59395 - Flamelet Versus Detailed Chemistry Les for a Liquid Fueled Gas-Turbine Combustor: A Comparison of Accuracy and Computational Cost 

Contemporary industrial computational fluid dynamics (CFD) simulations of gas-turbine combustors typically employ flamelet combustion models, in particular the Flamelet Generated Manifold (FGM) model or the Steady Laminar Flamelet (SLF) model. These flamelet models parameterize and tabulate the thermo-chemistry by a small set of variables such as mixture fraction, enthalpy, and progress variable or scalar dissipation. In contrast, the use of detailed chemistry combustion models is not common due to the large number of species and reactions in real fuel-based chemical mechanisms. The computational cost is prohibitive because several hundreds of species need to be transported and stiff chemical reactions need to be integrated in each computational cell. Recent advances in chemical kinetics modeling, such as hybrid chemistry (HyChem) [1], provide a less expensive and viable approach to modeling high-temperature combustion chemistry of multicomponent real fuels. In the HyChem approach, the large molecule fuel pyrolysis is modeled by semi-global lumped reaction steps, while the oxidation process is described by detailed chemistry of simpler hydrocarbon fuels. The approach has demonstrated very good accuracy (e.g. ignition delays, flame speeds, etc.) for real fuels such as Jet-A, with only about 50 species. Using a relatively small mechanism such as this, as well as CFD chemistry acceleration techniques such as optimized ODE solvers, and cell clustering, the computational run-time of detailed chemistry is only about 3 times that of flamelet models.

In the current paper, Large Eddy Simulations (LES) of a Honeywell liquid-fuelled gas turbine test combustor, at Idle conditions are simulated with FGM and detailed chemistry combustion models in Simcenter STAR-CCM+ [2]. The Pressure Implicit by Splitting of Operators (PISO) scheme is used, along with the sub-grid Wale model with LES [3]. To fully capture potential emissions, a soot moment model, and Zeldovich NOx model are employed along with radiation. A detailed comparison with experimental data including NOx, CO, Unburned Hydrocarbons, Soot, and the outlet radial temperature profile are presented. A comparison of results with and without chemistry acceleration techniques for detailed chemistry is included. Then, computational costs are assessed by comparing the performance and scalability of the simulations with each of the combustion models. It is found that the detailed chemistry case with clustering can reproduce nearly identical results to detailed chemistry without any acceleration, if CO is added as a clustering variable. With the Lagrangian model settings chosen for this study, the detailed chemistry results compared more favourably with the experimental data than FGM, however there is uncertainty in the secondary breakup parameters. By varying the breakup model settings, the FGM results could be brought much closer to the experimental data. The cost for running detailed chemistry with clustering was found to be only about three-times that of FGM on the same core count. 

References:

1)      R. Xu, H. Wang et al., A physics-based approach to modeling real-fuel combustion chemistry - II. Reaction kinetic models of jet and rocket fuels, Combustion and Flame 193 (2018) 520-537. https://web.stanford.edu/group/haiwanglab/HyChem/

2)      Simcenter STAR-CCM+ v2020.3, Siemens Digital Industries.

3)      Nicoud, F., and Ducros, F., 1999. “Subgrid-scale stress modelling based on the square of the velocity gradient tensor”. Flow, Turbulence and Combustion, 62, pp. 183–200.