mode 1, the reactor tripped due to partial loss of non-1E UPS. Two of l three UPS phases were lost because of a common fault in the associated inverters constant voltage transformer output windings. Jumpers were found from previous work but not related to this problem. The root-cause of the transformer failure was the breakdown of insulation between energized windings and grounded iron core. 02/03/89 Point Beach 2 50-301/89-001 With the unit a 100 percent, a turbine runback was experience when an operator repositioned both supply breakers in the bus. One of the instrument buses had previously been tagged out for service to correct a minor auxiliary switch indication problem. Root-cause is a personnel error. B 11 .j l l

.j t-t h 02/12/89 South Texas 1 50-498/89-008 [ With the unit in mode 5, a plant operator determined that an inverter which supplies UPS to a distribution panel was overheating. The operator transferred the inverter to its l alternate power source. The failure of the inverter was a short to ground on the secondary side of the inverter ferroresonant transformer. Corrective actions include i replacement and failure analysis of the failed transformer. i 03/12/89 Salem 2 50-311/89-005 With the unit at full power, the reactor tripped due to a vital inverter having a lose control power fuse. Cause was determined to be improper installation in the past. 03/29/89 Hatch 1 50-321/89-006 With the unit at full power, the HPCI was declared inoperable due to the failure of an inverter. The cause of this event is component failure. The Topaz static inverter tripped due to a failed diode. 08/14/89 Cook 2 50-316/89-014 With the unit at full power, the reactor tripped on RPS actuation when transferring control room inst,amentation distribution vital inverter power to its normal power supply. The inverter failure was due to a failed SCR in the static transfer switch. This ( also resulted in fuses blowing and other power supplies failing in the components fed by i the control room instrumentation distribution. 08/29/89 South Texas 2 50-499/89-020 j With the unit at full power, three turbine driven feedwater pumps tripped causing the i control room operator to manually trip the reactor. The cause of the event was a L momentary interruption of control power to the feedwater pump overspeed protection circuits due to the failure of an inverter. l l 09/22/89 South Texas 2 50 499/89-023 l With the unit in mode 1. a turbine / reactor trip occurred on loss of power to the four main turbine auto stop solenoids. The cause of this event was failure of a nonsafety-related inverter which interrupted power to the main turbine auto stop solenoids. The root-cause of inverter failure was not determined. I 10/11/89 South T2xas 1 50-498/89-020 With the unit in mode 3, an inverter to the class 1E vital ac distribution panel failed. The inverter had a failure of the bridge rectifier circuit on the de to de converter board. The circuit appeared to have failed due to excessive output voltage over an extended period of time which overheated the components. Preventive maintenance procedures B-t2 l

t l 1 will be revised to require periodic de to de converter board output voltage adjustments. Inverter will be added to the existing plant thermography program. 10/25/89 Limerick 1 50-352/89-055 With the unit at full power, ESFAs occurred during the transfer of an RPS inverter to its preferred power source, the RPS shunt breaker tripped. Root-cause of the RPS breaker trip is unknown. 03/05/90 Comanche Peak 1 50-445/90-002 With the unit shutdown and cold rod drop testing in progress, an inverter blew a fuse and caused some system actuations. The root-cause ofinverter failure was not determined. However, possible causes were the ferroresonant transformer and loose connections in the gating circuit. A detail visual inspection will be done on the other inverters at the next shutdown. This inspection will include the verification of all bolted and soldered connections in the inverters. l 03/30/90 Limerick 2 50-353/90-007 With the unit in cold shutdown, a failure of the '2B' RPS/UPS inverter inductor caused i' a loss of power to the '2B' RPS/UPS power distribution panel. ESF occurred due the various automatic actuations. The proximate cause of the loss of power to the i distribution panel was a gross failure of an inductor in the '2B' RPS/UPS inverter. 06/24/90 Seabrook 1 50-443/90-016 With the unit hot standby, the normal power supply to vital 120 V ac power panel PP-1E experienced a capacitor failure and a blown fuse. This failure initiated an inverter automatic swap from the normal power supply to the maintenance power supply. In investigating the problem an operator mistakenly tripped open the maintenance power supply breaker causing more system actuations. The root-cause of this event is attributed l to personnel error. Corrective actions include more descriptive labels on the inverters and power panels, and one line drawings including ac and de feeds being placed on inverters. j 09/03/90 Byron 2 50-455/90-006 With the unit shutdown, SI surveillance was being performed. During the work, due to j miscommunications, two instrument inverters were de-energized. This is a personnel error. Corrective actions include procedure revisions to the annual SI surveillance. i 09/09/90 South Texas 1 50-498/90-021 i ESFAs occurred while the ur.it was at full power when an inverter which feeds the class j 1E ac vital distribution panel failed. Unable to fix the problem within the TS LCO time. a unit shutdown was commenced. The cause of the inverter failure was a power filter l B.13 b

