Available online at www.sciencedirect.

com

Nuclear Engineering and Design 238 (2008) 1410–1420

Improvement of the seismic safety of existing nuclear power plants by

an increase of the component seismic capacity: A case study
Young-Sun Choun ∗ , In-Kil Choi, Jeong-Moon Seo
Integrated Risk Assessment Center, Korea Atomic Energy Research Institute, 1045 Daedeok-daero, Yuseong-gu,
Daejeon 305-353, Republic of Korea
Received 1 September 2006; received in revised form 15 October 2007; accepted 31 October 2007

Abstract
This case study produces the scenario earthquakes for an example nuclear power plant (NPP) site and suggests the effective seismic capacity
of safety-related equipment and components which significantly contribute to a core damage to improve the seismic safety of an existing NPP by
using a probabilistic safety assessment. The response spectra for the scenario earthquakes show greater spectral accelerations than those for the
design response spectrum in the frequency range higher than about 12 Hz. In order to improve the seismic safety of an example NPP, the effects
of the seismic capacity of safety-related equipment and components on the core damage frequency (CDF) are investigated, and their effective
seismic capacities are determined. The results of the case study show that an increase of the seismic capacity of the equipment reduces the CDF
considerably. The effective seismic capacities for the diesel generator, offsite power, condensate storage tank and battery rack are determined as
0.84, 0.35, 0.63 and 0.63 g, respectively.
© 2007 Elsevier B.V. All rights reserved.

1. Introduction the seismic safety of the NPPs near the faults must be reevalu-
ated by using the modified design spectra. Then, if necessary,
The safety-related structures, systems and components some modification may be added to the vulnerable SSCs to
(SSCs) in a nuclear power plant (NPP) which are designed to be increase their seismic resistance capacities and to ensure the
safe for a design basis earthquake may be damaged or failed by seismic safety of the plants. Actually, since a NPP consists of
strong ground motions greater than a design basis earthquake numerous systems and components, the selection of the SSCs
as well as a particular earthquake of which the frequency con- important to the seismic safety of a plant is not easy. Therefore,
tents are different from those of a design input motion. Due to more efficient procedures are necessary to evaluate and improve
these uncertainties from a magnitude and frequency contents of the seismic safety of a plant.
the earthquake ground motions, it is necessary to maintain the Probabilistic safety assessment (PSA) can be effectively used
seismic margin of the safety facilities high enough to ensure the for evaluating the seismic safety of a NPP. The use of a seismic
seismic safety of a plant against probable strong-earthquakes PSA cannot only provide the core damage frequency (CDF)
at a plant site. When new faults are found near a plant site, the associated with earthquakes, but also identify the dominant seis-
design basis earthquake should be determined by considering the mic risk contributors and the range of a peak ground acceleration
new faults and then the seismic safety of the safety-related SSCs that contributes significantly to a plant risk as summarized in
should be reevaluated under the revised design basis earthquake. the IAEA Technical Document, IAEA-TECDOC-724 (1993),
Recently, in Korea, Quaternary fault segments were found ‘Probabilistic Safety Assessment for Seismic Events’. In the
near NPP sites and a geological survey was performed to identify United States, the NRC issued Generic Letter 88-20, ‘Individ-
whether they were active faults or not. When they are identified ual Plant Examination for Severe Accident Vulnerabilities, 10
as active faults, the seismic design spectra must be modified and CFR 50.54(f)’ (USNRC, 1988) to conduct an individual plant
examination (IPE) of a severe accident risk for internally ini-
tiated events. Subsequently, the NRC issued Supplement 4 to
∗ Corresponding author. Tel.: +82 42 868 2036; fax: +82 42 868 8256. Generic Letter 88-20, ‘Individual Plant Examination of Exter-
E-mail address: sunchun@kaeri.re.kr (Y.-S. Choun). nal Events (IPEEE) for Severe Accident Vulnerabilities, 10 CFR

0029-5493/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2007.10.008
 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420 1411

