ML17332A802

EDG Load-Run Performance & Reliability During Short & Long Duration Test Periods TER

ML17332A802
Person / Time
Site:	Cook
Issue date:	06/15/1995
From:	AMERICAN ELECTRIC POWER CO., INC.
To:
Shared Package
ML17332A800	List: ML17332A799 ML17332A801 ML17332A802
References
NUDOCS 9506200467
Download: ML17332A802 (27)

Text

ATTACHMENT 4 TO AEP'NRC'0896V COOK NUCLEAR PLANT EMERGENCY DIESEL GENERATOR LOAD-RUN PERFORMANCE AND RELIABILITY DURING SHORT AND LONG DURATION TEST PERIODS 9506200467 950615 PDR ADQCK 05000315 PDR to AEP:NRC:0896V page 1

ABSTRACT The following study compares Cook Nuclear Plant emergency diesel generator performance during relatively short test runs to performance during significantly longer test durations.

A statistical treatment of data over a twenty year period is used to develop an estimate of cumulative failure probability with respect to test run time.

In addition, a review ofprevious emergency diesel. generator performance studies is used to provide an industry wide comparison ofload-run failure probabilities. The study concludes that an endurance test duration ofeight hours in lieu of the current twenty four hours would be acceptable for the emergency diesel generators at Cook Nuclear Plant.

American Electric Power to AEP:NRC:0896V page 2 TABLEOF CONTENTS Section 1.0 Introduction

~Pa e

2.0 Cook Nuclear Plant Diesel Generators 2.1 Specifications 4 Major Components 2.2 Operations Overview 3.0 Review ofCook Nuclear Plant Test Data 3.1 Valid Load-Run Criteria 3.2 Invalid Load-Run Criteria 3.3 Long Duration Load-Run Data 3.4 Short Duration Load-Run Data 3.5 Failure Characteristics 3.6 Data Analysis 4.0 Review ofSelected Studies 4.1 BNLTechnical Report ¹A-3134 1-85 4.2 NSAC-108 4.3 NUREG/CR-5994 15

'.0

.Indu'stry Surveys 17 6.0 Conclusions 17

. 7;0, References

~ '

18 American Electric Power to AEP:NRC:0896V 1.0 Introduction page 3 This report focuses on emergency diesel generator (EDG) load-run performance at Cook Nuclear Plant.

The report considers both long duration load-runs (greater 60 minutes) and short duration load-runs (60 minutes or less).

The report does not consider EDG starting reliability.

This report has four primary objectives:

1) Review the load-run performance ofemergency diesel generators (EDGs) at Cook Nuclear Plant with emphasis on the long duration test runs.

2)

Review existing studies, evaluations, and data to determine a measure of EDG load-run performance on an industry wide basis.

3)

Sample EDG 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> endurance test run performance at other nuclear power plants to provide comparison data.

4)

Construct a meaningful translation of this data in conjunction with existing studies, evaluations, and other data to determine a qualitative benchmark for evaluating the acceptability ofreducing the 18 month EDG endurance test duration Rom the current 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> requirement for Cook Nuclear Plant.

Atnericnn Electric Popover to AEP:NRC:0896V page 4 2.0 Cook Nuclear Plant Diesel Generators The data in sections 2.1 and 2.2 summarize important EDG specifications and operating requirements.

Both units at Cook Nuclear Plant are Westinghouse, 4-loop, pressurized water reactors.

Unit 1 is rated 3250MWt and Unit 2 is rated 3411MWt.

Four EDGs, two per unit.

Each diesel engine is a Worthington Type SWB-12, 12

cylinder, heavy duty turbocharged diesel engine, with a continuous rated output of4900BHP at 514RPM.

Each generator is a General Electric, 4375KVA, 3500KW at 0.8 P.F., 514RPM, 3-

phase, 60-cycle, 4160V, 25% voltage regulation, direct engine-driven synchronous type generator.

Each generator is equipped with a GE brushless exciter type SAR with a field rated 5.8A and 100VDC and, a static voltage regulator with its associated potential and sensing transformers.

