In-Situ MVA of CO2 Sequestration Using Smart
Field Technology FE - 0001163
Shahab D. Mohaghegh Petroleum Engineering & Analytics Research Lab (PEARL)
West Virginia University U.S. Department of Energy
National Energy Technology Laboratory Carbon Storage R&D Project Review Meeting
Developing the Technologies and Infrastructure for CCS August 20-22, 2013
2
Presentation Outline
• Introduction • Reservoir Simulation Model • Intelligent Leakage Detection System (ILDS) • Accomplishments • Summary
Objective • Develop an in-situ CO2 leak detection technology based on
the concept of Smart Fields. – Using real-time pressure data from permanent downhole gauges to
estimate the location and the rate of CO2 leakage.
CO2 Leakage(X,Y,Q) Artificial
Intelligence & Data Mining
Industrial Advisory Committee (IAC)
• Project goes through continuous peer-review by an Industrial Review Committee.
• Meetings:
– November 6th 2009 : • Conference call • Site selection criteria
– November 17th 2009: • A meeting during the Regional Carbon Sequestration Partnership Meeting in Pittsburgh • Selection of a suitable CO2 sequestration site
– November 18th 2011: • Reporting the modeling process to IAC
– February 16th 2012: • Reporting the modeling process to NETL/DOE
– April 18th 2013: • Reporting project’s progress to NETL/DOE
Name Affiliation Neeraj Gupta Battelle Dwight Peters Schlumberger George Koperna ARI Grant Bromhal DOE-NETL Richard Winschel CONSOL
Background
Citronelle
Injected Fluid: Carbone Dioxide Depth of Injection Well:11,800ft Depths & Geological Name of Interval: 9,400-10500 ft (Paluxy Formation)
Injection Volumes: 500 ton/day(9.48 Bcf/day) Injection Duration: 3 Years(2012-2015)
Geological Model 3 Cross Sections Sand Layers-D-9-7 Grid Thickness
Porosity from 40 Well Logs Permeability Realizations
Reservoir Simulation Model
7
17 Layers( 10 Injection Layers) 51 Simulation Layers Porosity Distribution from 40 Well Logs Permeability Distribution: Conductive 1,147,500 Grid Blocks
Plume extension: 500 years after injection ends.
Plume extension is shown
only for the blocks with
CO2
Impact of Trapping Mechanism
8
Srg = 0.30 Srg = 0.29
Srg = 0.26
Srg = 0.11
Residually Trapped CO2 vs. Time
Impact of Trapping Mechanism
9
Dissolved CO2 vs. Time
Srg = 0.30
Srg = 0.29
Srg = 0.26
Srg = 0.11
Impact of Trapping Mechanism
10
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Cardium #1 Cardium #2 VikingSandstone
Nisku #2
Mineral Trapping 0.000% 0.000% 0.000% 0.000%Structural Trapping 17.101% 55.176% 33.482% 80.237%Solubility trapping 28.250% 10.238% 28.064% 9.254%Resiudual traping 54.649% 34.586% 38.455% 10.510%
Con
trib
utio
n Pe
rcen
tage
, %
Trapping Mechanism Contribution to the Storage
Process (After 500 years)
Srg = 0.30 Srg = 0.11
Total CO2 Injected (MMCF) 15,045 Total CO2 Injected (TONS) 550,596
Seal Quality
11
Realization Thickness (ft) Permeability (Darcy)
1 150 10^-3 2 150 10^-5 3 150 10^-7 4 200 10^-3 5 200 10^-5 6 200 10^-7 7 250 10^-3 8 250 10^-5 9 250 10^-7
150 ft < h < 250 ft 10-3
darcy < k < 10-7 darcy
Basal Shale
Danztler Sand
Two additional geological layers where included in the model corresponding to the Washita-Fredericksburg interval (on top of the Paluxy formation):
• Basal Shale (Seal) • Danztler Sand (Aquifer)
Seal Quality
12
Grid refinement of the basal shale simulation layers: Grid was refined vertically into 75 to 125 simulation layers to generate grid-blocks with thickness of 2 ft.
