recognition and overcome of uncertainty in seismic hazard ...€¦ · recognition and overcome of...

Recognition and Overcome of Uncertainty in Seismic Hazard Evaluation

K.Ebisawa1), K.Irino1), H.Nanba1), H.Tsutsumi1), M.Sakagami1), M.Hirano1), H.Kameda2)

1) Japan Nuclear Energy Safety Organization (JNES), Japan 2) National Research Institute for Earth Science and Disaster Prevention (NIED), Japan

Abstract

As a part of seismic probabilistic safety assessment of nuclear power plants, Japan Nuclear Energy Safety Organization (JNES) have been advancing the probabilistic seismic hazard evaluation method. The characteristic of JNES’s method is to use two ways for calculating ground motions. One is based on the empirical attenuation model. The other is based on the fault model and applied for near large active faults around a site. The uncertainty evaluation is carried out using logic tree. The procedure to elicit and integrate the expert’s opinions and to generate the logic tree was developed. Through the uncertainty evaluation in seismic hazard at several sites, important insights concerning elicitation and integration of expert opinions were obtained JNES held the workshop on seismic hazard assessment with the focus on recognition and overcome of the uncertainties in seismic hazard evaluation on April 27, 2004. The important common recognition was obtained from the workshop. This paper describes the above contents.

1

1. Introduction As a part of seismic probabilistic safety assessment of nuclear power plants, Japan Nuclear Energy

Safety Organization (JNES) have been developing and advancing the probabilistic seismic hazard evaluation method. The characteristic of JNES’s method is to use two ways for calculating ground motions, based on the empirical attenuation model and the fault model for near large active faults around a site.

In JNES’s procedure for evaluating the uncertainties in seismic hazard, the uncertainty is divided into aleatory uncertainty and epistemic uncertainty. The later uncertainties are presented in a logic tree. This procedure was applied to seismic hazard analyses for several sites1). In the process of determination of the analysis models and conditions, expert opinions were elicited through a series of questionnaire investigation and discussion among several experts.

JNES held the workshop on seismic hazard assessment with the focus on recognition and overcome of the uncertainties in seismic hazard evaluation on April 27, 20042). In this workshop, we had presentations of the JNES activities and by two invited experts and also panel discussion, and several important consensus and insights were obtained.

This paper describes the method to evaluate seismic hazard using attenuation and fault models and the example of evaluation result, the procedure to generate the logic tree on JNES, and the overview of the above workshop. 2. Methodology for evaluating uncertainty of seismic hazard on JNES (1) Methodology to evaluate seismic hazard on JNES

Japan Nuclear Energy Safety Organization has been advancing the probabilistic seismic hazard evaluation method. The characteristic of JNES’s method is to use two ways for calculating ground motions as shown in Fig.1. One is based on the empirical attenuation model. The other is based on the fault model and applied for near large active faults around a site. The seismic hazard curve at a site is obtained by combining the two hazard curves which are evaluated separately based on these ways.

In this method, the fault parameters are determined by the recipe proposed by Irikura et al. (Irikura recipe3)), and uncertainty of these fault parameters are taken into account as a logic tree. The seismic motions for each path in the logic tree are calculated by the fault model. (2) Classification of uncertainty factor

In JNES’s procedure for evaluating the uncertainty in seismic hazard, the uncertainty is divided into aleatory uncertainty for the randomness and epistemic uncertainty for the knowledge lack, as shown in Fig.1. The later uncertainty is presented as branches in the logic tree.

The aleatory uncertainty is caused by the intrinsic property of the nature itself and that is generally

irreducible randomness Examples of this uncertainty and its treatment are shown in the following: - Variability of seismic motion and size of next earthquake, etc. - Incorporated into one hazard curve

2

・source parameter・earthq. motion parameters・source parameter・earthq. motion parameters

deterministic parameters


undeterministicparameters


Logic treeLogic tree

seismic hazard evaluationseismic hazard evaluation

Collection and Analysis of related Seismic Information(Source characteristic, historical earthquake data, active fault data, trenching investigation etc.)