50.54(f)’ (USNRC, 1991) and NUREG-1407 (1991), ‘Procedu- exceedance and the ground motion intensity such as a peak
ral and Submittal Guidance for the Individual Plant Examination ground acceleration, velocity or spectral acceleration. Since a
of External Events (IPEEE) for Severe Accident Vulnerabilities: probabilistic seismic hazard analysis (PSHA) can determine the
Final Report,’ to perform a complementary assessment to iden- annual probability of exceedance for the intensity parameters
tify plant-specific vulnerabilities under severe accidents caused of a ground motion, it is very useful to define the scenario
by external events. Finally, the NRC issued Supplement 5 to earthquakes (Ishikawa and Kameda, 1993).
Generic Letter 88-20, ‘Individual Plant Examination of Exter- The concept of the probability based scenario earthquake
nal Events for Severe Accident Vulnerabilities’ (USNRC, 1995) originates from McGuire (1995). The probability based scenario
to provide guidance on modifications in the scope of a seismic earthquakes, represented by particular sets of an earthquake
IPEEE for certain plants. One of the IPEEE objectives is to source magnitude and the distance for a specified probabil-
reduce the overall likelihood of a core damage and radioactive ity level, can be obtained from a de-aggregation of the PSHA
material releases by modifying, where appropriate, hardware results. Currently, two typical methods are usually used for defin-
and procedures that would help prevent or mitigate severe acci- ing the probability based scenario earthquakes at a NPP site.
dents. The first one was developed by USNRC. The scenario earth-
This paper investigated the effect of the seismic capacity of quakes are called controlling earthquakes in the Regulatory
safety-related components on the seismic risk of a NPP and sug- Guide 1.60 (1973). The second one was developed by Japan
gested their effective seismic capacities by using a seismic PSA. Atomic Energy Research Institute (Hirose et al., 2002) base on
A case study produced the scenario earthquakes for a nuclear a study by Ishikawa and Kameda (1993). Takada et al. (1999)
plant site and the corresponding response spectra by considering adopted these two procedures to define the scenario earthquakes
the potential active-fault effect, and then conducted a compari- for an example site and showed that both methods generated
son with the current design response spectra. For an evaluation similar results. However, the authors noted a methodological
of the seismic safety of the nuclear plant, this study selected the difference between them. The USNRC’s procedure cannot iden-
components important to the CDF from the results of a seismic tify an earthquake source location for the scenario earthquakes.
PSA, and then performed a seismic PSA by using different seis- This procedure uses coarser bins for an earthquake magnitude
mic capacities of the selected components. Finally, this study and distance, thus all the results including the earthquake source
suggested the effective seismic capacities of the selected com- regions and faults are mixed up in the bins. On the other hand,
ponents for improving the seismic safety of the nuclear plant. Ishikawa’s procedure has the advantage that it produces source
contribution factors that are an effective indicator for identifying
2. Evaluation of the design response spectra which earthquake sources and/or which fault(s) are most influ-
ential, and one or more scenario earthquakes can be determined.
Most of the NPPs in Korea have been designed by using Finally, the authors concluded that the USNRC’s procedure
the design response spectra proposed in the USNRC Regulatory gives a global view of the scenario earthquakes, while Ishikawa’s
Guide 1.60 (1973) instead of the site-specific design response procedure provides a more precise view of the scenario earth-
spectra due to the lack of strong ground motion records. quakes.
However, recently, since many earthquake records have been In order to develop the scenario earthquakes for a Korean
accumulated and several new faults have been found near NPP NPP site, this study used the USNRC Regulatory Guide 1.165
sites, it is necessary to evaluate whether the use of the NRC procedure because it was the original approach adopted for deter-
design response spectra as their design bases could ensure mining the controlling earthquakes at the site. The procedure
an adequate level of a conservatism for the safety of these is based on a de-aggregation of a probabilistic seismic hazard
plants. This chapter proposes the scenario earthquakes and their in terms of an earthquake magnitude and distance. A simple
response spectra at one of the plant sites and compares them description of the procedure is as follows (USNRC Regulatory
with the NRC design response spectra. Guide 1.165, 1997):

2.1. Procedures for determining the scenario earthquakes Step 1: Perform a site-specific PSHA. The hazard assessment
(mean, median, 85th percentile and 15th percentile) should be
To define the scenario earthquakes at a site, in general, performed for spectral accelerations at 1, 2.5, 5, 10 and 25 Hz,
the deterministic and the probabilistic seismic hazard analysis and the peak ground acceleration. A lower-bound magnitude
approaches are used. The deterministic methodology estimates of 5.0 is recommended.
the strong motion parameters for the maximum possible earth- Step 2: Using the reference probability, determine the ground
quake assumed to occur at the closest area around a site, while the motion levels for the spectral accelerations at 1, 2.5, 5 and
probabilistic methodology integrates the effect of all the earth- 10 Hz from the total median hazard obtained in Step 1, and
quakes expected to occur at different locations around a site calculate the average of the ground motion level for the 1 and
during a specified life period. The seismic hazard analysis aims 2.5 Hz and the 5 and 10 Hz spectral acceleration pairs.
at evaluating the annual probability of exceedance of various Step 3: Perform a complete probabilistic seismic hazard anal-
earthquake sizes at a selected site. The seismic hazard of a NPP ysis for each of the magnitude–distance bins.
site is usually represented by a series of seismic hazard curves, Step 4: From the de-aggregated results of Step 3, the median
which shows a relationship between the annual probability of annual probability of exceeding the ground motion levels of
1412 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420

Step 2 (spectral accelerations at 1, 2.5, 5 and 10 Hz) are deter-

mined for each magnitude–distance bin. Using the median
annual probability, the fractional contribution of each mag-
nitude and distance bin to the total hazard for the average of 1
and 2.5 Hz and 5 and 10 Hz, are computed, respectively.
Step 5: Review the magnitude–distance distribution for the
average of 1 and 2.5 Hz to determine whether the contribution
to the hazard for distances of 100 km or greater is substan-
tial. If the contribution to the hazard for distances of 100 km or
greater exceeds 5%, additional calculations are needed to deter-
mine the controlling earthquakes using the magnitude–distance
distribution for distances greater than 100 km.
Step 6: Calculate the mean magnitude and distance of the
controlling earthquake associated with the ground motions
determined in Step 2 for the average of 5 and 10 Hz.
Step 7: If the contribution to the hazard calculated in Step 5for
distances of 100 km or greater exceeds 5% for the average of 1
and 2.5 Hz, calculate the mean magnitude and distance of the
controlling earthquakes associated with the ground motions
determined in Step 2 for the average of 1 and 2.5 Hz.
Step 8: Determine the SSE response spectrum.

