Simulation of urban flooding using 3D computational fluid dynamics with turbulence model
Muhammad Waqar Saleem, Imran Rashid, Sajjad Haider, Mohiq Khalid, Amro Elfeki
Abstract
• 3D CFD simulations accurately predicted urban flood dynamics, achieving a high Nash-Sutcliffe Efficiency (NSE) of 0.83. • Building Hole (BH) method proved most efficient for mesh treatment, reducing computational time significantly. • Sensitivity analysis highlighted the minimal impact of surface roughness on model accuracy. • Advanced mesh refinement correlated directly with improved model accuracy, as evidenced by lower RMSE values. • Employed sophisticated turbulence models to enhance the precision of flood risk assessments in urban settings. An alarming increase in the frequency of extreme rainfall events necessitates advanced flood risk reduction methods. This study uses ANSYS Fluent to simulate the 'Model urban city' flow experiment, comparing aligned and staggered urban layouts. The simulation's goodness of fit, evaluated by flow depth at probe locations, achieved a Nash-Sutcliffe Efficiency (NSE) of 0.83, indicating a 'very good' performance. Sensitivity Analysis (SA) indicated that increasing grid coarseness from 1.7 million to 0.1 million cells raised the Root Mean Square Error (RMSE) by 15%. This Research also assessed different mesh treatments for building incorporation in the computational mesh-building hole (BH) and Building Block (BB). The BH method was computationally more efficient, taking about 1 Day compared to 2-3 days for BB. The Building Resistance (BR) method yielded poor results. Employing a sophisticated 3D CFD model, this study provides detailed inputs for high-precision flood risk assessments in urban areas prone to rapid flooding, enhancing the precision and applicability of urban flood models.