A variety of combustion models for Large Eddy Simulation (LES) of premixed turbulent flames have been developed and validated over the years. Validation studies concentrate on relevant mean quantities and turbulent fluctuations, however the prediction of flame dynamics is typically not taken into account. Furthermore, it is difficult to meaningfully compare the computational efficiency of model formulations due to different compute resources, meshes, code bases and numerics. The present study compares turbulent combustion models on the same code base, keeping boundary conditions, meshes and numerical settings constant. The reliability and versatility of two turbulent combustion models, i.e. the artificially thickened flame and flame surface density formulations, is assessed by applying them to a variety of operating conditions and burner configurations. In particular, for a premixed methane burner we consider three power ratings by changing the inflow velocity, which increases the demand on the sub-grid scale model due to increased sub-grid scale wrinkling. A change in swirler position modifies the interference of swirl and acoustic perturbations, with a significant impact on flame dynamics. Changes in thermal boundary condition and combustion chamber size provide insights into the consequences of quenching effects resulting from heat losses on flame anchoring and flame topology.

The present work compares the respective advantages and disadvantages of compressible and incompressible computational fluid dynamics (CFD) formulations when used for the estimation of the acoustic flame response. The flame transfer function of a turbulent premixed swirl-stabilized burner is determined by applying system identification (SI) to time series data extracted from large eddy simulation (LES). By analyzing the quality of the results, the present study shows that incompressible simulations exhibit several advantages over their compressible counterpart with equal prediction of the flame dynamics. On the one hand, the forcing signals can be designed in such a way that desired statistical properties can be enhanced, while maintaining optimal values in the amplitude. On the other hand, computational costs are reduced and the implementation is fundamentally simpler due to the absence of acoustic wave propagation and corresponding resonances in the flame response or even self-excited acoustic oscillations. Such an increase in efficiency makes the incompressible CFD/SI modeling approach very appealing for the study of a wide variety of systems that rely on premixed combustion. In conclusion, the present study reveals that both methodologies predict the same flame dynamics, which confirms that incompressible simulation can be used for thermoacoustic analyses of acoustically compact velocity-sensitive flames.

The present study combines Large Eddy Simulation (LES) with System Identification (SI) to determine the Flame Transfer Functions (FTF) of technically premixed flames, which respond to fluctuations of upstream velocity as well as equivalence ratio. Two variants to obtain the corresponding FTFs from numerically determined time series data are reported and compared with experimental results. The experiment does not measure heat release rate directly, but instead CH* chemiluminescence. This is insufficient for FTF identification of technically premixed flames, but can be used for validation of the simulation. We implemented a CH* post-processor in the simulation and validate with experiment. After validation the simulation is used to identify the contributions of velocity and equivalence ratio to the FTF of technically premixed flame dynamics. We propose and compare two approaches for the identification of FTFs. The direct approach via Multiple Input Single Output system identification requires one simulation with simultaneous excitation of fuel and air inlets and carefully chosen input signals. The second approach reconstructs the FTF decomposition from two separate simulations, one perfectly premixed and one technically premixed, with reduced requirements on signal quality. We compare both approaches and discuss the FTFs of perfectly and technically premixed flames. Overall the LES/SI approach proved to be flexible and reliable for technically premixed flames.

No abstract

Transient cavitation induced loads are investigated in a flow region subject to severe cavitation erosion at an ultrasonic horn test facility operated in indirect mode, both experimentally and via CFD. A variation of gap width and sensor position is used to vary flow aggressiveness. The experimental technique utilizes piezoelectric sensors based on Polyvinylidene fluordide (PVDF). Sensors are employed at the stationary specimen in close proximity to the oscillating horn tip and wall load spectra are obtained. Complementary CFD studies emulate the experimental sensor. Flow aggressiveness decreases when the gap is widened and increases towards the stationary specimen center. Results from experiment and simulation show good quantitative agreement.