Adding Waterlines in Rhino

Hello Everyone

this would be my first post. Just found this forum and after reading so many fascinating comments feeling bit intimidated. I’m not a professional NA, just an enthusiast of all things Naval.

I’d like to ask for your advice on a project I’m working on in Rhino 5. It's a battleship hull. I traced all the station over hand drawn body plan, plus the main deck outline. Now I’d like to add waterlines. My stations are reasonably fair but when I use < curve through points > to get waterlines rhino returns, not surprisingly rather badly faired lines.

My idea is to take those waterlines, fair them, and than intersect with stations. But what is the most efficient way of accomplishing it in 3d...? If there is one. Appreciate any good ideas..

File:https://www.dropbox.com/s/bnnf7ki1enij604/Hull_Battleship_1.3dm

 

Turning a lines plan into a fair 3D model of a surface is very tricky. It's much harder than starting from scratch. There's no obvious way to do it and it's very, very easy to make something that is extremely unfair.

From the model you posted I'm a little confused what you mean when you talk about adding your waterlines. Are you tracing those as well, or are you intersecting the profile lines to get the points? Your Rhino file does not seem to have the points you're using.

These stations seem to loft pretty well with the loft command -- sometimes it doesn't work so well, but it seems to work relatively well for your hull. It's likely you can get most of the ship by tweaking the loft options a bit and tease the ends in by direct NURBS control point manipulation. I can't tell exactly what's going on at the stern, but that might be tricky.

Have you read up / watched much about hull modeling?

 

Fair your sections by using diagonals approximately perpendicular to the sections. Once your diagonals are reasonably fair and intersecting all your stations you can build your hull surface using networksrf command. You'll acheive the best result using the command on your profile lines (sheer and keel), your stations and a few diagonals in between. Depending on how important accuracy is for your project you can achieve a fairer surface by checking the rebuild usin xx control points box when using networksrf, at the expense of accuracy. If you have done a good job on fairing your diagonals and stations you should achieve a good end result even without rebuilding.

Good luck

 

I prefer normally to loft a Surface through section curves to get the hull shape. If you then add rectangular surfaces and use Curve from surface Intersection you get the resultant line. If this is where you have drawn the waterlines from the original lines drawings, you have a match.

The diagonals are useful to ensure fairness as well, and you must choose them judiciously. If you get odd results, you may have to rebuild and adjust the section curves. The tools are there, try and use the minimum number of points and complexity in the original curves. This gives the best result. It also helps sometimes to 2D draft the section curves and 'flat' parts and import into Rhino. Even 2D analysis and drafting can eliminate some future modelling problems. Better done sometimes in a 2D package as Rhino is a 3D modeller not a drafting package.

Network Surface command often gives a rather too complex result. You should have a minimal number of U and V isolines if you have a good clean surface. Note you should start at tight tolerance in the initial model say 0.001mm. A deviation of a mm in one section may cause an unfair surface, and often old paper drawings are not totally fair. One of the niceties of 3D design packages....

 

Thank you guys for your contributions. I appreciate it.

Normally in such case when I have stations only I would use about half of them for loft ( loose option ) with history hooked up. Than manipulating station’s control points massage the surface shape until it closely matches the real thing. It works okay for most cases. This is not working hulls after all, it’s for display only. But I’m at the stage when approximating is not enough. I’d like to take it a step higher. I’d like to recreate the hull lines, with stations, waterlines, buttocks and diagonals, all nice and fair first, and than build 3D hull. I guess I have to go back to school again...

I am considering doing it first in 2d. Do you think rhino drafting tools are not enough. Rhino can handle Bezier curves in form of Handle curves. I feel a lot more comfortable working in rhino than CorelDraw.

Mike, I don’t have waterlines here. The one I was taking about were arbitrarily crated by making flat planes intersecting all stations thus obtaining points. Then using <Curve through point> to draw waterlines. They not in the file because being ugly they got deleted...

 

If you have data ie drawings with the X and Y points for a Curve for all the stations/sections and a side elevation you have enough information to create the hull and all other lines ie buttock lines.

Corel Draw is not a proper 2D drafter, Autocad LT, FastCad, Intellicad etc are much more able to deal with 2D drafting. There are a load of decent 2D packages, even Illustrator is better than Corel if you are familiar with it. All of these handle bezier curves or NURB splines. Personally I would only enter the waterlines and buttocks as reference. The end surface will give almost indentical lines if it was drafted correctly. There is a hardly any shape that cannot be 2D drafted, but it may take (shape dependent) a lot of time either on a drawing board or 2D CAD system..... and yes I have done production drawings on tracing and film....

 

Rhino is perfectly adequate for your 2D needs. In fact it has an advantage over a pure 2D CAD program in that you can lay out your curves in 3D space. Looking at them in the top, front or side view, they will look like 2D drawings but you will be able to see them as a three dimensional wireframe in perspective view. This saves you huge amounts of time as you won't have to manually project your curve intersections. If you want to create a true 2D drawing from it you simply project the whole thing down to your construction plane using the ProjectToCPlane command.

 

Here's an example

 

This is very nicely executed drawing. Easy on the eye too. Thanks. Did you build it..?

I’m definitely sticking with rhino tools. I’ve tried CorelDraw and it’s a mess... Not for me. At least not in this particular case.

