Transactions of the 17th International Conference on Structural Mechanics in Reactor Technology (SMiRT 17) Prague, Czech Republic, August 17 –22, 2003

Paper # K03-2

1

Scenario Earthquakes for Korean Nuclear Power Plant Site Considering Active Faults In-Kil Choi1), Young-Sun Choun1) , Jeong-Moon Seo1)

1) Korea Atomic Energy Research Institute, Daejeon, Korea

ABSTRACT

Scenario earthquakes have been used for the design and safety assessment of nuclear power plant structures and equipments. Scenario earthquakes can be obtained from the deaggregation of Probabilistic Seismic Hazard Analysis (PSHA) results. In this study, the probability based scenario earthquakes were developed for the Korean nuclear power plant site using the PSHA data. The magnitude and distance bins of the scenario earthquakes for the example Korean nuclear power plant site were M6.4, 9km and M6.2, 13km. The ground response spectra for the scenario earthquakes were developed using the attenuation equations adopted in PSHA (Probabilistic Seismic Hazard Analysis). To incorporate the near-fault earthquake effects, the ground response spectra were modified by the correction factor. KEY WORDS : scenario earthquakes, probabilistic seismic hazard analysis, hazard de-aggregation, control earthquakes, near-fault, ground response spectrum INTRODUCTION

The seismic analysis and design have been performed based on a design basis earthquake. The design basis earthquake ground motion is generally specified as the design response spectrum. However the seismic safety of a nuclear power plant can not be secured by considering only the design basis earthquake, since the seismological situation of the nuclear power plant site is changed during the development of geosciences.

A survey on some of the Quaternary fault segments near the Korean nuclear power plants is ongoing [1]. It is likely that these faults would be identified as active ones. If the faults are confirmed as active ones, it will be necessary to reevaluate the seismic safety of nuclear power plants located near the fault.

This study is to develop the scenario earthquakes for the reevaluation of seismic safety of the nuclear power plant near the active fault. The probabilistic seismic hazard analysis for most of the Korean NPP sites have been completed. The hazard consistent earthquake scenario is developed as the probability based earthquake scenario using the existing results of probabilistic seismic hazard analysis.

The scenario earthquake is specified in terms of the earthquake magnitude, M, and its distance, R, from the site under consideration. The probability based scenario earthquake is developed by the de-aggregation of the probabilistic seismic hazard analysis results according to the procedures of the US NRC R.G. 1.165 [2]. The spectral shape for the scenario earthquake is developed using the attenuation equations adopted in PSHA. The near-fault ground motion effect is incorporated into the response spectra, since the potential active fault is located near the nuclear power plant site. Near-fault ground motions have caused much damage in the vicinity of seismic sources during recent earthquakes. This is due to the pulse-type ground motion, which has a large amount of input energy. Finally, this study proposes the scenario earthquake for the nuclear power plant site and corresponding response spectra considering the potential active fault effect. SCENARIO EARTHQUAKE DEVELOPMENT METHODS Scenario Earthquakes

Tow typical method, probabilistic seismic hazard analysis and deterministic seismic hazard analysis, are generally used to define the scenario earthquakes. The purpose of the seismic hazard analysis is to evaluate the annual probability of exceedance of various earthquake sizes at a given site, and to develop the spectral shapes of the motion from these earthquakes. This is called a probabilistic seismic hazard analysis to emphasize that its results are intrinsically probabilistic in nature. The seismic hazard of a nuclear power plant site is generally described by a series of seismic hazard curves, that is, a plot of the probability of exceedance vs. the ground motion intensity, such as peak ground acceleration, velocity or spectral acceleration. PSHA is very useful to define the scenario earthquakes, since it can determine the annual probability of exceedance for a ground motion intensity parameter [3].

