document preview

Abstract
In this work, we extend our previous work on thermal streamline simulation to steam floods. Steam floods exhibit large volume changes and compressibility associated with the phase behavior of steam, strong gravity segregation and override, and highly coupled energy and mass transport. To overcome these challenges we implement a pressure update along the streamlines and a Glowinski -scheme operator splitting. We tested our streamline simulator on a 2D horizontal quarter- five-spot steam flood, a 2D vertical cross section steam flood with gravity override, and a 2D multi-well example. We compared our thermal streamline results with those computed by a commercial finite-volume thermal simulator on both accuracy and efficiency. For the cases investigated, we are able to retain solution accuracy while reducing computational cost and gaining connectivity information from the streamlines, which is useful for reservoir engineering purposes.

Introduction
A large part of the world oil reserves exists in the form of heavy oils. The development of such reserves by traditional methods (primary depletion, water flood) is often inefficient due to the high viscosity of the heavy oil. Thermal recovery processes, which rely on viscosity reduction of the oil through heat that is injected (steam or hot water) or generated in-situ (in-situ combustion), are well suited for unlocking these resources. Planning and management generally make use of finite volume (FV) thermal reservoir simulations. However, the computational cost is usually high because of the strongly nonlinear and coupled nature of the governing equations. Running optimization and/or sensitivity studies for realistic reservoir settings on grid resolution fine enough to capture critical physics becomes prohibitive.

In other applications, streamline-based flow simulation has been successfully applied to reduce computational costs whilst retaining accuracy. Applications include the simulation of water floods (Batycky et al., 1997; Martin and Wegner, 1979), compositional gas injection (Crane et al., 2000; Gerritsen et al., 2005; Thiele et al., 1997) and polymer flooding (Clemens et al., 2010). Streamline simulation has also been extended to account for compressible fluid and rock properties (Beraldo et al., 2009; Cheng et al., 2006). The computational efficiency of streamline simulation is especially suitable for workflows involving many simulations as might be the case in history matching and optimization problems, and commercial streamline simulation is applied for such purposes (Thiele and Batycky, 2006). In our previous work (Zhu et al., 2010) we developed a streamline simulation framework for hot water flooding.  Here, we explore extension of streamline simulation to steam flooding. If possible, it will give a fast and effective simulator that gives sufficient accuracy for common reservoir simulation studies.