163559 research outputs found
Sort by
Impact of exciton fine structure on the energy transfer in magic-sized (CdSe)<sub>13</sub> clusters
No full textMagic-sized (CdSe)13 clusters (MSCs) represent a material class at the boundary between molecules and quantum dots that exhibit a pronounced and well separated excitonic fine structure. The characteristic photoluminescence is composed of exciton bandgap emission and a spectrally broad mid-gap emission related to surface defects. Here, we report on a thermally activated energy transfer from fine-structure split exciton states to surface states by using temperature dependent photoluminescence excitation spectroscopy. We demonstrate that the broad mid-gap emission can be suppressed by a targeted Mn-doping of the MSC leading to the characteristic orange luminescence of the 4T1 -> 6A1 Mn2+ transition. The energy transfer to the Mn2+ states is found to be significantly different than the transfer to the surface defect states, as the activation of the dopant emission requires a spin-conserving charge carrier transfer that only dark excitons can provide.Y
Plain language summary: 5-year results from the CROWN study of lorlatinib vs crizotinib in non-small-cell lung cancer
No full textWhat is this study about?: This is a summary of the results of an ongoing study called CROWN. In the CROWN study, researchers looked at the effects of two medicines called lorlatinib (Lorbrena) and crizotinib (Xalkori) for people with advanced non-small cell lung cancer (NSCLC) who had not been treated yet. Everyone in the study had changes in a gene called anaplastic lymphoma kinase, or ALK, in their cancer cells. The changes in the ALK gene can make cancer grow. This analysis looked at how well lorlatinib and crizotinib worked and their side effects in people with advanced ALK-positive NSCLC after 5 years. What did this study find?: After observing people for an average of 5 years, researchers found that more people who took lorlatinib were still alive without their cancer getting worse than the people who took crizotinib. At 5 years, the probability of being alive without their cancer getting worse was 60% in people who took lorlatinib compared with 8% in people who took crizotinib. Fewer people who took lorlatinib had their cancer spread within or to the brain than the people who took crizotinib. In more than half of the people who took lorlatinib, tumors that had spread to the brain did not get worse, and no new tumors spread to the brain after 5 years. In contrast, in about half of the people who took crizotinib, tumors that had spread to the brain got worse or new tumors spread to the brain after 16.4 months. More people who took lorlatinib (115 out of 149, or 77%) had severe or life-threatening side effects than people who took crizotinib (81 out of 142, or 57%). These side effects were like the ones reported in the earlier 3-year analysis. What do the findings of the study mean?: The 5-year results from the CROWN study showed that more people who took lorlatinib continued to benefit from their treatment than those who took crizotinib. The 5-year benefit of lorlatinib in people with ALK-positive NSCLC has never been seen before. Clinical Trial Registration:NCT03052608 (Phase 3 CROWN study) (ClinicalTrials.gov).Y
Fragmented Investment and Investment Contract Securities—The Critical Analysis of the South Korean Regulatory Bodys Attempt to Correlate the Two