The concept of the probability based scenario earthquake originates from McGuire studies [4]. The probability

2

based scenario earthquakes can be obtained from the de-aggregation of the PSHA results. It is intended to obtain particular sets of earthquake source magnitude and distance for the specified probability level. Two typical methods for defining the probability based scenario earthquakes are generally used. The first one is the method developed by U.S. NRC. PSHA can provide important information to define the safe shutdown earthquakes (SSE) in the recent revision of U.S. NRC Regulatory Guide 1.165 [2]. The scenario earthquakes are called controlling earthquakes in the Regulatory Guide. The second one was developed by Japan Atomic Energy Research Institute [5,6] base on the Ishikawa and Kameda study [7].

The two procedures mentioned above were adopted to define the scenario earthquakes for the example site by Takada et al. [8]. This study shows that both methods generate the similar results. But the authors point out the methodological difference between them. The U.S. NRC Regulatory Guide 1.165 procedure can not identify the earthquake source location of the scenario earthquakes. And this procedure uses coarser bins for the earthquake magnitude and distances, and all results including earthquake source regions and faults are mixed up in the bins. The Ishikawa’s procedure [7] has the advantage that it produces the source contribution factors that is an effective indicator for identifying which earthquake sources and/or which fault(s) are most influential, and one or more scenario earthquakes can be determined. Finally, it is concluded that the U.S. NRC Regulatory Guide 1.165 procedure gives a global view of the scenario earthquakes, while the Ishikawa’s procedure provides a more precise view of the scenario earthquakes. US NRC Regulatory Guide 1.165 Procedure

This procedure is originally an approach for determining the controlling earthquakes. In this study, the scenario earthquakes for a Korean nuclear power plant site were developed using this procedure. This procedure is based on a de-aggregation of the probabilistic seismic hazard in terms of earthquake magnitudes and distances. This section gives a simple description of the procedure [2]. Step 1 : Perform a site-specific PSHA. The hazard assessment (mean, median, 85th percentile, and 15th percentile)

should be performed at 1, 2.5, 5, 10, and 25Hz, and the peak ground acceleration. A lower bound magnitude of 5.0 is recommended.

Step 2 : Using the reference probability, determine the ground motion levels for the spectral accelerations at 1, 2.5, 5, and 10Hz from the total median hazard obtained in Step 1. Calculate the average of the ground motion level for the 1 and 2.5 Hz and the 5 and 10 Hz spectral acceleration pairs.

Step 3 : Perform a complete probabilistic seismic hazard analysis for each of the magnitude-distance bins. Step 4 : From the de-aggregated results of Step 3, the median annual probability of exceeding the ground motion

levels of Step 2 are determined for each magnitude-distance bin. Using the median annual probability, the fractional contribution of each magnitude and distance bin to the total hazard for the average of 1 and 2.5Hz, and 5 and 10Hz, 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 substantial. If the contribution to the hazard for distances of 100 km or greater exceeds 5%, additional calculations are needed to determine 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. SCENARIO EARTHQUAKES FOR KOREAN NPP SITE

In this study, the probability based scenario earthquakes for a Korean NPP site were developed using the method proposed by the U.S. NRC Regulatory Guide 1.165[2]. The example NPP site is located in the southeastern part of Korean peninsular. Probabilistic Seismic Hazard Analysis

The probabilistic seismic hazard analysis was performed for the NPP site. The team approach developed by EPRI (Electric Power Research Institute) was adopted for the hazard analysis. Three seismicity expert teams and one attenuation team were composed to obtain the PSHA input parameters. At least one non-seismologist was included in each seismicity team. However, in the attenuation team, only one expert recommended several different attenuation equations with weight [9].

A questionnaire was made and distributed to the team. The contents of the questionnaire are as follows.

3

- Seismicity

1) Matrix of physical characteristics 2) Assessment of tectonic features according to the matrix of physical characteristics 3) Seismic source (source zone) and their inter-dependency 4) Maximum magnitude of each zone 5) Seismic parameters of each zone 6) Backup data (or interpretation) on the given figures

- Attenuation (Strong ground motion)

1) Equations and their weights 2) Background Fig. 1 shows one of the seismic source maps which was used for the evaluation of seismicity by the expert team.