Earthquake(intraplate earthq., interplate earthq. ）


modelingmodelingmethodmethod

Condition of E.Q.motionCondition of E.Q.motion

Select of method Near large fault

aleartoryuncertainty for

randomness

epistemic uncertainty for knowledge lack

Uncertainty factor

Seismic hazard Evaluation based on LTSeismic hazard Evaluation based on LT

All pathTI & Expert team･questionnaire

investigation･elicitation of opinions･generation of LT･weighting

Important parameterImportant parameter

Fractile hazard curveFractile hazard curve

PGA

Exc

eeda

nce

Freq

. Sensitive studySensitive study

Fault model

LTLTHazard evaluation for all path of LTHazard evaluation for all path of LT

tentative hazard

・frequency・source, propagation,

site amplification characteristics

・analysis method

Attenuation model・magnitude, location,

frequency・attenuation model,

logarithmic standard deviation, upper limit of earthquake motion

Area source

Exc

eeda

nce

Freq

.

Exc

eeda

nce

Freq

.PGA PGA

・source parameter・earthq. motion parameters・source parameter・earthq. motion parameters





Logic treeLogic tree

seismic hazard evaluationseismic hazard evaluation





modelingmodelingmethodmethod

Condition of E.Q.motionCondition of E.Q.motion

Select of method Near large fault

aleartoryuncertainty for

randomness

epistemic uncertainty for knowledge lack

Uncertainty factor

Seismic hazard Evaluation based on LTSeismic hazard Evaluation based on LT

All pathTI & Expert team･questionnaire

investigation･elicitation of opinions･generation of LT･weighting

Important parameterImportant parameter

Fractile hazard curveFractile hazard curve

PGA

Exc

eeda

nce

Freq

. Sensitive studySensitive study

Fault model

LTLTHazard evaluation for all path of LTHazard evaluation for all path of LT

tentative hazard

・frequency・source, propagation,

site amplification characteristics

・analysis method

Attenuation model・magnitude, location,

frequency・attenuation model,

logarithmic standard deviation, upper limit of earthquake motion

Area source

Exc

eeda

nce

Freq

.

Exc

eeda

nce

Freq

.PGA PGA

Fig.1 Procedure for evaluating seismic hazard on JNES

On the other hand, the epistemic uncertainty is caused by a lack of knowledge on modeling shown as follows. That is reducible with increasing knowledge. An example of this uncertainty and its treatment are:

- Presence of a fault - Usually shown as multiple hazard curves

The modeling of the epistemic uncertainty is most difficult and important to obtain reliable seismic hazard curves.

(3) Procedure to evaluate epistemic uncertainty of seismic hazard using logic tree In the seismic hazard evaluation using logic tree as shown in Fig.1, the important parameters are identified by the sensitivity study. On these parameters, the logic tree is generated. Then the seismic hazard curves are calculated for each path in the logic tree. The fractile seismic hazard curve is estimated by statistical analysis.

3

3. Procedure for eliciting and integrating of expert’s opinions and generating logic tree on JNES

The procedure to elicit and concentrate the expert’s opinions and generate the logic tree consists of 11 steps as shown in Fig.2.

1. Collection of related seismic information 2. Identification of important issues

To identify the important issues within the epistemic uncertainties based on the above results 3. Selection of experts

To select the experts concerning the above important issues 4. Selection of technical integrator

To select the technical integrator to elicit and integrate the expert’s opinions 5. Selection of methods to elicit the expert’s opinions

The methods to elicit the expert’s opinions are as follows. - Questionnaire - Interview - Exchange of opinions within group - Workshop - Combination of the above methods

Identification of important items Elicitation and integration of expert’s opinion

Review by experts

Generation of LTSet up of site

Basic analysis

Sensitive analysis

Calculations of hazard for each path based on LT

Combination of seismic hazard curve

Identification of important items

selection of experts

Elicitation of expert’s opinion by questionnaire

Opinion exchange among experts

Selection of method to read expert’s opinions

Tentative LT

Completion of LT

Review by experts

Review by experts

Selection of TI

Sensitive analysis

・questionnaire

・interview

・discussion in group

Estimation of weight by TI

Opinion exchange

Collections of related seismic information and various analyses

*TI：ＴｅｃｈｎｉｃａｌＩｎｔｅｇｒａｔｏｒ

2～3times

Identification of important items Elicitation and integration of expert’s opinion