.j ? capacitor which interrupted power to the inverter controller card and blew two main j power fuses. Corrective actions include replacement of de to de converter board, j trending and analysis of the other similar de to de converter boards, and revision to the l maintenance manual. 09/14/90 Sequoyah 1 .50-327/90-021 l With the unit in mode 1, a feedwater transient generated by the loss of a vital inverter tripped the plant. The failure occurred after the completion of inverter maintenance during the transfer of inverter maintenance power supply to its normal power supply. During the transfer, the inverter output voltage dropped to zero because of the failure of the SCRs. 11/01/90 Palisades 50-255/90-019 With the unit in cold shutdown, an ESFA occurred due to personnel error. A technician j error caused a short which tripped the inverter. 11/01/90 Limerick 2 50-353/90-019 With the unit at full power, various actuations and a half-scram occurred due to the tripping of '2B2' RPS static inverter resulting in the loss of the RPS/UPS power distribution panel. The cause of this event was a damaged connector in the '2B2' RPS static inverter circuitry coupled with troubleshooting being performed. All static inverters were inspected for damaged connectors an no other problems were identified. 11/17/90 Summer 1 50-395/90-010 With the unit at full power, several spurious actuations occurred with Protection Set I. A local inspection found the associated Protection Set I inverter output voltage was low and slowly degrading. The inverter was transferred to its alternate power which caused certain instrument and control anomalies but the operators were able to stabilize the plant. Maintenance repaired a shorted ferroresonant transformer in the inverter. 3 12/06/90 San Onofre 2 50-361/90-016 With the unit at full power, a non-1E UPS momentary de-energized non-1E instrument bus. A failure of a capacitor in the non-1E UPS inverter output caused the loss of. power on instrument bus. The manufacturer of the non-1E UPS determined that the design of the capacitor which failed was defective. All like capacitors in SONGS Units 2 l and 3 have been replaced with an upgraded model. Instrument buses powered by the i UPSs will be modified to prevent total loss of critical systems. 03/03/91 Trojan 50-344/91-006 [; With the unit a full power, an inverter alarm was received in the control room.' One of the 125 V de-backed inverters supplying a 120 V preferred instrument panel had i B-14 5 L

~ } u automatically transferred to bypass source. The instrument panel was declared 1 inoperable and a TS LCO was entered. The automatic transfer was initiated by a j random component failure of an integrated circuit chip on an inverter logic board. It is postulated that the automatic transfer was initiated by a current surge which damaged inverter circuit components that may have been stressed by transfers occurring at excessive phase angle differences. Synchronization checks were performed on all the ~ remaining inverters. An adjustment were made on the inverter and will be made on the other inverters during the 1991 refueling outage. t 05/10/91 Zion 1 50-295/91-008 With the unit in hot standby resistance temperature detector cross-calibration was in .i progress when an ac inverter failed. The exact cause of the Westinghouse inverter failure was not determined. However, the investigation found several oxidized breakers ( and some SCRs were replaced. SCRs and breakers will be included in the preventive maintenance program. Currently, thermography is done periodically and the capacitors are checked for leaks. 05/30/91 Point Beach 1 50-265/91-005' With the unit at full power, the reactor tripped when power was lost to red vital instrument bus. Three diodes failed and a number of fuses blew. Two of the diodes were associated with mas'.er board power supply. Red and blue inverters are original Westinghouse. White and yellow are newer Elgar units. 06/18/91 Limerick 1 50-352/91-018 With the unit at full power, an ESFA occurred due to inverter failure. Root-cause was ) inadequate design of inductors. Preventive maintenance previously written following other inductor failures to replace the newer inductors for all the inverters on a annual basis will be expanded to include the older inductors pending thermographic inspections. Thermographic inspections will continue to be performed on all inverter inductors on a j quarterly frequency. 06/29/91 Point Beach 1 50-266/91-008 I With the unit at full power, the reactor tripped when power was lost to white vital l instrument bus. Cause of the trip was not mentioned. .l 07/31/91 Susquehanna 1 50-387/91-008 With Units 1 and 2 at full power, loss of an offsite ac source coincident with a Unit 1 pre-existing conditions, resulted in a RPS scram and MSIV isolation on Unit I and a RPS half-scram on Unit 2. The reactors responded as designed, but TS were violated - 1 before ' units reached Cold Shutdown. Among the problems, there was a loss of ac to an - instrument panel on Unit 2 due to the backup power supply. Upon loss of the normal power supply, the UPS automatically transferrsd to its self-contained battery bank B-15 i

