[Athos_e]

Hello everyone,

I've been trying to simulate a shock tube detonation using SU2 NEMO and MUTATION++. I will outline the procedure first:

My shock tube is composed of two zones, the left one that is a short zone with high pressure and temperature in order to initialise the detonation, and the right zone that has gas in ambient conditions. Following another thread I was advised not to use multizone simulation as this is not very efficient. So what I did was to create two different meshes and 2 different cfg files for the two sections I described above. Having ran each sim for 2 iterations (since I use 2nd order dual time stepping, two iterations are required to have a restart solution) I obtained two restart files that I later on combined into one using a script. This restart file basically includes the points of the whole domain, both the left and right section. I validated that this procedure works as I ran the same case with Euler as a solver and I obtained similar results to mutizone approach and Euler. Therefore I can confirm this apprach works.

However the problem arises when I use nemo_navier_stokes as solver. In particular when I wan to include chemical kinetics and so set FROZEN_MIXTURE to NO in the cfg file. I noticed that by using nemo_navier_stokes with mutayion++ but setting FROZEN_MIXTURE to YES i.e no chemical kinetics, the expansion of the gas is as it should. Since my shock tube has 3 walls and an outlet, the high pressure packed on the left side of the tube will naturaly epand towards the outlet, which is what happens when ignoring the chemical reactions.

However, as soon as I include chemical reactions with frozen_mixture= NO, the simulation gives wrong results. Instead of having a wave moving towards the right, all the parameters (pressure, temperature, mass fractions etc) change along the whole domain at the same time gradually. Please see the attached pictures to give you a better undertstanding.

Has anyone experienced anything similar, and does anyone know why this might be the case? Is this mutationpp related or?

Thanks in advance and best regards,
Athos

[wallym]

Hi Athos,

Thanks for your question and for using SU2-NEMO! I am glad to hear that it is getting used. I took a look at your config file and noticed that your species are listed as (.77,.23,0,0,0).

I believe this would be setting monoatomic nitrogen and oxygen at these levels. So when the chemistry is turned back on - everything recombines and things go crazy.
I haven't been developing the code for a while, so the above may not be true. But when I last left off, the native library and mutation++ had a different order of inputs.

Let me know if this helps.

[Athos_e]

Hello Wally and thanks for the quick reply.

This makes a lot of sense now that I think about it. And yes you are right, the first two entries in air_5 for mutation++ is monoatomic nitrogen and oxygen. I was gradually building this test case testing other things and totally forgot to check again my mixture.

In the end I would like to simulate this pulse using hydrogen and air mixture. Would you suggest me to first check a one step reaction (create my own) for hydrogen-air and then move on? Just to make sure everything works.

Also from the papers I read and try to replicate the results, on the left zone they have a mixture that are only reaction products, whereas on the right zone it's just hydrogen-air mixture at ambient conditions and set to stoichiometric ratio. I think I will be able to specify the left zone to have products and the right just the mixture (since I firstly run these two zones separately). But in the end when I combine the two and run a cfg file for the whole domain like the one I posted, what should I use as GAS_COMPOSITION since the two zones don't have the same composition?

Thanks again for all the help.

Best reagards,
Athos

[wallym]

Awesome, I'm glad I didn't lead you astray.

I think in your case, creating a mutation++ reaction file wouldn't be too difficult and sounds like it would be useful. You can also try using an N2 mixture to simplify the problem. I would try an inviscid solution here first.

To answer your second question -> it shouldn't matter. Because you will be using a restart file, the flow state will be extracted from that and not the input parameters.
The issue will be in your boundary conditions (farfield, outlet, etc.). If you do include a farfield/outlet, I would use the gas composition of that side (I recommend using walls though). Make sure all species on both the left and right side are accounted for, even if you specify the concentrations to be zero.

Let me know if that helps!

[Athos_e]

Hello again,

Thanks for the reply. I had already implemented a single step reaction with hydrogen and air and then also moved to a multi step case. Everything seems to work much better than before and the behavior on the right part is as expected. Nothing happens until the wave comes from the left Also i have a steady consumption of my initial mass fraction that turns into the products i set using the reactions. So this seems correct.

However now i encountered another problem.
The issue seems to be that the reactions after some small distance from the initial hot zone seem to stop or slow down by a lot. The temperature also drops so maybe this is what it causes it. But i don't understand why. Since we have some reactions I would expect to have also some heat generated (provided they are exothermic). Whereas here it appears that there is not a self sustained reaction. This is also the case for pressure. If we have a reaction then pressure would theoretically increase, whereas here it is not. Am i missing something? Or could it be that my reactions do not actually provide any exothermic energy and maybe i should include even more? This is the case for both single and multi step reaction by the way.

Any help is greatly appreciate
Best regards,
Athos

P.s. The pictures are for multi-step reactions. In particular i have 4 reaction and 6 species. You can see that the 'pressure' wave is in front of the temperature or the mass fraction (this mass fraction is the one that is used up with time, not the product). Velocity mach number and other variables also match the pressure. And also Mach number doesn't really reach =1 as it would be expected if the detonation was successful.

[wallym]

Hmmm,

I think there are a lot of potential questions arising from your simulation.
Are we sure that the pre- and post-shock conditions are correct? If they aren't you will see information propagating at differing characteristic speeds.

Your reaction files look good, but I don't have a good grasp on these reactions. Could you post curves of the species concentrations through the wave? This may provide insight to the reaction behavior.

This could potentially be a numerical issue, but I don't think so.

[Athos_e]

Hello again,

Sorry for the late reply. Since I last posted on the forum I've read some more papers. I believe the reason my reactions stop is because a reflective wave bouncing of the left wall, reaches my primary shock wave. Due to this interaction it cause the reactions to stop. When I implemented an outlet on the left so that the reflective wave exits the domain, the chemical reactions seem to continue.

However, now I experienced some other issues that hopefully I will solve. When I manage to get the results I want I will post here so that it's in the forum in case someone has the same doubts.

Thanks for your help again!