Review by experts

Generation of LTSet up of site

Basic analysis

Sensitive analysis

Calculations of hazard for each path based on LT

Combination of seismic hazard curve

Identification of important items

selection of experts

Elicitation of expert’s opinion by questionnaire

Opinion exchange among experts

Selection of method to read expert’s opinions

Tentative LT

Completion of LT

Review by experts

Review by experts

Selection of TI

Sensitive analysis

・questionnaire

・interview

・discussion in group

Estimation of weight by TI

Opinion exchange

Collections of related seismic information and various analyses

*TI：ＴｅｃｈｎｉｃａｌＩｎｔｅｇｒａｔｏｒ

2～3times

Fig.2 Procedure for evaluating uncertainty of seismic hazard on JNES

4

6. Elicitation of expert’s opinions based on questionnaire investigation

In order to elicit the expert’s opinions, the following devices need to the questionnaires. - To set up the concrete questions - To offer the results of sensitivity analyses - To set up the answer columns including answers of weighting for each model and

parameter and free discussion 7. Exchange of expert’s opinions among experts

To repeat about 2 or 3 times the exchange of expert’s opinions and to integrate them 8. Generation of tentative logic tree

The tentative logic tree is generated by the following ways. - That is generated by the technical integrator. - That is generated through the exchange of expert’s opinion.

9. Review of tentative logic tree among experts 10. Generation of final logic tree 11. Review of final logic tree among experts

4. Examples of uncertainty evaluation of seismic hazard and knowledge obtained through evaluation (1) Example of uncertainty evaluation of seismic hazard using attenuation model a. Evaluation conditions

A site located at east Japan was selected to estimate. The important items were identified based on sensitivity analysis results for the factors of epistemic uncertainties as shown in Table 1. They are the following four items.

- Item 1: Modeling of seismic zone and source depth - Item 2: Length, magnitude and frequency of specific fault - Item 3: Selection of attenuation model, setting of deviation and upper limit of seismic motion - Item 4: Combination of historical earthquake data and active fault data

The experts were selected from committee members of Seismic Hazard assessment WG (11persons of

JNES. Their specialty field and organization are as follows. - Specialty field: Geology, topography, seismology, earthquake engineering and reliability

engineering etc. - Organization: University, research organization and utility

The team was set up as follows.

- Team for modeling seismic source: 7 persons - Team for modeling seismic motion propagation: 5 persons

5

Table 1 Uncertainty factors in the case estimated seismic hazard using attenuation model

1) Selection of attenuation model2） Set up of logarithmic standard deviation

The same as left contentsThe same as left contents○

○

○

3) Selection of truncate deviation

○ Seismic motion propagation model

○4) Set up of seismic zone○4) Set up of seismic zone

○

○

○

○

3) Magnitude, location, frequency① Magnitude（Ｍ）・frequency（ν）

・ Set up of magnitude・ Set up of minimum magnitude・ Set up of maximum magnitude

② Location（epicenter, depth）

・Plain (longitude, latitude): uniformdistribution

・Depth: α times for fault length

○

○

○

○

○

○

3) Magnitude, location, frequency① Magnitude（M）

・Set up of Magnitude

② location（epicenter, depth）


・Depth: α times for fault length③ Frequency（ν）: based on non- poison

process・Selection of non-poison model・Set up of mean return period (Ts=D/Vs)・Set up of Average slip rate (Vs) and

displacement (D)・Set up of last occurrence time（Ｔo）

○

○

○

○

○

○

○

3) Magnitude, location, frequency①Magnitude（M）

・Set up of fault length (way of segmentation)

・Selection of fault length –magnitudeequation

②location（epicenter, depth）


・Depth: α times for fault length③Frequency（ν）: based on non- poison



○

○○

○

2）Set up of historical earthquake in catalogs・Set up of area around site (longitude,

latitude )・Set up of historical time and magnitude・Avoidance of after shocks・Avoidance of double count fault and

historical data

○

○

○

○


latitude )・Set up of historical time and magnitude・Avoidance of after shocks・Avoidance of double count fault and historical data

○

○

○

2) Set up of active fault in catalogs・based on area around site (longitude,

latitude )・based on confidence (I, II, III) of active

fault

○1) Selection of historical earthquake datacatalog


○1) Selection of active fault data catalogs

○ Hypocenter model○ Hypocenter model○ Hypocenter model

UＲB value modelUＲMaximum moment modelURMaximum moment model

Area hypocenterSpecial hypocenter

The same as left contentsThe same as left contents○

○

○

3) Selection of truncate deviation

○ Seismic motion propagation model

○4) Set up of seismic zone○4) Set up of seismic zone

○

○

○

○

3) Magnitude, location, frequency① Magnitude（Ｍ）・frequency（ν）

・ Set up of magnitude・ Set up of minimum magnitude・ Set up of maximum magnitude

② Location（epicenter, depth）


・Depth: α times for fault length

○

○

○

○

○

○

3) Magnitude, location, frequency① Magnitude（M）

・Set up of Magnitude

② location（epicenter, depth）


・Depth: α times for fault length③ Frequency（ν）: based on non- poison



○

○

○

○

○

○

○

3) Magnitude, location, frequency①Magnitude（M）

・Set up of fault length (way of segmentation)