1 s l backup but with three failed cells, the backup power was not available. High ambient l temperature inside the UPS unit shorted the battery life from 10 years to 3 years. Correct actions are increased battery maintenance with replacement every other fuel cycle, and performance of a semi-annual load test. i 08/13/91 Nine Mile Point 2 50-410/91-017 With the unit at full power. the reactor tripped due to a main transformer failure. j Generated transient caused loss of five non-1E UPS units. The site declared a Site Area Emergency, Backup logic batteries in the UPS units were found dead. Cause of failure was a design error in the preferred logic power source. UPS units investigations found deficient documentation, inadequate maintenance of logic batteries, and no in-place vendor support program. NUREG-1455 discusses the details of the event, j 08/31/91 Palo Verde 3 50-530/91-007 Unit was shutdown from hot. standby to fix failed inverter. Failure was due to a cognitive personnel error by a technician who improperly attached a variable power i supply to an energized control element assembly calculator circuit causing failure j inverter failure. 11/01/91 Three Mile Island 1 50-289/91-005 1 W.th the unit in a refueling mode, a vital inverter was being returned to service after I preventive maintenance when an safety assessment system actuation occurred. The root-cause was a personne! error by the control room operator who pushed the wrot.g button. ? 1 11/07/91 Zion 1 50-295/91-016 1 1 With the unit at full power, the reactor tripped when one ac inverter failed. An SCR in the master section of the inverter misgated causing a reduction in the inverter output voltage. Corrective actions were replacing all Unit 1 inverter SCRs prior to startup, j including the SCRs in the preventive maintenance program, and ensuring all Unit 2 SCRs will be replaced during the next Unit 2 refueling outage. 01/17/92 Brunswick 1 50-325/91-003 With the unit at full power, there was a reactor trip due to loss of an UPS. The UPS sustained momentary voltage losses due to failure in the primary static switch. The problem prevented automatic transfer of the UPS loads to the alternate powers source. 02/19/92 Wolf Creek 1 50-482/92-002 With the unit a full power, a reactor trip occurred when power was lost to a safety-related bus. The bus power failure was due to loss of a gating / sequencing card in the i inverter supplying the instrument bus. A review of the operating procedure," Loss of {. B-16

1 e Vital 120 V ac Instrument Bus," has been conducted and while the procedure provides the necessary guidance when responding to this kind of event, enhancements are being incorporated into the procedure. ~ 02/21/92 Limerick 2 50-353/92-004 Unit at 100 percent power had to initiate shutdown when HPCI declared inoperable due to an inverter failure. -TS actions required since another RCIC was already inoperable due to maintenance. Incident involved two inverter failures at different times during this event sequence. Inverter failures were caused by a power supply blown fuses. 03/06/92 Diablo Canyon 1 50-275/92-002 i With the unit at full power, the reactor tripped due to the failure of a non-safety related inverter. The inverter provides power to the speed sensing probes of the main feedwater pump (MFP). Failure of inverter was due to poor design and manufacturing. Corrective actions included testing the other MFP inverters. Also, there will be a new design that provides a separate power supply to each of the two speed probes on each MFP. 03/23/92 Nine Mile Point 2 50-410/92-0 % At 0 percent power with core partially offloaded, the site lost offsite power. As a result, numerous emergency safety features actuated, control room annunciation was lost, a RPS scram signal was received, and the high-pressure core spray EDG tripped following the loss-of-offsite power. The site declared an ALERT. Among the other problems, the UPS failure that caused the loss of annunciators was due to bad internal logic batteries. These same logic batteries were a topic of discussion of an IIT (NUREG-1455). An i AIT, Inspection Report No. 50-410/92-81, was sent to investigate this event. The corrective actions were to replace the logic batteries and to perform failure analysis on the failed batteries. 03/26/92 Nine Mile Point 2 50-410/92-007 With the unit in a refuel mode, ESFAs occurred caused by the loss of output power from an UPS unit. The ESFA was a Group 9 Primary Containment Isolation. The immediate operator response was to manually close the failed UPS circuit breaker. This circuit breaker on the UPS failed to automatically close upon transfer of loads to the UPS maintenance supply. The root-cause was determined to be dried grease inside the breaker. Similar break.ers to the failed one are scheduled to be replaced on the other UPS units. 03/27/92 Crystal River 3 50 302/92-001 t With the unit in mode 1, while performing maintenance on a vital inverter, a personnel l error caused a transient that tripped the reactor and started the EDGs. l P B-17 i

