NUCLEAR SAFETY & SEISMIC RISK MANAGEMENT IN FRANCE: OVERVIEW

Transcription

1 NUCLEAR SAFETY & SEISMIC RISK MANAGEMENT IN FRANCE: OVERVIEW Catherine BERGE-THIERRY, Seismologist & Seismic Risk Expert at CEA SEPTEMBER 28, 2016, SCIENTIFIC & TECHNICAL SEMINAR AT THE CANADIAN NUCLEAR SAFETY COMMISSION 21 OCTOBRE 2016 CEA 10 AVRIL 2012 PAGE 1

2 CONTENT 1. Introduction 2. Seismology & French Approach to define the seismic hazard for nuclear facilities 3. French Approach and acceptance criteria for the design and assessment of nuclear facilities 4. Seismic risk management in light of Fukushima action items 5. The research project 6. Conclusions et Discussions CEA SEP., 28 th, 2016 PAGE 2

3 5. THE RESEARCH PROJECT CEA 10 AVRIL 2012 PAGE 3 21 OCTOBRE 2016

4 5. THE RESEARCH PROJECT 5.1. Framework & Motivations of the project (SEISM Institute), 5.2. Objectives, Resources and Partnership SMIRT-2015 presentation 5.3. Scientific Structure 5.4. Synthesis of identified limitations of the current regulatory nuclear practice 5.5. Overview of the 5 scientific work packages 5.6. Focus on some specific key issues NED post- SMIRT-2015 special issue Paper accepted for WCEE2017 control point where is defined the hazard? Toward an outcropping bedrock seismic hazard definition instead of current free field SHA including partially site effects Non-linear interactions between seismic wave field, soil and foundations CEA SEP., 28, 2016 PAGE 4

5 TOWARD AN INTEGRATED SEISMIC RISK ASSESSMENT FOR NUCLEAR SAFETY IMPROVING CURRENT FRENCH METHODOLOGIES THROUGH THE RESEARCH PROJECT CATHERINE BERGE-THIERRY 1, P.Y. BARD 2, T. CHARTIER 3, R. COTTEREAU 4, E. BERTRAND 5, F. LOPEZ-CABALLERO 4, D. CLOUTEAU 4, S. GRANGE 6, S. ERLICHER 7, F. HOLLENDER 8, P. KOTRONIS 9, M. LANCIERI 3, A. LAURENDEAU 8, A. LE MAOULT 8, N. MOUSSALLAM 10, M. NICOLAS 8, F. RAGUENEAU 11, J.F.SEMBLAT 12 AND F. VOLDOIRE 13 1 ATOMIC ENERGY COMMISSION (CEA) SENIOR RESEARCHER & SINAPS@ COORDINATOR, FRANCE 2 ISTERRE, 3 IRSN, 4 ECP, 5 CEREMA, 6 INP-GRENOBLE & UJF, 7 EGIS, 8 CEA, 9 ECN, 10 AREVA, 11 ENS CACHAN, 12 IFSTTAR, 13 EDF, Paper 571, Division VII PAGE 5

6 OUTLINE Context project motivations Focus on the «Risk Assessment» Work Package 4 In short WP1, WP2, WP3, WP5 and WP6 Conclusion CONSULTANCY MEETING EBP PHASE 2 CEA 19 JUILLET 2012 WORKING AREA 2, TASK 2.1. FEB TH, 2016 PAGE 6

7 CONTEXT - SEISMIC MARGINS (S.M.) Several international programs on S.M. Niigataken Chuetsu Oki, M6,6 EQ 2007 : Kashiwazaki-Kariwa NPP experience IAEA Karisma benchmark Extension life duration of NPP s (subject raised in France since 2009), 2011, Tohoku, M9, EQ & Fukushima accident : Underestimation of Seismic & Tsumani Hazards Consequences on French NP s : Complementary Safety Studies (C.S.S.) ( ) French Nuclear Safety Authority asked all Nuclear Operators to assess the capacity of the existing NPP to sustain seismic levels higher than the one considered for design and/or during safety reassessment reviews. requires to assess the Seismic Margins of existing plants PAGE 7

