Factors Affecting Miscible Flooding Dispersion Coefficients

Abstract

Listen

Translate article

Abstract Miscible solvent slug size, and therefore cost, is dependent on the mixing or dispersion taking place in the reservoir. Fluid mixing can also be important in the interpretation of laboratory simulations of miscible floods. An experimental program was conducted to study the effects of velocity, viscosity ratio, rock type and core length on dispersion (mixing) coefficients measured in short cores, with the objective of scaling laboratory measurements to field systems. Statistical analysis of the results of the tests, matched with the capacitance-dispersion ("dead-end core volume") model, shows that an effective dispersion coefficient derived from the model is the most consistent measure of mixing in the systems studied. Viscosity ratios differing by ±4%from unity had no significant effect on the effective dispersion coefficient. The effect of system length on the effective dispersion coefficient Is shown. Because of the length dependence, dispersion coefficients measured in short laboratory systems may underestimate the true dispersion taking place in the reservoir. The length dependence of the dispersion coefficient may also have important implications in the interpretation of laboratory results where mixing is a factor. INTRODUCTION Economical enhanced oil recovery requires injection of a relatively small slug of solvent or oil mobilization fluid, e.g., CO2 rich gas or micellar fluid. A chase fluid (N2 flue gas or polymer water), which is miscible with the solvent, is generally used to maximize solvent utilization. A physical phenomenon referred to as dispersive mixing acts to reduce the effectiveness of the solvent. Dispersive mixing also decreases the effectiveness of chase fluid in displacing the solvent. Therefore, the effect of dispersive mixing on the process must be included in numerical models for the design of enhanced oil recovery projects. Various mathematical models have been propsoed to simulate the dispersive mixing which occurs in oil reservoirs(1). These models include the simple dispersion model and the capacitance-dispersion model. The dispersive mixing simulated by these models is characterized by adjustable parameters. The parameters must be determined or estimated for a porous medium in order to simulate the dispersive mixing which will take place between miscible fluids flowing through it. In previous work on dispersive mixing(2), a technique was developed to determine the dispersion coefficient for a consolidated core. This technique consists of conducting displacements in the core of interest with equal-viscosity-contact miscible fluids and matching the effluent concentration data with a dispersive mixing model. This work concluded that for many systems the simple dispersion model was not adequate. The capacitance-dispersion model was shown(2) to better simulate the dispersive mixing in these systems. This model includes additional parameters which allow it to fit more irregular effluent concentration profiles with greater accuracy. The work described in this paper is a continuation of the miscible-displacement, dispersion-coefficient study. The prinicpal objectives of this work were to:determine the physical significance of the parameters of the capacitance-dispersion model;determine the validity of using an "effective dispersion coefficient" defined by this model to simulate dispersive mixing; andshow that as system length increases the dispersive mixing for heterogeneous systems can be simulated by a simple dispersion model with an "effective dispersion coefficient."