No full textIn South Korea, the advent of fragmented investment has attracted many young adults who were looking for new investment opportunities with limited capitals. Musicow's business model has been the epitome of the new phenomenon, as it introduced an idea of royalty participation right and offered it to the public instead of copyright itself. However, some started to complain about the lack of or the insufficiency in investor protection. As a response, the Korean regulatory body labeled the royalty participation right as an investment contract security and ordered Musicow and other service providers of similar fragmented investment products to establish systems for bankruptcy remoteness and other means of investor protections. During the process, however, the regulatory body erred in labeling royalty participation right as an investment contract security as such a decision conflicted with the existing law. While restrictions on the issuance of investment contract security could be no more than the submission of registration statement and avoidance of unfair tradings, the demands to Musicow and other service providers were above such a limitation. In theory, it seems to be more fit to consider such products as derivatives-linked securities rather than investment contract securities. In addition, such an inconsistency could have been avoided if the regulatory body had i) waited for an appropriate amendment of the related law or ii) guided Musicow and other service providers in restructuring their products as entrusted beneficiary certificates
Legal Oversight of AI, Part Two: Discriminative Model
No full text판별모델(discriminative model)은 주로 지도학습 기반의 분류기 또는 채점기이다. 그간 판별모델의 법적 통제는 불투명성과 자동화에 의한 인간소외, 존엄성 훼손에 초점이 맞춰졌다. 이는 2020년 신용정보법상 자동화평가 설명요구⋅이의제기권, 2021년 행정기본법상 자동적 처분의 재량행위 활용 금지와 기속행위 활용 법률유보, 2023년 개인정보법상 자동화된 결정 거부권⋅설명등요구권⋅공개의무 조항으로 이어진다. 그러나 소외⋅존엄은 자동성보다는 자율성의 문제이고, 불투명성⋅편향이 사람의 판단에 현저한 경우도 많으니, 자동화의 금지⋅제약이 제대로 된 해법이 될 수 없다. 특히, 취업준비생 등 판별대상자들이 현실에서 겪는 어려움을 덜려면, 소외⋅존엄 등 추상 개념에 천착할 것이 아니라, 완전자동⋅부분자동⋅수동을 불문하고 판별 작업의 정확성, 가치정렬성, 비차별성, 다원성을 기해야 한다. 따라서 입법적 과제는 의사결정 자동화의 억지⋅지연에 있는 것이 아니라, (1) 기존의 사람에 의한 결정의 성실성까지 성찰하면서, (2) 귀납⋅실증⋅데이터에 근거한 판별모델의 도입을 오히려 촉진하고, (3) 이로써 채점기준(rubric)의 공개 등 더 나은 설명을 기하되, (4) 판별대상자들이 부당하게 차별받고 배제되지 않도록 안전장치를 설정함에 있다. 이 관점 하에 다음 단계를 거쳐 기존 입법의 대안을 강구한다. 첫째, 판별 모델을 사람과의 상호작용 방식에 따라 유형화하여 (1) 배분모델(채용⋅입학사정⋅대출심사⋅신용평가⋅보험계약인수⋅사회보장⋅랭킹 등)은 차별, (2) 제재모델(부정계정탐지⋅사기탐지⋅보험손해사정⋅재범예측 등)은 오판(정확성의 부족), (3) 인지모델(의료영상진단⋅센서⋅생체인증⋅추천 등)은 안전을 해하거나 배분⋅제재모델의 특성값에 오류를 초래할 정도의 정확성(견고성 포함)의 부족(품질편차 문제 포함), 프라이버시를 해할 정도의 정확성의 과잉을 핵심 위해로 지목한다. 모델에 대한 설명⋅공개는 배분⋅제재모델에서 판별대상자에게 대칭적 정보 하에서의 개선의 기회 또는 차별에 문제제기할 근거를 부여하는 장치로 구성한다. 둘째, 법적 개입을정당화할 만큼 (1) 공적 영역에서 활용되거나 판별대상자의 삶을 좌우하는 배분⋅제재모델, (2) 안전, 프라이버시 리스크가 있거나 배분⋅제재모델에 투입될 특성값을 매기는 인지모델을 식별한다. 셋째, 이같이 식별된 모델들에 대해 앞서 지목된 핵심 위해를 측정⋅경감하는 방법과 이를 뒷받침할 법적 수단을 검토한다. 넷째, 상기 검토를 종합하여 현 자동화 평가⋅처분⋅결정 규제의 대안을 제시한다.
A discriminative model is largely a supervised learning-based classifier or regressor/interpolator. Legal responses thus far have primarily targeted bias and opacity arising from black boxes or algorithmic alienation. These notions have led to regulations such as the explanation requirement for automated evaluation under the Credit Information Act of 2020, the prohibition of discretionary actions in automated disposition under the Framework Act on Administration of 2021, and the explanation and disclosure requirement for automated decision-making under the Personal Information Protection Act of 2023. However, bias, opacity, and alienation are not exclusive to AI. They already exist in human judgment processes that discriminative models seek to replace or complement, sometimes to a greater extent. Viewing the emergence of AI as an opportunity to reflect on existing human decisions, the legal system must be revamped to ensure fairness and acceptability in decisions impacting others lives, regardless of being automated or not. From this perspective, the following steps are taken to examine the problems of existing legislation and seek alternatives. First, discriminative models are categorized based on how they interact with humans, and different harms are mapped for each category: (1) discrimination