...The EDG system for'each unit consists of two redundant, Clhss 1E, identical 'diesel":

generators which are irtdividuallycapable ofsupplying sufficient power to operate one'omplete redundant train of engineered safety features (ESF) and protection systems required for safe shutdown ofthe unit.

. The EDGs are designed ta start automatically upon receipt ofa safety injection signal

'nd/or a loss ofoffsite power signal and'be ready to accept loads within 10 seconds of

'eceiving a start signal.

During normal plant operation, the EDGs are on standby and are automatically available ifo6site power is lost.

American Electric Popover to AEP:NRC:0896V page 5 3.0 Review ofCook Nuclear Plant Test Data This section considers load-run test data from 1974 to 1994.

The objectives of this section include examination of load-run data specifically associated with 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> surveillance tests.

Accordingly, the study could have focused strictly on 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> surveillance test runs. However, it was felt that doing so could eliminate, or mask, insightful observations about the "quality" of EDG performance during other extended full load-runs and the characteristic behavior ofload-run failures. In addition, the limited data available from the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> surveillance tests was not statistically significant. Even by including all long duration (greater than 60 minutes) full load-runs, the database was still limited to 128 total tests. Therefore, data from short duration (60 minutes or less) load-runs is provided below for comparison and to facilitate analysis of the long duration fullload-runs.

In evaluating EDG test data, the boundary and support systems of the EDG system were considered to include the diesel engine, the generator, the exciter and voltage regulator system, the control and protection system, the EDG lubricating oil system, EDG fuel oil system, EDG jacket water system, EDG starting air system, EDG intake air system, EDG exhaust gas system, and the control circuitry up to the immediate control power source.

The criteria in Sections 3.1'and 3.2 were used to evaluate the validityofEDG test runs.

The followingcriteria +ere based on references 5, 10, and 11. A'load-run was counted.

hs a valid demand, success, or failure ifitsatisfied either. one ofthe followingcriteria:

P

~

a.

Termination of the load-run prior to mission completion, due to abnormal conditions that would. ultimately have;resulted in the failure of.the EPG, was

'counted as a valid demhnd and failure..

I b.

Termination after'completion of the intended mission was counted as a valid demand and success.

American Electric Popover

to AEP:NRCt0896V page 6 The following criteria were based on references 5, 10, and

11. A load-run was not counted as a valid demand,

success, or failure when the load-run was prematurely terminated for any ofthe followingreasons:

a.

Aspurious operation ofa trip that is bypassed in the emergency operating mode.

b.

Amalfunction ofequipment that is not operative in the emergency operating mode.

c.

An operating error that would not have prevented the EDG fiom being restarted and loaded within a few minutes and without corrective repairs.

d.

Observable abnormal conditions that would not have prevented the EDG from completing its mission during an actual emergency demand.

33 n

r ionL a -R nD This section examined EDG load-run data which met the followingcriterion:

The EDG was successfully started with an intention to operate at full load for a duration greater than 60 minutes.

The above criterion was used.to.eliminate data fi'om the 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> monthly rurts, which are petforn1ed at halfload ('1750KW), and to capture full(3500KW)'load-'run missions of significant duration (greater than 60 minutes).

Also, the criterion distinguishes between EDG load-run data and EDG.starting data by defining the load-run phase to begin only after'a successfuf start.

'For. comparison and analysi's purposes, the data

from sh'ort duration (60'minutes or less) load-runs are pYoviged separately'in Sectio'n 3.4 below.'.

Using the above criterion, a population of 128 load-runs was examined.

From this population, 2 load-runs were deleted because they were actual demands which operated the EDG at less than full load.

Also, 3 additional load-runs were deleted because the EDG was not operated at full load continuously during the surveillance duration.

American Electric Poorer

to AEP:NRC:0896V page 7 Examination of the remaining data showed that 16 full load-run missions were prematurely aborted.

