George Guthrie Presentation

‹#›

Geologic Storage Risk Assessment George Guthrie (US DOE National Energy Technology Laboratory)

•  buoyant relative to brine; mobile – must be trapped beneath impermeable seal

•  chemically reactive – dissolves in brine, forming carbonic acid – changes aqueous chemistry through pH,

complexation, solvent properties, etc. •  large volumes of injected mass

– potential geomechanic responses of the reservoir (seal integrity, ground motion, etc.)

“Observations from engineered and natural analogues as well as models suggest that the fraction retained in appropriately selected and managed geological reservoirs is very likely to exceed 99% over 100 years and is likely to exceed 99% over 1,000 years.” “With appropriate site selection informed by available subsurface information, a monitoring program to detect problems, a regulatory system, and the appropriate use of remediation methods to stop or control CO2 releases if they arise, the local health, safety and environment risks of geological storage would be comparable to risks of current activities such as natural gas storage, EOR, and deep underground disposal of acid gas.”

RECS 2011 Birmingham, AL June 9, 2011

‹#›


Schematic evolution of trapping mechanisms over time (IPCC, 2005) Multiple trapping mechanisms reduce CO2 mobility over time

• structural/stratigraphic • residual • solubility • mineralization; sorption (coal; shales?)

Risk profiles are expected to decline over time Site characterization, site operations, and monitoring strategies work to promote storage security (e.g., DOE Best-Practices documents)

Schematic profile of environmental risk (Benson, 2007)

‹#›

Risk relates to the probability that an event will occur as well as the consequence of that event; risk can vary over time.

Risk = Pevent x Cevent

consequence of an event

Ris

k (L

iabi

lity)

injection stops

Liability can occur when a consequence is declared a harm.

development phase

post-operation site-care

phase

long-term stewardship

phase

site closure (e.g., 40 CFR 146) probability that an

event will occur

injection begins

operational phase

?

? ?

?

? ? ? ? ?

? ? ?

‹#›

CSLF task force identified potential risk-related factors in assessment storage security at a site.

•  impingement on pore space not covered under deed or agreement •  impingement on other subsurface resources •  change in local subsurface stress fields & geomechanical properties •  impact on the groundwater and/or surface water •  elevated soil-gas CO2 in terrestrial ecosystems •  accumulation in poorly ventilated spaces or in low lying areas

subject to poor atmospheric circulation •  CO2 or other displaced gases (e.g., CH4) return to the atmosphere

Ø  Importance of direct impacts from CO2 vs. indirect impacts (e.g., brines, pressure fronts)

cslforum.org

‹#›

US-EPA Vulnerability Evaluation Framework for Geologic Sequestration of Carbon Dioxide

‹#›

Predicting a site’s behavior for risk assessment requires predicting the fate and impact of fluids.

•  Will the site have sufficient capacity & injectivity over time?

–  reservoir pore space –  fluid movement (CO2, brine, other)

•  What will the impacts be from introducing CO2 into the reservoir?

– pressure distribution; chemical reactions –  fluid movement (CO2, brine, other)

•  What is the chance that all of the CO2 will remain in the reservoir?

– non-reservoir pore space (wellbores; faults/fractures; fastpaths; other subsurface resources)

–  fluid movement (CO2, brine, other) •  What might the impacts be of CO2 release

from the reservoir? – e.g., CO2 & brine impact on groundwater –  fluid movement (CO2, brine, other)

k ρ g

µ A dh

dl

Q

injection phase

post injection phase

Darcy’s Law

= –

k — permeability ρ — density g — gravity µ — dynamic viscosity dh/dl — hydraulic head/length

Risk = Pevent x Cevent

‹#›

Predictions must also account for effects of CO2 dissolving into water, which changes brine chemistry.

Predicted dissolved CO2 concentration in Oriskany brine (from Goodman et al.). Comparison of experimental and predicted CO2

solubilities as a function of pressure.

CO2 + H2O = H2CO3

H2CO3 = HCO3– + H+

HCO3– = CO3

2– + H+

pH is a major factor in determining chemical reactions between rocks, brine, and CO2.

