Numerical simulation of autoignition under gas turbine operating conditions

Abstract

This work aims to determine the performance of the numerical modelling of autoignition processes under realistic gas turbine operating conditions and to identify promising optimization potential for the simulation. The focus is on the combustion of hydrogen-containing fuel in the second combustion chamber of a sequential gas turbine. In this system, the reliable prediction of autoignition is of central importance to ensure a high level of operational safety and to keep pollutant emissions low. New and optimized combustion chamber systems can be designed using numerical simulation methods. In order to evaluate the numerical predictions and derive a target-oriented development strategy for the combustion chamber design, it is necessary to assess the quality of the simulation results. The quality can be determined by means of validation studies in which simulation results are compared with suitable experimental reference data. Over the last 15 years, many studies on the validation of autoignition simulations have been published. The majority of these studies were performed on a generic free jet flow at atmospheric pressure. Until now, however, no detailed validations have been available for autoignition simulations under gas turbine typical conditions such as increased pressure and complex flow geometry. In order to close this gap in research, the reliability of modern simulation methods under realistic gas turbine conditions is analyzed in this work. In addition, to what extent the results, which were obtained under simplified laboratory conditions (atmospheric pressure, low turbulence and simple flow geometry), can be reliably transferred to real operating conditions is investigated. The validation study focuses on the fuel premix section of staged gas turbines at a pressure of 15 bar. The fuel is injected transversely into the hot oxidizer at temperatures of more than 1000 K and Reynolds numbers of up to 1 million. Under these conditions the demands on simulation and measurement technology are very high. Furthermore, the close coupling of different submodels for turbulence, chemistry and turbulence-chemistry interaction makes it difficult to identify specific model deficits. For this reason, the methodology of the validation hierarchy according to Oberkampf is applied in this work. In addition to the application-oriented high-pressure experiment, selected subsystems were defined and validated in order to obtain information on specific submodels. The comparison of the numerical simulation with the high-pressure experiments shows that the combustion phenomena, such as the formation of the ignition kernels, flame propagation and flame stabilization, can be reproduced very well. In the quantitative comparison, however, significant differences occur, which are mainly due to the high pressure and the complex flow geometry. It has been shown that the accuracy and modeling quality achieved under laboratory conditions cannot be directly transferred to real operating conditions. Although the ignition of hydrogen at low pressure can be determined very well, the uncertainties at gas-turbine-relevant pressure are relatively high. A further challenge results from the application-oriented flow configuration. In contrast to the free jet flow, a backflow occurs in the jet-in-crossflow configuration investigated here. Individual ignition kernels that form near this recirculation zone can stabilize there and anchor in the form of a steady flame. In order to model this safety-relevant phenomenon correctly, the entire spatial distribution of potential ignition kernels must be accurately reproduced. In this work two different causes for the variation of the ignition location were identified: On the one hand, temperature fluctuations in the hot oxidizer have a direct influence on the ignition delay time. Since autoignition is very temperature-sensitive, even small temperature fluctuations cause a broad spread of the ignition locations. On the other hand, turbulence also has a significant influence on the ignition variation, which is mainly caused by large-scale turbulent structures. For the first time, numerical simulations of autoignition processes under application-oriented gas turbine operating conditions were validated with suitable experimental reference data. It has been shown that the results from simplified autoignition experiments at atmospheric pressure cannot be directly transferred to real engine operating conditions, as the requirements for modeling differ significantly. Using a system-specific validation hierarchy, it was also possible to systematically identify the cause of the modeling uncertainties and to investigate them in detail. Based on the knowledge gained, a guideline for the simulation of autoignition processes for the design of new combustion chamber concepts is provided and the most promising optimization potential for the numerical prediction are deduced.