ML20205M639
| ML20205M639 | |
| Person / Time | |
|---|---|
| Site: | Maine Yankee |
| Issue date: | 03/26/1987 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML20205M614 | List: |
| References | |
| NUDOCS 8704020447 | |
| Download: ML20205M639 (13) | |
Text
. l o ueg'o UNITED STATES g
NUCLEAR REGULATORY COMMISSION WASHINGTON. D. C. 20$55 g p
%*...+/
SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION MAINE YANKEE ATOMIC POWER COMPANY MAINE YANKEE ATOMIC POWER STATION DOCKET NO. 50-309 SEISMIC MARGIN PROGRAM
1.0 INTRODUCTION AND BACKGROUND
All seismic " Class 1" structures and components of the Maine Yankee Atomic Power Station (MYNPS) which are important to nuclear safety, and could affect the health and safety of the public, are designed based on a minimum horizontal ground acceleration of 0.05g and a safe shutdown earthquake (SSE) horizontal acceleration of 0.19 (Ref. 1). However, the occurrence of two seismic events since the beginning of commercial opera-
+ tion at the plant in 1972 has caused some concerns about the adequacy ,of the seismic design basis for the plant.
An earthquake of approximately magnitude 4 occurred about 10 kilometers west of the MYNPS site on April 17, 1979. A second event was the occurrence of an earthquake of approximately magnitude 5 3/4 in central New Brunswick, Canada, on January 9,1982. This second event is of particular significance since it is the largest event known to have occurred in the New England-Piedmont tectonic province which has not been associated with a particular tectonic structure or seismogenic zone (Ref. 2). In the staff's judgment the MYNPS is in the New England-Piedmont tectonic province. Appendix A to 10 CFR Part 100 (Seismic and Geologic Siting Criteria for Nuclear Power Plants) requires that in determining the SSE in current licensing actions, the largest reported earthquake in a tectonic province which cannot reasonably be related to tectonic structure should be assumed to occur near the site.
As a result of the above factors, the staff held a meeting with Maine Yankee Atomic Power Company (licensee) on May 7,1982. At this meeting the staff expressed its concerns about the seismic hazard and seismic design adequacy of MYNPS. As a result of this meeting, the licensee undertook geological /
seismological studies and also instituted a series of voluntary reviews and walkdowns to find any potential weak areas and identify the inherent conservatismintheplantseismicdesign(Ref.3). Several cost-beneficial upgrades, associated with equipment anchorage, were identified and implemen-ted (Table 1). Based on these upgrades and the inherent design capacity of the plant, in its March 14, 1986 letter (Ref. 4), the licensee concluded that structures, systems and components at the MYNPS had sufficient strength to withstand a seismic event of at least 0.2g with a Regulatory Guide 1.60 spectrum.
In the interim, based on the review of the licensee's studies, results of the staff sponsored studies, and the staff practice in reviewing the !
design basis for older operating nuclear power plants under the Systematic Evaluation Program (SEP), the staff assessed that a spectrum obtained using the 50th percentile amplification factors in NUREG/CR-0098 (Ref. 5) and a peak acceleration of 0.189 represent an acceptable level to account for the April 17, 1979 and January 19, 1982 earthquakes. l 8704020447 870326 PDR ADOCK 05000309 P PDR
Because of this ongoing uncertainty in the seismic design basis, the licensee agreed to participate in an NRC-sponsored Seismic Design Margins Prograr (SDMP) (Ref. 6). The SDMP is a research program to develop and demonstrate a nethod for assessing safety margins at nuclear power plants with respect to seismic events. An expert panel developed such a margins assessment method (Refs. 7 and 8). This nethod was to be tested on a trial plant review. Thus, the MYNPS participation in this program addressed both research and licensing needs.