2.2. Probabilistic seismic hazard analysis

The PSHA was performed for a selected NPP site. The team
approach developed by EPRI (Electric Power Research Insti-
tute) was adopted for the hazard analysis. Three seismicity
expert teams and one attenuation team were composed to obtain Fig. 1. Example seismic source map used for the PSHA.
the PSHA input parameters. At least one non-seismologist was
included in each seismicity team. However, in the attenuation Fig. 1 shows one of the seismic source maps which was
team, only one expert recommended several different attenuation used for the evaluation of the seismicity by the expert team, and
equations with a weighting factor (Seo et al., 1999). Table 1 lists the six attenuation equations for the peak ground
A questionnaire was compiled and distributed to the team. acceleration and the spectral acceleration recommended by the
The contents of the questionnaire are as follows: expert.
Using the input data proposed by the expert teams, the PSHA
- Seismicity was performed for the example NPP site. Fig. 2 shows the seis-
(1) Matrix of the physical characteristics. mic hazard curves for the site.
(2) Assessment of the tectonic features according to the
matrix of the physical characteristics. 2.3. Scenario earthquakes for a Korean NPP site
(3) Seismic sources (source zone) and their inter-dependency.
(4) Maximum magnitude of each zone. The scenario earthquakes, which are specified in terms of
(5) Seismic parameters of each zone. the magnitude and the distance from the site under considera-
(6) Backup data (or interpretation) on the given figures. tion, can be obtained by a de-aggregation of the PSHA results
- Attenuation (strong ground motion) according to the USNRC Regulatory Guide 1.165 procedure
(1) Equations and their weights. summarized in Section 2.1. In this case study, the seismic hazard
(2) Background. was de-aggregated to determine the dominant magnitudes and

Table 1
Ground motion attenuation models
Ground motion measure Model Description Minimum distance (km) Weight

Peak ground acceleration Baag et al. (1998) South Korea 0 0.5

Toro et al. (1997) Central and Eastern North America 0 0.3
Xu et al. (1984) North China 0 0.2
Spectral acceleration Toro et al. (1997) Central and Eastern North America 0 0.5
Baag et al. (1998) South Korea 0 0.3
Atkinson and Boore (1995) Eastern North America 0 0.2
 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420 1413

Fig. 2. Seismic hazard curves for the example plant site.

distances at 1, 5 and 10 Hz at the median 1.0E−5 annual proba-

bility of exceedance level because the ground motion attenuation
equations proposed by the experts did not include an equation
for 2.5 Hz. The fractional contribution of the magnitude and
distance bin to the total hazard for 1 Hz was used for the devel-
opment of a low frequency scenario earthquake. Because the
contribution of the distance bins greater than 100 km contained
Fig. 3. Contribution factors for 1 Hz.
less than 5% of the total hazard for the 1 Hz, additional calcu-
lations to consider the effects of distant and larger events were
not needed.
The contribution of the magnitude–distance bins for 1 Hz
and the average of the 5 and 10 Hz are shown in Figs. 3 and 4,
respectively. The characteristics of the scenario earthquakes
which were determined based on their contributions are shown in
Table 2. The magnitudes and distances of the two scenario earth-
quakes are very similar. It may be due to the small contribution
of the distant earthquakes for the 1 Hz scenario earthquake.

2.4. Response spectra for the scenario earthquakes

The spectral shape for the scenario earthquakes were devel-

oped by using the attenuation relationships derived by Toro et
al. (1997), those by Baag et al. (1998) and those by Atkinson
and Boore (1995) as shown in Table 1. Fig. 5 shows the spec-
tral shapes for the scenario earthquakes normalized to 0.2 g
ZPA (zero period acceleration) together with the design based
response spectrum proposed in the USNRC Regulatory Guide
1.60 (1973) and the site-specific response spectrum to compare
their spectral shapes.

Table 2
Probability based scenario earthquakes
Frequency (Hz) Scenario earthquake Remarks

Magnitude Distance (km)

1 M6.4 9.0 Scenario I

5–10 M6.2 13.0 Scenario II
Fig. 4. Contribution factors for an average of 5 and 10 Hz.
1414 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420

Similarly, spectral accelerations for the response spectra of the

scenario earthquakes are smaller than those for the site-specific
response spectrum in the frequency range lower than about
10 Hz, while greater in the frequency range higher than about
10 Hz. Since there is a considerable amount of equipment and
components with a high natural frequency greater than 10 Hz in
NPPs, their seismic safety may be threatened under the scenario
earthquakes. Considering this circumstance, therefore, a seismic
reevaluation for the safety-related equipment and components at
the example NPP should be conducted and, if necessary, some
modification should be made to the vulnerable equipment and
components to increase their seismic resistance capacities and
to ensure the seismic safety of the plant.

