This paper describes a modeling technique for reservoir development using full physics and proxy simulations. Traditionally, full physics simulation involves gridding and discretization, fluid property calculations, displacement modeling, assembling matrix equations, and solving the equations over time. In the gridding step, the reservoir is divided into a collection of cells on which the governing equations for multiphase flow are rewritten in discrete form. Fluid property and displacement modules compute various quantities and their derivatives needed for assembling the matrix equations during time stepping. For reservoir simulation, the system of matrix equations often is quite large and may contain hundreds of thousands of unknown variables. Solving such a system is a computationally intensive task. As a result, a simulation run may take days or even weeks for large models to complete. To improve the computational time, proxy simulations have been considered in recent years as an alternative to full-physics simulation. In general, proxy simulations achieve higher computational efficiency at the expense of accuracy, and are constructed by either reducing the model size or simplifying the physical behavior of fluid flows.
In this work, we combine full physics and proxy simulations to take advantage of the strength of each approach. The proxy technique employed is known as profile generator. It simplifies the reservoir into material balance tanks and well source/sink terms into a set of tables known as type curves, which relates production GOR, WCUT, etc. to EUR and other parameters. With full physics or proxy simulation, flows in wells and surface facilities can be modeled and production/injection rates allocated using rule-based or optimization-based well management.
Integration of grid-based simulation and profile generator in a single modeling system makes it possible to easily generate type curves and validate results from proxy simulations. For large assets containing multiple reservoir systems or fields, this technique allows engineers to build a mixed simulation project containing grid-based models for some reservoir systems and type curve-based models for the rest, thus obtaining a balance between accuracy and computational efficiency. Using this new technique, engineers are able to more efficiently evaluate facility constraints and facility upgrade options, quantify impact of changing a field's startup time on facility capacity, and optimize production from multiple fields.