・Selection of fault length –magnitudeequation

②location（epicenter, depth）


・Depth: α times for fault length③Frequency（ν）: based on non- poison



○

○○

○


latitude )・Set up of historical time and magnitude・Avoidance of after shocks・Avoidance of double count fault and

historical data

○

○

○

○


latitude )・Set up of historical time and magnitude・Avoidance of after shocks・Avoidance of double count fault and historical data

○

○

○

2) Set up of active fault in catalogs・based on area around site (longitude,

latitude )・based on confidence (I, II, III) of active

fault



○1) Selection of active fault data catalogs

○ Hypocenter model○ Hypocenter model○ Hypocenter model

UＲB value modelUＲMaximum moment modelURMaximum moment model

Area hypocenterSpecial hypocenter

1) Selection of attenuation model2） Set up of logarithmic standard deviation

b. Evaluation results

As example of uncertainty evaluation of seismic hazard using attenuation model, the examination process concerning item 2 is shown in Fig.3. In this figure, the first questionnaire was carried out to the experts with the introduction of specific active fault data and seismic zone map etc. The results of questionnaire were integrated into the following four opinions.

- Opinion 1-1: Length (l)- 80 km, magnitude prediction equation (M)- Utsu equation , average slip rate (V)- no description

- Opinion 1-2: L- 30 km, M- Matsuda equation, V- S1 - Opinion 1-3: L- 21 km, M- Matsuda equation, V- S2 or S3 - Opinion 1-4: L- 18 km, M- Matsuda equation, V- S4 or S5

After the discussion was carried out among the experts based on the above four opinions, the second

questionnaire was carried out again. The results of questionnaire were integrated into the following two opinions.

- Opinion 2-1: L- 70 km, M- Matsuda equation, A- S1 - Opinion 2-2: L- 18 km, M- Matsuda equation, A- S2 or S3 or S4 or S5

The discussion was carried out among the experts about the details of reference materials and sensitivity analysis results for the above two opinions

The technical integrator finally estimated the following weight taking account of the above discussion. - Opinion 2-1: weight -0.2 for L- 70 km, M- 7.9 by Matsuda equation, A- 1/11300 years

6

- Opinion 2-2: weight- 0.8 for L- 18 km, M- 6.9 by Matsuda equation, A- 1/6000 years The fractile of seismic hazard of before and after the opinion exchange among experts is shown in Fig.4.

Introduction of reference materials

・Active fault data

・Establishment permit

・Seismic zone map

etc

Result of first questionnaire

･L： 80km

･M：Utsu eq.･－

･L： 30km

･M：Matsuda eq.･Average slip rate S1

Discussion of experts

・Location and length of specific fault・Validation of fault segment

・Magnitude and frequency of specific fault

・Average slip rate・Relationship between

validation and frequency・Addition to Matsuda eq.・Hypocenter mechanism

･L： 70km

・M： Matsuda eq・Average slip rate S1

Result of second questionnaire

Judgment of TI

・Out of branch frequency (ν)（Small impact for hazard)