‹#›

Predicting fluid flow hinges on characterizing the nature of pores and fractures in the subsurface over time.

must know: •  porosity

–  distribution of pores –  volume of pores

•  permeability –  connectivity of pores –  topology of pores –  materials lining pores –  fluids in pores

•  changes over time

‹#›

Fluid flow in the subsurface is controlled by permeability. Challenge: Data limitations require inferring permeability, which varies over space (and time). Thus, predictions embody uncertainty.

SACROC Core Data

Seismic image through SACROC reservoir (from M. Holtz previous RECS presentation).

Permeability is typically inferred from porosity logs, despite a poor relationship between the two. And, permeability can vary by orders of magnitude within a reservoir.

Direct data typically represent only <10–10 of the total reservoir volume.

‹#›

North American CO2 Storage Resource Estimates (Gt or 1012 kg)

Sink Type Low High Saline Formations 1653 22,281 Unmineable Coal Seams 60 117 Oil and Gas Fields 143 143

Available for download at http://www.netl.doe.gov/technologies/carbon_seq/refshelf/atlasIII/index.html

National Atlas (3rd Edition)

Emissions ~ 3.8 Gt CO2/yr point sources

saline

oil/gas coal

basalt shale (organic rich)

GCO2 (mass) = A • h • φ • ρCO2 • Esaline

‹#›

Classes of Reservoirs Tested by Regional Carbon Sequestration Partnerships

Sandstone

Coal

Basalt

Carbonate

Eolian*

Fluvial*

Coal

Basalt

Wave Delta-‐Marine

Tidal Delta

Shallow Shelf Open*

Reef

Strandplain*

Shallow Shelf

Restricted

Gross lithology Class Distribution

*Classes being tested in Phase III

‹#›

Risk assessment requires predicting fluid fate & impact from the reservoir to receptor.

Challenge: Most industrial experience ties to production of fluids from the reservoir, whereas risks often tie to other conditions…

•  injection of fluids

•  release of fluids through “open” pathways (wellbores, fractures, etc.)

•  long-term (post-production) processes

•  fate and impact of fluids outside of the reservoir

Large vertical volume to consider “Reservoir to Receptor”

(sometimes defined by regulator via an “area of review”)

‹#›

Conceptual hydrogeologic models are simplifications/approximations of the true system.

must recognize:

•  heterogeneity –  natural systems vary

•  uncertainty

–  will never know all subsurface characteristics

–  may not know all physical and chemical processes

Challenge: The complexity of natural geologic systems controls the fate of mobile fluids, yet it is never fully known.

‹#›

Monitoring/measurement are key elements of reducing uncertainty in CO2 storage projects.

Unc

erta

inty

on

Orig

inal

Oil

in P

lace

Number of Wells Drilled

Cooper (2009)

Case study of uncertainty over time in an oil project (Cooper, 2009)

Schematic illustration of evolution of uncertainty over time (from BPM on Risk Analysis; DOE/NETL-2011/1459)

‹#›

Schematic evolution of trapping mechanisms over time (IPCC, 2005)

Multiple trapping mechanisms reduce CO2 mobility over time • structural/stratigraphic; residual; solubility; mineralization; sorption (coal; shales?)

Risk profiles are expected to decline over time

Site characterization, site operations, and monitoring strategies work to promote storage security (e.g., DOE Best-Practices documents)

Schematic profile of environmental risk (Benson, 2007)

‹#›

Risks to a project’s success must be considered at all stages (from injection to long-term stewardship).

①  Fluid Flow in Reservoir Ø  fluid flow (porous flow & discrete fractures) Ø  trapping mechanisms Ø  heterogeneous and uncertain permeability

③  Wellbore/Seal Integrity Ø  flow in pipe/fracture

(coupled to reservoir flow) Ø  wellbore/fracture locations & geometries Ø  effective permeabilities Ø  cement behavior (stemming/completion) Ø  geochemical and geomechanical effects

④  Groundwater Protection Ø  fluid flow and fluid–rock geochemistry Ø  contaminant transport

⑤  Atmospheric Emissions Ø  release(s) from subsurface Ø  emissions from surface activities

②  Ground Motion Response Ø  geomechanical response to pressure changes

(induced seismicity; hydraulic fracturing; etc.)

