Monitoring Geological CO2 Storage: Quantitative CO2 Prediction with Uncertainty from Physical Modeling and Multiple Time-Lapse Data Types

The project's main goal is to develop new and improved methods for quantitative prediction of the distributions of CO2 in subsurface storage sites. We build statistically consistent models that take into account uncertainty at all levels, and use geophysi cal modeling and multiple time-lapse data sources. The main goal is being achieved by combining the partners' joint expertise on theoretical as well as experimental geophysical research, and stochastic modeling and uncertainty handling. We have develo ped the theoretically framework for how to join several data types, each sampled at various time steps over a period of years, with an underlying time dependent rock physics model. Uncertainties in the rock physics model parameters as well as in the sampl ed data are consistently taken into account. This ensures that data inversion by the means of this model will provide both predictions and estimates of the prediction uncertainties. The theoretical model is documented in internal work documents. The fu ll framework has further been verified in a synthetic case study. In this case study we start with a stochastic description of the reservoir parameters, and use this model to generate synthetic data, for the base case (before injection starts) and two mon itoring surveys. We then apply our inversion to assess the underlying reservoir properties. The results of the synthetic case study were presented as a poster at the CLIMIT Summit 2013. The results for Gravimetric inversion was presented on the EAGE in Amsterdam in June 2014. The methodology has further been tested on a Sleipner data. Late November last year a project meeting was held. At this meeting an analysis of the Sleipner data was presented, this included prediction of the CO2 saturation. Th e results were discussed and the partners gave relevant feedback and the finalizing work was agreed on. An updated inversion for the injection region was presented in project meeting march this year. The results are promising. The general description o f the rock physics library has been a major part of the collaboration between UiB and NR. This has resulted in a highly general framework for integration of rock-physic models into the software prototype for 4D joint inversion. The framework is such tha t it is easy to include new rock models and easy to combine existing rock-models to make complex composite rocks. The few rock models that have been implemented so far are quite basic, additional models will be included if this is required to obtain bett er rock physics models of the geological formation where the CO2 is injected from the Sleipner platform. The work is documented in an internal note, and there is ongoing work to make this into a publication. The collaboration with Stanford has been fr uitful and has resulted in a paper that has been published in Geophysics (top level journal) in 2013. The topic of the current collaboration is to investigate whether CO2 dissolve the minerals in the rock where it is injected, and to make mathematical mo dels for how this will influence the rock stiffness. This collaboration will continue after the project is finished. Statoil has provided the seismic data (Seismic amplitudes, RMS velocities, horizons, gravimetric) that are used for the inversion. Wel l data are used to calibrate the rock-physics model. To improve the confidence in the model, Statoil has gotten partners consent to release two additional wells going through the same geological formation as where the CO2 is injected. Carbon Capture and Storage is currently a major concern for the Norwegian energy and petroleum industry, and remains a key component in the plan of IEA to for reach the 2degree target. With CO2 storage gaining international focus, trustworthy uncertainty estimates base d on physical modeling and stringent mathematical reasoning will be increasingly important. The competitive edge of the Norwegian industry is facilitated by this project.

The project's main goal is to develop new and improved methods for quantitative prediction of the distributions of CO2 in subsurface storage sites. We will build statistically consistent models that take into account uncertainty at all levels, and use geo physical modeling and multiple time-lapse data sources. The main goal will be achieved by combining the partners' joint expertise on stochastic modeling and uncertainty handling, and theoretical as well as experimental geophysical research. Monitoring CO 2 volumes is a challenge. Main geophysical challenges include establishing proper rock physics relations between seismic velocity and saturation; interaction between injected CO2 and original fluid-rock; and chemical changes to the host rock. From a stati stical point of view it is a challenge to consistenly combine all elements of the rock physical models and appropriate data forward models into an overall scheme for 4D data inversion. We will develop appropriate rock physics relations to explain the sei smic effects of injection of CO2 in subsurface rocks of various reservoir qualities; develop a library of realistic rock physics models suitable for CO2 ; develop consistent stochastic models that jointly take into account various time-lapse data types, a nd find spatially and temporally correlated seismic parameters constrained by data; develop statistical inverse rock physics models; develop a software prototype for 4D joint inversion of multiple data sources; test the new methods on various synthetic sc enarios; and carry out a case study with real data from Statoil's CO2 storage project at Sleipner. Carbon Capture and Storage is currently a major concern for the Norwegian energy and petroleum industry. With CO2 storage quickly gaining international foc us, trustworthy uncertainty estimates based on physical modeling and stringent mathematical reasoning will be increasingly important. The competitive edge of the Norwegian industry is facilitated by this project.