안정적인 이산화탄소 저장을 위한 물-가스 교대 주입 시나리오 최적화 연구 = Optimization of water-alternating gas injection scenarios for stable carbon dioxide storage

다국어 초록 (Multilingual Abstract)

Optimization of water-alternating gas injection scenarios for stable carbon dioxide storage Lim, Sookyung Advisor : Prof. Jang, Ilsik, Ph.D Department of Advanced Energy & Resources Engineering Graduate School of Chosun University. This study aims to analyze CO2 storage patterns under various conditions and develop optimized injection scenarios to advance carbon dioxide capture and storage (CCS) technology. Stable underground storage is essential to prevent CO2 emissions into the atmosphere. Therefore, this research focuses on optimizing injection strategies to reduce CO2 mobility in structural traps while enhancing stable storage through residual and dissolution trapping mechanisms. Three injection methods were compared in this study: continuous injection, periodic injection, and water-alternating gas (WAG) injection. Continuous injection, used as the baseline scenario, involved injecting CO2 continuously for 20 years followed by 180 years of monitoring to analyze CO2 trapping mechanisms. Sensitivity analysis revealed that the WAG injection method provided the most stable results. Among the WAG scenarios, a water injection duration of six months demonstrated the highest stability. Additionally, when comparing injection frequencies ranging from 1 to 20 cycles, the 20-cycle scenario yielded the most effective storage performance. Based on these findings, a model combining ensemble artificial neural network (ANN) and particle swarm optimization (PSO) was applied to optimize the WAG injection scenario. The optimized scenario involved 10 cycles of CO2 and water injection with varying durations to maximize storage stability and efficiency. As a result, the storage efficiency of the optimized WAG scenario reached 9.89%, compared to 7.80% for continuous injection, representing a significant improvement. The optimized WAG scenario, derived through ensemble ANN and PSO, exhibited superior injection stability and higher storage efficiency compared to continuous injection, with reduced CO2 migration toward the upper aquifer. This study provides an effective and stable CO2 injection strategy, contributing to the advancement of CCS technology.

목차 (Table of Contents)

• 제1장 서론 1
• 제2장 이론적배경 7
• 제1절 CCS 개요 7
• 제2절 CO2 트랩 메커니즘과 이력현상 10
• 제3절 머신러닝 개요 17
• 1. 인공 신경망(Artificial Neural Networks, ANN) 19
• 2. 앙상블 기법 20
• 3. 군집화 알고리즘 22
• 제4절 최적화 기법 23
• 1. 입자군집최적화(Particle Swarm Optimization, PSO) 23
• 2. Designed Exploration and Controlled Evolution(DECE) 25
• 제3장 연구 방법 26
• 제1절 CO2 주입 시나리오별 민감도 분석 설계 26
• 1. 대상 저장층 정의 27
• 2. 주입 방식별 시나리오 30
• 제2절 WAG 주입 시나리오 최적화 설계 33
• 제4장 연구 결과 40
• 제1절 CO2 주입 시나리오별 민감도 분석 결과 40
• 1. 저장층 조건에 따른 주입 결과 40
• 2. 휴식 기간과 물 주입 기간에 따른 결과 56
• 3. 반복횟수별 주입결과 59
• 제2절 WAG 주입 시나리오 최적화 결과 63
• 1. 주입 횟수에 따른 훈련데이터 결과값 분석 63
• 2. 주입 시나리오별 최적화 결과 64
• 3. 주입 횟수별 최적화 결과 73
• 제5장 결론 75
• 참고문헌 77