ML20084N941

Reliability of AC Power at Grand Gulf Nuclear Station, Interim Assessment Rept

ML20084N941
Person / Time
Site:	Grand Gulf
Issue date:	05/14/1984
From:	Unione A ABB IMPELL CORP. (FORMERLY IMPELL CORP.)
To:
Shared Package
ML20084N916	List: AECM-84-0123, Forwards Reliability of AC Power at Grand Gulf Nuclear Station, Interim Assessment Rept ML20084N941
References
NUDOCS 8405170298
Download: ML20084N941 (44)

Text

--

RELIABILITY OF AC POWER AT GGNS An Interim Assessment A. J. Unione Impell Corporation May 14, 1984 l

5 I

t r

I i

L t

?

I IMP _EL.LPr pg.ggg ,. g -

P i PDR 1

TABLE OF CONTENTS

1.0 INTRODUCTION

1-1 1.1 Sunnary of Purpose and Scope 1-1 1.2 Conclusions of Study 1-1

.1.3 Report Organization 1-3 2.0 ESTIMATION OF LOSP FREQUENCY AND RECOVERY TIMES 2-1 2.1 Frequency of Total Grid Loss Assuming Independent Outages 2-1 2.2 Frequency of Coupled Outages 2-2 2.3 Estimated Time to Recovery of Offsite Power Links 2-6 2.4 Seasonal Affects on the Frequency of LOSP 2-7 3.0 AC POWER SYSTEM RELIABILITY MODEL 3-1 4.0 QUANTIFICATION OF AC POWER SYSTEM RELIABILITY 4-1

'4.1- Reliability Data 4-1 4.2 Base Case Reliabilty Evaluation 4-7 4.3 Sensitivity of AC Power Reliability 4-7 4.4 Affect of Seasonal Variations in Adverse Weather 4-10 5.0 RESULTS ANALYSIS 5.1 Comparisons of GGNS AC Power Reliability with Other Studies 5-1 5.2 Discussion of AC Power Loss as an Incremental Risk 5-4

= -

1.0 INTRODUCTION

1.1 Summary of Purpose and Scope The purpose of this document is to provide preliminary (l) reliability calculations which can be used to evaluate the acceptability of AC power system performance at GGNS. This includes an evaluation of the system design as proposed in the GGNS FSAR, along with estimates of reliability under a variety of interim configurations and conditions. Specifically, the following information is supplied.

1. Preliminary estimates of the reliability of offsite power sources.

Included are estimates of the frequency of losing all offsite power supplies to the site simultaneously (LOSP), the conditional probability of recovering at least one power source to the site as a function of time, the conditional probability that a LOSP is tornado caused, and the sensitivity of the frequency of a LOSP to the relative likelihood of tornadoes at various times of the year.

2. Reliability of AC power at the site for Division 1 and 2 loads. This includes the frequency of losing offsite sources and the unavailability of onsite AC power from diesel generators (D/G).

3. Reliabilityofabackupgasturbinegenerator(GTG)emergencyACpower system to start up and assume ESF bus loads for the duration of a LOSP given failures of the D/Gs. This includes an assessment of inherent unit reliabilities as well as sources of human error.

4. Sensitivities of- AC power reliability to variations in expected D/G reliability and to modified operating scenarios (i.e., assuming that one D/G is out for inspection).

5. The comparative reliability of AC power for an integrated system consisting of D/Gs and a backup system of GTGs to provide highly reliable power to key ESF loads.

6. The comparative reliability of AC power at GGNS to that at other plants based on information from reliability review.

7. The potential importance of AC power reliability as an impact on public safety at GGNS (as measured by core melt frequency), along with a discussion of the potential effects of modified AC power reliability on risk to the public.

1.2 Conclusions of Study Conclusions which can be drawn from the analysis in this document are as follows: ,

(1) A more detailed reliability evaluation of the AC power system at GGNS is currently being performed by PF&L.

1-1 4

1.0 INTRODUCTION

1. The GGNS AC power system is a reliable source of power for Division 1 and 2 ESF loads. Based on the estimated frequency of complete offsite LOSP (all three sources), median time to recovery of at least one offsite source..and reasonable reliability of the TDI D/Gs to start and bear load for the duration of the LOSP, unavailability of AC power is expected to occur with a frequentcy of approximately 5x10-4/R-yr using generic industry data. This estimate does not include availability of the GTG AC power system. This frequency compares well with NRC estimates of the frequency of AC power loss for a cross section of nuclear power plants and power system design configurations, using plant data and plant specific D/G experience where possible. Current GGNS starting experience with the TDI D/Gs indicates a higher starting reliability than that used in the calculations. In addition, a conservative estimate of the time to recovery of at least one offsite source was used. Therefore the predicted AC power system reliability is a conservative estimate of expected system performance.

2. The GTGs represent a reliable additional source of emergency power for the AC power system. For a system success criteria requiring at least 2 of 3 GTGs to start and bear load for the duration of the LOSP, system reliability is estimated to be approximately 967, using current industrial reliability experience with the particular GTG design used at GGNS. The GTGs do not reside inside an area protected from tornado damage.

However, it is not expected that a single event such as a tornado could fail all offsite sources and the GTGs simultaneously, since anticipated paths of any tornados going through the site would be unlikely to intersect both the GTGs and the switchyard.

3. An enhanced AC power system with GTGs used as a backup for the existing TDI D/Gs provides substantial additional protection against a loss of AC  !

power in the event of a LOSP. For postulated situations in which the AC power system is operating under additional constraints including a) higher anticipated failure rates for the TDI D/Gs during operation, or b) unavailability of one TDI D/G; addition of the GTGs decreases overall AC power system unreliability to a level of 2.7x10-5/R-yr and 1.5x10-4/R-yr respectively, which exceeds the reliability of the original system configuration. For these preliminary estimates it was assumed that a loss of all AC power to Division 1 and 2 ESF buses for 25 minutes was acceptable, based on station blackout studies.

4. Removal of 1 D/G from service for up to three months during the portion of the year when tornadic action is at a minimum in Mississippi, further assures the reliability of AC power at the GGNS site, provided that the GTGs are available as a source of power to the ESF bus.

5. Reliability of the AC power system at GGNS compares favorably with AC power reliability at other plants given that similar assumptions are used in the reliability assessment. Reliability of the AC power system, even with degraded D/G reliability, does not pose a significant contribution to public risk; especially when the GTGs are utilized as a backup power source.

1-2

l.0 INTRODUCTION 1.3 Report Organization Supporting technical arguments for these conclusions as provided in this document are organized in the following manner.

1. Section 2.0 evaluates an expected frequency for a LOSP at GGNS and estimates the time needed to recover at least one source of offsite power. Estimates are prepared using MP&L data..

2. Section 3.0 provides a preliminary reliability model which is used to evaluate AC power reliability in the event of a LOSP.

3. Section 4.0 provides a review of reliability information and a data base which is used to quantify AC power reliability. Section 4.0 also provides a sensitivity evaluation of AC power reliability under a variety of conditions and assumptions.

4. Section 5.0 provides a comparison of AC power reliability results for GGNS with those derived in other studies. In addition, a perspective on the contribution of AC power to public risk at GGNS is provided.

l l

l-3

2.0 ESTIMATION OF LOSP FREQUENCY AND REC 0VERY TIES Establishing a reasonable estimate of the likelihood of a complete loss of offsite power for the GGNS site is a key element of an analysis of the frequency of AC power loss. The RSSMAP for GGNS suggests a frequency of

.2/R-yr for a complete loss of grid. The following arguments are given to illustrate that for the GGNS site, a lower LOSP frequency than that used in the RSSMAP can be supported based on historical experience for W&L's grid.

