Real-time or faster-than-real-time simulation of airflow in buildings

Abstract

Real time flow simulation is crucial for emergency management in buildings, such as fire and accidental or intentional release of chemical/biological agents (contaminants). The simulation results can then be used to impose proper measures to minimize casualties. The computational fluid dynamics (CFD) is accurate, but too time consuming. Nodal models are fast, but not informative. To obtain a fast and informative solution, this study proposes an intermediate approach between nodal models and CFD by introducing a Fast Fluid Dynamics (FFD) method. This investigation used the FFD methods with and without turbulence treatments to study systematically four basic flows in buildings and compared the numerical results with the corresponding CFD results and the data from the literature. The results show that, on one side, the FFD can offer much richer flow information than nodal models do, but less accurate results than CFD does. On the other side, the FFD is 50 times faster than the CFD. The results also show that the FFD with the laminar assumption has the best overall performance on both accuracy and speed. It is possible to conduct faster-than-real-time flow simulations with detailed flow information by using the FFD method. Practical ImplicationThe paper introduces a Fast Fluid Dynamics (FFD) method, which can simulate airflow and contaminant dispersion in buildings with real time or fast-than-real-time speed and provide informative solutions. As an intermediate approach between nodal models and the CFD, the FFD can be a very useful tool for emergency management in case of fire and accidental or intentional release of chemical or biological agents in a building or around the buildings. The FFD can also be used as a preliminary test tool for fast assessment of indoor airflows before a detailed CFD analysis.

Numerical optimisation of thermal comfort improvement for indoor environment with occupants and furniture

“…The Reynolds-Averaged Navier-Stokes (RANS) equations are adopted together with the two-equation renormalized group RNG k-ε turbulence model, due to its capability of accurate prediction of turbulent indoor airflows at low-Reynolds number with and without flow swirl, as previous studies have shown good model performances in terms of accuracy, numerical stability and short computing time [36,39]. In the energy equation, radiation heat generation from a heating source is also included through a Discrete Ordinates (DO) model already implemented in ANSYS Fluent software and it is applied with various angular discretisation and sub-iteration parameters to control solid angles in discretising each octant of the angular space and volume overhang on each surface respectively, so that radiative conditions can be applied to each individual faces and fluid elements within the computational domain.…”

Reduction of Numerical Diffusion in FFD Model

“…On the other hand, the numerical viscosity is not turbulent viscosity so that the FFD model has difficulties to precisely predict the turbulent flows. In addition, simply adding turbulence models into the FFD cannot improve the prediction for turbulence (Zuo and Chen, 2009) since the flow will become too viscous with both numerical and turbulent viscosities. To improve the accuracy of the FFD model, it is necessary to reduce the numerical viscosity.…”

“…However, most indoor airflow is turbulent (Lariani et al, 2009). Our previous work (Zuo and Chen, 2009) tried to predict turbulent flow by adding turbulence models in the FFD. Unfortunately, it did not provide satisfactory results due to the large numerical viscosity in the linear interpolation.…”

“…The computational fluid dynamics (CFD) (Chen, 2009) can provide the information that are more accurate than the requirements, but its computing speed is too slow. As an intermediate method between the multizone models and the CFD, the FFD method (Zuo and Chen, 2009) was proposed to fill the gap by providing more detailed flow information than the multizone models at a speed much faster than the CFD. On the other hand, the FFD is slower than the multizone models and less accurate than the CFD.…”

“…Since his goal was to generate plausible effects of flow motion in real time, Stam (1990) mainly adopted low order numerical schemes due to their simplicity. By scarifying some accuracy, the FFD can solve the Navier-Stokes equations at a speed of 50 times faster than the CFD (Zuo and Chen, 2009). It can provide detailed flow information, such as flow velocity, air temperature and contaminant concentration with sufficient accuracy for building conceptual design and emergency management.…”