in allocative models (such as hiring, admission, loan review, credit scoring, insurance underwriting, public assistance, ranking, etc.); (2) misjudgment (lack of accuracy) in punitive models (such as fraud detection, insurance claim adjustment, recidivism prediction, etc.); and (3) either lack of accuracy (including robustness) jeopardizing safety or excessive accuracy compromising privacy in cognitive models (such as medical imaging and diagnosis, sensors, biometric authentication, recommendations, etc.). Instead of conceptualizing opacity as a stand-alone harm, explanations and disclosures about the model are structured as means to provide affected individuals with opportunities for improvement under symmetric information or grounds for challenging discrimination in allocative models or quality disparity in punitive models. Second, (1) allocative or punitive models either deployed in the public domain or decisive of individual lives and (2) cognitive models posing safety or privacy risks are identified to justify legal intervention. Third, tools for measuring and mitigating the identified core concerns of these models and appropriate legal means are reviewed. Fourth, alternatives to existing automated decision-making rules are explored
Efficient Production of the Colorless Carotenoid Phytoene in Yarrowia lipolytica through Metabolic Engineering
No full textPhytoene, a colorless carotenoid with unique ultraviolet (UV)-B absorption properties, offers potential for applications in functional food, cosmetics, and therapeutics. However, their low natural yield poses a challenge for large-scale production. This study aims to enhance phytoene production in the oleaginous yeast Yarrowia lipolytica by introducing a heterologous phytoene synthase gene combined with metabolic engineering approaches. We enhanced phytoene synthesis by overexpressing key genes in the mevalonate pathway and compartmentalizing the biosynthetic pathway within peroxisomes. Moreover, we inhibited the glyoxylate cycle to increase the accumulation of peroxisomal acetyl-CoA available for phytoene production. Our engineered strains demonstrated a significant increase in phytoene production, reaching up to 1.34 g/L titer and 58.74 mg/gDCW yield in the flask-scale fed-batch culture, which are the highest levels reported to date. These results underscore the potential of Y. lipolytica as a robust platform for producing phytoenes and other terpenoids on an industrial scale, offering valuable insights for future efforts in metabolic engineering.N
애나 로웬하웁트 칭(저), 노고운(역), 2023, 『세계 끝의 버섯: 자본주의의 폐허에서 삶의 가능성에 대하여』, 현실문화
No full text삶이 엉망이 되어갈 때 여러분은 무엇을 하는가? 나는 산책을 한다. 그리고
운이 좋으면 버섯을 발견한다.(21)
이 책의 프롤로그 도입부에 나오는 문장이다. 책 표지를 둘러싼 밝은
색 띠지에 별도로 인쇄되어 강조되어 있기도 한 이 문장은 언뜻 보면 산
책과 치유에 관한 식물인문학류 에세이집을 읽고 있나 싶어질 만큼 서정
적이다. 책 제목에 드러나 있는 세계 끝과 폐허라는 말이 사뭇 종말론적
세계관을 반영하고 있는 듯 느껴지기도 하고, 무엇보다 자본주의의 끝과
폐허이기에 경제 시스템에 관한 다소 무거운 내용이 다뤄지고 있으리라
생각하면 짐짓 기이한 불협화음이 느껴지기도 한다. 어쩌면 이 불협화음
이 500여 쪽이라는 적지 않은 분량의 학술서가 학계 안팎에서 다양한 독
자들의 마음을 당겨 움직였던 것일까?
불협화음은 저자인 인류학자 애나 로웬하웁트 칭이 말하고 있는다 운
율(polyphonic) 또는 다성 음악(polyphony)과 맞닿는 것이기도 하다
A Multi-Omics Profiling
No full text학위논문(석사) -- 서울대학교 대학원 : 보건대학원 환경보건학과, 2024. 8. 김성균.배경: 비스페놀 A 의 대체물인 비스페놀 S (BPS)는 내분비 교란 물질로서 건강 문제를 일으킬 수 있다. BPS 가 태반 발달과 기능에 미치는 영향은 잘 알려져 있지 않다. 태반은 태아 발달과 산모의 건강에 중요한 역할을 하기 때문에, BPS 로 인한 변화를 이해하는 것이 중요하다. 이 연구는 BPS 노출이 태반 분자 경로에 미치는 영향을 다중 오믹스 접근법을 사용하여 조사하여 환경 스트레스 요인에 대한 생물학적 반응을 종합적으로 이해하고자 했다.
방법: 임신한 C57BL/6N 마우스를 배아 발생 6.5 일부터 15.5 일까지 저용량(0.5 mg/kg/day) 또는 고용량(50 mg/kg/day)의 BPS 에 노출시켰으며, 태반은 16.5 일에 수집하여 분석했다. RNA 시퀀싱을 통해 전사체 분석을 수행했다. 대사체학 및 지질체학 분석은 양이온 및 음이온 모드 모두에서 UPLC-QTOF-MS 를 사용하여 수행했다. 조직병리학적 검사는 H&E 염색을 사용하여 수행했다. 태반 내 BPS 농도는 HPLC-MS/MS 를 사용하여 정량화했다. 차등 유전자 발현 분석은 edgeR 을 사용하여 수행했으며, 대사물질 변화는 Progenesis QI 및 Metabanalyst 6.0 을 사용하여 평가했다.