Seal Quality
13
0
50
100
150
200
250
1 2 3 4 5 6 7 8 9
Dep
th o
f inv
asio
n (ft
)
Realizations
Depth of invasion of CO2 within the Basal Shale (all realizations)
Realization Thickness (ft) Permeability (Darcy)
1 150 10^-3 2 150 10^-5 3 150 10^-7 4 200 10^-3 5 200 10^-5 6 200 10^-7 7 250 10^-3 8 250 10^-5 9 250 10^-7
150 ft < h < 250 ft 10-3
darcy < k < 10-7 darcy
Seal Quality
14
Conductive Seal 150 md-ft< k*h< 250 md-ft
Tight Seal 1.5 md-ft< k*h< 2.5 md-ft
Very Tight Seal 0.015 md-ft< k*h< 0.025 md-ft
Seal Quality
15
Pressure gain – all scenarios Seal Conductivity
Scenario Permeability of the Confining Unit (md)
K*h range of the confining unit (md-ft)
Conductive 1 150 250 Tight 0.01 1.5 2.5
Very Tight 0.0001 0.015 0.025
0
20
40
60
80
100
120
140
ConductiveTight
Very Tight
Ave
rage
Pre
ssur
e ga
in (p
si)
Scenario
Pressure Gain vs Scenario
0.43 % and 3.18% of the initial average reservoir
pressure
Pressure Gain = Avg. P @ 500 years – Initial Avg. P
D--7
1,259,000 1,261,000 1,263,000 1,265,000 1,267,000 1,269,000 1,271,000 1,273,000
1,259,000 1,261,000 1,263,000 1,265,000 1,267,000 1,269,000 1,271,000 1,273,000
11,271,00011,273,000
11,275,00011,277,000
11,279,00011,281,000
11,283,000
11,2
73,0
0011
,275
,000
11,2
77,0
0011
,279
,000
11,2
81,0
0011
,283
,000
0.00 0.25 0.50 0.75 1.00
0.00 0.50 1.00 km
3.15 miles
3.15 miles
Impact of Boundary Conditions
16
Pressure Behavior in Observation Well(D-9-8)
1
2
3
1 2 3 4
4
Post Injection Site Care (PISC)
17
Yearly Pressure Difference distribution Threshold Verification
1 ye
ar p
ost-
inje
ctio
n
2 ye
ar p
ost-
inje
ctio
n
Sensitivity Analysis
18
Reservoir Pressure @ Observation Well
• Kv/Kh • Maximum Residual
Gas Saturation • Brine Density • Brine Compressibility • Boundary Condition
Permeability Relative permeability
CO2 Plume Extension
History Matching
19
D-4-13 and/or D-4-14In-zone monitoringAbove-zone monitoringFluid sampling
D-9-11Neutron Logging
Primary Injector (D-9-7#2)Injection surveysPressureSeismicGroundwater
Backup Injector(D-9-9#2)Neutron loggingSeismicGroundwater
Characterization Well (D-9-8#2)Neutron loggingPressureFluid samplingSeismicDistributed Temperature
Model Plume Extent
Injection Data PDG Data
4365
4375
4385
4395
4405
4415
4425
4435
4445
4455
4465
8/17/2012 9/6/2012 9/26/2012 10/16/2012 11/5/2012 11/25/2012 12/15/2012 1/4/2013 1/24/2013 2/13/2013
Pres
sure
Psi
Date
PDG(5109) PDG(5108)
History Matching
20
17 Layers( 10 Injection Layers) 51 Simulation Layers Porosity Distribution from 40 Well Logs Permeability: 460md 125*125*51 (800000) Grid Blocks (∆x = ∆y =133.3 ft) Relative Perm: Mississippi Test site (sg=7.5%)
Operational Constraints (actual rate +Max 6300 psi) Ρbrine = 62 lb/ft3 Cbrine = 3x10-6 (1/psi) at 14.7 psi Preference = 4393psi at 4015 ft. Kv = 0.1Kh
History Matching
21
Matching Parameter Absolute Permeability Brine Density Reference Pressure Transmissibility Multiplier Reservoir Boundary Relative Permeability
CO2 Leakage Modeling
22
CO2 Leakage Modeling
23
CO2 Leakage Modeling
24
00.10.20.30.40.50.60.70.80.9
0 500 1000 1500 2000 2500
∆P(P
si)
Time(hour)
Leaking Well(D-9-2) Leakage rate = 30MSC/day
Observation Well
Pressure Behavior @ Observation Well
AI Model Development
25
00.10.20.30.40.50.60.70.80.9
0 500 1000 1500 2000 2500
∆P(
Psi)
Time(hour)
-Leakage Rate -Leakage Location(X,Y)
Descriptive Statistics Mean 0.091 Standard Error 0.0047 Median 0.092 Mode 0 Standard Deviation 0.062 Sample Variance 0.0038 Kurtosis -1.31 Skewness 0.029 Range 0.195 Minimum 0 Maximum 0.195 Sum 15.38 Count 168
