Strategy Of Unsteady Numerical Simulation In Predicting Air Distribution In A Real Aircraft Cabin

Commercial airliners are increasingly used by travelers However, the flyingpublic has demanded for a higher comfort level, because they encounter acombination of environmental factors, including low humidity, low air pressure, andsometimes, exposure to air contaminants, such as ozone, carbon monoxide, variousorganic chemicals, and biological agents. Besides, it has been found that cabinenvironment is associated with the international spread of some severe infectiousdisease such as SARS and HINI in the recent decades. A well designed airliner cabinenvironment is to ensure travelers’ safety, health, and comfort. Hence it is essential tostudy how the air are transported and distributed in cabins. It could be achieved byeither experimental approach or computer modeling. Because the experimentalmeasurements are expensive and slow, computer modeling is increasingly popular.The computer modeling technic of Computational Fluid Dynamics is widely appliedin analyzing the air distribution in enclosed environments, while the normally usedCFD models adopted many assumptions. Thus it is necessary to make comparison ofdifferent turbulence models and figure out the proper boundary condition and timerelated settings to ensure the reliability of CFD simulations.This research demonstrated the numerical simulation approach in predicting theair distribution in the real aircraft cabin. By comparing with the experimental data,this research analyzed the influence from grid-dependency. Then this researchconducted CFD simulation with three kinds of unsteady turbulence models: unsteadyRNG k-ε model, DES with Realizable k-ε model in boundary layers and LES withSmagorinsky-Lilly SGS model. The results showed that unsteady RNG k-ε modelperformed no worse than other two models. Based on the validated grid scale andturbulence model, this research then analyzed the boundary condition settings on thefront and back boundary of the cabin section. Compared with zero-sheared wall andperiodic boundary condition, velocity-inlet with experimental data turned out to be themost accurate in this case. As for the time related settings,0.5τ (50s) and1τ (100s)is enough for the flow field to get stable and gather data for statistics respectively.0.04s would be recommended for the time step setting in the CFD simulation withunsteady RNG k-ε model by balancing the consumption of computational sources andaccuracy of predicting air distribution.