결과: 태반의 BPS 농도는 저용량군에서 0.0045 ± 0.0046 mg/kg, 고용량군에서 0.0272 ± 0.0152 mg/kg 로 나타났다. BPS 노출은 고용량군에서 체중 대비 태반 무게의 유의한 감소를 초래했다. 조직병리학적 분석에서는 고용량군에서 junctional zone 두께 비율의 감소와 labyrinth 두께 및 면적 비율의 증가가 나타났다. 전사체 분석에서는 저용량군에서 1,199 개, 고용량군에서 2,618 개의 differently expressed genes(DEGs)가 확인되었으며, 주요 영향을 받은 경로로는 Tnfa signaling via Nfkb, G2M checkpoint 및 mTORC1 signaling 이 포함되었다. 대사체학 및 지질체학 분석에서는 저용량군에서 99 개, 고용량군에서 183 개의 significantly altered metabolites(SAMs)이 나타났으며, 주로 glycerophospholipid metabolism 영향을 받았다.
결론: 본 다중 오믹스 접근법은 임신 중 BPS 노출이 태반 무게, 조직 형태, 유전자 발현 및 대사물질 프로필에 중대한 변화를 보여준다. 이러한 변화는 특히 지질 대사 및 세포 신호전달 경로에서 나타나며, BPS 가 태반 기능을 방해할 수 있는 잠재적 메커니즘을 시사한다. 우리의 발견은 BPS 로 인한 태반 변화의 종합적인 분자적 기초를 제공하며, 임신 중 BPS 노출의 장기적인 결과에 대한 추가 연구의 필요성을 강조한다.Background: Bisphenol S (BPS), a substitute for Bisphenol A, raises health concerns due to its potential as an endocrine disruptor. Its effects on placental development and function are not well understood. Given the placenta's role in fetal development and maternal health, understanding BPS-induced changes is crucial. This study used a multi-omics approach to investigate BPS exposure's impact on placental molecular pathways for a comprehensive understanding of biological responses to environmental stressors. Methods: Pregnant C57BL/6N mice were exposed to low (0.5 mg/kg/day) or high (50 mg/kg/day) doses of BPS from embryonic day 6.5 to 15.5. Placentas were collected at day 16.5 for analysis. Transcriptomics was performed using RNA sequencing. Metabolomics and lipidomics were conducted using UPLC-QTOF- MS in both positive and negative ion modes. Histopathological examinations were carried out using H&E staining. Placental BPS levels were quantified using HPLC-MS/MS. Differential gene expression analysis was performed using edgeR, and metabolite changes were assessed using Progenesis QI and Metabanalyst 6.0. ResultsPlacental BPS levels were 0.0045 ± 0.0046 mg/kg and 0.0272 ± 0.0152 mg/kg in low and high-dose groups, respectively. BPS exposure led to a significant decrease in placental weight relative to body weight in the high-dose group. Histopathological analysis revealed a decreased junctional zone thickness ratio and increased labyrinth thickness and area ratio in the high-dose group. Transcriptomics identified 1,199 and 2,618 differentially expressed genes in low and high-dose groups, respectively, with key affected pathways including Tnfa signaling via Nfkb, G2M checkpoint, and mTORC1 signaling. Metabolomics and lipidomics analyses showed 99 and 183 significantly altered metabolites in low and high-dose groups, respectively, with glycerophospholipid metabolism being predominantly affected. Conclusion: The present multi-omics approach reveals that BPS exposure during pregnancy induces significant alterations in placental weight, tissue morphology, gene expression, and metabolite profiles. These changes, particularly in lipid metabolism and cellular signaling pathways, suggest potential mechanisms by which BPS may disrupt placental function. Our findings provide a comprehensive molecular basis for understanding BPS-induced placental alterations and emphasize the need for further investigation into the long-term consequences of gestational BPS exposure. Keyword : Bisphenol S, placenta, metabolomics, multi-omics, lipid metabolism, endocrine disruptor. Student Number : 2022-27370Chapter 1. Introduction 1
Chapter 2. Materials and Methods 5
Chapter 3. Results 22
Chapter 4. Discussion 39
Chapter 5. Conclusion 44
Chapter 6. References 46
Chapter 7. Supplementary 51
Abstract in Korean 76석
DNA 인터컬레이터를 이용한 DNA-금 나노조립체의 비틀림 변형 제어
