Research objectives and hypotheses

Seismic techniques provide the key monitoring capability for verifying site performance at the wide spatial scales required (tens of square kilometres and upwards) and will form the core of the proposal. Our industrial partners, BP and Statoil, have provided access to monitoring datasets from the three global large-scale CO2 storage operations: Sleipner, In Salah and Snøhvit. These include active (time-lapse 3D, multi-azimuth data) and passive seismics, ground displacements from satellite interferometry and reservoir pressure measurements. Monitoring results from these sites are outlined briefly below.

Sleipner has stored more than 14 million tonnes of CO₂ in the Utsira Sand, a very large, shallow, saline aquifer that perhaps typifies the 'perfect' reservoir in terms of storage capacity and monitorability (Arts et al., 2008). The CO₂ plume shows striking reflectivity (Figure 1a) and large time shifts. High resolution mapping of individual spreading CO₂ layers has been history matched against both numerical (Chadwick and Noy, 2010), and analytical flow models (Bickle et al., 2007). More recently, statistical analysis of very small time shifts has been used to constrain pressure increase in the reservoir (Chadwick et al., 2012).

At In Salah, more than three million tonnes of CO₂ have been stored in a complex fractured aquifer. Seismic time shifts are evident, aligned to likely fracturing, but diagnosis of these, in terms of fluid saturation or pressure changes, is difficult using conventional analysis. Passive seismics have shown evidence of induced micro-seismicity and satellite interferometry has revealed striking ground displacements above the injection wells (Figure 1b) that have been related to mechanical inflation and fracturing of the reservoir (e.g. Morris et al., 2011).

Snøhvit has stored 1.2 million tonnes of CO₂ in a very deep (approx. 2800 m), faulted aquifer. However, during injection, downhole pressures rose significantly and injection was terminated in the preferred storage reservoir (Gilding et al., 2012). Intriguing time-lapse 3D seismic time shifts and reflectivity changes appear to show both CO₂ saturation and pressure signatures in the reservoir (Figure 1c). Discriminating between these effects remains challenging.

Figure 1. (a) View of the Sleipner plume in 2006 showing the brightly reflective CO₂ plume beneath the reservoir topseal (coloured surface). (b) Shaded contour map of millimetre-scale surface displacements above the In Salah injection wells. (c) Reflectivity changes at the level of the storage reservoir at Snøhvit. Strong changes (red colours) close to the injection point likely correspond to the CO₂ plume; weaker, more spatially extensive changes may indicate pressure increase in the surrounding aquifer [black areas correspond to faults]. N.B. white discs correspond to CO₂ injection points/wells and all views and maps show areas several kilometres across.

Hypotheses

Seismic responses, essentially signal amplitude changes, frequency changes and time shifts, arise from a larger number of causative processes. Disentangling these, particularly fluid saturation and pressure effects, is notably challenging. Our research plans to improve discrimination between and characterisation of fluid changes, pressure changes and induced geomechanical effects, and rests upon four key hypotheses:

1. hydromechanical modeling of CO₂ injection can be used to predict changes in seismic properties, induced seismicity and ground displacements and fracture initiation
2. statistical analysis of very small seismic time shifts can be used to detect and map pressure changes over a range of potential storage reservoir types
3. azimuthal variation in seismic attributes can be used to map fracture patterns and fluid saturation changes within storage reservoirs
4. the thickness, velocity and saturation properties of thin layers of CO₂ can be constrained by simultaneous application of different analytical procedures