fusij |
March 19, 2016 07:42 |
Case diverging - mesh quality or something else
Hello,
I am trying to solve a case that involves a fairly complex geometry (riverbed with additional structure). I received the riverbed in .stl format and it was of a rather low quality so I used SolidWorks (SW) to clean it up and model the additional structure that was needed for the geometry model. The cleaning of the riverbed model was basically just deleting triangles that involved sharp corners and unrealistic representation of the real surface, using "Filled Boundary" in SW to introduce smoother surfaces.
I export the surfaces to separate .stl files and then use "cat" or "surfaceAdd" to concatenate the files together in a one .stl file with regions. This is the output of my surfaceCheck:
Code:
Reading surface from "total.stl" ...
Statistics:
Triangles : 8365
Vertices : 4497
Bounding Box : (1e-06 1e-06 100) (122.14 249.327 114)
Region Size
------ ----
inlet 2256
walli 4944
wall-river 50
wall-fleyti 253
atmo-top 240
atmo-horiz 119
outlet-river 2
outlet-fleyti 501
Surface has no illegal triangles.
Triangle quality (equilateral=1, collapsed=0):
0 .. 0.05 : 0.161626
0.05 .. 0.1 : 0.0676629
0.1 .. 0.15 : 0.048416
0.15 .. 0.2 : 0.0420801
0.2 .. 0.25 : 0.0369396
0.25 .. 0.3 : 0.0425583
0.3 .. 0.35 : 0.0361028
0.35 .. 0.4 : 0.032636
0.4 .. 0.45 : 0.0276151
0.45 .. 0.5 : 0.0310819
0.5 .. 0.55 : 0.0213987
0.55 .. 0.6 : 0.0218769
0.6 .. 0.65 : 0.0237896
0.65 .. 0.7 : 0.0518828
0.7 .. 0.75 : 0.128512
0.75 .. 0.8 : 0.154094
0.8 .. 0.85 : 0.0228332
0.85 .. 0.9 : 0.01841
0.9 .. 0.95 : 0.0160191
0.95 .. 1 : 0.014465
min 2.13856e-11 for triangle 5393
max 0.999792 for triangle 2342
Edges:
min 0.000280529 for edge 10494 points (96.3715 69.6728 105)(96.3717 69.6726 105)
max 90.1672 for edge 11019 points (11.8551 135.061 114)(98.2565 109.275 114)
Checking for points less than 1e-6 of bounding box ((122.14 249.327 14) metre) apart.
Found 0 nearby points.
Surface is not closed since not all edges connected to two faces:
connected to one face : 902
connected to >2 faces : 11
Conflicting face labels:935
Dumping conflicting face labels to "problemFaces"
Paste this into the input for surfaceSubset
Number of unconnected parts : 9
Splitting surface into parts ...
Writing zoning to "zone_total.vtk"...
writing part 0 size 7562 to "total_0.obj"
writing part 1 size 31 to "total_1.obj"
writing part 2 size 72 to "total_2.obj"
writing part 3 size 53 to "total_3.obj"
writing part 4 size 25 to "total_4.obj"
writing part 5 size 93 to "total_5.obj"
writing part 6 size 15 to "total_6.obj"
writing part 7 size 13 to "total_7.obj"
writing part 8 size 501 to "total_8.obj"
Number of zones (connected area with consistent normal) : 22
More than one normal orientation.
End
The "problemFaces" where edges are connected to either one or >2 faces is where the SW surfaces are adjacent to the original riverbed surface or other SW surfaces. This is because SW exports the surfaces as separate surfaces, so the surfaceMesh of the surfaces do not conform to each other. But I am not sure how this really affects the meshing procedure since the geometry is watertight. An example of this is shown below:
https://dl.dropbox.com/s/9p4tjwosg8y...small.png?dl=0
How critical are the "problemFaces" from the surfaceCheck output for the meshing process when originating from the situation I show in the image above, that is when adjacent faces do not share the same edge?
https://www.dropbox.com/s/c27cr34k21...metry.png?dl=0
|