3. Contribution of the safety-related equipment and

components to a core damage

Since many of the safety-related equipment and components

have a high natural frequency, a considerable difference between
the response spectra of the scenario earthquakes and the design
Fig. 5. Ground response spectra for the scenario earthquakes. response spectra in the high frequency range may threaten their
seismic safety. To cope with this problem appropriately, first of
Mean ground response spectra obtained from 270 earthquake all, it is important to elucidate the effect of the failure of each
records with magnitudes of 3–5 which occurred in the Korean safety-related equipment or component on the CDF of a plant
Peninsula are shown in Fig. 6. It is found that the spectral shapes under a seismic event.
for the scenario earthquakes shown in Fig. 5 are very similar to
the mean response spectrum developed from the real earthquake 3.1. Evaluation methodology for a CDF
data. This shows that the three attenuation relationships used in
this paper reflect the ground motion attenuation characteristics This study used the computer program EQESRA (1995)
and the site soil conditions relatively well. for evaluating the CDF of a NPP. EQESRA was developed
Spectral accelerations for the response spectra of the scenario to evaluate the probability distribution of a system failure fre-
earthquakes are smaller than those for the NRC design response quency from information about component fragilities (seismic or
spectrum in the frequency range lower than about 12 Hz, while non-seismic failures). The program performs component com-
greater in the frequency range higher than about 12 Hz. This binations in accordance with the Boolean expressions, and then
means that even though the SSCs were designed with a sufficient convolves the system fragility with the seismic hazard to yield
seismic margin for the NRC design response spectrum, the SSCs a probability distribution for a failure frequency. The EQESRA
with a natural frequency higher than 12 Hz become unsafe under program uses the methodology proposed by Kaplan (1981) and
the scenario earthquakes. Conversely, the SSCs with a natural Kaplan and Lin (1987).
frequency lower than 12 Hz become more conservative under The methodology for evaluating the seismic fragility of
the scenario earthquakes than under the design based spectrum. a structure or equipment, which is defined as a conditional

Fig. 6. Mean response spectra for the earthquake records for the Korean Peninsula. (a) NS component and (b) EW component.
 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420 1415

probability of its failure at a given value of a peak ground acceler- Table 3

ation, is described by Kennedy and Ravindra (1984). The ground Occurrence frequency and CDF for the seismic-initiating events
acceleration capacity is modeled as Initiating event Occurrence frequency CDF (ry−1 )

A = A m εR ε U (1) Loss of essential power 3.68E−06 3.68E−06

Loss of secondary heat removal 1.16E−06 1.16E−06
in which Am is the median ground acceleration capacity, εR and Loss of component cooling 2.48E−06 5.25E−08
εU are the random variables with unit medians representing an water/essential chilled water
Small LOCA 3.82E−08 3.82E−08
inherent randomness about the median and the uncertainty of a Loss of offsite power 1.12E−04 1.20E−06
median value, respectively. In this model, it is assumed that both General transient 2.79E−03 8.73E−07
εR and εU are lognormally distributed with logarithmic standard
Total 6.96E−06
deviations, βR and βU , respectively.
With perfect knowledge, the conditional probability of fail-
ure, f0 , for a given peak ground acceleration level, a, is given sions. The family of plant level fragility curves is evaluated
by: by combining the component fragility curves according to the
  Boolean expression for an accident sequence.
ln(a/Am )
f0 = φ (2) Each of the n plant level fragility curves for an accident
βR
sequence are convolved with each of the m seismic hazard curves
where φ[·] is the standard Gaussian cumulative distribution func- for the site; the convolution is expressed by the
tion.  ∞
When the modeling uncertainty βU is included, the fragility −dH(a)
S(a) da (4)
can be represented by a subjective probability density function 0 da
at each acceleration value as given by where −dH(a)/da is the frequency with which earthquakes occur
  in the size range da about a and S(a) is the conditional probability
 ln(a/Am ) + βU φ−1 (Q)
f =φ (3) of accident sequence.
βR
Assuming that pi (i = 1, 2, . . ., n) and hk (k = 1, 2, . . ., m) are
where Q = P[f < f |a] is the subjective probability (confidence) the probabilities associated with n plant level fragility curves
that the conditional probability of failure, f, is less than f for and m seismic hazard curves, respectively, n × m convolutions
a peak ground acceleration a and φ−1 [·] is the inverse of the are performed, resulting in n × m frequencies of a failure, with
standard Gaussian cumulative distribution function. associated probabilities pi hk . This assumes that the uncertainty
The accident-sequence definition and system modeling in a seismic hazard is independent of an uncertainty in fragility.
include the identification of accident initiating events, seis-
mic component failures, random failures, human errors and 3.2. Seismic-initiated events and core damage
dependent failure mechanisms that could cause these accident
sequences to occur. Event and fault trees are constructed to In the event of a strong earthquake, the failure of a safety
describe the accident sequences in the form of Boolean expres- system in a NPP brings the plant to a safe shutdown condition.
Table 4
Equipment and components selected through the fragility and screening analysis
Equipment/component Natural frequency Am (g) βR βU HCLPF Failure mode Related initiating event
(Hz) (g)

