Quantitative monitoring of CO2 injection at Sleipner using seismic full waveform inversion in the time lapse mode and rock physics modeling
IPGP - Îlot Cuvier
Soutenances de thèses
Carbon capture and sequestration is a technology to achieve a considerable deceleration of CO2 emission promptly. Since 1996 one of the largest CO2 storage projects is taking place at Sleipner in the Norwegian North Sea. In order to monitor injected CO2, time lapse seismic monitoring surveys have been carried out. Estimating subsurface parameters from the Sleipner seismic data is a challenging problem due to the specific geology of the storage reservoir, which is further complicated by injected CO2. Most seismic imaging methods enable only qualitative insights into the subsurface. Motivated by the need for a quantitative seismic monitoring of the injected CO2, I have applied 2D seismic full waveform inversion to seismic data sets from Sleipner from 1994 (baseline), 1999 and 2006 along three seismic lines to infer subsurface parameters and parameter changes in the storage reservoir. The P-wave velocity is the major parameter, as it is the most sensitive to CO2 injection. An energy preconditioning of the gradient has been implemented. The usual source wavelet calibration did not prove to be reliable. An alternative source calibration has been successfully applied. By comparing seismic images with inversion results, I found that using seismic images to locate CO2 accumulations in the subsurface may be misleading. The quantitative imaging approach using full waveform inversion resulted in a consistent evolution of the model parameter with time. Major reductions in P-wave velocity and hence the CO2 accumulations could be quantitatively imaged down to a resolution of 10 m. Observed travel time shifts due to CO2 injection are comparable to those derived from the inversion result. In order to estimate CO2 saturations, rock physical concepts have been combined and extended to arrive at a rock physical formulation of the subsurface at Sleipner. I used pseudo Monte Carlo rock physics modeling to assess the influence of lithologic heterogeneity on the CO2 saturations as well as to generate pseudo well logs to estimate confidence intervals of the inverted parameters. The rock physics modeling has been used to relate inverted parameters to CO2 saturations. The injected CO2 is buoyant. The highest CO2 saturations are in the upper half of the storage reservoir, but not necessarily at the top. Non-uniqueness of the saturation maps associated with the density scenario has been assessed. As a result, the distribution of the maximum saturation values remains the same. The quantity of dissolved CO2 in the reservoir water is a key parameter from both a security and optimization point of view. A quantitative estimation of dissolved CO2 by seismic means has not been undertaken yet to our knowledge. Based on the seismic inversion result of a seismic line, I found that along the line at least 20% of the injected CO2 mass was dissolved in 2006, after 10 years of injection. Such a high value indicates enhanced solubility trapping, which is very advantageous for storage safety at Sleipner. The results of this work represent a further step towards ultimate goals of quantitative monitoring, such as the estimation of the injected CO2 in-situ volume.