6/14/10

Seismic  Safety  of    

Nuclear  Power  Plants  
Bozidar  Stojadinovic,  Professor  
CEE  Department,  UC  Berkeley  
[email protected]  

What  is  Safety?  

•  The  state  of  being  safe  
•  The  state  of  being  
protected  from  
consequences  of  
undesirable  events:  
–  Accidents  
–  Errors  
–  Failures  

1  
 6/14/10  

Nuclear  Power  Plants  

•  Complex  energy  conversion  
machines  
•  Engineered:    
–  Not  naturally  occurring  
–  Made  and  operated  by  humans  
•  Introduce  a  layer  of  hazard  not  
present  before  we  invented  
them:  
–  RadiaVon  hazard  
–  Explosion  hazard  
–  Other  hazards  (environmental,  
occupaVonal…)    

Safety  Goal  
•  Protect  the  public  from  empirically  detectable  
harm:  
–  No  more:  protecVng  against  what  is  not  
empirically  detectable  is  near-­‐impossible  
–  No  less:  causing  a  detectable  change  in  the  hazard  
environment  is  not  acceptable    

2  
 6/14/10  

Safety  Goal  
•  Focus  on  radiaVon  dose  at  exclusion  area  
boundary:  

An#cipated  opera#onal  
       occurrences  

Design  basis  events  

Beyond  design  basis  

                   events  

Hazard  and  Risk  

Risk  =  P(event  occurring)  x  (Impact  of  event  occurrence)  

3  
 6/14/10  

Seismic  Safety  of  NPPs  

•  Earthquakes   •  Effect  of  earthquake  
•  CharacterisVcs  of   ground  moVon  on  real  
ground  moVons   structures  
•  Effect  of  ground  moVon   •  Demand,  damage  and  
on  simple  structures   decisions:  performance  
•  Measures  of  earthquake   •  Probability-­‐based    risk-­‐
hazard   informed  technology  –
neutral  design  
framework    

Cause  of  Earthquakes  

In Japan, people
believed that a
catfish that lived
under the land
caused
earthquakes every
time it wiggled.

The people in the

picture are striking
the catfish to stop it
from shaking the
earth.

4  
 6/14/10  

Tectonic  Plates  

View  of  a  Geo-­‐ScienVst    

•  An  earthquake  is  the  result  of  the  sudden  
release  of  energy  in  the  Earth's  crust  that  
creates  seismic  waves  

5  
 6/14/10  

Earthquake  Faults    
•  Earthquake  faults  
occur  at  the  
edges  of  tectonic  
plates  (where  
they  slip  by  each  
other)  
•  These  are  
complex  rock  and  
soil  fracture  
phenomena  

How  Do  Faults  Slip?  

•  An  earthquake  is  caused  by  a  build  up  of  strain  on  
the  edges  of  the  tectonic  plates.    
•  The  strain  becomes  so  great  that  rocks  give  way  and  
slipping  occur  along  the  fault.  

6  
 6/14/10  

Seismic  Wave  PropagaVon    

•  Waves  reflect  from  and  
refract  through  crust  
layers  of  different  
density  

hgp://www.uwgb.edu/DutchS/EarthSC202Notes/quakes.htm  

Wave  PropagaVon  Through  Soil  

•  Body  wave:  
–  Pressure  
–  Shear  
•  Surface  waves:  
–  Lowe  
–  Rayleigh  
–  They  are  slowest,  but  
they  do  most  of  the  
damage  when  they  
arrive  at  the  site  of  the  
structure  

7  
 6/14/10  

Ground  MoVon  at  the  Site  

•  Depends  on:  
–  Energy  released  at  the  
source  (fault)  
–  Path,  distance  the  waves  
travel  to  the  site  
–  Local  condiVons  at  the  
site:  
•  Soil  
•  Focusing  
•  Other  structures  

Ground  MoVon  at  the  Site    

of  an  NPP  Structure  

Surface  
waves  

16  

8  
 6/14/10  

Types  of    
Ground  MoVon  
•  Great  variety!  

Principal  Types  of  Ground  MoVon  

•  Delineated  by:   •  Near-­‐Fault  
–  Magnitude     –  Pronounced  direcVvity  
–  Distance  between  the   •  Perpendicular  
hypocenter  and  the  site   •  Parallel  
–  Local  soil  condiVons     •  Near-­‐Field  
–  Strong  pulse,  fling  
–  High-­‐frequency  content  
Same    
distance  
•  Far-­‐Field  
–  Longer  duraVon  
–  Moderate  and  low  
frequency  content  

9  
 6/14/10  

Magnitude:  Measure    
of  Energy  Release  
•  Many  versions:  

ObservaVons  
•  Intensity  of  
ground  shaking  
decreases  with  
increasing  
distance  from  the  
epicenter  
 (1994  M6.7  
Northridge  
Earthquake,  USGS  
Shake  Map)  

10  
 6/14/10  

How  Ground  MoVons  Affect  

Structures?    

