Performance prediction model of Miscible Surfactant-CO2 displacement in porous media using support vector machine regression with parameters selected by Ant colony optimization

• We determined dimensionless scaling groups that characterize Miscible Displacement in heterogeneous media.
• The relationships between Recovery efficiency and each scaling groups have been investigated using SVM.
• Ant Colony Optimization has been used to determine SVM parameters.
• ACO has better performance than GA, PSO and ANN to optimize SVM parameters for desired recovery process.

Hybrid system is a potential tool to deal with nonlinear regression problems. This paper presents an efficient prediction model for Surfactant-Water Solution Alternating CO2 injection recovery process based on support vector regression and dimensionless groups. A number of experiments and simulations has been carried out under a wide range of the operational and physical parameters to provide sufficient data set for training, validating and testing prediction model. Different sodium dodecyl sulfate (SDS) concentrations were used as the surfactant. The simulation core models were optimized and validated with core flood experiment. Since the selection of SVM's parameters is an optimization issue, Ant Colony Procedure (ACO) is applied to optimize the parameters. Comparative simulations with details are performed to present the performance (the time response and the predictive capability) of ACOR–SVM in comparison to other optimizing and predicting techniques (Genetic Algorithm, Particle Swarm Optimization and Artificial Neural Network). The accuracy obtained by ACO method is higher than those got by GA, PSO and ANN while the cost of time does not increase and computation time is less. The results proved that the ACO–SVM method may serve as a powerful complementary tool to other existing approaches in this area.