Title: Issues, Guidance and Recommendations for: Tsunami Risk Assessment and Reduction for Nuclear Facilities
1Issues, Guidance and Recommendations
forTsunami Risk Assessment and Reduction for
Nuclear Facilities
- Dr. Robert T. Sewell
- Dr. George Pararas-Carayannis
- Fifth International Tsunami Symposium ofThe
Tsunami Society International - European Commission Joint Research Centre
- Ispra - Italy, 5 September 2012
2Topics
- Definition and Importance of External Hazards
(Including Tsunamis) for NPPs - External Hazards Assessment for NPPs
- Special Considerations Unique Challenges
- Applicable Existing Safety Guidance for NPPs
- Screening and Evaluation Approaches and Bases
- Deterministic Assessment for External Hazards
- Probabilistic Assessment for External Hazards
- Recommendations and Discussion
3External Hazards IAEA Definition
- External hazards originate from sources located
outside of the site of the nuclear power plant. - Examples of external hazards include
- Seismic hazards
- High winds and wind-induced missiles
- External floods including tsunamis
- Other severe weather phenomena (e.g., snow, ice)
- Off-site transportation accidents
- Off-site explosions
- Releases of toxic chemicals from off-site storage
facilities - External fires (e.g. fires affecting the site and
originating from nearby forest fires)
4Importance External Hazards
- External Hazards can often be the dominant
contributor to the risk of nuclear plant failure
(e.g., core damage, or significant radiological
release) - For example, seismic events (earthquakes) are a
particularly severe challenge to NPPs, and
typically cannot be ruled at any location for
return periods of interest (i.e., up to 10
million years) - Recent experience at the Fukushima-Daiichi NPP
has demonstrated tsunamis to be an external
hazard requiring improved awareness and risk
management
5Special Considerations and Unique Challenges for
External Hazards
- High Severity Common Cause
- Scenarios have the potential to adversely affect
many components or, often, the entire plant - As in the Fukushima catastrophe
- High Uncertainty
- Experience data is often lacking uncertainties
must be systematically quantified. - Broad and Diverse Phenomena
- Covers several disciplines and areas of expertise
- Leads to temptation to minimize or ignore the
threat (e.g., it is common that analysts or
decision makers prematurely eliminate or
screen-out events outside of their immediate
expertise)
6Special/Unique Considerations (contd)
- Spatial relationships (e.g., Locations and
Proximities) usually are of particular Importance
in the assessment of external hazards.
(Knowledge of threats outside of the site
boundary of an NPP needs to be obtained.) - For tsunamis, the source event can be caused by
substantially different phenomena (e.g.,
earthquakes, landslides, meteor impacts, large
releases of gases) - As events of very long return period must be
considered (i.e., 10 million years median
hazard, or 1 million years mean hazard),
analysis (with associated uncertainties) is
typically required to fill-in for data gaps
7Example External Hazard (Tsunami) at NPP
(Fukushima Daiichi)
8Example External Hazard (Seismic, No Tsunami) at
NPP (Kashiwazaki)
9Applicable Safety Guidance External Hazards
(incl. Tsunamis)
- IAEA Safety Guides
- Several applicable guides to overview
- ASME/ANS Standard
- Update to New ASME / ANSI Standard for Use in
USNRC Programs in Response to Fukushima - US NRC
- Best Practices from IPEEE Implementation
Experience
10Applicable IAEA Safety Guides Periodic Safety
Review Process
- NS-G-2.10
- Logistics of Safety Assessment in the Context of
the Periodic Review Cycle - Relevance of External Hazards to the PSR
11Applicable IAEA Safety Guides General Safety
Assessment
- NS-G-1.2
- Guidance for Safety Assessment for Design and
Design Modifications - Relevance of External Hazards in a Design or
Retrofit Safety Assessment - Overviews Deterministic and Probabilistic
Approaches for Safety Assessment
12Applicable IAEA Safety Guides Deterministic
Safety Assessment
- SSG-2
- Detailed Guidance on the Deterministic Safety
Assessment Approaches - Treatment of External Hazards in a Deterministic
Safety Assessment
13Applicable IAEA Safety Guides Probabilistic
Safety Assessment (L1)
- SSG-3
- Detailed Guidance on the Probabilistic Safety
Assessment (PSA) Approaches for a Level-1
Analysis - Treatment of External Hazards in a Level-1 PSA
14Applicable IAEA Safety Guides Probabilistic
Safety Assessment (L2)
- SSG-4
- Detailed Guidance on the Probabilistic Safety
Assessment (PSA) Approaches for a Level-2
Analysis - Treatment of External Hazards in a Level-2 PSA
15Applicable IAEA Safety Guides External Hazards
(Except Seismic)
- NS-G-1.5
- Guidance on the Safety Assessment of External
Hazards, Excluding Seismic
16Applicable IAEA Safety Guides Seismic Hazard
Analysis
- NS-G-3.3
- Guidance on Procedures for Seismic Hazard Analysis
17Applicable IAEA Safety Guides Flood Hazards
- NS-G-3.5
- Guidance on Procedures for Flood Hazard Analysis
- Includes Tsunamis, Seiche, Storm Surges, in
Addition to River Flooding
18Applicable IAEA Safety Guides Geotechnical
Hazards
- NS-G-3.6
- Guidance on Geotechnical Considerations in NPP
Siting and Design - Relevant to Identifying and Evaluating External
Hazards Associated with Landsliding, Settlements,
Ground Collapses, Etc.
