Application of Artificial Intelligence for Mechanistic Modeling and Probabilistic Forecasting of Hybrid Low Salinity Chemical Flooding

After nearly thirty years of research and development, it is now widely agreed that Low Salinity Waterflooding (LSW) provides better oil recovery than High Salinity Waterflooding (HSW). Past studies also showed that there are significant advantages in combining LSW with other conventional EOR methods such as chemical flooding (polymer flooding and surfactant flooding) or miscible gas flooding to benefit from their synergies and to achieve higher oil recovery factor and project profit. This paper presents a study of Hybrid Low Salinity Chemical Flooding as a novel EOR approach with: (1) development of hybrid EOR concept from past decades; (2) implementation of an efficient modeling approach utilizing artificial intelligent technology for mechanistic modeling of these complex EOR processes; (3) systematic validation with laboratory data; and (4) uncertainty evaluation of LSW process at field scale. The phase behavior of an oil-water-microemulsion system was modeled without the need of modeling type III microemulsion explicitly. The approach has been successfully applied to model both conventional Alkaline-Surfactant-Polymer (ASP) flooding and emerging EOR processes (LSW, Alkaline-CoSolvent-Polymer, and Low-Tension-Gas Flooding). The new development allows the mechanistic modeling of the benefits of combining LSW and chemical EOR. One of the main challenges for mechanistic modeling of these hybrid recovery processes is that several factors, e.g. polymer, surfactant, and salinity, can change the relative permeability simultaneously. To overcome this problem, Multilayer Neural Network (ML-NN) technique was applied to perform N-dimensional interpolation of relative permeability. The model was validated with coreflooding data and the effectiveness of hybrid processes were compared with conventional recovery methods. The proposed model showed good agreements with different coreflooding experiments including HSW, LSW, and Low Salinity Surfactant flooding (LSS). This model efficiently captures the complex geochemistry, wettability alteration, microemulsion phase behavior, and the synergies occurring in these hybrid processes. Results indicated that LSS is an economically attractive hybrid EOR process since it increases the ultimate recovery factor compared to the conventional approaches and reduces surfactant retention. Bayesian workflow using ML-NN algorithm is efficient to capture the uncertainties in history matching and production forecasting of LSW.