‹#›

Simulation of trapping mechanisms suggests significant residual & solubility trapping on the decade-to-century time scale.

WestCarb Investigation – 1,000,000 metric tons over 4 yrs – candidate storage formation

(dipping formation; interbedded high permeability sand & low permeability shale)

– brine + supercritical CO2

Trapping Mechanisms – effectively immobilized by 25 yrs – significant solubility trapping in

years to decades, with complete dissolution in ~103–4 yrs

Simulated distributions of CO2 over time (from BPM on Risk Analysis; DOE/NETL-2011/1459)

‹#›

Simulation suggests that presence of residual oil in reservoir can be a rapid mechanism for trapping a significant portion of CO2.

Southwest Partnership Investigation – 7,000,000 metric tons over 30 yrs – mature CO2-EOR operation (SACROC)

(pinnacle reef formation; limestone overlain by thick shale sequence)

– brine + supercritical CO2 + oil Trapping Mechanisms

– without oil, significant residual and solubility trapping in ~101–2.5 yrs

– with oil, CO2 is largely trapped in oil versus residual CO2 or solubility

– both scenarios have a significant mobile plume at 200 yrs


‹#›

Simulation suggests that presence of residual oil in reservoir can be a rapid mechanism for trapping a significant portion of CO2.

Southwest Partnership Investigation – 7,000,000 metric tons over 30 yrs – mature CO2-EOR operation (SACROC)

(pinnacle reef formation; limestone overlain by thick shale sequence)

– brine + supercritical CO2 + oil Trapping Mechanisms

– without oil, significant residual and solubility trapping in ~101–2.5 yrs

– with oil, CO2 is largely trapped in oil versus residual CO2 or solubility

– both scenarios have a significant mobile plume at 200 yrs


‹#›









‹#›

Science-based system models tie system-level descriptions to process-level phenomena.

•  probability distribution functions (PDFs)

•  simplified analytical expressions

•  detailed physical & chemical processes

•  System-level models (probabilistic) – allow the treatment of heterogeneity and uncertainty – can be too general or limited in site-specific applicability

•  Process-level models (deterministic) – allow explicit treatment of physics and chemistry – can be too detailed and/or difficult to apply at large scale

‹#›

Wellbore integrity is central to storage integrity.

Gasda, Bachu, Celia (2004)

Two main issues to consider… •  Flow through the cement

(porous flow; fracture flow)

•  Flow along interfaces (open fractures, channels, etc.)

‹#›

For predicting movement of CO2 in wellbore at the system level, one must know the change in wellbore permeability over time.

time

Wel

lbor

e P

erm

eabi

lity

time

i ii

iii iv Possible scenarios for wellbore fate

i.  wellbores degrade rapidly ii.  wellbores degrade slowly iii.  wellbores improve over time iv.  wellbores are unaffected by CO2+brine

‹#›

CO2-EOR operations routinely utilize wellbore technology to place (and to contain) fluids within the reservoir.

SACROC is one of several industrial-scale examples in the Permian Basin

•  ~13.5 million tonnes of CO2/yr injected •  (~6-7 million t/yr of new CO2) •  ~ 70 million tonnes CO2 accumulated

(>30 million tonnes anthropogenic) •  CO2 injection since 1972

Carey et al., 2007!

49-6!

‹#›

SACROC Well 49-6 has experienced a complex history.

Additional core taken at 5160’ (adjacent geological formation is limestone)

v Drilled/completed 1950 Ø  water used as drilling fluid Ø  portland cement

v Water flood initiated 1954 Ø  Colorado River source supplemented

by Big Spring, TX, effluent

v First direct CO2 exposure 1975 Ø  Began with Val Verde gas plant CO2

(~45% of CO2 injected to date) Ø  Shift to include additional sources

(McElmo & Bravo Domes, Sheep Mtn) Ø  10 yrs as injector; 7 yrs as producer Ø  2.2 Bscf CO2 produced/injected

Top of

Pay

‹#›

Whipstock drilling at SACROC 49-6 provided recovery of core through cemented annulus to within 7’ of top of pay.