19Other Guidance / Standards
- ASME/ANS RA-S-2008
- Freely Available on Internet
- http//www.engineeringcodes.com/download/ASME_RA-
S-2008_cont_0670.pdf - Covers Internal Hazards and External Hazards
- This document is current being updated for use in
a US External Hazards Safety Program in Response
to the Fukushima Event - http//cstools.asme.org/csconnect/pdf/CommitteeFi
les/27524.pdf
20Other Guidance / Standards
- NUREG-1407
- Not much guidance for tsunamis
- Implementation Experience in the IPEEE Program
Has Been Important to Defining and Improving the
State of the Art - Event Categories and Screening Criteria in the
Earlier Guidance of NUREG/CR-2300 Has Become
Substantially Obsolete
21External Hazards Safety Assessment Basic
Approach (incl. Tsunamis)
- Comprehensively and Conservatively Identify and
List All General Types of External Hazards - This is not a screening step every general
hazard should be included - See Spreadsheet for typical list for NPPs
- Identify and Characterize all Plant-Unique
External Hazards - Perform an initial review of plant information
- Perform a GIS survey of the plant zone, vicinity,
region - Perform an initial walkdown of the plant and
vicinity - This is not a screening step every
plant-specific hazard should be included
22- Research and Collect the Necessary Plant,
Engineering and Scientific Data Sufficient to
Perform a Conservative Qualitative Screening
Analysis of Each External Hazard - Past experience reveals that most analysts tend
to be overly aggressive in this Qualitative
Screening phase. Some common problems - Insufficient research is done to rigorously
defend the reasons for screening out a hazard - e.g., Scientific literature and the relevant body
of scientists or experts having expertise with
the particular hazard are not fully consulted - Analysts tend to baseline their judgment on the
types of extreme events observed during a
generation or two (what they have heard about
anecdotally or experienced themselves in their
lifetimes), rather than on sufficiently rare
phenomena - Insufficient attention to uncertainties and
diverse viewpoints within the Informed Technical
Community (ITC)
23- For Each Hazard Not Qualitatively Screened-Out
- Perform Additional Walkdowns, Research and Data
Collection Sufficient to Perform a Conservative
(Simplified) Quantitative Screening Analysis - Hazard-based (or location-based) screening
- Fragility-based (or vulnerability-based)
screening - Exposure-based (or Conditional Core-Damage
Probability CCDP-based) screening - Simplified Risk-Based Screening (combining two or
more of the preceding) - The effective screening threshold in each case
should be no less stringent than - Mean annual frequency of 10-6 (1,000,000-yr RP)
- Median annual frequency of 10-7 (10,000,000-yr RP)
24- For Each Hazard Not Screened-Out with a
Simplified Conservative Quantitative Analysis - Perform Additional Walkdowns, Research and Data
Collection Sufficient to Perform a Detailed
Analysis - Detailed hazard analysis
- Detailed fragility analyses
- Detailed CCDP analysis or hazard-specific plant
logic analysis and/or - Detailed risk analysis
- Always Perform a Detailed Analysis for
Earthquakes at Every NPP (and Tsunamis for
coastal NPPs) - Report the Total Risk from External Hazards
- All contributions for hazards analyzed inSteps 4
through 6
25Deterministic Approach
- Conceive and Apply Maximum Credible Scenarios
- Often conceived as (or believed to be) a worst
case - Significant disadvantage of this approach is that
the event frequency (and its level of
conservatism) are not known, and can be
significantly misunderstood and misrepresented - Use of Conservative Assumptions and Relationships
- For example, equivalent to a mean one-sigma
(or 84th-Fractile) level - Expert Uncertainty Is Often Not Formally
Addressed - Significant disadvantage of this approach
- Could be somewhat mitigated by using the more
conservative / stringent expert interpretations
26Framework of Probabilistic Approach
- Conceptual Elements of Risk for External Hazards
- Risk Evaluation