Table 1 shows the attenuation equations recommended by the expert. As shown in this table, the attenuation expert recommended three attenuation equations for the peak ground acceleration and three attenuation equations for the spectral acceleration with individual weight.

Using these PSHA input data proposed by the expert teams, the PSHA was performed for the site. Fig. 2 shows the seismic hazard curves for the example NPP site. Scenario Earthquakes

The seismic hazard was de-aggregated to determine dominant magnitudes and distances at the prescribed exceedance level. In this study, the seismic hazard was de-aggregated at 1Hz, 5Hz, and 10Hz at the 10-5 exceedance level according to the U.S. NRC Regulatory Guide 1.165 [2]. According to the guide, the seismic hazard should also be de-aggregated at 2.5Hz. But the ground motion attenuation equations proposed by the expert did not include the equation for 2.5Hz. The fractional contribution of magnitudes and distance bin to the total hazard for 1Hz was used for the development of a low frequency scenario earthquake. Because the contribution of the distance bins greater than 100km contained less than 5% of the total hazard for the 1Hz, additional calculations to consider the effects of distant and larger events were not needed.

Fig. 3 and 4 show the contribution of magnitude and distance bins for 1Hz and the average of the 5Hz and 10Hz, respectively. The scenario earthquakes for the example site were determined based on the contribution. Table 2 shows the magnitude and distance of scenario earthquakes for the example Korean NPP site. As shown in this table, the magnitudes and distances of the scenario earthquakes are very similar. It may be due to the small contribution of distant earthquakes of the 1Hz scenario earthquake. Near-fault Ground Motion Effects

Near-fault ground motions are ground motions that occur near an earthquake fault. In general, the near-fault ground motion records exhibit a distinctive long period pulse like time history with very high peak velocities. These features are induced by the slip of the earthquake fault. Near-fault ground motions, which have caused much of the damage in recent major earthquakes (Northridge 1994, Kobe 1995, Chi-Chi 1999), can be characterized by a pulse-like motion that exposes the structure to high input energy at the beginning of the motion. The recorded acceleration response spectra of recent major earthquakes are well enveloped by the design response spectra of the codes in the medium to high frequency range, but not in the low to medium frequency range [10].

The near-fault effects such as pulse-like motions can dramatically influence spectral content in large earthquakes. Some of these effects are most pronounced within about 10 km. The fault normal component is about 30% larger than the fault parallel component in the frequency range 0.2 to 0.5 sec due primarily to rupture directivity. The rupture directivity effects are strongest for strike slip motion on vertical faults but can also be significant for cases of directivity for sites located near dipping faults. Other factors, perhaps strongest at close distances, include hanging wall/foot wall site location as well as thrust verses strike slip or normal slip mechanisms. These additional factors can have significant impacts on spectral composition [11]. RESPONSE SPECTRA FOR SCENARIO EARTHQUAKES

The spectral shape for the scenario earthquakes were developed using the attenuation equations proposed in the PSHA study. The spectral shapes for the scenario earthquakes normalized to 0.2g ZPA (Zero Period Acceleration) are shown in Fig. 5. The spectral shapes for WUS (Western US) and CEUS (Central and Eastern US) are also shown in Fig. 5 for comparison. The seismic response spectral shapes for design and analysis for WUS and CEUS sites were proposed by McGuire et al. [11,12]. The response spectral shapes for the WUS site were developed from empirical attenuation equations in the WUS. For the CEUS, the WUS spectral shapes were modified with a transfer function based on the random vibration model of strong ground motion that accounts for differences in source parameters,

4

crustal damping, and near-surface damping. The spectral shapes for the WUS and CEUS sites can be obtained from the following equations, respectively.

