[pianke]

Dear sir,
Recently I've had a problem with a hard ignition. In order to reduce the calculation time, I deleted the intake and exhaust ports of the engine, leaving only one cylinder. At the same time, the initial condition of premixed methane was set in the cylinder. I then added a hydrogen nozzle to the cylinder head. My purpose is to do hydrogen direct injection and methane premixed combustion. The combustion model selects SAGE, the ignition model selects Energy, the energy is 40mj, and L-type ignition. But there was a situation where it couldn't be ignited. Considering the sparse grid, I have encrypted the ignition model to level 6, but it still cannot be ignited. The strange thing is that when I choose the G equation combustion model and use passive-G-transport to ignite, it can ignite.
The CONVERGE manual says that the G equation is mostly suitable for premixed combustion. I think my calculation example will be more accurate with the SAGE combustion model. Is there any way to solve it?
If possible, I would like to provide my CVG files to CONVERGE engineers.

[angela.wu]

Hello,


Is the energy source some type of shape? What's the size of it? Try reducing it's size/increasing energy deposited to increase the energy density. You should be able to ignite the mixture with SAGE model if it can reach the temperature needed to start combustion.


Thanks,


Angela

[pianke]

Hello Angla.
I've tried reducing the size, increasing the energy, and raising the equivalence ratio to 1. The specific settings are shown in the picture, but the ignition is still not successful.

[angela.wu]

What is your embedding settings around the spark source?

[pianke]

My base grid is 8mm, fixed embedding as shown in the picture

[angela.wu]

It's possible that the embed level is not fine enough to capture the onset of ignition. An embed level of 5 with base grid of 8 mm means there is only 0.25 mm grid around the ignition source. Please check example cases (Internal Combustion Engines --> Gasoline Spark Ignition Premix --> SI8 engine premix SAGE transient RANS) for best ways to set up the embedding around the source. That example uses a base grid of 4 mm and the highest embed level is 5, equal to a grid size of 0.125 mm around the source.

[pianke]

Originally Posted by angela.wu

It's possible that the embed level is not fine enough to capture the onset of ignition. An embed level of 5 with base grid of 8 mm means there is only 0.25 mm grid around the ignition source. Please check example cases (Internal Combustion Engines --> Gasoline Spark Ignition Premix --> SI8 engine premix SAGE transient RANS) for best ways to set up the embedding around the source. That example uses a base grid of 4 mm and the highest embed level is 5, equal to a grid size of 0.125 mm around the source.

Thank you very much Angela, your advice was very useful.
But after I used SAGE+ENERGY to ignite and burn, I encountered a problem. The flame of the SAGE model cannot spread to the entire cylinder, but only concentrates on the middle part, as shown in the figure below. But when I switch to the G equation combustion model + ENERGY ignition, the entire cylinder can be ignited. My methane equivalence ratio is set to 0.625 and the initial conditions are the same for both cases. May i know what is this all about? Or my situation is more suitable for G equation + ENERGY?

[angela.wu]

The SAGE model is more accurate since it's using the actual chemical kinetics. There are many reasons why your combustion simulation may not be accurate or what you expected. The mechanism that you're using could be inaccurate, or perhaps your mixture preparation is not good enough to combust throughout the entire cylinder. I would first verify that cold flow simulations, i.e. without combustion, is correct. If that's accurate, then turn on combustion modeling, and see what the results look like. If the results are still not correct, then there might be some issues with your mech.dat for a premixed CH4+H2 DI simulation.

[pianke]

Originally Posted by angela.wu

The SAGE model is more accurate since it's using the actual chemical kinetics. There are many reasons why your combustion simulation may not be accurate or what you expected. The mechanism that you're using could be inaccurate, or perhaps your mixture preparation is not good enough to combust throughout the entire cylinder. I would first verify that cold flow simulations, i.e. without combustion, is correct. If that's accurate, then turn on combustion modeling, and see what the results look like. If the results are still not correct, then there might be some issues with your mech.dat for a premixed CH4+H2 DI simulation.

Hello Angela.
Now I haven't injected hydrogen into the cylinder, it's just methane burning. Since my model doesn't have an intake, my initial conditions are set to premixed methane, so I think my mixture configuration is fine. The mechanism file I am using is GRI30. I'm surprised that the G-equation combustion model works better than the SAGE model for premixed methane. I can provide my model to you if needed.

[angela.wu]

Is your base mesh size still 8 mm? It's recommended that the in-cylinder mesh size is 1 mm. That could also affect your combustion process. Also, you can try AMR for combustion (AMR, temperature SGS of 2.5 K).

[pianke]

Originally Posted by angela.wu

Is your base mesh size still 8 mm? It's recommended that the in-cylinder mesh size is 1 mm. That could also affect your combustion process. Also, you can try AMR for combustion (AMR, temperature SGS of 2.5 K).

Hello Angela. Thanks for your advice. I ran a case according to your suggestion. The base grid is 4mm, AMR is turned on, and the maximum encryption level is 2, which should be 1mm after conversion. The settings are shown in the figure below. It is indeed a great improvement compared to before, but the flame still did not spread to the entire cylinder. Is it because my embedding level is not enough? Or should it be better to fix the grid size of the entire cylinder to 1mm? Besides, why can the G equation completely burn methane in the cylinder without embedding?

[MFGT]

What kind of flow motion do you have in your cylinder? Maybe there is just not enough TKE or Swirl?

[pianke]

Originally Posted by MFGT

What kind of flow motion do you have in your cylinder? Maybe there is just not enough TKE or Swirl?

Thanks for your advice. The airflow movement is something I didn't think of, probably because my cylinder model has no intake and exhaust passages, so the airflow movement is weak. Hence insufficient combustion.