Flamelet Model with Epsilon Tracking in a Turbine Stator

Combustion within a two-dimensional turbine stator passage is numerically investigated in the context of the turbine-burner concept using a Reynolds-Averaged Navier-Stokes framework coupled with a novel flamelet model. The formulation links resolved-scale turbulence quantities with subgrid flamelet dynamics through the local turbulent kinetic energy dissipation rate, $ε$, which determines the flamelet inflow strain rate. For the first time, combustion of JP-5 is considered in a turbine stator passage as a practical fuel. This is achieved by solving transport equations for 14 major species on the resolved scale, while chemical source terms are obtained from precomputed flamelet libraries based on the HyChem A3 mechanism comprising 119 species and 841 elementary reactions. Model performance is assessed against methane combustion using both a one-step kinetics model and an $ε$-based flamelet formulation employing a 13-species skeletal mechanism. The $ε$-based formulation predicts lower peak flame temperatures due to dissociation effects and approximately 50% lower net chemical energy addition per unit mass compared with the one-step model, as a result of flame stand-off and downstream strain-rate-induced quenching. For JP-5, the simulations capture combined endothermic pyrolysis and exothermic oxidation processes, leading to vertically displaced reaction zones, increased near-wall temperatures, and larger resolved-scale reaction regions due to the higher flamelet flammability limit relative to methane.

💡 Research Summary

**
This paper presents a novel flamelet model for simulating combustion in a two‑dimensional turbine stator passage, a configuration relevant to the turbine‑burner concept. The governing framework is the Reynolds‑averaged Navier‑Stokes (RANS) equations coupled with a turbulence model (k‑ω SST). The key innovation lies in linking the resolved‑scale turbulence quantities to sub‑grid flamelet dynamics through the local turbulent kinetic energy dissipation rate, ε. By defining the flamelet inflow strain rate χ as a function of ε, the model directly couples the macroscopic turbulence field (k, ε) with the microscopic flamelet behavior, enabling a physically consistent representation of turbulence‑chemistry interaction.

The combustion chemistry is handled by solving transport equations for 14 major species (fuel, oxidizer, key intermediates, and products) on the resolved scale. Chemical source terms are not computed directly; instead, they are interpolated from pre‑computed flamelet libraries generated using the HyChem A3 mechanism, which comprises 119 species and 841 elementary reactions. This approach retains detailed chemistry while keeping the computational cost manageable.

Model performance is evaluated against methane combustion using two reference models: (1) a conventional one‑step global kinetics model and (2) an ε‑based flamelet formulation employing a 13‑species skeletal mechanism. The ε‑based formulation predicts significantly lower peak flame temperatures (approximately 200 K lower) and about a 50 % reduction in net chemical energy release per unit mass of fuel compared with the one‑step model. The reduction is attributed to flame stand‑off and downstream strain‑rate‑induced quenching, phenomena that the ε‑based model captures through the dependence of χ on ε. In contrast, the one‑step model over‑predicts heat release because it neglects the damping effect of high strain rates on reaction rates.

For the first time, the model is applied to JP‑5, a practical aviation fuel, within a turbine stator passage. JP‑5 combustion exhibits combined endothermic pyrolysis and exothermic oxidation. The simulations reveal vertically displaced reaction zones relative to methane, increased near‑wall temperatures, and a larger resolved‑scale reaction region. These effects stem from JP‑5’s higher flamelet flammability limit, which allows the flamelet to survive in higher strain‑rate environments and to extend further downstream. Consequently, the flame is less prone to early extinction, leading to broader heat release zones and higher wall heat fluxes.

The turbulence model supplies dynamic turbulent viscosity (μ_T) and turbulent Prandtl numbers, which are adjusted based on the local strain rate χ, ensuring that the flamelet inflow conditions reflect the instantaneous turbulent state. The flamelet library is parameterized by a non‑dimensional strain rate S* and pressure p; χ is approximated analytically as χ = 2 S* π exp(2 erfc⁻¹(2 Z)²), allowing rapid evaluation during the CFD run.

Overall, the ε‑based flamelet model offers a more accurate representation of turbulence‑chemistry coupling than traditional global kinetics, especially under high‑strain, high‑temperature turbine conditions. It successfully extends detailed chemistry to complex fuels like JP‑5 without prohibitive computational expense. The findings suggest that incorporating ε‑driven strain rate into flamelet models can improve predictions of flame location, temperature fields, and heat release, which are critical for the design and optimization of turbine‑burner systems. Future work should address three‑dimensional effects, multi‑fuel mixtures, and experimental validation to further establish the robustness and applicability of the approach.