A GPU-Accelerated Hydrodynamic Model for Urban Rainstorm Inundation Simulation: A Case Study in China

• Published in: KSCE journal of civil engineering 2022, 26(3), , pp.1494-1504
• Main Authors: Han, Hao, Hou, Jingming, Xu, Zongxue, Jing, Haixiao, Gong, Jiahui, Zuo, Depeng, Li, Bingyao, Yang, Shaoxiong, Kang, Yongde, Wang, Run
• Format: Article
• Language: English
• Subjects: Catchment area; Civil Engineering; Construction planning; Disaster management; Disasters; Engineering; Environmental risk; Exact solutions; Flood predictions; Flooding; Floods; Geotechnical Engineering & Applied Earth Sciences; Graphics processing units; Hydrodynamics; Industrial Pollution Prevention; Performance assessment; Rain; Rainfall; Rainstorms; Risk analysis; Shape; Simulation; Spatial variations; Urban areas; Water depth; Water Resources and Hydrologic Engineering; 토목공학
• ISSN: 1226-7988; 1976-3808
• DOI: 10.1007/s12205-021-2158-3

Frequent urban rainstorm inundations can cause serious damages to human life and social economy. Reliable simulation of urban rainstorm inundation is an effective approach for performing flood risk analysis to reduce losses. In this work, a full 2D high-performance hydrodynamic model for urban rainstorm inundation simulation based on Graphic Processing Unit (GPU) was developed. The idealized V-shape catchment and sponge city district of Fengxi New City were selected as the study areas to assess performance of the model. The model was validated against analytical benchmark results of the idealized V-catchment test which show good agreement between the modelled flow and analytical solutions. The model was then applied to simulate actual urban rainstorm inundation process under measured rainfall. The results indicated that the model can be applied to high-resolution urban region simulations. Specifically, the validation results indicated that the proposed model had a good performance, with an accepted error of less than 15%. When applied to compute the Fengxi New city under design rainstorms with different annual return periods of 1, 5, 10, 20, 50, and 100 years, it was found that the model effectively evaluated the temporal and spatial variation process of urban inundation, and quantitatively investigated flood risks according to the water depth change. The model has been substantially accelerated on GPU to quickly predict urban inundation. The accuracy and rapid simulation speed of the model were verified based on the actual study area. Therefore, the proposed model can help to predict the dynamic processes of urban flood inundation and thus reduce flood inundation disasters. This is essential for future optimal sponge urban construction planning in China.