The GGNS site has three incoming transmission lines, two 500 KV lines and a single 115 KV line. The possibilities for losing all three lines at the same time include:

independent and simultaneous outage of all three lines

-. loss of the grid due to an instability (such as that which occurred in the northeast U.S. in 1967)

. coupled loss of all power sources locally due to conditions such as adverse weather or a seismic occurrence.

Loss of all offsite power links to a plant as a result of a grid-related

. instability has never occurred on W&L's grid or other grids in the local power pool. It is not considered a reasonable contribution to the frequency of LOSP.

Of the events which can cause a coupled loss of power, the tornadic events represent the principal vulnerability and their effects are evaluated herein. A review of transmission line outage data for the W &L grid reveals that of the approximately half are tornado-induced events. All of the events which have caused outage of more than one major transmission line simul-taneously are tornado-caused.

Using W&L's transmission line outage experience and the assumption that failure rates are constant over time, estimates of LOSP frequency are produced along with measures of uncertainty.

2.1 Frequenc.y of Total Grid Loss Assumina Independent Outages From the Data Base in W&L submittal AECM-84/0241, dated 4/18/84 there were:

- 39 outages in 500 KV lines in 12 years, and

- 8 outages in the 115 KV line which connects Natchez, Port Gibson, GGNS and Baxter-Wilson PS in 12 years.

2-1

2.0 ESTIMATION OF LOSP FREQUENCY AND RECOVERY TIMES i I

For the 115 KV line this corresponds to P3 = .727 outages per year.

For the 500 KV lines,

- there are 520.3 total miles of 500 KV line in the system

- the outage rate per mile per year can be assumed to be constant, and

- the distances from GGNS to the two power sources which are in L turn supplied by multiple feeders is 21.96 mi. and 43.53 mi.

respectively.

Therefore the failure rate /500 KV mile - year is:

f 500KV=szu.jxiz = .006 outages / mile - year The independent failure rate for each of the two segments connecting GGNS l with other plants is: l P j=f500KV x 43.53 mi = .272 outages / year j x 21.46 mi = .134 outages / year. l P2"IS00KV The frequency of all three lines into GGNS failing due to independent  !

failures can be approximated by:

{

P=PjxP2 xP3.C12 . Cl3 i

= 5.5 x 10-5/R-yr

=.134x.272x.727x(M)

Where C 12 and C l3 represent the conditional " time at risk" when loss of additional lines would represent a threat. For the purposes of this  !

study,.the " time at risk" is essentially a conservative estimate of the time needed to fully restore a downed 500 KV line. ,

2.2 Frequency of Coupled Outanes GGNS has 3 independent powerline interties running in diverse paths from )

the plant to the rest of the grid. Reviewing the data base from the W&L  ;

submittal, there have been no cases on the W&L grid over the last 12 t years in which 3 independenTand diverse incoming lines have been lost to i a plant. There are 3 plants in W &L's grid with similar offsite power i configurations, GGNS with 6 years, the Baxter-Wilson PS with 12 years and  ;

the Ray-Braswell PS with 12 years. i I

Using the information that no simultaneous failures of three incoming power lines to a plant have occurred in 30 plant years of observation, an  ;

exponential estimator technique can be applied by assuming I failure in i the near future. Thus, 2-2  !

i i

_ _ _ --.- _ -_ __.__-----~__---_ _ _---___._ -.__ _ ___ _ _ _ _ . - -

2.0 ESTIMATION OF LOSP FREQUENCY AND RECOVERY TIES  :

t l

t I

i

- a mean outage rate of 0.3/R-yr is produced }

- a 905 confidence upper bound on the outage rate of .16 is l produced.

l t

As a second estimating technique, the data on 2 simultaneous line outages i to a plant is used. There were 4 instances in the HL data base which 2  ;

lines were out at the same time; l

- 3 of these have occurred at GGNS since 1978

- 1 of these occurred at the Ray-Braswell Station. I Of the 3 GGNS instances 2 can be removed since

]

- they involved a 500 KV line which was placed out of service i voluntarily because of the danger of high water, and

- the line has been moved and would no longer be threatened by high .!

4 water.  !

l There are therefore 2 valid outages of.2 incoming lines in 30 plant years i or .067 outages / plant year. ,

1 A third line can be presumed to be lost, either as a result of the same  !

event which caused the other two outages, or as a result of an indepen- i dont failure occurring during this time. j r

For an independent failure of a third transmission line to occur, an 'F estimate of the time at risk is required. For the failure which occurred ,

at the Ray Braswell Station, the outage time for recovery of at least one more line was 388 hours0.00449 days <br />0.108 hours <br />6.415344e-4 weeks <br />1.47634e-4 months <br />. For the failure at GGNS, an estimate of outage ,

time is more difficult to assess since the plant was under construction l and incoming power was not needed. It is not anticipated that GGNS would i be operated with two incoming lines unavailable, therefore, the limiting l case at risk for at risk time is the approximately 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> needed to attain cold shutdown given that two incoming lines have been lost. f

' Considering the worst exhibited reliability, that of the 115 KV line, i

P 3 = .727 outages / year = 8.3 x 10-5 outages / hour Therefore the conditional probability of a third outage given that two j lines are already out is .006.  !

To estimate the conditional likelihood that an third power source is l i.

failed from the same event which caused the first two failures, the two l i recorded cases can be reviewed.

h 2-3  !

I

2.0 ESTIMATION OF LOSP FREQUENCY AND RECOVERY TIMES i

i l

1 For the Ray Braswell event, accounts are that the two incoming lines were downed by separate tornadoes which struck the switch yard. Thus a third line would not necessarily be damaged and the conditional probability of i a third line failure is low. Conversely, for the GGNS event, a single  !

tornado hit the switch yard. This implies that the third line could  ;

easily have been damaged, althcugh the 115KV line was not damaged. The  !

data suggests that there is not a clear basis, even qualitatively, for assessing a conditional likelihood of failure. Therefore, a value of .5 is used.  ;

i The frequency for a complete loss of grid for GGNS can be stated in equation form as

1. LOSP " II*IC i where fg is the occurrence frequency for three independent failure and fC  ;

is the frequency of coupled failures. In turn  !

IC*#2 [Pg3+PC3 3 where 2f is the occurrence frequency for a loss of two lines, Pj3 is the l conditional probability of a third independent failure and PC3 is the j coupled probability of a third failure as a result of the same event  ;

which caused the first two.  ;

Using the estimates discussed above f

LOSP = 5.5 x 10-5 + .067 [.006 + .5] = .034 If the data on other plant sites were excluded from consideration leaving ,

only GGNS specific events, only one failure since 1978 is coanted. In this  ;

case f2 = .167. Also, if it were assumed that the likelihood of a third l failure given two failures is unity the LOSP f requency would become. l t

l f LOSP

= 5.5 x d + .167 = .167 f which is consistent with the frequ::ncy used in the NRC's RSSMAP of GGNS. l i

EPRI data NP2301 Loss of Offsite Power at Nuclear Plants. Data and l

Analysis) (illustrated in the accompanying Table 1 shows a wide variation l

!- in LOSP frequencies, principally due to site specific experiences.  !

l Arkansas Nuclear One (ANO) is the only site in SPP, and although ANO 1 i

! & 2 have never lost offsite power, the low experience base of 7 years i

! indicates using the EPRI approach a of no better than .15. f l

2-4 6 l [

f

,_. . , . , _ , _ . . . _ . ~ _ _ _ _ . . . _ . . _ - , _ _ , . . . _ _ _ . . _ . _ - . . . , _ , . . - _ . _ , - , _ _ _ _ . _ . _ _ _ . _ _ _ . _ , _ _ - _ _ _

2.0 ESTIMATION OF LOSP FREQUENCY AND REC 0VERY TIMES Table 1

, EPRI LOSP FRQUENCY AND RECOVERY TIME ESTIMATES BY REGION MEDIAN LOSP FREQUENCY REC 0VERY TIME REGIONAL COUNCIL (EVENTS / SITE YEAR) (HRS: MIN)

NPCC .153 :19 MAAC .061 1:24*

ECAR .338 1:11 SERC .046 1:24*

MAIN .076 1:23 MARCA .204 :29 SPP .149 --- **

WSCC .090 :06

• MAAC and SERC aggregate estimate

• • No recovery time data for SPP sites 2-5

x .

