abstract: This thesis focuses on an improved understanding of the dynamics at different length scales of wind farms in an atmospheric boundary layer (ABL) using a series of visualization studies and Fourier, wavelet based spectral analysis using high fidelity large eddy simulation (LES). For this purpose, a robust LES based neutral ABL model at very high Reynolds number has been developed using a high order spectral element method which has been validated against the previous literature. This ABL methodology has been used as a building block to drive large wind turbine arrays or wind farms residing inside the boundary layer as documented in the subsequent work. Studies conducted in the thesis involving massive periodic wind farms with neutral ABL have indicated towards the presence of large scale coherent structures that contribute to the power generated by the wind turbines via downdraft mechanisms which are also responsible for the modulation of near wall dynamics. This key idea about the modulation of large scales have seen a lot of promise in the application of flow past vertically staggered wind farms with turbines at different scales. Eventually, studies involving wind farms have been progressively evolved in a framework of inflow-outflow where the turbulent inflow is being fed from the precursor ABL using a spectral interpolation technique. This methodology has been used to enhance the understanding related to the multiscale physics of wind farm ABL interaction, where phenomenon like the growth of the inner layer, and wake impingement effects in the subsequent rows of wind turbines are important owing to the streamwise heterogeneity of the flow. Finally, the presence of realistic geophysical effects in the turbulent inflow have been investigated that influence the flow past the wind turbine arrays. Some of the geophysical effects that have been considered include the presence of the Coriolis forces as well as the temporal variation of mean wind magnitude and direction that might occur due to mesoscale dynamics. This study has been compared against field experimental results which provides an important step towards understanding the capability of the mean data driven LES methodology in predicting realistic flow structures. / Dissertation/Thesis / Turbulent flow past a large periodic wind farm / Doctoral Dissertation Mechanical Engineering 2018