Premixed Flame Model

Category: Fluid Analysis (CFD) | Integrated 2026-04-06
CAE visualization for premixed flame theory - technical simulation diagram
Premixed Flame Models

Premixed Flame: Theoretical Foundations

Overview

๐Ÿง‘โ€๐ŸŽ“

Teacher, how is a premixed flame different from a diffusion flame?


๐ŸŽ“

A premixed flame is a form where the flame propagates in a state where fuel and oxidizer are sufficiently mixed before combustion. Examples include gasoline engines, lean-burn combustors in gas turbines, and household gas stoves. The flame front has a distinct boundary, separating unburned mixture and burned gas.


๐Ÿง‘โ€๐ŸŽ“

The flame propagation speed is an important parameter, right?


๐ŸŽ“

Correct. The laminar burning velocity $S_L$ is the fundamental parameter for premixed flames. For methane/air (equivalence ratio 1.0, room temperature and pressure) $S_L \approx 0.36$ m/s, and for hydrogen/air $S_L \approx 2.1$ m/s.


Progress Variable $c$ and Governing Equations

๐Ÿง‘โ€๐ŸŽ“

What variables are used in CFD for premixed flames?


๐ŸŽ“

For premixed flames, the progress variable $c$ is used to track the flame front. $c=0$ represents unburned mixture, $c=1$ represents burned gas.


$$ \frac{\partial(\rho c)}{\partial t} + \nabla\cdot(\rho\mathbf{u}c) = \nabla\cdot(\rho D\nabla c) + \dot{\omega}_c $$

Here, $\dot{\omega}_c$ is the reaction source term, which becomes non-zero only near the flame front.


๐Ÿง‘โ€๐ŸŽ“

How is the source term $\dot{\omega}_c$ modeled?


๐ŸŽ“

This is the core of premixed flame modeling. There are three main approaches.


Major Premixed Combustion Models

ModelPrincipleAdvantagesDisadvantages
G-equation (Level Set)Tracks flame front as iso-surface $G=0$Geometrically clearNo internal flame structure
TFC (Turbulent Flame Closure)Zimont model. $S_T = A(u'/S_L)^n S_L$Easy to implementDepends on empirical $S_T$ correlation
FSD (Flame Surface Density)Transport equation for flame surface density $\Sigma$Physics-basedModel constants for $\Sigma$ equation
c-equation + reaction rate$\dot{\omega}_c = \rho_u S_L\nabla c$DirectNumerical issues with flame thickness

Turbulent Burning Velocity

๐Ÿง‘โ€๐ŸŽ“

What happens to the flame speed in turbulence?


๐ŸŽ“

The turbulent burning velocity $S_T$ increases with turbulence intensity $u'$. Zimont's correlation is widely used.


$$ S_T = A\,(u')^{3/4}\,S_L^{1/2}\,\alpha^{-1/4}\,l_t^{1/4} $$

Here, $\alpha$ is the thermal diffusivity, $l_t$ is the turbulent integral scale, and $A$ is a model constant ($A \approx 0.52$).


๐Ÿง‘โ€๐ŸŽ“

So, the stronger the turbulence, the more wrinkled the flame front becomes, increasing the apparent burning speed, right?


๐ŸŽ“

Exactly. Turbulence wrinkles the flame front, increasing its area, which enhances the burning rate per unit cross-sectional area. This is the classical picture by Damkohler (1940), and modern CFD models are also based on this concept.


๐Ÿง‘โ€๐ŸŽ“

So, unlike diffusion flames, the core of premixed flames is "tracking the flame front."


๐ŸŽ“

Yes. The fundamental difference is describing the flame with a progress variable rather than a mixture fraction.


Coffee Break Casual Talk

"Flame Thickness" is Less Than 1mmโ€”How Difficult It Is to Measure Laminar Burning Velocity

The "laminar burning velocity $S_L$," which is fundamental to premixed flame theory, is actually a very difficult quantity to measure. There are multiple experimental methods like the advanced stagnation flame method, counterflow method, and spherical propagation method, but even for the same gas, values can vary by 10-20% depending on the method. The reason lies in differing opinions among researchers on "how to correct for strain effects." Furthermore, the flame thickness itself is only about 0.1-1mm, and inserting a thermometer to measure it disturbs the flame. This problem of "the measurement itself disturbing the subject" is different from the uncertainty principle in quantum mechanics but shares a similar structure in terms of measurement difficulty. The validation of reaction constants in GRI-Mech 3.0 uses such carefully measured data.