2.0 SEISMIC DESIGN MARGIN PROGRAM (SDMP)
The margin review process involves the screening of components based on their importance to safety and their seismic capacity. The products of the review are high confidence of low probability of failure (HCLPF) carecities for components, systems, accident sequences and the plant. The HCLPF is a conservative representation of capacity and in simple terms corresponds to
& the earthquake level at which it is extremely unlikely that core damage will occur. Mathematically, the HCLPF can be thought of as an estimate of the E7 failure probability point with 95% confidence. These HCLPF capacities, expressed in terms of peak ground acceleration (pga), are compared to the pga predicted for the earthquake against which the plant is to be assessed, called the review level earthquake. This review level earthquake is chosen at some level above the design basis (safe shutdown earthquake (SSE) and a favorable comparison of its pga to the HCLPF capacity of the plant indicates that there is high confidence of a low probability of failure (core damage) of the plant with respect to that review level earthquake. The extent tn which this plant HCLPF is above the pga predicted for the SSE is a measure of the seismic margin of the plant.
System analysis is used to determine those plant systems and components (including structures) t:aat are important contributors to plant seismic safety and thus allow focusing of effort on components. requiring a margin review. By studying previous seismic probabilistic risk assessments (FRA) made on pressurized water reactors (PWR's), it was found that only systems and components needed to assure reactor subcriticality and early emergency core-coolant injection (by definition this includes Auxiliary Feedwater System) needed to be considered. Therefore, only systems and components related to these two functions (with a few exceptions) and associated system interactions were considered at MYNPS.
Capacities of generic sets of components are estimated in NUREG/CR-4334 (Ref. 7) based on capacities estimated in previous PRA's, experience data gained from studying earthquake effects on industrial facilities and engineering analysis. These generic capacities are used to screen out fror further consideration those components identified as important~by systems analysis if the generic capacity is found to be higher than the review level earthquake pga.
The components remaining after the systems and fragility screenines, plus the systems interaction and plant-unique features, are then subjected to a margins quantification. Prior to this cuantification, each remaining component is thoroughly inspected and studied, and systems models are developed to describe the possible seismic-initiated accident behaviour of the plant. The quantification is accomplished by calculating the HCLPF capacities for each of these components using structural / mechanical analyses and then analyzing the minimal cut sets derived from the systems analysis using the rules of Boolean algebra and Discrete Probability Distribution methods to arrive at system, accident sequence and plant HCLPF capacities.
Component HCLPF capacities are calculated for the important components remaining after screening, using the Fragility Analysis (FA) metted or Conservative Deterministic Failure Method (CDFM). The FA method recuires
, estimating the median failure capacity.for a component, and its random and modeling uncertainties. Assuming a log normal failure probability distri-3 bution, the HCLPF (5% failure with 95% confidence) capacity can be calcula-ted. The CDFM uses a deterministic, more design oriented method of calcu-lating component HCLPF's. Random failures, test and maintenance outages, and operator actions are also included in the analysis.
The application of the above SDMP review process to the MYNPS and resulting findings are described in the next section.
- 3.0 APPLICATION OF SDMP REVIEW PROCESS TO MYNPS The seismic margin evaluation at the MYNPS was carried out in the following manner. Two review teams, a systems team and a fragility team, were choser.
A Peer Review Group (PRG) was selected and chartered to review the technical
- adequacy of this study. The objective of the PRG was to assure that the seismic margin review was executed in a fully competent and professional manner using the guidelines established by the expert panel (Refs. 7 and 8).
The seismic margin review was then organized, allowing for participation by the licensee and its representative (Yankee Atomic Electric Company), members of andthetheNRC Working Lawrence GroupNational Livermore on Seismic Design Laboratory Marg LLNL) (ins, and technical andthe appropriate NRC program managers. The LLNL is the primary contractor to the NRC for the SDMP.
The composition of the review teams, details of the study, results of the study, MYNPS related findings, general nethodology related findings, and suggested improvements for the future reviews are documented in a three volume report (Refs. 9, 10, and 11). The staff has reviewed these reports (as discussed earlier, also participated in the review process) and the PRG's comments (Appendix B of Ref. 9) on the study and concurs with the technical findings as they relate to the MYNPS HCLPF estimates. In the following safety evaluation, only the results and' findings which affect the MYNPS HCLPF estimates are discussed.
1 A flow chart of the margin review process is shown in Figure 1. As dis-cussed earlier, this process involves the screening of components based on their importance to plant safety and their seismic capacity. Inspection of Figure 1 indicates that Steps 2, 5, and 7 are prinarily concerned with plert safety functions and systems, and are performed by a team of systems analysts.
