A Cartesian Grid Multilevel Multigrid Method for Steady-State Turbulent Flows with Irregular Geometries.
ORAL
Abstract
More than 55% of the global population currently resides in urban areas, a figure projected to rise to 68% by 2050. High population density and limited evacuation routes significantly elevate the risk of human exposure to airborne contaminants during accidental or deliberate chemical or biological releases. Fast and accurate urban wind and dispersion modelling is therefore essential for effective emergency response.
Achieving the fast turnaround required for emergency response necessitates models that are both highly automated and computationally efficient. Over the past two decades, Cartesian grid methods with embedded boundaries have emerged as one such approach for solving the Navier-Stokes equations over complex geometries. These methods offer key advantages, including automated mesh generation, high accuracy, and rapid simulation turnaround. On the computational side, multilevel solvers remain among the fastest iterative techniques for solving PDEs.
To meet these demands, we present the development of a Cartesian grid solver that combines a cut-cell embedded boundary method with a multilevel algorithm to achieve both efficiency and automation. Developed using the AMReX framework, the solver computes steady-state, incompressible RANS equations using the SIMPLE algorithm. Turbulence is modelled with the k-omega model with robust wall treatment adapted for cut-cell grids. We present the solver’s numerical methods, validation, and scaling results.
Achieving the fast turnaround required for emergency response necessitates models that are both highly automated and computationally efficient. Over the past two decades, Cartesian grid methods with embedded boundaries have emerged as one such approach for solving the Navier-Stokes equations over complex geometries. These methods offer key advantages, including automated mesh generation, high accuracy, and rapid simulation turnaround. On the computational side, multilevel solvers remain among the fastest iterative techniques for solving PDEs.
To meet these demands, we present the development of a Cartesian grid solver that combines a cut-cell embedded boundary method with a multilevel algorithm to achieve both efficiency and automation. Developed using the AMReX framework, the solver computes steady-state, incompressible RANS equations using the SIMPLE algorithm. Turbulence is modelled with the k-omega model with robust wall treatment adapted for cut-cell grids. We present the solver’s numerical methods, validation, and scaling results.
–
Presenters
-
Ankith Anil Das
University of Sydney
Authors
-
Ankith Anil Das
University of Sydney
-
Nicholas J Williamson
University of Sydney
-
Steven W Armfield
University of Sydney