Offsite power – 0.30 0.22 0.20 0.15 Functional failure Loss of offsite power
Diesel generator >33 1.13 0.36 0.30 0.38 Concrete coning Loss of essential power
Essential chilled water compression tank >33 1.00 0.35 0.20 0.40 Anchorage Loss of essential chilled water
Battery charger 11.5 1.03 0.28 0.28 0.41 Functional failure Loss of essential power
1.54 0.33 0.33 0.52 Structural failure Loss of essential power
Condensate storage tank 9.95 0.91 0.21 0.27 0.41 Sliding Loss of secondary heat removal
Essential chilled water chiller 8 1.08 0.28 0.27 0.44 Structural failure Loss of essential chilled water
Regulating transformer 9.9 1.30 0.33 0.30 0.46 Functional failure Loss of essential power
Essential service water pump 33.98 1.20 0.29 0.28 0.47 Anchorage Loss of component cooling water
Component cooling water surge tank 17.6 2.00 0.41 0.47 0.47 Concrete coning Loss of component cooling water
4.16 kV switchgear 6 1.33 0.33 0.29 0.48 Functional failure Loss of essential power
Inverter 13.8 1.37 0.33 0.30 0.49 Functional failure Loss of essential power
Battery rack 23 1.46 0.33 0.31 0.51 Structural failure Loss of essential power
480 V load center 5.5 1.50 0.32 0.29 0.54 Functional failure Loss of essential power
Switches – 2.33 0.41 0.45 0.55 Functional failure Loss of essential power
Instrumentation tube (primary system) – 1.50 0.30 0.30 0.56 Piping break Small LOCA
125 V DC control center 8 1.58 0.33 0.29 0.57 Structural failure Loss of essential power
HVAC ducting and supports – 2.06 0.32 0.41 0.62 Functional failure Loss of essential power
Essential chilled water pump 37.2 1.85 0.36 0.27 0.65 Heavy duty bolt Loss of essential chilled water
1416 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420

Since the shutdown is caused through several paths of the sys- or developed on a plant-specific basis for components not fit-
tem, the possible paths that a plant would follow are identified. ting the generic component descriptions. Fragility functions for
These paths involve a seismic-initiated event that causes a shut- the generic categories are developed based on a combination
down, and a success or failure designation for the plant systems of experimental data, design analysis reports and an extensive
affecting the course of the events. Typically, the minimum set of expert opinion survey. A generic fragility for any particular
initiating events includes both loss of coolant accidents (LOCA) component can be estimated by selecting a set of site-specific
and transient events. In addition, the site-specific failure events, fragilities for that component.
which act as initiating events, may be added to a minimum set. For Younggwang Nuclear Units 5 and 6, after the seismic
For the Yonggwang Nuclear Units 5 and 6 in Korea, the hazard and seismic fragility analysis, a screening analysis was
seismic-initiated events include the following six events, by performed to identify the structures, components and equipment
considering the results of a seismic fragility analysis for their important for the seismic risk of the plants, and to minimize the
structures and equipment, and a result of the evaluation of a number of calculations for the significant contributors to the
relay chatter and its effect analysis (KEPCO, 2001). The seismic- seismic CDF. The result of the seismic fragility analysis for the
induced medium and large LOCAs were excluded through the site-specific spectrum shown in Fig. 5 included 7 civil structures
preliminary evaluation for the initiating events. and 67 items for the equipment and components whose fail-
ure during seismic events might conceivably affect the safety
- Loss of essential power (LEP). shutdown function of the plants. The screening criterion was
- Loss of secondary heat removal (LHR). based on HCLPF (high confidence of a low probability of fail-
- Loss of component cooling water/essential chilled water ure) value, which has a 95% confidence of not exceeding a 5%
(LOCCW). probability of producing a failure and indicates the seismic resis-
- Small loss of coolant accidents (SLOCA). tance of the equipment in terms of a gravitational acceleration
- Loss of offsite power (LOOP). calculated by
- Seismic induced general transient (GTRN).
HCLPF(g) = Am (g) × exp[−1.65(βR + βU )] (6)
In computing the frequency of the initiating events, a hier- where Am is the median seismic capacity and βR and βU are
archy between them must be established. The order of this the lognormal standard deviation for the randomness and uncer-
hierarchy is defined such that, if one initiating event occurs, the tainty, respectively.
occurrence of other initiating events further down the hierarchy For the screening analysis, the HCLPF value of 0.65 g was
has no significance in terms of a plant’s response. The seismic selected as a screening criterion under the assumption that the
event trees should be taken directly from those developed for effects of seismic-induced failures or the combined effects of
the internal events analysis, with modifications to include any their failures with a loss of offsite power directly lead to core
seismically induced failures. damage. This corresponds to a ground acceleration with a mean
The occurrence frequencies for the initiating events are cal- frequency higher than 1.0E−5 per year. Thus, this criterion
culated as