Steps 3, 6, and E are mainly concerned with capacity assessment and are i performed by a team of fragility analysts. Step 4 is performed by both teams of analysts. The process requires close cooperation and interaction between the two teams of analysts and the utility.
The selection of the review level earthquake for the MYNPS review is dis-cussed below. The details for other strps discussed above een be found in Refs. 9, 10, and 11.
3.1 REVIEW LEVEL EARTHOUAKE
+
The review earthquake level chosen was a spectral shape defined by the 50%exceedancespectrumspecifiedinNUREG/CR-0098(Ref.5)andanchored at 0.3g pga for the initial screening. The seismic margin for the compo-nents and plant is referenced to this spectrum but anchored to the pga cer-responding to the HCLPF capacity. This review level is considered sufficiently high to identify plant vulnerabilities and is also convenient to use with the screening criteria given in NUREG/CR-4334 (Ref. 7). This definition of spectra used to determine a HCLPF capacity does not in any way refer to the probability of occurrence of an earthquake. It is a spectrum used to define the HCLPF capacity that recognizes the dependency of a con-ponent capacity on the frequency content of the spectrum and not just the pga.
The results of the seismic margin study are interpreted as follows. The HCLPF capacity of the structures, equipment and plant are conditional on the actual site-specific spectrum not exceeding the target spectrum; exceedance is defined as the event when 16 percent of the spectral ordinates exceed the target spectrum over the frequency range of interest. It is assumed that the spectrum peak-to-peak and earthquake direction variabilities are removed from the hazard analysis leading to the selection of the review earthquake. The review earthquake is specified by the same spectrum in two horizontal directions and two-thirds of the horizontal spectrum in the vertical direction. It is also assumed that the review earthquake level is specified as the higher of the response spectra from the two orthogonal horizontal directions.
3.2 RESULTS OF MARGIN REVIEV For the MYNPS, twc important accident sequences were identified. Both are initiated by a seismically induced loss of offsite power (LOSP) assumed to always occur at the review earthquake level. In one sequence a small loss-uf-coolant accident (LOCA of 3/8 in. to 2 in. diameter equivalent area) is conservatively assumed to occur as a result of the review level earthquake.
HCLPF capacities for components which might cause other types of small
LOCA's (pump seal or power operated relief valve LOCA's) were sufficiently J high so they could be easily screened out. The other accident secuence :
assured no small LOCA.
The small LOCA accident sequence involved seismic failures only end resulted in a plant HCLPF of 0.219 Since the analysts involved in this review could not get inside the MYNPS containment to inspect the small primary l system piping, they chose not to screen.small LOCA out. If they had screened out this ft.ilure mode, this sequence would not be considered and the plant HCLPF would be above 0.30g.
The smell LOCA accident sequence was composed of three singleton (failure of a single component would cause plant damace state to occur) cut sets with the dominant contributor failure of the Refueling Water Storage Tank (RWST) which had a HCLPF of 0.219 Other singletons in the sequence had HCLPF's greater than 0.30g (based on proposed upgrade as discussed later).
f Failure of the RWST results in no coolant being available for reactor vessel injection following a LOCA. Thus, the plant level hCLPF is approximately 0.219 based on the RWST capacity.
The second accident sequence, LOSP with no small LOCA,- involved no singic-ton cut sets but a number of doubletons, some combining seismic and non-seismic (random, test and maintenance, human error) failures. The most important doubletons are the Demineralized Water Storage Tank (DWST)
(HCLPF=0.179) and the Circulating Water Pump House (HCLPF = 0.30g). The HCLPF capacity for the no LOCA core damage sequence is estimated as 0.339 The plant level HCLPF capacities are summarized in Table 2. This table also includes results of several sensitivity studies to evaluate the effects of assumptions made in the study. The calculations in this study were per-formed assuming perfect independence between seismic failure of different components; i.e., the seismic capacities are assumed to be statistically independent, both in randomness and uncertainty. For the small LOCA sequence, as discussed earlier, the core damage HCLPF capacity is estimated to be 0.21g (governed by the capacity of RWST.) When the Boolean expression is dominated by singletons, the assumption of perfect independence is more severe than the assumption of perfect dependence between failures if the fragilities are approximately equal; if there is a single component with a very low capacity compared to the rest of the components in the Boolean expression consisting of singletons, both the assumptions give about the same plant level HCLPF capacity (cases 1 and 2 of Table 2).