- Direct (statistical analysis of data) and derived
(analytical modeling of phenomena) approaches - Random Process Modeling
- Rate, occurrence model, and time
- Aging, forecasting, time dependencies
- Uncertainty Analysis Framework
- Aleatory variability and epistemic uncertainty
27Conceptual Elements of Riskfor External Hazards
Hazard
Vulnerability
Risk
Location /Proximity
Exposure
28Generalized Direct Formulation of Hazard (or Risk)
TH Hazard State e.g., X?x TC Component
Damage State e.g., RWST Fails TS
System-Level Risk State e.g., CCW System
Fails TP Plant-Level Risk State i.e.,
Plant Fails
29Generalized Derived Formulation of Probabilistic
Hazard (or Risk)
30Risk Evaluation (for a Failure State)
31Common Hazard Fragility Formulation
Hazard Curve
Fragility Curve (Component, System, or Plant)
32Random Process Modeling
- Rates, ?, are long term values (events per
year) - Assumes that the current snapshot of processes
continue indefinitely (e.g., millions to hundreds
of millions of years, not only 50 years) - Does not consider changes in processes themselves
that can occur over these long time frames - A stochastic occurrence model (e.g., Poisson,
Renewal, Time-Predictable, etc.) is needed to
evaluate probabilities for each category of
external event - In general, different occurrence models may be
appropriate for different external hazards
33Poisson Occurrence Model (Common)
34Aleatory and Epistemic Variations
- Aleatory Variation (inherent randomness) Derives
from differences, even under seemingly identical
conditions, in the way nature behaves and is
manifest. - Usually addressed by classical methods of
probability and statistics - Epistemic Variation (professional uncertainty)
Derives from data limitations and differences in
models, judgments, and use of data by credible
experts - Usually addressed by methods of subjective
probability and Bayesian analysis
35Accounting for Aleatory and Epistemic Variations
- Develop an epistemic logic tree (E-LT) model to
describe possible epistemic variations - Requires development of multiple, alternative
aleatory models and their subjective
probabilities - Simulate epistemic scenarios from the E-LT
- For each epistemic scenario, develop an aleatory
logic tree (A-LT) model to describe possible
aleatory variations - Requires a behavioral model, its basic random
variables and their probability distributions
36Uncertainty Analysis Framework
EpistemicModel ofHazard
Aleatory Model ofScenarios
Epistemic Model ofOutcomes
Aleatory Model ofOutcomes
UncertainRiskScenarios
HazardAnalysis
Vulnerability Analysis
RiskResults
37Elements of Tsunami Risk Assessment for NPPs
- Seismic Probabilistic Safety Assessment
(PSA)a.k.a. Seismic Probabilistic Risk
Assessment (PRA) - Fragility analysis approach for capacity
assessment of components and structures - Full event-tree / fault-tree quantification of
plant systems response from component responses - Full treatment of random failures and human
errors - Point-estimate or full uncertainty analysis
- Tsunami core-damage frequency (CDF)
- Tsunami large-early release frequency (LERF)
38Tsunami Hazard fromSubmarine Landsliding
- l(W?w) ? nLS ?? PTypeSize ? PVelocity ?
PTrajectory ?PW?w Type, Size, Velocity,
Trajectory - (Lack of Site-Specific Empirical Data Suggests
the Need forModeling and Paleo-Tsunamic Data,
Which Are Readily Accommodated Here)
39Tsunami Hazard Scenario
- One scenario (numerical simulation) is
illustrated here - Multiple such scenarios and their likelihoods can
be estimated to evaluate the probabilistic hazard - Latin Hypercube simulation considering aleatory
and epistemic variations is an efficient approach
for this type of hazard study
40Location
Carmel AreaNear Pt. Lobos(CA State Reserve Area)
Latitude36.51965 N
Longitude121.92126 W
41Location
42Location
Site
43About 160 Miles Further South
44LandslideHazard
45Submarine Landslide Scenario Animation
46T0.00 min.
47T0.50 min.
48T2.00 min.
49T4.00 min.
50Tsunami Scenario Animation
51T0.0 min.
52T0.5 min.
53T2.0 min.
54T4.0 min.
55T9.0 min.
56T15.0 min.
57T20.0 min.