This data required engineering evaluation based on the criteria presented in sections 3.1 and 3.2.

In general, premature termination was associated with minor fuel oil leaks, personnel error, or spurious operation or malfunction of equipment that would not be operative during an emergency condition.

Out ofthe 16 premature terminations, zero terminations were determined to be valid load-run failures. The remaining 107 load-runs provided approximately 2,010 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> ofload-run data for an average run time duration ofapproximately 19 hours2.199074e-4 days <br />0.00528 hours <br />3.141534e-5 weeks <br />7.2295e-6 months <br />.

3.4 Short Duration Load-Run Data Tliis section examined EDG load-run data which met the followingcriterion:

The EDG was successfully started with an intention to operate for a duration of not more than 60 minutes.

As part of the Individual Plant Examination (IPE), data collection and analysis were completed for EDGs over the period January 1983 through December 1992.

The test data primarily examined monthly surveillance runs and other short duration tests which involve loading the diesel to at least 1750 KW. The study reviewed a total of 929 demands and found 12 start failures and 3 load-run failures.

Per the above criterion, the 12 start failures were subtracted from the 929 demands to yield 917 valid load-run demands.

Also, based on review ofthe data and the cumulative run time hours, it was determined that.some of the 917. demands iqcluded runs over 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> (some of the demands'were actually 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> fullload-'runs).

In Section 3.3 above, the long'duration full load-run missions were examined.

Therefore, to prevent double counting and to condition these demands to a one hour average duration, 44 successful demands were deleted from the 917 remaining demands.

~ 'I I

.. 'Thus;-873 demands provided a total of'873 hours0.0101 days <br />0.243 hours <br />0.00144 weeks <br />3.321765e-4 months <br /> ofoperation.

The 3 load-run failures occurred witliinthe first 30 minutes ofoperation during routine testing.

American Electric Power to AEP:NRC:0896V page 8 35 Filre h

ri i

Prior to analyzing the above

data, it is important to review the underlying characteristics ofEDG failures. In general, EDG failures may occur as a result ofthe following:

a)

End ofnojvnal lifeofa component orpiece ofeqnPtment.

In general, preventative maintenance programs will reduce the number of age-related component failures.

However, there is some statistical uncertainty associated with EDG component lifetimes. This uncertainty creates the possibility of age-related failures.

Thus, EDG components are subject to failure aAer a requisite number of cumulative service hours, regardless of the length of the surveillance run during which the failure occurs.

Therefore, to properly interpret the observation ofa component failure during any given surveillance, it is important to carefully study the nature ofthe failure to determine ifthe component failure was precipitated by the actual surveillance duration or, rather, by the normal expected lifetime of a component (Appendix B ofreference 12, provides useful information on age-related failure mechanisms).

If the failure distribution of a particular component is known, then the time offailure can be predicted.

Therefore, it can be concluded that excessive testing increases the probability of such failures and that the surveillance test duration should be minimized to reduce the risk of failure during an actual demand.

\\

b) Rctndommalfimction ofcompone'nts, desigIt anomalies, or'incorject.mainte6 cmce.

These problems may surface over any time interval. In particular, it is intuitively obvious that the effects of small deviations from maintenance procedures may be.

.'ore dificult to identify and tnay'equire a ion'ger time'to appehr Tins', suggests

'hat, to identify all possible problems of this nature, the'surveillance test duration should be maximized.

Note that items (a) and (b) present competing requirements.

Accordingly, the objective of EDG surveillance testing should be to expose as many of the type "b" failures as possible without accelerating the type "a" failures. If, however, the test time exceeds an optimal duration, T, then the number oftype "a" failures being introduced could theoretically increase beyond the exposure rate of type "b" failures. Ideally, the surveillance test duration would be selected to minimize type "a"

American Electric Pou er to AEP:NRC:0896V page 9 failures.

In this case, it can be shown that a good approximation of the optimal test duration, T, can be determined by using an exponential probability density function for the EDG failure probability. The form ofthis function is shown below and is further developed in Section 3.6.

where t = time and, A, and a are constants.

