Wang et al. (2026) Flood simulation and risk assessment in urban underground spaces based on 3D laser scanning: capacity–depth–damage curves and computational fluid dynamics-based flood response

Identification

• Journal: Frontiers in Water
• Year: 2026
• Date: 2026-03-09
• Authors: Yan Wang, Zhao Cai, Shaozhi Chu, Peng Liu, Hongwei Liu, Jie Lin, Li Tang
• DOI: 10.3389/frwa.2026.1777013

Research Groups

• Nanjing Hydraulic Research Institute, Nanjing, China
• State Key Laboratory of Hydrology, Water Resources and Hydraulic Engineering Science, Nanjing, China
• Water Conservancy Development Research Center, Taihu Basin Authority, Shanghai, China

Short Summary

This study develops and cross-validates an integrated framework for flood simulation and risk assessment in urban underground spaces, combining a rapid rainfall-informed capacity–depth–damage (C–D–D) curve method with a detailed computational fluid dynamics (CFD) inundation model. The framework provides actionable risk indicators, demonstrating that higher rainfall intensities significantly reduce evacuation windows, with both methods showing strong consistency in threshold timing while CFD offers superior spatial resolution for early risk identification.

Objective

• To develop and cross-validate an integrated framework combining a rainfall-informed capacity–depth–damage (C–D–D) curve method and a 3D computational fluid dynamics (CFD) inundation model to translate inundation dynamics in urban underground spaces under extreme rainfall into actionable risk indicators (e.g., depth thresholds, arrival times, and spatial risk maps).

Study Configuration

• Spatial Scale: A single-level urban underground parking garage (approximately 4,000 m²) in Tongzhou District, Beijing, China, was used as a case study. The contributing surface watershed area was 0.1 km², with an effective underground-space catchment of 0.0538 km².
• Temporal Scale: Flood dynamics and risks were evaluated under design rainfall scenarios with annual exceedance probabilities of 1%, 2%, and 5% (corresponding to 100-year, 50-year, and 20-year return periods, respectively). Simulations tracked inundation processes over periods of up to 41 hours.

Methodology and Data

• Models used:
  • Rainfall-informed Capacity–Depth–Damage (C–D–D) curve method: A rapid, screening-level tool linking rainfall-driven inflow to depth evolution and warning indicators, based on empirical inflow relationships and geometric storage capacity.
  • Computational Fluid Dynamics (CFD) inundation model: A physics-based, 3D-geometry-resolved model solving the continuity and Navier–Stokes equations, coupled with the Volume of Fluid (VOF) multiphase model and Realizable k-ε turbulence model, using the SIMPLE algorithm for pressure-velocity coupling.
• Data sources:
  • 3D Laser Scanning: A GeoSLAM handheld 3D Laser Scanning system (ZEB-Horizon with Velodyne VLP-16 LiDAR sensor and industrial-grade IMU) was used to reconstruct high-resolution, as-built 3D geometry of the underground space.
  • Design Rainfall Scenarios: Inflow boundary conditions for the models were derived from design rainfall scenarios with annual exceedance probabilities of 1%, 2%, and 5%.
  • Experimental Data: A custom-designed laboratory experimental apparatus was used to validate the CFD model's accuracy in simulating water level variations under controlled inflow conditions.
  • Geographic Data: Local topography and road network data were used to delineate the watershed and underground-space catchment areas.

Main Results

• Stronger rainfall intensities significantly advance the attainment of critical water-depth thresholds and compress evacuation windows. For instance, under a 1% annual exceedance probability (P = 1%), the 0.2 meter alert depth occurs 1.5 hours earlier than under P = 2% and 5.3 hours earlier than under P = 5%.
• Under the P = 1% scenario, water intrusion begins at 22.6 hours, reaching the 0.2 meter depth threshold by 23 hours, the 1.23 meter electrical equipment threshold by 25.4 hours, and a maximum depth of 3.052 meters by 40 hours. The Waterlogging Disaster Risk (WDR) enters the high-risk level after 28 hours.
• The C–D–D curve method and the CFD model demonstrate strong consistency in predicting threshold attainment times, with biases typically within 0–1 hour and isolated cases not exceeding 2 hours.
• The CFD model, due to its ability to resolve spatial heterogeneity, identifies localized medium-to-high risk zones near the entrance and ramp approximately 1 hour earlier in the intrusion stage compared to the C–D–D curve method, which uses cross-section-averaged velocities.
• Risk levels monotonically increase with inundation progression; under the P = 1% scenario, approximately 5 hours after inflow onset (when depths exceed 1 meter), the underground space is effectively classified as entirely extreme risk.
• The study suggests that with a 0.2 meter entrance baffle, approximately 24 hours after rainfall onset represents the optimal window for occupant evacuation under the P = 1% scenario, as high- and extreme-risk zones rapidly pervade the garage by 25 hours.

Contributions

• Integration of handheld 3D Laser Scanning with a novel "curve-first, CFD-refine" dual-model workflow for comprehensive flood risk assessment in urban underground spaces.
• Cross-validation of rapid C–D–D-based warning thresholds against geometry-resolved CFD dynamics, demonstrating the consistency and complementary nature of the two approaches.
• Delivery of actionable time-to-threshold warnings and spatially explicit risk maps, providing crucial information for emergency planning, evacuation window assessment, entrance protection design, and drainage capacity sizing.
• Overcoming the limitation of idealized geometric layouts in previous studies by utilizing high-resolution 3D laser scanning data to enable geometry-specific inundation and warning-threshold predictions.

Funding

• National Natural Science Foundation of China (42201048)
• Postgraduate Thesis Fund of Nanjing Hydraulic Research Institute (Yy525022)
• National Key R&D Program of China (2022YFC3090601-2)

Citation

@article{Wang2026Flood,
  author = {Wang, Yan and Cai, Zhao and Chu, Shaozhi and Liu, Peng and Liu, Hongwei and Lin, Jie and Tang, Li},
  title = {Flood simulation and risk assessment in urban underground spaces based on 3D laser scanning: capacity–depth–damage curves and computational fluid dynamics-based flood response},
  journal = {Frontiers in Water},
  year = {2026},
  doi = {10.3389/frwa.2026.1777013},
  url = {https://doi.org/10.3389/frwa.2026.1777013}
}