Large-Eddy Simulation of a Lifted High-Pressure Jet-Flame with Direct Chemistry

ABSTRACT

A large-eddy simulation is presented of a challenging high-pressure jet flame case that is representative of state of the art, dry low-NO_x and low-CO real gas turbine combustion. A reaction scheme is developed for lifted lean premixed high pressure methane jet flames, and tested by three-dimensional large-eddy simulation of an experiment, for which very detailed data are available. Auto-ignition-delay times of different mixtures of fresh gas and products have been introduced as a novel optimization criterion for the mechanism development. The new mechanism has been developed by a genetic algorithm-based reduction and optimization, and consists of 15 species and 18 reactions. The large-eddy simulations are performed using a finite rate chemistry (FRC) approach and the dynamic thickened flame (DTF) model to investigate a lifted jet flame at high pressure in a gas turbine model combustor. In the simulations, the novel mechanism is compared to a similar mechanism that was generated without this criterion, and the well-established Lu19 mechanism. With the new mechanism, the LES predicts the flame as accurately as with Lu19, at a significantly lower cost. Further post processing with Lagrangian tracer particles confirmed that ignition events occur in the region corresponding to the liftoff height estimated in the experiment, which is corroborated by a chemical explosive mode analysis (CEMA). Overall, the newly developed mechanism with the novel optimization criterion was found to provide a better agreement with the experiments than previous mechanisms of similar cost, or a comparable agreement to a mechanism of significantly higher cost.

KEYWORDS:

• Large-eddy simulations
• finite rate chemistry
• mechanism development
• lifted high pressure flame

Acknowledgments

The authors gratefully acknowledge the financial support through BMWi and CEC3 (funding reference number 03ET7073D) and computing time granted on magnitUDE (DFG grant INST 20876/209-1 FUGG). We would like to thank Sandeep Jella for helpful discussions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Bundesministerium für Wirtschaft und Energie [03ET7073D]; Deutsche Forschungsgemeinschaft [INST 20876/209-1 FUGG].