ML20126D299
| ML20126D299 | |
| Person / Time | |
|---|---|
| Site: | 05000447 |
| Issue date: | 09/21/1983 |
| From: | GENERAL ELECTRIC CO. |
| To: | |
| Shared Package | |
| ML20126D269 | List: |
| References | |
| FOIA-84-175, FOIA-84-A-175, FOIA-84-A-66 NUDOCS 8506150015 | |
| Download: ML20126D299 (20) | |
Text
. - . . - _ . _ , -,
- r ,s-9/21/83 1.etter to D. G. Eisenhut from J. F. Quirk re: GESSAR-II Seismic Event Analysis (152 pages) 8506150015 PDR FOIA 850215 CURRAN 84-A-66 PDR
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION
2.1 INTRODUCTION
In a seismic hazard analysis, the focus is to obtain a complementary cumulative distribution function (CC0F) defining the annual probability of occurrence of an earthquake with magnitude greater than a specified ground motion parameter (e.g., peak ground acceleration, effective peak ground acceleration, sustained peak ground acceleration, etc.).
This CCDF is also known alternatively as the seismic hazard y a
/
For a specific site, this CCDF can be developed through a systematic evaluation of:
a) Source zonation; b) Frequency-magnitude and/or frequency-intensity relationship; lf c) Ground motion attenuation relationship; #
da d) Saturation values of magnitude and ground motion; and e) Near-field effects.
Upon suitable manipulation ( ) , this information is cast into mathematical form and then synthesized by~means of the following equation:
P[A>a] = ffP[A>a/m,r]fMI") R(r)dmdr (1.0) where P[A>a] is the total probability of the ground motion parameter A exceeding a given value a, andMf (m) andR f (r),are the probability j distributions of the magnitude and the distance from source-to-site respectivel f This probability, P[A>a], when evaluated over a series 9
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION 2.2 TECHNICAL BASIS To develop a realistic, upper-bound seismic hazard curve that will
- envelope a majority of the potential nuclear plant sites in the United States, at a consistent probability level, two separate, ideal approaches can be utilized. The first is known as the " Demand" approach and the second, the " Capacity" approach. In the Demand approach, the bounding seismic hazard curve is obtained by repeating the deterministic or Baysian process outlined in Section 2.1 for a number of strategic site locations to generate the necessary site-specific seismic hazard curves. The envelope of these site-specific curves defines the bounding curve to be used. In this manner, a family of bounding curves, at various probability levels can also be established using Baysian techniques. In the " Capacity" approach, the seismic hazard demand is increased incrementally until the system capacity is exceeded. In this case, the bounding seismic hazard curve is represented by the most limiting seismic hazard demand placed upon the system and is definitely site-independent.
In other words, no matter where the nuclear plant site is located, so long as it is constructed according to the GESSAR requirements, the seismic hazard capability, represented here as a seismic hazard curve, is unique.
However, neither of these approaches is considered to be practical in view of the general lack of quality ear _thquake data and quality structural and component fragility data.){Therefore, for the evalua- h
' 33 g' tion of seismic impact on the GESSAR II PRA, a pseudo-demand4a
- app 12
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION was devised. To utilize this approach, the existing seismic hazard (
analyses available in the open literature (7 12) , were carefully reviewed, consolidated, and extrapolated to obtain a realistic, median-centered upper-bound seismic hazard curve. The median-centered, upper-bound seismic hazard curve developed in this manner relies heavily on existing data and partially on subjective judgment.
Realistically, for a majority of the potential GESSAR sites, this upper tound seismic hazard curve will bound the site-specific curve fnotonlyatthe50%probabiiitylevelbutalsoathigherprobability levels. This is desirable and is consistent with the standard plant design concept.(13) -
32.
om 13
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION relationship used for converting peak ground acceleration (PGA) values to EPGA is of the form PGA = (K)(EPGA) where the constant K ranged from 1.23 to 1.25. The GESSAR II seismic hazard curve is defined in terms of EPGA.
2.3.2 ANNUAL FREQUENCY OR EXCEEDANCE AT HIGH g LEVEL It was also noted that for the hazard analyses utilizing EPGA, upper bound cutoff values were used for a given earthquake source and intensity.
This is generally understood to be an attempt to correlate the historical earthquake data in Modified Mercalli Intensity (MMI) to EPGA. In the information presented in reference 10, the upper bound EPGA is conserva-tively estimated to be no more than 0.8g for MMI scale of IX. Other PRAs have followed the concept of defining an upper bound cutoff.