2.0 ESTIMATION OF LOSP FREQUENCY AND REC 0VERY TIMES i

I A summary of the estimates and uncertainty bounds which are produced  :

using the existing W &L grad data is shown in Table 2. Based on the size l of the data sample for the W &L grid a best estimate of l f

LSOP

= .1/R-yr is used in the reliability analysis in order to acknowledge the good grid j experience, and to conservatively account for uncertainties in the data j base.  !

l 2.3 Estimated Time to Recovery of Offsite Power Links {

To estimate time of recovery of at least one offsite power link, two i sources of data can be used:  !

- EPRI NP2301 data on recovery times, and Data specific to the W &L system.

A sunnary of the EPRI data for regional outage data is shown in Table 1.

Notably, the estimates of recovery time are short. The EPRI report  !

' noted, however, that recovery time for events caused by adverse weather  ;

conditions were much longer. On a national basis, an average of 7.2 hrs. j for time to recovery was estimated. j For the W &L grid the probability of losing single lines due to tornadoes  !

is high and the probability that multiple lines failures are due to  :

tornado effects is overwhelming. W&L data on outage times for specific events includes data on 500 KV and 115 KV lines. Using both data sources l and weighting the outage times by .67 and .33 respectively, an average recovery time of_68 hours. ,

Although this average includes the 115 KV line data it is highly skewed  :

by the 500 KV data, since recovery times for the 115 KV line are more j consistent with EPRI data.  !

It seems likely that the 115 KV line would be recovered first. A best

• estimate of mean recovery time is made by assessing the shortest single f' line recovery time which in this case is the 115 KV line. ,

On May 3, 1984 the 115 KV line was lost due to adverse weather. Recovery I time for the May 3 incident is estimated to be less than 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br />. l Including the May 3 incident in the data base for recovery times produces i an average recovery time of less than 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />. Based on l

- W &L data on recovery of the 115 KV lines, and

- EPRI estimates of recovery time for a LOSP i 2-6

'V 2.0 ESTIMATION OF LOSP FREQUENCY AND RECOVERY TIMES a conservative median recovery time of 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> to re-establish one incoming line to the plant is used.

Figure 1 illustrates a log-normal model for recovery, with a median recovery time of 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> and an error factor of 5. The assumption of a log-normal model is consistent with existing risk assessments, and is used to assess the conditional probability of not recovering at least a single incoming line for specific time periods following a LOSP.

2.4 Seasonal Affects on the Frequency of LOSP The frequency of loss of offsite power at GGNS is dominated by failures as a result of tornadic activity. This activity is most pronounced in the early spring as shown in the accompanying Figure 2. Using data for Mississippi compiled over a 32 year period, a reduction in tornado and subsequent offsite power' loss frequency of 45% can be established by restricting consideration to a 90-day window beginning on June 1.

2-7

2.0 ESTIMATION OF LOSP FREQUENCY AND REC 0VERY TIMES TABLE 2

SUMMARY

OF LOSP FREQUENCY DATA REVIEW FREQUENCY OF LOSP (1/R-yr)

MLE(Note 1) LB(Note 1) UB(Note 1)

DATA SOURCE f f.05 f.95

. RSSMAP For GGNS (Note 2) .2 - -

. EPRI Estimate For SPP (Note 3) .15 .01 .45 Based on W&L Experience (Note 4) a) 3 simultaneous failures .03 .002 .16 only (Note 5) b) 2 simultaneous failures with conditional third .034 .006 .11 failure (Note 6) c) GGNSonly(Note 7) .17 .009 .79 MP&L Case b) with " Quiet .019 .0005- .044 Season Impact" 2-8

2.0 ESTIMATION OF LOSP FREQUENCY AND RECOVERY TIMES Note 1 the technique used for statistical estimation is that utilized in EPRI NP-2301. As such the time occurrences of a LOSP are assumed to be exponentially distributed, the maximum likelihood estimate (MLE) is M/T where M is the number of occurrences in T years, and the 90% confidence bounds are a = .025/2T and a = .95/2T (the a's are chi-square fractiles for .05 with 2m degrees of freedom and .95 with 2m + 2 degrees of freedom respectively). For no failures experienced, MLE = 1/T is used as a conservative estimate.

Note 2 a conservative estimate with no supporting analysis is used.

Note 3 EPRI data for the Southern Power Pool is used. At the time of the EPRI analysis there was only one site in the region (Arkansas Nuclear) with 7 years of operating experience and no recorded incidents.

Note 4 three plants in the MP&L grid have similar intertie designs (ie. 3 incoming lines and diverse incoming paths) GGNS with 6 years of experi-ence, Baxter-Wilson PS with 12 years of experience and Ray-Braswell PS with 12 years of experience.

Note 5 no recorded cases of three failure simultaneously have occurred in 30 plant years observed experience.

Note 6 two recorded cases of two simultaneous failures have occurred in 30 plant year with the conditional probability of a third failure assessed as .51.

Note 7 only 6 years of observation at GGNS are used with single failure of two lines and the conditional probability of a third failure assessed as .51.

G 2-9 l

-- , - - , - - ,. -, - - - - - - . . - - - , - . - , a , ..,,,- - ,.,-...n--, - +..a- - - - , - --

Figure 1.0

. Log-Normally Distributed Recovery Tim Expected Recovery Curve

/

95% - I X

50% -

P R [

'essim stic lecovery, curve 5% - - /

2.5 10 20 30 40 50 60 70 Hours After LOSP l

l l

l l

l l

t

i

. i iia i 1 . _Jj Iw3--

I IIF I _i -

i

,I FIGURE 2.0 Annual distribution of tornado frequency in Mississippi, 1950 - 1982.

I l I

J ht i

I

• i s i

l l

, i I i .. _

i ._-

' i I i i

1 I I I

i E

120 i ,

i t

! i I I i i e g j, '

g -

I 11

1. '

i i '

4 A 15

{

~

It 1 i '

n ' '

ri 100 S , ee ti .-

i

(- i ,

44 I 1 i o ____- -

7 i i k i

}f' 2 r:

L 1

i , t g i i l i 1 x -

-+. ti i1 i n .

r m 80 .

i 1 1 I I I

,r 't

\ l !

J

\

I 1 1 f

J 'l I 1  !

1 I i I r

- i 1 63 i 1 L

I g '

.f 1 i

/ ) i i i # _

/ L i

I 2 P 1 6 .

-r i

\ '

j' I

/

t 40 / I

/ \

./ h -

,r l I 1  ;

f i -

f

\ I,

'# f (-- 1

( 1 ,J _

20 b q-- /, x 1 /-

__C% SF/

e hk g i I )

' I I j ,

-1.

~ dor,th -

t . '.

I1 3I I

i IIIIIiillIIIIIIi I I I j

'hy i ,

1 I I I I I I I11I 1 !I1' ' I I t I C A J F M M J J A 1

. S 0 N D A

I 3.0 AC POWER SYSTEM RELIABILITY MODEL  !

i The purpose of developing an AC power reliability model is to provide .1 preliminary forecasts of AC reliability for- the GGNS site and the logical ,

, relationship between major components and AC power system reliability at  ;

the site. Modeling of the AC power system is not intended to provide a j detailed reliability evaluation and analysis of failure modes. l A fault logic model for AC power reliability is presented in Tables 3 and  :

4. Dependence models are used for both the diesel generators and the gas  !

turbine generators to provide more realistic reliability estimates. For i starting reliability of diesel generators there is experience indicating  !

