서브메뉴
검색
Direct Pore-Scale Modeling of Foam Transport in Porous Media and Related Machine Learning
Direct Pore-Scale Modeling of Foam Transport in Porous Media and Related Machine Learning
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260311091536.5
- ISBN
- 9798270231989
- DDC
- 532
- 저자명
- Ma, Xuesong
- 서명/저자
- Direct Pore-Scale Modeling of Foam Transport in Porous Media and Related Machine Learning / Xuesong Ma
- 발행사항
- [Sl] : The University of Texas at Austin, 2025
- 형태사항
- 1 electronic resource (222 pages)
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-06, Section: B.
- 주기사항
- Advisors: Prodanović, Maša Committee members: Daigle, Hugh; Mohanty, Kishore; DiCarlo, David; Nojabaei, Bahareh.
- 학위논문주기
- - Ph.D. : The University of Texas at Austin, 2025.
- 초록/해제
- 요약Foams are dispersions of gas bubbles within a liquid. They are often generated in porous and fractured media during co-injection of two fluids in the presence of a surfactant. Surfactants can lower surface tension to form gas bubbles and generate disjoining pressure to stabilize foam lamellae against bubble coalescence. Since foam viscosity is much larger than its constituent fluids, foam has many applications in subsurface engineering for controlling the mobility of fluids or carrying particulates. Enhanced oil recovery (EOR) and carbon storage processes assisted by foam benefit from its diversion effects, because foam blocks the preferential flow paths in high-flow zones, redirecting the injected fluids toward regions with low permeability and ultimately improving the swept volume by gas. Pore geometry, fluid properties, molecular structure, and injection conditions are some important factors affecting the stability and regeneration of foam in porous media. In this dissertation, foam transport through fractures and a pore-throat structure is investigated under a wide range of injection conditions and fluid properties. We explore foam morphological, interfacial, and rheological changes with detailed evolution of gas bubbles.As a result of the dynamic process of foam generation and coalescence in porous media, it is important to understand pore-level foam mechanisms to describe and predict foam texture and foam rheology during foam flow. Thus far, there have been no pore-scale numerical models that incorporate the micrometer-scale geometry details from state-of-the-art microscopy or X-ray tomography imaging. In this dissertation, we explicitly develop the first such effort in modeling foam transport and capturing changes in foam texture with both two-dimensional (2D) and three-dimensional (3D) imaged geometry.Moreover, we focus on foam viscosity quantification as well as further validation against microfluidic experiments in the literature (Liontas et al., 2013). We conduct foam flow modeling by periodically reloading bubbles from a bubble pool, which controls foam quality and texture. The apparent viscosity of foams is a function of six factors, i.e., pore geometry (capillary radius), foam quality, bubble size, flow rate, interfacial tension, and bubble morphology. Our simulation results illustrate the effect of geometry on bubble flow resistance, namely, the addition of tortuosity and the decrease in aperture are conducive to increasing flow resistance and reducing mobility. The foam apparent viscosity obtained in this work increases with increasing foam quality and then decreases above the critical foam quality, supported by a variety of laboratory data. It is important to note that the interfacial tension, flow rate and bubble size can be combined into one variable, capillary number. Foam apparent viscosity scales as a power law of the capillary number, with negative exponents between −0.33 and −0.64.We establish a separating surface in parametric space that delineates the transition between bubble splitting and deformation (wiggling) when passing through a pore throat. Further, we quantify the difference in foam texture between 2D and 3D modeling at the same fluid properties and injection conditions. For bubbles passing through a pore throat, 2D simulation results overestimate the occurrence and severity (the number of resulting bubbles after exiting the throat) of splitting, compared with 3D counterparts with a relative channel height as 1.In addition, we investigate the capability of machine learning (ML) to learn from the pore scale simulations and to predict interface evolution. Specifically, we implement the most up-to-date Fourier neural operators for solving partial different equations (PDE) and forecasting time series with the objective of accelerating LBM simulations. Results show that deep learning-based surrogate models accelerate time evolution of LBM simulations and provide predictions in close quantitative agreement with simulation results in pre-specified geometries. Hence, we set a stage for scaling pore-scale insights into foam mechanisms up to their continuum-scale implications through significant pore-scale simulation speedup. In the future, the combined direct simulation and machine learning can provide required parameters for population balance models at core scale by dynamically capturing pore-scale foam texture changes.
- 언어주기
- English
- 일반주제명
- Morphology
- 키워드
- Gas bubbles
- 기타저자
- The University of Texas at Austin Petroleum and Geosystems Engineering
- 기본자료저록
- Dissertations Abstracts International. 87-06B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260311s2025 us eng d■001000017361212
■00520260311091536.5
■006m o d
■007cr|nu||||||||
■020 ▼a9798270231989
■040 ▼aMiAaPQD▼beng▼cMiAaPQD▼erda
■082 ▼a532
■1001 ▼aMa, Xuesong▼eauthor.