The GESSAR II seismic hazard curve was truncated approximately 20% beyond ,
=a
. 0.8g at a valueaof 0.95g. The annual frequency of exceedance at 0.95g 4 9-.
l l l.
was specified at a value of 10.s, and a smooth curve was arbitrarily l
/
drawn from an EPGA of 0.8 to 0.95. j/
15
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION 2.3.3 ANNUAL' FREQUENCY OF EXCEEDANCE, 0.lg-0.8g i t In order to define the GESSAR II hazard curve for EPGAs of 0.19 to 0.8g, '
L L9t
,1 plant specific best estimate hazard curves were plotted and an arbitrary t. M
!l 33 bounding curve was constructed.
4.s '
Figure 1-1 is the GESSAR II seismic 4a l hazard curve defined by this approach and the assumption in 2.3.2.
f Figure 1-2 provides a comparison of the GESSAR II curve with best estimate ,
hazard curves from other plaid, specific studies.
~ _. . - - - - .. .
2.3.4 COMPARIS0N WITH USGS DATA Reference 11 provides recent national seismic hazard information. Even though.there may be disagreements with information provided in Reference 11, it provides an information base to compare to a generic hazard curve.
The seismic hazard information in Reference 11 was presented as a set of three isoacceleration zone maps. Using the potential GESSAR sites identified in Reference 13, raximum ground accelerations were identified for each of the time isoacceleration maps representing 10, 50 and 250 ;
year return periods. The return periods were translated into an annual frequency of exceedance using the approach described in Appendix A. Thus l
for annual frequencies of exceedance greater than 4.2x10 4, the seismic 1
hazard curve in Figure 1-1 is expected to bound more than 80% of the I l
potential GESSAR sites identified in Reference 13.
I l
16 i
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION x Figure 2-1 10 GESSAR II SEISHIC HAZARD CURVE 10~4 -
f 10-5 S
U U
b U
5 Sd 5 3d y10~6 -
4d
~d i e \
T- t l
t i
10-7 -
10 l 0
I I 0.1 0.2 I I I I i 0' 3 0.4 0.5 0.6 0.7 0.8 0.9 i i 1.0
/
EFFECTIVE PEAK GROUND ACCELERATION (g) 19
~ ~
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION i
10-3 _ g\
LIMERICK --- -
\ IllDIAtl POINT -
\ \
ZION --
\s g OYSTER CREEK
\ Y 10~4 -
s
\
s g -
' i s
% \
\\ s
\'
\
\
s GESSAR
] 3ct I Id 10-5 _ s-d- s 5 E \ \
S OYSTER ' N d CREEK
\
.b s
\ s w LIMERICK b \ -
W 10'6 ._
\
w ZION s s
s\
g ^
i INDIAN E POINT .
E 4
1C' -
. I i
/
/
10-8 t I I I I I I I l l 1 f' O 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 EFFECTIVE PEAK GROUND ACCELEPATIOil(g)
Figure 2-2: GESSAR II Seismic Hazard Curve Compared to Hazard Curves from other Studies.
20
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION f
I 4.0 EVALUATION OF SEISMIC IMPACT ON CORE DAMAGE FREQUENCY ; .
This section describes the methodology used to estimate the annual core damage frequency for the GESSAR II plant due to seismically-induced accident initiators. The general procedure was to integrate the seismic hazard analysis (section 2.0) and the plant structure and component fragility data (sections 3 and 4) and assess the plant system response.
4,1 Introduction
[ M vent tree / fault-tree analysis was performed to identify the possible ways in which an earthquake could initiate severe accident sequences.
The plant system responses were obtained in terms of boolean expressions.
The plant structure and component fragility data were used to quantify the plant system failure probabilities which were combined with the seismic hazard analysis to obtain earthquake-induced core damage frequency.
v 84
GENERAL ELECTRIC COMPANY i PROPRIETARY INFORMATION l 4.2 Accident Sequence Analysis l
The event trees that were developed for t'he BWR/6 Standard Plant internal ;
event PRA were studied in order to construct seismically-induced accident
)
sequence event trees. The plant structure and component fragility data in Table 4-1 was used to guide the process for developing the potential
~ 1 accident sequences that could result from seismic events. .
J cs
)
l i . l l
l l
l l
f U
85
GENERAL ELECTRIC CDMPANY PROPRIETARY INFORMATION s.
e l
The fault trees in the BWR/6 Standard Plant were modified to include both seismic and non-seismic failures. Each individual fault tree was further reduced by eliminating those seismically-induced failures that would be negligible contributors to the total core damage frequency. Figures 4-6 through 4-13 describe the fault tree for accident mitigating systems.
The a and 0 shapes depict the random independent failures and seismically-induced failures respectively. The data used for the random independent failures of safety related systems was derived from the BWR/6 Standard Plant PRA and is listed in Table 4-3. Seismically-induced failure probabilities of components were calculated from the component and 86
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION
- i. structural fragility curves. Median capacities and coefficients of variations corresponding to the lognormal fragility curves are listed in Tabli 4-4.