(5)

Seismic CDF for Younggwang Units 5 and 6 is calculated as eliminates all the equipment and components that can contribute
6.96E−06 per reactor year (ry) as shown in Table 3. The loss of no more than 5.0E−7 per year to the seismic CDF with a con-
essential power is a governing initiating event in calculating the fidence level of 95%. As a result of the screening analysis,
total CDF. The seismic CDF due to the loss of an essential power eighteen equipment and component items, which have HCLPF
occupies more than half of the total value. The loss of an essential values lower than 0.65 g, were finally selected from the results
power, loss of a secondary heat removal, and a small LOCA of the site-specific fragility analysis. Table 4 lists the selected
directly induce a core damage, whereas the loss of a component equipment and component items, together with their natural
cooling water/essential chilled water, loss of an offsite power frequencies, median fragilities, uncertainty parameters, HCLPF
and a general transient are coupled to the secondary event trees. values, failure modes and related initiating events. The offsite
power, switches, instrumentation tube and HVAC ducting and
3.3. Safety-related equipment and components important supports listed in Table 4 used the generic values.
for a seismic risk The comparison between the response spectra of the scenario
earthquakes and the site-specific response spectrum, shown in
Component seismic fragilities are obtained from a data base Fig. 5, reveals that there may be a change in the fragility value of
of generic fragility functions for seismically induced failures the equipment and components with a natural frequency between
 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420 1417

Fig. 7. Contribution of the equipment and components to a core damage.

10 and 50 Hz. This paper excludes a detailed evaluation of the response spectra of the scenario earthquakes in the frequency
effect of the scenario earthquakes on the fragility of the SSCs. range between 10 and 50 Hz. In a severe case, the scenario earth-
quakes may threaten the seismic safety of the equipment and
3.4. Contribution of the safety-related equipment and components with a high natural frequency greater than 10 Hz.
components to a core damage Thus, in order to improve the seismic safety of the plant under
this circumstance, a seismic reevaluation for the safety-related
Fig. 7 shows the contribution of a failure of the equipment equipment and components should be conducted and, if nec-
and component item to a seismic core damage in Youngg- essary, some modification should be made to the vulnerable
wang Units 5 and 6. It is found that the high contributors to equipment and components to increase their seismic resistance
a seismic core damage of the plant are the diesel generator capacities.
(29.8%), offsite power (18.3%) and condensate storage tank
(17.7%). 4.1. Effect of the seismic capacity of the equipment and
components on a seismic core damage
4. Determination of an effective seismic capacity
To investigate the effect of the seismic capacity of the
As shown in Fig. 5, the spectral accelerations for the site- equipment or components on the frequency of a seismic core
specific response spectrum are smaller than those for the damage of the plant, a seismic PSA is performed by using dif-
ferent seismic capacities of the four selected equipment and
components—diesel generator, offsite power, condensate stor-
age tank and battery rack. Although the contribution of the
batter rack to a core damage is not high, it is also considered

Table 5
Decrease of the CDF with an increase of the equipment seismic capacity
Equipment Increase ratio CDF (ry−1 ) CDF decrease
of seismic ratio (%)
capacity (%)

Diesel generator 25 5.83E−06 16.2

50 5.41E−06 22.3
Offsite power 25 6.36E−06 8.6
50 5.93E−06 14.8
Condensate storage tank 25 6.02E−06 13.5
50 5.86E−06 15.8
Battery rack 25 6.78E−06 2.6
50 6.74E−06 3.2
All 25 3.77E−06 45.8
50 2.47E−06 64.5
Fig. 8. CDF with an increase of the seismic capacity of the selected equipment 75 2.30E−06 67.0
and components.
1418 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420

