Learning Forum

[krany3]

The temperature contour from premixed combustion using finite-rate/eddy dissipation model looks like this. I don't understand why the flame is propagating far downstream. Can anyone please explain this?

[klu]

Hi,

Do you have any experimental data or theoretic estimation for the flame propagation based on the model conditions? Would you please provide more details why you think the flame is propagating too far? Thanks.

[krany3]

The above temperature was obtained using LES under the same conditions that I'm using for modeling. The flame has stabilized right after the nozzle exit in LES. But, in my case using RANS, I don't see that. I'm assuming because of slow combustion, the flame is propagating far downstream.

[klu]

Hi, 

Although the conditions are the same, how about the mesh density and quality, convergence, etc? Would you think that it is an apple-to-apple comparison? Please also compare the results for example the turbulence variables, reaction rate, etc. so that we may better analyze the difference. In addition, would you run a URANS model in case the flame has high unsteadiness? 

[DrAmine]

Valuable suggestions. 

[krany3]

Thank you. The meshes are different. I don't have anything to compare with, except the temperature contour. Maybe, I'll try the uRANS and will see how it's varying.

[krany3]

I tried with uRANS. It didn't give me the swirling flame. What could be the reason?

Thanks

[DrAmine]

For swirling flames and generally swirling flows two equation models or any boussinesq approach eddy viscosity based model will behave wrong. Flow anistropy and secondary flows are well capture by second momemts of closure RANS  models as well  as  scale resolving methods. Please post your model settings deployed boundaries and material properties. 

[DrAmine]

Also moving the thread to Fluid dynamics category. 

 

[krany3]

Model Settings:

-For turbulence modeling

realizable k-epsilon model with standard wall functions and default model constants

-For combustion

Species transport with volumetric reactions. 2 step chemical mechanism of CH4-Air was used. 

Turbulence-chemistry interaction was modeled using finite-rate/eddy dissipation model

Material Properties:

CH4-air mixture properties:

Density -  calculated using incompressible ideal gas equation

Specific heat - using mixing law

Thermal conductivity -0.0454 w/m-K

Viscosity - 1.76 e-05 kg/m-s

Diffusivity - 2.88e-05 m2/s

Boundary Conditions:

INLET 

mass flow rate - 2.0579 g/s

turbulent intensity - 5%

turbulent viscosity ratio - 10

temperature - 300 K

species mass fractions - CH4-0.0376, O2 - 0.2239, rest is N2

OUTLET

turbulent intensity - 5%

turbulent viscosity ratio - 10

temperature - 300 K

species mass fractions - O2-0.0739, CO2-0.1032, H2O - 0.0844

Chamber walls were set at a temperature of 600 K

For the pressure-velocity coupling SIMPLE algorithm was used and gradient method used was least squares cell based. All the discretizations were modeled using 2nd order upwind scheme

Default under-relaxation factors were used except for species and energy, which are 0.8 

[DrAmine]

And do you know what has been used for LES run? I guess a high-end combustion model like EDC with detailed mechanisms.  As LES is always transient I assume the picture you posted has been time-averaged over a certain time intervall. If you want to make a comparison you need to use the same methods or at least use the same combustion model and boundary conditions.

[krany3]

Yes Amine. The posted picture was the time-averaged temperature. For LES, they used thickened flame model for modeling turbulence-chemistry interaction. I'm confused on which model to be used. Is it EDC or the flamelet generated manifold (FGM)? 

Thanks

[krany3]

Hello Amine, which model in Fluent is considered as the thickened flame model?