[alenp07]

I have couple of questions:

1. How critical is to maintain Yplus values in combustion simulation?
Being an starCD user, I have seen a presentation from one or my colleagues showing the impact of Wall Yplus layers on Tumble and TKE using a trimmed mesh. But I got this doubt, if there is near wall treatment (Standard) why someone is really worried about generating extra layers and increasing the run time? Also if you use a refined mesh why would someone use a near wall treatment?

2. Does it make sense to ensure the yplus values within a range (say 30-150) in Converge by using boundary AMR option?
I just did the setup and imposed the yPlus criteria for head, port, cylinder and piston for a cold flow without injection throughout the simulation and what I could observe was the simulation time went almost double with identical results.

3.What is effect of turbulence model on predicting the wall Yplus? What is recommended for Converge?

4. Do we need to be really worried about yPlus values for engine simulation? both Cold flow and Combustion?

5. For generic flow simulations, maintaining a yplus within a certain range can ensure the the independent nature of the mesh. In Converge how am I ensuring that? (if yplus is not a criteria).

Thanks in advance

[jetcheve]

Originally Posted by alenp07

I have couple of questions:

1. How critical is to maintain Yplus values in combustion simulation?
Being an starCD user, I have seen a presentation from one or my colleagues showing the impact of Wall Yplus layers on Tumble and TKE using a trimmed mesh. But I got this doubt, if there is near wall treatment (Standard) why someone is really worried about generating extra layers and increasing the run time? Also if you use a refined mesh why would someone use a near wall treatment?

2. Does it make sense to ensure the yplus values within a range (say 30-150) in Converge by using boundary AMR option?
I just did the setup and imposed the yPlus criteria for head, port, cylinder and piston for a cold flow without injection throughout the simulation and what I could observe was the simulation time went almost double with identical results.

3.What is effect of turbulence model on predicting the wall Yplus? What is recommended for Converge?

4. Do we need to be really worried about yPlus values for engine simulation? both Cold flow and Combustion?

5. For generic flow simulations, maintaining a yplus within a certain range can ensure the the independent nature of the mesh. In Converge how am I ensuring that? (if yplus is not a criteria).

Thanks in advance

Good morning!

1 - The importance of Yplus is dependent on what quantities you are interested in. In a simulation, for example, where you are tracking heat transfer, or doing a CHT in conjunction with combustion, it is recommended to maintain Yplus values in a sensible range. I do not think resolving the viscous sublayer as opposed to using a wall treatment/model is practical due to the fact that this would required sub-micron size cells in many cases leading to extremely large computational overhead.

2 - Boundary AMR is a good solution to this if you feel maintaining Yplus in a tight range is important. I would recommend that you are releasing cells as necessary as well, as over resolving the viscous sublayer while using a wall model may lead to, for example, over predicting heat transfer.

3 - I'm not certain your question here. Could you clarify? However, we do recommend using RNG k-eps for ICE applications.

4 - See answer to question one. This is highly dependent on what you are interested in.

5 - If you are interested in grid independence, you can easily use the grid scale parameter in inputs.in to scale the dx, dy, and dz values of the grid by factors of two allowing you to run simulations to see if grid independence is achieved.

[alenp07]

Originally Posted by jetcheve

Good morning!

1 - The importance of Yplus is dependent on what quantities you are interested in. In a simulation, for example, where you are tracking heat transfer, or doing a CHT in conjunction with combustion, it is recommended to maintain Yplus values in a sensible range. I do not think resolving the viscous sublayer as opposed to using a wall treatment/model is practical due to the fact that this would required sub-micron size cells in many cases leading to extremely large computational overhead.

2 - Boundary AMR is a good solution to this if you feel maintaining Yplus in a tight range is important. I would recommend that you are releasing cells as necessary as well, as over resolving the viscous sublayer while using a wall model may lead to, for example, over predicting heat transfer.

3 - I'm not certain your question here. Could you clarify? However, we do recommend using RNG k-eps for ICE applications.

4 - See answer to question one. This is highly dependent on what you are interested in.

5 - If you are interested in grid independence, you can easily use the grid scale parameter in inputs.in to scale the dx, dy, and dz values of the grid by factors of two allowing you to run simulations to see if grid independence is achieved.

Hi,

Thank you for the detailed clarifications

I think I have a conclusion then. In my simulations (cold flow / combustion) if I am not really worried about the boundary phenomena then I don't need to be worried about the wall Yplus.

Regarding point 2, can you give me some more details on releasing cells? I have seen that option but couldn't understand how this can be useful.

Point number 3, is there an impact on the prediction of yplus if I change my turbulence model? What's the recommend model for wall heat transfer and near wall treatment for premixed combustion along with RNG k-eps model?

Also can we quantify the grid convergence by looking at the turbulent length scale?

Thank you once again!