Cities are classified as the urban categories in the 24-class USGS land-use classification with high roughness length as the characteristics and Monin Obunkov Similarity Theory (MOST) is traditionally applied to simulate boundary layer dynamics based on surface energy fluxes resulted in over-prediction in wind speed and under-prediction in temperature in night time in the weather forecast. Momentum and thermal drag effects together with a building energy model are crucial to represent the presence of buildings impact on momentums and thermals in the boundary layer. In this thesis, a realistic multi-layer approach will be introduced to the ACM2 model in the WRF model. Sensitivity tests to the urban morphological parameters in idealized simulations are analysed and the results are consistent to the physics. Simulations based on real urban morphological data are also performed. We will show there are significant improvements on wind prediction.