Dynamic  Equilibrium:    
EquaVon  of  MoVon  
•  Total  displacement:  
–  Ground  displ.  
–  RelaVve  displ.  
•  Forces:  
–  InerVa  
–  Damping  
–  Structural  resistance  
•  Equilibrium    

11  
 6/14/10  

ElasVc  Structural  Response    

•  Characterized  
by:  
–  Natural  
period  
(frequency)  
of  vibraVon  
–  Damping:  
energy  
dissipaVon  
during  
vibraVon  

ElasVc  Response  Spectrum  

12  
 6/14/10  

ElasVc    
Response  Spectrum  
•  Displacement:  D  
•  Pseudo-­‐velocity:  V  

•  Pseudo-­‐acceleraVon:  A  

•  Note  on  effecVve  force  

•  Note  on  strain  energy:  

Spectra  Vary,  too!  

•  El  Centro  1940,   •  El  Centro,  different  
different  damping   earthquakes  

•  More  damping:  less  

deformaVon  

13  
 6/14/10  

ElasVc  Design  Spectra  

•  Average  of  a  large   •  Describes  elasVc  
number  of  similar   seismic  response  of  
ground  moVon   simple  structures  
spectra  
•  Smoothing  occurs  
along  the  way  
•  Note:  PGA  on  the  
T=0  axis  
•  Note:  Cap  at  the  
maximum  values  

Seismic  Hazard  and  Risk  

•  Seismic  Hazard  Analysis:   •  Seismic  Risk  Analysis:  

 Describes  the  potenVal    Assesses  the  probability  

for  dangerous   of  occurrence  of  losses  
earthquake-­‐related   (human,  social,  
natural  phenomena   economic)  associated  
(such  as  ground   with  seismic  hazard  
shaking)  

Risk  =  P(event  occurring)  x  (Impact  of  event  occurrence)  

14  
 6/14/10  

ProbabilisVc  Seismic  Hazard  Analysis  

1.  IdenVfy    
Sources  

4.  Combine  to  compute  

probability  of  exceedance  

2.  Examine    
Magnitude   3.  Examine  
Recurrence   Ground    
MoVon  
AgenuaVon  

Earthquake  Source:  Faults  

•  Faults  (localizing  
structures):  
–  Known/Unknown  
–  AcVve/Passive  
•  Seismotectonic  
province:  
–  There  is  a  known  seismic   •  Max.  possible  eq.  
hazard,  but  there  are  no   •  Max.  credible  eq.  
known  localizing  
•  Max.  historic  eq.    
structures  

15  
 6/14/10  

Probability  of  Recurrence:  

Magnitude  and  Likelihood  of  Occurrence  

Physical  Limits  on  Magnitude  

16  
 6/14/10  

Ground  MoVon  AgenuaVon  

•  Reasons:  
–  Geometric  spreading  of  
waves  
–  AbsorpVon  (damping)  in  
the  rock/soil  

Empirical  AgenuaVon  RelaVons  

hgp://peer.berkeley.edu/products/nga_project.html  

17  
 6/14/10  

Model-­‐Based  AgenuaVon  RelaVons  

•  Complex,  regional  
models:  
–  Sources  
–  Faults  
–  Geographic  features  
–  Rock  and  soil  layers  
•  Huge  computer  
resources  
•  hgp://www.scec.org/  

Seismic  Hazard  Curve  

•  For  a  given  site,  

provide  the  
probability  that  a  
ground  moVon  
intensity  parameter  
will  be  exceeded  
•  P(IM>im)  

18  
 6/14/10  

Use  of  Seismic  Hazard  Curves  

Seismic  Risk  from  NPP  Structures  

•  Losses  to  society  due  to  a  large  radiaVon  
release  induced  by  earthquake  ground  moVon  

JNES  web  site  

19  
 6/14/10  

How  Ground  MoVons  Affect    

Real  Structures  

How  Ground  MoVon  Affects  Structures  

20  
 6/14/10  

How  Ground  MoVon  Affects  Structures  

How  Ground  MoVon  Affects  Structures  

21  
 6/14/10  

Measure  of  Seismic  Demand  

Engineering  Demand  Parameters  (EDP)  
Deforma#on  duc#lity   Energy  absorp#on  ability      

•  Story  drip:  relaVve  moVon   •  Quality  of  structural  system  

between  top  and  bogom  of   and  structural  detailing  
a  story  

Seismic  Demand  Model  

•  A  relaVon  between  ground  moVon  intensity  and  
demand(s)  on  the  structure  
•  Method:  
–  Develop  a  computer  model  of  the  structure  
–  Develop  a  porrolio  of  site-­‐specific  ground  moVons  
scaled  to  reflect  different  hazard  levels  
–  Conduct  (a  possibly  large  number  of)  analyses  to  
determine  demand  on  the  structure  imposed  by  each  
ground  moVon  
–  Do  a  staVsVcal  fit  