The minimal cut-sets for accident mitigating systems were derived from the fault trees in Figures 4-6 through 4-13 and are listed in Table 4-5.
4 The booleon expressions for dependent systems were derived from their individual cut-sets and simplified by using a boolean reduction technique.
The boolean expressions for accident sequences which were obtained from the event trees in Figures 4-1 through 4-5 are listed in Table 4-6.
These expressions were further reduced into minimal cut-set form before being converted.into a probabilistic expression. I i
I 4
g m
9 f
E w
4 L, , _ _ _
' ^
87
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION l 1
5.0 EVALUATION OF SEISMIC IMPACT ON OFFSITE CONSEQUENCES This section describes the methodology used to assess the fission product transport and release to the environment for the seismic initiated events. Offsite consequences are evaluated and the results are presented in this section.
5.1- Introduction This section describes the potential fission product transport pathways for the seismic initiated core damage sequences determined in Section 4.0. The various release pathways are combined with accident sequence characteristics such as timing and flow rates resulting in accident specific release sequences. These release sequences are further grouped into the appropriate release category which is input to the consequence analysis code to determing, potential health effects. Results of the offsite consequence evaluation are also presented in terms of a risk curve and mean risk value for internally and externally initiated core damage events.
5.2 Release Frequency Analysis '
5.2.1 Release Pathways The pathways for release of radionuclides following seismic events have -
been evaluated with a methodology similar to that employed in the internal events PRA (GESSAR 150.3). The potential pathways for release are !,
t 109
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION -
identified and probablistically assigned to release categories as evaluated by the CORRAL /CRAC codes (GESSAR Section 15D.3, Appendix F). .For the Seismic PRA assessment the same release models are used for assessment of offsite consequences, but the Release Category frequencies are modified to reflect the release frequencies corresponding to seismic events rather than random failures.
The potential release pathways have been identified from a review of the Containment and Drywell penetrations in the 238 Nuclear Island design.
These pathways were grouped as to release location. The release locations were:
Drywell Containment Secondary Containment Turbine Building (or other areas outside secondary containment)
In each area the lines which entered the area were reviewed to identify the barriers to fission product transport. These barriers included the number and types of isolation valves, the line size for estimation of flow splits with the suppression pool, the degree of ' aerosol plugging which may occur in the smaller lines, and the location of any high pressure to low pressure interfaces. Emphasis was placed in this eeview on potential pathways which could bypass the suppression pool and thereby not receive the benefit of fission product scrubbing by the suppression pool. The results of the investigation were expressed in terms of an
- 110 ,, /
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION
\
Iodine / Particulate release fraction which is the product of the release pathway probability and the flow split fraction. Table 5.1 is a summary ,
of the results of this evaluation based on random failure probabilities.
5.2.2 Flow Splits Many accident sequences have mass flow rates from various ' compartments which are split among two or more outlets. For example, during a small-break LOCA, the mass ficw from the RPV is split between the linebreak and the safety-relief valves. Additionally, flows from the drywell airspace can be split between the horizontal vents which release into the suppression pool and a potential path through the drywell structure directly into the containment airspace. Generally, for the purposes of this risk analysis, the term flow split is taken to mean the fraction of flow directed to the suppression pool versus that which bypasses the pool.
The flow splits were estimated based on the limiting flow conditions for compressible fluids in pipes or orifices. The Darcy's formulael were used with the appropriate gas densities, pressures, and temperatures as calculated by the MARCH code for various events considered in the internal events PRA.
The degree of aerosol plugging in potential release pathways was also considered. The methods described by Morewitz and summarized in NUREG-0772 were applied to the smaller size release pathways using the total mass of fission product aerosols released as calculated by the fuel release model in NUREG-0772, Appendix 8. The ratio of aerosol release prior to plugging 111
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION l
over the total aerosol generated represents a fractional release limited
. by the aerosol plugging effect.
The larger calculated release fraction determined by either the flow split or plugging effects was used in the evaluation of the potential release pathway and is included in the values shown in Table 5-1. The .
I*
remaining effluents from the reactor pressure vessel or drywell which were restricted from taking the bypass pathway are directed to the ,
suppression pool. '
- 5. 2. 3 Failure Rates Failure rates for piping, valves, and control systems in each of the potential release pathways following seismic events were based on the fragility curves for structures and components. The basis for these fragility curves is described in section 3.0. The fragility curves which were used as a basis for failures in the release locations (identified in section 5.2.1) are shown in Table 5-2.
1 Flow of Fluids through valves, fittings, and pipe, Crane Company Technical Paper No. 410, 1972.
112
' ~ GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION i
The probabilities of failure at various values of ground acceleration
[
were determined from the mean values and coefficient of variation sum-marized on Table 4-1. These fragility curves are shown in Figure 5-1.