'that failures of multiple units are not totally independent.

Failures of instrumentation and controls associated with supplying startup signals for the diesel generators are not modeled directly, since l these are usually identified as relatively minor contributors to the j unavailability of a single diesel generator. Likewise failure of support i 3

systems associated with cooling the diesel generators are not considered  :

directly.

Gas turbines are modeled as being directly equivalent to at least one diesel j choice. generator in terms of being able to supply power to an ESF bus ofSinc!

ability to start and bear load in 10 seconds and the gas turbines are  ;

! expected to be on line in approximately 22 minutes, the assumption of  !

equivalency requires some explanation. l t

For the loss of offsite power initiator, loss of all AC power to Division l 1 & 2 loads would not impair coolant makeup to the core, since the HPCS l i

and its attendant diesel generator are still available. The major l requirement for Division 1 and 2 loads given the occurrence of a LOSP, is  !

i associated with removal of decay heat. Therefore the gas turbine system i starting time does not impair its usefulness for this purpose provided j that the Division III HPCS 0/G is operable. It should be noted that W &L i intends to continuously operate the GTGS under certain conditions such as l adverse weather. .No credit'for this procedure is taken in this analysis. j Although not considered as an initiator in the AC reliability model, $

there is the possibility that a design basis LOCA can occur  !

simultaneously with a loss of offsite power. The combination of a LOCA l and a conditional loss of grid is estimated to be near the threshold of t 10-7, below which accident sequences are not considered credible. Thus ,

even though the GTGs do not provide a viable response capability for this event, the requirement for AC onsite power reliability is minimal; and  ;

the AC power system would be sufficiently reliable even with one D/G out i of service. -

l

\

i 3-1 l l

1

. .. , 1

- !~

3.0'ACiOWERSYSTEMRELIABILITYMODEL

J .,

f 4

k 4 g

! s.

Failures to start the gas turbine generators can result from failure of the gas turbine ignition system, failure of the APU start units or

, failure of the operating staff to effectively follow the starting procedure. Since the APU units are interchangeable, each unit has the capability of starting all three gas generators.

In Table 4 running reliability ef the D/Gs is modeled as b' eing coupled as such failure of one unit during operation could alter the likelihood of the other successfully operating for the duration of the LOSP. For ease of calculation and because corrmon cause failure experience for D/Gs consists mainly of startup problems, a simplified boolean expression for the reliability of both D/Gs is defined in Table 5. I i

. I l

((

l t' >

j lT 4

8' ,. ,

k

.[ i *

,. / - -

r l.

s '

t

+ -

f

_ .f q' Y p

1. a

~

>- Y 4. # r i g. -1 a a

j .  : , .s -

1. =(,,., ~r /

h*, #

.] , '. , .-  ;

ly , ,

, y - -

, , -t p N ;. v

-} ,

' p ,. a . . ,

27 , d! j - j <j p '

.i '$, s j,? ' ,

l

, c, t -

l . _

q .

r, .. .. , ,

h-e

, c l - ?,

,, p y '

... L

.s' 4*/.

y e

_ 'N4, p! ' ,

• \. (  %, .f .

/ .:l ' ,1 f s , V' ,

. 1. m '

.;j , ;) p ,n s - , >

/

, -,e.

1. e -

, ,.3-2  !

~

c t

/- /q [

^

f w p,.c -

l

. 2 o . .. _ _ _ _ -

3.0 AC POWER SYSTEM RELIABILITY MODEL Table 3 DESCRIPTION OF TERMS IN AC POWER SYSTEM RELIABILITY MODEL GATES AC loss of AC power to Division 1 and 2 ESF loads DG12 failure of both D/Gs to supply power to at least one division of ESF loads for the duration of the LOSP GT failure of the backup GTGs to supply power to and at least one division of ESF loads for the duration of the LOSP GTI2 Failure of GTGs 1 and 2 (1 and 3, 2 and 3) to supply power for (GT13, duration of the LOSP.

GT23)

GTIS failure of GTG 1 (2,3) to start or bear load (GT25, GT3S)

BASIC EVENTS LOSP loss of all offsite power sources to ESF transformers NRE conditional probability of not recovering at least one offsite power source from the time of the occurrence of the LOSP to the time when the reliability calculation is made DGIS independent failure of D/G 1 (2) to start or bear load (DG2S)

(DG2S)

DGlR independent failure of D/G 1 (2) to continue running for the (DG2R)

(DG2R) duration of the LOSP DGlM unavailability of D/G 1 (2) due to maintenance and testing (DG2M)

DG2SC failure of a second D/G to start or bear load given that a first has already f ailed to do so ,

3-3

e . ,. _ _.

I 3.0 AC POWER SYSTEM RELIABILITY MODEL I

i I

Table 3 DESCRIPTION OF TERMS IN AC POWER SYSTEM RELIABILITY MODEL U,ontinued)

BASIC EVENTS (Cont'd) 2 G2RC failure of a second D/G to run for the duration of a LOSP given that a first has already failed to do so CBDIV, - f ailure of circuit breakers connecting the GTG system with ESF buses CB1906 to energize HEGTCB f ailure of control room operators to correctly follow procedure for connecting GTGs with the one of the ESF loads LOSPTC conditional likelihood that a complete LOSP has been caused by one or more tornadoes from the same storm.

TORNGT condition 51 likelihood that GTGs have been rendered inoperable by the same event which caused the LOSP GTIR failure of GTG 1 (2,3) to continue running for the duration of the

.(GT2R,- LOSP GT3R)

HEGT1 f ailure of GTG operating staff to successfully execute starting (HEGT2, ' procedure for GTG 1 (2,3)

HEGT3)

GTIl starting failure of GTG 1 (2,3)

.(GTI2,

'GTI3)

'APul  : failure of auxiliary power units to provide sufficient starting (APU2, torque to GTGs APU3).

APU12 conditional failure of both alternate APUs to provide starting

-(APU13, torque to GTG given -that the preferred unit has not

=.

APU23) 3-4 e

3.0 AC POWER SYSTEM RELIABILITY MODEL 1

i Table 3 DESCRIPTION OF TERMS IN AC POWER SYSTEM RELIABILITY MODEL (continuea; BASIC EVENTS (Cont'd)

HESW12 failure of. GTG operating staff to successfully switch sources of l (HESW13, startup power to a second APU given that one has failed i i

HESW23)

GTIM GTG 1 (2,3) unavailable due to maintenance and testing.

(GT2M, GT3M HESYNC Failure of GTG operator to successfully synchronize and align a <

second gas turbine.to an ESF bus.

GT2SC Conditional failure of a second GTG to start given that a first has (GTISC,- failed to do so (this uncludes failures of operators to start a GT3SC)- second GTG given that they have not used a correct procedure for the

-first).

s l

I i

3-5

r. ,-,.-m....,.- ,..~%,,.._.,.-,, -. _ . - - - - - , - , _ ,y_-,e.-.-- -.. ... . . y -

c -..