8 R&D collaborative project Early 2012 the French Government published a call to initiate research projects to improve NP s Safety regarding internal and external events. SINAPS@: «Earthquake and Nuclear Plants Improving and Sustaining Safety» Main Issue: Identification and propagation of uncertainties (epistemic and aleatory) on data and methods in the assessment of seismic risk (deterministic & probabilistic approaches) in moderate to low seismic areas (considering extreme seismic levels) : Critical Opinions on the French and International Practices, Improve Methodologies Contribute to Seismic Margins Assessment, & Formulate Recommendations Strong Partnership ensuring the completeness of skills 13 teams, ~60 researchers/eng., 12PhD s, 19 post-docs Full Cost ~13M, National Funding 5 M PAGE 8

9 - 6 WP S «WP3» Structural & SSC s response «WP2» N.L. Site Effects & SSI «WP1» Seismic Hazard Demonstrative test case: Kashiwazaki Kariwa NPP site «WP4» Seismic Risk Assessment CEA 19 JUILLET 2012 PAGE 9 + WP 5 BBI & WP 6 Knowledge Dissemination

10 : WP4 RISK ASSESSMENT & KK DEMONSTRATIVE CASE (D.C.) Ground motion parameters devoted to seismic risk assessment Parameters adapted to the vulnerability analysis, involving SSI. WP1, WP2 & WP3 Simulation and propagation of uncertainties Statistical meta-models built on best-estimate finite element simulations and sampling techniques, Performance of Bayesian methods and extreme statistics for fragility curves computing. Demonstrative numerical case study Kashiwazaki-Kariwa BWR unit 7 Objective: from the fault to probabilistic floor spectra, including N.L. Site Effect simulation, N.L. Soil-Structure Interaction, RC N.L. structural behavior. KK Test case based on: Data from the July 16 th 2007 Niigata-ken-Chuetsu-Oki earthquake, TEPCO measurements available for the former International benchmark KARISMA, [ near-field natural seismic signals, soil & site geotechnical data, nuclear island structural parameters...] PAGE 10

11 WP4 THE KK D.C. First Step : Estimating the Plant Fragility using current tools & practices as required by French deterministic regulation uncertainties treated through standard coefficients Seismic scenario : Chuetsu Oki 2007 EQ, Seismic Level [ Spectral Approach / GMPE(s) ], SSI, structural seismic & floor responses [elastic- linear - BEM-FEM] Accelerograms selected from Japanese databases. Fragility curves computed for 2 SSC s Damage criteria assessed using EPRI simplified approach Structural model of RB Unit 7, [Banci and Zentner 2015]. Second Step : Re-Assess the Plant Fragility using state of art knowledge & innovative methods [SINAPS@], including variabilities and N.L. propagating uncertainties through probabilistic approaches General 2D stratigraphy [ V. Pavlenko and K. Irikura]. Seismic inputs from a PSHA study, 3D accelerograms from UHS (WP1), Use more «realistic» methods including variabilities & uncertainties, N.L., SSI and Site Effects (WP2), Use of N.L. structural models provided by (WP3), Fragility curves assessed using probabilistic approaches [uncertainties accounted],(wp4). KK D.C. (i) get forward S.M. through identification of key parameters & assumptions uncertainties treatment & methods used in the whole seismic chain. CEA 19 JUILLET 2012 PAGE 11 (ii) validate and disseminate methodologies for practitioners and structural engineers.

12 WP1 MAIN ISSUES & PROGRESS «SEISMIC SOURCES CARACTERISATION, GROUND MOTION PREDICTION & UNCERTAINTIES» French Metropolitan territory - Diffuse seismicity, - Uncertainties in meta data of seismic catalogs, - Difficulty to identify faults and associate deformation rates, - Poor or lack of knowledge on soil properties (site effects) ( ) - Which Seismic Scenario? (Magnitude? Distance?) - Which Maximum magnitude to account for (PSHA)? - Large dispersion in Ground motion predictions ( ) Great Uncertainties Large dispersion in SHA Major challenges : (i) Analysis of appropriate methods for seismic hazard evaluation vs seismicity knowledge and the uncertainties. (ii) Provide to WP2-3-4&5 accurate seismic outputs Improving Characterization of «French data» & Metadata and their uncertainties; Sensitivity of methods (probabilistic or deterministic) towards data / assumptions Hierarchy in data/parameters used in seismic hazard process and evaluation of impact of uncertainties; Interface between hazard and vulnerability of structures - " relevant indicators «/ selection of time series? PAGE 12