Nonseismic failures were found not to be important contributors (cases 3 and 4 of Table 2). They made no contribution to the small LOCA accident sequence and only a 3% contribution to the HCLPF for the transient (no small LOCA) sequence. The most important nonseismic failure found was a common cause failure binding (median of the Auxiliary unavailability Feedwater per demand of 1.2 x System,)
10~ .
caused by steam
As discussed earlier, the plant level HCLPF is based on the conservative assumption of occurrence of a small LOCA (with LOSF) given a seismic initi-ator. However, if one explicitly considers the conditional probability of a seismic induced small LOCA, by a split fraction p, then the two accident sequences (their Boolean expressicr.) will be combined in the following can-ner to estimate the plant level HCLPF capacity:
Core camage = p. [small LOCA] + (1-p). [No LOCA).
Case 5 in Table 2 include results of sensitivity studies performed on this split fraction by assuming different values. This study illustrates the conservatism involved and the fact that the conclusico regarding the domi-nance of RWST failure in the HCLPF capacity estimation is a function of the split fracticn assuned, i
It should be noted that during the review prccess, important components were i found for which HCLPF was either anticipated to be low or not determined.
These componer.ts were the lead-antinony station batteries, the station service transformers, a block wall near HVAC equipment, parts of the Primary 1 Com,nenent Cooling Water (PCCW) and Secondary Component Cooling Water (SCCW) air conditioning heat exchangers, and anchorage of the diesel generator day tank. These components are being replaced or upgraded and the results of this review are based en the upgraded configurations.
The above plant HCLPF estimates do not include explicit censideration of design and construction errors, aging, and relay chatter. However, with regard to the design and construction errors, it is noted that this plant has undergone a number of walk-downs and recent evaluations, and, therefore, the likelihood of the design and construction errors which may affect the HCLPF estimates is much less compared to some other situations.
The aging effects are ccosidered to the extent that the components are evaluatert in their present conditions. The relay chatter issue is currently )
under a separate NRC investigation and also a subject of an industry spon-sored study. ;
A list of upgrades accomplished and planned is included in Table 3. Since j the completion of the study, by a letter dated February 26, 1987 (Ref. 12), 1 the licensee has indicated that it will also upgrade the RWST anchorages. )
This modification is estimated to increase the HCLPF value for the tank from 0.219 to 0.27g (other than RWST modifications, as discussed earlier, were assumed in the review). This will, in turn, also increase the plant's HCLPF in the range of 0.279 . t 4.0 EVALUATION Based on the review of the detailed reports, above discossed results, and the active participation by the NRC staff during the review process (the staff attended most of the working meetings among review teams and the
PRG, and also performed limited walk-downs), it is concluded that the' sef snic margin methodology, as suggested by the expert panel, was properly applied in the MYNPS review. This view is strongly supported by the PPG evaluation of the effort. As noted earlier, the PRG had a specific charter to assure that the trial review was executed in a fully competent manner usinc state-of-the-art method. The PRG reported the following finding in its final report regarding the trial plant review (Appendix B of Ref 9):
(1) The review teams (collectively identified as the project tean) followed the guidance established in NUREG/CR-4334 and NUREG/CR-4d82 as fully as could be expected; (2) The. project team executed the study in a fully competent and professional manner; (3) The project team used state-of-the-art methods, and in few cases advances were rade in the state-of-the-art; and (4) The project team seems to have taken cognizance of all relevant information.
These comments provide assurance regarding the adequacy of the study. The PRG comments are also supplemented by comments from some of the individual PRG members. These comments provide further confidence in results of the MYNPS review.
