| Abstract: | New enhanced oil recovery techniques, such as in situ combustion and chemical flooding, occur with the simultaneous flow of three phases (water, oil and gas) in the reservoir. Accurate estimates of three-phase relative permeability and capillary pressure functions (flow functions) are necessary to predict reservoir behavior with numerical simulators. Although several methods exist to obtain estimates of three-phase relative permeability functions, they have several assumptions that make their use limited and their results subject to controversy. In this work, a parameter estimation method for simultaneously estimating accurate three-phase relative permeability and capillary pressure functions is presented. This method has few of the assumptions made by other methods. Also, the mathematical model used by the method has more flexibility to represent displacement experiments than the models used by other methods. In the model, core sample heterogeneities can be considered in estimating the flow functions. The method is an extension of a regression based method developed for estimating accurate two-phase flow functions using data from unsteady-state displacement experiments. In situ saturation data, obtained with CT scanning techniques, are also used in the history matching method. This type of data is a great source of information to predict correctly multiphase flow through porous media and to obtain accurate estimates of multiphase flow. To accomplish this work, a simulator of three-phase flow through anisotropic porous media was developed. The simulator can model systems in one, two or three dimensions. The flow functions were represented as functions of two saturations with B-splines due to their great flexibility to represent any smooth function. Monte Carlo simulation is used to analyze the accuracy of the estimated flow functions. To demonstrate the use of the parameter estimation method, results for estimated two-phase flow functions using simulated and experimental data are presented and discussed. In the experimental results, heterogeneities of the sample are considered in estimating the flow functions. Results for estimated three-phase flow functions for a simple simulated displacement experiment are presented to illustrate how the method can be used in this case. Since the estimation procedure requires a great amount of computer work, the Cray supercomputer was used to perform the estimation procedure fast and efficiently. The results obtained show that the parameter estimation method is reliable for obtaining accurate estimates of three-phase flow functions. |
|---|