13 WP 1 : S.H. practices & outputs Deterministic scenario Current practice in France Seismotectonic Zone Fault Probabilistic : UHS IAEA main practice / post-fukushima Assessments Gutenberg-Richter Recurrence Characteristic earthquake Seismicity dmin Prediction of ground motion and variability «Reference earthquake(s)» Prediction of seismic motion = response spectr(a)um PSA Great regional earthquake Small local earthquake Envelop Spectrum WP2, WP3 and WP4 : - Spectra «levels» / Uncertainties Need of time series (T.S.) Uniform Hazard Spectrum Hz -Methodologies to select T.S.? - Impact of selected T.S. (natural or synthtics) on structural response? 13

14 WP2 MAIN ISSUES & PROGRESS NON LINEAR SITE EFFECTS / SOIL-STRUCTURE INTERACTION (1/3) KEY ISSUE: Track & propagate the uncertainties, avoiding double counting from WP1 to WP2! SINAPS@ strategy: WP1 : providing Hazard at the Bedrock condition reference, WP2: Assess Non-linear effects from the bedrock to the plant foundations, & SSI Objectives of WP2 : Improvement of current-practice methods defining the input motion at structure base Based on results obtained from WP1 Including spatial variability of seismic motions quantification of the effect of uncertainties of various soil materials Development of new methods From the fault to the equipment's : including non-linear behaviour and variability Coupling of the seismic source models, wave propagation, and structural codes New seismic data acquisition to validate numerical developments In high seismic activity countries (Japan) and a European seismic framework (low to moderate) <-> Greek test site PAGE 14

15 WP2 MAIN ISSUES & PROGRESS NON LINEAR SITE EFFECTS / SOIL-STRUCTURE INTERACTION (2/3) Argostoli Greece - Test Site Vertical and Horizontal accelerometric networks, Spatial variability/coherency of seismic motions, Validation in 2D or 3D conditions 60*60*30 km 3 mesh of Argostoli Island and the surrounding sea Argostoli Test site Regional model to site effects / spatial variability and SSI studies (data constrained) Seismic network installed in the framework of SINAPS@ after the Argostoli January 26, 2014 M6 EQ Continuous Data acquisition 3D Spectral Elements Simulation Spring 2015 : Release of the Linear Soil behavior, To be completed by a kinematic source model, N.L. soil behavior and structural code Development of a large-scale non-linear probabilistic model from source to structure enable to account for variabilities and propagate uncertainties.

16 Models reduction WP 3 STRUCTURAL AND COMPONENTS SEISMIC BEHAVIORS (3/4) Major challenge : Enhance the modeling relevance regarding the structural vulnerability assessment (under severe seismic loadings) Task 1 : Model calibration and Experimental Comparisons 1. Calibrate the dynamic model 2. Estimate the Modeling to Experiment Gap 3. Modeling update Task 2 : Nonlinear Dynamic behaviours Modal superposition, Multifiber beams models, Plates and shells 2D models, Full 3D models, Multi-scales analysis Task 3: Seismic Isolation Reinforcement / Design Optimization Strong interactions WP1 : S.H. outputs relevance / time series selections/seismic damage indicators (PhD) WP4 : provide relevant structural models from simplified to complex one s All WP : check uncertainties propagation/treatment CEA 19 JUILLET 2012 PAGE 16

17 WP 5 BUILDING / BUILDING INTERACTION (BBI) Objectives: Evaluation and reduction of pounding of existing adjacent nuclear buildings during earthquakes. Topic highlighted in French (3/4) Post-Fukushima C.S.S. Pounding can create Stresses in local areas and collapse Floor spectra modification Work package resources: Access to CEA/TAMARIS shaking table facility 1 PhD (Sept-2015), 2 scholarships 50 man.months Ressources for mockup, instrumentation, etc. One of the very few ST study (California) Work plan: State of the art (2014) Design of the experimental setup (2015) Preliminary numerical investigation (2015) Tests (2016 & 2017) Analyses (2017 & 2018) CEA 19 JUILLET 2012 PAGE 17 Azalée CEA shaking table From A. Le Maoult et al., 2015.

