Combined GIS, CFD and Neural Network Multi-Zone Model for Urban Planning and Building Simulation

Abstract

This study developed a methodology using Geographic Information System (GIS), Computational Fluid Dynamics (CFD) and neural network to help predict the microclimate of the building. The geographic representation of an urban area in Syracuse generated in GIS was converted to the computational domain used in CFD simulation. The flow field around the building was simulated using the CFD model under different wind speeds and directions. Results from CFD simulation could be well presented in GIS using anchored coordinate system. The flow patterns were very similar when the wind speed was varied, while they were highly dependent on the wind directions. However, predicting the flow fields of different wind directions requires running the CFD simulation for each case. A neural network for machine learning was adopted to help predict the microclimate around the building so that much time can be saved. The results show that the proposed neural network has the potential to help predict the microclimate. The predicted microclimate could be used for further study of the building performance.