・Degree of detail of referencematerials

・Distribution of expert’s opinion

Weight

L=70km（M 7.9）

ν1=1/11300 0.2

L=18km（M 6.9）

ν5=1/6000 0.8

TI：ＴｅｃｈｎｉｃａｌＩｎｔｅｇｒａｔｏｒ

･L： 18km

･M： Matsuda eq･ Average slip rate S4

･L： 21km

・Ｍ： Matsuda eq･Average slip rate S2

S3

･L：18km

･M： Matsuda eq･average slip rate

S2, S3, S4 , S5

200 400 600 800 1000 1200 1400010-7

10-6

10-5

10-4

10-3

10-2

10-1

100

最大地動加速度(Gal）

年超

過確

率

基本

CASE 1

CASE 2,3,4,5

〔Sensitivity analysis〕

S5

・Degree of detail of reference materials ・Results of sensitivity

analysis


analysis

Discussion of expertsIntroduction of reference materials

・Active fault data

・Establishment permit

・Seismic zone map

etc

Result of first questionnaire

･L： 80km

･M：Utsu eq.･－

･L： 30km

･M：Matsuda eq.･Average slip rate S1


・Location and length of specific fault・Validation of fault segment

・Magnitude and frequency of specific fault

・Average slip rate・Relationship between

validation and frequency・Addition to Matsuda eq.・Hypocenter mechanism

･L： 70km

・M： Matsuda eq・Average slip rate S1

Result of second questionnaire

Judgment of TI

・Out of branch frequency (ν)（Small impact for hazard)

・Degree of detail of referencematerials

・Distribution of expert’s opinion

Weight

L=70km（M 7.9）

ν1=1/11300 0.2

L=18km（M 6.9）

ν5=1/6000 0.8

TI：ＴｅｃｈｎｉｃａｌＩｎｔｅｇｒａｔｏｒ

･L： 18km

･M： Matsuda eq･ Average slip rate S4

･L： 21km

・Ｍ： Matsuda eq･Average slip rate S2

S3

･L：18km

･M： Matsuda eq･average slip rate

S2, S3, S4 , S5

200 400 600 800 1000 1200 1400010-7

10-6

10-5

10-4

10-3

10-2

10-1

100

最大地動加速度(Gal）

年超

過確

率

基本

CASE 1

CASE 2,3,4,5

〔Sensitivity analysis〕

S5


analysis


analysis


Fig.3 Process concerning Location, length, magnitude and frequency of the specific fault (items 2)

100

10-7

10-6

10-5

10-4

10-3

10-2

10-1

0 500 1000 1500

0.84 fractile0.50 fractile

Before opinionexchange

After opinion exchange

Exce

edan

cefr

eque

ncy

(1/y

ear)

Maximum acc.（Gal)

100

10-7

10-6

10-5

10-4

10-3

10-2

10-1

0 500 1000 1500


10-7

10-6

10-5

10-4

10-3

10-2

10-1

0 500 1000 1500


Before opinionexchange

After opinion exchange

Exce

edan

cefr

eque

ncy

(1/y

ear)

Maximum acc.（Gal)

Fig.4 Comparison of hazard curve before and after opinion exchange

7

(2) Example of uncertainty evaluation of seismic hazard using fault model The method to evaluate the uncertainty of seismic hazard using fault model was applied to a site in

Japan. The hypocenters around a site consist of those based on (1) historical earthquake, (2) active fault data and (3) near large active fault.

The seismic hazards of the hypocenters based on (1) and (2) were evaluated using attenuation model. The hypocenters of (3) were defined as the fault parameters; asperity, stress drop, etc., determined by the Irikura receipt. The variability of these fault parameters was considered by referring expert opinion and the probable combinations of their parameters were represented as a logic tree. The seismic motions for each path in the logic tree were calculated using fault model and the hazard of (3) was evaluated as shown in Fig.5. The seismic hazard curves of (1), (2) and (3) were combined as shown in Fig.6.

（）: Weight of path

Rupturepoint

Num. of asperity in segments

Stress dropof asperity

Stress of elementGround motions

Maximum acceleration for each pathsMaximum acceleration for each paths

Calculation of ground motions by fault model


AnalysisCondition each path

ｾｸﾞﾒﾝﾄ 1



破壊開始点

ｻｲﾄ

Acceleration time histories for each paths

10-5

10-4

10-3

10-2

10-1

100

0 500 1000 1500

年超過

確率

最大地動加速度 [cm/s/s]

10-5

10-4

10-3

10-2

10-1

100

0 500 1000 1500

年超過

確率


Example of hazard by fault model

Rupture point

segment1

segment2

segment3

Exce

edan

cefr

eque

ncy

Maximum acceleration (Gal)

（）: Weight of path

Rupturepoint

Num. of asperity in segments

Stress dropof asperity

Stress of elementGround motions

Maximum acceleration for each pathsMaximum acceleration for each paths



AnalysisCondition each path




破壊開始点

ｻｲﾄ

Acceleration time histories for each paths

10-5

10-4

10-3

10-2

10-1

100

0 500 1000 1500

年超過

確率


10-5

10-4

10-3

10-2

10-1

100

0 500 1000 1500

年超過

確率


Example of hazard by fault model

Rupture point

segment1

segment2

segment3

Exce

edan

cefr

eque

ncy

Maximum acceleration (Gal)