58(No Transcript)
59Recommendations (Discussion)
- In the US, there exists a PMEL-USGS-NRC(SITAG)
joint program on tsunami hazard for NPPs, which
is a step in the right direction, provided that
the program embraces an epistemic assessment
similar to the rigorous Level-4 (or, in some
cases, Level-3) guidelines of the Senior Seismic
Hazard Analysis Committee (SSHAC) that have set a
precedence and standard for probabilistic hazard
assessment at NPP sites. - i.e., The diverse viewpoints within the entire
Informed Technical Community should be addressed,
requiring the participation of additional experts
60Recommendations (Discussion)
- Beyond-design margin for tsunamis needs to be
better understood. - A 10,000-yr design basis (as has been applied as
a standard for NPP design for other external
hazards) has not been systematically applied for
the tsunami threat. - Even the 10,000-yr design level may not be
sufficient if there exist "cliff edge" effects
with the tsunami margin (i.e., if failure is
incipient with minimal exceedance of the design
level).
61Modern Tsunami Evaluation Methods
Recommendations (Discussion)
- Probabilistic safety/risk assessments (PSAs/PRAs)
is the modern approach for safety assessment of
NPPs. Correspondingly, the probabilistic nature
of the tsunami hazard and tsunami fragility of
NPPs need to be better evaluated or understood. - The major knowledge base developed over the past
30 years in improvement of methodology for
seismic risk assessment of NPPs should be
leveraged for rapid improvement in tsunami risk
assessment of NPPs
62Recommendations (Discussion)
- Consistent with screening criteria for external
hazards (e.g., in an external events PSA for an
NPP), probabilistic tsunami hazard studies should
cover hazard values down to 1E-6/yr mean
exceedance frequency and 1E-7/yr median
exceedance frequency. - This can be expected to represent a significant
challenge for some tsunami experts, who may have
focused in the past on characterizing the hazard
for much shorter return periods (e.g., typical of
inundation studies or studies for public warning
and evacuation zoning, rather than extremely rare
events).
63Recommendations (Discussion)
- Development and application of procedural
methodology for epistemic assessment should be
undertaken in a manner similar to a SSHAC Level-4
(or Level-3) approach for tsunami hazard, with
tsunami source, wave propagation and run-up being
analogous (respectively) to seismic source
modeling, ground-motion modeling, and site
response analysis.
64Recommendations (Discussion)
- Implementation of a robust and rigorous
probabilistic tsunami hazard assessment (PTHA)
methodology for the aleatory assessment is also
needed. Such a PTHA methodology was developed by
the first author (in concert with Dr. C. Mader)
in 2002 for an LNG facility, and adapted the
well-established PSHA (probabilistic seismic
hazard analysis) methods originated by Cornell
(1968), as well as generalized the probabilistic
treatment of tsunamigenic sources for all tsunami
causes (not just earthquakes). - Based on the authors experience, such a
methodology can be successfully applied and
improved. Although similar approaches have since
been used by others, it may still take
significant time and effort to fully acquaint the
ITC of tsunami scientists with these methods.
65Recommendations (Discussion)
- Sufficient coverage and undertaking of the
marine, geophysical, geological, paleo, etc.,
studies is needed to properly understand and
characterize the tsunami threat, bathymetry, etc - Every coastal NPP should perform these according
to a rigorous consistent standard (which itself
needs to be developed) - Adaption of physical models (finite elements, 3D
slip surface having minimum factor of safety,
etc.) to define the extent and behavior of
submarine landslides in consideration of physical
geological/geotechnical properties, rather than
reliance solely on geometric models that are
related primarily to marine geomorphology, is
needed for describing the landslide-tsunami
threat to NPPs.
66Recommendations (Discussion)
- A sufficient description/characterization of the
full scope of possible tsunami threats
including characterization for both direct
effects and collateral effects and damages from
tsunamis is needed. - Development of local tsunami warning systems for
NPPs is needed. - Whereas this step may seem a severe measure to
some, oil companies have sought such measures for
protection of LNG plants and other facilities.
Also, local tsunami warning systems are
considered for warning communities of silent
tsunamis from causes unrelated to earthquakes
(e.g., Skagway, AK). It can be easily argued that
the need to protect NPPs in areas of significant
tsunami hazard is at least as great as for
petroleum facilities and for communities having
high visibility as vacation destinations.
67Recommendations (Discussion)
- A reproducible set of policies and procedures
(consistent with overall safety management)
pertaining to how to conduct site-specific
tsunami hazard studies, tsunami risk studies, and
reviews of such tsunami studies for NPPs should
be developed. - Procedures are needed including comprehensive
walkdown methods and field studies of NPPs to
suitably understand and assess the deterministic
and probabilistic vulnerability, design, and risk
evaluation, and to ensure adequate
fragility/resistance of NPPs to tsunamis
(including impacts on structures, systems,
components and operators).