[yufan]

Dear all,

I'm currently working on a numerical simulation of deflagration to detonation transition (DDT) in a straight tube. A premixed hydrogen and oxygen gas fills the tube, which is closed at the left end and open at the right end, as shown in the .png file below.

The literature defines a small burned zone of 2mm × 6mm at the left end as the initial condition. Then the flame gradually accelerates until it catches up with the shock wave.

The problem is how to define the burned zone. I've tried using the equilibrate function in Cantera. What parameters ['TP', 'TV', 'HP', 'SP', 'SV', 'UV', 'UP'] should I choose to define this weak ignition condition?


Yours
Yufan



Reference:
[1]Ivanov, M. F., Kiverin, A. D., Yakovenko, I. S., & Liberman, M. A. (2013). Hydrogen–oxygen flame acceleration and deflagration-to-detonation transition in three-dimensional rectangular channels with no-slip walls. International journal of hydrogen energy, 38(36), 16427-16440.
[2]Zhou, F., Liu, N., & Zhang, X. (2018). Numerical study of hydrogen–oxygen flame acceleration and deflagration to detonation transition in combustion light gas gun. International Journal of Hydrogen Energy, 43(10), 5405-5414.

[bigfootedrockmidget]

Are you setting up a CFD simulation for DDT and you would like to create an accurate ignition sphere? You can indeed use Cantera's 'equilibrate' for this, and the exact choice does not matter that much, as explained here, they are equivalent when choosing equivalent values: https://cantera.org/3.1/userguide/python-tutorial.html


The main point of importance is that the flame kernel contains an amount of energy that is higher than the ignition energy, and that the setup is represented accurately.



We noticed in simulations that the results are sensitive to the placement of the flame kernel (if ignition is in the domain, then do not ignite at the inlet/wall), the initial size of the flame kernel (should be close to minimum flame kernel size) and the reflection of the pressure wave at the outlet (proper outlet boundary condition). Results of flamefront location X(t) were correct with < 1% error, just using Fluent.






Some setup for you, copy-pasted from internal report:

Turbulence model [1]:

„RNG k-ε turbulence model.

„Non-equilibrium wall functions.

„

Solver:

„Ansys Fluent 2024 R2.

„Unsteady.

„Convergence:
- Energy equation à1⋅10^(-6)
-All other equations à1⋅10^(-5)
- Max 350 inner iterations.

„Timestep:
- Acoustic CFL standard.
- C_A=0.5.

Combustion Model:

„Detailed chemistry model.

„Stanford H2/O2 Mechanism version 1.2
- Argon species and reactions are neglected.

„9 Species.

„20 Reaction mechanisms.

Eddy Dissipation Concept (EDC) Model.

[yufan]

Originally Posted by bigfootedrockmidget

Are you setting up a CFD simulation for DDT and you would like to create an accurate ignition sphere? You can indeed use Cantera's 'equilibrate' for this, and the exact choice does not matter that much, as explained here, they are equivalent when choosing equivalent values: https://cantera.org/3.1/userguide/python-tutorial.html


The main point of importance is that the flame kernel contains an amount of energy that is higher than the ignition energy, and that the setup is represented accurately.



We noticed in simulations that the results are sensitive to the placement of the flame kernel (if ignition is in the domain, then do not ignite at the inlet/wall), the initial size of the flame kernel (should be close to minimum flame kernel size) and the reflection of the pressure wave at the outlet (proper outlet boundary condition). Results of flamefront location X(t) were correct with < 1% error, just using Fluent.






Some setup for you, copy-pasted from internal report:

Turbulence model [1]:

„RNG k-ε turbulence model.

„Non-equilibrium wall functions.

„

Solver:

„Ansys Fluent 2024 R2.

„Unsteady.

„Convergence:
- Energy equation à1⋅10^(-6)
-All other equations à1⋅10^(-5)
- Max 350 inner iterations.

„Timestep:
- Acoustic CFL standard.
- C_A=0.5.

Combustion Model:

„Detailed chemistry model.

„Stanford H2/O2 Mechanism version 1.2
- Argon species and reactions are neglected.

„9 Species.

„20 Reaction mechanisms.

Eddy Dissipation Concept (EDC) Model.

Heyyy Nijso,

Nice to see you again here! Thank you for your reply!

In my understanding, it is important to give a reasonable weak ignition condition at the left end of the smooth tube, and I found that almost all of the literature donnot specified this condition clearly, they state as "burned region is set at the left end","The laminar flame was ignited at left end".