Fig.5 Example of seismic hazard evaluation using fault model

8

Exce

edan

cefr

eque

ncy

(1/y

ear)

Maximum ground motion (Gal)

Combination hazard

Hazard by fault model

Hazard of zone x by attenuation model

Hazard of zone y+z by attenuation model

10-5

10-4

10-3

10-2

10-1

100

0 500 1000 1500

Exce

edan

cefr

eque

ncy

(1/y

ear)

Maximum ground motion (Gal)

Combination hazard

Hazard by fault model

Hazard of zone x by attenuation model

Hazard of zone y+z by attenuation model

10-5

10-4

10-3

10-2

10-1

100

0 500 1000 1500

Fig.6 Example of evaluating seismic hazard using attenuation and fault models

(3) knowledge obtained through evaluation

Through carrying out such trial process, the following important consideration items were identified

concerning elicitation of expert opinions and integration of discussion points. a. Selection of expert

- To select from various specialty fields and organizations - To select the experts for important items

b. Concerning Technical Integrator (TI)

- To maintain the fairness - To avoid the compulsory opinions - To maintain the fairness in management - To promote the positive discussion

c. Questionnaire investigation

- To offer the high qualities information - To avoid the vogue answer by showing suitable sentence

9

- To elicit the various opinions by showing column of free answer d. Concerning elicitation of expert pinions

- To carry out the opinion exchange several times by suitable members (5-6 persons), - To delete the unimportant opinion based on the results of sensitivity study, - To carry out the opinion exchange under no compulsion, - To agree the process of technical decision making based on scientific information, - To maintain the transparency of opinion exchange process and - To clarify the items on which the opinions are not in consensus finally.

4. Overview of international seismic hazard workshop organized by JNES (1) Program of workshop

JNES held the workshop on seismic hazard assessment with the focus on recognition and overcome of the uncertainties in seismic hazard evaluation on April 27, 2004. The program of this workshop is shown in Table 2.

In this workshop, the above JNES’s activity was introduced. Then the presentations of Dr. Leon Reiter (U.S. Nuclear Waste Technical Review Board) and Dr. Paul G. Somerville (URS Corporation) invited from U.S.A. were made. Furthermore the panel discussion with the title of recognition and overcome of uncertainty in seismic hazard evaluation was carried out. Several important insights were obtained at the workshop.

(2) Important common consensus obtained by workshop

The important common consensus obtained from the workshop is as follows. - Seismic PSA is effective method to evaluate the seismic safety of nuclear plants and to identify their

weakness. - The factors of uncertainty in seismic hazard are divided into the aleatory and epistemic uncertainties. - The elicitation and integration of various expert opinions concerning the epistemic uncertainties are

very important in order to utilize the logic tree. In the same way, the transparency and accountability are also very important.

- The practical experiences of logic tree have to be accumulated. - It is important to approach from both science and technology sides and to promote the mutual

cooperation and understanding - It is also important to study based on the analyses with taking account of new knowledge.

10

Table 2 Program of workshop organized by JNES on April 26, 2004.

Japan N

Workshop on

Time／Date： 9:30－18:00 ／ Place： Tokyo Palace HoContents：

・Opening Address (9:30～・Introduction －Purpose of

(National R

Japan

Workshop on

Time／Date： 9:30－18:00 ／Place： Tokyo Palace Contents：・Opening Address (9:30～

・Introduction －Purpose (National Research Institu・Presentation on Seismic

・Invited Presentations (1・Uncertainty in Seismic H

・Treatment of UncertainCharacteristics at Yucca

＜Lunch（13:00～1・Panel Discussion（14:00:～・Title: Recognition and Ove・Coordinator: Tsuyoshi Ta・Panelists:

Yoshihiro Kinugasa (Takashi Kumamoto (

Hiroyuki FujiwarTakao Kagawa (Geo Shizuo Noda (Tokyo Tadashi Annaka (TokDr. Leon Reiter（USDr. Paul G. SomervillKatsumi Ebisawa (JN

(1) Short Presentations (6 Pa＜Coffee Break (20 min

(2) Discussions including flo・Summary (17:30-17:55) ：M・Closing (17:55～18:00)：M

Incorporated Administrative Agency uclear Energy Safety Organization (JNES)