l 11/18/92 Farley 2 50-364/92-015 With the unit at full power, the reactor tripped due to a turbine trip. Turbine system digital electrohydraulic control inverter failed. Failure of the inverter initiated a voltage j transient. The root-cause of the inverter failure was not mentioned. r h DATE PLANT 50.72 NOTIFICATIONS 04/04/92 Comanche Peak 2 50-446/ ENS 23451 During testing of the Elgar inverters, the licensee discovered several discrepancies on the IE inverters. Some of these were wiring errors, incorrect parts, missing parts on the circuit boards, incorrect wound transformers, poor crimp connections, and failed control circuits. Corrective actions were to correct the discrepancies before placing the inverters in service. L 12/08/92 Waterford 3 50-382/ ENS 24719 I From the 50.72 report:- Reactor tripped when unit lost 'B' vital instrument bus causing RPS actuation. He event was initiated when the static UPS SM-B de-energized. The loss of the 'B' vital instrument bus caused partial actuation of the 'BC', 'BD', and 'CD' RPS matrices which resulted in a reactor trip. The licensee entered a TS I.CO. The cause of the inverter failure has not been determined. I DATE PLANT R III DAILY REPORT 11/26/92 Callaway 1 50-483/MR 3-92-0332 From the morning report: The output breaker for vital instrument bus NN14 opened. l The instrument bus was re-energized from the alternate source. The cause of the breaker trip was the inverter which had a faulty ferroresonant transformer. A TS LCO was entered but time ran out and the licensee requested a 24 hour extension to the TS. The inverter was restored to its safety-related power supply within the 24 hour extension. t i i t b B-18 l 1

.A--, i 0 i f APPENDIX C - INVERTER FAILURES BY PLANT (1985-1992) l P P s I t i f 1 i

APPENDIX C Inverter Failures by Plant (1985-1992) PIANT UNIT NO. OF FAILURES Byron 1 1 ByTon 2 2 Braidwood 2 1 Brunswick 1 -1 Bnmswick 2 1 Callaway 1 1 Catawba 2 2 Comanche Peak 1 1 Comanche Peak 2 1 Cook 1 1 Cook 2 2 Crystal River 3 4' Diablo Canyon 1 2 Duane Arnold 2 Farley 1 1 Farley 2 2 Fort Calhoun 1 Grand Gulf 1 1 Hatch 1 2 Hatch 2 1 i Hope Creek 1 1 L Indian Point 3 1 Kewaunee 1 Lacrosse 2 Limerick 1 6 Limerick 2 3 Maine Yankee 1 c1 i

I e Q APPENDIX C (cont.) Inverter Failures by Plant (1985-1992)- PLANT UNIT NO. OF FAILURES Millstone 2 1 Monticello 1 Nine Mile Point 2 3 Oconee 1 1 Oconee 2 1 Palisades 1-Palo Verde 2 1 Palo Verde 3 3 Peach Bottom 3 1 Point Beach 1 4 Point Beach 2 3 Quad Cities 1 1 River Bend 1 1 Salem 2 3 San Onofre 2 3 San Onofre 3 1 Seabrook 1 1 Sequoyah 1 1 Shoreham 1 South Texas 1 4 South Texas 2 2 St. Lucie 1 1 Summer 1 1 Surry 2 1 Susquehanna 1 3 Three Mile Island 1 1 Trojan 1 C-2