■24510▼aDirect Pore-Scale Modeling of Foam Transport in Porous Media and Related Machine Learning ▼cXuesong Ma
■260 ▼a[Sl]▼bThe University of Texas at Austin▼c2025
■264 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a1 electronic resource (222 pages)
■336 ▼atext▼btxt▼2rdacontent
■337 ▼acomputer▼bc▼2rdamedia
■338 ▼aonline resource▼bcr▼2rdacarrier
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-06, Section: B.
■500 ▼aAdvisors: Prodanović, Maša Committee members: Daigle, Hugh; Mohanty, Kishore; DiCarlo, David; Nojabaei, Bahareh.
■5021 ▼bPh.D.▼cThe University of Texas at Austin▼d2025.
■520 ▼aFoams are dispersions of gas bubbles within a liquid. They are often generated in porous and fractured media during co-injection of two fluids in the presence of a surfactant. Surfactants can lower surface tension to form gas bubbles and generate disjoining pressure to stabilize foam lamellae against bubble coalescence. Since foam viscosity is much larger than its constituent fluids, foam has many applications in subsurface engineering for controlling the mobility of fluids or carrying particulates. Enhanced oil recovery (EOR) and carbon storage processes assisted by foam benefit from its diversion effects, because foam blocks the preferential flow paths in high-flow zones, redirecting the injected fluids toward regions with low permeability and ultimately improving the swept volume by gas. Pore geometry, fluid properties, molecular structure, and injection conditions are some important factors affecting the stability and regeneration of foam in porous media. In this dissertation, foam transport through fractures and a pore-throat structure is investigated under a wide range of injection conditions and fluid properties. We explore foam morphological, interfacial, and rheological changes with detailed evolution of gas bubbles.As a result of the dynamic process of foam generation and coalescence in porous media, it is important to understand pore-level foam mechanisms to describe and predict foam texture and foam rheology during foam flow. Thus far, there have been no pore-scale numerical models that incorporate the micrometer-scale geometry details from state-of-the-art microscopy or X-ray tomography imaging. In this dissertation, we explicitly develop the first such effort in modeling foam transport and capturing changes in foam texture with both two-dimensional (2D) and three-dimensional (3D) imaged geometry.Moreover, we focus on foam viscosity quantification as well as further validation against microfluidic experiments in the literature (Liontas et al., 2013). We conduct foam flow modeling by periodically reloading bubbles from a bubble pool, which controls foam quality and texture. The apparent viscosity of foams is a function of six factors, i.e., pore geometry (capillary radius), foam quality, bubble size, flow rate, interfacial tension, and bubble morphology. Our simulation results illustrate the effect of geometry on bubble flow resistance, namely, the addition of tortuosity and the decrease in aperture are conducive to increasing flow resistance and reducing mobility. The foam apparent viscosity obtained in this work increases with increasing foam quality and then decreases above the critical foam quality, supported by a variety of laboratory data. It is important to note that the interfacial tension, flow rate and bubble size can be combined into one variable, capillary number. Foam apparent viscosity scales as a power law of the capillary number, with negative exponents between −0.33 and −0.64.We establish a separating surface in parametric space that delineates the transition between bubble splitting and deformation (wiggling) when passing through a pore throat. Further, we quantify the difference in foam texture between 2D and 3D modeling at the same fluid properties and injection conditions. For bubbles passing through a pore throat, 2D simulation results overestimate the occurrence and severity (the number of resulting bubbles after exiting the throat) of splitting, compared with 3D counterparts with a relative channel height as 1.In addition, we investigate the capability of machine learning (ML) to learn from the pore scale simulations and to predict interface evolution. Specifically, we implement the most up-to-date Fourier neural operators for solving partial different equations (PDE) and forecasting time series with the objective of accelerating LBM simulations. Results show that deep learning-based surrogate models accelerate time evolution of LBM simulations and provide predictions in close quantitative agreement with simulation results in pre-specified geometries. Hence, we set a stage for scaling pore-scale insights into foam mechanisms up to their continuum-scale implications through significant pore-scale simulation speedup. In the future, the combined direct simulation and machine learning can provide required parameters for population balance models at core scale by dynamically capturing pore-scale foam texture changes.
■546 ▼aEnglish
■590 ▼aSchool code: 0227
■650 4▼aMorphology
■650 4▼aWater resources management
■653 ▼aEnhanced oil recovery
■653 ▼aGas bubbles
■653 ▼aMicrofluidic experiments
■653 ▼aBubble morphology
■7102 ▼aThe University of Texas at Austin▼bPetroleum and Geosystems Engineering.▼edegree granting institution.
■7201 ▼aProdanović, Maša▼edegree supervisor.
■7730 ▼tDissertations Abstracts International▼g87-06B.
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17361212▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


