Personal tools

SC²S Colloquium - July 24, 2015

From Sccswiki

Jump to: navigation, search
Date: July 24, 2015
Room: 02.07.023
Time: 9:00 am, s.t.

Shashank Sharma: CFD based design optimization for a cooling jacket of an electric machine

In electric drive machines (E-machines) high energy losses are partly dissipated as heat during the conversion from electrical into mechanical energy. If the heat cannot be sufficiently and effectively removed, the components integrated within the electric drive machines can be irreversibly damaged and can lead to a strong reduction in the lifetime of the thermal critical components. In order to manage the thermal output of the E-machines, cooling jackets which are commonly integrated in housings, are effectively used. The important design objectives of a cooling jacket are to achieve uniform flow with minimal pressure-drop and to maximize the convective heat transfer. Manual shape optimization, which often used to obtain an optimized shape is often time and resource consuming. In order to eliminate the manual design overhead and reduce the computational cost for design optimization process a study is performed in this thesis to explore the viability of adjoint methods for the design optimization process. The other objectives are to obtain the design optimization for the geometries under consideration with respect the aforementioned design objectives and to evaluate the effects of adjoint solver parameters on the optimization procedure. To achieve the above stated objectives, a careful study of the adjoint methods is done. It is followed by an intensive evaluation of ANSYS FLUENT adjoint solver in terms of usability, numerical robustness and performance. Various single- and multi-objective simulations are performed on three geometries, i.e. a generic U–bend tube, a cooling jacket segment geometry and cooling jacket full geometry. The simulations are analyzed to identify the regions of change and high sensitivities. The design optimization of U-bend and segment model shows improvement in the design variables. The simulations for the complete geometry are under way.