Predicting and optimizing the performance of the expanding solvent steam assisted gravity drainage (ES-SAGD) process using an improved semi-analytical proxy model
MetadataShow full item record
Steam Assisted Gravity Drainage (SAGD) is a commonly used EOR/IOR method for improving recovery in heavy oil reservoirs. However, continued research for a more energy efficient method has led to the development of an improved version called Expanding Solvent (ES)-SAGD, which has the potential to replace conventional SAGD method for production from some heavy oil reservoirs. This thesis provides some insights into determination of the reservoir performance of ES-SAGD process using an improved semi-analytical method. This model is then used for optimizing the solvent requirement while minimizing the steam injected. The semi-analytical model is determined by combining Butler’s oil drainage analytical model and solvent dilution effect of VAPEX process. The predictive ability of this model was improved by accounting for concentration and viscosity dependent solvent diffusion process. Results from this extended model in terms of solvent injection, oil production and Cumulative Steam to Oil Ratio (cSOR) were compared with that of reservoir simulation at various levels of grid resolution. Furthermore, the results from simulation were analyzed using response surface methodology including gradient based optimization technique to determine optimum operating conditions, which was then compared with more robust multi-objective optimization based on Non-dominated Sorting Genetic Algorithm II (NSGA-II) and Pareto-optimality. Both the optimization techniques were used within the improved semi-analytical formulation to come up with optimized operational parameters. Modeling solvent diffusivity as a function of solvent concentration gives better results than those obtained using a constant value for diffusivity. Moreover, results for some key performance factors are in good agreement between the semi-analytical model and the numerical simulation, rendering this model suitable for performing solvents-screening studies. The multi-objective optimization framework within the semi-analytical model is demonstrated to be a feasible option for determining optimum ranges of key operating parameters that would result in success of the project. Intermediate values of solvent fraction ranging 0.1 to 0.2 for almost the entire range of injection pressures result in high bitumen recoveries and relatively low cSOR. The results indicate that higher values of solvent fraction at low operating pressures and lower values of solvent fraction at high operating pressures lead to optimized oil recovery rate and lower steam-oil-ratio. The multi-objective optimization process results in several combinations of control parameters that yield solutions along the Pareto-optimum front. These combinations are all viable solutions to the optimization problem.