Scan-and-Solve for Rhino

Simulate Early, Simulate Often... In Rhino

# Scan & Solve for surface and poly-surface.

Hello,
I did some tests on a roof of a board of a yacht of 12 tons.
During the construction of the glass-resin shell I found it very convenient to draw a pin surface, to which I added the skeleton cross.
I tried to use Scan & Solve, but could not calculate the stress, because it was not a solid.
Proposal: only for the structures 'leaf', it could allow Scan & Solve to calculate the stress, for instance by including the thickness of the shell (in mm or inch)?
This would save time in modeling the object, because it would be enough to draw a single surface (surface or poly), avoiding the complications of modeling of a solid.

Yours sincerely
Germano Pecoraro - Italy

Views: 633

### Replies to This Discussion

I second this. A thin shell model should be able to work with a surface and a given thickness, since SnS works with just querying points inside a volume, the volume can be defined by a distancefield around the surface.
Yes,
I solved this problem in the following mephod:
1 - design of the roof surface,
2 - from the Command window "means land", I selected "offset surfaces,"
3 - I selzionato joystick in the "solid closed",
4 - I typed the thickness;
so I got a solid closed.

After this long, but simple preparation I adoperatp SnS.
Yes, but unfortunately this doesn't work for all kinds of surfaces or meshes... but the distance field approach could probably be implemented inside SnS.
Yes I agree with you, Rhino does not always respond properly to these commands.
For those who know how to use, you can use "Grosshopper", or some plug in reverseenginnering ...
Another solution is to derive simplified models from the real.
I think SnS in the near future you will have to implement.
Germano and Kristopher,  thank you for you very insightful comments.  You are both correct of course.  Offsetting is one of the more sensitive operations in most boundary-based geometric modeling systems -- such as Rhino.  Bug issues aside,  the difficulty has to do with surface properties.   When the offset distance exceeds the radius of curvature, the offset surface becomes self-intersecting (in a fairly complex way).   When the surface is not smooth and has "seams", the offset direction may not be well defined.  Etc.   So there is a good reason why Rhino has problems with offsets.     You are also correct that "thickening" by the distance is always well defined (though this may not always produce the same results as offsetting),  and that S&S should be able to compute on such model directly.    We have added this capability on our "to do" list.   It is not as trivial as it may sound, and may slow down the solution procedure.     Also,   which surfaces/faces  do you restrain or load, if the model is defined implicitly by an offset?

I'm happy to hear it is on the TODO list.

The scenario I'm in, and which I assume is a common one, is to model for example a thin shell, hence the offset must be smaller than the radius of curvature anyway, right?

There would then be the possibility to specify an offset thickness as well as direction for which the anisotropic direction field would be calculated.

Presumably, the forces would then be specified by boundary edges? I realize that it differs from the way SnS works at the moment, but on the other hand since Rhino will not change it's handling of solids in the near future, it would still be the fastest way to get results.

Ironically, I'm in the situation where I turn to mesh-based methods (OASYS) where you can specify a surface and its thickness as a variable. This presumably goes straight into the thin-shell equations.

Perhaps there is a way of doing mesh-less methods in 2D on the surface, and then specifying the thickness as a parameter for the equations also in SnS?

The offset SHOULD be smaller than the radius of curvature,  but sometimes the user is not aware of the imperfections in the surface due to geometric construction methods.    And things will get much more complicated if the offset direction is not normal to the boundary.

We will probably add point and edge restraints to S&S soon.   We have not decided on point/edge loads. While all of these are used widely in FEA and are convenient abstractions,  they are not really physical (you can't really restrain a point or have a point load), and the users must know how to interpret the results.

Shell (2d curved surface) elements are fundamentally different from 3D solid elements, and should give you more accurate results for very thin-shell objects ... as long as you do not mind meshing of course.   S&S can handle thin-shell models of course,  but high accuracy would require a lot of elements.    We have a number of ideas on how to improve handling of such models in the future,  but it will take some time (and resources!).

Yes,
in the case of the roof of a vehicle or a boat, you can lock the edges of connectionbetween the two surfaces, while the load (vector x, y, z) is applied on the outer surface.
This model is not very far from reality.
Complicating the model, you can draw the solid roof deck attachment, so as to lock theroof at these points.
I take this occasion, however, indicating that it would be appropriate to include the option to not block the piece and keep it in balance with the only forces acting.
I wonder how I can use SnS for solving the simple static exercises type supported beam-supported, cantilever beam, with various load configurations (concentrate,distribution, moments, etc.)?
The passage "block"is not always necessary.
For example, a chair may act four forces under the feet (ground) and a pressure on theseat (per person).
SnS I have simply changed the style of design with Rhino and I'll explain.
In the past when I learned to model, first with AutoCAD, 3DStudio, etc. and then the pair-Rhino Flamingo, 3D modeling was designed to call them for "theater" to do the rendering.
For several months now, SnS has made me evolve approach to design, say "a pattern of substance (thickness).
In summary, the point now when I use Rhino to get two results:
1 - a form rendered with Flamingo and Bongo, the small errors of rendering them correct with Corel Draw,
2 - closed volumes to be checked with SnS.
I'm used to this way and I have not big problems, even SnS does go beyond mere "concept"of the item.