The study, as intended, identified few areas (Table 3) where the seismic ruggedness of the plant could be effectively enhanced. The pre-tank modification HCLPF of the plant, 0.21g, is conservatively calculated (0.219 represents high confidence, low probability of failure (95-5) capacity provided an earthquake with this acceleration level does not exceed the specified spectrum shape for more than 167 of the spectral frequencies).
This capacity is well beyond the design level earthquakes of 0.19 , and it also exceeds the earthquake defined by a spectrum obtained usina the 50th percentile amplification factors in NUREG/CR-0098 (Ref. 5) and a peak acceleration of 0.18g (the staff assessment of the earthquake level accounting for April 17, 1979 and January 19, 1982 events). Of course, with the planned upgrade of the RWST anchorages, the HCLPF is estimated to be about 0.279 - well beyond the staff-assessed earthquake level. It ,
must be emphasized that the HCLPF is a conservative representation of the I capacity and the median capacity is at least a factor of 2 greater than the HCLPF capacity and thus no proverbial " cliff" or sudden failure is expected to occur immediately upon exceeding this capacity.
5.0 CONCLUSION
Based on the above evaluation and the licensee's committment to upgrade items identified in Table 3, it is concluded that the upgraded MYNPS will have the HCLPF capacity in the range of 0.27 9 . This capacity is
6-significantly higher than the earthquake event defined by NUREG/CR-0098 (Ref. 5) 50% spectrum anchored to a peak acceleration of 0.18g. Even without the tank modification, the plant HCLFF is estimated to be 0.21g; higher than the staff assessment of the appropr.iate earthquake level at the MYFPS.- Therefore, all the issues associated with the design basis (becapre of the occurrence of April 17, 1979 and January 9,1982 earthquakes) for the MYNPS and hence the seismic design adequacy of the plant are considered resolved.
F 4
l t I f
(
i
-r- - , , , e < - - . . -. - -, .,
REFERENCES 1.0 Maire Yankee Atomic Power Company, Final Safety Analysis Report (FSAR), Section 2.5, Seismology.
2.0 Letter from Frank J. Miraglia, NRC, to J. B. Randazza, Maine Yankee Atomic Pcwer Company, " Maine Yankee Participation in Seismic Design Fargins Program," March 31, 1986.
3.0 Appendix C of Peference 9.0 4.0 Letter from G. D. Whittier, Maine Yankee Atomic Power Company, to Director of Nuclear Reactor Regulation, NRC, " Seismic Assessment Program" March 14, 1986. ,
5.0 Newmark, N.M. and W. J. Hall, " Development of Criteria for Review of Selected Nuclear Power Plants, "NUREG/CR-0098 (May 1978).
6.0 Letter from G. D. Whittier, Maine Yankee Atomic Power Company, to Director of Nuclear Peactor Regulation, NRC, " Maine Yankee Seismic Review Program," March 4, 1986.
7.0 Budnitz, R. J., P. J. Amico, C. A. Cornell, W. J. Hall, R. P. Kennedy, J. W. Reed, and M. Shinczuka, "An Approach to the Quantification of Seismic Margins in Nuclear Power Plants," NUREG/CR-4334, 0C1D-20444 (August 1985).
8.0 Prassir.os, P. G., M. K. Ravindra, and J. B. Savy, " Recommendations to the Nuclear Regulatory Comission on Trial Guidelines for Seismic Margin Reviews of Nuclear Power Plants," NUPEG/CR-4482, UCID-20579 (March 1986).
9.0 Prassinos, P.G., R.C. Murray, and G.E. Cummings. " Seismic Margin Review of the Maine Yankee Atomic Power Station - Volume 1.
Summary Report," NUREG/CR-4826, UCID-20948, Vol. 1 (March 1987) 10.0 Moore, D.L., D.M. Jones, M.D. Quilici, and J. Young, " Seismic Margin Review of the Maine Yankee Atomic Power Station - Volume 2. Systems Analysis," NUREG/CR-4826, UCID-20948, Vol. 2 (March 1987) 11.0 Ravindra, M.K., G.S. Hardy, P.S. Hashimoto, and M.J. Griffin, " Seismic Margin Review of the Maine Yankee Atomic Power Station - Volume 3.