3.0 AC POWER SYSTEM RELIABILITY MODEL

)

[

t Table 4 ,

BOOLEAN EXPRESSIONS FOR AC POWER SYSTEM RELIABILITY MODEL

.l AC = LOSP o NRE o DGl2 o GT DG12 = DGIS o DG2SC oMoM ,

+ DGlS o DG2R o DGlR o DG1M

+ DGlS o DG2M oMoM

+ DGlR o DG2S o Mo M i-

+ DGlR o DG2RC o Mo M ,

r

+ DG1R o DG2M o Mo E

+ DGlM o DG2S o Mo'M ,

+ DGlM o DG2R o Mo M GT = CBDlV + HEGTCB + CB1906 + LOSPTC o TORNGT

+ GT12 + GTl3 + GT23 + HESYNC  ;

i 3-6

e 3.0 AC POWER SYSTEM RELIABILITY MODEL 5

Table 4 BOOLEAN EXPRESSIONS FOR AC POWER SYSTEM RELIABILITY MODEL (continued) t i

GT23 = GT25 'o GT3SC oMo mom  ;

+ GT2S o GT3R o MoMoM

+ GT2S o GT3M o MoMo R

+ GT2R o GT3S o MoMo N ,

l + GT2R o GT3R o MoMoM ,

+ GT2R o GT3M o MoMo M

+ GT2M o GT3S o MoMo M t

+ GT2M o GT3R o MoMoM GT12 = GT15 o GT2SC oNoMo M

+ GTlS o GT2R oRoMoM

+ GTIS o GT2M o Mo GTIR o GT3M ,

i

+ GTlR o GT2S o MoMoM

+ GTIR o GT2R o MoMo M i

+ GTIR o GT2M o NoMo M  ;

+ GTIM o GT2S oMoMo D

+ GTIM o GT2R o GTIR o GTIS o GT3M i

, 3-7 i

__._ . _ . . . . . ~ - _ _ _ _ . _ . _ _ . . . , _

I F

3.0 AC POWER SYSTEM RELIABILITY MODEL  ;

i I

Table 4  :

BOOLEAN EXPRESSIONS FOR AC POWER SYSTEM RELIABILITY MODEL u,ontinuea)

GT13 = GTlS o GT3SC o Mo Mo GT2M ,.

+ GTlS o GT3R oMo GTIM o GT2M [

+ GT15 o GT3M o MoRo M

+ GTlR o GT3S o EoMoM i

+ GTlR o GT3S oEoMoM

+ GTIR o GT3M o NoEo GT2M

+ GTIM o GT3S o N o ET1T o N .

+ GTIM o GT3R o No UTTY o M GTIS = HEGT1 + GTil + APU1 o HESW1 + APUI o APU23C

+ APU1 o APU2 o APU3 GT2S = HEGT2 + GTI2 + APU2 o HESW2 + APU2 o APU13C i

+ APU1 o APU2 o APU3 ,

i

'GT3S = HEGT3 + GTI3 + APU3 o HESW3 + APU3 o APU12C

+ APUI o APU2 o APU3 i

L e

3-8

., ,. m...,_.. . _ _ . - ., _ _ . . _ . - _ -

3.0 AC POWER SYSTEM RELIABILITY MODEL p

i Table 5 BOOLEAN EXPRESSION FOR LOSS OF D/Gs GIVEN ASSUMPTION OF INDEPENDENT l RUNNING FAILURES DG12 = DGlS o DG2SC o DGlR oM ,

+ DGIS o DG2R oMo M ,

+ DG1R o DG2S oMo M

+ DGlR o DG2R o EoM

+ DGlM o DG2S o Mo R ,

+ DGlM o DG2R o Mo M  :

i

+ DGlS o DG2M o Mo R

+ DGlR o DG2M o Mo R i-c i

3-9 2

4 y em , - - - . - - - -

-n-- -,-,-

r-e -,w,w,ene - m-, .

n- r-r,, ,,,-, , ,-w,,-- -- -- --- ,- -

f 4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY  !

l i

y  !

I 4.1. Reliability Data  !

The AC power system reliability model discussed in section 2 was ,

quantified using a combination of site specific and generic data taken  !

froni a variety.of sources.' A summary of data sources is shown in Table

-6, with. details for some data sources and derivation procedures given in j the accompanying notes.

The following considerations were used in developing a working data base: l

1. diesel generator unreliabilities (start and run) could  !

conservatively be estimated using generic reliability data summaries  !

such as Wash 1400. Also diesel test and maintenance unavailability  !

could be estimated conservatively in the same manner. ,

! 2. cas turbine running reliability could be extracted from industrial experience for the same or similar gas turbine units.

., 3. . gas turbine starting reliability as well as APU reliability could be t inferred from telephone interviews.

4. unreliability of components such as circuit breakers is estimated i using generic reliability data.

!' '5. human error estimates for gas turbine operation are developed using

! NUREG/CR-1278 as a source of failure rates for individual ,

j . activities, gross estimates of the number and description of i l

activities involved, and the assumption that procedures will be i l- fully described and rehearsed by the GTG operating srew.  !

6. Common cause couplings for failure of more than one diesel generator i j are estimated from compilations of industry data. Common cause  !

running failures for diesel generators and running' failure of gas turbines are estimated only on the basis of engineering judgement.  ;

l.

i' r I

g .

I t

f P

l t

[

l.  :

f  !

l 4-1  ;

l  :

i

?

4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY i

Table 6 i RELIABILITY DATA FOR AC POWER RELIABILITY MODEL I

RELIABILITY FREQUENCY OF AC POWER, UNAVAILABILITY  !

DESIGNATOR AS A FUNCTION of TIME AFTER LOSP (hrs) C0tEENT  ;

0-2 2-6 6-24 24-72 LOSP, LOSPTC .1, .95 Section 1  ;

NRE .98 .85 .40 .08 Section 1 DG15, DG2S .3 x 10-2 - - -

GGNS RSSMAP DG1R, DG2R 3 x 10-3 1.2 x 10-2 4.5 x 10-2 1.3 x 10-1 Based on MTBF i of 333 hrs.  !

WASH 1400, Note 1 DG1M, DG2M 6 x 10-3 - - -

GGNS RSSMAP

-CBD1V 10-3 - - - WASH 1400 CB1906 10-3 - - - WASH 1400 l L

HEGTCB 5 x 10-4 - - - NUREG/CR-1278, Note 3 i

TORNGT 10-3 - - -

Qualitative Argument

- GT1R,' GT2R, 3 x-10-4 9 x 10-4 4.5 x 10-3 1.4 x 10-2 Note 2, Note 1 >

( GT3R GTI1, GTI2, , 5 x 10-2 - - -

AECM 84/0241 i- GTI3 ,

HEGT1, HEGT2, 5 x 10-3 - - - NUREG/CR-1278, HEGT3, HESYNC Note 3 l APUl, APU2, 6 x 10-2 - - -

Note 4 APU3, HESWl, HESW2, 1 x 10-2 - - - NUREG/CR-1278, HESW3 Note 3 l g

APU23C,APU13C 10-3 - - - Interview, '

APU12C Note 4 l

4-2 I

t 4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY Table 6 RELIABILITY DATA FOR AC POWER RELIABILITY MODEL (Continued) j a

RELIABILITY FREQUENCY OF AC POWER, UNAVAILABILITY l DESIGNATOR AS A FUNCTION of TIME AFTER LOSP (hrs) COMMENT 0-2 2-6 6-24 24-72 i GT2C, GT3C 10-1 - - - Interview, l GT1C Note 4 i DG1SC, DG2SC .19 - - -

Shoreham PRA t DG1RC, DG2RC .1 .1 .2 .2 Qualitative Argument GT15C, GT25C .15 - - - Comparison of  !

Meenanical .

Complexity with  :'

D/Gs GT1M, GT2M .02 Interview, Norez i

l t

t h

4-3

-.-.-l

.s ._

4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY  !

I

-Notes for TABLE 6

~

1. For the range of reauired operating hours average operating time was l used as an operational demand ,

2. gas turbine failures to start can be estimated in either of 2 ways l 1

-. data from Allison indicates a forced 6 outage rate of.08/1000 hrs I of unit operation, based on 7.6 x 106 hours0.00123 days <br />0.0294 hours <br />1.752645e-4 weeks <br />4.0333e-5 months <br /> of experience with-.the 50lk unit. This corresponds to a failure rate i

fFTR = 8 x 10-5/hr .

- other data also from Allison indicates a unit running availability of approximately 95%. ' Running availability -

is interpretable as the amount of time that power could be '

generated given that it was wanted all the time. As such,  :

scheduled maintenance and other outage sources would be included along with forced outages. . Without acknowledging scheduled  ;

outages, assuming an average outage time (T) of 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br />,

" 0.5 -3 f

FTR

= + = 1 x 10 /hr  !

