Implement fractures

[jirihnidek]

• [x] Get testing data from Jakub.
• [x] Implement reading of fractures from input files.
• [ ] Split volume to several volumes using fractures.

[jirihnidek]

Fractures can be loaded in BREP file format:

[jirihnidek]

Try to use this method section = BRepAlgoAPI_Section(surface, face): http://api.pythonocc.org/OCC.BRepAlgoAPI.html#OCC.BRepAlgoAPI.BRepAlgoAPI_Section for computing intersection between face and B-Spline surface.

Edit: no this method is not faster than boolean operations.

[jirihnidek]

It is possible to split volume into the two compatible volumes:

[jirihnidek]

It is possible to create volume defined by box, which is created from 7 faces:

It means that further splitting of volume is in theory possible.

[jbrezmorf]

Seems great. Try further splitting and b-spline faces. What about interior faces, i.e. face with some or whole boundary inside.

On Mon, Feb 13, 2017 at 12:29 PM, Jiri Hnidek notifications@github.com wrote:

It is possible to create volume defined by box, which is created from 7 faces:

[image: netgen-7_faces_cube] https://cloud.githubusercontent.com/assets/2057012/22881682/fc2faa30-f1e7-11e6-8cb6-7b787f0a2b3c.png

It means that further splitting of volume is in theory possible.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/GeoMop/bapprox/issues/10#issuecomment-279361469, or mute the thread https://github.com/notifications/unsubscribe-auth/AAT6PO74uxXNXq_sD2w54KOqlGTZDOFBks5rcD6jgaJpZM4IQsQQ .

[jirihnidek]

I just work on further splitting. Results you already see is created by code using b-spline faces. It seems flat, but everything is defined by b-spline faces. First NetGen screenshot has interior face.

Face partially dived into volume or face with whole boundary inside volume is something more complicated and I don't know how to solve it ATM.

[jbrezmorf]

What about subtracting a degenerate flat volume given by oppositely oriented faces of a single surface? While the hierarchy of subdivisions may be used to specify different volumes with different properties the subtracting may be used to specify compatible fracture network. Backup solution is adding artificial fractures to convert it to the previous case (subdivision).

On Mon, Feb 13, 2017 at 2:40 PM, Jiri Hnidek notifications@github.com wrote:

I just work on further splitting. Results you already see is created by code using b-spline faces. It seems flat, but everything is defined by b-spline faces. First NetGen screenshot has interior face.

Face partially dived into volume or face with whole boundary inside volume is something more complicated and I don't know how to solve it ATM.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/GeoMop/bapprox/issues/10#issuecomment-279395158, or mute the thread https://github.com/notifications/unsubscribe-auth/AAT6PKgIXQKvKNtN4N4wYgn39STFwa9Uks5rcF1UgaJpZM4IQsQQ .

[jirihnidek]

I can confirm that It is possible to create one compound containing two volumes with shared face. GMSH is not able to mesh volume, but NetGen Mesher creates compatible meshes for volumes and shells. Notice that volumes are visualized with different colors:

[jirihnidek]

It seems that splitting volume to more parts is complicated. The hard part is replacing one face with more faces:

https://www.opencascade.com/content/compound-made-2-solids-shared-face

At least it is described.

[jirihnidek]

Replacing face using BRepTools_ReShape() is just not good enough, when you try to "join" more then two shells/solids. Netgen cries, because there is duplicity in edges. It is interesting that netgen doesn't care, when there are two shells/solids.

[jirihnidek]

WIP: It is more complicated, than I thought. 4 new edges orthogonal to removed face have wrong direction. Good news is: I was able to manually edit generated BREP file and create correct BREP file (I replaced 4 - characters with 4 + characters). It was possible to mesh final BREP file in netgen:

I hope I will have working algorithm fixing wrong direction of edges tomorrow.

I did also some experiments with Glue* functions. It can remove common face of two solids, but it produces only one solid. Thus it does almost the same thing as sewing and it is not useful for our purpose.

[jirihnidek]

This is mindblowing. The only one problem was missing switch of orientation of vertex. In what universe does have single point orientation? Crazy library!

[jbrezmorf]

Congratulation. It's a kind of magic.

On Mon, Mar 20, 2017 at 4:09 PM, Jiri Hnidek notifications@github.com wrote:

This is mindblowing. The only one problem was missing switch of orientation of vertex. In what universe does single point orientation? Crazy library!

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/GeoMop/bapprox/issues/10#issuecomment-287789172, or mute the thread https://github.com/notifications/unsubscribe-auth/AAT6PM1PBDI5XUYHyg3lq5uqOVbbpc__ks5rnpa8gaJpZM4IQsQQ .

[jirihnidek]

Screenshot with three solids:

[jirihnidek]

I did some experiments with unusual shapes of cutting solids objects and results are OK. Meshing of final BREP files is OK too:

[jbrezmorf]

Looks good. What about more cuttings, e.g. more levels of recursive cuttings?

On Wed, Mar 22, 2017 at 4:58 PM, Jiri Hnidek notifications@github.com wrote:

I did some experiments with unusual shapes of cutting solids objects and results are OK. Meshing of final BREP files is OK too:

[image: bapprox-triangle] https://cloud.githubusercontent.com/assets/2057012/24207204/a17afb70-0f20-11e7-9954-f137608d343c.png [image: bapprox-odsek] https://cloud.githubusercontent.com/assets/2057012/24207212/a584fc16-0f20-11e7-96ce-f89819767122.png

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/GeoMop/bapprox/issues/10#issuecomment-288446674, or mute the thread https://github.com/notifications/unsubscribe-auth/AAT6PAjGNo6dqmxntegzfal9oEtwhdWSks5roUUhgaJpZM4IQsQQ .

[jirihnidek]

I will test more cutting today. Some more experiment from yesterday:

[jirihnidek]

Some non trivial examples with more levels of splitting. Due to second example I had to fix one bug.

Of course, I was able to do compatible meshing of all compounds.

[jirihnidek]

WIP: splitting of splitted solid works as well glue it together. What remains is glue following shapes with other two solids.

[jirihnidek]

Almost done. Screenshot from GMSH to show compatible meshing of solids:

[jirihnidek]

My task in this project is finished: