[home] [Personal Program] [Help]

---

Session: Session 10: Turbine design II
Session starts: Tuesday 13 October, 10:30
Presentation starts: 12:10
Room: 1A Europe

---

Pablo Rodriguez-Fernandez (Politecnico di Milano)
Giacomo Persico (Politecnico di Milano)

Abstract:
In this paper, an automated design tool for Organic Rankine Cycle (ORC) turbines is presented. Supersonic flows and real-gas effects featuring ORC turbines complicate significantly their aerodynamic design, which may benefit significantly from the application of systematic optimization methods. This study proposes a complete method to perform shape optimization of ORC turbine blades, constructed as a combination of a generalized geometrical parametrization technique, a high-fidelity Computational Fluid Dynamic (CFD) solver (including real gas and turbulence models) and an evolutionary algorithm. As a result, a non-intrusive tool, with no need for gradients definition, is developed. The high computational burden typical of evolutionary methods is here tackled by the use of a surrogate-based optimization strategy, for which a Gaussian model is applied. % Application to ORC turbines has been proved to be successful, resulting in a comprehensive method for a very wide range of applications. In particular, the present optimization scheme has been applied to the re-design of the supersonic nozzle of an axial-flow turbine. In this design exercise very strong shocks are generated in the rear blade suction side and shock-boundary layer interaction mechanisms occur. Optimization aiming at a more uniform flow at the blade outlet section is shown to minimize the shock losses, resulting in a significant improvement in the nozzle efficiency. The optimal configuration determined with the present design tool is also successfully validated against the outcome of a previous optimization performed with a gradient-based method, demonstrating the reliability and the potential of the design methodology here proposed.