Confidence in Flame Impulse Response Estimation From Large Eddy Simulation With Uncertain Thermal Boundary Conditions

Abstract

Thermoacoustic stability analysis is an essential part of the engine development process. Typically, thermoacoustic stability is determined by hybrid approaches. These approaches require information on the flame dynamic response. The combined approach of advanced System identification (SI) and Large Eddy Simulation (LES) is an efficient strategy to compute the flame dynamic response to flow perturbation in terms of the Finite Impulse Response (FIR). The identified FIR is uncertain due in part to the aleatoric uncertainties caused by applying SI on systems with combustion noise and partly due to epistemic uncertainties caused by lack of knowledge of operating or boundary conditions. Carrying out traditional uncertainty quantification techniques, such as Monte Carlo, in the framework of LES/SI would be computationally prohibitive. As a result, the present paper proposes a methodology to build a surrogate model in the presence of both aleatoric and epistemic uncertainties. Specifically, we propose a univariate Gaussian Process (GP) surrogate model, where the final trained GP takes into account the uncertainty of SI and the uncertainty in the combustor back plate temperature, which is known to have considerable impact on the flame dynamics. The GP model is trained on the FIRs obtained from the LES/SI of turbulent premixed swirled combustor at different combustor back plate temperatures. Due to the change in the combustor back plate temperature the flame topology changes, which in turn influences the FIR. The trained GP model is successful in interpolating the FIR with confidence intervals covering the "true" FIR from LES/SI.

Incompressible versus compressible large eddy simulation for the identification of premixed flame dynamics

“…The combustor walls and the bluff body plane are set to no-slip isothermal walls with a fixed temperature of 600 K. The combustor wall temperature is an estimated value based on measurements of a similar burner configuration as mentioned in Tay-Wo-Chong et al 48 , Previous studies [61][62][63] show that wall heat transfer in the combustion chamber is very important for correct modeling of the flame shape (and therefore the flame dynamics) in swirl-stabilized combustors. Uncertainty in wall temperature in the experiment might cause deviations in the flame response evaluation with LES 64 . The burner section including the swirler is treated as an adiabatic no-slip wall, which is sufficient as the reaction only takes place inside the combustion chamber.…”