Savonius-type wind turbines (SWT) have low efficiency compared to other wind turbine types, but they have some advantages for urban application which arise from their ability to capture wind from every direction and relatively low rotation speed. In an urban environment, the SWTs are usually mounted near each other thus it is of interest to investigate the SWT wake and its interaction with another turbine. In [1], an SWT performance and wake was studied in a wind tunnel experiment at different Reynolds numbers and inlet turbulence intensities. The comparison of the wake shape with numerical RANS 3D CFD from [2] shows that RANS underpredicts the size of wake while 2D CFD significantly overpredicts. An interaction between multiple SWT turbines was studied in [3] but only using 2D CFD. The objective of the current paper is to evaluate the Hybrid RANS-LES turbulence model for prediction of SWT performance and the prediction of wake behind the turbine. The current paper will use conventional SWT and also for novel optimized design from [2]. The results show that the prediction of the individual SWT performance is not significantly changed compared to RANS CFD model. However, the length of the wake is increased by ~20% brings the numerical model closer to the experimentally measured wake. Another SWT was than placed 5D (D – turbine diameter) downstream at several lateral positions (from -5D to +5D). The SWT performance in the middle of the wake is reduced 4 times while it can be increased by 8% if positioned at +2D. The results show the importance of application of hybrid RANS-LES models for modeling the SWT wake.