| dc.description.abstract | This thesis presents a numerical modeling study of two gas-based, same-well, enhanced oil recovery (SWEOR) processes—huff-and-puff (HnP) and fracture-to-fracture (F2F)—applied to tight unconventionals. The processes consist of cyclic injection and production through the same well, where increased recovery of hydrocarbons may be achieved from repeated injections of gas that acts as a solvent for the in-situ hydrocarbons in the near-fracture region. The mechanisms contributing to increased recovery of the hydrocarbons will mainly be dilution and vaporization for the HnP process, and displacement for the F2F process.
The study focuses mainly on the phase-behavior aspects of gas injection. This includes a thorough review of key concepts, such as miscibility, molecular diffusion, numerical dispersion, and PVT formulations, and how these concepts affect the HnP and F2F processes.
The extensive fracture network created by modern hydraulic fracturing of wells in tight unconventionals provides a large surface area through which the injection gas enters the matrix during the injection period. When the injection gas contacts the in-situ hydrocarbons, new mixtures are created that may either exist as a single phase (recovery by dilution), or as two phases where the vapor phase has been enriched with in-situ hydrocarbons (recovery by vaporization). The new mixtures are produced back during the production period, ideally resulting in an increased recovery of oil relative to an established baseline representing continued production by regular pressure depletion.
Recent studies of the fractured volume, often referred to as the stimulated rock volume (SRV), show that the fracture network induced by hydraulic fracturing contains some primary fractures (propped, conductive), and many secondary fractures (unpropped, less conductive). It is found that many secondary fractures are beneficial for the HnP process. Two different symmetry-element models of the fractured reservoir are suggested, the shattered-volume (SV) model, and the slab model. The SV model is designed to include a volume of significant secondary fracturing in-between two primary fractures. The primary fractures are explicitly gridded in the model, and the secondary fractures are represented by the dual-porosity model. The slab model more closely reflects recent observations reported in the literature of fracture networks and is numerically represented by explicitly gridding both primary- and secondary fractures. Both models share the same near-zero incremental recovery by the HnP process when only primary fractures are included, and the grid-refinement level is sufficiently high. Increasing intensity of secondary fractures improves the recovery performance of the HnP process, and the incremental uplift in oil recovery relative to the pressure-depletion baseline may become positive if sufficient secondary fracturing is achieved.
The F2F process has been suggested in the literature as an improved alternative to HnP, where the idea is to direct the flow of injection and production through alternating perforations (entry points from wellbore to the fracture network) such that displacement, rather than slow diffusion-driven mixing, is achieved between the points of injection (high-pressure zones) and points of production (low-pressure zones). The directional flow control is achieved by installing a completion string in the perforated interval consisting of alternating one-way valves and packers. The intention of the F2F process is to utilize the high displacement efficiency of miscible displacement, that may yield large recoveries in the intra-fracture region if the conformance (areal- and vertical sweep efficiency) is good. Numerical modeling shows that shorter primary-fracture spacings (<50 ft), together with a matrix permeability of >0.1 millidarcy may yield much larger ultimate recovery of oil than HnP, but with a delayed response. For larger primary-fracture spacings and tighter matrix permeability, F2F is less applicable.
A key finding of this study is that the predicted recovery from numerical modeling of the HnP process is highly dependent on the grid-refinement level of the model. Coarse gridding (large grid cells) causes artificial mixing of the injected gas and the in-situ reservoir fluid, which in turn may greatly exaggerate the recovery performance. For the F2F process, the effect of coarse gridding is opposite. The sharp displacement front formed by multi-contact miscibility is smeared by the coarse gridding, leading to earlier breakthrough of gas at the producing fracture, and less recovery of oil. | en_US |
