Anda di halaman 1dari 34

1st Kashiwazaki International Symposium on

Seismic Safety of Nuclear Installations

Current Status of Seismic Margin


Assessment, Risk Evaluation, and Key
Issues
Dr. James J. Johnson
James J. Johnson and Associates
7 Essex Court
Alamo, CA 94507 USA

Dr. Nilesh C. Chokshi, Deputy Director


Division of Site and Environmental Reviews
Office of New Reactors, US NRC

24-26 November 2010


1
Topics of Discussion

• Seismic Margin Assessment Methodologies


– Deterministic SMA (Success Path) (EPRI) (termed
“Deterministic SMA” in this presentation)
– PRA (or PSA) – Based SMA (NRC Seismic Margin
Method) (termed “PSA-based SMA” in this presentation)
• Current Status
• Current Trends
• Key Issues

2
High Confidence of Low Probability of
Failure (HCLPF) Concept
• Plant and SSC Seismic Capacity can be
described by a measure termed High Confidence
of Low Probability of Failure (HCLPF)
• HCLPF can be interpreted as: “approximately” a
95% confidence of about 5% (or less) probability
of failure or a 1% probability of failure of the mean
fragility curve
• HCLPF calculations
– Probabilistically - fragility analysis (FA) method -
HCLPF capacities are defined to be 95-5 or 99
depending on treatment of uncertainties
– Deterministically – Conservative Deterministic Failure
Margin (CDFM) method – 1% probability of failure

3
Seismic Margin Assessment
Methodologies: PSA-Based SMA

• Based on Seismic PSA Methods


– Review Level Earthquake (RLE) (Probabilistic
Seismic Hazard Assessment (PSHA) – not
required)
– Systems models – Event trees/fault trees
– HCLPF values for SSCs by Fragility Functions (or
in some cases CDFM)
– Quantification of Accident Sequences by
Integration (or in some cases by min-max
method)
• Easily expanded to a seismic PRA
4
Seismic Margin Assessment
Methodologies: Deterministic SMA

• Review Level Earthquake (RLE) (Deterministic or


Probabilistic Seismic Hazard Assessment (PSHA) –
not required)
• Systems models – Success Path(s)
– US – two success paths to achieve safe shutdown
– One path mitigates small LOCA
• HCLPF values for SSCs
– Conservative Deterministic Failure Margin (CDFM)
– Fragility Analysis (FA)
• Quantification of Plant HCLPF by Min-Max Method
• Not easily expanded to a seismic PSA
5
SMA:
Summary Methodological Differences

• System Models
– Success Paths vs. Event/Fault Tree Approach
• High Confidence of Low Probability of Failure
(HCLPF) Values of SSCs
– Conservative Deterministic Failure Margin (CDFM) (or
Fragility Analysis methods) vs. Fragility Functions
• Plant HCLPF
– Min-Max of Success Paths vs. Quantification of Accident
Sequences and Plant HCLPF by Integration

6
SMA Selection of Methodology:
End Products of Interest and Future USES

• End Products
– Risk Metrics vs. Plant Seismic Capacity (HCLPF, other)
– Complete System Review (PSA-based SMA) rather than
limited review of systems in Success Paths
• Future Strategy
– Development of Risk Metrics (CDF, LERF, other) –
expansion to Seismic PSA
• Availability of internal event system models (event/fault trees from
PSA)
• Availability of PSHA
– Uses of results (risk informed decision-making, cost-
benefit analysis where benefit is change in risk metric, )

7
Fragility Model

• A = Am R U
Am is median seismic capacity
R and U represent aleatory (randomness) and
epistemic (modeling) uncertainties respectively; are
lognormal variables with median value of unity and
lognormal standard deviation R & U ;
C composite uncertainty
C = R2 + U2

• HCLPF capacity = Am exp [-1.65 (R + U)]


= Am exp [-2.33C]
8
Variability of Parameters

• Inherent Randomness, aleatory uncertainty, R


– Primarily associated with earthquake characteristics
(i.e., response spectra shape and amplification,
duration, number and phasing of peaks)
– Not considered possible to significantly reduce
randomness by additional analyses or tests
– Recent changes in procedures to assure no double
counting of aleatory uncertainty in ground motion
• Modeling Uncertainty, epistemic uncertainty, U
– Associated with lack of knowledge of model and
parameters
– Generally could be reduced by additional analyses
and tests
9
Seismic Fragility Curves and
HCLPF Value

1 Am = 0.87 g
R = 0.25
U = 0.35
PROBABILITY OF FAILURE

0.8
0.79
95% Median Mean
CONDITIONAL

Confidence
0.6

0.4

5%
0.20
Confidence
0.2

0.068
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
HCLPF 0.32g

PEAK GROUND ACCELERATION (g)