Actually this kind of statement is quite tricky I think, I think ignition definition in DDT problem is really important, which affect the velocity of the flame front and the position of the DDT process happens. For very explicit example if you give a extremely high temperature and high pressure to the ignition region, there DDT won't happen, the detonation phenomenon happens immediately. Could you specified the ignition condition more specified?

Anyways, the 'equilibrate' in cantera for the different condition, the temperature, pressure, composition varies a lot, ex, the temperature may varies from the 1500K to 3700k for H2:O2=2:1. I use SV equilibrium condition to set the ignition condition(p = 1.60380Pa; T = 3057.1K; massf={H2=0.017,H2O=1-0.017-0.093-0.004-0.101-0.023,O2=0.093,H=0.004,OH=0.101,O=0.023})compared to the premixed flame (p=101325Pa, T=298K) the temperature profile is quite different from the literature shown, DDT happens real quick

Lastly, thank you for your FLUENT setting, unfortunately, i donnot use FLUENT anymore cuz i donnot have license anymore right now, i use the SU2-NEMO and Eilmer5.
Did the "Stanford H2/O2 Mechanism version 1.2" refer to the paper "Hong, Z., Davidson, D.F. and Hanson, R.K. (2011) An improved H2/O2 mechanism based on recent shock tube/laser absorption measurements.Combustion and Flame, 158, pp. 633--644" ? I did use this mechanism and Eilmer5 to do the simulation. And I think maybe DNS and LES might be better choice for DDT problem although they are very consuming.

THANK YOU for your time reading my reply!!!


Yours
Yufan

[bigfootedrockmidget]

Hi Yufan,
Yes, the exact ignition process is important, but what we found is that the actual burnt condition does not matter that much, as long as you provide sufficient energy to get the reactions started. And we stay close to the burnt conditions of a constant pressure laminar premixed flame, so Tb<2200 K.
We ignite using a small ignition kernel, so there is a small amount of 'incorrect' burnt mixture present that is supposed to have a negligible effect on the rest of the simulation.
In the initial phase, the flame kernel is still a laminar expanding spherical flame, so a constant pressure equilibrium mixture is a good assumption for the burnt conditions in the flame kernel imo.
Yes, we used the mechanism from the paper of Hong et al. Indeed, DNS might be a better approach, but it will be much more expensive, especially if you want to take into account the preferential diffusion accurately using a multicomponent diffusion approach.



Are you already on our SU2 slack channel? Link is on our website.
We have weekly developer meetings, schedule is on slack.

[yufan]

Originally Posted by bigfootedrockmidget

Hi Yufan,
Yes, the exact ignition process is important, but what we found is that the actual burnt condition does not matter that much, as long as you provide sufficient energy to get the reactions started. And we stay close to the burnt conditions of a constant pressure laminar premixed flame, so Tb<2200 K.
We ignite using a small ignition kernel, so there is a small amount of 'incorrect' burnt mixture present that is supposed to have a negligible effect on the rest of the simulation.
In the initial phase, the flame kernel is still a laminar expanding spherical flame, so a constant pressure equilibrium mixture is a good assumption for the burnt conditions in the flame kernel imo.
Yes, we used the mechanism from the paper of Hong et al. Indeed, DNS might be a better approach, but it will be much more expensive, especially if you want to take into account the preferential diffusion accurately using a multicomponent diffusion approach.



Are you already on our SU2 slack channel? Link is on our website.
We have weekly developer meetings, schedule is on slack.

Dear Nijso,

Greetings!


Thank you for your reply. I fully understand what you mean.
what I mean is that I found different ignition conditions will affect DDT position and deflagrating flame front velocity vt diagram especially exponential acceleration process, so the model validation process for DDT in smooth tube is very roughly, I can only use CJ- velocity, pressure, temperature to validate my case’s setting rather than the temperature /pressure contour at different time if the ignition condition is not giving in details, which confused me a lot .But recently I realized that using the CJ property and empirical DDT position equations for model validation is fairly enough.

Yes, I have joined the slack channel, and If I remember correctly, you also helped me to update the H2-O2 database of mutationpp. I was originally going to compare the Eilmer 5 results to the SU2-NEMO results, but I haven't gotten around to doing that yet.

And thank you for inviting me join the developer meeting, I am glad to do so, I did not notice the meeting information before, I will check it again later!


Best regards
Yufan