It is also worth noting that it is reasonable to expect that, for a machine that is designed for extended operation at fully loaded conditions, a majority of the primary failure modes would occur before the engine reaches thermal equilibrium (usually within 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />). Engine equilibrium temperature is defined as the time at wliichthe jacket water and lube oil temperatures are both witliin+10'F oftheir normal operating temperatures established by the engine manufacturer (reference 11). The expectancy of a higher failure probability during the first two hours of operation is consistent with Cook Nuclear Plant actual performance data, as indicated in Sections 3.3 and 3.4.

3~6D A d In consideration of the failure characteristics discussed in Section 3.5, the following basic questions were addressed in this study:

a) Hev does the failuie probability duiing slrort dkuation runs conipare with long duration runs?

r '

b)'ver ivhat test diuatioii ivor/d the inajorily of-EDG load-iviii failures be.

expected?

To answer the above questions, it is necessary.

to develop a technical approa'ch for

~ ev'aluating'EDG unreliability as a function Ofsurveillance test 'time.

Therefore, the purpose ofthis section is to use actual performance data to establish a statistical model that conservatively characterizes EDG failure probabilities over time.

American Electric Poivcr to AEP:NRC:0896V page 10 Table I: Data Sananary Short Duration Load-Run Data 1983 - 1993 Total Demands 873 Total Failures Average Duration Hours Total Hours 873 L'ong Duration Load-Run Data 1974 - 1994 Total Demands 107

'appmzhute balue Total Failures Average Duration Hours19'otal Hours 2,010 The above data can be reasonably characterized as a set of Bernoulli trials.

That is, each valid test (or trial) can either be a success or failure, each success has a constant probability p, and each test is independent of the previous test.

It is recognized that some of the tests may not be independent and identical, however, in many cases the assumptions of a series of Bernoulli trials will provide a good approximation.

Therefore, the binomial distribution is a reasonable selection for evaluating this type of data.

Binomial Distribution Z =(n!/[(n-x)!*x!]}*p'"*(I-.p)"'I where Z equals the probability that out ofn trials x failures willbe observed given the failure probability, p.

~'ls'o, it can be p'roven, that thl;.most probable. value ofp is p=x/n For example, for the short duration runs, ps~ = 3/873 = 0.0034 and Z~ = 0.2244 American Electric Poorer to AEP:NRC:0896V page 11 However, for the long duration runs with xL s =0, a difFerent approach must be taken to calculate p (i.e., the result pt,& is not statistically meaningful to answer the above questions). By utilizing the binomial distribution function and the above data, a range for the parameter, p, can be determined forthe long duration load-runs.

For any given value ofn, it can be shown that the inequality, Z(x=0) > Z(x=1) is a bounding case and results in p < 1/(n+1)

Therefore, for n = 107, the most probable value for the parameter pt is less than 0.0093.

Also, based on the discussion in Section 3.5, it would be reasonable to expect that pL, > ps~ and, therefore, 0.0034<pL s<0.0093 For conservatism, assume pL

=0.0093, then the ratio p~p~ = 2.74 To answer question "a" above, this result means that the failure probability associated with the long duration load-runs is 2.74 times greater than the'failu're probability for short duration load-runs.

r P

Table 2: SNovnary ofIailare Probabklides

. Short Duration Load-Run Data Probabilities 1983 - 1993

¹Trials n

Failures X

Failure Failure Probability Rate short

/Hour Zshort 873 0.0034 0.0034 0.2244 Long Duration Load-Run Probabilities 1974 - 1994

¹Trials n

Failures X

Failure Probability pLong Failure Rate

(/Hour)

ZLong 107

~calcutated value.

0.0093'.0005*

0.3680 American Electric Popover to AEP:NRC:0896V page 12 As discussed in Section 3.5, the optimal test duration would produce failures predominately due to type "b" occurrences (random),

as opposed to type "a" (component wear).

