Abstract
Computational fluid dynamics can be time consuming for predicting indoor airflows and pollutant transport in large-scale problems or emergency management. Fast fluid dynamics (FFD) is able to accomplish efficient and accurate simulation of indoor/outdoor airflow. FFD solves the advection term of the Navier–Stokes equations either by a semi-Lagrangian (SL) scheme or an implicit upwind (IU) scheme. The SL scheme can be highly efficient, but its first-order version is not conservative and introduces significant numerical diffusion. To improve its accuracy, a high-order temporal and interpolation scheme that not only reduces dissipation and dispersion errors but also guarantees the convergence speed should be applied. Otherwise, an IU scheme instead could be used to solve the advection term. The IU scheme is conservative and introduces minor numerical diffusion, but it may increase the computation time. Therefore, this study investigated the performance of FFD with SL scheme using high-order temporal and interpolation schemes and that with IU scheme. The comparisons used experimental data of two indoor airflows and one outdoor airflow. The results showed that FFD with IU scheme was overall more accurate than FFD with SL scheme. In simulating indoor airflow, both methods were robust and the predictions were independent of time step sizes if the Courant number was less than or equal to one. In simulating the outdoor airflow, the FFD with SL scheme performed better than the FFD with IU scheme for large time step sizes. The FFD with IU scheme consumed 44%–61% computing time of the FFD with SL scheme.
Highlights
The computational fluid dynamics (CFD) has been introduced to the indoor environment community for the prediction of air distribution since 1970s [1]
This investigation compared the performance of Fast fluid dynamics (FFD) with SL scheme and FFD with implicit upwind (IU) scheme in predicting two typical indoor airflows and one outdoor airflow
For simulating the indoor airflow, the predictions by both methods were independent of time step size when the Comean was less than or equal to one
Summary
The computational fluid dynamics (CFD) has been introduced to the indoor environment community for the prediction of air distribution since 1970s [1]. In the indoor environment community, researchers placed significant efforts on improving the computational speed in terms of parallel computing [15], reduced-order models such as recurrence CFD [16,17], and developing new numerical algorithms [18]. This study implements a non-incremental pressure-correction scheme with SL scheme using third-order backward and forward method (FFD with SL scheme) and that with an IU scheme (FFD with IU scheme) in OpenFOAM This investigation compares the performance of these two solvers in terms of accuracy and efficiency in predicting two typical indoor airflows and outdoor wind. The impact of the time step size was studied
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
