단순 대기확산 모델의 염소 확산 예측 및 국내 적용 연구

다국어 초록 (Multilingual Abstract)

Chlorine (Cl2) is a widely used industrial chemical with high acute toxicity, and reliable pr ediction of its atmospheric dispersion is essential for emergency response planning and off -site risk assessment. In practice, dispersion analysis for chemical accidents often relies on simple and readily accessible models that allow rapid scenario evaluation rather than high- fidelity numerical simulations. However, the predictive performance and limitations of such models under large-scale dense-gas release conditions have not been sufficiently validated using full-scale experimental data. This study quantitatively evaluates the performance of commonly used simple atmospheric dispersion models—ALOHA (Gaussian and DEGADIS modes), SLAB, and SCREEN3—for c hlorine releases, using field measurements from the Jack Rabbit II large-scale chlorine rele ase experiments (2015, Trials 1–5). Model predictions were compared with observed maxi mum concentrations over distances ranging from 0.2 to 11 km. Model performance was as sessed using standard statistical metrics, including geometric mean bias (MG), geometric v ariance (VG), and the fraction of predictions within a factor of two of observations (FAC Across the five experimental trials, the SLAB model showed the most consistent and stabl e performance, with an overall MG of approximately 0.9, VG around 2, and the highest F AC2 score (20 out of 30 paired comparisons). ALOHA operating in DEGADIS mode produ ced generally acceptable results but tended to underpredict concentrations in the low-conce ntration mid-to-far field, resulting in shorter predicted impact distances. In contrast, ALO HA (Gaussian mode) and SCREEN3 exhibited larger inconsistencies under dense-gas condi tions; SCREEN3, in particular, underestimated near-field peak concentrations while overesti mating concentrations in the far field, leading to increased scatter and reduced reliability. To examine domestic applicability, two representative chlorine accident scenarios at a Kore an industrial facility were analyzed: (i) a conservative worst-case continuous release over 10 minutes and (ii) a realistic short-duration transfer-line failure with emergency isolation. Toxic impact distances were evaluated using ERPG-1, ERPG-2, and ERPG-3 endpoints. F or the worst-case scenario, SLAB predicted a substantially larger ERPG-2 distance (10 k m) than ALOHA (DEGADIS) (6.8 km), while the regulatory KORA model yielded a shorte r distance (6.34 km). In the alternative scenario, ERPG-2 distances were 1.8 km (SLAB), 1.6 km (ALOHA), and 1.02 km (KORA), with comparable ERPG-3 core regions between S LAB and ALOHA. Overall, the results indicate that SLAB provides the most conservative and reliable basis f or protective distance estimation under dense chlorine release conditions, while ALOHA (D EGADIS) is suitable for near-field assessment and practical risk communication. KORA re sults should be interpreted with caution, as differences in meteorological processing and im plementation can lead to systematically shorter predicted impact distances compared with t he standalone SLAB model. These findings provide practical guidance for model selection and interpretation in domestic chlorine accident risk assessment and emergency planning.

목차 (Table of Contents)

• 제 1 장. 서 론 — 1
• 1.1 연구 배경 및 목적 — 1
• 1.2 연구 방법 — 4
• 제 2 장. 이론적 배경 — 6
• 2.1 염소(Cl₂)특성과 유해성 — 6
• 2.2 대기확산 모델의 분류와 적용 범위 — 8
• 제 3 장. 확산 모델 개요 및 비교 기준 — 11
• 3.1 ALOHA 모델 — 11
• 3.2 SLAB 모델 — 13
• 3.3 SCREEN3 모델 — 14
• 3.4 모델 비교 및 적용상 고려사항 — 15
• 제 4 장. Jack Rabbit II 기반 모델 예측 정확도 분석 — 18
• 4.1 Jack Rabbit II 실험개요 — 18
• 4.2 실험 조건 및 데이터 구성 — 20
• 4.3 모델링 입력 설정 — 24
• 4.4 모델별 예측 결과 — 29
• 4.5 실측값과의 비교 분석 — 33
• 4.6 정량적 정확도 평가 — 36
• 제 5 장. 국내 염소 취급소 누출·확산 시나리오 해석 — 52
• 5.1 사고 시나리오 설정 — 52
• 5.2 입력자료 구성 및 기상조건 설정 — 55
• 5.3 최악의 사고 시나리오 비교 결과 — 58
• 5.4 대안의 사고 시나리오 비교 결과 — 61
• 5.5 KORA 모델링 결과 비교 — 63
• 제 6 장. 결 론 — 67
• 참 고 문 헌 — 71