Problems simplifying a lattice design for CFD from Fusion

[andrea.mazzoleniR2DKQ]

Hello everyone, and hope you are well.

It is my first post on the forums so forgive me if it's too wordy or, well, not good. I am happy to receive any feedback as to how I should post technical questions here.

I've been using Fusion 360 to design a scaffold for cell culture. An important feature of this design is that we need a porous disk with porosity changing along the radial direction of the disk, meaning that the scaffold will have higher porosity in the center and lower porosity at its edges.

To achieve this, I used the new lattice generation function, but I could not find an way to apply a variable solidity along the radial direction on a single solid disc body. To solve this I extruded a core cylinder in the center of my design, and surrounded it with concentric shells. All extrusions were treated as new components. I then generated a lattice from each of these layers, using a different solidity value for each, the way I explained in the previous paragraph. I then modelled, as an additional component, a tube surrounding the scaffold so that I could then patch it to import it to Autodesk CFD to run computational fluid dynamics.

The layers were then converted to a mesh body. The same was done to the tube adaptor. Finally all the meshed components were merged together in a single meshed body. I will attach two screenshots, one with the scaffold alone, the other with a section view of the scaffold in the adaptor, just to give an idea.

Now when I try to simplify the design for simulation IN Fusion 360 I cannot for the life of me manage to select the internal edges of the adaptor to patch them to generate the internal fluid volume of the design. Would you know how I might solve this issue, or in general, find a way to do CFD on such designs? Is it an inherent issue with my workflow or with the lattice generator function?

I will attach also an stl of one of my scaffolds. I unfortunately cannot attach an stl of the scaffold plus the adapter as it goes beyond the attachment limit.

Thank you very much for any feedback you might have, and I wish you a great day.

[TrippyLighting]

First you need to investigate what from of geometry is accepted by the CFD software.

My guess  is that that it accepts one form of geometry, either a mesh, or a solid but unlikely a mixture of both.

Then we can discuss how to make that happen using Fusion 360, but more likely other tools in combination with it or even without it.  

[andrea.mazzoleniR2DKQ]

Thank you, you raise a great point. I never used Autodesk Ultimate CFD before and I just assumed that it would import my meshed bodies no problem. So farI only managed to import my model from the simplify tab of Fusion 360, but the end result was the import of the solid bodies themselves (the full cylinders I generated the lattices from), not the meshes. 

The closest I've got to was importing the mesh as a .obj file, but then the software got stuck in generating the surface wrap indefinitely, even on the fastest setting. Trying the same with a .stp file got me an error during import for surface wrap.

I read that Ultimate CFD can accept mesh bodies as .unv .nas and .dat but i cannot find those options in the Fusion 360 exporter. 

So I think I'm stuck with using solid bodies. Is there a way you know that I can convert a lattice in a solid body in Fusion?

[TrippyLighting]

---

@andrea.mazzoleniR2DKQ wrote:

Thank you, you raise a great point. I never used Autodesk Ultimate CFD before and I just assumed that it would import my meshed bodies no problem.

Those meshes likely are volumetric meshes created by software aimed at FEA simulation, such as gmesh. 

.stl and .obj files describe surface meshes.

---

@andrea.mazzoleniR2DKQ wrote:

The closest I've got to was importing the mesh as a .obj file, but then the software got stuck in generating the surface wrap indefinitely, even on the fastest setting. Trying the same with a .stp file got me an error during import for surface wrap.

---

What are you referring to with "surface wrap" ?

---

@andrea.mazzoleniR2DKQ wrote:

...

I read that Ultimate CFD can accept mesh bodies as .unv .nas and .dat but i cannot find those options in the Fusion 360 exporter. 

...

---

.nas is a mesh file format originating from Nastran. It is a volumetric mesh format.

---

@andrea.mazzoleniR2DKQ wrote:

So I think I'm stuck with using solid bodies. Is there a way you know that I can convert a lattice in a solid body in Fusion?

---

Yes, I believe so. Fusion 360 has built-in tools (in the paid subscription version and perhaps also the educational license) that can re-mesh a triangulated  surface mesh into a quad mesh, then T-Spline and ultimately a NURBS surface, or BRep (Solid Body).

You'll find those in the Mesh tab in modify->Convert. InstantMeshes also does a fabulous job re-meshing triangulated mesh data into quad meshes.

[andrea.mazzoleniR2DKQ]

Thank you again!

For surface wrap I mean that it's the default option Autodesk CFD was giving me for importing .obj files. I honestly do not know what it involves.

And thank you for enlightening me about the option for mesh convertion. I am trying now and I'll update the post if it works. Unfortunately converting my meshes seems to be taking quite a lot of time due to the sheer amount of elements even of low mesh density, but that's a positive step for sure.

[laughingcreek]

can't help with the CFD bit.  bit it does seem like starting with a single disk instead of multiple bodies would help all around.  can you get enough of a gradient by playing with the parameters in the offset tab?

[andrea.mazzoleniR2DKQ]

Hello, thank you for the advice! I agree, doing the whole thing from single disk would be more streamlined. So far though I have two issues:

- I do not think I can locally regulate the size of the cell while keeping the porosity the same in that particular area. Since I would eventually need to print these designs (and I already tried to print a few, both with filament and resin printers), the solid beams of the cell would need to be of a certain size, and assuming the design you share in your screenshot is a disk of 10x4 mm I do not think that is any way for a printer to go down to that resolution. Do you think it would be possible to increase the cell size while keeping the same solidity locally?

- I may be doing something wrong here but i cant seem to reach the same results as you. Are you just using the offset of the lateral surface of the disk using those settings or are you combining with gradient along path solidity (which path would that be in that case? Because I can only select the circular edge of the disk) and nonuniform cell size?

Thank you again for the help 🙂

[karol.suchon]

Hello Andrea,

I am the CFD Specialist so I will try to give you some hints, probably you want to after creating the geometry check the pressure drop per the structure. 

Solving the entire geometry will be problematic, the amount of the mesh will be huge so the solution time will be extremely high. If we do not prepare a good geometry we could get problems with mesh which generates sharp edges so in the simulation we also get a higher pressure drop. 

So we need to clean up the geometry in Fusion, create the solid bodies and later transfer them to the CFD.

There is one good thing, you do not need to solve the entire geometry, to compute the pressure drop parameters we only need to know the data of a piece of the structure(there is some basic structure which is repeatable in the flow direction). After that, by computing the resistance model we could generate the pressure curve per entire structure.

So if you handle a problem with the geometry I could try to help you with the CFD part.

BR

Karol

[andrea.mazzoleniR2DKQ]

Hello Karol and thank you for the many hints. I am indeed still struggling with generating the solid bodies from the geometry as the computer I'm working on is too weak to handle the mesh conversion, even with low resolution meshes. Thankfully I will have access to a much more powerful machine soon enough, and in the meantime I will try to find a workaround to generate a mesh with even fewer elements.

I think the structure is not entirely symmetric, do you thing that even in this case I might just be able to compute half of the design in CFD? Or you are talking about another specific kind of mesh/geometry simplification?

Thank you again and have a good one.

Andrea

[karol.suchon]

You could try to calculate those pieces of geometry.

That should be enough to measure pressure drop.

Br 

Karol