PRA Technology: Historical Perspectives and Challenges for the

Assessing the Seismic Risk of Small Modular Reactors

By:

Karl Fleming

KNF Consulting Services LLC

[email protected]

Presented to:

CRA’s 7th Risk Forum Statford-on-Avon UK October 5 and 6 2016

Discussion Topics

Seismic PRA basics

Technical challenges faced

NuScale SMR basic safety features

Technical approach to SMR seismic PRA Risk metrics for multi-module accidents

Seismic PRA modeling

Seismic fragility correlation

Example results

Path forward

NuScale Multi-Module Seismic PRA 2

Seismic Probabilistic Risk Assessment

Component-Fragility

Evaluation

Seismic Hazard Analysis

Seismic Motion Parameter

Fre

qu

en

cy o

f E

xce

ed

an

ce

Event Trees Fault Trees

Containment Analysis

P i

P i

P i

Systems Analysis

Seismic Motion Parameter

Release Frequency Consequence Analysis

Frequency Damage

Fre

qu

en

cy o

f E

xce

ed

an

ce

1 3 2

Release

Category

Pro

babili

ty D

ensity

•Weather Data

•Atmospheric

Dispersion

•Population

•Evacuation

•Health Effects

•Property Damage

Risk

Conditio

nal P

robabili

ty

of

Fa

ilure

Key Elements of Seismic PSA

Seismic Hazard Analysis: to develop frequencies of occurrence of different levels of ground motion (e.g., peak ground acceleration) at the site.

Seismic Fragility Evaluation: to estimate the conditional probability of failure of important structures and equipment whose failure may lead to unacceptable damage to the plant (e.g., core damage).

Systems/Accident Sequence Analysis: modeling of the various combinations of structural and equipment failures that could initiate and propagate a seismic core damage sequence including combinations of seismic and non-seismically induced failures and unavailabilities.

Technical Issues for SMRs

Previous seismic PRAs at multi-unit sites performed on one reactor at a time

Risk metrics such as core damage frequency fail to capture contributions from multi-unit (multi-module) accidents

Importance of seismic fragility correlation is magnified

Seismic risk expected to be reduced and determined by less likely extreme earthquakes with larger uncertainties


Seismic Fragility Correlation

Problem occurs when there is a single fragility curve for a set of identical components or structures.

Sources of fragility correlation

Seismic intensity variability correlation

Soil and structure amplification correlation

Component capacity correlation

Problem occurs in single reactor PRAs but is amplified for SMRs

Should the probability of multiple seismic failures be calculated as independent or common cause failures?

What to do if there are many identical components, say 12 modules?

How does correlation affect the calculation of the frequency of core damage involving 1, 2, 3, …, 12 reactor modules?


Correlation Involves Two Types of Dependency

NUSCALE SMR BASIC SAFETY FEATURES


NuScale Reactor Module


NuScale Module Features


NuScale SMR Plant Parameters


NuScale SMR Containment


Passive Decay Heat Removal via Steam Generators


Passive Decay Heat Removal via Containment


NuScale Long Term Station Blackout Capabilities


Technical Approach


Technical Approach Summary

Selection of seismic intensity bins and cut-off values

Selection of seismically induced initiating events and accident sequences

Treatment of multi-unit accidents and end states

Treatment of seismic fragility correlation

Within a single unit seismic PRA

Within a multi-unit seismic PRA

Quantification of mean risk metrics, uncertainties, and sensitivities


Candidate NuScale Multi-Module Risk Metrics Level 1 PRA

TCDF = Total frequency of a core damage involving one or more modules per MM plant year

PCDF(n) = frequency of core damage involving exactly n modules concurrently per MM plant year


12

1

)(n

nPCDFTCDF

Candidate NuScale Multi-Module Risk Metrics Level 2 PRA

TSRF = Total frequency of a significant release from one or more modules per MM plant year

PSRF(n) = frequency of a significant release from exactly n modules concurrently per MM plant year

Significant defined as a release that approaches or exceeds the EPZ selection criterion


