CO2 mass estimation visible in time-lapse 3D seismic data from a saline aquifer and uncertainties

Thursday, 18 December 2014
Alexandra Ivanova1, Stefan Lueth1, Peter Bergmann1 and Monika Ivandic2, (1)Helmholtz Centre Potsdam GFZ German Research Centre for Geosciences, CGS, Potsdam, Germany, (2)Uppsala University, Uppsala, Sweden
At Ketzin (Germany) the first European onshore pilot scale project for geological storage of CO2 was initiated in 2004. This project is multidisciplinary and includes 3D time-lapse seismic monitoring. A 3D pre-injection seismic survey was acquired in 2005. Then CO2 injection into a sandstone saline aquifer started at a depth of 650 m in 2008. A 1st 3D seismic repeat survey was acquired in 2009 after 22 kilotons had been injected. The imaged CO2 signature was concentrated around the injection well (200-300 m). A 2nd 3D seismic repeat survey was acquired in 2012 after 61 kilotons had been injected. The imaged CO2 signature further extended (100-200 m). The injection was terminated in 2013. Totally 67 kilotons of CO2 were injected. Time-lapse seismic processing, petrophysical data and geophysical logging on CO2 saturation have allowed for an estimate of the amount of CO2 visible in the seismic data. This estimate is dependent upon a choice of a number of parameters and contains a number of uncertainties. The main uncertainties are following. The constant reservoir porosity and CO2 density used for the estimation are probably an over-simplification since the reservoir is quite heterogeneous. May be velocity dispersion is present in the Ketzin reservoir rocks, but we do not consider it to be large enough that it could affect the mass of CO2 in our estimation. There are only a small number of direct petrophysical observations, providing a weak statistical basis for the determination of seismic velocities based on CO2 saturation and we have assumed that the petrophysical experiments were carried out on samples that are representative for the average properties of the whole reservoir. Finally, the most of the time delay values in the both 3D seismic repeat surveys within the amplitude anomaly are near the noise level of 1–2 ms, however a change of 1 ms in the time delay affects significantly the mass estimate, thus the choice of the time-delay cutoff is crucial. In spite of these uncertainties, the close agreement (over 85%) between the injected and observed amount is encouraging for quantitative monitoring of a CO2 storage site using seismic methods. The missing 15% (maximum) is completely within the simulated dissolved CO2 quantity.