T 48 P

- clearly the contribution of planned outages is therefore an L average estimate of FFTR = 3 x 10-3/hr is used along with l a maintenance outage unavailability of .02.  ;

l i

L I

f I k l

! [

i L  ;

l i

i i  !

f 4-4 j I

4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY  !

Table 6  !

RELIABILITY DATA FOR AC POWER RELIABILITY MODEL l i

(Continued)

{

.(3) A preliminary analysis of human' errors associated with gas turbine f operation, based on NUREG/CR-1278 is given below. l l

l;  ;

l~

Parameter Description of Basis for Base No TOTAL  :

HE Interpretation Failure Recovery l of NUREG/CR-1278 Rate ,

HEGTCB Failure of Oper- sin 31 e activity, 1 x 10-3 5 x 10-2 5 x 10-5 (R)  !

i ator to locate at least 1 check i and properly rack written procedure 5 x 10-3 1 x 10-1 5 x 10-4 (C) [

.in CB1906 from local indication  ;

-gas turbines l 1: I l

HEGT1 failure of remote multiple activities 1 x 10-2 5 x 10-2 5x10-4(R) >

l HEGT2 . operator to at least 1 check .

l HEGT3' correctly start,- written & rehearsed 5 x 10-2 1-x 10-1 5 x 10-3 (C) or align GTG procedure, local [

.with load indicators  ;

HESW1 failure of remote small number 5 x 10-2 5 x 10-2 2.5x10-3(R)  !

'HESW2 operator to of activities, ,

l j HESW3 correctly realign checked possible,  !

i l APU to different procedure, no gas turbine local indication i l

HESYNC Failure of Oper- some Gasis as 1 x 10-2 5 x 10-2 5 x 10-4 (R) rator to Synchro- -HEGTI 5 x 10-2 1 x 10-1 5 x 10-3 (C) nize at'least two .

i L GTGs to ESF' bus l

I f R - realistic  ;

C - conservative f t

Because of the relatively low amount of nuclear industry experience with GTG  ;

L operations. conservative estimates of human error rate are used in the reliability  !

[

analysis. ,

4-5

4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY Notes for Table 6 (Continued)

(4) Dan Maranacci of Obrien Machinery indicated by telephone conversation on

. May 2, 1984 the following information utilized in assessing the reliability of APU based starting for the GTGs. For three different starter systems used with the jet engine, the reliabilities are as ,

follows:

o hydraulic starter system close to 100%

with small diesel and hydraulic transmission o AW starter system with approximately 95%

compressed air reservoir for,startup o jet fuel based starting approximately 90%

system The Garrett Ind. APUs are not precisely equivalent to any of the above

. systems. However, Maranacci believed that the APUs would be only slightly less reliable than the compressed air reservoir system. To accommodate this data an unavailability of 6 x 10-2/d is used for the ,

APUs. The. potential for common cause failures was discussed. However, l

further discussion with Mr. Maranacci did not reveal any reasonable l

common failure modes.- Mr. Maranacci did feel that there may be a strong common mode coupling for gas turbines based on their vulnerability to a variety of fuel and air particulate problems.. Hence a conditional  ;

failure rate of .15 is accessed for a second GTG failure to run given that a first one has occurred.

l.

i e

I 4-6 -

V I

4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY l t

4.2 Base Case Reliability Evaluation

' The AC power system reliability model for GGNS was evaluated using the f data presented in Table 6. To estimate the potential impact of assuming i coupled diesel generator running failure an assessment of the probability  :

of losing both diesel generators as a function of time was performed and is summarized in Table 7. Based on the fact that reliability differences were only noted at the longest time periods and that data on common cause  ;

' failures to run are not generally available, the semi-independent diesel l l

failure model in Table 5 was used in the AC pcwer reliability evaluation.

Reliability of AC power for the base case conditions is summarized in  !

- Table 8. The base case power system reliability quantification was i performed, with and without consideration of the impact of gas turbines. i

. The resulting.AC unreliability does not vary significantly over time, j principally because the failure to run probabilities for equipment '  !

increase with time while the likelihood of recovering offsite power i t increases.  ;

The use of only point estimates of reliability data were could have a substantial effect when compared with estimates produced by simulation j models. t i

4.3 Sensitivity of AC Power Reliability

+

(

.A way to assess the potential importance of equipment reliability l problems and modified conditions'of operation is to perform a systematic ,

sensitivity evaluation using a reliability model. In truth,' the major i usefulness of the reliability model is to accommodate the identification of potential impacts of changes in equipment design or performance. i A set of sensitivity evaluations were performed using the AC power  !

reliability model developed for GGNS to assess the potential impact of: l l

1. adverse reliability-of'the diesel generators in the running mode  :

(i.e., using a failure rate to run fFTR = 10-2/hr. This is done i in order to accomodate the fact that due to design or fabrication  :

problems the diesel generators may be experiencing " infant deaths; i thus experiencing a failure rate of approximately three times the ,

industry average);

P i

I 4-7  :

4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY O

Table 7  !

COMPARIS0N OF INDEPENDENT DG RUNNING VS. DEPENDENT RUNNING MODELS

' t UNAVAILABILITY (1/d) 0F BOTH D/Gs IN TIME INTERVAL .

DG Reliability AFTER LOSP (br) '

Assumptions 0-2 2-6 6 - 24 24 - 72 Independent I Failures to Run .006 .007 .011 .30 .

Dependent * .006 .007 .018 .040  :

Failures to Run ,

• . A factor of 4 increses in fFTR for a second D/G given failure of a first D/G before 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, followed by a factor of 2 increase up to 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />. t l

l l

l l

r l 4-8

- - ._ .c __ , , . _ . . _ . . _ . . _ , . _ ..,.. .__ .-. ., - , _ _ _ . . -

4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY Table 8

SUMMARY

FOR BASE CASE EVALUATION OF AC POWER UNRELIABILITY AT THE GGNS SITE PARAMETER DESCRIPTION FREQUENCY (1/R-yr) OR PROBABILITY (1/d)

IN-TIME INTERVAL AFTER LOSP (hrs) 0-2 2-6 6 - 24 24 - 72 LOSP Occurrence of loss .1 .1 .1 .1 of grid NRE Likelihood that power .98 .98 .45 .08 to site on at least 1 incoming line has not been recovered DG12 Probability of losing .006 .007 .011 .030 power from both Div 1 and Div 2 D/Gs GT- Inprobability of not .039 .039 .041 .044 obtaining continuous source of power from GTGs

' AC ' Loss of AC Power to  ;

Division 1 & 2 ESF loads .

- D/Gs only 5.9x10-4 6.9x10-4 5.0x10-4 2.4x10-4 i

- D/Gs With GTGs 2.3x10-5 2.7x10-5 2.0x10-5 1,1x10-5 ,

as backup

• 4-9 l t

- - , , ,-- - . - - , - - - - -_o, .

4.0 QUANTIFICATION of AC POWER SYSTEM UNRELIABILITY

2. removal of one diesel generator from service in order to accommodate a thorough inspection and maintenance while the plant is on line; i

. 3. higher than anticipated frequencies of LOSP as a result of uncertainties on the present data base;

4. longer offsite power recovery times as a result of uncertainties in the present data base, and;

5. combinations of all of the above.

The results of the sensitivity evaluation are shown in Tables 9 and 10.

Table 9 is compiled for the AC power system, assuming no backup by GTGs; while Table 10 includes the affect of the GTGs.