+=

63

)exp()cosh(

]/)(ln[ 54

2

1CC ffCC

fCCPGAfSA (1)

2/1

8754

2

1963

)exp()exp()cosh(

]/)(ln[

++= CCC f

fCCf

fCCfC

CPGAfSA (2)

where, )( fSA and f are the spectral accelerations and frequencies, respectively. iC is a statistical coefficient defined as a function of magnitude and/or distance by creating a data set of response spectral shapes.

The spectral shapes for scenario earthquakes using the WUS and CEUS spectral shapes are also shown in Fig. 5. As shown in this figure, the spectral shapes from the proposed attenuation equations are similar to the shapes from the CEUS 1 corner frequency model.

Mean ground response spectra obtained from 270 earthquake records with magnitudes 3 to 5 which occurred in Korea are shown in Fig. 6 [9]. The fluctuation in the spectrum for magnitude 5 earthquakes is due to the very limited number of data. It is noted from Fig. 5 and Fig. 6 that the spectral shapes from the proposed attenuation equations are very similar to the mean response spectrum developed from the real earthquake data. The uniform hazard spectrum for the Korean nuclear power plant site was also very similar to the mean response spectrum developed from the real earthquake data [13]. This result shows that the ground motion attenuation equations used in the seismic probabilistic hazard analysis reflected relatively well the ground motion attenuation characteristics and the site soil condition.

These features of near-fault ground motion are generally not considered in the seismic design of nuclear power plant structures and components. Many researches have been performed to identify the characteristics of near-fault ground motion [14,15,16]. Ohno et al. [17] showed the range that the near-fault rupture directivity effect is dominant and proposed a method to correct the predefined response spectrum considering this effect. Fig. 7 shows the FN (Fault Normal) to FP (Fault Parallel) response spectral ratio from 37 records of the 11 strong earthquakes. These strong motion records were strongly affected by the near-fault directivity effect. Based on this study, Nishimura et al. [18] propose a correction factor to modify the response spectrum. The correction factor, )( iTλ , can be obtained from the following equations.

1)( =iTλ for Di TT ≤ (3)

)/log(/)/log()5.2log(10)( DHDi TTTT

iT⋅=λ for iD TT < (4)

where, iT denotes the period. DT (= 0.33sec) and HT (= 5sec) are the control points of the design ground response spectrum. These equations express that only the spectral acceleration in long period range greater than 0.33 sec is increased due to the near-fault rupture directivity effect. And, it is assumed that the spectral acceleration amplification due to the near-fault rupture directivity effect does not appeare in short period range shorter than 0.33 sec.

Fig. 8 shows the correction factor to incorporate the near-fault directivity effect on the response spectrum shape. As shown in this figure, the correction factors from the equation can express the near-fault rupture directivity effect which appeared in the strong earthquake records.

Using the proposed equation, the spectral shapes for the scenario earthquakes were modified to incorporate the near-fault rupture directivity effect. Fig. 9 shows the modified response spectral shape for the two scenario earthquakes.

CONCLUSIONS

In this study, the probability based scenario earthquakes for the Korean NPP site were developed using the PSHA

results. The magnitude and distance bins of the scenario earthquakes for the example Korean nuclear power plant site were M6.4, 9km and M6.2, 13km. The magnitude and distance of the two scenario earthquakes for 1Hz and the average of the 5Hz and 10Hz are very similar, since the contribution of distant earthquakes to the total hazard is very small.

5

The spectral shapes for the scenario earthquakes were developed using the ground motion attenuation equations proposed by an expert in PSHA and empirical spectral shapes for WUS and CEUS sites. The spectral shapes from the attenuation equations were very similar to the spectral shape from the CEUS 1 corner frequency model. And the characteristics of the spectral shape from attenuation equation were similar to that of the mean response spectrum of the real earthquake records. This means that the ground motion attenuation equations used in the seismic probabilistic hazard analysis reflected relatively well the ground motion attenuation characteristics.