As such, selecting an optimal test duration would mean that non-failure in previous test periods would not significantly change the failure probability in the future periods and also that the failure rate would be approximately constant over time.

To determine the optimal test duration, the integral ofthe probability density function, introduced in Section 3.5, can be used to model the probability ofEDG failures that could be observed over a time period, fO,t}.

EDG Failure Probability Function, P(t)

P(t) = (o/A)[1-e']

where n and X are constants.

The constant X, is dependent on the failure probability of the short and long duration'oad-runs which were defined as short dumtion runs 6 60 minutes < long duration rtms From Tables 1 and 2 above, at t = 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />, P = 0.0034 and at t = 19 hours2.199074e-4 days <br />0.00528 hours <br />3.141534e-5 weeks <br />7.2295e-6 months <br />, P = 0.0093.

These values can be substituted into. the above expression to first calculate A,'and then

~ u'. The resulting v'alues are shown below.

et=0.004294 A, = 0.4637 Figure 1, below, showsia >lot ofP(t) using the above values ofa and X'ove'r'.a'time period, t, froin'zero to twenty-four hours.:

Atnerican Electric Power to AEP:NRC:0896V page 13 EDG Failure Probability 0.0095 0.0090 0.0085 0.0080 0.0075 0.0070 0.0065 0.0060 0.0055 P(1) 0.0045 0.0040 0.0035 0.0030 0.0025 0.0020 0.0015 0.0010 0.0005 0.0000 0

2 4

6 8

10 12 14 16 18 20 22 24 TIME(hour$ )

~

Figure l:

Plot ofP(t)

Figure 1 shows that aAer approximately ten hours the EDG failure probability would increase at a very slow rate (i.e., dP(t)/dt = 0). To better understand this result and to answer question "b" above, a plot ofP(t) normalized by P(t&4 hrs) is shown below.

American Electric Pc>vcr to AEP:NRC:0896V page 14 1.0000 0.9500 0.9000 0.8500 0.8000 0.7500 0.7000 0.6500 0.6000 0.5500 P(i)IP(24) 05I 0.4500 0.4000 0.3500 0.3000 0.2500 0.2000 0.1500

.0.1000 0.0500 0.0000 Cumulative Failure Probability 0

2 4

6 8

.10 12 14 16 18 20 22 24 TIME(hours)

~

Figare 2: EDG Casnalative Failure Probability From figure 2, it can be seen that approximately 98% of the failures occur within the first 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> while only 60% occur within the first 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> ofoperation.

This may be an overly conservative estimation from the standpoint that, at Cook Nuclear Plant, all ofthe actual observed failures occurred within the first 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> ofoperation.

American Electric Popover

to AEP:NRC:0896V page 15 4.0 Review ofSelected Studies The purpose of this section is to provide a measure of EDG load-run performance on an industry wide basis.

The studies reviewed below were selected because they provided discernible EDG performance data during the load-run phase ofoperation.

(reference 7)

In 1985, Brookhaven National Laboratory completed a study to examine EDG performance at Nuclear Power Plants in the United States during a four year period, from 1980 through 1983.

The report did not distinguish between start and load-run failures but, rather, calculated overall failure rates normalized by total EDG years of operation.

Data was compiled from LER files, 10CFR50.55(e) reports, Nuclear Plant Reliability Data System, and Electric Power Research Institute document files.

The study included 158

EDGs, produced by 12 difFerent manufacturers (including Wortliington), for a total of588 EDG years.

The reported total composite number of failures was 396 which corresponds to a failure rate of approximately 0.7 failures per EDG year.

The report also studied the failure modes of di6erent EDG manufacturers, with the exception of Trans-America Delaval.

Also, the study provided a

qualitative presentation of predominant failure modes based on the number of reported failures.

Review of the report indicates that the five most dominant failure modes included instrumentation and controls systems, lubricating oil systems, speed and load control, cooling water systems, and starting systems.

In 1986, EPRI completed a comprehensIve study ofEDG success/failure experience at Nuclear Pow'er'Plants in the United States during.a.three year period from'983

'. through 1985.