Computational Methods for Premixed Flame

Details of Numerical Methods

๐Ÿง‘โ€๐ŸŽ“

Please tell me about the numerical challenges when solving premixed flames with CFD.


๐ŸŽ“

The biggest challenge is resolving the flame thickness. The thickness of a laminar premixed flame is $\delta_L \approx \alpha/S_L$, about 0.5 mm for methane/air and about 0.2 mm for hydrogen/air. Directly resolving this with RANS meshes on the order of millimeters is impossible.


Thickened Flame Model (TFM)

๐Ÿง‘โ€๐ŸŽ“

How is that solved?


๐ŸŽ“

In LES, the widely used method is the Thickened Flame Model (Colin et al., 2000). It artificially thickens the flame to make it resolvable by the mesh.


๐ŸŽ“

Increase the diffusion coefficient by a factor of $F$, and decrease the reaction rate by a factor of $1/F$.


$$ D_{\text{eff}} = F \cdot D, \quad \dot{\omega}_{\text{eff}} = \frac{\dot{\omega}}{F} $$

This increases the flame thickness to $F\delta_L$, but $S_L$ remains unchanged. $F = 5-20$ is typical.


๐Ÿง‘โ€๐ŸŽ“

But doesn't thickening the flame change its interaction with turbulence?


๐ŸŽ“

Sharp observation. A thickened flame cannot resolve small-scale turbulent wrinkling. Therefore, an efficiency function $E$ is introduced for correction.


$$ \dot{\omega}_{\text{eff}} = \frac{E}{F}\dot{\omega} $$

The Charlette efficiency function is representative, given in the form $E = E(\Delta/\delta_L, u'/S_L)$.


Implementation in Fluent

๐Ÿง‘โ€๐ŸŽ“

How do you set up a premixed flame in Fluent?


๐ŸŽ“

The following models are available in Fluent.

1. Premixed Combustion (Zimont TFC model): c-equation based. For RANS.

2. Partially Premixed Combustion: Hybrid of premixed + non-premixed.

3. FGM (Flamelet Generated Manifold): Progress Variable + mixture fraction.


๐ŸŽ“

Zimont TFC model settings:

  • Models > Species > Premixed Combustion
  • Turbulent Flame Speed model: Zimont
  • Laminar Flame Speed: Input value or calculated value (equivalence ratio dependent)
  • Flame Stretch Factor: Default 0.26

Implementation in OpenFOAM

๐Ÿง‘โ€๐ŸŽ“

What about in OpenFOAM?


๐ŸŽ“

XiFoam is the solver for premixed combustion. It solves a transport equation for the flame wrinkling factor $\Xi$ (= $S_T/S_L$).


SolverTargetModel
XiFoamPremixed compressible$\Xi$-equation
reactingFoam + PaSRPremixed/Partially premixedSpecies Transport
fireFoamFireEDM/Diffusion flame
๐Ÿง‘โ€๐ŸŽ“

Is the Thickened Flame Model included in the standard OpenFOAM distribution?


๐ŸŽ“

It is not in the standard distribution, but community versions (like TFM4OpenFOAM) are available. It is widely used in gas turbine LES research.


๐Ÿง‘โ€๐ŸŽ“

So the core of the numerical method for premixed flames is the bold idea of "thickening the flame."


๐ŸŽ“

Yes. TFM is a physically sophisticated trick and has become the de facto standard for LES premixed combustion.


Coffee Break Casual Talk

The True Nature of the "Progress Variable c"โ€”A Story About the Most Troublesome Variable in Premixed Flame Models

In the numerical implementation of premixed flame models, many people stumble over the definition of the reaction progress variable c. c is a variable representing 0 (unburned) to 1 (burned), but "which chemical species' mass fraction is used to define c" varies by model and researcher. There are schools that define it with CO2, schools that use normalized temperature, schools that define it as a linear combination of multiple componentsโ€”each yielding slightly different results. Fluent's default uses a combination of products, but depending on the fuel or equivalence ratio, other definitions might yield better accuracy. Stories like "the results suddenly matched when I changed the definition of c" are often told at CAE conferences.

Related Simulators

Experience the theory firsthand with the interactive simulator for this field

All Simulators

Related fields

Thermal AnalysisV&V ยท Quality AssuranceStructural Analysis
Rate this article
Thank you for your feedback!
Helpful
More details
Report error
Helpful
0
More details
0
Report error
0
Written by NovaSolver Contributors
Anonymous Engineers & AI โ€” Sitemap
About the Authors