22  
 6/14/10  

Seismic  Demand  Model:  

Log-­‐log  Linear  
•  A  condiVonal  probability:  P(EDP>edp|Im=im)    

Damage  to  Structures  

•  Local  damage  affects  
global  loss  of  stability  

23  
 6/14/10  

Damage  to  Non-­‐Structural  Content  

•  Affects  ability  to  
conVnue  using  the  
structure  

Seismic  Damage  Model  

•  A  relaVon  between  demand  on  the  structure  
and  structural  and  non-­‐structural  damage  
•  Method:  
–  Gather  data  (experimental,  empirical,  from  
manufacturers)  on  the  types  of  damage,  when  
they  occur  and  how  they  affect  the  structure  
–  Do  a  staVsVcal  fit  
–  Determine  probabiliVes  of  excessive  damage  

24  
 6/14/10  

Seismic  Damage  Model:  

Fragility  Curves  
•  CondiVonal  probability  P(DM>dm|EDP=edp)  

Performance  of  the  Structure  

•  HolisVc  evaluaVon  of  risk  

25  
 6/14/10  

Discrete  Performance  Levels  

Seismic  Decision  Model  

•  A  relaVon  between  structural  and  non-­‐
structural  damage  and  the  performance  ability  
(or  lack  there  off)  of  a  structure  from  the  
standpoint  of  its  intended  funcVon  
•  Method:  
–  Establish  Decision  Variables    
–  Establish  threshold  values  and  associate  them  
with  acceptable  recurrence  intervals  

26  
 6/14/10  

Seismic  Decision  Model:  

Fragility  Curves  
•  CondiVonal  probability:  P(DV>dv|DM=dm)  

Risk  EvaluaVon  
•  Given  a  seismic  hazard  environment  and  a  structure,  
the  probability  that  a  performance  objecVve  is  not  
achieved  (DV  exceeds  a  threshold)  is:  

•  Consider  probability  distribuVons  of  seismic  hazard,  

of  demand,  damage  and  decision  variables  due  to:  
–  Lack  of  knowledge  (epistemic  uncertainty)  
–  Record-­‐to-­‐record  randomness  (aleatory  uncertainty)    

27  
 6/14/10  

Performance  ObjecVve:  
ProbabilisVc  DescripVon  of  Risk  
Performance  Level   Performance  Recurrence  

Performance  ObjecVve  Table  

Recurrence  Interval  

28  
 6/14/10  

Acceptance  Criteria  
•  A  comparison  of  Demand  and  Capacity:  
–  Failure  (to  saVsfy  a  performance  objecVve)  occurs  
when  (some  staVsVcal  expression  of)  demand  is  
larger  than  (some  staVsVcal  expression  of)  
capacity  

Demand  
distribuVon   Capacity    
distribuVon  

C  mean  
D  mean  

threshold  

Risk-­‐Informed  Design  
•  Formulate  design  acceptance/rejecVon  criteria  
such  that  there  is  High  Confidence  in  Low  
Probability  of  Failure  (HCLPF)  to  saVsfy  a  
performance  objecVve    
•  Example:  
–  99%  confidence  that  the  probability  of  collapse  is  
less  or  equal  1%  in  any  50-­‐year  interval    
•  Accounts  for  first  and  second  moments  of  the  
interim  model  probability  distribuVons  

29  
 6/14/10  

Structural  Engineering  of  NPPs  

ASCE  43-­‐05  
•  Two  acceptance  criteria:  
–  Less  than  1%  probability  of  unacceptable  
performance  for  the  Design  Basis  ground  moVon  
–  Less  than  10%  probability  of  unacceptable  
performance  for  150%  of  Design  Basis  ground  
moVon  
•  Both  must  be  saVsfied:  
–  Trying  to  control  the  shape  of  the  fragility  curve  by  
these  two  points    

Safety  Goal  
•  Again  10-­‐6  annual  probability  of  seismic  core  
damage  

An#cipated  opera#onal  
       occurrences  

Design  basis  events  

Beyond  design  basis  

                   events  

30  
 6/14/10  

Common  Design  Framework  

•  Performance-­‐based  design  is,  in  essence,  
technology-­‐neutral  design:  
–  Define  what  we  want  to  achieve,  not  how  
•  Acceptance  criterion  units  must  be  selected  to  
enable  use  of  different  technologies:  
–  Ability  to  model  and  analyze  is  crucial  
•  Design  must  be  risk-­‐informed:  
–  HCLPF  to  perform  as  desired    
–  Basis  for  comparaVve  evaluaVon  of  different  
technologies  

Seismic  Safety  of    

Nuclear  Power  Plants  
•  Complex  energy  conversion  
machines  
•  Engineered  by  teams  
represenVng  all  branches  of  
engineering  
•  Engineered  to  perform  their  
funcVon  in  a  manner  that  is  
safe  even  under  most  
extreme  hazards  

31  
 6/14/10  

Thank  you!  

32