In addition, when the condensate storage tank has an increased

seismic capacity, the CDF will decrease by more than 10%. The
effect of the seismic capacity of the battery rack on the CDF
is not considerable. When all of the selected equipment items
have a 25, 50 and 75% increased seismic capacity, the CDF will
decrease by 45.8, 64.5 and 67%, respectively.
Fig. 9 plots the ratios of the CDF for the equipment and
components with an increased seismic capacity to that with
an original capacity according to the peak ground accelera-
tion. It is found that the ratios of the CDF are considerably
influenced by the value of the peak ground acceleration, up
to 1.0 g. The effect of the seismic capacity of the equipment
on the CDF is remarkable in the PGA range of 0.2–0.6 g. If
the seismic capacities of all the selected equipment items are
improved, the CDF may be decreased by about 5 and 30% at 0.2
and 0.3g, respectively. At 0.4 g, increasing the seismic capac-
ity of the offsite power will be more effective, and, under 0.6 g,
increasing both the seismic capacities of the offsite power and
Fig. 9. CDF ratios with an increase of the seismic capacity of the selected the diesel generator will be more effective. In the case of the
equipment and components.
offsite power, at 0.4 g, an increase of its seismic capacity of
25 and 50% leads to a reduction of 33 and 45% in the CDF,
respectively.
in the investigation because it has a natural frequency of 23 Hz
and its fragility may be considerably affected by the scenario
earthquakes. 4.2. Effective seismic capacity of the selected equipment
Fig. 8 shows the relations between the cumulative mean fre- and components
quency of the failure and the peak ground acceleration for the
selected equipment items with different increased ratios of 25, Fig. 10 shows the relations between the HCLPF of the equip-
50 and 75%, and Table 5 summarizes the CDF and their ratios to ment and the CDF according to the median value of the seismic
the original value 6.96E−06 ry−1 shown in Table 3. It is found capacity and Table 6 summarizes the CDF for the different
from Fig. 8 and Table 5 that the failure of the diesel generator HCLPF values of the selected equipment and components. From
has a greater influence on the CDF than the other equipment. Fig. 10 and Table 6, the effective HCLPF values for the diesel
In other words, increasing the seismic capacity of the diesel generator, offsite power, condensate storage tank and battery
generator can improve the seismic safety of the plant consider- rack are determined as 0.84, 0.35, 0.63 and 0.63 g, respec-
ably. As shown in Table 5, when the diesel generator has a 25 tively. For a larger HCLPF value than the effective value, even
and 50% increased seismic capacity, the CDF will decrease by though the seismic capacity increases significantly, there is small
16.2 and 22.3%, respectively. This indicates that an increase of decrease in the CDF. In the case that all the four selected
more than 25% in the seismic capacity of the diesel generator equipment items have an increased seismic capacity, the CDF
can improve the seismic safety of the plant by more than 16%. decreases to 2.30E−06 ry−1 from 6.96E−06 ry−1 .

Table 6
Variation of the CDF for different HCLPF values of the equipment and components
Median seismic capacity (g) Diesel generator Offsite power Condensate storage tank Battery rack All

HCLPF (g) CDF (ry−1 ) HCLPF (g) CDF (ry−1 ) HCLPF (g) CDF (ry−1 ) HCLPF (g) CDF (ry−1 ) CDF (ry−1 )

0.05 0.02 2.79E−03 0.03 3.58E−05 0.02 2.89E−03 0.02 2.82E−03 2.91E−03
0.1 0.03 1.67E−03 0.05 2.38E−05 0.05 1.68E−03 0.03 1.66E−03 2.39E−03
0.3 0.10 1.49E−04 0.15 6.98E−06 0.14 1.14E−04 0.10 1.39E−04 2.33E−04
0.5 0.17 3.94E−05 0.25 6.04E−06 0.23 2.78E−05 0.17 3.65E−05 6.14E−05
0.7 0.24 1.65E−05 0.35 5.86E−06 0.32 1.10E−05 0.24 1.58E−05 2.34E−05
1.0 0.34 8.13E−06 0.50 5.82E−06 0.45 6.44E−06 0.35 8.67E−06 8.10E−06
1.2 0.40 6.59E−06 0.60 5.82E−06 0.54 5.99E−06 0.42 7.50E−06 5.00E−06
1.4 0.47 5.95E−06 0.70 5.82E−06 0.63 5.88E−06 0.49 7.06E−06 3.76E−06
1.6 0.54 5.65E−06 0.80 5.82E−06 0.72 5.86E−06 0.56 7.02E−06 3.01E−06
1.8 0.61 5.52E−06 0.90 5.82E−06 0.82 5.86E−06 0.63 6.79E−06 2.67E−06
2.0 0.67 5.45E−06 1.00 5.82E−06 0.91 5.85E−06 0.70 6.76E−06 2.48E−06
2.5 0.84 5.39E−06 1.25 5.82E−06 1.13 5.85E−06 0.87 6.73E−06 2.33E−06
3.0 1.01 5.38E−06 1.50 5.82E−06 1.36 5.85E−06 1.04 6.72E−06 2.30E−06
 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420 1419

Fig. 10. Relationship between the HCLPF values of the equipment and components and the CDF. (a) Diesel generator, (b) offsite power, (c) condensate storage tank,
(d) battery rack and (e) all.

5. Conclusions by more than 16%. In the case of increasing the seismic capac-
ities of the equipment which reveals a high contribution to a
This paper evaluated the effects of the seismic capacity of core damage, the CDF may be decreased by more than 50%.
safety-related equipment on the CDF of an existing NPP which The effective HCLPF values for the diesel generator, offsite
should be reevaluated for the modified seismic design spectra power, condensate storage tank and battery rack were deter-
and suggests the effective seismic capacities of the selected mined as 0.84, 0.35, 0.63 and 0.63 g, respectively. In the case that
safety-related equipment and components by using a probabilis- all four selected equipment and components have an increased
tic safety assessment. seismic capacity, the CDF decreases to 2.30E−06 ry−1 from
For the NPP considered in this case study, the failures of the 6.96E−06 ry−1 .
diesel generator, offsite power, condensate storage tank and bat- The seismic safety of existing NPPs can be improved easily
tery rack contribute remarkably to the CDF. When the diesel by an increase of the seismic capacity of the dominant equip-
generator or condensate storage tank has an increased seismic ment or components. Since the relationship between the seismic
capacity, the CDF will be decreased considerably, while in the characteristics of the facilities and the seismic risk of a nuclear
case of the battery rack the CDF does not decrease signifi- plant can be obtained by using a probabilistic safety assessment,
cantly. Increasing the seismic capacities of the diesel generator an improvement of the seismic safety of existing NPPs can be
by more than 25% can improve the seismic safety of the plant achieved effectively.
1420 Y.-S. Choun et al. / Nuclear Engineering and Design 238 (2008) 1410–1420

