Abaqus - mesh

[mechabaqususer]

Hello guys,

I have a question about meshing cylinder parts in Abaqus. I know, that for these problems, commonly use practise is next: use of partitions, core is meshed with sweep meshing technique and outside cells with structured technique (seeds are concentrated near the wall). But this is for cylinder with constant diameter. What if we have an example of cylinder with diameter that vary (change) - cylinder with narrowing. How to mesh is such example to provide quality mesh?

My english isn't perfect, so i hope you will understand my question Thank you for all answers!

 

[mechabaqususer]

Anyone?

 

[corus]

Never heard of meshing the core using a swept mesh and structured on the outside. If you need to have a finer mesh near the surface, say for thermal transients, then just bias the mesh to one or both surfaces. For a cylinder in which the diameter reduces then there is no set technique. You just have to look at the overall geometry and partition and bias lines appropriately. Perhaps a picture of the geometry might generate more responses for your particular case.

 

[mechabaqususer]

Hm...about meshing the core using a swept mesh and structured on the outside - my opinion is, that this is one possible option. For example, such technique is also used in Abaqus Documentation, Getting Started with Abaqus: Interactive Edition, Flow through bent tube. This is in version 6.12. I think, that if we don't use this technique, we won't be able to use hexaedral elements. Am I wrong?

Such geometry for example:

https://www.google.si/search?q=sten...=nb7d7XNNq4SZmM:;Pe3bK1xwOn6ByM;http%3A%2F%2F

http://www.aquaphoenix.com%2Flectur...%2Flectures%2Fideas-flow%2Fpage2.html;293;142

Thank you for answer

 

[corus]

For the narrowing tube in the example I'd have used the same number of elements in the narrow section as the wide section so that the elements 'flowed' through the narrow section. This would give you a naturally higher mesh density in the narrow section where you'd expect higher stresses to occur. Alternatively, use natural points in the geometry to form partition lines, for instance at tangent points on the outer surface. The partitioned regions would ideally form rectangular regions, as best you can.

 

[mechabaqususer]

I've used partitions: i defined cutting planes, partition face-sketch (offset curves for narrowing part - i create outer and inner (core) section)...i've tried to use swept/structured technique, as mentioned before. But i still have problems with some areas - where the 'curvature' is large. Problem is, that i don't have experience with meshing such geometry, so i'm not sure how appropriate mesh must look.

If i don't use partitions and just use seed part and mesh part i succeed. But i think that this is not a suitable approach.

 

[corus]

Where the curvature is large (presumably meaning that the radius is small) partition the geometry using radial lines. The best mesh would have element shapes that followed contours of the expected stress distribution. For small radii it's best to partition that region off from the main geometry, say at one radial distance away, then a fine mesh can be imposed in that region. Partition the remainder of the geometry using larger blocks and try and blend in the mesh density using a bias on lines so that there is a gradual change in mesh density between regions.