Difficulty meshing a pipe tee

I’m trying to decompose the domain in a way so that way I can sweep but not sure if that’s the right strategy. I decompose at every angle but I get to my last piece and it’s not sweepable. Is there a better way to do this?

Thank you

Hi @d.weaver,
could you share your geometry, so that i can take a closer look?

Yeah, here is the general geometry that I’m trying to do. I also tried to decompose in a circular direction.
Thank you
tee_geometry.sat (18.7 KB)

Well, I had some time to burn while waiting for a debugging session to hit a breakpoint… so I gave it a shot… how does this look?

Colored by scaled Jacobian:

Here’s the journal file I used:

# PREPARE GEOMETRY
## IMPORT GEOMETRY
reset
import acis "tee_geometry.sat" nofreesurfaces heal attributes_on  separate_bodies 

## CUT IN HALF TO UTILIZE SYMMETRY
webcut volume 1 with plane normal to curve 5  fraction .5 from start 
delete volume 2

# PREPARE FOR MESHING
## WEBCUT FOR MESHABILITY
### SEPARATE CENTRAL BAR
webcut volume 1  with sheet extended from surface 13
### SEPARATE CENTRAL HOLE
webcut volume 1  with sheet extended from surface 14
### SEPARATE BACK SECTION FROM FRONT SECTION
webcut volume all with plane normal to curve 26  fraction .5 from start
### SPLIT THE 90-DEGREE BEND REGION
webcut volume 1 with plane vertex 18  vertex 43  vertex 7
### SPLIT THE CENTRAL BAR WITH THE BASE OF THE HOLE
webcut volume 6 3  with plane from surface 71

## COMPOSITE SURFACES FOR MESHABILITY
composite create surface 54 53 keep angle 15

## IMPRINT AND MERGE TO ENSURE CONTIGUOUS MESH
imprint all
merge all

# GENERATE MESH
## MESH FRONT SECTION
volume 1 scheme polyhedron
mesh vol 1

## MESH UPPER PORTION OF HOLE
surface 50 scheme map
volume 5  redistribute nodes off 
volume 5  scheme Sweep  source surface 50    target surface 52   sweep transform least squares 
volume 5  autosmooth target on  fixed imprints off  smart smooth off 

mesh vol 5 8

## MESH LOWER PORTION OF HOLE
volume 4  redistribute nodes off 
volume 4  scheme Sweep  source surface 69    target surface 67   sweep transform least squares 
volume 4  autosmooth target on  fixed imprints off  smart smooth off 

volume 7  redistribute nodes off 
volume 7  scheme Sweep  source surface 71    target surface 74   sweep transform least squares 
volume 7  autosmooth target on  fixed imprints off  smart smooth off 
surface 71 scheme map

mesh vol 4 7

## MESH REMAINING VOLUMES
mesh vol all

## RECOVER FULL-3D MODEL
## COPY-REFLECT GEOMETRY ON ORIGINAL CUT-PLANE
volume all copy reflect vertex 51 53
### MERGE THE GEOMETRY TO ENSURE CONTIGUOUS MESH
merge all

The key to getting this mesh was to not extend the cut that separates the 90-degree bend into the back-half of the model, and then to composite the remaining small surface with it’s larger neighbor (see the dashed green line):

image

Wow, this is great. Thank you. This helps me a lot moving forward.

Do you know of a different way to decompose this domain? I’m experiencing clustering in this location:
image

Hi @d.weaver,
what exactly are you simulating? Can you show what happens to the mesh?