ON THE ACCURACY OF REACTIVE-TRANSPORT MODELS FOR LARGE EDDY SIMULATIONS OF HIGH SPEED FLOWS

Date

2026-8-14

Author

Sakaoğlu, Sergen

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

50
views

32
downloads

Cite This

High-fidelity simulation tools are essential for modeling multi-species reacting flows, where temperature and species concentrations vary sharply and small-scale transport governs accuracy. Chemical kinetics and molecular transport strongly influence ignition, flame propagation, extinction, and pollutant formation, and understanding these processes is increasingly important under stricter environmental regulations. However, accurate simulation of turbulent reacting flows requires detailed transport models, reliable kinetics, and efficient chemistry-reduction strategies to remain feasible. Tabulation methods, such as the Flamelet Generated Manifold approach, address this challenge by using precomputed reduced-order flame solutions to represent complex reacting flows. Nevertheless, multidimensional tabulation can require substantial memory, and conventional lookup and interpolation procedures may become computationally expensive in high-dimensional manifolds. In this study, a Flamelet Generated Manifold model is developed and implemented within the OpenFOAM framework to simulate three-dimensional turbulent reacting flows, with multidimensional lookup procedures incorporated into the solver for manifold-based evaluation of thermochemical quantities. The formulation is derived in both low-Mach and fully compressible forms; in the latter, pressure is retained as an additional control variable and a density-based solver captures acoustic wave propagation and its coupling with unsteady heat release. The model is further extended to multiphase combustion through a Lagrangian–Eulerian description, in which droplet dispersion, evaporation, and interphase heat and mass transfer are coupled to the gas-phase manifold, while an enthalpy-based control variable removes the adiabatic restriction and accounts for evaporative cooling and heat losses. Finally, tabulated retrieval is replaced by runtime neural network inference, reducing memory requirements and enabling nonlinear retrieval from the manifold.

Subject Keywords

Reactive Flow, Multiphase, Flamelet Generated Manifolds, Computational Fluid Dynamics, Neural Networks

URI

https://hdl.handle.net/11511/120643

Collections

Graduate School of Natural and Applied Sciences, Thesis