The study'reviewed start and load-run reliability. The criter'ia used to evaluate load-run data was similar to that used in Section 3.0 above.

The load-run database included both planned and unplanned (actual) demands that involved EDG operation at greater than 50% ofplant essential safety function load rating for one hour or longer. For the period under study, a database of 13,808 load-runs was assembled.

The study reported 138 load-run failures wliich corresponded to an EDG unreliability of0.0100 failures per demand.

American Electric Poivcr to AEP:NRC:0896V page 16 43

/ R-5 4 (rc(erence 4)

In 1994, NUREGICR-5994 presented recent operating experience to assess EDG unavailability due to testing, maintenance, and failures during reactor power operation and during plant shutdown.

Section 3 of the report provided EDG industry-wide performance data over four years, 1988 through 1991, and used the empirical Bayes method to analyze EDG failure data.

The data covers 195 EDGs at 63 plant sites and includes both actual and test load-run data.

For plant sites, the mean load-run failure probability was 9.5E-03 failures per demand with a variance of2.9E-05.

Tnble 3:

EDGIndustry-wide F'ailare Probability History 1980 -1983 (reference 7) number of Failures 396'umber of.

EDGs 158 number of EDG ears 588 failures per EDG ear 0.7

-1983 - 1985 (reference 5) number of Failures:

number of EDGs.

number of load-runs failures per demand 138

. 154

.13,808:

.0.01 1988 - 1991 (reference 4)

. number number

'of of Failures EDGs number of load-runs failures

'er demand'82 195 19,520 0.0095

'Includes both start ttt load-ntn failures.

Atnerican Electric Popover

to AEP:NRC:0896V page 17 5.0 Indust Surve s

In an e6ort to provide more recent EDG performance data, a limited informal industry survey was conducted to specifically determine the approximate number offailures occurring before 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> and the number of failures occurring aAer 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> during the 18 month 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> surveillance test. Review ofthe data indicated that the industry-wide EDG failure rates during the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> surveillance tend to support the previous studies discussed in Section 4.0 (i.e.,

failures/load-run demand < 0.01).

6.0 Conclusions Examination ofthe full load run data on the EDGs at Cook Nuclear Plant revealed that 3 valid load-run failures occurred during short duration runs (t < 60 minutes) and zero failures occurred during long duration runs (t > 60 minutes).

A conservative, statistically based, mathematical model was developed to determine the reliabilityofthe EDG during the load-run phase oftesting as a function of time.

The model predicated that 60% of the valid load-run failures would occur witliinthe first two hours of operation, 95% in the first 6.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />, and 98% in the first eight hours.

Also, the more recent industry surveys, including the informal random surveys discussed in Section 5.0, are very consistent with the results found at Cook Nuclear Plant (i.e., failures/load-run demand < 0.01).

As explained in Section 3.5, the objective ofthe endurance test should be to expose as many of the rando'm type'failures as possible and to iMnimizeEDG component wear: Thus, the optimal.

surveillance test duration, T,'hould be selected to provide reasonable'ssurance that'the majority ofincipient random failures are exposed without adversely impacting.EDG availability for an actual emergency demand.

Also, consistent with actual performance data and engineering judgment, the period of higliest stress would eccur during startup and.before

'quilibriumconditions are established (0'5 T < 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />).".

Therefore, based on actual performance data and related industry-wide surveys, an endurance test duration of T = 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> is recommended to provide the necessary insight about EDG reliability. Test durations greater than eight hours would not provide a significant operating safety benefit and would only serve to increase cumulative run time and the likelihood ofage-related component failures.

American Electric Popover to AEP:NRC:0896V 7.0 References page 18 1)

"A Reliability Program for Emergency Diesel Generators at Nuclear Power Plants-Program Structure," NUREG/CR-5078, SAND87-7176, Vol.1, April 1988.

2)

"A Reliability Program for Emergency Diesel Generators at Nuclear Power Plants-Maintenance, Surveillance, and Condition Monitoring," NUREG/CR-5078, SAND87-7176, Vol.2, December 1988.

