Textured thrust bearings are capable of providing higher load capacity and lower friction torque compared to non-textured bearings. However, most previous optimization efforts for texturing geometry were focused on rectangular dimples and employed Reynolds equation. Limited studies have been done to investigate the effects of partially textured thrust bearings with elliptical dimples. This study proposes a new optimization approach to find the optimal partially texture geometry with elliptical dimples, which maximize the loading capacity and minimize the friction torque. In this study, a 3D-CFD model for a parallel sector-pad thrust bearing is built using ANSYS-CFX. Mass conserving cavitation model is used to simulate the cavitation regions. Energy equation for Newtonian flow is also solved. The results of the model are validated by the experimental data from the literature. Based on this model, the flow pattern and pressure distribution inside the dimples are analyzed. The geometry of elliptical dimple is parameterized and analyzed using design of experiments. The selected geometry parameters include the length of major and minor axis, dimple depth, radial and circumferential space between two dimples, the radial and circumferential extend. A multi-objective optimization scheme is used to find the optimal texture structure with the load force and friction torque set as objective functions. The results show that the shape of dimples has a crucial effect on the performance of the textured thrust bearings. This optimization approach proposed is expected to be useful in typical texture design process of thrust bearing.