No full text학위논문(박사) -- 서울대학교 대학원 : 공과대학 기계공학부, 2024. 8. 김도년.DNA origami nanotechnology can produce various shapes of nanostructures by folding and fixing a long scaffold DNA strand with multiple staple DNA strands. This technology has been used to fabricate DNA-gold nanoassemblies where gold nanoparticles (AuNPs) were precisely positioned on the DNA origami structure through the complementary binding between their bases of DNA strands. However, altering the position of AuNPs in DNA-gold nanoassemblies required modification of the staple design within the DNA origami structure. To solve this, although the methods to control the connection of DNA strands have been studied, the reconfiguration of the DNA-gold nanoassemblies has been restricted to limited states through the on-off mechanism of the DNA strand. Here, we propose a method to induce gradual and continuous reconfiguration of chiral arrangement in the DNA-gold nanoassembly by controlling the torsional deformation via the DNA intercalators for various responses. By binding the DNA intercalator to the DNA strand, the unit block of the DNA-gold nanoassemblies between the two AuNPs was twisted, changing the relative position between the neighboring AuNPs on the unit block. The axial repetition of these unit blocks allowed for the control of the chiroptical responses by varying the chiral arrangement of AuNPs through the torsional deformation. The sensitivity of the variation could be controlled depending on the DNA intercalator types, their mixtures, and the rigidity of the DNA origami structure. The lateral connection of unit blocks enabled the achievement of the hinge motion responses with various angles by controlling the chiral arrangement of the nanohinge arms via torsional deformation. The sensitivity of the nanohinge was controlled by the DNA intercalator types and their mixtures, while its operating range was adjusted according to the distance between its two AuNPs. Additionally, the polymer connected by the nanohinges exhibited bending motions as it accumulated the rotational angles. In summary, the response could be finely controlled through gradual and continuous reconfiguration of chiral arrangement in the DNA-gold nanoassembly. This was achieved by adjusting the torsional deformation of the unit block within the nanoassembly induced by the DNA intercalator. This approach has the potential to efficiently fabricate nanodevices requiring various states such as metasurfaces, nanosensors, and nanorobots.DNA 종이접기 나노기술은 긴 스캐폴드 DNA 가닥을 여러 개의 스테이플 DNA 가닥으로 접고 고정함으로써 다양한 형태의 나노구조를 만들 수 있습니다. 이 기술은 DNA 가닥의 염기 서열 사이의 상보적 결합을 통해 금 나노입자(AuNPs)가 DNA 종이접기 구조 위에 정확히 위치하는 DNA-금 나노어셈블리를 제작하는데 사용되었습니다. 그러나 DNA-금 나노어셈블리에서 AuNPs의 위치를 변경하려면 DNA 종이접기 구조 내의 스테이플 디자인을 수정해야 했습니다. 이를 해결하기 위해 DNA 가닥의 연결을 제어하는 방법들이 연구되었지만, DNA-금 나노어셈블리의 재구성은 DNA 가닥의 on-off 메커니즘을 통해 제한된 상태로 한정되었습니다. 여기서, 우리는 다양한 반응을 위해 DNA 인터칼레이터를 통한 비틀림 변형을 조절함으로써 DNA-금 나노어셈블리 내 카이랄 배열의 점진적이고 연속적인 구조변경을 유도하는 방법을 제안합니다. DNA 인터칼레이터가 DNA 가닥에 결합함으로써 두 AuNP 사이의 DNA-금 나노어셈블리의 단위 블록이 뒤틀려 단위 블록 상의 이웃하는 AuNPs 간의 상대적인 위치가 변경되었습니다. 이러한 단위 블록의 축 방향 반복은 비틀림 변형을 통해 AuNP의 카이랄 배열을 변경함으로써 키롭티컬 반응을 제어할 수 있었습니다. AuNP의 카이랄 배열에서 이러한 변화의 민감도는 DNA 인터칼레이터 종류, 그들의 혼합물, 그리고 DNA 오리가미 구조의 강성에 따라 제어될 수 있었습니다. 단위 블록의 측면 연결은 비틀림 변형을 통한 나노힌지 팔의 카이랄 배열을 제어함으로써 다양한 각도로 힌지 운동 반응을 달성할 수 있게 했습니다. 나노힌지의 민감도는 DNA 인터칼레이터 종류와 그들의 혼합물에 의해 제어되는 반면 작동 범위는 두 AuNP 사이의 거리에 따라 조정되었습니다. 또한 나노힌지로 연결된 폴리머는 회전 각도를 축적함에 따라 구부러지는 움직임을 보였습니다. 요약하면, DNA-금 나노어셈블리 내 카이랄 배열의 점진적이고 연속적인 구조변경을 통해 반응을 미세하게 제어할 수 있었습니다. 이는 DNA 인터칼레이터에 의해 유도된 나노어셈블리 내 단위 블록의 비틀림 변형을 조정함으로써 달성되었습니다. 이 접근 방법은 메타표면, 나노센서, 나노로봇와 같이 다양한 상태가 필요한 나노디바이스를 효율적으로 제작할 수 있는 잠재력을 가지고 있습니다.Abstract 1
Table of Contents 3
List of Figures 5
List of Tables 17
Chapter 1. Introduction 18
1.1. DNA-gold nanoassembly 18
1.2. Reconfiguration of DNA-gold nanoassembly 20
1.3. Research objective and outline 23
Chapter 2. Chemo-Mechanical Deformation of DNA 25
2.1. Torsional deformation induced by DNA intercalator 25
2.2. Unit block in DNA-gold nanoassembly 27
Chapter 3. Controlling Chiral Arrangement of AuNPs 30
3.1. Introduction 30
3.2. Coupling of nanoparticles for various chiroptical responses 32
3.3. Reconfiguration of AuNP arrangement 35
3.4. Ag enhancement 44
3.5. Sensitivity depending on DNA intercalator type 49
