Non-Deterministic CFD Simulation of a Transonic Compressor Rotor

Abstract

The compressor is one of the most sensitive components in a gas turbine. Small variations in geometry or operating conditions can have a detrimental effect on component performance, efficiency and life. During the past few years, significant effort has been invested in modelling the propagation of input uncertainties for CFD simulations using stochastic methods. Due to the large number of variables involved in typical industrial applications, problems often involve intensive computations, making the use of stochastic methods impractical. Therefore in addition to accuracy issues, the desire to reduce the computational overhead is also a key consideration in industrial applications. This work investigates the propagation of uncertainty within a transonic compressor rotor (NASA Rotor-37), using a Non Intrusive Polynomial Chaos methodology. Extensive computational research of this geometry has previously been undertaken and provides comparative data sets. The Non Intrusive Polynomial Chaos methodology is an inexpensive approach based on the spectral representation of the uncertainty parameters. The polynomial coefficients are evaluated using the Probabilistic Collocation method providing an exponential convergence for arbitrary probability distributions. Results will be shown for variations in inlet total pressure. The resulting performance parameters, including total pressure ratio and adiabatic efficiency, are presented along with their uncertainties. The paper serves as a description of the application of the Polynomial Chaos methodology, within a general purpose CFD software, to a gas turbine aerodynamics problem.