Defining absolute permeability for CO2 injection projects: scope for increased economic storage capacity

Single phase brine permeability, also reported as “absolute brine permeability”, is typically used as reference value in CO2 drainage experiments to normalise relative permeability data. However, using the brine absolute permeability to normalise relative permeabilities creates an unexpected problem in the event that the (Klinkenberg-corrected) gas permeability is larger than the absolute brine permeability. To explain: late in the displacement of brine by a non-wetting phase, at low brine saturation, the effective gas permeability will necessarily approach the single phase gas permeability giving rise to gas relative permeabilities above unity when normalised to the absolute brine permeability. This indeed also has been observed in the laboratory. In itself, such relative permeability values are a viable result. However, commercial reservoir simulators do not allow relative permeabilities to be input with values above unity. In our experience, in line with data reported by many authors, the absolute gas permeability often is larger than the brine permeability. Low brine saturations particularly do occur close to the injector in the field. Therefore, the issue around relative permeabilities being above unity is very relevant estimating CO2 well injectivity and therefore for simulating CCS field behaviour on the whole.
To determine whether lower absolute brine permeabilities are a laboratory artefact or representative of field conditions, we conducted flow experiments on ceramic plugs and clean lithology outcrop Obernkirchener (OBKN) sandstone samples and conducted a rigorous statistical analysis on the results. We found the brine absolute permeabilities reproducible and insensitive to brine composition and equilibration time of the plugs in brine. Ceramic plugs exhibited brine permeabilities equal to the gas permeabilities with high statistical probability. The OBKN samples had brine permeabilities with a factor of around 0.7 lower than the gas permeabilities. We attribute this behaviour to ionic forces active between the rock surface and the ions in the brine.
Subsequently, we adapted the SCAL simulator SCORES to accept relative permeabilities larger than 1 and found that the generated production, pressure drop and saturation profiles with input normalised to the brine permeability, duplicated the profiles generated with the relative permeabilities normalised to the gas permeability. This tallies with what one finds analysing the flow equations in terms of re-normalising relative permeabilities. Therefore, our recommendation is to use the Klinkenberg-corrected gas permeability as the absolute permeability to normalise CO2-brine relative permeabilities. We show that this may well have a positive effect on the economic storage capacity.