How to Overcome Uncertainties in Seismic Hazard ―Application of Logic Treeー

Agenda

April 27 (Tuesday), 2004 tel

9:45) ：Hideki Nariai (President, JNES) the WS－ (9:45～10:00)：Hiroyuki Kameda

Incorporated Administrative Agency Nuclear Energy Safety Organization (JNES)

How to Overcome Uncertainties in Seismic Hazard - Application of Logic Tree -

Agenda

April 27 (Tuesday), 2004 Hotel

9:45) ：Hideki Nariai (President, JNES) of the WS－ (9:45～10:00)：Hiroyuki Kameda te of Earth Science and Disaster Prevention) Hazard Evaluation Methodology in JNES －Logic Trees－

（10:00～11:00）： Katsumi Ebisawa（JNES） 1:00～13:00)

azard Analysis: Some Lessons Learned： Dr. Leon Reiter (U.S. Nuclear Waste Technical Review Board)

ty in Earthquake Source and Strong Ground Motion Mountain： Dr. Paul G. Somerville (URS Corporation)

4:00）＞ 17:30） rcome of Uncertainty in Seismic Hazard Evaluation kada (University of Tokyo)

Tokyo Institute of Technology) Okayama University) a (National Research Institute of Earth Science and Disaster Prevention) Research Institute) Electric Power Company) yo Electric Power Services Co., Ltd.) NWTRB） e（URS Corporation） ES)

nelists) and Discussions utes) > or asaharu Sakagami (JNES)

esearch Institute of Earth Science and Disaster Prevention) itsumasa Hirano (JNES)

11

5. Summarization and future plan (1) Summarization

The main examination results are summarized as follows.

- The characteristic of the method to evaluate seismic hazard by JNES is to use two ways for calculating ground motions. One is based on the empirical attenuation model. The other is based on the fault model and applied for near large active faults around a site.

- In the method of uncertainty evaluation on JNES, the uncertainty is divided into aleatory uncertainty and epistemic uncertainty. The later uncertainty is presented in the logic tree. The procedure to elicit and integrate the expert’s opinions and generate the logic tree was developed.

- This method was applied to the evaluation for several sites. Through the evaluation, the important consideration items concerning elicitation and integration of expert opinions were identified.

- JNES held the workshop on seismic hazard assessment with the focus on recognition and overcome of the uncertainties in seismic hazard evaluation on April 27, 2004. The important common consensus was obtained from the workshop. The main items are as follows.

- Seismic PSA is effective method to evaluate the seismic safety of nuclear plants and identify their weakness.

- The elicitation and integration of expert opinions related the epistemic uncertainties are very important in order to utilize the logic tree. In the same way, the transparency and accountability are also very important.

- It is important to approach from both science and technology sides and to promote the mutual cooperation and understanding

(2) Future plan

In order to advance the method of seismic hazard evaluation, the following issues will be carried out.

- To examine the factors of the uncertainty and to set up the appropriate standard deviation by considering that the deviation depends on the amplitude of seismic motion.

- To examine the limited value of seismic motion and to set up the truncation value such as 3 standard deviation.

- To generate the standard manual of logic tree considering the elicitation and integration of expert’s opinions, which is suitable to our country.

References 1) Hiroyuki Kameda, K.Kosaka, Kenji Hagio and Masaharu Sakagami: STUDYONMETHODOLOGY

OF SEISMIC HAZARD EVALUATION IN JAPAN,, Proceedings of the OECD/NEA Workshop on Seismic Risk, Nuclear Safety NEA/CSNI/R(99)28, pp.77-98, 2000.

2) Katsumi Ebisawa: Seismic hazard evaluation methodology in JNES –Logic trees-, Workshop on how to overcome uncertainties in seismic hazard assessment –Application of logic tree method- organized by JNES, pp.2-1 - 2-30, April 27, 2004.*

3) Kojiro Irikura, Hiroe Miyake, Tomotaka Iwata, Katsuhiro Kamae, Hidenori Kawabe, Luis Angel Dalgure: Recipe for Predicting Strong Motion from Future Large Earthquake, Annuals of Disas. Prev. Res. Inst., Kyoto Univ., No.46 B, 2003.*

*: In Japanese

12

recognition and overcome of uncertainty in seismic hazard ...€¦ · recognition and overcome of...

Documents