3.6. Sensitivity depending on stiffness 54
3.7. Conclusion 61
3.8. Materials and methods 62
Chapter 4. Controlling Chiral Arrangement of Nanohinge Arms 70
4.1. Introduction 70
4.2. Coupling of torsional deformation for rotational motion 73
4.3. Rotational motion of the nanohinge 76
4.4. Performance control of the nanohinge 88
4.5. Polymerization of nanohinges for bending motion 96
4.6. Conclusion 105
4.7. Materials and methods 106
Chapter 5. Conclusion 113
Appendix 117
A1. CD and absorbance spectra data 117
A2. AFM images and histograms of nanohinge monomers 134
A3. AFM images of nanohinge polymers 154
Bibliography 162
Abstract in Korean 170박
고갈 가스전에서의 이산화탄소 주입에 대한 복합거동 지오메카닉스 해석
No full text학위논문(박사) -- 서울대학교 대학원 : 공과대학 에너지시스템공학부, 2024. 8. 민기복.The safe and effective implementation of CO2 injection, for both geological carbon storage and minimization of depletion-induced seismicity, relies on a thorough understanding of reservoir compaction and mechanical stability. This dissertation presents a comprehensive numerical investigation into coupled geomechanical responses, including stress state evolution, reservoir compaction, and induced seismicity potential, during natural gas production and CO2 injection in a depleted gas reservoir.
This work first employs a generic model of two-dimensional multi- layered sedimentary system that includes two sealed bounding faults. Coupled thermo-hydro-mechanical simulations were conducted for a hypothetical study on geological carbon storage in a depleted gas reservoir, incorporating the inelastic component in reservoir compaction characterized by the modified Cam-clay model. Based on the reshaped recognitions on irreversible compaction impact, the methodology was then applied to a three- dimensional site-specific model, using the geological setting and available data from the Groningen gas reservoir. The geomechanical investigation has been integrated with the reservoir simulation, to provide important insights into a potential solution involving CO2 injection for depletion-induced seismicity.
In the generic two-dimensional multi-layered sedimentary system, simulations exhibited various levels of reservoir compaction during gas extraction under different pre-consolidation conditions of the sediments, highlighting the essential role of reservoir compaction in geomechanical analysis. The results demonstrated the occurrence of strain-hardening compaction behavior in the reservoir during post-yield reservoir depletion. This irreversible compaction was accompanied by a significant reduction in porosity and permeability, as well as irreversible surface subsidence. Hysteresis in the stress state was induced by irreversible reservoir compaction through two primary mechanisms: thermo-poro-elastoplastic stressing and differential compaction on each side of the sealing fault. Moreover, caprock deformation and fault reactivation were impeded and delayed by the irreversible reservoir compaction, owing to the resulting alteration in magnitude and orientation of stress inside and outside the depleted reservoir.
The three-dimensional application study in the Groningen gas field includes numerical modeling of historical gas production, capturing the evolution of pore pressure and subsidence. Successful modeling of production-induced seismicity demonstrated changes in the Coulomb Failure Function on fault elements, and the irreversible reservoir compaction contributed to stabilizing nearby faults. The capacity to reproduce the correlation between production, compaction, and seismicity enabled predictions of the effect of CO2 injection. Results revealed that the effectiveness of CO2 injection in mitigating seismicity and subsidence was attributed to pressure maintenance and rebalancing. Uncertainties existed due to hydraulic heterogeneity; for example, the less permeable claystone layer exhibited mechanical instability during the shut-in period, persisting even during CO2 injection.
