Probabilistic Seismic Risk Assessment for CCS

Induced Seismicity Working Group National Risk Assessment Partnership

NRAP’s Induced Seismicity Working Group Lawrence Berkeley National Laboratory C. Bachmann T. Daley B. Foxall L. Hutchings T. Kneafsey J. Rutqvist H. Murakami-Wainwright Lawrence Livermore National Laboratory S. Carroll L. Chiaramonte S. Johnson W. Trainor-Guitton J. Wagoner J. White Los Alamos National Laboratory C. Bradley B. Carey D. Coblentz R. Lee

National Energy Technology Laboratory D. Crandall E. Lindner H. Siriwardane (WVU) Pacific Northwest National Laboratory Z. Hou C. Murray External Collaborators J. Savy, Savy Risk Consulting J. Dieterich, UC Riverside

Contact: Joshua White, jawhite@llnl.gov LLNL-­‐PRES-­‐639617.    Por1ons  of  this  work  were  performed  under  the  auspices  of  the  U.S.  Department  of  Energy  by  Lawrence  Livermore  Na1onal  Laboratory  under  Contract  DE-­‐AC52-­‐07NA27344.  

Typical scenario of concern •  Injection creates relatively small CO2

plume, surrounded by larger plume of pressurized brine.

•  Pressure increase along a well-oriented fault could trigger seismic (or aseismic) slip.

•  The fault is sufficiently large to produce concerning earthquakes.

•  Fault is sufficiently small that it may have been unobserved or poorly characterized during site selection.

•  Overall risk is controlled by several components in a complex system.

Typical scenario of concern

•  Overall risk is controlled by several

components in a complex system. *  Figure  not  to  scale  

Four key risks associated with induced seismicity

①  Damage Risk

•  Induced ground motions can damage nearby infrastructure

②  Nuisance Risk

•  Induced ground motions can annoy nearby populations

③  Brine Leakage Risk

•  Slip-enhanced leakage pathways can allow brine to contaminate protected groundwater.

④  CO2 Leakage Risk

•  Slip-enhanced leakage pathways can allow CO2 to contaminate protected groundwater.

Helpful  to  consider  each  separately.    Though  related,  they  have  different  physics,  1mescales,  likelihoods,  impacts,  poten1al  mi1ga1on,  etc.  

A series of conditions must line up for an impact to occur:

start

Will the pressure plume encounter a sufficiently large fault?

Is the fault capable of generating significant seismicity, based on its

dimensions, orientation, tectonic loading, and frictional properties?

Significant seismicity could occur.

Could resulting ground motion exceed building code standards?

Infrastructure could be damaged.

Could resulting ground motion exceed nuissance standards?

Local population could be scared and/or annoyed.

Does the fault fully or partially penetrate the caprock seal(s)?

Can new leakage pathways be created along the fault,

based on slip distance, slip area, and fault lithology?

Is the duration and magnitude of pressure drive sufficient to allow

brine leakage to protected drinking water?

Brine contamination could occur.

Does mobile CO2 ever reach the fault?

Is the duration and magnitude of pressure and buoyancy drive

sufficient to allow CO2 leakage to protected drinking water?

CO2 contamination could occur.

A series of conditions must line up for an impact to occur:

start

Will the pressure plume encounter a sufficiently large fault?

Is the fault capable of generating significant seismicity, based on its

dimensions, orientation, tectonic loading, and frictional properties?

Significant seismicity could occur.

Could resulting ground motion exceed building code standards?

Infrastructure could be damaged.

Could resulting ground motion exceed nuissance standards?

Local population could be scared and/or annoyed.

Does the fault fully or partially penetrate the caprock seal(s)?

Can new leakage pathways be created along the fault,

based on slip distance, slip area, and fault lithology?

Is the duration and magnitude of pressure drive sufficient to allow

brine leakage to protected drinking water?

Brine contamination could occur.

Does mobile CO2 ever reach the fault?

Is the duration and magnitude of pressure and buoyancy drive

sufficient to allow CO2 leakage to protected drinking water?

CO2 contamination could occur.

Yes.  

Challenging  to  assess  pre-­‐injec1on  (irreducible  uncertain1es)  .  

Offshore?  Remote?  Tokyo,  San  Francisco?  Basel?  

•  Natural  tendency  to  focus  on  early  part  of  the  chain,  when  later  safeguards  may  exist  (or  be  put  in  place).  

start

Will the pressure plume encounter a sufficiently large fault?

Is the fault capable of generating significant seismicity, based on its

dimensions, orientation, tectonic loading, and frictional properties?

Significant seismicity could occur.

Could resulting ground motion exceed building code standards?

Infrastructure could be damaged.

Could resulting ground motion exceed nuissance standards?

Local population could be scared and/or annoyed.

Does the fault fully or partially penetrate the caprock seal(s)?