18 Objectives : SINAPS@ WP 6 TRAINING & KNOWLEDGE DISSEMINATION 2 training sessions during the project, (3/4) All topics of Seismic Risk Assessment, particularly for NPs, State of the Art of knowledge, data, modelings and methodologies Current seismic risk approaches, but also other innovative. Session 1 Summer School June 2016 (in French) for Master, Ph D s, post doctorates & young researchers/engineers. Session 2: in 2017 in collaboration with IAEA-ISSC (in English) for researchers (academic, safety authorities, technical supports, gov.org., design offices ). Participation of international experts. CEA 19 JUILLET 2012 PAGE 18

19 CONCLUSION aims To prioritize parameters and Assess Impact of Uncertainties (data & methods) on all key steps: seismic hazard, site effects, soil-structure interaction, seismic behavior of structures and equipment's, risk assessment. To Identify & quantify potential seismic margins from data / meta data / assumptions/ methods / uncertainties treatment WP4 KK D.C. To disseminate Knowledge/Practices on Seismic Risk Assessment [2 training sessions 2016 & 2017] To Formulate Recommendations for future R&D actions and efforts on data acquisition, regulatory developments and updates Achieving these goals calls for strong inter-disciplinarity in and between all WP s Currently 6 complementary WPs whose experimental & numerical work is now well engaged ~ 60 researchers/engineers involved, 12 PhDs, 19 post-docs already working or beginning soon Next SINAPS@ progress point during SMIRT24! CEA 19 JUILLET 2012 PAGE 19

20 PAPER PUBLISHED SOON IN NUCLEAR ENG. & DESIGN Paper which presents, for each step of the seismic risk analysis: The state of practice in the French Nuclear Approach, The advantages and limits identified from return experience, Illustrates the gap with recent R&D results and improvments, The specific (on data) or more generic (methods) objectives of CEA 19 JUILLET 2012 PAGE 20

21 PAPER ACCEPTED FOR THE WCEE2017 WCEE-2017-SINAPS Paper which presents a practicle application on the Kashiwazaki-Kariwa site. On the influence of the «control point» where the seismic input (from SHA) is transferred to the SSI computation, On soil non-linearity / outcropping bedrock condition On the use of the equivalent linear method to deconvolve the seismic input from the surface down to foundations or basement reactor, Check the impact on the fragilty curves estimates, A two steps study : Initial Phase: «French current practice» (free field seismic input & deconvolution) Final phase: defining the seismic motion at the outcropping bedrock, CEA 19 JUILLET 2012 PAGE 21

22 GLOBAL SEISMIC RISK ANALYSIS CEA 19 JUILLET 2012 PAGE 22

23 - CASE STUDY Assumptions: (see KARISMA, IAEA 2010 benchmarck) Seismic scenario: the 2007 NCOE earthquake (Mag 6.6 / Epicentral dist 16 km), Reactor building N 7 Soil column: very low Vs 30m (250 m/s) at the near surface, Bedrock found at 167m in depth (Vs 30m = 720 m/s) RB7 model: simplified but including the embedment Structural behavior : elastic linear Soil : non linear properties SSI : equivalent linear method Linear transfer from structure to equipment CEA 19 JUILLET 2012 PAGE 23

24 CASE STUDY SCHEME OF GEOLOGY/ RB7 Simplified scheme of the KK DC. The RB7 is embedded over 25 m. Beneath the soil the bedrock is found at ~167 m in depth. Stars indicate location of different control points. CEA 19 JUILLET 2012 PAGE 24

25 STRUCTURAL RB7 MODEL CEA 19 JUILLET 2012 PAGE 25

26 CASE 1: SEISMIC INPUT AT THE FREE SURFACE (STAR 1) SOFT SOIL VS 30 =250 m/s NCOE 2007 scenario (Mw=6,6 and epic.dist. of 16 km), 50 synthetic ground motions have been generated whose mean response spectrum fits the target scenario spectrum assessed using the Campbell and Bozorgnia (CB) GMPE). Figure 6 presents the initial 50 strong motions set. To increase the seismic inputs number, the classical engineering scaling process is applied on the set (with factors of 0,5, 1, 2, 2.5 and 3). CEA 19 JUILLET 2012 PAGE 26