This dissertation underscores the important role of irreversible reservoir compaction in evaluating the sealing integrity for geological carbon storage in a depleted gas reservoir. In the context of managing depletion-induced seismicity risks, the findings emphasize the need for incorporating both reservoir and geomechanical modeling to address the challenges posed by reservoir heterogeneity. This work provides valuable insights for safe and effective CO2 injection in a depleted gas reservoir, either for long-term geological storage or induced-seismicity mitigation.본 연구는 천연가스 개발 및 이산화탄소 주입 기간 동안 저류층의 역학적 반응, 응력 상태의 진화, 압축 및 유발 지진 가능성에 대해서 열-수리-역학적 복합거동 수치해석을 통해 검토하였다. 이 연구에서는 경계 단층을 포함하는 다층 퇴적 시스템에 열-수리-역학 복합거동을 고려한 모델을 사용하였으며, 이는 Groningen 저류층의 지질학적 조건 및 이용 가능한 데이터를 기반으로 한 3차원 현장 특화 모델을 통해 보완되었다.
본 연구를 통해 항복 이후 고갈로 인한 평균 유효응력 증가 및 편차 응력 감소와 함께, 가스 생산 후 비가역 압축으로 인한 공극률 및 유체 투과도의 감소가 확인되었다. 응력 크기와 방향의 변화는 다공성-탄소성 응력과 차등적인 압축에 의해 영향을 받는 것으로 나타났다. 비가역 압축은 표면 침하를 악화시키고 단층 주위의 전단 응력을 증가시키지만, Groningen 예측 연구에서 관찰된 바와 같이 이산화탄소 주입을 통한 부분적인 완화도 관찰된다.
비가역 압축은 약 2년간 덮개암 압축을 지연시키고 그 정도를 줄여, 최대 2 MPa까지의 덮개암 균열 압력의 과대 추정을 방지할 수 있는 것으로 나타났다. 비가역적 압축을 고려하지 않을 경우, 단층 미끄러짐을 유발하기 대한 임계 압력이 4.9 MPa까지 과소 추정될 수 있다. Groningen 지역에서는 주입중단 기간 동안 침투성이 낮은 점토층이 역학적 불안정성을 보이며, 이는 이산화탄소 주입 중에도 지속된다. 나아가, 비가역 압축은 주변 단층의 안정화에 기여하는 것으로 나타났다.
CO2 주입은 다상 및 다성분 동적 흐름을 통해 부력 효과를 활용하면서 효과적으로 압력을 유지하고 조절할 수 있음을 확인되었다. 대규모 유체 주입을 통해 가스 생산 후 유발 지진의 위험을 완화할 수 있지만, 위험성을 완전히 배제할 수는 없다.
이 연구는 이산화 탄소 주입의 안전한 운영을 위해 비가역 압축의 고려 및 주입 전략 조정의 중요성을 강조하며, 지진 위험을 관리하기 위한 방법을 제공한다.Chapter 1 Introduction 1
1.1. Background 1
1.2. Motivation 5
1.3. Purpose of Research 9
1.4. Outline of Dissertation 11
Chapter 2 Literature review 13
2.1. Compaction and subsidence 13
2.2. Induced seismicity 21
2.3. Production, compaction and seismicity in Groningen 32
Chapter 3 Theory and Methodology 38
3.1. Irreversible reservoir compaction 38
3.2. Caprock integrity 40
3.3. Fault reactivation 42
3.4. Numerical code 43
3.4.1 TOUGH2 43
3.4.2 FLAC3D 48
3.4.3 TOUGH2-FLAC3D 50
Chapter 4 Effect of irreversible compaction on CO2 storage in a depleted reservoir 61
4.1. Model descriptions 61
4.1.1 Model setup 61
4.1.2 Material properties 66
4.2. Effects of irreversible reservoir compaction 72
4.2.1 Reservoir compaction 72
4.2.2 Hysteresis in stress state 76
4.2.3 Surface deformation 83
4.3. Geomechanical impacts during CO2 injection in depleting reservoirs 85
4.3.1 Caprock integrity 85
4.3.2 Fault reactivation 88
4.4. Thermal effects 92
4.5. Summary 94
Chapter 5 CO2 injection into gas reservoirs in Groningen, Netherlands 98
5.1. Overview of Groningen gas reservoirs 98
5.1.1 Lithostratigraphy 99
5.1.2 Fault system: sealing capacity and contribution to seismic moment 101
5.2. Model descriptions 104
5.2.1 Model setup 104
5.2.2 Material properties 112
5.3. Results of gas production modeling 117
5.3.1 Pressure evolution 117
5.3.2 Stress state evolution 120
5.3.3 Ground subsidence 123
5.3.4 Fault stability 125
5.3.5 Comparison with historical measurements 129
5.4. Prediction of CO2 injection in depleting reservoirs 135
5.4.1 CO2 injection 135
5.4.2 Ground subsidence recovery 136
5.4.3 Risk of induced seismicity 138
5.5. Summary 142
Chapter 6 Discussions and Conclusions 144
6.1. Discussions 144
6.1.1 Phase of injected CO2 144
6.1.2 Trigger factor in fault reactivation analysis 146