,y-APPENDIX C (cont.) inverter Failures by Plant (1985-1992) PLANT UNIT NO. OF FAILURES Turkey Point 3 4 Turkey Point 4 2 Vogtle 1 2 WNP 2 1 Waterford 3 1 Wolf Creek 1 2 Zion 1 2 Zion 2 1 D C-3 i

4 e i UNITED STATES DRAFT NUCLEAR REGULATORY COMMISSION OFFICE OF NUCLEAR REACTOR REGULATION WASHINGTON, D.C. 20555 ,1994 NRC INFORMATION NOTICE 94-XX: INADEOUATE MAINTENANCE OF UNINTERRUPTIBLE POWER SUPPLIES-AND INVERTERS Addressees All holders of operating licensees on construction permits for nuclear power rectors. i Purpose The U.S. Nuclear Regulatory Commission (NRC) is issuing this information notice to alert addressees of failures of uninterruptible power supplies (UPS) and inverters installed in systems important to safety. Many of these failures could be averted by appropriate maintenance in accordance with manufacturers guidelines. Suggestions contained in this information notice are not NRC requirements; therefore, ne ecific action or written response is required. Past Related Correspondence Tne NRC has issued the following correspondence on this subject: 1 IN 91-64, Supplement 1," Site Area Emergency Resulting From A 1.oss of Non-Class IE Uninterruptible Power Supplies," October 7,1997 IN 91-64," Site Area Emergency Resulting From A Loss Of Non-Class 1E l Uninterruptible Power Supplies," October 9,1992 IN 87-24, " Operational Experience Involving Imsses of Electrical Inverters," June 4,1987 IN 84-80, " Plant Transients Induced by Failure of Non-Nuclear Instrumentation Power," 4 November 8,1984 Bulletin 79-27," Loss of Non-Class 1E Instrumentation and Control Power System Bus During Operation," November 30,1979 IE IN 79-29, " Loss of Nonsafety-related Reactor Coolant System Instrumentation During Operation " November 16,1979 IE Circular 79-02," Failure of 120 Volt Vital AC Power Supplies," January 11,1979 i

k DRAFT IN 94-XX ,1994 Page 2 Description of Circumstances On August 13, 1992, at the Nine Mile Point Nuclear Station Unit 2 (NMP-2), an internal failure in the main transformer resulted in degraded voltage which caused the i simultaneous loss of five UPS. The five UPS units provide the power to the main control room annunciator system and to other systems important to safety. On March 26,1992, a nonsafety-related UPS failed to supply power to a radiation monitoring cabinet which caused an engineered safety feature actuation (ESFA). On July 31,1991, at the Susquehanna Steam Electric Station Unit 1 (SSES-1), a relay in the switchyard was misoperated, causing one of the offsite ac circuits to de-energize. The results included a loss of power to instrument ac panels after the UPS transfer to its backup power supply was unsuccessful due to internal battery problems. This loss of power to instrument panels resulted in the loss of numerous instruments, control room indications, and some plant support equipment. Discussion The root-causes for the UPS failures at NMP-2 on August 13,1992, and March 26,1992, were determined to be inadequate maintenance of the batteries which supply power to the UPS logic, and inadequate or incorrect lubrication of a UPS circuit breaker respectively. The root-cause of the UPS failures at SSES-1 was determined to be three failed cells in the UPS battery bank. UPS units provide uninterruptible ac power by rectifying the station power from ac to de and inverting the de back to ac. The UPS design combine a rectifier and inverter as one unit. Some configurations do not use a rectifier but only an inverter. Either a station [ battery or vendor supplied battery-pack located in separate cubicles provides backup to the inverter. Offsite ac provides backup to the inverter ac output through a transfer switch. A recent NRC engineering evaluation by the Office of Analysis and Evaluation of Operational Data (AEOD) was prepared to assess the industry inverter operational experience over the last 7-% years. AEOD/E93_. concludes that the number of inverter failures have decreased but some problems still exist. The significant findings were: 1. Inverter failures are still causing a large number of inadvertent ESFAs, reactor trips, and turbine runbacks. 2. Component failures are still the number one root-cause of the failures, followed by human error.

