Learning Forum

[sarbakhshian]

The geometry I attached below is the one I'm trying to mesh. It is not very complicated and I'm guessing a good Hex mesh is possible for this geometry. I'm trying the multizone method with ansys meshing but I'm facing the error in the picture. Anybody has any suggestion how to have a structured Hex mesh for this geometry?

[peteroznewman]

You don't say what you are meshing it for. Is it a Structural model of the duct or are you going to mesh the air in the duct for a CFD model?
If its a structural model of the duct, then open the Geometry in SpaceClaim and extract the midsurface and mesh that with quad shell elements. Don't mesh a thin walled structure like that with hex elements.
If its for a CFD model, you don't have a solid body of the air. You can get the fluid volume in SpaceClaim.

[sarbakhshian]

Actually I'm gonna do FSI analysis for both solid and fluid domains. The fluid body in the picture is suppressed so it is not shown here. I'm gonna try shell elements, I hope it works for me.
thank you very much for your time

[sarbakhshian]

Hello Sir
The idea of shell elements for my geometry helped me a lot. I managed to mesh the solid part of my model but I just faced a new problem with my fluid domain. As you see in the picture, there is a triangular area
in the geometry which ansys mesher wont detect it's edges and twists the elements around that edge. I couldn't understand what is the problem with that edge!!
I'm also attaching the picture of my fluid domain, Do you have a general idea how can I have hex elements with infilation layers for this geometry?
Thanks a lot for your attention

[peteroznewman]

Here is one way to get the shell element nodes to line up with the solid element nodes.
Stop using a Cartesian mesh method that you have used in the example above.
If you are willing to give up on hex mesh and use a tet mesh for the fluid, you will get triangles for the structure.
To keep hex mesh in the fluid, you will have to figure out how to slice the solid up into six-sided objects.
Say more about the FSI, is it one-way or two-way? What quantity is being sent to the structural solver? Pressure?
For a one-way FSI, ANSYS has data mapping software that can take results on one mesh and apply it to a different mesh. This would let you solve the fluids using the cartesian mesh method and have a non-congruent mesh of quad shell elements on the surfaces for the structural analysis.

[sarbakhshian]

peteroznewman
The model is a duct made of thin walled st37 steel with maximum thickness of 12mm. Air flow will be going through inlet with a varying velocity 0.3 to 0.7 Mach and the outlet condition will be varying outflow. Structural wall displacement is required as my solution results. Based on the material properties I think one-way FSI would be accurate enough for this analysis. As long as I know, in one way FSI fluent transfers pressure data into mechanical solver. Coupled pressure-velocity equations will be chosen for the analysis and k-w model will be used for boundary layer area.
I'm also thinking of doing the FSI analysis just with fluent, because as long as I remember fluent has this capability for one-way FSI.
I'll appreciate any guide and suggestions for this simulation from you

[peteroznewman]

You can't do structural stress calculations in Fluent only. Fluent computes the pressure, Mechanical computes the stress due to the pressure.

[Erik Kostson]

and
There are two ways to do a FSI with Ansys.

The first option is using the standard 1 or 2 way coupling within mechanical between fluent and mechanical system (Static say). This is what is referred to in the help manual as the extrinsic FSI method.

The other option (is probably less known than the first one), is to model and simulate 1-way and 2-way fluid-structure interaction completely within a Fluent session, something referred to as intrinsic FSI.

See the help manual for more info on this.
Chapter 24: Modeling Fluid-Structure Interaction (FSI) Within Fluent (ansys.com)
For any questions on those ways, use the fluid dynamics forum.

Thank you

Erik