Data Summarization
AI Model Development
26
Leakage Rate
Mcf/day
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
Leakage Location(X) Leakage Location(Y)
Leak
age
Rat
e
Hou
rly
Pres
sure
on
e w
eek
afte
r Lea
kage
(168
dat
a po
ints
fo
r eac
h le
akag
e ra
te)
Leakage Locations
Output Input
Validation – Blind Runs
27
Leakage Rate
Mcf/day 26 52
88
11276000
11276500
11277000
11277500
11278000
11278500
11279000
11279500
1268500 1269000 1269500 1270000 1270500
Lati
tude
(Y)
Longtitude(X)
Actual Leakage Location
Neural Network PredictionLeakage
Location(X) Actual
Leakage Location(X)
N.N
Leakage Location(Y)
Actual
Leakage Location(X)
N.N Run 1 1268902.53 1268903.05 11277566.74 11277569.97 2 1268902.53 1268902.78 11277566.74 11277565.13 3 1268902.53 1268902.55 11277566.74 11277567.57 4 1270359.37 1270359.03 11279158.24 11279157.46 5 1270359.37 1270359.11 11279158.24 11279157.51 6 1270359.37 1270359.17 11279158.24 11279157.44 7 1270184.29 1270184.53 11276221.98 11276223.47 8 1270184.29 1270185.16 11276221.98 11276224.14 9 1270184.29 1270183.81 11276221.98 11276222.66
Nine new leakage Scenarios
Validation – Blind Runs
28
0
20
40
60
80
100
120
1 2 3 4 5 6 7 8 9
26
52
88
41
77
103
33
61
94
24.8
51.5
88.6
43.5
78.0
101.5
33.8
62.3
22.3Le
ak
ag
e R
ate
(MS
cf/
da
y)
Run
Actual Rate ILDS Prediction
ILDS Leakage Rate Prediction
PDGs at Citronelle Site
29
D-4-13 and/or D-4-14In-zone monitoringAbove-zone monitoringFluid sampling
D-9-11Neutron Logging
Primary Injector (D-9-7#2)Injection surveysPressureSeismicGroundwater
Backup Injector(D-9-9#2)Neutron loggingSeismicGroundwater
Characterization Well (D-9-8#2)Neutron loggingPressureFluid samplingSeismicDistributed Temperature
Model Plume Extent
Ref: ARI
Noise Analysis - PDGs
30
𝑁𝑁𝑁𝑁 = 𝑃𝑃𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 − 𝑃𝑃𝑓𝑓𝑓𝑓𝑎𝑎𝑎𝑎𝑓𝑓𝑓𝑓 → 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝐿𝐿𝑁𝑁𝐿𝐿𝑁𝑁𝐿𝐿 =1
𝑛𝑛 − 1�𝑁𝑁𝑓𝑓2𝑛𝑛
𝑓𝑓=1
1/2
Noise Level = 0.08 Psi Distribution = Normal (Gaussian)
De-noising Process
31
Training with De-Noised Data
32
Leakage Location(X) Leakage Location(Y)
Leakage Rate
33
0
0.2
0.4
0.6
0.8
1
0 1000 2000 3000
∆P(
Psi)
Time(hour)
0
0.2
0.4
0.6
0.8
1
0 500 1000 1500 2000 2500
∆P
(Psi
)
Time(hour)
De-noising
Summarization
Descriptive Statistics Mean 0.091532 Kurtosis -1.31344 Standard Error 0.004755 Skewness 0.029047 Median 0.091797 Range 0.194824 Mode 0 Minimum 0 Standard Deviation 0.061636 Maximum 0.194824
Sample Variance 0.003799 Sum 15.37744
Leakage Location
Leakage Rate
Noisy Pressure Data
The Interface Development
35
Accomplishments to Date
• Geological model was developed. • Reservoir Simulation Model was developed. • Impact of Relative Perms of Trapping Mechanism was
determined • Seal Quality and Integrity was studied • Sensitivity analysis was performed • Reservoir Simulation Model was history matched • Intelligent Leakage Detection System (ILDS) was designed
and developed. – Initial Design – Validated for Simple Reservoir System – Validated for Simple Leakage System
• High Frequency data was cleansed and summarized • ILDS interface was developed
Summary
• Key Findings: - Location and amount of CO2 leakage can be detected and quantified, rather quickly, using continuous monitoring of the reservoir pressure. - Pattern recognition capabilities of Artificial Intelligence and Data Mining may be used as a powerful deconvolution tool.