12

1

)(n

nPSRFTSRF

Seismic Fragility Correlation Approach

Single Module Seismic PRA

Identical or similar components within a module subject to correlation

Multi-Module Seismic PRA

Second order correlation for components within a module

Module to Module correlation

Assumed that multi-module dependencies will be explicitly modeled (e.g. reactor building failure)


Introduce Seismic “Common Cause” Model for Correlation


Basic Event Probabilities


)()1( GfGP jjGkj

)(GfGP jjGCORj

Where:

kj GP Probability of independent seismic induced failure of Component Gk at seismic intensity j

CORj GP Probability of seismic induced failure of all components in correlation group G at seismic intensity j due to seismic fragility correlation

jG Seismic correlation split fraction, fraction of seismic events with common cause failure of similar components due to seismic fragility correlation

)(Gf j Seismic fragility of components in Group G at seismic intensity j

Seismic Correlation Between Two Components


Level 1 MM Seismic PRA Model


N

j

jnjeffj

n

j

n

jeffjjS pFFFn

ShnPCDF1

121

121)(

Where:

)(nPCDFS Frequency of Seismically Induced Multiple Module Core

Damage involving exactly n, modules, per MM plant year

effj Effective seismic correlation split fraction for the plant level fragility at seismic intensity j; conditional probability that two or modules will fail due to seismic correlation given seismic failure of one module at seismic intensity j; computed using Equation (3.13)

jnp Conditional probability that a seismic event causes core damage on exactly n reactor modules due to seismically correlated failures; pj0 = pj1 = 0 because correlated failure must involve at least two failures

Sensitivity Studies

Used most recent (2014) mean hazard curve for Seabrook – bounds all east coast sites

Used fragility and alpha estimates for a “reactor vessel” to simulate a NuScale reactor module

Different models for correlated failure distributions


Seabrook Hazard Curves


Fragility Estimation for NPM Variable Median BetaR BetaU Notes

Strength

Formula for shear failure 2.00 0.00 0.20 EPRI 103959

Material Strength (reinf. Steel) 1.20 0.00 0.10

Inelastic Energy Absorption 1.80 0.10 0.20 ASCE 4-05

Spectral Shape 1.38 0.20 0.19

Damping 1.00 0.00 0.14

Modeling 1.00 0.00 0.15

Modal Combination 1.00 0.00 0.00

Earthquake Component

Combination

1.00 0.05 0.00

Soil structure interaction 1.00 0.00 0.10 Site specific, median could be

1.2

Ground Motion Incoherence 1.00 0.00 0.05 Depends on plan dimensions;

ignore

Horizontal Direction Peak

Response

1.00 0.13 0.00

Total Safety Factor 5.96 0.26 0.43

Ground acceleration Capacity, g 2.98 0.26 0.43

HCLPF capacity, g

0.96

Fragility and Alpha Based on Ravi’s Estimate for a NuScale Module


Fragilities of Modules

Correlated Failure Distributions for Sensitivity


Plant Level Core Damage Frequency PCDF(n) – No Correlation


Plant Level Core Damage Frequency PCDF(n) – No Correlation vs. Uniform


Plant Level Core Damage Frequency PCDF(n) – No Correlation vs. Triangle 2


Plant Level Core Damage Frequency PCDF(n) – No Correlation vs. Four Correlation Models


Plant Level Core Damage Frequency PCDF(n) – No Correlation vs. Four Correlation Models Frequency of 4 or More Combined


Plant Level Core Damage Frequency PCDF(n) – No Correlation vs. Four Correlation Models (IAEA Example with increased capacity)


Summary

Basic safety features of NuScale SMR described

Technical approach to multi-module (MM) seismic PRA defined

Risk metrics and seismic correlation parameters proposed

Methods for quantifying seismic correlation within and among modules developed

Sensitivity studies provide insights on the relative contributions of independent and correlated combinations of seismic failures

Additional refinement is expected when seismic design and single module seismic PRA is more advanced


Documents

PRA Technology: Historical Perspectives and Challenges for the