Iodine and particulate release fractions were determined for seismic levels of .4g, .6 9, .8g and .959. In this evaluation failure at any of the fragility values was conservatively taken to indicate failure to accomplish the action to prevent release of fission products from the plant. In the case of valve operation there may be situatio'ris where the failure mode which leads to core damage (valve failure to open) and that which allows release from the plant (valve fails to close) may be incon-sistent and thus a deterministic approach has not been strictly followed.
5.2.4 Release Categories The assignment of the frequency of core damage following seismic events to various release pathways was accomplished through the development of release pathway event trees. These trees (Figures 5-2 through 5-6) show the assignment of core damage frequencies and the releases through the potential _ pathways. Conditional fractions (discussed below) corresponding to particular release pathways are also shown. Table 5-3 summarizes the consolidation of the various core damage sequences determined in Section 4.0 into release pathway event trees. The release frequencies are assigned to release categories used in the evaluatiori of the offsite consequences, as was done in the internal events PRA (GESSAR 15D.3).
113 l
GENERAL ELECTRIC COMPANY j PROPRIETARY INFORMATION '
The individual event trees represent different combinations of seismic core damage accident sequences, as noted on the figures, which have similar release characteristics. Since the loss of offsite power is the most likely type of core damage initiating event following a seismic event, the models developed for the internal events PRA are used as a basis for assigning release pathways to this accident sequence.
In Figure 5-2 accident sequences which do not take one of the pathways which bypass the suppression pool are routed to release sequence 13 (see GESSAR section 150.3, Appendix C for definitions of release sequence nomenclature). Pathways which provide a potential release from the reactor pressure vessel which does not pass through the suppression pool are routed to release sequence El since the seismic event can be presumed to develop these paths prior to the onset of core damage. Pathways which pass through the drywell structure are routed to sequence 12 to account for their contribution after reactor pressure vessel meltthrough.
Potential releases through the drywell following breaks in the drywell concurrent with the seismic event are not shown independently since their contribution would be very small in comparison with other early release frequencies. These low frequencies are combined into the El release sequence frequency. Figures 5-3 and 5-4 are variations on the loss of offsite power event which account for particular common mode failures (e.g., building failures) such that the probability of bypass pathways are higher than those given in the event tree in Figure 5-2.
114
' GENERAL ELECTRIC' COMPANY !
PROPRIETARY INFORMATION '
i
, Figure' 5-5 presents an event tree for the loss of containment heat removal following a seismic event. Pathways which do not bypass the suppression pool are routed to release sequence 83. Pathways from the reactor pressure vessel which bypass.the pool are routed to release sequence L1 since core damage for this type of event would occur after a considerable period of time (con,tainment failure is due to overpressuri-zation rather than hydrogen burning because of steam inerting). Similarly the pathways which pass through the drywell structure are routed to sequence 12.
Figure 5-6 represents the release pathways for seismic initiated core damage events including a failure to scram the reactor. Pathways which do not bypass the suppression pool are routed to release sequence F3.
1 Other pathways are routed similarly to the loss of offsite power case to release sequences 12 or El.
5.2.5 Conditional Release Fractions The conditional release fractions shown on Figures 5-2 through 5-6 are developed from the Iodine / particulate release fractions discussed above.
An average release fraction (weighted by the hazard frequency in a manner similar to equationiin Section 4.0 was determined for each of the
. potential release areas. For the particular case (Figures 5-3 and 5-4) ,
where common fragility values were used in both the core damage frequency evaluation and the release sequence evaluation, the average release fraction was weighted by the hazard frequency and the common fragility value to avoid double-counting in the conditional release fraction.
115
GENERAL ELECTRIC COMPANY PROPRIETARY INFORMATION Because the same release category models were used for evaluation of offsite consequences (except for frequency) in both this assessment and the internal events PRA, the flow splits included in the CORRAL models for the internal events study were taken into account in conditional release fractions. In this way, the offsite consequence evaluation properly accounts for the flow splits determined by this assessment.
5.3 Release Category Frequency Summary Table 5-4 summarizes the total release frequency for each of the release sequences shown in Figures 5-2 through 5-6. The release categories developed for the internal events and their corresponding frequencies are also shown in Table 5-4.
-1.
In comparison with the release category frequencies determined in the internal events study, the total frequencies are limited for seismic events due primarily because the hazard frequency (Figure 2-1) de reases faster than the corresponding increase in failure rates in valve isolatior and piping from larger ground acceleration (Figure 5-1). More detailed modelling of bypass pathways in this study as compared with the internal events PRA also contributes to lower release frequencies. The loss of containment heat removal sequences (B3) are higher for the seismic events than for the internal events because of the conservative assumption of inability to restore ac power or to repair RHR system components.
~
116 -
__ __ __ _____ __,_