3)

"Improvements to Technical Specification Surveillance Requirements," NUREG-1366, December 1992.

4)

"Emergency Diesel Generator: Maintenance and Failure Unavailability, and Their Risk Impacts," NUREG/CR-5994, BNL-NUREG-52363, November 1994.

5)

"The Reliability ofEmergency Diesel Generators at U.S. Nuclear Power Plants," NSAC-108, September 1986.

6)

"ASimulation Study: Emergency Diesel Generator Availability,"EPRI NP-1759, Interim Report, March 1981.

7)

"A Review of Emergency Diesel Generator Performance at Nuclear Power Plants,"

Brookhaven National Laboratory, Technical Report A-3134 1-85, January 1985.

8)

"Guidelines and Technical Bases for NUMARC Initiatives Addressing Station Blackout at.Light Water Reactors," NUMARC87-.00, Rev.1, August 1991.

'I 9)

"Periodic Testing ofDiesel Generator Units Used as Onsite Electric Power Systems at Nuclear Power Plants," Regulatory Guide 1.108, Revision 1, August 1977.

.10)

"Selection, Design, Qualification, and Testing of Emergency Diesel Generator.Onits Osed as Class 1E Onsite Electric Power Systems at Nuclear Po'wer Plants," Regulatory Guide 1.9, Revision 3, July 1993.

11)

"IEEE Standard Periodic Testing ofDiesel-Generator Units Applied as Standby Power Supplies in Nuclear Power Generating Stations," IEEE Std 749-1983.

12)

"IEEE Standard Criteria for Diesel-Generator Units Applied as Standby Power Supplies in Nuclear Power Generating Stations," IEEE Std 387-1984.

Americnn Electric Popover to AEP:NRC:0896V page 19 13)

"IEEE Guide to the Collection and Presentation of Electrical, Electronic, Sensing Component, and Mechanical Equipment Reliability Data for Nuclear-Power Generating Stations," IEEE Std 500-1984 14)

"Resolution ofGeneric Safety Issue B-56, Diesel Generator Reliability," SECY-93-044, 02/22/93.

American Electric Popover

ATTACHMENT 5 TO AEP:NRC:0896V DONALD C.

COOK NUCLEAR PLANT COST BENEFICIAL LICENSING ACTION EDG SURVEILLANCE TESTING TECHNICAL SPECIFICATION CHANGES

Attachment 5 to AEP:NRC:0896V Page 1

Regulatory Requirement:

Technical Specification T/S 4,8.1.1.2.e.7 requires scheduling of EDG 18 month surveillance during shutdown (refueling outages) and performing a 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> endurance test run.

Effect of Requirement:

The present 18 month 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> EDG endurance test does not provide a safety benefit commensurate with its cost. In addition, performing the required testing causes critical path complications and delays during the outage.

Rationale for Regulatory Change:

The rationale for the proposed changes is that the intent of regulatory surveillance requirement to demonstrate the ability of the EDG to operate for an extended period of time under fully loaded conditions willbe preserved by maintaining an 8 hour9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> endurance test run.

At the same

time, these changes will better utilize plant resources and prevent critical path complications during the outages.

Unnecessary diesel generator stress and wear will also be reduced.

Additionally, the reduction from 24 to 8

hours will increase diesel availability and,

thereby, reduce shutdown risk.

Approximate Cost of Require'me6t:

The cost savings associated with reducing the 18 month 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> EDG surveillance test duration is provided below.

The cost savings was calculated.using'n estimate of.labor and materials required for"

. perfo'iming the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> endurance run (on two diesels per unit) and' multiplyixig this estimate'y the proposed reductibn in surve'illance duration.

~

Unit 1: remaining cycles 13 Unit 2: remaining cycles 15 2*13*16,000*16/24

$277,333 2*15*16,000*16/24

$320,000 Total plant lifetime savings is approximately

$600,000.

.0