Whitney Trainor, Jef Caers, & Tapan Mukerji 21 st SCRF Meeting May 9, 2008

Transcription

1 Whitney Trainor, Jef Caers, & Tapan Mukerji 21 st SCRF Meeting May 9, 2008

2 Value of Information driven by interplay of 3 components 1. Reliability of Information Seismic ability to resolve reservoir size 2. Magnitude of decision that can be impacted Different possible asset value (size of reservoir)? 3. Chance of making a poor/suboptimal decision for the true state of nature Proposed actions different suburface scenarios? (Coopersmith, 2006)

3 $450 M Value PRIOR No Information How to develop field? No action Plan A Plan B $0 M $450 M $300 M Reservoir Size? 300 Million Barrels 150 Million Barrels 300 Million Barrels 150 Million Barrels 300 Million Barrels $ 0 M $ 0 M $ 600 M $ 300 M $ 750 M 150 Million Barrels $ -150 M Value Info $525 M Decision Node Interpret 300 Possibility Node Million Barrels Value Node No action Plan A Plan B $ 0 M $ 600 M $ 750 M With Perfect Information Interpret 150 Million Barrels No action Plan A Plan B $ 0 M $ 300 M $ -150 M

4 !" Petroleum Reservoir Models Predict future production Optimize oil recovery Uncertainty measure necessary Heterogeneity Freshwater Basin Models Predict drinking/ irrigation water supplies / reserves Optimize placement of supply wells or recharge wells Uncertainty measure Heterogeneity

5 !"# Flow property heterogeneity affect spatial decisions Managed Recharge (Injection) Seawater Intrusion Seawater - Constant Pressure SALT (lb) 12 0 Natural Recharge - Constant Flux Subsurface information many challenges! Acquisition & Interpretation errors

6 $% # Given that managed recharge will become a customary or necessary practice: Does the risk of poor predictions justify the cost of acquiring higher quality data? How can we quantify the value of geophysical data for hydrogeologic models? How does geophysics change our present view of the uncertainty of subsurface heterogeneity?

7 # Modeled after coastal Californian water basin Decision Locations for managed recharge Are costs of recharge justified given uncertainty of heterogeneity? Value = Gridblocks below salinity threshold translated to potential $ in crop production Seawater - Constant Pressure Natural Recharge - Constant Flux

8 &! 1. Reliability of Information Airborne Time-domain Electromagnetic Data Interpreting channel orientation 2. Magnitude of decision that can be impacted Perform Managed Recharge? If yes, where? Value at time = 10 years 3 channel scenarios 3. Chance of making a poor/suboptimal decision for the true state of nature Perform No Recharge (No Initial Cost) Different Recharge Locations (Initial Cost)

9 " ' 1. Create prior model space fluvial realizations (3 scenarios) 2. Geophysics forward and inverse modeling 3. Measure of reliability - Image Analysis (classified into the 3 scenarios) 4. VOI calculation - flow simulations of managed recharge

10 $&() Original Angle Maps Training Image Zone 1 Zone 2 Zone 1 Zone 1 Zone 2 Zone 2 Northeast East Southeast Zone 3 Zone 3 Zone 3 Sand Clay

11 *+ For all realizations (from 3 Scenarios) Sand Clay Invert for Resistivity Map m Northeast Mixed Southeast Add overburden and basement Lithology to Resistivity: modified Archie s Law Compute Airborne EM Forward Response Add Background Noise (proportional t -0.5 )

12 ,) Inverted Resistivity Image Image analysis to obtain map with the direction (angle) of maximum correlation m Northeast East Southeast

13 # / - +. /0#0 Zone 1 Zone 2 Zone 3 #, Northeast East Southeast Zone 1 Zone 2 Zone 3

14 1),(2 Mixed Interpreted Scenario Northeast Southeast True Scenario Mixed Northeast 64% 2% 0% 88% 36% 10% Southeast 0% 2% 98% Frequency calculated from 125 realizations & inverted images

15 ) Pumping Schedule (in stb/day) 4 Options (Branches): 1. No recharge 2. Recharge 1 (central) 3. (north) 4. Recharge 3 (south) Cost of recharge: $4 Million capital $100,000 operational Recharge No K 23K 7K 9K 16KRECHARGE 7K 2 9K 23K RECHARGE 1 16K 7K 12K RECHARGE 16K3 7K 7K 23K 16K 7K 7K SALT (lb) 12 0

16 No Information Value Info $12.8 M With Perfect Information Value PRIOR $12.47 M Mixed Interpreted Northeast Interpreted Southeast Interpreted $12.47M No Recharge $11.6M Recharge 1 $12.1M $11.2M Recharge 3 No recharge Recharge 1 No recharge $12.15 M Recharge 1 $11.1 M Recharge 3 Recharge 3 No recharge Recharge 1 Recharge 3 $12.9 M $11.3 M $13.66 M $13.8 M $13.9 M $12.8 M $11.6 M $ 9.89 M $ 9.66 M 33.3% each $ 9.44 M

17 Value PRIOR $12.47 M 22% VOI = Buy Imperfect Information $ 200K Mixed Interpreted $2.8M =$12.9M x 0.22 No Recharge No Recharge $12.2M $12.9M $13.6M Probability Interpretation is Correct: 97 % 3 % 0 % 30% Northeast Interpreted $4.14 =$13.8M x 0.3 $13.8M 0 % 97.8 % 2.2 % Value_FE = $ M 48% Southeast Interpreted No Recharge $11.9M $10.8M 25 % 7 % 68 % $5.7 M =$11.9M x 0.48

18 , 3 Injectors at each of the previous 3 locations How does this effect VOI? RECHARGE 2 RECHARGE1 RECHARGE3

19 No Information Value Info $23.1 M With Perfect Information Value PRIOR $23.1 M Mixed Interpreted Northeast Interpreted Southeast Interpreted $12.47M No Recharge $23.1M Recharge 1 $22.2M $19.5M Recharge 3 Recharge 1 Recharge 3 No recharge Recharge 1 No recharge $12.15 M Recharge 3 No recharge Recharge 1 Recharge 3 $22.2 M $21.1 M $18.5 M $13.66 M $25.7 M $25.3 M $24.1 M $11.6 M $ 21.4 M $20.2 M 33.3% each $ 15.8 M

20 ) Reliability of Information Perfect Information measure is unrealistic for earth sciences Subsurface scenario - quantifiable way to measure misinterpretation rate Risk of suboptimal decision No value in information if optimal choice is the same for the heterogeneity represented in the tree

21 " ' Neutral Risk Attitude Equal risk-weighted Risk-adverse attitude More sophisticated ways of defining the 3 branches P10, P50 and P90 Land based TEM surveys 3D Model flow and geophysical modeling Apply workflow to real hydrogeologic case Workflow could be applied to petroleum case More rigorous economics