10
Conservative Deterministic Failure
Margin (CDFM) Approach
• Procedures for calculating HCLPFs in EPRI NP-
6041 use Conservative Deterministic Failure
Margin (CDFM) method
– Analyses similar to design analyses using different
factors
– CDFM procedures - understandable to design
engineers (expertise in probabilistic techniques not
required)
– Procedures in EPRI NP-6041 have been calibrated to
give similar results compared to using the FA method
– Recently revised (EPRI, 2009)

11
Modified CDFM Approach (EPRI, 2009)

SEISMIC DEMAND

Load Combination Normal Operating Loads + RLE


Ground Response Spectrum Anchor CDFM Capacity to defined response
spectrum shape without consideration of spectral
shape variability
Seismic Demand Perform seismic demand analysis in accordance
with latest version of ASCE 4 (80% NEP)
Damping Conservative estimate of median damping
Structure Model Best Estimate (median) + Uncertainty Variation in
Frequency
Soil-Structure Interaction Best Estimate (median) + Parameter Variation
In-Structure (Floor) Response Spectra SSC response defined at 80%NEP by combination
(ISRS) Generation for Subsystem Input of subsystem input definition (ISRS) and
and Response subsystem response analysis procedures and
parameter values (frequency shifting vs. peak
broadening and enveloping to account for
uncertainties, conservative estimate of median
damping, etc.)

12
Modified CDFM Approach (EPRI, 2009)

SEISMIC CAPACITY

Material Strength Code specified minimum Strength or 95%


exceedance actual strength if test data are
available
Static Strength Equations Code ultimate strength (ACI), maximum strength
(AISC), Service Level D (ASME), or functional
limits. If test data are available to demonstrate
excessive conservatism of code equation, then
use 84% exceedance for strength equation.
Inelastic Energy Absorption For non-brittle failure modes and linear analysis,
use appropriate inelastic energy absorption
factor from ASCE 43-05 to account for ductility
benefits, or perform nonlinear analyses and go to
95% exceedance ductility levels.

13
Steps in Seismic Margin Assessment
Methods

• Selection of Review Level Earthquake (RLE)


• Selection of Assessment Team
• Pre-walkdown and Pre-Calculation Preparation
• Selection of Success Paths and Elements or
modify ET/FTs of internal event PSA for PSA-
based SMA
• Ruggedness Screening
• Seismic Capability Walkdown
• HCLPF or Fragility Function Calculation
• Enhancements
• Documentation
14
Review Level Earthquake (RLE)

15
Review Level Earthquake (RLE)

• Purposes of RLE
– Defines ground motion for which HCLPF capacities
of SSCs are evaluated
– Defines initial screening levels – SSCs screened out
because HCLPF capacities greater than RLE
• RLE – generally larger than the Design Basis
Earthquake (SL-2, S2 (or Ss), SSE)
– Sufficiently larger so that SSCs are challenged –
plant and SSC HCLPFs determined
– Weak links identified

16
Review Level Earthquake (RLE)

• RLE may be based on Site Specific PSHA


results – current trend for existing NPPs
– Mean 1e-04 to 1e-05 UHS
– PGV, PGD, duration, time histories
– Energy content and damage potential
– Deaggregated contributors to UHS may be treated
separately
• RLE is not acceptance criteria and is not a new
design earthquake ground motion

17
Review Level Earthquake (RLE):
History of RLE

• Ground response spectra - Broad-banded


frequency content, generally site independent
(rock, soil), anchored to PGAs appropriate to
screening levels (0.3g or higher)
• Weakly correlated to SHA results (DSHA,
PSHA)
• US NRC IPEEE Program
– US NRC NUREG/CR-0098 median rock or soil
spectra anchored to assigned PGA values (SSE,
0.3g, 0.5g)

18
Review Level Earthquake (RLE):
Current and Future Definitions of RLE

• New Nuclear Power Plants (US +)


– All new NPPs required to perform a seismic PSA
after site is identified and before fuel load
– All new NPPs required to demonstrate seismic
margin of 1.67 times the DBE
• RLE is 1.67 times CSDRS or GMRS
• PSA-based SMA to be applied
• Acceptance criteria: Nuclear Island (licensed under
10CFR52) (HCLPF >1.67 x CSDRS); Site specific/plant
specific elements HCLPF > 1.67 x GMRS

19
Review Level Earthquake (RLE):
Current and Future Definitions of RLE

• New Nuclear Power Plants (outside US)


– European Utility Requirements document - All new
NPPs required to demonstrate margin of at least 1.4
times DBE (RLE = 1.4 times DBE)
• Specifically excludes performance of a seismic PSA
• Likely means performance of a deterministic SMA
– Japan – NSC requires consideration of “residual
risk” – methodology to do so should be clear after
this conference