Fragility Analysis," NUREG/CR-4826, UCID-20948, Vol. 3 (March 1987).
12.0 Letter from G. D. Whittier, Maine Yankee Atomic Pcwer Company, to Director of Nuclear Reactor Regulation, NRC, " Seismic Margins Program,"
February 26, 1967.
TABLE 1 Maine Yonkee Eculte nt An chc r.u >_ ar.c S u u:.o r t Uncrades Emergency Buses 5, 6, 7, and S Battery Chargers 1, 2, 3, and 4 Battery Inverters 1, 2, 3, 4, and 5 Floor Mounted Diesel Generator Control Panels Main Control Board Control Room Auxiliary Cabinets Electrical Control Board Radiation Monitoring System 4 Heat Tracing Cabinet Air Conditioning Control Panel EHC Panel Reactor Regulating System Feedwater Regulating System Vibration and Loose Parts Monitoring Cabinet Reactor Protective System Core Loading Panel Radiation Monitoring System Meteorological Survey Cabinet Safety-Related Instrument Racks Masonry Wall Reinforcement Service Water Piping Support 1
i l
1 I
l l
l
Start Time axis N
1 Select an earthquake rev:ew levci
- Ga ther information on systems and sort . . * . :* Gather information on Group A function:;. #*
~
- the plant. Determint which
.. 2 . : l.* Interaction
- broad classes or groups of
. . . f ....-
- ** ** + - - - - - + , 3 components have HCLPF
"*/*.**- / values greater than the rewew
/>
level. Possibly identif y plant-unique features.
First plant walkdown: o Concentrate on identification of problems.
'. . 4 :. Emphasize systems interaction. Confirm
..,,-
- apphcabihty of screening tools. Complete identification of plant-unique features.
i K EY: Task is performed by:
- o V ***
. : f. *.-l:['*' Revision of systems
- 1.l.* . Systems Analyst relationships established h, *. .*. *.1
, .
- f, *, *,
in ,Step
- f' 2. Develop fault
- , .* /. , *.l,y*, .*y, trees and event trees.
Fragility Analyst u
Both Second plant walkdown:
[ Primarily fragility analyst for checks.
6
~ Collect specific data (size and other
/
/ physical characteristics) of components requiring detailed analysis o
- I.. ;*/J :.. /j Determine minimal * !*I*
~* '*
- ~*
cut sets for end- -
- ~
.-l '
Finalize HCLPF value for components in final cut sets
.[* *.7...-
point core melt. *
- 8
. , .l,-l ... (components not screened out).
,l., : :. . .' .: :'. : .
/>-
lE Margin assessment Complete Figure 1 Graphic representation of the screening operations.
TAELE 2
SUMMARY
OF PLANT LEVEL HCLPF CAPACITIES Case Description HCLP Capacity (g) 1 Small LOCA 0.21
- Independent Seismic Failures 2 Small LOCA 0.21
- Dependent Seismic. Failures s 3 No LOCA 0.32
- Independent Seismic Failures with Norssismic Failures 4 No LOCA 0.33
- Independent Seismic Failures without Nonseismic Failures 5 Core D3,7,3ge
- Split Fraction p=0.01 0.32 p=0.10 0.28 p=0.50 0.23 i
1
.j l
. i 1
TABLE 3 - MODIFICATION VADE AND PLANNED AS A PESULT OF MARGIN REVIEV Descripton Schedule Completion Change Date o Diesel fuel day tank anchorage upgrade done o Control Roer cooler. anchorage upgrade done
- e Welding cart / gas bottle tiedown- done o Security lighting tiedown done 4
o Main control board alarn tiedown done F
~
o Strengthen blockwall VE 21-1 1987 Outage o Upgrade anchors for fans 44A & B 1987 Outage o Install internal anchors for transformers 507 & 608 1987 Outage o Replace safety class batteries I & 3 1987 Outage o Replace safety class batteries 2 & 4 1988 Outage 1
- o Refueling water storage tank anchorage 1988 Outage upgrade
- Note: This modification was not included in the estimated HCLPF of 0.219 With this modification, the plant HCLPF is estimated to be in the range of 0.279 1
_ _ - . . _