6.1.3 Viscosity effects 147
6.2. Conclusions 150
References 155
초 록 173박
OSA Diagnosis with Multi-night Self-Sleep Data
No full text학위논문(석사) -- 서울대학교 대학원 : 데이터사이언스대학원 데이터사이언스학과, 2024. 8. Hyung-Sin Kim.본 연구는 가정 기반의 수면 데이터를 이용한 폐쇄성 수면 무호흡증 (OSA) 진단의 가능성을 탐구한다. 세계 최대 규모의 147명의 환자로부터 여러 밤의 가정 기반 수면 데이터를 총 1,147건 수집하여 해당 데이터의 심전도 데이터를 활용해 폐쇄성 수면 무호흡증 중증도 여부를 예측한다. 비지도식 도메인 적응 기법의 적용으로 임상 다중수면기록 (PSG) 데이터와 가정 기반 데이터를 연결하여 가정 기반 데이터의 라벨링 한계점을 극복하고 모델의 성능 향상을 도모하였다. 또한 테스트 시간 적응을 통해 해당 모델을 개인화하고 정확성을 높여 진단 변동성을 감소시켰다. 본 연구 결과는 가정 기반 수면 모니터링 체계의 잠재성을 보여주며, 여러 밤 동안의 가정 기반 진단 프레임워크의 실현 가능성을 처음으로 입증하고 실용적인 OSA 진단의 가능성을 제시한다.This study explores the feasibility and potential of home-based sleep data for diagnosing sleep disorders, particularly Obstructive Sleep Apnea (OSA). Leveraging the Self Sleep Dataset, which includes ECG data collected from natural home environments over multiple nights, we employed Unsupervised Domain Adaptation (UDA) techniques to bridge the gap between clinical Polysomnography (PSG) data and home-based data. Our results demonstrated significant improvements in model performance, with notable increases in accuracy and F1 scores, underscoring the robustness of our approach in handling environmental and demographic shifts. To further enhance model performance and personalize the diagnostics for individual patients, we implemented Test-Time Adaptation (TTA) using Domain Unsupervised Adaptation (DUA). This approach led to additional performance gains and reduced individual fluctuations, highlighting the importance of multi-night data collection in mitigating the first-night effect and providing consistent diagnostic results. Despite the promising findings, the study also acknowledges limitations such as the inherent noise in the dataset and the challenges posed by labeling home-based data with clinical PSG standards. Future work will focus on simpler binary classification tasks to demonstrate the dataset's value more effectively. By targeting specific cutoffs corresponding to mild, moderate, and severe OSA thresholds, we aim to develop screening models that can accurately identify patients needing further medical evaluation. This approach aligns with the primary goal of home-based apnea testing and provides a practical path for validating the dataset's utility in real-world applications. Overall, this research supports the potential of home-based sleep monitoring systems, offering a convenient, cost-effective, and accurate alternative to traditional clinical PSG, and underscores the necessity of multi-night data collection to ensure reliable sleep disorder diagnostics.국문 초록 i
Abstract ii
Chapter 1 Introduction 1
Chapter 2 Background 6
2.1 Home Sleep Apnea Test (HSAT) 6
2.2 ECG-based OSA detection 7
2.3 Unsupervised Domain Adaptation 7
2.4 Test-time Adaptation 9
Chapter 3 Method 11
3.1 Overview 11
3.2 Dataset 11
3.3 Supervised model training with PSG Dataset 13
3.4 UDA model training with Home-based Dataset 14
3.5 Patient-wise Adaptation via TTA 15
Chapter 4 Evaluation 18
4.1 Domain Shifts between PSG and Self Sleep Dataset 18
4.2 PSG limitations observed with Multi-night Home Dataset 20
4.2.1 Night-to-Night Variability 20
4.2.2 First-Night Effect 22
4.3 Demographic Shift Analysis and TTA 23
4.4 Performance Comparison 26
Chapter 5 Conclusion 27
5.1 Conclusion 27석