Monitoring and Modelling CO2 Injection at Sleipner

Abstract

CO2 has been injected into the Utsira Sand at Sleipner since 1996, with more than 11 million tonnes currently in the reservoir. Seismic monitoring surveys to follow the migration of the CO2 in the reservoir have been carried out in 1999, 2001, 2002, 2004, 2006 and 2008. The CO2 plume is imaged on the seismic data as a prominent multi-tier feature, comprising a number of bright sub-horizontal reflections, growing with time and interpreted as arising from up to nine discrete layers of high saturation CO2, each up to a few metres thick. Quantitative seismic interpretation of the time-lapse data has included synthetic seismic modelling to derive CO2 distributions in the reservoir (Skov et al, 2002, Arts et al., 2004). Convolution-based modelling has shown that seismic reflection amplitudes are broadly related to layer thickness via a tuning relationship. However acquisition geometry, lateral velocity changes, mode conversions and intrinsic attenuation are all likely to affect amplitudes, especially at the lower levels in the reservoir, and need to be incorporated within a rigorous quantitative analysis. Moreover, the predicted response heavilly relies on a rock physics model (Arts et al., 2004) to convert simulated saturation data to seismic impedance data. In this study we present a reservoir simulation sensitivity study based on the existing models of the Utsira Sand to mimic the spreading behavior of the CO2 as observed on the time-lapse seismic data. The transmissibility of each intra-reservoir shale layer has been estimated by matching the simulated quantity of CO2 with the seismically observed (estimated) quantity of CO2 per layer. The corresponding spreading patterns per level have been compared to the seismically observed accumulations.