– Lessons Learned(proof of concept): - Development of an Intelligent Leakage Detection System (ILDS) is initiated for detection and quantification of CO2 leakage.
– Future Plans: - Increase the robustness of ILDS by: + Using history matched model + Examining impact of different boundary conditions, + Including more sources of leakage(like Cap rock Leakage) + Examining detection of simultaneous multiple leakages.
Bibliography List peer reviewed publications generated from
project per the format of the examples below • Journal, one author:
– Gaus, I., 2010, Role and impact of CO2-rock interactions during CO2 storage in sedimentary rocks: International Journal of Greenhouse Gas Control, v. 4, p. 73-89, available at: XXXXXXX.com.
• Journal, multiple authors: – MacQuarrie, K., and Mayer, K.U., 2005, Reactive transport modeling in fractured
rock: A state-of-the-science review. Earth Science Reviews, v. 72, p. 189-227, available at: XXXXXXX.com.
• Publication: – Bethke, C.M., 1996, Geochemical reaction modeling, concepts and applications: New
York, Oxford University Press, 397 p.
38
Appendix Benefit to the Program
• Program goals : – Develop technologies to demonstrate that 99 percent
of injected CO2 remains in the injection zones.
• Benefits statement: – This project is developing the next generation of
intelligent software that takes maximum advantage of the data collected using “Smart Fields” technology to continuously and autonomously monitor and verify CO2 sequestration in geologic formations. This technology will accommodate in-situ detection and quantification of CO2 leakage in the reservoir.
Appendix Project Overview:
Goals and Objectives • Goals and objectives in the Statement of Project:
– This project proposes developing an in-situ CO2 Monitoring and Verification technology based on the concept of “Smart Fields”. This technology will identify the approximate location and amount of the CO2 leakage in the reservoir in a timely manner so action can be taken and ensure that 99 percent of the injected CO2 remains in the injection zone.
– Success Criteria and Decision Points:
– There are a total of 10 milestones (and 4 sub-Milestone) in this project. – Decision points come at the end of quarters 4 (Milestone 2.2) and 15
(Milestone 6). At the decision points a “go” or “no go” decision on the continuation of the project is made based on the accomplishments of the project up to that point.
Appendix Organization Chart
Main Contributors (Research & Development): Alireza Haghighat, Alireza Shahkarami, Daniel Moreno, Najmeh Borzoui, and Yasaman Khazaeni. Full Time Research Associate: Vida Gholami,
Appendix Gantt Chart
Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Q13 Q14 Q15 Q16Program Management Task OneSite Selection Task Two
2.12.22.3
Base Model Development Task Three3.13.23.33.43.53.6
Sensitivity Analysis Task Four4.14.24.3
CO2 Leakage Modeling Task Five5.15.25.35.4
High Frequency Data Handling Task Six6.16.26.3
Pattern Recognition Analysis Task Seven7.17.27.37.47.57.6
Application to Homogeneous Reservoir Task Eight
History Matching Task Nine9.19.29.3
Application to Heterogeneous Reservoir Task Ten
Interface Development Task Eleven
Budjet Period 1 Budget Period 2Task Title
Project Tasks
Task 1: Program Management and Reporting Task 2: Site Selection Subtask 2.1: Establishing the Industrial Review Committee Subtask 2.2: Developing Site Selection Criteria Task 3: Reservoir Data Collection and Base Reservoir Model Construction Subtask 3.1: Selection of Reservoir Modeling Software
Task 2: Site Selection Subtask 2.3: Selecting a Site Task 3: Reservoir Data Collection and Base Reservoir Model Construction Subtask 3.2: Collect data
Task 3: Reservoir Data Collection and Base Reservoir Model Construction Subtask 3.3: Use Collected Data to Develop a Geological Model Subtask 3.4: Assessing the need for Up-scaling the Geological Model
Task 3: Reservoir Data Collection and Base Reservoir Model Construction Subtask 3.5: Import the Geological Model into the Flow Model Subtask 3.6: Flow Model Testing
Task 4: Sensitivity Analysis of the Reservoir Simulation Model Subtask 4.1: Building multiple heterogeneous porosity maps based on logs from existing wells in the formation. Subtask 4.2: Defining different Porosity Permeability correlations and building different geological realizations of the reservoir. Subtask 4.3: Comparing different realizations of the reservoir and ranking them based on injectivity.
