Recent progress in optimization of RANS turbulence models for accurate urban airflow and contaminant dispersion simulations

Modeling urban airflow and contaminant dispersion is critical for maintaining a healthy and safe outdoor environment. Reynolds-Averaged Navier-Stokes (RANS) turbulence models, while computationally efficient, often lack accuracy, limiting the reliability of airflow and pollutant dispersion predictions. This literature review explores the latest optimization techniques for improving RANS model accuracy in urban airflow and contaminant dispersion simulations. The paper discusses the challenges with traditional RANS models, including uncertainties in the model structure, parameters, and dispersion sub-models. It also reviews recent optimization strategies, such as machine learning methods, surrogate models, hybrid approaches, and Physics-informed neural networks (PINNs). Through case studies and applications, the paper assesses the effectiveness of these techniques and highlights future research directions. By synthesizing existing research, this review offers insights into optimizing RANS models and enhancing urban environmental modeling.