Recent strong earthquakes that occurred in near-fault regions showed unique characteristics that distinguish them from the far-field earthquakes. The rupture directivity effect is a dominant factor that impacts on spectral shape. In this study, using the correction factors, the response spectra for the scenario earthquakes considering the near-fault rupture directivity effect were developed. The correction factor causes the enrichment of the spectral acceleration in low frequency region.

A survey on some of the Quaternary fault segments near Korean nuclear power plants is ongoing. If the faults are confirmed as active ones, it will be necessary to reevaluate the seismic safety of nuclear power plants located near the fault. The scenario earthquakes and their response spectra can be used for reevaluation of seismic safety of the nuclear power plants near the active faults.

ACKNOWLEDGEMENT

This research was supported by the Mid- and Long-Term Nuclear Research & Development Program of the

Ministry of Science and Technology, Korea.

REFERENCES [1] Korea Institute of Nuclear Safety, Development of Seismic Safety Evaluation Technology for NPP Sites,

KINS/GR-206, 2000. [2] US NRC Regulatory Guide 1.165, Identification and Characterization of Seismic Sources and Determination of

Safe Shutdown Earthquake Ground Motion, 1997. [3] Yutaka ISHIKAWA and Hiroyuki KAMEDA, “Scenario Earthquakes vs. Probabilistic Seismic Hazard,”

Proceedings of ICOSSAR ’93, Paper No. 410/4/27, 1993. [4] McGuire, R. K., “Probabilistic Seismic Hazard Analysis and Design Earthquakes : Closing the Loop,” Bulletin of

Seismological Society of America, Vol. 85, No. 5, 1995. [5] J. Hirose, K. Muramatsu, T. Okumura, and S. Taki, A Procedure for the Determination of Scenario Earthquakes

for Seismic Design Based on Probabilistic Seismic Hazard Analysis, JAERI-Research 2002-009, 2002. [6] K. Muramatsu and J. Hirose, "The Use of Seismic Hazard Analysis for Determining Scenario Design Earthquakes

- An Overview of a Recently Completed Work at JAERI -," Procedings of the Seventh Korea-Japan Joint Workshop on Probabilistic Safety Assessment, 2002.

[7] Ishikawa, Y. and H. Kameda, “Scenario Earthquakes vs. Probabilistic Seismic Hazard,” Proc. of Fourth International Conference on Structural Safety and Reliability, 1993.

[8] T. Takada, T. Okumura, J. Hirose, K. Muramatsu, S. Taki and K. Ishii, “Probabilistic Scenario Earthquakes for Seismic Design –Comparison of Two Identification Procedures,” Proceedings of the OECD/NEA Workshop on Seismic Risk, NEA/CSNI/R(99)28, 1999.

[9] Jeong-Moon Seo, Gyung-Shik Min, Young-Sun Choun, and In-Kil Choi, Reduction of Uncertainties in Probabilistic Seismic Hazard Analysis, KAERI/CR-65/99, 1999.

[10] NEA, Lessons Learned from High Magnitude Earthquake with Respect to Nuclear Codes and Standards, NEA/CSNI/R(2002)22, 2003.

[11] R. K. McGuire, W. J. Silva and C. J. Costantino, Technical Basis for Revision of Regulatory Guidance on Design Ground Motions : Hazard- and Risk-consistent Ground Motion Spectra Guidelines, NUREG/CR-6728, 2001.

[12] R. K. McGuire, W. J. Silva, and C. J. Costantino, Technical Basis for Revision of Regulatory Guidance on Design Ground Motions : Development of Hazard- and Risk-consistent Seismic Spectra for Two Sites, NUREG/CR-6769.

[13] In-Kil Choi, Young-Sun Choun, Jeong-Moon Seo, and Kwan-Hee Yun, “Reevaluation of Seismic Fragility Parameters of Nuclear Power Plant Components Considering Uniform Hazard Spectrum,” Journal of the Korean Nuclear Society, Vol. 34, No. 6, 2002.

[14] Paul Somerville, “Characterization of Near-Fault Ground Motions,” Proc. of the US-Japan Workshop on the Effects of Near-Field Earthquake Shaking, 2000.

