Generative Design of Internal Cooling Channels for Gas Turbine Blades

Abstract

Conventional cooling channels inside air-cooled turbine blades were mainly constructed by regular geometric features such as pin-fins, ribs or dimples. The heat transfer potential of such features were already well exploited by the literature in the past few decades, and the cooling performance of turbine blades has reached a bottle neck. Instead of manually crafting regular geometry shapes and making combinations, a novel concept arising in recent years advocated to generate complicated geometries by rules, which was named generative design. This study proposed a geometry generation method based on bionic rules and partial differential equations. The concept was inspired by the self-organization phenomenon of natural objects. Heat transfer evaluation was conducted on multiple geometries generated in different domains, with different length scale and with different equation parameters. The results indicated a wide range of heat transfer performance that generatively designed cooling channels could provide. Geometries with low porosity or high connectivity of the solid tended to have better heat transfer performance. Complex geometries were generated successfully under the control of the initial value of the phase filed. This method was expected to produce new possibility for enhancing cooling performance of turbine blades.