The results indicate that the AC power system could experience a r significant decline in reliability (as much as an order of magnitude) with a single D/G out. Notably the evaluation indicates that maintaining '

the plant on line with two D/Gs exhibiting early failure tendencies is better than pulling one out of service if the other is good. Most important, however, the evaluation indicates that under all but the most pessimistic of assumptions concerning the frequency of LOSP and recovery times, the addition of the GTGs produces an overall AC power reliability i no worse than the base case estimate with no GTGs.

4.4 Affect of Sea _sonal Variations in Adverse Weather A basic task of this study was to determine the potential reliability  ;

affects of having a plant at power during a period of time when a single  !

D/G is unavailable, compensating for this by having a gas turbine system available.

The principal vulnerability of AC power is to tornado damage. Thus, the effect of seasonal variations in tornadic activity could potentially be important. ,

In Section 2.0, it was noted that the frequency of tornado damage comprises most of the frequency of offsite power loss; and that the several months, beginning in June, are a relatively benign period for tornadic activity. y The effect of having GGNS at power during the " quiet period" with only 1 D/G available is evaluated in Table 11. Notably, with best estimates of LOSP frequency and time to recovery for this period, the addition of gas turbines produces a significant impact on system reliability; such that if the conditional likelihood of GTG outage due to tornado (along  :

with LOSP) were underestimated by as much as an order of magnitude, the resulting conclusions about the adequacy of AC power would not be affected. ,

4-10

Table 9 SENSITIVITY RESULTS, AC POWER RELIABILITY FOR GGNS (N0 GAS TU'RBINES)  !

SENSITIVITY FREQUENCY OF LOSP AND AC POWER LOSS (1/R-yr)

CASE DESCRIPTION FOR INCREASING TIME AFTER LOSP (hr)

P 0-2 2-6 6-24 24-72 l

0 - BASE CASE, AVG. DGs 5.9 x 10-4 6.9 x 10-4 5.0 x 10-4 2.4 x 10-4

- Site specific fLOSP ,

i

- Best estimate PNRE .

1 - 1 DG out 3.8 x 10-3 4.7 x 10-3 3.5 x 10-3 1.3 x 10-3  ;

4

- Site specific fLOSP -

- Best estimate PNRE 2 - 1 DG out 4.5 x 10-3 7.3 x 10-3 7.7 x 10-3 3.2 x 10-3 ,

- 1 DG high FFTR~ i

- Site specific fLOSP

- Best estimate PNRE 3 - 2 DGs high fFTR 6.9 x 10-4 9.8 x 10-4 1.5 x 10-3 1.3 x 10-3 :

- Site specific fLOSP l'

- Best estimate PNRE 4 - AVE DGs 1.2 x 10-3 1.4 x 10-3 9,9 x 10-4 4.8 x 10-4

-fLOSP = .2 .

- Best estimate PNRE 5 - 1 DG out 9.0 x 10-3 1.5 x 10-2 1.5 x 10-2 6.5 x 10-3  ;

- 1 DG high fFTR $

-fLOSP = .2

- Best estimate PNRE 6 - AVG DGs 1.2 x 10-3 1.4 x 10-3 2.0 x 10-3 3.0 x 10-3

-fLOSP = .2

- Pessimistic PNRE 7 - 1 DG out 4.6 x 10-3 7.3 x 10-3 1.5 x 10-2 2.0 x 10-2 [

- 1 DG high fFTR I

- Site specific fL0sp -

- Pessimistic PNRE 1 8 - 1 DG out 9.2 x 10-3 1.5 x 10-2 3.1 x 10-2 4,0 x 10-2

- 1 DG high FFTR

-fLOSP = .2 f

- Pessimistic PNRE l

l 4-11  !

~

Table 10 .

SENSITIVITY RESULTS, AC POWER RELIABILITY FOR GGNS (WITH GAS TURBINES) l k

SENSITIVITY FREQUENCY OF LOSP AND AC POWER. LOSS (1/R-yr)

CASE DESCRIPTION SHUTDOWN WITH INCREASING TIME AFTER LOSP (hr) {

- l 0-2 2-6 6-24 24-72  !

F O - AVG DGs 2.3 x 10-5 2.7 x 10-5 2.1 x 10-5 1.1 x 10-5 l

- Site specific fLOSP -

- Best estimate PNRE j 1 - 1 DG out- 1.5 x 10-4 1.8 x 10-4 1.4 x 10-4 5.8 x 10-5  ;

- Site specific fLOSP  !

- Best estimate PNRE ,

2 - 1 DG out 1.8 x 10 2.8 x 10-4 3.1 x 10-4 1.4 x 10-4

- 1 DG high FFTR

- Site specific fLOSP t

- Best estimate PNRE  !

3 - 2 DGs high fFTR 2.7 x 10-5 3.8 x 10-5 6.1 x 10-5 5.8 x 10-5 j

- Site specific fLOSP

- Best estimate PNRE 4- - AVG DGs 4.6 x 10-5 5.4 x 10-5 4.1 x 10-5 2.1 x 10-5

-fLOSP = .2  !

- Best estimate PNRE l 5 - 1 DG out 3.5 x 10-4 5.7 x 10-4 6.3 x 10-4 2.8 x 10-4

- 1 DG high fFTR

-fLOSP = .2  :

- Best estimate PNRE i 6- - AVG DGs 4.7 x 10-5 5.4 x 10-5 8.1 x 10-5 1.3 x 10-4

-fLOSP = .2 t

- Pessimistic PNRE [

i

-7 - 1 DG out 1.8 x 10-4 2.9 x 10-4 6.3 x 10-4 8.8 x 10-4 l l- - 1 DG high fFTR I I- - Site specific fLOSP  !

- Pessimistic fNRE 8 - 1 DG out 3.6 x 10-4 5.7 x 10-4 1.3 x 10-3 1.8 x 10-3 i

- 1 DG high FFTR  !

l

-fLOSP = .2 i

- Pessimistic PNRE-t 4-12 I

Table 11 EFFECT of SEASONAL VARIATIONS on TORNADIC ACTIVITY on RELIABILITY of AC POWER of GGNS FREQUENCY OF LOSP AND LOSS OF AC POWER (1/R-yr)

SENSITIVITY CASES FOR INCREASING TIME AFTER LOSP (hr) 0-2 2-6 6-24 24-72  !

t

<- l l

O' f LOSP = .1, PNRE = Expected f

- AVG DG performance 5.9 x 10-4 6.9 x 10-4 5.0 x 10-4 2.4 x 10-4  !

t

- 1 DG out,1 with high fFTR 4.5 x 10-3 7.3 x 10-3 7,7 x 10-3 3.2 x 10-3

- 1 OG out,1 with high fFTR 1.8 x 10-4 2.8 x 10-4 3.1 x 10-4 1.4 x 10-4 I GTGs added l l i

o fLOSP = .019, P NRE = Expected l

- AVG DG performance 1.1 x 10-4 1.3 x 10-4 9.5 x 10-5 4.6 x 10-5 l

- 1 DG out, 1 with high fFTR 8.6 x 10-4 1.4 x 10-3 1.5 x 10-3 6.1 x 10-4 1 DG out,1 with high fFTR 3.4 x 10-5 5.3 x 10-5 5.9 x 10-5 2.7 x 10-5 GTGs added 4

I

[

t 4-13  !

l

c ..

5.0 RESULTS ANALYSIS t

5.1 Comparisons of GGNS AC Power Reliability with___0ther Studies j The results of this evaluation are preliminary and are useful for i indicating trends and order of magnitude impacts rather than detailed i

comparisons. It is useful, however, to compare results of this study j with other AC-power reliability calculations in order to place the ,

resultir.g AC power reliability at GGNS in perspective as an impact on (

plant risk.-  ;

t Results of the preliminary assessment are compared with power system i reliabilities in the Shoreham PRA in Table 12. The reliabilities compare  !

reasonably well for the first few hours after the LOSP. At later times, #

however, the use of a faster offsite power recovery rate and the  ;

consideration of repairs for diesel generators reduces the comparative  !

