This paper reports core flood interpretation for two scenarios: oil displacements by low salinity water (LSW) and by LSW combined with a boosting agent, di-ethyl ketone (DEK). The target reservoir was an offshore carbonate oil field in UAE. In the laboratory scale, the two scenarios showed notable incremental oil recovery compared with the result of sea water (SW) injection. Therefore, the objective of this work was to derive the representative relative permeability curves and the range of their uncertainties in order to incorporate these promising results into field scale simulation studies, e.g., designing a field pilot test.
The interpretation was conducted using a one-dimensional black oil simulation model in conjunction with the algorithm of Ensemble Smoother with Multiple Data Assimilation (ES-MDA). Parameters related to relative permeability were treated as variables, and those of the initial ensemble members were decided by Latin Hypercube Sampling from a uniformly distributed parameter space. Subsequently, the parameters in ensemble members were iteratively updated by the ES-MDA algorithm so that the misfit between the observed data and the calculated results was reduced. Finally, the representative relative permeabilities and the range of their uncertainty were determined from a misfit-degree viewpoint. The simulation model contained the salinity dependent relative permeability controlled by the weighting factor that is the function of salinity. In this study, the weighting factor was modelled based on the results of the fluid-fluid interaction test, which was conducted as the initial screening to identify oil reservoirs suitable for LSW flooding. This weighting factor setting is consistent with the theory that fluid-fluid interaction is a key mechanism of oil recovery in LSW flooding.
The conclusion of this study is summarized as follows: The representative relative permeability curves were obtained by the ES-MDA algorithm more effectively and systematically than manual history matching because it did not require trial and error to find the model that reproduced the experimental results.In addition to the efficiency enhancement, the ES-MDA algorithm iteratively reduced the misfit of the ensemble members and gave the range of uncertainty associated with the interpretation. This workflow is efficient in that history matching and its uncertainty quantification are conducted simultaneously. This evaluation will be utilized to make subsequent simulation studies more rigorous.In comparison with the results of pure LSW and DEK-assisted LSW, the latter changed the wettability nature to more water-wet, which is indicative of the boosting effect.
Though various LSW EOR-boosting additives have been experimentally investigated, there has been no study that covered the numerical interpretation of DEK as the EOR-boosting additive. This paper is motivated to conduct the numerical interpretation of the DEK-assisted LSW core flood experiment. The results indicated the efficacy of the additive boosting from the numerical perspective. This study also demonstrated the effectiveness of the ES-MDA algorithm, which handles multiple models, to conduct uncertainty evaluation associated with the core flood interpretation and future prediction.