열차 무정차 통과는 역사 군중 밀집을 얼마나 완화하는가 : 10⋅29 참사 당일 이태원역의 시뮬레이션 기반 재구성과 정량 평가

초록

This study quantitatively evaluates how train non-stop pass-through operationsmitigate crowd congestion in an urban railway station, using Itaewon Station on the dayof the 10⋅29 crowd crush as a case study. Urban railway stations face crowd risks dueto excess population movement in narrow vertical corridors over short intervals, yet theeffect of operational interventions such as train control has rarely been quantified. Anintegrated three-dimensional model was constructed using minute-level fare-gateboarding/alighting records and CCTV flow analysis, linking the station interior (platformfare gate–underpass–exit) with the surrounding terrain. An agent-based simulationreproduced the flow during the accident period (18:00–23:00). Three scenarios—normaloperation, alternating-interval non-stop pass-through, and every-interval non-stoppass-through—were compared using the occurrence frequency of risk levels, cumulativeduration, and first-occurrence time under sustained-duration criteria, alongside averageand peak density. As non-stop pass-through frequency increased, the occurrencefrequency and cumulative duration of caution-and-above levels decreased byapproximately 45%–67% at two monitoring points under the every-interval condition. Theonset of high-density states was delayed, leading to an estimated reduction of 1,800–3,600cumulative moving persons over five hours. Although average density remained relativelystable, the occurrence structure and persistence of risk-level density emerged asimportant management indicators. Additionally, crowd risk was governed more by astation's flow-convergence structure than by absolute ridership. The findings supportlinking the current discretion-based non-stop pass-through decision to quantitativedensity indicators.

키워드