Accelerating flow simulations in the built environment by using the fast fluid dynamics initializer

Abstract

This study introduces the Fast fluid dynamics (FFD) based initializer as a novel approach to accelerate steady-state flow simulations. Three modes - Coarse-Mesh Mode (CMM), Large Timestep Mode (LTM), and Hybrid Mode (HM) - are developed on the OpenFOAM platform for efficiently simulating airflow in the built environment. All three modes exhibit faster computations than standard initialization methods by enabling quicker attainment of desired flow states and reducing the required matrix iterations. These initializers demonstrate better performance in flow problems characterized by higher Reynolds numbers. The CMM initializer achieves a nearly six-fold increase in speed by mapping initial solutions from a coarse mesh to a fine mesh. The LTM initializer allows for larger timesteps during initialization, resulting in a two-fold speed improvement in fluid simulation. The HM initializer combines the advantages of both CMM and LTM. However, it does not surpass the speed-up performance of the previous two modes. Apart from the benchmark test case, the proposed initializers can be applied to other indoor and outdoor flow scenarios. Furthermore, the simulation process is further streamlined by automating the execution of the initializer using scripts, eliminating the need for manual step-by-step inputs.