A Fully Adaptive Streamline Framework for the Challenging Simulation of Gas-Injection Processes

Abstract

Abstract We propose a novel fully adaptive computational framework for the efficient simulation of gas injection processes. For flow, we employ a Cartesian Cell-based Anisotropically Refined (CCAR) grid method together with single-phase Multi-Level Local-Global (MLLG) upscaling. The combination of adaptivity and MLLG reduces process-dependency because it ensures that high permeability paths are well resolved. For transport, we use the Compositional Streamline Simulator (CSLS) that was designed for compositional processes. CSLS incorporates mesh adaptivity along streamlines. It adapts the streamline coverage to concentrate streamlines where needed to resolve transport, e.g. near solution fronts, and to reduce errors introduced by the mappings between the pressure grid and the streamlines. These mappings are the primary sources of error if streamlines are updated sufficiently frequently and appropriate higher-order schemes are used for propagation of components along streamlines.We reduce the smoothing errors introduced by the mappings by a higher order mapping from pressure grid to streamlines. We use a Kriging algorithm to map from the streamlines to the background grid. Mass balance errors resulting from the mappings are effectively controlled by increasing the streamline density based on a local error indicator. The flexibility of Kriging allows us to include partially traced streamlines that are not required to begin and end at wells. This is not possible with the commonly used mappings that are based on flux weighting. To reduce errors associated with fixing the pressure field between pressure updates, we developed a higher order global time-stepping method. We demonstrate the potential of the framework with several calculation examples including 2D continuous CO2 injection, water alternating gas (WAG) displacements, and 3D field scale CO2 injection into a saline aquifer. Introduction Gas-injection processes are widely used for enhanced oil recovery (EOR) throughout of the world. In the United States alone, EOR production by gas injection now accounts for approximately 45% of total EOR production and has been steadily increasing (tripled since 1986) [1,2]. The understanding of the multiphase, multicomponent flow taking place in any displacement process is essential for successful design of gas-injection projects. Due to complex reservoir geometry and reservoir fluid properties, the design of accurate and efficient numerical simulations of the flow processes is extremely challenging. In this paper, we present the framework of our novel compositional streamline simulator (CSLS) designed specifically for (near-) miscible gas injection processes. The design of CSLS was non-trivial: straightforward extension of the traditional streamline method that was primarily developed for different classes of reservoir fluid flow processes, such as water-flooding or tracer flows is not possible. We explain the difficulties in compositional simulation for gas injection processes in this introduction and motivate the development of our streamline method. After discussion of the governing equations, we then present the main steps in our framework, discuss specific challenges and our approaches to resolve them. Several components of our framework were previously discussed in [4,5,15,21,26,28,32,35,37]. New work is discussed in more detail in the later sections of this paper. We also present applications of our current compositional streamline simulator to 2D and 3D processes. Challenges in compositional simulation for gas injection processes When gas displaces oil at a sufficiently high pressure, the local displacement efficiency can be high. Miscibility is said to develop when an optimal, piston-like, local displacement efficiency is achieved as can often be achieved using CO2 as injection gas. Due to heterogeneity in reservoirs and the low viscosity of gas, however, the injected gas may contact only a small portion of a reservoir as it finds the high permeability flow paths. That is, the global sweep efficiency of a gas flood may not be high. The process performance of gas injection schemes depends on this balance between local displacement efficiency and global sweep efficiency and both need to be captured accurately by a performance prediction tool.