Multi-objectives Assisted Inversion of Surfactant-polymer Flooding Experiments on Sandstone Cores

Reservoir modeling is a powerful tool to thoroughly assess chemical EOR performances and to design the economically optimal injection sequence (chemical concentrations, slugs sizes). A careful model calibration from laboratory tests is necessary to provide meaningful results. However, while building a model from laboratory tests requires a large amount of data, laboratory tests only provide a limited number of oil recovery experiments. Some essential data besides remain costly or time-consuming to acquire: the set of relative permeabilities for different capillary numbers, interfacial tension dependency to salinity and surfactant concentration, chemicals adsorption, amongst others. The calibration process is thus not straightforward and typically results in underconstrained models. To increase the model robustness, we propose to carry out a multi-objectives assisted calibration, further tested on an additional differentiating experiment. The multi-objectives feature refers to the calibration of various production data acquired from a consistent series of experiments. The assisted calibration is preferred over the common trial and error approach both to quantitatively appreciate the calibration quality via the objective function and to exhaustively scan the uncertainty range in input parameters. The calibration is done using our Assisted History Matching software, CougarFlow. The methodology is applied to five surfactant-polymer displacements performed on sandstone cores, differing from the surfactant slug concentration, its size and the brine salinity. The experiments set has been built to provide a wide range of tertiary recoveries, from 50% to 90% of residual oil in place, which is wider than found in similar studies. The oil recovery, the pressure difference and the adsorbed mass of surfactant has been history matched with a limited number of adjustable parameters among which the surfactant adsorption, the water-oil interfacial tension and the extrapolation of Corey exponents along flooding history. The adjustable parameters effect was thoroughly explored by more than 1,000 simulations to finally get a model which successfully reproduces the main features of the experiments. As expected, calibration on a few experiments gives more confidence in the model than on a single experiment. Its ability to reproduce the physics occurring in the core and to further predict the optimal sequence has been reinforced by the additional differentiating experiment which segregated between two consistent models. This additional experiment specifically better constrained the surfactant adsorption. Additionally, this study revealed the positive effect for tertiary oil recovery of a salinity decrease between the main surfactant-polymer slug and the polymer postflush.