An Efficient Workflow to Quantify the Value of Information and Uncertainty Reduction from Time-Lapse Seismic Data for Monitoring Optimization and Reservoir Characterization in CCS Projects

Design of time-lapse (4-D) seismic monitoring in CO2 sequestration projects should be optimized not only by regulations and risk management but also by accounting for reservoir characterization and the value-of-information (VOI). However, it is challenging to quantify posterior distributions of reservoir parameters and project values given multiple possibilities of future information. This study presents an efficient workflow to analyze the VOI of 4-D seismic data and the expected uncertainty reduction resulting from these data. Our workflow adopts Ensemble Variance Analysis (EVA), which utilizes covariance matrices to estimate the posterior probability distributions of reservoir parameters and project values after assimilating 4-D seismic data with prior models. Flow simulations need to be run only for prior models, and thus no model calibration to data realizations is required. Additionally, incorporating a parameterization technique can reduce dimensions of 4-D seismic data and improve computational efficiency. The posterior distributions of project values from EVA are further utilized in decision tree analysis to determine the VOI of 4-D seismic data. We demonstrated robustness and efficiency of our proposed workflow using an illustrative case example. The posterior variances of reservoir parameters and project values estimated from EVA were validated by comparing with those estimated from the posterior models that were actually history-matched to each data realization. For the presented CO2 sequestration scenario, it was shown that the future acquisition of 4-D seismic data exhibited uncertainty reduction in reservoir parameters and the VOI which corresponds to improvement of project values by updating selection of decision alternatives based on additional information. Furthermore, optimal timings for future seismic surveys that maximize the VOI were identified through this quantitative assessment, which offers essential insights for justifying and improving a monitoring plan. The novelty of this approach is that flow simulations are needed only for prior models to estimate the posterior distributions of model parameters and project values after assimilating 4-D seismic data, which enables efficient analysis of the VOI. This offers significant time savings as calibrating prior models to each data realization is not required. Incorporating a parameterization technique to reduce dimensions of 4-D seismic data also makes it feasible for field-scale application.