Task 5: Simulating CO2 Leakage and Realistic Downhole Pressure Data Subtask 5.1: Simulation of CO2 Leakage
August 22, 2013
Task 5: Simulating CO2 Leakage and Realistic Downhole Pressure Data Subtask 5.2:High Frequency Data Streams Generation Task 6: Developing Techniques for Handling High Frequency Data Subtask 6.1: Processing of High Frequency Data Streams – Data Cleansing
Task 5: Simulating CO2 Leakage and Realistic Downhole Pressure Data Subtask 5.3:Transmission of Data from Modeled Pressure Gauges Task 7: Performing Pattern Recognition Analysis Subtask 7.1: Key Performance Indicators Using Fuzzy Set Theory
Task 5: Simulating CO2 Leakage and Realistic Downhole Pressure Data Subtask 5.4:Emulation of Field Data Using Data Stream Distortion Task 7: Performing Pattern Recognition Analysis Subtask 7.2: Data Partitioning for Neural Network Modeling
Task 6: Developing Techniques for Handling High Frequency Data Subtask 6.2: Processing of High Frequency Data Streams – Data Summarization
Task 6: Developing Techniques for Handling High Frequency Data
Subtask 6.3: Preparation of High Frequency Data for Pattern Recognition
Task 7: Performing Pattern Recognition Analysis
Subtask 7.4: Subtask 7.3: Neural Network Architecture Design
Task 8: Testing and Validation of CO2 Leak Detection in a Homogeneous Reservoir
Task 7: Performing Pattern Recognition Analysis
Subtask 7.4: Neural Network Training and Calibration Subtask 7.5: Neural Network Validation Subtask 7.6: Neural Network Model Analysis
Task 9: Integrating CO2 Injection and History Matching the Model
Subtask 9.2: In Situ CO2 Behavior Validation Subtask 9.3: Model Integrity Verification
Milestone Timelines Milestone log
Title Description Related task or subtask Completion Date
Budget Period 1
Milestone 1.1 Advisory Board Meeting Advisory board should get together for a meeting (or conference call) to select a site for the project. Subtask 2.1 End of First Quarter
Milestone 1.2 Site Selection A site must be selected for the project. Subtask 2.2, 2.3 End of Second Quarter
Milestone 2.1 Data collection Completion of geologic and production data collection Subtask 3.2 End of Third Quarter
Milestone 2.2 Completion of geological model Completion of geologic/geo-cellular model Subtask 3.3 End of Fourth Quarter
Milestone 2.3 Completion of the base model Completion and testing the base flow model Subtask 3.6 End of Fifth Quarter
Milestone 3 Sensitivity Analysis Completion of the sensitivity analysis on the reservoir model Subtask 4.3 End of Sixth Quarter
Budget Period 2 Milestone 4.1 CO2 Leakage Modeling Model realistic CO2 leakage from the formation Subtask 5.1 End of Eighth Quarter
Milestone 4.2 Downhole pressure modeling Model realistic real-time downhole pressure measurements. Subtask 5.2, 5.3, 5.4 End of Eleventh Quarter
Milestone 5 Handling High Frequency Data Developing techniques for handling high frequency data Subtask 6.1, 6.2, 6.3 End of Thirteenth Quarter
Milestone 6 Pattern recognition Completing pattern recognition analysis Subtask 7.1, 7.2, 7.3, 7.4, 7.5, 7.6 End of Fifteenth Quarter
Milestone 7 Application to Homogeneous system Completing of analysis and application to Homogeneous system Task 8 End of Fifteenth Quarter
Milestone 8 CO2 Injection Modeling Completion of modeling the CO2 injection. Subtask 9.3 End of Fifteenth Quarter
Milestone 9 Application to Heterogeneous system Completing of analysis and application to Heterogeneous system Task 10 End of Sixteenth Quarter
Milestone 10 Build Program Interface Completion of Software Package Task 11 End of Sixteenth Quarter
