Learning Forum

[bbcn]

Hi everyone,

I have created and parameterized the below geometry and created design points via Design of Experiments:

I have drawn the geometry using DesignModeler. Fluent analysis runs flawlessly. Below, you can see the created design points and the related errors:

I cannot set the created design points as current due to the error. However, when I manually change the dimensions according to the given design points on DesignModeler -after removing them from parameters because otherwise it will not let me-, the changed geometry looks okay. 

I have done this process with a simpler geometry and I did not encounter such an issue. I'm not sure if this occurs because my geometry contains flaws even though I have made sure the constraints are logical and fully created. Let me know if you require further info.

Thank you.

[Rob]

You've used a lot of edge sizing, once you've updated the geometry manually, how does the mesh look?

[bbcn]

Geometry looked okay for a few dimensional changes. However, when I removed all the parameters and modify them according to DP1 the geometry glows yellow:

Therefore, I was wrong to say that geometry looks okay. It does not when I modify all the parameters. Is this because constraints conflict?

[Rob]

Looks that way. Depending on how you link dimensions things can get tangled up. 

If I have 3 points arranged in a line, x, y & z. 

Case 1:  x=0, y=2 & z=4

Case 2: x=0, y=x+2 & z=y+2

 

Case 1: If I move x so x=2 then x & y are in the same place, so the line breaks. 

Case 2: If I move x so x=2 then y=4 and z=6 so the line is OK. 

 

I'd focus on how you defined all of the curves & dimensions for the "hollow" part and check you've not accidentally got some constraints or are trying to turn the surface inside out. It's easily done, and why parametric runs always need a lot more care when setting up. 

[bbcn]

You are right, thank you. Since all of my parameters change within vertical direction I removed horizontal and tangent constraints and I was able to update the geometry manually. Below is the new geometry and the mesh:

However, I'm not able to open the setup through WB. Fluent opens and tries the load the geometry but then closes and I recieve the following error:

My named selections remains healthy if that matters. Let me know if you require more info.

Thank you.

[Rob]

Not sure about that, I assume you've rebooted and renamed the .ansys folder(s) in %appdata%? 

Re the geometry. Is that a multibody part? The jump in cell size isn't ideal, so you may want to review pave as an option. 

[bbcn]

I'm sorry I did not get it quite well. I'm not sure what "pave" is, I can try to optimize the meshing though.

[Rob]

Pave is an unstructured format with mostly hex cells, it's more flexible than map, and should also mean you don't need so many edge mesh constraints. 

[bbcn]

Hi,

Sorry for the late response. I have ben trying different things to overcome the occuring errors. I have updated the geometry to be flexible for parametric changes and the mesh to have less sizing. The solution runs well. However, I still recieve the following errors:

The erros about the geometry stem from over-flexed dimensions, I checked it by setting the related DP's to current.

For instance, DP2 cannot be solved by the DoE, hence I tried to solve it by setting it as current:

And the solution converges and the flow finds the desired regime:

So, I do not understand why the update solution fails for DP2 when I use DoE to solve for it instead of manually solving it. What is the difference even though I have used the same setup.

 

Thank you.

[Rob]

What are you monitoring as the output parameter? The mesh could do with a bit more refinement, but otherwise looks OK. I'd also review how much free space you need, but that won't alter the result. 

[bbcn]

The output parameter is maxVal(Water.Mass.Flow)@inlet . It seems to be working for the DP0 because I can see the value when I write this expression and apply.

In addition, I have 630 gb free space, I don't think this is the issue. Let me know if you need more info.

Thanks again.

[Rob]

Sorry, free space in the outlet region, it's 2d so I'd hope the disk is big enough! 

That should work. I've never tried a parametric run with a transient flow before as they tend to take ages to run. It's always been easier to build several meshes & run all of the cases at once using a set-up file to speed up the Fluent part. 

[bbcn]

I have solved and obtained the result with DP2, I still do not understand why DoE cannot do it. I can see solving for each DP manually. Setting the related DP as current and doing the whole process one by one. This is the long way which optimization cannot be transferred to to response surface step. Hence, optimization can be done on another software with the data obtained here. Also, geometrical error that should be troubleshoot before the solution.

Thank you for all the responses. Let me know if you can come up with a solution, I will also do so. 

Regards,

Babacan