[dguti]

Dear fellow ANSYS ICEM users,

I am trying to hexa-mesh a relatively simple geometry. It's just a cylinder with a truncated sphere at the top, similar to a a bullet or a dome (see Snap_01 attached). The mesh is treated in ANSYS CFX as a solid body with a HTC and a reference temperature boundary conditions for the dome. The rest of the parts are treated as adiabatic. The goal of the computation is to evaluate how the dome is heating up in a transient solid body simulation

The procedure for the mesh generation looks as follows:
1. Create one block and split it into two blocks.
2. Generate o-grid mesh
3. Set the mesing parameters
4. Mesh!

After checking the quality parameters, everything is looking quite good (Q > 0.6). I then export the mesh to ANSYS CFX and let it run for a simulation time of 5 seconds with a timestep of dt = 0.01s. I export the temperature results every 10th iteration, that is, every 0.1s.

What I found out is that at some vertices of the mesh I obtain unphysical temperatures after a couple of computation seconds, which can only be attibuted to a meshing problem. Futhermore, the residuals for the energy equation increase drastically to values higher than 10e-3, which is not usual/acceptable for a solid body simulation, in which only the energy equation is being solved.
In these vertices, even if the quality parameters are looking fine, one can observe that the cells look quite "stretched" (see Snap_06).

I tried to play a bit with the meshing parameters (resolution) and moved some of the inner vertices of the o-grid but I couldn't find an optimum solution.

Any hint about what could be going wrong?

Many thanks for you time!

Best regards,
David

[dguti]

Attached Snap_06.