Design of Fully Cooled Turbine Blade Using Shape and Film Cooling Combined Optimization

Description

Gas turbines are widely used in power generation and aircraft propulsion. The efficiency of a  gas turbine can be improved by increasing the turbine inlet temperature, which now far exceeds  the melting temperature of the blade material. Therefore, turbine blades must be protected by  cooling, typically by injecting cooler air through hundreds of pores on the blade (fully cooled  turbine) to create a protective film over the blade surface. In conventional turbine design, the  blade shape and the film cooling system are designed sequentially based on the assumption that  cooling flow passively attaches to the blade surface without interfering with the mainstream.  Recent studies, especially those in the transonic turbine, showed that the coolant could interact  with and change the mainstream, which challenges the fundamental assumption of the  sequential design process. Moreover, the sequential design process needs costly interactions  between the aerodynamics and heat transfer departments and is susceptible to being trapped in  the local optimization. Thus, the concurrent design, which simultaneously optimizes blade  shape and film cooling, is proposed to eliminate the interactions between the two departments  and achieve global optimization with better aerothermal performance (aerodynamic efficiency  and temperature distribution) in the fully cooled turbine blade.  Challenges in concurrent optimization of the fully cooled turbine include parameterization  (mathematical modelling), mesh generation, and high computational cost. First, based on the  PI’s previous work, a parameterization system and a mesh generator ad hoc for the fully cooled  turbine will be developed and released to the turbomachinery community as a free software.  Second, we will implement our novel algorithm, the multi-scale method, to provide high-fidelity  cooling simulations with affordable computational costs using source terms added to  the coarse mesh. These methods, looped by the optimizer, lead to concurrent optimization. If  our proposed concurrent optimization has substantially better aerothermal performance than the  sequential optimization would, a novel direction to improve the efficiency and extend the  lifespan of the gas turbine blade will be achieved. The flow physics analysis will be used to  quantitatively study the underlying mechanism for the extra gains obtained from the concurrent  optimization and provide design guidelines for fully cooled turbine blades. Moreover, the  source terms used in the multi-scale method will have several novel applications in cooling-related  simulations, for instance, the cooling injection patch using contoured source terms and  the machine learning–accelerated computational fluid dynamics simulations using source terms  as physics-informed training objectives.