Turbomachines form the main components of thrust producing aircraft engines and power producing power plant systems. They are categorized as power absorbing (compressors) or power producing (turbines) machines depending on the energy conversion to or from a fluid stream. These turbomachines are composed of a set of aerodynamic blade profiles arranged along the periphery of the shaft.
The efficiency of these machines is dependent on the nature of flow passing over these aerodynamic blade profiles. Even a slight improvement in their efficiency contributes to a significant reduction in the specific fuel consumption and enhancement in the power production. One of the ways to improve the efficiency is to model the aerodynamic profiles with a smooth pressure distribution on their surfaces. To this end, manufacturers of turbomachinery have switched to sophisticated blade design algorithms based on inverse methods rather than using conventional aerodynamic shapes such as NACA and DCA profiles.
Design of blade profiles for a prescribed pressure distribution is an iterative process. Two iterative algorithms are proposed in this paper. Each iterative step requires a guess blade profile and consists of three modules, namely, flow solver, blade modification algorithm and smoothing procedure. In the case of turbine blade profiles, due to high blade turning and blade curvature, special attention is given to design the initial guess blade profiles using an eleven blade parameter approach. A commercial flow solver based on finite volume approach is used to evaluate the flow field around the guess blade profile. The guess blade profile and the results of the validated flow solver are given as inputs to the blade modification algorithm, based on virtual velocity approach, to modify the blade profile. These blade profiles are subjected to a smoothing procedure to obtain C2 continuous realistic blade profiles. The design formulation is validated for a single airfoil, compressor and turbine cascade blade profiles. The methodology not only converged to the target pressure distribution but did so in lesser number of inverse design iterations when compared to the existing techniques in the literature.