v  Drilled/completed 1950 Ø  water used as drilling fluid Ø  portland cement

v  Water flood initiated 1954 Ø  Colorado River source

supplemented by effluent

v  First direct CO2 exposure 1975

Ø  10 yrs as injector; 7 yrs as producer

Ø  2.2 Bscf CO2 produced/injected

v  Primary core taken at ~6550’ Ø  within ~7–19’ from top of pay

v  Additional core taken at 5160’

wireline data from KinderMorgan

‹#›

Observations suggest initial flow along interfaces followed by precipitation of silica and carbonate phases.

casing shale fragment zone

carbonated cement

“pristine” cement

silica-carbonate precipitation

shale caprock w

ellb

ore

v  fluid flow along interface into sandy unit in shale

v  precipitation of silica and carbonate from brine along interfacial zones

v  diffusive carbonation of cement to form orange “popcorn” zone

Guthrie et al., 2005, Midland CO2 Conf.; Carey et al., 2007, Int. J. GHG Cont.!

‹#›

NRAP is developing predictive basis for wellbore integrity.

experiment

35-yr field exposure (SACROC)

Key Research Needs • Reaction mechanism(s) for various

conditions – P, T, brine/reservoir composition – Cement admixtures – Co-constituents in CO2

• Diffusion vs. advection – conditions for opening/closing of flowpath

• Coupled flow-reaction-geomechanics

Simple diffusion controlled model can

explain observed reaction mechanisms driven by gradients in

chemistry and pH

dense CaCO3 + silica

silica + Ca(OH)2 depleted zone with minor CaCO3

unaltered cement

porous silica (CaCO3 depleted)

Ca 2+, OH – CO3 2–, H +

pH~12.4

pH~4–5

CT images of cement core show bridging of internal flow pathway during experiment due to dissolution, transport, and precipitation in a CO2 saturated brine.

bridging of flow path

advective alteration

Field evidence suggests wellbore integrity can be controlled by both diffusive and advective processes.

diffusive alteration

‹#›

Airborne measurements are being used to identify wellbore locations and gas flux.

• RD 100 Award for Abandoned Well Detection Technology

–  Magnetic Survey •  Locates steel-cased wells by

detecting their distinctive monopole magnetic anomaly

–  Methane Detection •  Uses Differential absorption

lidar system •  Measures methane by

reflecting two lasers off of the ground

Methane DetectorMethane Detector

MagnetometersMagnetometers

Methane DetectorMethane Detector

MagnetometersMagnetometers

100 m

N

Magnetic response of steel cased wells

‹#›









‹#›

National Risk Assessment Partnership (NRAP) is developing science-based methodology for quantifying site-specific risk profiles.

Storage Reservoir

Release and Transport

Potential Receptors or

Impacted Media

Risk Profiles are being developed for four potential consequences ① pH in groundwater

② TDS in groundwater (total dissolved solids)

③ return of CO2 to atmosphere

④ pressure-induced ground motion (induced seismicity; fracturing of seal)

‹#›

What tools will NRAP exploit and develop for risk profiles?

•  research simulators for groundwater chemistry

•  analytical expressions

•  research simulators & analytical representations for wellbores, fractures, porous media

•  pressure & saturations from reservoir simulators (research & commercial)

•  research simulators & scaling relationships for stress/geomechanics

•  system model(s) for integration •  statistical methods/models

Tools •  statistical analysis of industrial

and natural analogs

Validation Strategy

•  field data from industrial and natural analogs (groundwater; atmosphere)

•  field data from small tests (e.g., ZERT, EPRI, other)

•  field and historical data from industrial and natural analogs

•  proposed field test for injection into fault/fracture

•  field data from RCSPs and other large scale storage

•  field data from industrial analogs •  proposed field test for stress-

response relationship

Storage Reservoir



Impacted Media

‹#›

Preliminary Integrated Assessment Model for Risk Profiles in Groundwater Systems

CO2/brine leakage rates used as boundary conditions in detailed reactive-flow models to calculate dynamic evolution of pH & TDS • equilibrium-geochemistry, continuum-scale reactive flow; based on two real aquifers

•  High Plains aquifer in LLNL’s NUFT •  A Coastal Sandstone aquifer in LBNL’s TOUGH2

Wellbore-release model used to calculate CO2/brine leakage rates based on predicted reservoir pressure and saturation • abstraction based on continuum-scale multiphase-flow model plus Monte Carlo analysis

•  Multiple realizations using wellbore cement characteristics •  CO2/brine leak rates calculated in LANL’s CO2-PENS using

abstraction for wellbore flow output from reservoir model

Detailed reservoir model used to predict pressure & saturation at reservoir–caprock interface • continuum-scale multiphase-flow model

•  based on real site •  used to predict CO2:brine ratio (saturation), pressure

Approach assumes that mass transfer across sub-system boundaries does not significantly affect mass balance within individual sub-systems.

