[farojas]

Hello,


I am doing RANS 2D axisymetric simulations on a CD nozzle in starccm+


I have a nozzle with a NPR= 34.7, the boundary conditions that I am using are stagnation inlet for the inlet, outlet pressure for the outlet, wall with no-slip conditions for the farfield; and of course I defined the axis.


However, I am struggling doing the mesh, in particular with the prism layer at the nozzle wall. Right now I am using the automated 2D mesh option and selected the triangular mesh. I have used this calculator https://volupe.com/support/all-y-cal...-cfd-analysis/ to estimate the prism layer characteristics and the base size. I am looking for y+ <1 at the wall.


These are the outputs I got from the calculator using 33 layers and a stretching factor of 1.2 :


Prism layer total thickness [m]: 0.005347760676980807
First cell height [m]: 0.000002613851473445454
Last prism layer cell height [m]: 0.000002613851473445454


It is worth mentioning that the Throat radius is 13 mm and the exit radius is 75 mm.


The preliminary results quite match with a paper which I am using as a reference, but they are not accurate enough. The residuals are not converging and I think is because the quality of prism layer cells is not good. I have a quality of 0.005 for the nearest cell to the wall using the values that I have.


The problem is that when I refine the mesh to get a better cell quality the number of cells that I need is too large (The cells after the prism layer should have a size of the order of 1e-6 m for an aspect ratio of 0.5)


My question is, how can I improve the prism layer cells quality and what value of cell aspect ratio would be enough?



I would really appreciate any help.

[Marc_462]

Could you provide some mesh images and the models are you using to configure the flow physics?

[farojas]

Hello Marc,


Sorry for my late response.



Here are some images form my mesh and physics

[Marc_462]

I don't think the meshing of the prism layer is causing those problems. Maybe the last layers of the prism layer are too high (lowering the stretching factor to 1.1 or 1.15 would fix it). But overall, the prism layer looks good. Perhaps the meshing with tetrahedra, or in this case triangles, isn't the most suitable for this type of problem, and a trimmed or polyhedral mesh might work better.

[farojas]

Thank you so much, I have changed to a trimmed mesh and now it seems to work.


Thank you so much again!


Best regards