CFD Analysis of Pressure Instabilities in Stator-Rotor Disc Cavity Systems

Sammanfattning: The continuous demand to improve turbine performance has led manufactures to focus on aspects that have been previously considered of secondary importance such as the secondary air system. The purpose of this system is to cool the components and prevent ingestion of hot gas into the stator-rotor cavity that could lead to low frequency pressure fluctuations called Cavity Flow Instabilities. These instabilities could cause unpredictable rotor vibrations and damage several components. A CFD method capable of detecting cavity flow instabilities in a rotor-stator disk cavity system is investigated, based on the 360◦model of the cavity without anystator vanes and rotor blades. Boundary conditions are simplified by considering steady and uniform flow in the main gas path. Different turbulence models are tested such as Realizable k−ε,k−ω SST, DDES, and SAS. In order to test the performance of the method, different purge flow levels are simulated. The most successful results, are predicted by the Realizable k−ε turbulence model. This model predicts two rotating low pressure structures in the cavity, for low purge flow levels. These pressure structures rotate at approximately 80% of the rotor speed. Furthermore, the spectra analysis of the pressure shows a reasonable agreement with the experimental results in terms of the frequency, showing a distinct region of low frequencies pressure instabilities. Nonetheless, this method overpredicts the amplitudes by a factor of 3-7 depending on the frequency. In addition, regions of one order of magnitude higher in frequency is also predicted. The DDES model shows similar findings but the amplitudes in the pressure spectra associated to the low frequencies are lower. Additionally, SAS also predicts the pressure in-stabilities but, in this case, the amplitudes are closer to RANS simulations, yet the high frequencies disappear. Unfortunately, k−ω SST, did not predict these pressure instabilities. Further research is still needed in many of the aspects of this work, from the simplifications up to the turbulence models. However, it is concluded from this work that this method could be a useful tool for turbine design as it decreases the need for testing and prototype manufacturing.