Files in this item

FilesDescriptionFormat

application/pdf

application/pdfKOHANPUR-DISSERTATION-2020.pdf (17MB)
(no description provided)PDF

Description

Title:Improving capabilities of pore-scale modeling of multiphase flow for geological storage of CO2
Author(s):Kohanpur, Amir Hossein
Director of Research:Valocchi, Albert
Doctoral Committee Chair(s):Valocchi, Albert
Doctoral Committee Member(s):Werth, Charles; Druhan, Jennifer; Makhnenko, Roman; Tahmasebi, Pejman
Department / Program:Civil & Environmental Eng
Discipline:Civil Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:Ph.D.
Genre:Dissertation
Subject(s):Pore-scale modeling
CO2 sequestration
Pore-network
Lattice-Boltzmann
Upscaling
Heterogeneity
Multiphase flow
Abstract:This dissertation presents a computational modeling framework to address current challenges in pore-scale modeling of two-phase flow with applications to sequestration of carbon dioxide (CO2) in deep saline geological formations. These formations are widely available and have relatively high storage capacity to host injected CO2 for long-term as a practical solution to reduce CO2 emissions from power plants. Due to the expense and complexity of experimental investigations, computational approaches have been developed to understand the physics of CO2-brine flow at the pore-scale. The dissertation considers both direct numerical simulation on real rock geometry measured by X-ray micro-CT scans, as well as pore-network (PN) models which simplify the pore space into interconnected idealized shapes. Both approaches are challenged in applications to large heterogeneous cores. A heterogeneous Mt. Simon sandstone sample is characterized in terms of morphology and CO2-brine flow properties. 3D rock images are investigated to assess the REV size and heterogeneity. Three distinct simulation approaches are applied to simulate the displacement of brine by CO2: PN modeling on the extracted network, and the lattice-Boltzmann (LB) method and the finite-volume method using OpenFOAM (OF) on the rock geometry. The relative permeabilities are computed and compared using different measurement choices: the steady-state approach for LB, unsteady approach for OF, and quasi-static approach for PN. All approaches are in close agreement with one another. The accuracy, computational efficiency, and the effect of grid resolution are also compared. A novel pore-network stitching method (PNSM) is developed that combines the inherent simplicity of PN modeling with statistical network generation to characterize the heterogeneity of cores. The method overcomes technical limits on sample size during X-ray scanning and computational limits on network extraction algorithms. The workflow is validated on various types of rock samples and applied on large domain problems based on pore structure and CO2-brine flow properties. In each sample, multiple realizations are generated and the average results are compared with properties from defined reference PNs. A new set of pore-level flow models in PN modeling are proposed to improve the prediction of residual trapping of CO2. This is important for assessing the long-term storage capacity and safety of geological sequestration. LB simulations are carried out on several PN configurations to investigate pore-body filling and snap-off events that are simplified in PN modeling. The threshold local capillary pressure is evaluated and modified equations are defined. The modified model is incorporated into a quasi-static PN solver and applied to Berea and Mt. Simon sandstone samples to obtain relative permeabilities and residual trapping of CO2 after a drainage-imbibition cycle. The modified model predicts residual trapped CO2 closer to experimental data than the conventional model. These studies together enable the current generation of PN models to be more accurate and applicable in practice. The PNSM enables study of large heterogeneous cores. The use of direct numerical simulation to study multiphase flow physics in PN configurations enables modification of rules implemented in PN models to improve accuracy of predicted residual trapped CO2.
Issue Date:2020-12-02
Type:Thesis
URI:http://hdl.handle.net/2142/109416
Rights Information:Copyright 2020 Amir Hossein Kohanpur
Date Available in IDEALS:2021-03-05
Date Deposited:2020-12


This item appears in the following Collection(s)

Item Statistics