Coupled Computational Fluid Dynamics (CFD) and Artificial Neural Network (ANN) for Prediction of Traffic-related Air Pollution Infiltration Effects in Hong Kong

Improper ventilation of buildings may lead to an accumulation of pollutants indoors. In the case of a room with forced air ventilation and external air intake like most centralized and some home air conditioning units, this study will show CFD simulations of various indoor air quality conditions based on different forced ventilation AC unit intake conditions like common in housing situations like Hong Kong. Especially when close to roadways or other external pollution sources, the positioning of the air intake shows up to have a high significance for the infiltration rate resulting as influence for the indoor air quality as previous research shows (e.g. Zheming Tong et al., 2016). The same is the case for a forced ventilation case like air conditioning units with outside air intake. Research like earlier referenced paper has not been conducted with higher buildings or forced air intake yet. Parametrized CFD-based air quality models with using OpenFoam will be employed to quantify the impact of the air intake location and rate in a 2-dimensional interface on the indoor air quality of a forced ventilated section of a building. The findings of the CFD simulation will be simplified as average indoor air pollution and other external factors. As an approach to predict the estimate indoor infiltration rate, an ANN (Artificial Neuronal Network) will be used, trained and validated with said data. The neural network is supposed to predict the pollutant intake based on fewer and as easier to obtain meteorological parameters and air pollution data. Finally, the ANN predictions of the models will be verified with real life data from other papers. Results will show that a major part of indoor pollutants may emerge indoors and cannot be neglected. In comparison with real life data, it seems the model lacks significant input to predict with high accuracy.