A comparative study of transient blade row and blade count scaling approaches for numerical forced response analysis in a transonic turbine

Abstract

Traditional forced response calculations for adjacent blade rows without a common divisor demand full annular transient simulations. This typically results in large and often impractical computational costs during development. The present article discusses an approach to reduce this cost using a signal-patching routine on a transonic axial turbine. Additionally, the potential for further reduction of computational cost by means of time domain flow transformation methods is investigated. The methods assessed are profile transformation, time inclination and Fourier methods, all within ANSYS CFX. Comparisons are made for the vector components of blade force harmonics, generalised force and computational costs. The computational cost can be reduced to 40% and 5% of the full annular simulation with signal patching or time domain flow transformation methods respectively. The signal-patching and transformation methods applied result in an error of 10% and 15% of generalised force respectively.