[15] Shizuo Noda, Kazuhiko Yashiro, Katsuya Takahashi, Masayuki Takemura, Susumi Ohno, Masanobu Tohdo, and Takahide Watanabe, “Response Spectra for Design Purpose of Stiff Structures on Rock Sites,” OECD/NEA workshop on the Relations Between Seismological Data and Seismic Engineering, Istanbul, 16-18 October 2002.

6

[16] Wilfred D. Iwan, “Near-Field Considerations in Specification of Seismic Design Motion,” Proc. of 10th European Conference on Earthquake Engineering, 1995.

[17] S. Ohno, M. Takemura, and Y. Kobayashi, “Effects of Rupture Directivity on Near-Source Strong Motion,” Proc. 2nd International Symposium on the Effects of Surface Geology on Seismic Motion, 1998.

[18] Isao Nishimura, Shizuo Noda, Katsuya Takahashi, Masayuki Takemura, Susumu Ohno, Masanobu Tohdo, and Takahide Watanabe, “Response Spectra for Design Purpose of Stiff Structures on Rock Sites,” Transactions of SMiRT 16, Paper No. 1133, 2001.

Table 1. Description of Ground Motion Attenuation Models [9]

Table 2. Probability Based Scenario Earthquakes

Ground Motion

Measure

Model # Description Var. Minimum

Distance Weighting References

1 South Korea 0.6 0 km 0.5 Baag, 1998

2 Central & Eastern North America 0 km 0.3 Toro, Abrahamson

and Schneider, 1997PGA

3 North China 0 km 0.2

Sum =1

Zhixin, Xiaobai and Jingru

4 Central & Eastern North America 0 km 0.5 Toro, Abrahamson

and Schneider, 1997

5 South Korea 0 km 0.3 Baag, 1998

SA (Spectral

Acceleration)

6 Eastern North America 0 km 0.2

Sum =1

Atkinson and Boore, 1995

1Hz 5-10Hz

M6.4, 9.0Km M6.2, 13.0km

Fig. 1. Example Seismic Source Map Used in PSHA

7

1.00E- 09

1.00E- 08

1.00E- 07

1.00E- 06

1.00E- 05

1.00E- 04

1.00E- 03

1.00E- 02

0 200 400 600 800 1000Acceleration(gal)

P(A>

a)/Year PGA

25Hz10Hz5Hz1Hz0.5Hz

Fig. 2. Seismic Hazard Curves for Example Site

0-10

10-25

25-50

50-100

100-150

150-200

>200

5-5.55.5-66-6.56.5-77-7.5

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350 5-5.55.5-66-6.56.5-77-7.5

0-10

10-25

25-50

50-100

100-150

150-200

>200

5-5.55.5-66-6.56.5-77-7.5

0.000

0.050

0.100

0.150

0.200

0.250

5-5.55.5-66-6.56.5-77-7.5

Fig. 3. Contribution Factors for 1 Hz Fig. 4. Contribution Factors for Average of 5 and 10 Hz

0.1 1 10 100

Frequency(Hz)

0.0001

0.001

0.01

0.1

1

Spec

tral A

ccel

erat

ion(

g)

Attenuation eq.EUS_1CORNEREUS_2CORNERWUS

0.1 1 10 100

Frequency(Hz)

0.0001

0.001

0.01

0.1

1

Spec

tral A

ccel

erat

ion(

g)

Attenuation eq.EUS_1CORNEREUS_2CORNERWUS

(a) 1Hz Scenario Earthquake (b) 5-10Hz Scenario Earthquake Fig. 5. Ground Response spectra for Scenario Earthquake

8

0.1 1.0 10.0Frequency(Hz)

1E-4

1E-3

1E-2

1E-1

1E+0

1E+1

Spec

tral A

ccel

erat

ion(

gal)

Magnitude

3

4

5

Fig. 6. Mean Response Spectra for Real Earthquake Records