27 CASE 1: RECORDED NCOE DATA VS SYNTHETICS CEA 19 JUILLET 2012 PAGE 27

28 DECONVOLUTION FROM SURFACE TO - 25m (EMBEDMENT FUNDATIONS LEVEL, CONTROL POINT 3 To account for the nonlinear soil behavior, a linear equivalent approach is used (similar to that used for the original KK benchmark): for each seismic input of the 250 strong motions (amplification factors of 0.5, 1, 2, 2.5 and 3 on initial 50 data), an iterative procedure is applied assessing the equivalent soil column properties (through the shear strain, G modulus reduction, damping ratio). Finally the seismic input at the reactor basement is obtained for every input signal, if the process converges. A significant number of seismic signals (among them, especially those coming from the scaling process with factors 2, 2.5 and 3) produced divergence in the linear equivalent deconvolution approach, highlighting the problem related to the use of the method above its own limitations (usually 0.1% shear strain, threshold also recommended in [3]) to a soil site which is highly nonlinear in such acceleration domains. In the following, we consider deconvolution results only if the maximal soil shear strain does not exceed 0,8% (as done in the IAEA Karisma benchmark). CEA 19 JUILLET 2012 PAGE 28

29 CASE 2: SEISMIC INPUT AT THE OUTCROPPING BEDROCK (STAR 2) - VS 30 = 720 m/s THEN DECONVOLUTION TO 25m Now 50 seismic signals are generated at the control point 2, at the ground surface for a bedrock site condition (outcropping bedrock, Vs30=720 m/s) in order to avoid the soil non linearity phenomenon, still fitting the CB2008 GMPE. Figure 9 : seismic motions re-assessed at the RB7 basement, after the deconvolution. In the left figure ( case 1 ), a large amplitude appears for one of the signal exhibiting the deconvolution failure with such a nonlinear soil using linear equivalent method. On right, generating the initial seismic input at the outcropping bedrock ( case 2 ) ensures the stability of the deconvolution. CEA 19 JUILLET 2012 PAGE 29

30 IMPACT ON THE FRAGILITIY CURVES CASES 1 & 2 a fictive equipment - resonance frequency postulated at 4 Hz: the failure criterion is the exceedance of its 5% damped PSA at 4 Hz of a level of acceleration; this is supposed unknown, and will be explored during the study. RB7 responses sets 1 and 2, RB basement motions transmitted to the equipment. Figure 10 presents in ordinates the PSA values corresponding to seismic inputs from set 1 (triangles) and to set 2 (circles) as function of PGA values at control point 3 - RB basement at -25 m: the color scale is related to the soil distorsion rate. CEA 19 JUILLET 2012 PAGE 30

31 IMPACT ON THE FRAGILITIY CURVES CASES 1 & 2 Finally, the fragility curves of the equipment have been determined from the 145 structural responses for the set 1 (excluded the 105 runs that do not converge or the soil shear strain is over 0.8%), and from the 157 inputs of set 2. Theses fragility curves have been approximated by the cumulative distribution function of a lognormal random variable. CEA 19 JUILLET 2012 PAGE 31

32 CONCLUSIONS OF THE STUDY Without a careful check at each step (and especially analyzing the physical meaning of incredible high acceleration values resulting from the deconvolution phase - using a methodology not adapted for such high nonlinear soil behavior) the fragility curve itself (from set 1) could be considered as acceptable, whereas this study demonstrated its unrealistic and unphysical bases. This study also illustrates the biases which can be introduced into the fragility assessment process. The median acceleration of the fragility curve and its uncertainty ( value) are different in case of using set 1 data or set 2, and finally the case 1 approach (which has to be proscribed) would be not conservative (Am-set1 systematically higher than Am-set2). To conduct the full seismic risk analysis, this fragility curve should be convolved to the seismic hazard curve: this latter step should be necessarily performed by seismologists at the control point 3 to assure the coherency of the whole process. Such SHA in depth at the outcropping bedrock site condition is clearly not the current practice in France (SHA is always performed at the free field level, including potential site effects, and most of the time SHA is given through response spectra, the time CEA series 19 JUILLET selection 2012 and generation being sensitive and complex problems). PAGE 32

33 THANKS TO ALL CONTRIBUTORS Contact & Consult THANK YOU FOR YOUR ATTENTION The work carried out under the project benefited SSRR French funding managed by the National Research Agency under the program Future Investments reference No. ANR-11-RSNR-0022].

34 Catherine BERGE-THIERRY CEA/DEN/DANS/DM2S Seismologist Seismic Risk Expert Centre de Saclay R&D project coordinator SEISM Institute Scientific Director PAGE 34 CEA 10 AVRIL OCTOBRE 2016 Commissariat à l énergie atomique et aux énergies alternatives Centre de Saclay Gif-sur-Yvette Cedex T. +33 (0) Etablissement public à caractère industriel et commercial RCS Paris B