Jump to content
Product Design Forums

Treasurebox
Sign in to follow this  
Guest rudis

How To Model A Lobster Nose?

Recommended Posts

Guest rudis

Hi all, I'm a complete modeling n00b looking for opinions on what the best software is to model a lobster nose (similar to an insect antenna)... See attached pics of a crude physical model of the object - it consists of a cylindrical stalk with zig-zag rows of straight "hairs" along with rows of curved hairs. I have dimensions, angles, and orientations from data taken from actual animals. I need this model for export into a CFD (computational fluid dynamics) code that will simulate the flow of water around the nose. The nose is rigid and I only need to model the surface geometry.

 

Here are my needs:

 

A very simple export format that I can low-level input to the CFD code. Ideally, a text file of the many vertex coordinates making up a highly resolved surface mesh. Sounds like STL format might work?

 

The ability to export an arbitrarily smooth surface mesh. Since I will run the CFD code at many different refinement levels, it would be great if I had a "perfect" mathematical representation of the surface in my modeling program, and then could subsample a specified (directly or indirectly) number of surface points from that representation to create a mesh of variable accuracy. An alternative would be just a single, extremely high resolution mesh but it's hard to know what is high enough beforehand. Sounds like NURBS would be best, but can I convert a NURBS surface into a mesh of specified smoothness? What about constructive geometry?

 

I need to specify or at least know precise dimensions and angles of the model, but hopefully this is a given for any CAD-type software. It would also be great to quickly change something like the diameter of all the curved hairs, without having to alter each hair individually, since they are all the same.

 

I was thinking about freeware such as Blender but I'm not sure Blender is best for the mathematical precision I'd like? Is NURBS the way to go? I am a PhD student at a large university, I think I have access to Solidworks and I can probably persuade my adviser to get anything less than about $1000. The school might even have access or be able to obtain access to NX, Pro/E, etc. I'm aware no matter what, the learning curve will probably be steep, but as long as it's not Mt. Everest that's OK. So, with cost not being a major issue, does anyone have advice for a n00b? Thanks!

post-18711-1198816933.jpg

post-18711-1198816944.jpg

Share this post


Link to post
Share on other sites

You could do this in Solidworks, Rhino or ProE as a series of sweeps. The question is how precise do you want this model to be. If you are modelling a flexible part, then accuracy or precision are not critical. If you are attempting a biomimicry exercise then you can get away with a poly model in Blender, Maya Silo etc:

Share this post


Link to post
Share on other sites
Guest csven

I'd recommend a parametric CAD modeler. You get the mathematical precision and the ability to pretty simply change things like diameter for all the stalks either with a "feature" modify or using a relational system.

 

The whole array could be patterned (probably two patterns, from the looks of the thing). The base surface could be relatively complex as well; mimic'ing a lobster (could be modeled in a more freeform tool or scanned/cleaned, and imported).

 

Exporting at various mesh densities shouldn't be a big problem.

 

Could do this in Pro/E. Educational version should be relatively cheap.

Share this post


Link to post
Share on other sites
Guest rudis

Thanks for help!

 

Parel - It is true that given biological variation, I'm not looking to exactly duplicate Joe Lobster's antennule structure. However, I do need to have an exact description of the particular smooth surface I create. Although I will output what is essentially a mesh model, I don't want to change the smooth surface it approximates if I decide to refine the mesh (not sure if this is making sense, recall my n00b status). I think I'd like to represent the surface by mathematical splines and then sample this truly smooth surface to create an exported mesh, so that I know the real surface is always the same? Would this be a NURBS method?

 

Csven - In parametric modelers like Pro/E, is the smooth surface inherently described "exactly" then, instead of a "piecewise triangular" mesh (I'm from a math background...)? This is what I'd want, I think.

 

Sounds like Pro/E might be a good choice then. Any potential problems with Pro/E? Are any of these programs significantly easier than others to learn?

 

Thanks again, this is very helpful!

Share this post


Link to post
Share on other sites

Parametric models usually are in true mathematical spline but presented on screen as triangle meshes. Just so it runs faster on the computer. And when you convert to stl, it becomes a mesh.

 

Regarding learning parametric software. I found Rhino to be best as it is flexible in it allowing free 3d rotation/panning, quick command box. If you just looking to model simple stuff in the future, then Rhino just pwns it, not to mension it is cheap and simple as well.

