Identification of Polymer Flooding Flow Channels and Characterization of Oil Recovery Factor Based On U-Net DOI
Jinxin Cao, Yiqiang Li, Yaqian Zhang

et al.

Published: April 22, 2024

Abstract Image identification is a major means to achieve quantitative characterization of the microscopic oil displacement process. Traditional digital image processing techniques usually uses series pixel-based algorithms, which difficult real-time large-scale images. Deep learning methods have characteristics fast speed and high accuracy. This paper proposes four-channel segmentation method based on RGB color rock particle mask. First, micro model mask divided together with component form input data through technology. Pixel-level training set labels are then created traditional techniques. Through U-Net semantic network, pixel-level water recovery factor calculation polymer process were carried out. Combined pore distance transformation algorithm, lower limit utilization for different media was clarified. The results show that can accurate division areas. Compared conventional three-channel images, improved proposed in this has significantly accuracy due addition constraints mask, global be Up 99%. Combining some post-processing methods, found flooding increased mobilization degree small pores basis lowered from 25 μm 16 μm. In experiments, by 24.01%, finally achieving rapid network article strong adaptability flow channels Quantitative movement during provides new processing.

Language: Английский

Progress of Gas Injection EOR Surveillance in the Bakken Unconventional Play—Technical Review and Machine Learning Study DOI Creative Commons
J. Zhao, Lu Jin, Xue Yu

et al.

Energies, Journal Year: 2024, Volume and Issue: 17(17), P. 4200 - 4200

Published: Aug. 23, 2024

Although considerable laboratory and modeling activities were performed to investigate the enhanced oil recovery (EOR) mechanisms potential in unconventional reservoirs, only limited research has been reported actual EOR implementations their surveillance fields. Eleven pilot tests that used CO2, rich gas, surfactant, water, etc., have conducted Bakken play since 2008. Gas injection was involved eight of these pilots with huff ‘n’ puff, flooding, injectivity operations. Surveillance data, including daily production/injection rates, bottomhole pressure, gas composition, well logs, tracer testing, collected from generate time-series plots or analytics can inform operators downhole conditions. A technical review showed pressure buildup, conformance issues, timely breakthrough detection some main challenges because interconnected fractures between offset wells. The latest operation co-injecting surfactant through same could be mitigated by careful design continuous reservoir monitoring. Reservoir simulation machine learning then for rapidly predict performance take control actions improve outcomes reservoirs.

Language: Английский

Citations

1
Ion types effect on oil sweep efficiency during engineered waterflooding; an experimental micro-scale study DOI
Hamideh Khajepour, Hossein Ali Akhlaghi Amiri, Shahab Ayatollahi

et al.

Geoenergy Science and Engineering, Journal Year: 2024, Volume and Issue: 241, P. 213175 - 213175

Published: July 26, 2024

Language: Английский

Citations

0
Synergistic Effects of SiO2-Rhamnolipid Nanofluid on Oil-Water Distribution in Low-Permeability Reservoirs DOI
Di Wang, Ying Zhang, Yijing Luo

et al.

Energy & Fuels, Journal Year: 2023, Volume and Issue: 37(22), P. 17250 - 17262

Published: Nov. 1, 2023

The application of nanofluid in enhancing oil recovery (EOR) is eliciting considerable interest from researchers. present study investigates the impact a novel environmentally friendly nanofluid, composed rhamnolipid and SiO2 nanoparticles, on oil-water distributions micro/nano pore spaces, suggesting potential EOR low-permeability reservoirs. We conduct comprehensive experiments, including contact angle measurement, micromodel tests, nuclear magnetic resonance (NMR) core displacement to research synergistic effects biosurfactant nanoparticles distribution changes. Firstly, liquid droplet rock surfaces provides quantitative characterization water phase spreading walls. results indicate that injection advantageous swept volume phase. Secondly, visualization fluid mobilization migration pores demonstrated by test. Residual mobilized stagnant zones with injection, subsequently recovered waterflooding. Moreover, observations exhibit can disperse trapped into small droplets displace them. Finally, NMR experiments clarify main mechanisms microporous spaces using nanofluid. As compared brine solution, more likely motivate oil, thereby increasing fluids. Besides, changes residual are also influenced reservoir's permeability. permeability experimental cores decreases 48.5 mD 0.1 mD, gradually expands larger smaller (from >1000 nm 1–10 nm). saturation three 3.97%, 4.97%, −1.65%, respectively, rhamnolipid-only solution. increase observed sample suggests re-aggregate activated induces distribution.

Language: Английский

Citations

1
Identification of Polymer Flooding Flow Channels and Characterization of Oil Recovery Factor Based On U-Net DOI
Jinxin Cao, Yiqiang Li, Yaqian Zhang

et al.

Published: April 22, 2024

Abstract Image identification is a major means to achieve quantitative characterization of the microscopic oil displacement process. Traditional digital image processing techniques usually uses series pixel-based algorithms, which difficult real-time large-scale images. Deep learning methods have characteristics fast speed and high accuracy. This paper proposes four-channel segmentation method based on RGB color rock particle mask. First, micro model mask divided together with component form input data through technology. Pixel-level training set labels are then created traditional techniques. Through U-Net semantic network, pixel-level water recovery factor calculation polymer process were carried out. Combined pore distance transformation algorithm, lower limit utilization for different media was clarified. The results show that can accurate division areas. Compared conventional three-channel images, improved proposed in this has significantly accuracy due addition constraints mask, global be Up 99%. Combining some post-processing methods, found flooding increased mobilization degree small pores basis lowered from 25 μm 16 μm. In experiments, by 24.01%, finally achieving rapid network article strong adaptability flow channels Quantitative movement during provides new processing.

Language: Английский

Citations

0