HR: 1340h
AN: GC13B-1238 [Abstracts]
TI: A Dynamic Programming Model for Optimizing Frequency of Time-Lapse Seismic Monitoring in Geological CO2
Storage
AU: * Bhattacharjya, D
EM: debarunb@stanford.edu
AF: Dept. of Management Science & Engineering, Stanford University, Stanford, CA 94305
United States
AU: Mukerji, T
EM: mukerji@stanford.edu
AF: Stanford Rock Physics Lab, Dept. of Geophysics
Stanford University, Stanford, CA 94305
United States
AU: Mascarenhas, O
EM: oscarmas@stanford.edu
AF: Dept. of Management Science & Engineering, Stanford University, Stanford, CA 94305
United States
AU: Weyant, J
EM: weyant@stanford.edu
AF: Dept. of Management Science & Engineering, Stanford University, Stanford, CA 94305
United States
AB:
Designing a cost-effective and reliable monitoring program is crucial to the success of any geological CO2 storage project.
Effective design entails determining both, the optimal measurement modality, as well as the frequency of monitoring the site.
Time-lapse seismic provides the best spatial coverage and resolution for reservoir monitoring. Initial results from Sleipner
(Norway) have demonstrated effective monitoring of CO2 plume movement. However, time-lapse seismic is an expensive
monitoring technique especially over the long term life of a storage project and should be used judiciously. We present a
mathematical model based on dynamic programming that can be used to estimate site-specific optimal frequency of time-lapse
surveys.
The dynamics of the CO2 sequestration process are simplified and modeled as a four state Markov process with transition
probabilities. The states are M: injected CO2 safely migrating within the target zone; L: leakage from the target zone to the
adjacent geosphere; R: safe migration after recovery from leakage state; and S: seepage from geosphere to the biosphere. The
states are observed only when a monitoring survey is performed. We assume that the system may go to state S only from state
L. We also assume that once observed to be in state L, remedial measures are always taken to bring it back to state R.
Remediation benefits are captured by calculating the expected penalty if CO2 seeped into the biosphere. There is a trade-off
between the conflicting objectives of minimum discounted costs of performing the next time-lapse survey and minimum risk of
seepage and its associated costly consequences. A survey performed earlier would spot the leakage earlier. Remediation
methods would have been utilized earlier, resulting in savings in costs attributed to excessive seepage. On the other hand,
there are also costs for the survey and remedial measures.
The problem is solved numerically using Bellman's optimality principal of dynamic programming to optimize over the entire
finite time horizon. We use a Monte Carlo approach to explore trade-offs between survey costs, remediation costs, and survey
frequency and to analyze the sensitivity to leakage probabilities, and carbon tax. The model can be useful in determining a
monitoring regime appropriate to a specific site's risk and set of remediation options, rather than a generic one based on a
maximum downside risk threshold for CO2 storage as a whole. This may have implications on the overall costs associated with
deploying Carbon capture and storage on a large scale.
DE: 1225 Global change from geodesy (1222, 1622, 1630, 1641, 1645, 4556)
DE: 1699 General or miscellaneous
DE: 6304 Benefit-cost analysis
DE: 6329 Project evaluation
SC: Global Climate Change [GC]
MN: Fall Meeting 2005