Share this post


Link to post
Share on other sites

rhino is not parametric and hardly even makes parent child relationships. You get what you pay for you know. I am surprised so many designers use it.... I mean we teach it too but only because of the demand. At the end of the class we always show Pro/E and the designers are like "why did we not learn that instead" I just laugh.

 

In Pro/E you might be able to use a graph feature to randomize the curve growth. So your sweeps would appear random.

 

I think what I would do after thinking about it for a sec.

 

In Pro/E: Pattern a group of curve sweeps combo. Then unpattern the pattern so you can go in and edit each trajectory a little different on each one so it appears random. I don't think you can unpattern in SW tho. :)

Share this post


Link to post
Share on other sites
Guest csven
Csven - In parametric modelers like Pro/E, is the smooth surface inherently described "exactly" then, instead of a "piecewise triangular" mesh (I'm from a math background...)? This is what I'd want, I think.

 

Sounds like Pro/E might be a good choice then. Any potential problems with Pro/E? Are any of these programs significantly easier than others to learn?

 

Mechanical engineering CAD modelers are not mesh-based. You'll natively get a "smooth surface". They're also supposedly designed to provide output for FEA and CFD applications, since that's partly why they're used. Whether they play nicely with 3rd party applications for those needs, I don't know.

 

I've only played with SolidWorks, but I'd venture they're now pretty comparable. Pro/E probably has the advantage in surfacing and perhaps in file translation, but it's about a $1000 extra for a single, commercial seat. Educational versions for both are probably cheap though.

 

I don't think you'll have problems with Pro/E. I'd venture you could use SW as well. My recommendation is to test both of them and decide for yourself. There are other good CAD apps, most notably CATIA and UGS's offerings, but afaik they're pricey and more difficult to learn. And I don't know if they offer educational versions.

Share this post


Link to post
Share on other sites
Guest rudis

Thanks for the advice guys!

 

 

Regarding the random patterning design-engine mentioned, I don't actually want to do that anyway. In contrast, I only want two types of hairs, with all the short straight hairs identical and all the curved hairs identical. I do plan to change properties (like diameter or length) of all the little hairs, or all the curved hairs, later and note the effects in the CFD analysis. Perhaps this would be difficult in Rhino if it's not parametric, then.

 

I suppose I'm deciding on Pro/E versus Rhino versus Solidworks. Perhaps I'll try Pro/E first, it is actually just $100 for a student (including Interactive Surface Design module), even less than a student Rhino license (though I'm not personally paying anyway).

Share this post


Link to post
Share on other sites
Guest JD Mather

Students can download Autodesk Inventor Professional for free from http://engineersrule.org

For what you want to do Inventor and SolidWorks would be essentially identical.

Both have easy learning curves.

In my opinion Pro/E has a significantly steeper learning curve.

Share this post


Link to post
Share on other sites
Guest chris nobles

pattern the larger & smaller hairs along the body, then trim them with a surface to get the different lengths/dynamics you require...

 

you could also have the hairs as individual parts, and create a random hole/point pattern on the main body. Then put the hairs together in an assembly, and move the holes/points around as you wish.

Share this post


Link to post
Share on other sites
Guest csven
pattern the larger & smaller hairs along the body, then trim them with a surface to get the different lengths/dynamics you require...

 

you could also have the hairs as individual parts, and create a random hole/point pattern on the main body. Then put the hairs together in an assembly, and move the holes/points around as you wish.

 

I'd actually do a group pattern for each set; so two here.

 

- Start with a point on a curve on the surface.

- Build a local reference system for the point on that curve on that surface.

- Generate the antenna as solid geometry using that reference system (can go normal to surface, at some angle to it, or something else).

- Group all the features for constructing one and then pattern the entire construction to get an antenna array.

 

It'll get feature-heavy, but for this model that shouldn't be an issue. Can add or subtract elements by simply changing the pattern number. Can change the location of them. Can even change the base surface and re-map the pattern onto a new one.

 

Plus, each antenna could be set at an angle using a relational equation (again, either using the surface-referencing coordinate system, or another system... e.g. freestream vector). So for example, you could change angle of all the antenna relative to a flow direction and a flow velocity. Then export the geometry into the CFD and use the same vector for calculation.

 

Also, each antenna could derive shape/length information from a separate database, but it's been too long since I've done anything like that so I can't describe how to do that here.

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
Sign in to follow this  

×

Important Information

By using this site, you agree to our Terms of Use.