Fire modelling using code_saturne7.0.5

[Guillaume THIRIET]

Dear all,

System information
My code_saturne version is 7.0.5.
My salome version is 9.9.0.

Description of my issue
I am trying to model a (very) simple gas fire using code_saturne and salome.

My aim is to make a kind of benchmark between FDS (which is a software I already used a lot) and code_saturne which is a new one. I know that by default, both softwares do not use the same turbulence model (LES for FDS / RANS for code_saturne) but I at least want to show that they can give similar results for a similar case.

Description of the model
The picture N1 describes the geometry of my model.

N1-Geometry-Fire.png

(7.62 KiB) Not downloaded yet

The picture N2 summarizes the boundaries of my model.

The mesh is composed of uniform squared cells of 10 cm length (similar to what can be done using FDS).

The main model assumptions are the following:
TURBULENCE
- Turbulence model: k-eps Linear Production
- Velocity scale: 1.0 m/s
GAS COMBUSTION
- Fuel: C3H8
- CO Yield = 0.005 / C Yield = 0.019 (SFPE Handbook)
- Thermochemistry file: fromGUI
GRAVITY
Gravity taken into account (-9.81 on Z axis)

VOLUME CONDITIONS
Oxydant reference temperature: 293.15 K
Fuel reference temperature: 293.15 K

INLET BOUNDARY CONDITIONS
Mass flow rate: 0.05 normal to the inlet

TIME SETTINGS
cf. picture N4

Description of the results
The picture N3 describes the evolution of the temperature on a slice in the middle of the model.

As you can see, the very beginning looks quite "normal". We can see the fire plume being created and the hot temperature remains close to the plume which is what we can expect in reality (convection neglectable in front of radiation at medium / far field).

But, very soon we see something like a reflux of hot gases within the domain which leads to an elevation of the temperature within the whole domain during a certain amount of time before things starts to cool down again.
In the end (after temperatures cool down), the plume also seems to me a bit "taller" - At least, I am not used to meet such kind of temperature that high under the fire source whatever it is using FDS or reading scientific papers on similar experiments - So my opinion is that there is an issue in my inputs because theses results are for me unphysical.

Should you have already run successfully a code_saturne combustion case, I would be glad to discussed with you the matter in order to understand where I can be wrong.
Yours faithfully.

Commentary
I also tried replacing the lateral boundaries(left, right, front and back borders) by boundaries of type Free inlet / outlet but it seems it just heavily slows down the whole calculation (for now the calculation is not over (as it is very slow) so I cannot say precisely that it won't change the results in the end ---> This conclusion should be taken with caution)

[Guillaume THIRIET]

Here, the N4 picture as I don't succeed to insert it into the previous text message...

[Guillaume THIRIET]

Good afternoon,

Description of my issue
Following my previous message, please find there an update regarding my fire simulation issues.
As a reminder, I am trying to model a (very) simple propane gas fire using code_saturne and Salome. The final aim is to make a small benchmark between FDS (CFD software dedicated to fire - developed by the NIST) and code_saturne.

Description of the model
IDEM

BOUNDARY CONDITIONS
In order to correct the issue met in my last message, I modified some of the boundary conditions from OUTLET to INLET/OUTLET.
The following picture describes the new boundary conditions applied on my model.

RESULTS
The picture below presents the temperature profile resulting from the code_saturne calculation.

The picture below presents the fuel concentration profile resulting from the code_saturne calculation.

According my experience modelling fires, the results of code_saturne does not seem physical.
The flame height is far higher than the one calculated by FDS (code_saturne --> > 5 m / FDS --> max 3.5 m).

The temperature profile is far higher than the one calculated by FDS (code_saturne --> max 2500 C / FDS --> max 1500 C).

Both these values are not physical but I do not have solution to solve this issue.

WHERE DOES THE ISSUE CAN CAME FROM ?

• I am wondering if a radiation model is implemented despite the specification of an absorption coefficient within the dpThermochemistry file as in the Thermal model, the option Thermal scalar is frozen by default while using Gas combustion.

• I applied a mesh of 10 cm as I did in FDS (following the FDS recommendation on that topic). In that respect, cells are quite big and I wonder if code_saturne is able to deal with combustion efficiently in such big cells

Thank you very much by advance for your help if you have any

Thanks for reading in any case.

Yours faithfully

[Yvan Fournier]

Hello,

I do not know much about fire modeling so my help will be limited here.

Regarding the thermal scalar, I think the choice of thermal scalar is frozen because fire modeling assumes the thermal scalar solved is the enthalpy. Choosing the Gas combustion option, if you choose the first option with enthalpy source terms, you will get to choose the radiative model (I do no know why this choice is not available for the enthalpy + mixture fraction option; I need to check whether this is a GUI bug or not).

Checking the sensitivity of results to mesh size could be interesting.

Checking the impact of the turbulence model may be important too (some of our test cases use classical, not linear production k-e, though I do not know why or if it is better or just an old setting).

Combustion cases usually set "idilat = 4".

In cs_user_parameters.c

Code: Select all

cs_velocity_pressure_model_t *vp_model
    = cs_get_glob_velocity_pressure_model();
vp_model->idilat = 4

Though I am not familiar with the matching theory. This is also available with the GUI, in the numerical parameters, as "algorithm to take into account the density variation in time",

I also know from experience with another combustion case (coal combustion) that the radiative model boundary conditions can have a large influence.

Most of the gas combustion cases I have seen have quite a bit of user code, and I do not know whether this is simply because they were setup before enough GUI features were available, or if the GUI is not sufficient...

Beyond that, we will need a combustion specialist to read this post...

Best regards,

Yvan

[Guillaume THIRIET]

Dear all,

Please find enclosed my case folder in order to allow you to manipulate my case. I removed the results as the latter where a bit to heavy to be upload.

@Yvan,
Regarding the question of the thermal scalar, I was also wondering about how the thermal losses radiation are taking into account. How to set them up.
I read on the Theory guide - "3 points chemistry" that radiation seems to be considered while running combustion but without too many details.
On the Practical users guide - "Radiative thermal transfers in semi-transparent gray media", there are also some information but it is said that

when a calculation is run using a specific physics module, the radiation is activated or not according to the parameter file related to the considered specific physics

--> So maybe it is required to pass by cs_user_parameters file.

I will run a fast case of 10 / 20 seconds with 0.05 m cell size mesh this week-end and provide additional information regarding this try latter.

Regarding the turbulence model, I have to admit that I simply chose the simplest model as I am not familiar with turbulence theory and I do not really know the difference between all models in details.
I can run some tests with changing that parameter too.

Though I am not familiar with the matching theory. This is also available with the GUI, in the numerical parameters, as "algorithm to take into account the density variation in time",

- Indeed - There is an Algorithm for fire I didn't know. This can heavily change the results so I will relaunch a simulation using that parameter.

Thank you for the information.
I will continue my tests and keep you informed.

Have a nice day.

[Guillaume THIRIET]

Please find enclosed my modelling files as it seems I get an issue uploading them