' unreliability of the Shoreham system. Normalizing AC reliability for Shoreham by removing D G repair, the reliabilty estimates are comparable. l i

An additional comparison is made with results of the NRC's study of l emergency AC power system reliability'(NUREG/CR-2989) in Table 13.  !

Results of the studies compared were not adjusted to account for  !

differences in the calculational techniques. However the resulting  ;

reliabilities can still be compared. What is relevant about the .

comparison with NRC results is that the use of the gas turbine system i maintains a comparatively good reliability for GGNS even with one D/G -

out. Thus even if a detailed analysis modifies the quantitative  ;

estimates, the assertion-that plant performance is acceptable with 1 D/G

~

out and the gas turbine system used as backings can still be made.

(

i 1

t l

l r

t 5-1 i i

,, , ~ - - , . ,

6 Table 12 COMPARIS0N WITH AC POWER RELIABILITY AT GGNS THE SHOREHAM PRA SNPS PRA AC POWER FREQUENCY OF LOSP AND LOSS OF AC POWER (1/R-yr)

FOR INCREASING TIME AFTER LOSP (hrs)

PARAMETER 0-2 2-4 4-10 10-24 Thoreman Unavailability of AC PWR 1.4 x 10-4 5.7 x 10-5 2.7 x 10-6 3.0 x 10-6 ivisions 1 & 2+

++

NRE .52 .28 .23 .06

'robability of No DG Repair

.88 .66 .47 .20 idjusted AC PWR Unreliability+++ 1.6 x 10-4 8.6 x 10-5 5.7 x 10-5 1.5 x 10-5 4C Calculations 9 GGNS 2.0 x 10-4 2.4 x 10-4 1.7 x 10-4 8.2 x 10-4 '

lith N3 DG Repair ++++

C Calculations 9 GGNS

. 7.8 x 10-6 9,2 x 10-5 7.1 x 10-6 3.7 x 10-6 lith Gas Turbine Backup fLOSP = .082 for the Shoreham site

'+ This is probability for not recovering offsite power for a given time.

'++ conditional likelihood of DG repair by given time is taken out

~+++ Calculated fLOSP of .034 is used. No gas turbine backup is assumed.

5-2

Table 13 COMPARIS0N GGNS AC POWER RELIABILITY WITH NRC FORECASTS (2)

FREQUENCY OF LOSP AND LOSS OF AC POWER (1/R-yr)

FOR INCREASING TIME AFTER LOSP (hrs)

STUDY 0 .5 8 i GGNS base case 5.9 x 10-4 5.9 x 10-4 5.0 x 10-4 (2) 1 DG OUT 1.8 x 10-4 1.8 x 10-4 3.1 x 10-4 (2) 1 DG high fFTR gas turbine backup i GenericPlant(1) 2.9 x 10-4 1.9 x 10-4 1.9 x 10-5

- 1 of 2 DG Required

- water cooled Nine Mile Point (1) 2.3 x 10-4 1.5 x 10-4 1.1 x 10-5

, ANO-1 (1) 9.7 x 10-4 7.5 x 10-4 8.6 x 10-5

> Davis-Besse (1) 2.5 x 10-3 1.5 x 10-3 7.1 x 10-5 (1) DG repair included (2) all data taken from NUREG/CR-2989, except GGNS 5-3

e - - _ _ _ _ _ _ _

5.0 RESULTS ANALYSIS ,

5.2 Discussion of AC Power Loss as an Incremental Risk It is appropriate to place AC power reliability at GGNS in perspective as a contributor to fisk, and to assess the following potential impacts on  !

risk:

1. expected AC power reliability at GGNS

2. decreased AC power reliability as a re5 ult of equipment reliability [

problems or taking a D/G out of service  !

3. modified AC power reliability due to placing one D/G out of service ,

and adding a system of GTGs in place.  ;

l To consider the effects of AC power on plant risk, the Grand Gulf RSSMAP '

is used since it provides a set of probabilistic core melt calculations ,

including core damage frequency estimates and event free sequence ,

contributions.

l In the Grand Gulf RSSMAP, the frequency of core melt was assessed as 3.7 x 10-5/R-yr. There were 9 core melt sequences which were initiated by  ;

a LOSP. Dominant contributing accident event sequences are the following TQUV - a loss of offsite power followed 1.5 x 10-6/R-yr-  !

by a loss of high and low pressure coolant makeup i TQW - a loss of offsite power followed 6.2 x 10-6/R-yr by a failure to remove decay heat .

from the suppression pool within (

30 hours3.472222e-4 days <br />0.00833 hours <br />4.960317e-5 weeks <br />1.1415e-5 months <br /> r r

TPQI - a loss of offsite power followed 1.6 x 10-6/R-yr ,

by a failure of 1 or more safety  !

relief valves to reseat, followed [

by a failure to remove decay heat  !

from the suppression pool  !

t Assuming that the remaining six core melt event sequences identified in the RSSMAP which were initiated by a LOSP contribute less than or equal to 10-7/R-yr to the core damage frequency; LOSP initiated events j contribute approximately 25% of the core damage frequency. }

5-4  :

en rw-,--- e ~ -me--,.e --ny -------vee n , a vw ,-e e- ,,nev,-,-,eme -- --- , - -wmwnm--na-. .-,-,-.,-- ,, ,,

5.0 RESULTS ANALYSIS l

4 Loss of. AC power was a major contributor to all three accident event sequences. Of the dominant sequences described, loss of one AC power train or complete AC power loss occurred in no less than 18%, 53%, and 34% of the cutsets for the three accident sequences respectively. l

. 1 The available cutset information in RSSMAP was used to establish an  !

importance measure for. subsystem failures. Although a preliminary review I of the RSSMAP cutsets was not sufficient to determine an accurate '

importance measure, a range of values of importance for one train of the onsite AC power can be estimated as ,

1 IACT = ( .11 .22 )(I)  !

Using the upper bound on the importance measure and the information i available in.the equipment failure cutsets presented in RSSMAP, the effects of_several operating configurations for GGMS can be evaluated.

The results of this evaluation are displayed as table 14.

The results on table 14 indicate  ;

5

1) that'a substantial loss in running reliability of the D/Gs can be tolerated without a significant impact on core melt frequency. i

2) that the plant being operated with one D/G out of service may be new l or above the design objective of the NUREG 0880 Safety Goals  ;

(1x10-4/R-yr),- but j

3) that the plant operated with 1 D/G out and GTGs available as a  !

backup is clearly acceptable from the standpoint of incremental  !

public risk.

i t

i l

l

-(l) Using the Vessely-Fussell measure IACT = ro o re 5-5  :

f

~

x ..

7 Table 14 IMPACTS'0F AC POWER RELD.BILITY ON CORE MELT FREQUENCY (I) l s )

i Condition '

Frequency of Core Melt (1/R-yr) ,

Base Case-Equivalent D/G perforinance, to 3.7 x 10-5 RSSMAP for GGNS._ assumed i Case where pl' ant l is operated with 1 D/G 1.9 x 10-4 out of service (2)  ;

Case where plant is operated with D/Gs 4.8 x 10-5 t exhibiting high rate of failures to run

~

Case where plant'is operated..with 1 D/G ,' 3.6 x 10-5 out of service and GTG syst'em available

^ for backup power supply ,

l(1) all data based on GGNS RSSMAP Models (2) normal performance from other D/G assumed i

3

.g.-

i e

t

.. ~

Q -d 4 I t

s f

t j

9 s

~ '

s, l

, s _.

O

! ~

.y _

n s  %

h*

! ,. 5-6 l -

x ,

$ s.

~n

s_ .

I' ,~ _ _ . . _ _ _ _ __ _ - . . _ . . . . , _ _ _ _ _ _ , _ _ . _ . _ , _ . . ._. _..