[krapic]

Hi everyone,

I'm simulating a bluff body stabilized flame (test HM1E in Sydney) using the rhoFlameletFoam solver made by Hagen Mueller.

I initialize the simulation with the solution made with rhoSimplecFoam (just air), and I set a field (setField utility) of fuel (methane hydrogen mixture 1:1) in the fuel tube. As you can see from the velocity plot at 0.2s, the flame doesn't stabilize. There were some issues with pressure waves in the fuel tube so the inlet, outlet, coflow and side are set to a waveTransmissive boundary - that solved the pulsating flow at the inlet tube but didn't affect the flame near the coflow.



I experimented with a wide variety of configurations including various mesh densities and BC values but nothing seems to help. Here is the link to the case folder on google drive (due to the size limit on cfd-online). Note - the cantera tables aren't included there. I also included a few videos showing the progress of the simulation.

Does anyone have an idea of what might be causing the instability and have a possible solution to the problem?

Thank you in advanced,
Kristijan

[Andrew_Sm]

Originally Posted by krapic

Hi everyone,

I'm simulating a bluff body stabilized flame (test HM1E in Sydney) using the rhoFlameletFoam solver made by Hagen Mueller.

I initialize the simulation with the solution made with rhoSimplecFoam (just air), and I set a field (setField utility) of fuel (methane hydrogen mixture 1:1) in the fuel tube. As you can see from the velocity plot at 0.2s, the flame doesn't stabilize. There were some issues with pressure waves in the fuel tube so the inlet, outlet, coflow and side are set to a waveTransmissive boundary - that solved the pulsating flow at the inlet tube but didn't affect the flame near the coflow.



I experimented with a wide variety of configurations including various mesh densities and BC values but nothing seems to help. Here is the link to the case folder on google drive (due to the size limit on cfd-online). Note - the cantera tables aren't included there. I also included a few videos showing the progress of the simulation.

Does anyone have an idea of what might be causing the instability and have a possible solution to the problem?

Thank you in advanced,
Kristijan

Lemme know when you resolve the issue
Thanx for sharing

[babakflame]

Greetings Kristijan

I downloaded your case and will take a look at it in a few days. I am busy at the moment. I hope to be capable of helping.

Best,
Bobi

[babakflame]

Greetings Kristijan

Normally, You should not experience this instability.

First of all, I think you should reconsider on your grid resolution (Just 13000 points )

In a 3 dimensional simulation with RAS, my recommendation is a grid with at least 1 million points.


Since you are using RAS, my recommendation is a wedge-typed grid with about 150,000 points.

I have done pretty much simulations with this solver, change your grid resolution and check the grid quality.
Afterwards, send me your email, I will send you my 0, constant and system folders (without polyMesh folder).

Your simulation must be stable. Don't worry my friend.

Best,
Bobi

[babakflame]

Greetings Kristijan

I will take a look at your grid tomorrow.

My recommendation is using RAS instead of URAS solvers for quick answers (without any noticeable difference in accuracy) refer to: http://www.cfd-online.com/Forums/ope...nsmoke-22.html

Yes, the proposed tuning for C1 is applicable in bluff-body stabilized flames.

I will email you the files tomorrow.

Best,
Bobi

[babakflame]

Greetings Kristijan

I took a look at your grid. Although I did not visualize it by paraview, the overall structure seems O.K.

I sent u the files via file2send. The files are for a LES simulation. Thereby, you need to change AlphaSgs to Alphat and muSgs to mut. The details of the files are similar. Take a look at what I have used for inlets, outlets and walls c.f. bluff-body.
However, for velocity inlet, use precursor simulation.
I think you have already done it by extending fuel injection to upstream so as to let the fuel stream become fully developed.

Let me know what happens this time

Best,
Bobi

[krapic]

Originally Posted by babakflame

Greetings Kristijan

I took a look at your grid. Although I did not visualize it by paraview, the overall structure seems O.K.

I sent u the files via file2send. The files are for a LES simulation. Thereby, you need to change AlphaSgs to Alphat and muSgs to mut. The details of the files are similar. Take a look at what I have used for inlets, outlets and walls c.f. bluff-body.
However, for velocity inlet, use precursor simulation.
I think you have already done it by extending fuel injection to upstream so as to let the fuel stream become fully developed.

Let me know what happens this time

Best,
Bobi

Hi Bobi,

I am running your case right now but since I have access to a computer with only 16 cores, it's going to take a while - 3 days at least.

In the meantime, maybe you could write your thoughts on this kind of burner configuration for everyone in the future to see. These are the possibilities as far as I know as of now - Hagens solver is URAS so it's computationally really expensive, libOpenSmokes steady state solver is fast but cannot capture the highly transient behavior of this type of flow (at least I tried it and cannot achieve a stable solution). LibOpenSmoke's PISO solver is no longer available (at least on OF23x) because of stability issues. Finally, I assume LES is even more computationally demanding then URAS and thus not really appropriate.

So what would you advise people doing this kind of problem to do?

Best regards,
Kristijan

[babakflame]

Greetings Kristijan

Providing hints for just a specific configuration is not a correct way of helping people unless the case be very special.
Instead, Tobi is writing a summary detailing solvers employing flamelet approaches based on OpenFOAM. I hope that in this way people would receive comprehensive helps instead of partial hints.

Best,
Bobi

[dimaCFD]

Hi Kristijan,

You can try my solver: http://www.edcpisofoam.decgroup.org.
with some support from me =)

In general, this application is computationally expensive for the detailed chemistry (etc., the full GRI3.0 mech), however it works fine for one-, two- or three global reactions.

The most stable release is for OF-2.1.x, however if people will be very interested I can provide it for OF-2.3.1 (testing it now).

Thx, Dmitry

[krapic]

Originally Posted by dimaCFD

Hi Kristijan,

You can try my solver: http://www.edcpisofoam.decgroup.org.
with some support from me =)

In general, this application is computationally expensive for the detailed chemistry (etc., the full GRI3.0 mech), however it works fine for one-, two- or three global reactions.

The most stable release is for OF-2.1.x, however if people will be very interested I can provide it for OF-2.3.1 (testing it now).

Thx, Dmitry

Hi Dmitry,

I am aware of your solver, I've read a few reports/papers from people successfully simulating this case with your solver.

Unfortunately, since this is a university project, my supervisor and I agreed to use Hagens solver. Maybe at some point in the future I'll check out yours.

Thanks for the tip though,
Kristijan

[krapic]

OK guys,

Glad to say the problem is solved.



To my big surprise, the solution was reasonable even with 4.000 cells (image above - comparison between C_e1=1.44 and 1.60). I'm not saying that number of cells is optimal, but it's a good start since simulations run relatively fast on such a coarse mesh.

The key is to use a rather low maximum Courant number, in the range of 0.05 to 0.1 Co. Also, use low under-relaxation factors, even as low as 0.01. Then, as the simulation progresses, you can gradually increase both the maximum Courant number and the relaxation factors. My flame would stabilise around 0.5 seconds so every 0.05-0.1s I would increase those numbers.

Hope this helps,
Kristijan

[babakflame]

Greetings Kristijan

Good to hear successful news from you.

I have simulated this flame with Smagorinsky and SpallartAlmaras Sgs models. At the moment, I am testing oneEqEddy on this flame.

Best