Boundary Layer Does Not Form - Except When Default Settings Used

Project Support
61

Great! Notepad++ is the best for manual concatenations of text files. By the way: MRF zone should be really in the vicinity of the propeller. Do not make it too big.

If it is a solution for your problem, please indicate it were appropriate.

Cheers,

Retsam

1 Like
62

Hey guys!

I have posted 2 scripts a while ago to merge STLā€™s that you can find here: Merging STL files - #2 by jousefm

Best,

Jousef

3 Likes
63

Hi all, so yes, this post is to confirm that using STL exclusively seems to result in higher prism layer formation! Thanks again @Retsam

Details:

  1. Objective of my current exploration was to conduct MRF simulation on propeller to see how well SimScale predicts the thrust and torque characteristics. Before I can do that, a decent simulation mesh needs to be generated and this post was a result of problems encountered at this stage.

  2. As you would read above, there were several problems along the way, perhaps due to my inexperience with OpenFoam and SnappyHex. All along, I was working with IGES files and the best prism layer formation was about ~70% I would say (optimistically). I have tried both y+=1 and y+=100 types, 5 layers to >10 layers etc since the cause of the problems were not clear.
    image

  3. Y+=1 mesh will be something like:
    Layer1------3.40E-06m
    Layer2------4.42E-06m
    Layer3------5.75E-06m
    Layer4------7.47E-06m
    Layer5------9.71E-06m
    Layer6------1.26E-05m
    Layer7------1.64E-05m
    Layer8------2.13E-05m
    Layer9------2.77E-05m
    Layer10----3.61E-05m
    Layer11----4.69E-05m
    Layer12----6.09E-05m
    Layer13----7.92E-05m
    Layer14----1.03E-04m
    Layer15----1.34E-04m
    Layer16----1.74E-04m

  4. After almost giving up, and with very few options left (havenā€™t tried cfmesh as I donā€™t think my laptop has enough RAM), I just thought of trying STL. But I was not able to import 2 solids in 1 file if I directly output from Autodesk Fusion360. Nevertheless, the prism layer formation was >90% for meshing without the MRF volume.

solid_0_stl_solid 293784 10 0.000309 93.1

Aside: Using STEP file, the formation was ~70+%.

A1_I1541_B2_TE141_A1_I1541_B2_TE43 193864 7.83 0.000262 78.9
A1_I1541_B2_TE141_A1_I1541_B2_TE50 277 0.17 1.3e-05 3.93
A1_I1541_B2_TE141_A1_I1541_B2_TE53 2524 1.19 4.62e-05 13.9
A1_I1541_B2_TE141_A1_I1541_B2_TE124 193794 7.83 0.000262 79
A1_I1541_B2_TE141_A1_I1541_B2_TE131 277 0.17 1.3e-05 3.93
A1_I1541_B2_TE141_A1_I1541_B2_TE134 2523 1.19 4.62e-05 13.9

  1. Luckily @Retsam replied my distress call on STL and I managed to combine 2 STL files by exporting them in ASCII format and copying and pasting one STL file to another using Notepad++. Great idea! You will have something like that:

solid PROPELLER [this is the 1st STL file]
facet normal -8.195655e-01 5.663210e-01 8.713685e-02
outer loop
vertex -1.877391e+01 1.240122e+01 8.000000e+00
vertex -1.874856e+01 1.238448e+01 8.347296e+00
vertex -1.796457e+01 1.354920e+01 8.151286e+00
endloop
.
.
.
endsolid
solid MRFVOLUME [this is the pasted-in 2nd STL file]
.
.
.
endsolid

And then save as ā€œAll Filesā€ with the .STL extension. Then proceed to import into SimScale.

  1. With the MRF volume inside too, it works! Again, >90% of the prism layers generated:

solid_0_PROPELLER 297213 9.97 0.000308 92.7

I used very fine tessellation on the propeller (and reminder to self to scale down by 0.001 to work in m).

image
image1366Ɨ768 205 KB

Now, I can finally run the simulation knowing that I have a decent mesh. Hopefully.

Thanks @DaleKramer, @Get_Barried and @Retsam for the help.

3 Likes
64

Your welcome! However credits should go to @Get_Barried, @anirudh2821998 as topic is about meshing. I simply took a look at STL file handling, which is nothing compared to meshing complexity! :man_student:

65

Different geometry, same problem (if not worse). Now, even the high y+ mesh PL is problematic (0.2% formed), following all the previous settings and tricks.

SimScale sales contacted me to ask about converting to Professional planā€¦ no news heard after I shared the pain points.

66

Problematic indeed.

Any significant differences between the new geometry and original working geometry?

67

Jousef, I wanted to confirm, exporting as STL instead of importing directly fom Onshape is still the recommended path? It would be great if Simscale did this by itself :wink:

68

Thatā€™s IMHO one of the best ways to go. What would you recommend, can you be a bit more explicit?

Feel free to add your request in this section: Feedback Section

Best,

Jousef