20
Review Level Earthquake (RLE):
Current and Future Definitions of RLE

• Existing nuclear power plant evaluations


– US +
• RLE based on PSHA results – mean 1e-04 to 1e-05 UHS,
etc.
• Absence of PSHA – based on generic site independent
ground response spectral shapes
– Europe +
• RLE based on hazard assessments
• Based on generic site independent ground response
spectral shapes
– Japan (to be verified)
• Based on results of new seismic hazard assessments

21
Uniform Hazard Spectra

22
Deaggregated Seismic Hazard

23
Seismic Demand for Review Level
Earthquake (RLE)

24
Pre-Walkdown Preparation:
Assemble Plant Documentation

• FSAR, drawings
• Site characteristics
• Design seismic response analysis
• Design criteria and reports
• Equipment stress reports or qualification test reports
• Equipment anchorage design and installation data

25
Pre-Walkdown Preparation:
Generate Seismic Demand for RLE

• Extremely important element to generate reasonable


SMA results
– Lessons learned from US IPEEE (NUREG-1742)
– Typically, very significant conservatism in seismic design
analyses
‒ Structure response (loads, stresses, ), in-structure
response spectra (ISRS)
• Methods
– Scale seismic design calculated responses
– Re-analysis by deterministic or probabilistic methods

26
Seismic Response as Input to SMA

• HCLPF Values developed by CDFM approach


– Seismic Demand to be at 80% NEP (ASCE 4-
2010, ASCE 43-05)
• HCLPF Values developed by FA approach
– Seismic Demand to be at 50% NEP (Median)
– Identical approach to SPSA Fragility Function
development
• All seismic response generation methods
– RLE or de-aggregated RLE
– Construct new models as necessary
– Treat variability in ground motion, soil and structure
properties (probabilistically - SMACS – US NRC SSMRP)
• 27
Seismic Response as Input to SMA:
Current Practice and Future Trends
• Soil sites and RLE defined by low or high frequency
ground motion
– Re-analysis of a sample or all safety significant
structures will be performed (scaling of in-structure
response is difficult)
• Rock sites and seismic hazard defined by low
frequency ground motion – scaling of seismic design
responses still viable approach
• Rock sites and seismic hazard defined by high
frequency ground motion (> 10 Hz.) – re-analysis
taking into account SSI for coherent and incoherent
ground motions
• Re-analysis by probabilistic analysis techniques
preferred 28
Screening of SSCs

29
Screening of SSCs: Apply Response and
Capacity Filters
• EPRI NP-6041 Screening Tables Re-Interpreted
for In-Structure Response Spectra (ISRS) (EPRI,
2009)
– Applied to ISRS – Screening levels increased by 1.5
factor
• Seismic Response
– Calculated by probabilistic or deterministic methods
targeted to 80% NEP conditional on RLE (assumed to
be defined by the mean ground motion)
– Deterministic – best estimate procedures apply,
including enveloping of soil cases (three as a minimum),
peak clipping, peak shifting instead of peak broadening,
etc.

30
Screening of SSCs:
Apply Response and Capacity Filters
• Screen components – assign HCLPF values
based on screening values and verification of
caveats and in-plant conditions
– Revisions based on recent generation SMA and SPSA
results (EPRI, 2009)
– Generic capacity information when shown applicable –
caveats, anchorage
– Seismic design criteria (used during design process)
– Mechanical components/piping/others
– All screened-out SSCs – verify caveats met, anchorage
capacity, and no vulnerabilities in the field - plant
walkdown
• Screening Tables for Systems and Components revised to
allow seismic demand to be calculated and compared at
support locations (ISRS) (EPRI, 2009)
31
SMA Current Status and Trends:
Summary

• Two methodologies
– Deterministic SMA and PSA-based SMA – both
being applied
– Selection dependent on end products desired
• Review Level Earthquake (RLE)
– Site specific seismic hazard where appropriate
– Standard designs – site independent definitions
– Can be treated in deaggregated form

32
SMA Current Status and Trends:
Summary

• Seismic demand
– Re-analysis by probabilistic or deterministic
approaches being performed
• Soil sites – all RLE definitions
• Rock sites – high frequency ground motion
• Screening of SSCs
– Screening tables of EPRI NP-6041 re-interpreted for
in-structure response spectra (ISRS) at support
locations of equipment and components – scaled up
by factor of 1.5
– Caveats and anchorage checks remain in place
• Configuration Control
33
1st Kashiwazaki International Symposium on
Seismic Safety of Nuclear Installations

Current Status of Seismic Margin


Assessment, Risk Evaluation, and Key
Issues
Dr. James J. Johnson
James J. Johnson and Associates
7 Essex Court
Alamo, CA 94507 USA

Dr. Nilesh C. Chokshi, Deputy Director


Division of Site and Environmental Reviews
Office of New Reactors, US NRC

24-26 November 2010


34

Anda mungkin juga menyukai