Prism layers and grid dependence study
 

Hi all,

I am conducting a grid dependence study for a free falling wedge impacting the water surface. Core mesh base size and time step are divided by the same factor (2 in this case) and the resulting surface average pressures are compared.

My question is, should the 'total' thickness of the prism layer be kept constant (i.e. using an absolute value for the prism layer thickness) or divided by the same factor (i.e. using the relative value setting)? In my opinion, I think that it is better to specify a relative size rather than an absolute one to keep the same thickness ration between the last prism layer and the core mesh. I have tried both options however when using the absolute size, it becomes very difficult to maintain the same CFL, which when doing a grid dependence study in a transient simulation I think is critical.

Also, when using the relative size option and keeping all other settings (number of prism layers and stretching factor) the same, the thickness of the 1st prism layer is also divided by 2. This is the correct approach right, I mean, if I am dividing the core mesh by a factor of 2 it makes sense that the thickness of the prism layers is also divided by 2?

Am I making sense?

Thanks and regards,

Josef

Using a relatively sized prismatic layer won't control the final layer/core aspect ratio any better than using a constant size. If you want to control that value, there's a setting for it in the prism layer expert properties.

I don't know what you're saying at all with the last paragraph.

The prism layer settings are partially dependent on physics. Your first layer should be sized correctly to get the wall y+ you need. This is independent of any grid study. The total thickness should be chosen to resolve the near wall gradients enough. This is minorly dependent on the grid study. For doing a grid study, you should really only be modifying the number of elements in the layer.

Hi me 3480,

thanks for your reply. Keep in mind that I am doing a grid dependence study so when I am saying that the prism layer/ core aspect ratio is better maintained what I mean is that when I divide the base size by 2 to go from coarse mesh to medium mesh and medium mesh to fine mesh everything else is divided by two.

That is was I was saying in the last paragraph also; when dividing the base size by two to get the medium mesh from the coarse mesh and keeping all other settings (stretching factor, relative size and number of PL) the same, the total prism layer thickness would be divided by two and also the thickness of all the PL. Therefore, if initially I had developed a coarse mesh with good prism layer/core cells ratio when I divide the core cells by two even the prism layers are divided by two and the ratio is maintained.

Would you suggest using a constant total prism layer thickness and vary the number of prism layers or?

Thanks and regards,

Josef

Oh, I see.

Yes, I would recommend you specify the prism layer height manually. The prism layer height is dependent on the boundary layer, and it doesn't make sense to change it with some arbitrary base size.

me3840,

Thank you!