Storage Reservoir



Impacted Media

‹#›

Storage Reservoir



Impacted Media

Integration of reservoir behavior through continuum-scale reservoir model to predict pressures and saturations at bottom of caprock.

TOUGH2 model of potential storage formations at a site in the Southern San Joaquin Valley

•  Lateral extent 53 km by 46 km with 3.85º dip; 22-layer model with total thickness = 540 m; depth 1805–2345 m

•  Hydrostatic pressure (~220 bars at Vedder); geothermal temperature gradient (T=71 ºC at Vedder)

Injection of 1 million metric tons CO2/yr for 50 yrs; followed reservoir evolution for 50 yrs post-closure

•  Pressures & saturations at the top formation layer at 20 time intervals

Reservoir Simulations for Preliminary Profiles: Curt Oldenburg (LBNL), Christine Doughty (LBNL)

‹#›

Integration of release processes through wellbore-release model based on abstracted multiphase physics and assumed wellbore permeability.

Wellbore leak-rate was treated as a stochastic variable using Monte Carlo analysis

•  Wellbore response surface generated from high-fidelity, multi-phase flow of CO2/brine through wells using LANL’s continuum-scale FEHM

•  Leak-rate variability was function of pressure, saturation, and permeabilities of reservoir, wellbore cement, and aquifer.

•  Monte-Carlo methods using LANL’s CO2-PENS system model and wellbore response surface

•  Coupling to storage reservoir via simulation results from TOUGH2 (time dependent pressure and saturation)

Wellbores were assumed to have spatial density of a typical EOR site (based on site in west Texas)

•  10 randomly distributed abandoned wells (injection well is not considered as potential flow path)

•  90% of wells with good (low permeability) cement; 10% wells with poor cement

•  Values used for wellbore cements based on preliminary assessment of one available field data set

•  good cement permeability – 10–17 m2 (10 µD) •  poor cement permeability – 10–10 m2 (100 D)

Time-dependent CO2 & brine leakage rates into shallow aquifer were based on multiple (but limited) realizations

Storage Reservoir



Impacted Media

Leakage Simulations for Preliminary Profiles: Phil Stauffer (LANL), Rajesh Pawar (LANL)

‹#›

Integration of aquifer processes through equilibrium-geochemistry, continuum-scale, reactive-flow model.

Two different sets of calculations with two models • High Plains aquifer using LLNL’s NUFT (Caroll et al., 2009) • Coastal sandstone aquifer using LBNL’s TOUGH2 (Zheng et al, 2009)

Both models used time-dependent CO2 & brine leakage rates as boundary conditions to predict time-dependent change in pH and TDS

Reactive transport calculations with assumed mineralogy and fluid compositions • quartz-calcite aquifer; quartz-feldspar-clay aquifer

Background flow to account for regional groundwater flow Storage Reservoir



Impacted Media

pH in Aquifer

log[a(Na+)] in Aquifer

Aquifer Simulations for Preliminary Profiles: Yue Hao (LLNL), Susan Carroll (LLNL), Liange Zheng (LBNL), Jens Birkholzer (LBNL)

‹#›

Preliminary* Risk Profiles for pH in Groundwater System

•  Probability of pH impact goes down with distance from release point •  Recovery initiates after injection ceases

* Although these profiles were derived from quantitative simulations, they embody limitations due to numbers of runs, level of detail, single release process, comprehensiveness with respect to uncertainty, etc. Rather, they form the basis for NRAP’s initial work to develop 1st generation, quantitative risk profiles.

Preliminary Risk Profiles Simulations: NRAP Systems Modeling Working Group

‹#›


Documents

George Guthrie Presentation