چكيده به لاتين
Abstract:
Naturally Fractured Reservoirs contain a significant proportion of the world's remaining oil reserves. Gas-Oil Gravity Drainage mechanism largely has the potential of recovering the residual oil in the porous medium of these reservoirs after conventional oil displacement methods. However, the production rate of this mechanism is low in some type of reservoirs due to oil reimbibition effect. Miscible gas injection is one of the most applicable enhanced oil recovery techniques improving the performance of this mechanism.
In this study, the performance of gas-oil gravity drainage mechanism has been evaluated under different gas enrichment conditions with considering molecular diffusion effect. To aim this goal, a single matrix block and a stacked matrix block numerical models has been compositionally simulated to investigate the mechanisms optimizing mass exchange between matrix and fracture. According to the numerical results, miscibility development between displacing and displaced fluids through repeated contacts can increase the ultimate recovery factor significantly by reducing capillary holdup and diminishing oil reimbibition effect. In addition, cross phase diffusion effect can increase the drainage rate by reducing oil viscosity. However, this phenomenon may not play an important role after a certain miscibility enrichment condition is achieved. Dimensionless analysis of the results reveals that the gravity forces overcome capillary forces during a miscible displacement and control the drainage process. Sensitivity analysis of the mechanism's ultimate performance with respect to different matrix parameters proves the dominant role of some factors, including matrix block height/dimension, gas/oil capillary pressure, residual oil saturation, and density difference between fluids. The random altering of these parameters simultaneously in the data file and recording the oil recovery factor values while simulating the files automatically by coupling the simulator with the Matlab engineering software causes a large amount of input and output data to be produced. These data has been then used to develop a surrogate intelligent model by use of neural network toolbox, as the study byproduct. This model can approximate the performance of gas-oil gravity drainage mechanism with an acceptable accuracy.
Keywords: Naturally Fractured Reservoirs, Gas-Oil Gravity Drainage Mechanism, Oil Reimbibition, Miscible Gas Injection, Molecular Diffusion, Neural Network, Matlab