Acknowledgement McGuire, R.K., 1995. Probabilistic seismic hazard analysis and design earth-
quakes: closing the loop. Bull. Seismological Soc. Am. 85, 1275– 1284.
NUREG-1407, 1991. June Procedural and Submittal Guidance for the Individual
This study was supported by the Ministry of Science and
Plant Examination of External Events (IPEEE) for Severe Accident Vulnera-
Technology, Korean government, through its National Nuclear bilities, Final Report. US Nuclear Regulatory Commission, Washington, DC.
Technology Program. Regulatory Guide 1.165, March 1997. Identification and Characterization of
Seismic Sources and Determination of Safe Shutdown Earthquake Ground
References Motion. US Nuclear Regulatory Commission, Washington, DC.
Regulatory Guide 1.60, December 1973. Revision 1, Design Response Spec-
Atkinson, G.M., Boore, D.M., 1995. Ground-motion relations for eastern North tra for Seismic Design of Nuclear Power Plants. US Nuclear Regulatory
America. Bull. Seismological Soc. Am. 85, 17–30. Commission, Washington, DC.
Baag, C.-E., Chang, S.-J., Jo, N.-D., Shin, J.-S., 1998. Evaluation of seismic Seo, J.-M., Min, G.-S., Choun, Y.-S., Choi, I.-K., 1999. Reduction of Uncer-
hazard in the southern part of Korea. In: Proceedings of the 2nd Interna- tainties in Probabilistic Seismic Hazard Analysis. Korea Atomic Energy
tional Symposium on Seismic Hazards and Ground Motion in the Region of Research Institute KAERI/CR-65/99 (in Korean).
Moderate Seismicity, Seoul, Korea, pp. 31–50. Takada, T., Okumura, T., Hirose, J., Muramatsu, K., Taki, S., Ishii, K., 1999.
EQESRA, 1995. Version 3.0. Reference Document. EQE International, Inc. Probabilistic scenario earthquakes for seismic design—comparison of two
Hirose, J., Muramatsu, K., Okumura, T., Taki, S., 2002. A Procedure for the identification procedures. In: Proceedings of the OECD/NEA Workshop on
Determination of Scenario Earthquakes for Seismic Design Based on Prob- Seismic Risk, NEA/CSNI/R(99)28.
abilistic Seismic Hazard Analysis, JAERI-Research 2002-009. Japan Atomic Toro, G.R., Abrahamson, N.A., Schneider, J.F., 1997. Model of strong ground
Energy Agency. motions from earthquakes in central and eastern North America: best esti-
IAEA-TECDOC-724, 1993. Probabilistic Safety Assessment for Seismic mates and uncertainties. Seismological Res. Lett. 68, 41–57.
Events. International Atomic Energy Agency, Vienna, Austria. USNRC, November 1988. Generic Letter 88-20, Individual Plant Examination
Ishikawa, Y., Kameda, H., 1993. Scenario earthquakes versus probabilistic for Severe Accident Vulnerabilities, 10 CFR 50.54(f). US Nuclear Regula-
seismic hazard. In: Proceedings of the Sixth International Conference on tory Commission, Washington, DC.
Structural Safety and Reliability, Innsbruck, Austria. USNRC, June 1991. Generic Letter 88-20, Supplement No. 4, Individual Plant
Kaplan, S., 1981. On the method of discrete probability distributions in risk and Examination of External Events (IPEEE) for Severe Accident Vulnerabil-
reliability calculations—application to seismic risk assessment. Risk Anal. ities, 10 CFR 50.54(f). US Nuclear Regulatory Commission, Washington,
1, 189–195. DC.
Kaplan, S., Lin, J.C., 1987. An improved condensation procedure in discrete USNRC, September 1995. Generic Letter 88-20, Supplement No. 5, Indi-
probability distributions calculations. Risk Anal. 7, 15–19. vidual Plant Examination of External Events for Severe Accident
Kennedy, R.P., Ravindra, M.K., 1984. Seismic fragilities for nuclear power plant Vulnerabilities. US Nuclear Regulatory Commission, Washington, DC.
risk studies. Nucl. Eng. Des. 79, 47–68. Xu, Z., Shen, X., Hong, J., 1984. Attenuation relation of ground motion in north-
KEPCO, 2001. External Event Analysis for Yonggwang Units 5 and 6 PSA. ern China. In: Proceedings of the Eighth World Conference on Earthquake
Korea Electric Power Company (in Korean). Engineering, vol. II, San Francisco, USA, pp. 335–342.