Browsing by Subject "Shape factor"
Now showing 1 - 1 of 1
- Results Per Page
- Sort Options
Item Matrix/fracture transfer function during counter-current spontaneous imbibition in naturally fractured reservoirs(2023-08-10) Al Homidi, Othman; Mohanty, Kishore Kumar; Balhoff, Matthew T.Naturally fractured reservoirs are abundant in the earth’s crust and host a substantial percentage of oil reserves globally. The main mechanism of oil recovery during waterflooding of these types of reservoirs is through spontaneous imbibition of water into the matrix and simultaneous counter-current flow of oil out of the matrix. Understanding the predominate recovery mechanism enhances reserves estimates, accurate simulation forecasts and overall sound development plans. Dual-porosity and dual-permeability simulations are used in the industry to simulate waterfloods in naturally fractured reservoirs. One of the key parameters in these simulations is the matrix-fracture transfer term, which is not well understood and modeled, especially in mixed-wet reservoirs. The same transfer term is used for primary, secondary and tertiary recovery processes, though it should change depending on the mechanisms of oil recovery. The key mechanism during primary recovery is depressurization, not spontaneous imbibition. The main goal of this research is to develop an accurate representation of the matrix-fracture transfer term in waterflooding for dual-porosity simulators. The analytical and semi-analytical solutions for 1D counter-current imbibition were studied for defining the exact solution in fractured porous media. Fine-grid, single-porosity numerical solutions were developed that are consistent with the 1D analytical solutions, in conjunction with coarse-grid single-porosity conceptual models. Both single-porosity models are used as reference against dual-porosity conceptual models to address the built-in matrix-fracture transfer terms through recovery of the matrix element. The error in simulation was defined as the difference in recoveries between the fine-grid single-porosity solution and the dual-porosity solutions. A detailed investigation of both rock and fluid inputs affecting transfer terms in dual-porosity was made in an effort to match the transient solution obtained from fine-grid single-porosity models. The inclusion of transient effect in dual-porosity requires optimizing the following inputs which are shape factor, capillary exponent and oil relative permeability exponent. Two main processes were proposed for optimization. Firstly, an accuracy-based Latin Hyper Cube sampling method was utilized that converged to the solution quickly. Secondly, utilizing a machine learning algorithm (specifically an Artificial Neural Net model) that predicts recovery accuracy based on the aforementioned chosen inputs. The machine learning model needed many iterations to converge to a solution.