Three-dimensional unsteady flow in the oscillating turbine blade row

Abstract

This thesis documents an experimental and computational study of the unsteady flow around oscillating blades in low-pressure turbines, with emphasis on the three- dimensional flow behaviour, intra-row interaction effects, tip clearance flow and part- span shroud influence. The research vehicles were a linear low speed oscillating turbine test cascade and a realistic low-pressure steam turbine rotor/stage. Systematic experimental measurements were conducted on the linear turbine cascade, which consists of seven, large scale, prismatic blades with the middle blade being driven to oscillate in a three-dimensional bending/flapping mode. Blades were instrumented with pressure tappings at six span-wise sections between 10% and 95% span to facilitate detailed three-dimensional steady and unsteady pressure measurements on the blade surface. Steady flow pressure was measured by using an inclined manometer bank, whilst the unsteady pressure measurements were obtained through off-board pressure transducers. The measured unsteady pressure was superposed to construct tuned cascade flutter data using a technique named the influence coefficient method. This study produced the first known set of 3D flutter data for tuned turbine cascade. On the computational side, a state-of-the-art, single-passage, three-dimensional, time- marching, Navier-Stokes flow solver has been adopted. The computational solutions of the linear cascade flow exhibits a consistently high level of agreement with the experimental data, which corroborates the experimental findings on the one hand and acts to validate the present flow solver on the other hand. The results, from several aspects, suggest a strong three-dimensional nature of the unsteady aerodynamic response to the blade first-bending/flapping and clearly demonstrate the inadequacies of the currently widely used two-dimensional and quasi-three-dimensional methods. turbine configurations. It was found that accurate flutter predictions require three- dimensional, multi-row flow solvers including tip clearance modelling.