Diffusion-Dominated Proxy Model for Solvent Injection in Ultratight Oil Reservoirs

Abstract

Summary Enhanced oil recovery (EOR) by solvent injection offers significant potential to increase recovery from shale oil reservoirs, which is typically between 3 and 7% original oil in place (OOIP). The rather sparse literature on this topic typically models these tight reservoirs on the basis of conventional-reservoir processes and mechanisms, such as by convective transport using Darcy's law, even though there is little physical justification for this treatment. The literature also downplays the importance of the soaking period in huff ’n’ puff. In this paper, we propose, for the first time, a more physically realistic recovery mechanism based on solely diffusion-dominated transport. We develop a diffusion-dominated proxy model assuming first-contact miscibility (FCM) to provide rapid estimates of oil recovery for both primary production and the solvent huff ’n’ soak ’n’ puff (HSP) process in ultratight oil reservoirs. Simplified proxy models are developed to represent the major features of the fracture network. The key results show that diffusion-transport considered solely can reproduce the primary-production period within the Eagle Ford Shale and can model the HSP process well, without the need to use Darcy's law. The minimum miscibility pressure (MMP) concept is not important for ultratight shales where diffusion dominates because MMP is based on advection-dominated conditions. The mechanism for recovery is based solely on density and concentration gradients. Primary production is modeled as a self-diffusion process, whereas the HSP process is modeled as a counter-diffusion process. Incremental recoveries by HSP are several times greater than primary-production recoveries, showing significant promise in increasing oil recoveries. We calculate ultimate recoveries for both primary production and for the HSP process, and show that methane injection is preferred over carbon dioxide injection. We also show that the proxy model, to be accurate, must match the total matrix-contact area and the ratio of effective area to total contact area with time. These two parameters should be maximized for best recovery.