Improving monitoring network design to detect leaks at hazardous facilities: Lessons from a CO₂ storage site

Exploring the challenges posed by uncertainties in numerical modeling for hazardous material storage, this study introduces methodologies to improve monitoring networks for detecting subsurface leakages. The proposed approaches were applied to the Korea CO₂ Storage Environmental Management (K-COSEM) test site, undergoing calibration, validation and uncertainty analysis through hydraulic and controlled-CO₂ release tests. The calibration phase involved inter-well tracer and multi-well pumping tests, leveraging the Parameter ESTimation (PEST) model to determine the aquifer flow and solute transport properties of the K-COSEM site. To tackle uncertainties with limited observation data, we adopted Latin Hypercube simulation. Our uncertainty analysis confirmed model accuracy in simulating observed CO₂ breakthrough curves. We also explored a probabilistic method to identify the environmental change point (EnCP) through correlation analysis with the distance from the CO₂ injection well, revealing a linear trend and pinpointed potential preferential flow pathways by assessing detection probabilities. Evaluating CO₂ detection capabilities was crucial for optimizing monitoring well placement, highlighting strategic well selection based on detection probabilities. This study advances managing uncertainties in hydrogeological modeling, underscoring the importance of sophisticated models in designing monitoring networks for hazardous leak detection in complex subsurface conditions.