Here is my dilemma though, as someone coming to the game from CAD background. When I build 3d surfaces I’m always looking to accomplish my task using one span curves. This is the best way for perfectly smooth and tangent surfaces. Yet your curves in the body plan are not one span, and curvature analysis returns rather messy graphs. In the reality though I’m sure you can build perfectly fair boat out of your drawings.

This discordance between CAD and the reality what confuses me. And BTW I doubt it would be possible to construct body plan from one span curves. I’m not that good with math but I suspect it might be even impossible.
I guess I need a step back and another look at the problem.

 

Thank you.

No, I didn't build her. I'm a yacht design student, only about a quarter of the way through the course, this was part of an exercise.

Generally I find that there's no problem in keeping the stations very fair, I don't know if I did that here. The rest of the lines are a bit trickier. I would guess that these lines would be adequately fair for manual lofting, where the loftsman would correct the designers mistakes. As for printing full size patterns or CNC milling parts I don't know. A few millimeters off would still be within what most builders would achieve on the jig. Maybe someone with more experience could shed some light here :?:

/Hampus

 

I do not expect a loftsman, toolmaker or fabricator to correct my mistakes in the original drawings or models. They may request additional data like views or changes, maybe find missing dimensions or other things, but nothing major. Yes, some small things slip through even when externally checked but I have never encountered any major fault (and no extra cost), and at best no corrections required at all. Depends on complexity and number of parts etc etc. Sometimes the material has to be changed as the original did not make the grade or perform as predicted, stuff like that.

If you 3D model a shape, I would expect to get that shape. If you CNC cut it, why should it be out by more than the machining tolerance? A few mm on a small say 4 meter hull is not acceptable at all. If you use 2D drafting you can get fair lines too and diagonals which can be lofted fine. Would you accept 1.5mm out of fairness on your car bonnet? In fact total tolerance build up on cars (external dimensions/shape) is around +/- 1.5mm, pretty impressive, though more slack is allowed by the designers.

In other design fields, the advent of 3D modelling has led to a reduction of areas where a note and arrow saying 'Blend' occur.... ie where several edges with different radius values meet.

The work now is on the model and this contains the master data. You can have lots of control drawings etc even 2D drawings for 2 1/2D type objects like extrusions. So now the designer has to do all the loftsmans main work and ensure fairness in the original data file. Effectively most of the work that used to be done downstream of the design and drawing is now done by the designer or drafter/modeller. It takes a bit longer but saves time downstream. Just the way things have gone with 3D CAD and CAM. Rhino is as accurate as any other 32 bit double precision CAD system out there no matter how much you pay for it.

 

My hull project is on hold while professional NA is taking look at the body plan. Hopefully I can learn something from how he goes about doing stuff...

 

It's one of those unfortunate facts of reality that it's actually very difficult (to the point of being practically impossible in many cases) to create a fair 3D-modeled surface which exactly matches a given set of fair offsets.

Refining and refining in 2D is often wasted work if the goal is to create the 3D surface, as it gives you more constraints. It seems as though it should actually give you more hints, more help, etc., but in reality more constraints make fairness even harder to achieve. The details of local fairness can be handled with enough care and attention, which is pretty common in areas like graphic art, but not so common in industrial design.

It's common for traditionalists to realize this and criticize our modeling tools like NURBS as limiting the shapes which we can create. In reality, I don't think that's fair -- all the basic ideas are easily represented; the problem is with exact recreation.

Why is this so? I don't know. But this is one of those problems that seems like it ought to be easy but in reality is somewhat difficult.

 

There is no guarantee that an arbitrary set of individually fair intersecting curves (stations, waterlines, buttocks and diagonals) will produce a surface which is fair everywhere to the same extent as each of the individual curves is fair. And that is true whether the surface is created digitally or physically. How many physical surfaces are created which are fair and are verified to contain the input curves to the same level of precision and accuracy as is possible digitally in good software?

When a very fair physical surface based on a set of curves is needed the usual procedure is to create a surface using the curves. The surface is then faired, and if needed the close of the surface to the "input" curves is monitored to ensure it remains "close enough". The same is possible in the digital world - create a surface using curves and then fair the surface while checking that the surface remains "close enough" to the curves.

Working in the opposite direction if a surface is fair everywhere then all curves on the surface will also be fair.

 

That is a fair comment Mike. I suppose I have the benefit of working in both worlds so to speak. Having drafted on paper, through 2D computer drafting then on to 3D modelling. One approach that I have used, is to make full size models/patterns and send them to toolmakers along with the drawings. This works very well for certain situations. Scale ones would be more applicable to boats in general. Even so, full size cardboard/foamcore interior mock ups are pretty useful and fast.

Sometimes with 3D it is a good idea to CNC/3d print a complex or critical area only to validate that. Done it with some furniture - injection moulded one piece chairs for example. Not forgetting the reverse engineering path, an IGES format point scan of surfaces, which can be re engineered in the CAD system. Hand grips for certain applications such as helicopter controls have been done this way. I have used it too, as long as you can simplify the point count enough and still keep integrity of the form it is OK. So in the sense of industrial design missing that fairness I would probably disagree. I would agree that you have to be a little 'looser' about some of the 2D stuff (except essentials) when working in 3D. Actually it is about being a little more open minded to what can be achieved and not being controlled by an algorithm.

I still use a fairing board and block on c/board and rudder blades to get maximum fairness in reality as very small deviations can be felt by hand. Also on any plugs (bucks) which I tend to spray gloss black and polish...that last 0.05mm 'roughness' goes pretty fast then.