Stochastic and Direct Combustion Noise Simulation of a Gas Turbine Model Combustor

Combustion noise of a gas turbine model combustor operating in partially premixed mode under atmospheric conditions is simulated with both, a hybrid stochastic approach and a direct, scale resolving method. Results from the hybrid ansatz are compared with experimental data and with results from incompressible and compressible CFD simulations. The hybrid time-domain method 3D FRPM-CN consists of a stochastic sound source reconstruction algorithm, the Fast Random Particle Method (FRPM) nd sound propagation by linearized Euler Equations. The method is herein evaluated for its capability of Combustion Noise (CN) prediction. Monopole sound sources are reconstructed by using an estimation of turbulence statistics from reacting, steady-state CFD-RANS.

As a direct approach, a Compressible Projection Method (CPM) is applied. It is an extension of conventional pressure-based methods for the treatment of compressible flows. This solution strategy is implemented as a fractional step scheme in the DLR Finite Volume based research code THETA. CFD results of CPM and RANS are furthermore compared to results from a conventional incompressible projection method (IPM). First, steady state and unsteady CFD simulations of flow field and combustion of the model combustor are compared to experimental data. Two equation modeling for turbulence and global chemistry treatment for combustion are employed. Turbulence in unsteady computations is depicted with a scale adaptive simulation (SAS). In a second step, the hybrid acoustics simulation setup for the model combustor is introduced. Selected results are presented and 3D FRPM-CN pressure spectra are compared to experimental data and results from CPM. Finally, computational turnaround times of hybrid and direct approach are evaluated and opposed.