| Popis: |
The successful design of a polymer flood relies on the ability to properly model the distribution of polymer concentration in the reservoir while accounting for effects on fluid properties such as water viscosity increases as a function of in-situ polymer concentration and loss of polymer due to adsorption. Despite advances in numerical techniques and computer hardware, the numerical modeling of polymer floods using Eulerian-based approaches such as finite differences remains a challenge: coarse grids tend to excessively smear concentration fronts masking the true benefit of polymers, yet introducing finer grids inevitably leads to excessive run times that make using modern reservoir engineering workflows unrealistic. This problem was already outlined in 1978 by Lake et al.. We revisit the same problem 30 years later in the context of modern streamline simulation techniques. In this work, we present the extension of modern streamline simulation to field-scale polymer flooding, which represents a step-change from the hybrid, 2D steady-state models used in the 1970's. We apply the well-established numerical modeling of polymer flooding to captured the displacement efficiency in 1D, and couple it with a three-dimensional (3D) streamline simulator to efficiently capture the inter-pattern sweep efficiency caused by well rate imbalances, reservoir architecture, and reservoir heterogeneity. Because modern 3D streamline simulators account for changing well rates, non-uniform initial conditions, and gravity, adding polymer functionality means that real-field polymer floods can be modeled with sufficient confidence to be useful and with the necessary speed to be used for modern reservoir engineering workflows that center on assessing uncertainty and risk associated with design parameters. In this paper we proceed to outline the basic architecture of a streamline simulator with a polymer option. We discuss streamline-specific issues, such as the data that must be exchanged between the Lagrangian streamline grid and the underlying Eulerian grid to allow streamlines to be updated as production/injection conditions change. We discuss advantages and disadvantages of the formulation and present numerical experiments in 1D, 2D, and 3D to illustrate our results. |
