[lachie]

hi everyone.
just as you see in the picture, i want to generate grid with ALL structured cells in this box, so the biggest challenge to me is the structured grid generation around the dimples. i have tried many times but failed in the end. So please gvie me some addvice!
i use Pointwise V16.02R3.

[jchawner]

Hello Xu Zhu.

The complicated answer to your question is that there many structured grid topologies that can be applied to the dimples in your attached image.

To choose one topology, you need to know whether the perimeter of each dimpled must be modeled accurately from a geometric standpoint. In other words, must there be a circular grid line around the edge of each dimple?

If the answer is no, make a simple H grid on the "floor" of your grid using as many points as you feel necessary. Then project that grid onto the database geometry that includes the dimples. When you do this you'll probably find that you'll want to cluster points in the region of the dimples and that's easily done with the subconnector feature in the Distribute command.

On the other hand, if you absolutely must have a well defined edge for each dimple, you are then driving to put an H grid inside each dimple or an O-H topology (an O grid at the center of each dimple surrounded by 4 H grids). With a grid in each dimple, you then build topology around and between the dimples out to the rectangular boundaries of the floor.

If you send me the dimensions of the floor and dimples or if you can send me your geometry file I might be able to illustrate a few techniques for you.

Otherwise, I suggest you contact our Tech Support folks directly and they'll help you get started.

Best Regards

[lachie]

Originally Posted by jchawner

Hello Xu Zhu.

The complicated answer to your question is that there many structured grid topologies that can be applied to the dimples in your attached image.

To choose one topology, you need to know whether the perimeter of each dimpled must be modeled accurately from a geometric standpoint. In other words, must there be a circular grid line around the edge of each dimple?

If the answer is no, make a simple H grid on the "floor" of your grid using as many points as you feel necessary. Then project that grid onto the database geometry that includes the dimples. When you do this you'll probably find that you'll want to cluster points in the region of the dimples and that's easily done with the subconnector feature in the Distribute command.

On the other hand, if you absolutely must have a well defined edge for each dimple, you are then driving to put an H grid inside each dimple or an O-H topology (an O grid at the center of each dimple surrounded by 4 H grids). With a grid in each dimple, you then build topology around and between the dimples out to the rectangular boundaries of the floor.

If you send me the dimensions of the floor and dimples or if you can send me your geometry file I might be able to illustrate a few techniques for you.

Otherwise, I suggest you contact our Tech Support folks directly and they'll help you get started.

Best Regards

thanks very much! John Chawner. Helps me lot.
ok, i will send you the geometry file to your email, please give me more details!

[jchawner]

Xu Zhu:

Thanks for sharing the IGES file of a dimpled wing. However, it has no surfaces - only curves. So the first thing I had to do was create surfaces onto which the grid could be projected to give it the right shape. I used simple Coons patches for that. I made one for the wing and four for each dimple (because each dimple was four wedges of curves).

The first picture (simple-project2.png) shows the simplest way to put a structured grid on this shape. Make a single domain on the wing surface then project it (using the Linear method) onto the dimple surfaces only. You can probably also see the drawback - the edges of the dimples are not reproduced cleanly. This effect can be minimized by increasing the number of points.

The second picture (dimple-topos.png) shows how to get a crisp edge around each dimple. You have to mesh each dimple with it's own grid topology and I've shown three ways to do that. The lower left what's called a butterfly or O-H topology - 4 H-grids wrapped around an O-grid. Upper right is a simple O-grid but it's drawback is a pole (line singularity) at the center. Finally, upper left is an H-grid. This is simplest to generate but you have to decide whether the cells in the corners meet whatever mesh quality criteria your solver requires.

The most difficult part of putting a mesh in each dimple is connecting them all with a structured grid topology on the wing surface itself. I think you can see how the grid lines would have to run from leading edge to trailing edge and connect all the dimples. That's just a little connector drawing, domain creating, then copying and pasting. You'd finish up with a little run of the solver with projection onto the wing shape and you're done.

I hope this helps.

[lachie]

Originally Posted by jchawner

Xu Zhu:

Thanks for sharing the IGES file of a dimpled wing. However, it has no surfaces - only curves. So the first thing I had to do was create surfaces onto which the grid could be projected to give it the right shape. I used simple Coons patches for that. I made one for the wing and four for each dimple (because each dimple was four wedges of curves).

The first picture (simple-project2.png) shows the simplest way to put a structured grid on this shape. Make a single domain on the wing surface then project it (using the Linear method) onto the dimple surfaces only. You can probably also see the drawback - the edges of the dimples are not reproduced cleanly. This effect can be minimized by increasing the number of points.

The second picture (dimple-topos.png) shows how to get a crisp edge around each dimple. You have to mesh each dimple with it's own grid topology and I've shown three ways to do that. The lower left what's called a butterfly or O-H topology - 4 H-grids wrapped around an O-grid. Upper right is a simple O-grid but it's drawback is a pole (line singularity) at the center. Finally, upper left is an H-grid. This is simplest to generate but you have to decide whether the cells in the corners meet whatever mesh quality criteria your solver requires.

The most difficult part of putting a mesh in each dimple is connecting them all with a structured grid topology on the wing surface itself. I think you can see how the grid lines would have to run from leading edge to trailing edge and connect all the dimples. That's just a little connector drawing, domain creating, then copying and pasting. You'd finish up with a little run of the solver with projection onto the wing shape and you're done.

I hope this helps.

Dear John Chawner
Thank you very much!! As you said, the most difficult is to connect all the dimples with a structured grid tolology on the wing surface, i will try to make it.
Thanks again! For your so detailed answer!
BTW, today i want to define a line( actually it is a parafoil) surrounded by unstructured grid the WALL boundary condition, but it seems the pointwise V16 can do this for me. i want to know if i can define a line the WALL boundary condition using pointwise V16.

[jchawner]

Xu Zhu:

I don't understand your question about "define a line the WALL boundary condition." You can create lines. You can apply wall boundary conditions. But I'm not exactly certain what you're asking. Also, you mention "unstructured" grid but we have been discussion structured grids.

[lachie]

Originally Posted by jchawner

Xu Zhu:

I don't understand your question about "define a line the WALL boundary condition." You can create lines. You can apply wall boundary conditions. But I'm not exactly certain what you're asking. Also, you mention "unstructured" grid but we have been discussion structured grids.

The structured gird generation around the dimple is already solved through your kind help. This is an another problem, . As you see in the picture, this is a parafoil section with a leading edge cut, i want to define it the WALL boundary condition, but it seems i can not do it using pointwise V16? Thanks!

[jchawner]

If you're asking whether you can set a flow solver boundary condition on the box outlined in light blue color which appears to me to be a connection to an outer domain, the answer is no - the version of Pointwise you have cannot over-write an interblock connection with a boundary condition. (You would have to repair this manually in the boundary condition file exported for your solver.)

However, Pointwise Version 16.04 will be released soon and it does let you overwrite a connection with a BC.

Is that your question?

[lachie]

Originally Posted by jchawner

If you're asking whether you can set a flow solver boundary condition on the box outlined in light blue color which appears to me to be a connection to an outer domain, the answer is no - the version of Pointwise you have cannot over-write an interblock connection with a boundary condition. (You would have to repair this manually in the boundary condition file exported for your solver.)

However, Pointwise Version 16.04 will be released soon and it does let you overwrite a connection with a BC.

Is that your question?

That's it! ' the version of Pointwise you have cannot over-write an interblock connection with a boundary condition'. i just want to make clear of it.
Thanks very much! You are really a very kind man.
Looking forward to using V 16.04!