Can new leakage pathways be created along the fault,

based on slip distance, slip area, and fault lithology?

Is the duration and magnitude of pressure drive sufficient to allow

brine leakage to protected drinking water?

Brine contamination could occur.

Does mobile CO2 ever reach the fault?

Is the duration and magnitude of pressure and buoyancy drive

sufficient to allow CO2 leakage to protected drinking water?

CO2 contamination could occur.

•  Should  consider  induced  risks  in  tandem  with  “background”  risks.  

•  Ac1ve  interven1on  and  mi1ga1on  should  also  be  included.  

High costs and large uncertainties suggest a phased approach to seismicity management Phase   Characteriza-on  &  

Monitoring  Modelling   Risk  Assessment  

•  Site-­‐screening   •  Regional  stress  es1mates  

•  Fault  density  es1mates  

•  Back-­‐of-­‐the-­‐envelope  

•  Red-­‐flags  •  Atlas  

•  Pre-­‐injec1on   •  3D  seismic  •  XLOTs  •  FMI  •  Limited  

microseismic  

•  Simple  models   •  Qualita1ve  Assessments  

•  PSHA  

•  Injec1on  &  PISC   •  4D  seismic  •  Full  microseismic  

array  

•  Sophis1cated  models  

•  Traffic-­‐light  •  PSRA  

-­‐-­‐  Cost/benefit  of  addi1onal  methods  assessed  based  on  evolving  project  condi1ons.  -­‐-­‐  Baselines  are  important.  -­‐-­‐  Timely  processing  and  interpreta1on  of  data  are  important.  

Probabilistic Seismic Risk Assessment

•  PSRA is commonly used for dealing with “natural” seismic hazards.

•  Framework is well-suited to dealing with induced seismicity, but must be modified to address differences between natural and induced events.

•  Three key ingredients to a PSRA:

•  Earthquake frequency/magnitude relationship …. very challenging

•  Ground motion hazard …. mostly standard (some issues)

•  Fragility curves …. mostly standard (some issues)

•  For CCS, framework also needs to be extended to capture leakage risks.

SIMRISK Framework

•  SIMRISK is a framework for PSRA, specifically adapted to induced seismicity.

•  Allows for flexible input, so that component modules may be easily swapped.

•  Currently testing an earthquake simulation module based on RSQSim (Dieterich 1995; Richard-Dinger & Dieterich 2012)

•  Validating against waste-water injection analogs.

Earthquake Frequency

•  Modeled data for a synthetic site in a seismically-active region [Foxall et al. 2012].

Background:  0-­‐200  y  Injec1on:  200-­‐250  y  Post-­‐injec1on:  250-­‐500  y  

Zoom  of  injec1on  period  

Ground Motion Intensity

For reference … •  2 cm/s/s is barely

perceptible by most people.

•  20 cm/s/s will cause light shaking but no damage.

•  200 cm/s/s can cause moderate to severe building damage.

Background  (0-­‐200  y)  

Injec1ng  (200-­‐250  y)  

Fragility Curves •  Fragility curves quantity likelihood of damage to a structure given a certain level

of shaking.

•  Same idea can be applied to a nearby population via nuisance curves [Majer et al. 2012]

NRAP is pursuing integrated, system-level models of damage, nuisance, and leakage risk.

Reservoir(flow(simula/on(

Basin2wide(fault(characteriza/on(

Earthquake((((simula/on*(

Ground(mo/on(calcula/on(

Regional,(in#situ(stress;(background(

seismicity(

Fault(leakage(simula/on(

Damage(&(nuisance(risk(

Leakage(probability(

Aquifer(response(simula/on(

Groundwater(impact(risk(

Δk(

pressure(

fault(slip(Δk(

Slip2related(fault(permeability(

evolu/on(model(

Ground'mo)on' Fault'leakage'

Basin-scale ground motion and earthquake-caused fault leakage risk assessment

Characteriza/on/research(studies(Simula/on/computa/on(Result(Downstream(calcula/on(

pressure(satura/on(

1

4(

6b(

6a(

6c(

7d(

8(

7a2c(

*Empirical((Task(3)(alterna/ve(

2,5(

1 Task/subtask(

Fault(slip2permeability(model(

Key  science  gap:  permeability  behavior  of  slipping  faults.  

Conclusions

1.  There will always be irreducible uncertainties associated with the seismic behavior of a field. That said, it is possible to choose sites that are robust with respect to seismic behavior.

2.  There are four key risks associated with induced seismicity, and each has nuances that should be considered separately.

3.  Seismicity deserves real attention when developing the characterization, monitoring, mitigation plans. A phased approach, combined with good contingency plans, can reduce cost while still addressing risk.

4.  Probabilistic seismic risk assessment provides a rigorous, quantitative framework. Significant progress has been made adapting it to induced seismicity, but some important science gaps still exist.