Prediction of Thermo-Acoustic Properties in Combustion Chambers

Doctoral thesis, 2013

Design and analysis methods for gas turbine combustors or afterburners need to be improved in particular regarding the prediction of combustion instabilities. A methodology, which combines the use of a linearized flow solver and the Arnoldi algorithm for eigenmode extraction, resolves eigenmodes which are candidates to these instabilities. The Arnoldi eigenmode extraction procedure is built on three different linearized solvers based on:1) the Linearized Euler Equations (LEE), 2) the Linearized Navier-Stokes Equations (LNSE) and 3) the Linearized Unsteady RANS equations (LURANS). Another eigenmode extraction technique, which is based on solver input data only, is also used for predicting the combustion instabilities: The Dynamic Mode Decomposition. The eigenmode extraction techniques are compared in terms of accuracy, CPU time requirements and robustness. The Validation Rig I combustor test rig is the experimental rig used for validating numerical results. Under certain conditions the rig is the stage of two well known combustion instabilities: the ''Buzz'' mode which has a frequency of about 120 Hz and the ''Screech'' mode which has a frequency of about 1200 Hz. Further studies are also carried out regarding the combustion models and the mode stability analysis which models the coupling between the aero-acoustic waves and the heat release fluctuations, the core mechanism behind the combustion instabilities. A novel dynamic porous wall model is investigated for the Validation Rig I to stabilize the Screech mode. The Validation of dynamic porous wall model is investigated via the 3D computations of a system composed by a orifice and a back sheet. Experiments performed on the Siemens Acoustic Rig, depict the ability of dynamic porous wall model to mimic the behavior of a perforated plate in presence of Grazing-Bias flows.