Digital twin-based multi-scale framework for seismic integrity assessment of deep underground caverns.

Deep underground caverns have attracted considerable interest in recent years due to their critical role in hydropower and energy infrastructure systems. However, reliable post-earthquake assessment remains challenging due to the complexity of multi-scale seismic response, uncertainty in rock mass properties, and difficulties in interpreting heterogeneous monitoring data. Despite the importance of seismic integrity assessment, there remains a paucity of systematic frameworks that integrate modeling and monitoring in a unified manner. This study therefore proposes a Digital Twin-driven multi-scale framework for seismic integrity assessment of underground caverns, integrating analytical, two-dimensional, and three-dimensional numerical modeling with real-time monitoring data. The aim of this study is to develop a comprehensive framework that enables consistent integration and interpretation of model predictions and monitoring observations. Within this framework, model outputs are defined as Project Intent Information, while monitoring data are defined as Project Status Information, allowing structured comparison through a unified data and knowledge ontology. This study introduces three performance metrics to assess the seismic response, namely the Seismic Integrity Index, Model Consistency Index, and Sensor Consistency Ratio, which collectively provide a quantitative basis for evaluating structural performance and model reliability. In addition, a rule-based multi-sensor consistency logic is developed to distinguish between local anomalies and global structural response, thereby improving the reliability of post-earthquake interpretation. Furthermore, a feedback-based model updating mechanism is implemented to refine numerical models using monitoring-informed interpretation within an integrated Digital Twin-based assessment framework. The proposed approach demonstrates the applicability of an integrated Digital Twin-based framework for model-monitoring-based seismic assessment under the investigated conditions. The results demonstrate that seismic-induced deformation remains limited under high confinement conditions, while the Digital Twin framework enables accurate interpretation of monitoring data and identification of local anomalies. These findings suggest that the proposed approach provides a systematic and scalable methodology for integrating modeling and monitoring in underground seismic assessment.