An industrial-scale CO2 storage project has been in operation in Algeria since 2004 as part of the In Salah Gas Joint Venture (comprising BP, Sonatrach and StatoilHydro). CO2 from several fields within the development is stripped from the production stream (using the amine gas sweetening process) and injected into the aquifer several km away from one of the gas reservoirs (the Carboniferous sandstone at Krechba). This storage site is an important demonstration case used by a number of research institutes, within the European Union (especially the CO2ReMoVe research consortium) the USA and elsewhere, to better understand CO2 storage processes and to demonstrate and monitor the viability of Carbon Capture and Storage (CCS).
Verification of long-term CO2 storage at this site involves a number of new challenges for understanding and predicting rock-fluid interactions and fluid mobility in CO2-aqueous-hydrocarbon systems. We summarise our current understanding of this site and show preliminary forecasts of the system performance, identifying the key reservoir, caprock and fluid flow characteristics. Six performance domains are identified for which expected volumes as a function of time may be estimated: i) CO2 solution and precipitation in the aquifer, ii) CO2 storage as a gas phase in the aquifer, iii) CO2 storage in hydrocarbon reservoir, iv) CO2 storage in the caprock, v) possible CO2 leakage through the caprock, and vi) possible CO2 leakage through wells.
In terms of the reservoir and cap-rock characterisation, a key challenge is the estimation of permeability in mudstones and strongly-cemented sandstones within a Carboniferous extensional basin, which was then inverted and uplifted during the Hercynian and Alpine orogenic phases. Under the present-day stress field, the Carboniferous rock sequence displays sets of open fractures which can be shown to have influenced the gas flow performance during 4 years of injection and production, and which are likely to be a key factor in forecasting the long-term performance.
Using fractured rock characterisation methods (image logs, mud-loss data, and core analysis) we construct effective property models and discrete fracture network models as a basis for flow performance. Flow modelling is performed using a variety of complimentary tools, including an invasion percolation tool (MPath), a compositional reservoir simulator (Eclipse 300) and single and multi-phase fracture-flow tools. Each storage and leakage scenario is modelled to investigate and illustrate the likely long-term destinations of CO2.
Given the large uncertainties in rock permeability and fluid-rock interactions, it is argued that a probabilistic approach, using plausible alternative scenarios, is the best way of demonstrating acceptable risks and long-term viability of CO2 storage at this site.