3 4 2 l 1 DRAFF IN 94-XX - ) ,1994 Page 3 3. Electrolytic capacitors are the most failed component followed by transformers, silicon-controlled rectifiers, and transfer switches. 4. Following vendor recommendations of minor and major maintenance would reduce the number of failures. The report found a marked improvement in electrical inverter failures as identified by licensee event reports when the inverter units are provided cooling, are in a preventive maintenance program, and the components are replaced periodically. If the vendor maintenance recommendations are followed rigorously, the predominate root-cause failure would be reduced due to identification of degradation before components fail. In Enclosure 1, the NRC staff compiled the recommendations of three vendors for short term and long term maintenance. Licensees need to refer to their own manuals provided by the manufacturer for more specific maintenance requirement. By communicatirig periodically with the vendors, licensees may learn of new ways to improve design and maintenance and reduce the number of maintenance related failures. This information notice requires no specific action on written response, if you have any questions about the information in this notice, please contact the technical contacts listed below or the appropriate Office of Nuclear Reactor Regulation (NRR) project manager.

Enclosure:

As stated Technical

Contact:

Jose Ibarra, AEOD (301) 492-4441

,, o 7 DRAFT - Page 1 IN 94-xx ,1994 Maintenance Recommendations for the Uninterruptible Power Supplies (UPS) Exide recommends to annually perform at least the following maintenance. Shut down the unit and inspect the physical condition of all components and t connections. i Inspect all electrolytic capacitors for leaks. e Operate unit with low load or no load, and check the logic functions. o Verify the setting to assure no changes which jeopardize UPS reliability. e Verify'the wave shapes wherever applicable. e Connect a variable UPS load and adjust load to verify corresponding settings e response. Check all the trip functions, fuse failures, and overload conditions. e Verify the transfer mechanisms for alternate power is operable. e Exide recommends minor maintenance on their UPS at a frequency that depends on the environmental conditions in which they operate. Customers should check and replace the air filters when dirty, check for discolored components such as resistors or capacitors and verify that the D-cell batteries function properly. ELGAR recommends the following maintenance: Monthly l Inspect intake air filters, cleaning or replacing them as required. e inspect batteries for overheating, tightness of all electrical connections, and leaks. e Visually inspect connectors and printed wire assemblies for overheating, tightness e of connection, and damage. Inspect fans for proper blade rotation and freedom from obstruction.

0., DRAIT . - Page 2 IN 94-xx ,1994 Measure to verify the following according to the specification o charger output voltage inverter output voltage and frequency inverter synchronizing circuitry transfer switch operation alarms and indicators inverter output over and under voltage settings Annually Verify specifications for the torque on the silicone controlled rectifier mounting bracket bolt and flatness. Biannually i i. Replace fans. Every 9 Years Replace all oil-filled and electrolytic capacitors as noted in manual. In some cases. replacement interval may be 7 years, refer to manual. To perform most of this work, the maintenance personnel must have a thorough knowledge of the unit and its system. ELGAR recommends contacting them before performing the task if the knowledge and equipment is not available at the site. All sites should have the recommended spare parts kit depicted in the unit manual delivered with the shipment. i

.9 4 3 DRAFT - Page 3 IN 94-xx ,1994 Solid-State Control Inverters Component Maintenance and Replacement Schedule Replace maintenance components at the required intervals to maintain the 1E e system integrity. Replace in accordance with the manufacturers recommendations or as determined by the Wyle Laboratory aging analysis. Include circuit boards in the component replacement schedule. Replace e components at the designated intervals in lieu of replacing the complete circuit board. Replace components by a qualified technician or manufacturer recommends sending him the boards for the replacement work. Also include in the required intervals the following : Inspect fan motor strain relief busing for wire entry on 2 year intervals and replace if deterioration is found. Lubricate fan motor bearings on 2 year intervals if bearing oil cups are present. Clean the system of dust and debris. Pressurized air to 40 psi maximum may be used to clean the system. Do not use cleaning agents containing hydrocarbons on the components or in the system. Components with a life less than the qualified system life are being operated at a margin below which component was analyzed for. Thus, the design life and ire replacement period are the same time periods. The fans are not required to mai.ataan the proper inverter operation.}}