I am trying to create a model for a book illustrations showing the layout of hull plating. The larger objective is to show how the physical hull can deviate significantly from the molded lines. I have surfaces for the molded lines and want to go to the physical.
In the attached example, I have started with the simplest case. I have a strake that sit on the molded line. I split the strake from the molded surface at the sight edges and broke that down into thickness zones (20# to 35#). I do not care about adjacent, individual plates of the same thickness. because you cannot see the divisions.
I did offsetsrf for individual thickness zones. Then I try to boolean union. That brings me down to the purple and blue polysrfs that will not come together due to small gaps the different offset directions.
I think created a third poly-surface (Red) and did offsetsrf at the minimum thickness (20#). Even that cannot be added to the purple or blue polysrfs
I am curious what approach works for this problem? It only gets harder after this strake because the thicknesses increase to 80#, there are laps, joggle, and lap reversals.
Hi Miano, In my experience going from a surface “molded lines” model to solid where all material thicknesses are represented, is best done using a dedicated solid modeler such as SolidWorks.
Shapr3D, Onshape or Plasticity may do the trick for Mac users. All these programs have implemented the Parasolid modeling kernel from Siemens (the one Solidworks is using as well).
Since your goal is visual rather than engineering accuracy, I wonder if the modeling process would be smoother if you did subtractions rather than additions.
Building a station of plating to the maximum hull thickness, and then creating offsets at the thickness of the plate you want to ‘cut off’ to create the thinner sections should always generate a significant intersection for Rhino to work with.
It should also leave no gaps between plates, visually or computationally, when it comes time to do a nice render.
The massive casting shown here is depicted in most books as framing. A problem here is that the aft edge and the cross section at the shaft curve follow the molded lines The rest of the casting has 30# added to the thickness, particularly at the top, bottom, and forward edges. Part of the exercise would be to illustrate that the molded lines are reference lines and not the inside of the plating as often claimed.
Another feature I would like to illustrate is at the bottom. The 45# “C” strake running along the bottom rests on top of a 35# A strake sitting on the molded line. The bottom of the C strake is then 2" below the molded line, which is enough to be quite visible this scale. However, the aft end of the C strake is at the molded line. The actual curve at the bottom is forward of where the molded lines have it shown here.
The E strake forward is also 45# resting on 35# plate at the ML. The keel, C, and E strakes sticking out 2" below the molded line provide the resting places for keel blocks when in drydock.
There are a lot of other oddities visible worth illustrating. The G strake at the right sits on the molded line while at the left it is lapped over the L strake. The upper edge of the G strake reverses the lap.
There are a lot of lapped joints that come to a T (e.g. the C1–D joint joint) where the lap tapers to zero and a rectangular plate is riveted over the taper.
Yet another detail with illustrating is the C strake is connect to the C1 strake using a flange at the right (welded at the left). For reasons that escape reason, the designers used riveting until the space got too tight so they switched to welding. Except that that laps of less than 4 1/2" were welded, so most of the lap joints shown in this excerpt were weld at the inside and outside faces.
Here are some thoughts at a hopefully close approximation- it should work well if the profile items are drawn to the moulded/reference surface but on rounded sections I think the overall positioning error could approach the plate thickness if you’ve only got the one drawing and are representing local sections that approach horizontal.
I would expect that originally there were very detailed structural drawings for construction.
If all you’ve got is that profile drawing, I guess you could make a copy of the hull surface so you have the original.
Then, construct polygons representing the outlines of each plate section at max beam plus a foot. Then, extrude those profiles in through the hull, perhaps to centerline. Use each of those to cut a piece out of the hull surface representing that plate section.
At that point, depending on the precision you need and local complexity of the surface, you could extrude the surface in the best approximation of a local normal to the surface (if it’s simple and pretty flat like the rectangular doublers) or offset the surface by plate thickness and then try to get a solid bent plate from the pair of surfaces.
As you’re cutting the sections, you could also consider just color coding the pieces by thickness and overlaps. Is the ultimate goal to cut transverse sections to illustrate, or to use 3D visualization from various angles?
I think it would be hard to estimate in advance how closely the resulting pieces would fit together. I think you could get a section that’s not bad as a sketch to illustrate concepts: looks about right, no ‘light shining through the holes’.
As you do this, I would strongly recommend driving it from a script that reads a simple text file containing whatever data you pull from the drawing so that you can easily experiment with approaches without having to do minutes/hours of rework each time.
For educational purposes here, there is both detail and lack of detail. I suspect some of the lack of detail was made up for with tribal knowledge that has been long lost.
You get:
Molded data points in X/Y/Z coordinates.
Radii at the keels to correct molded sharp edges as being rounded.
Sight edges showing the molded location for the edge of a plate in X/Y/Z coordinates
Expansion diagrams showing how the plates should be cut. In my case here, these are in bad shape. 5. Lapping diagrams. These show how the strakes are overlapped. However, they show the plates in a vertical straight line.
You get from the plan a mixture of mathematically correct placements where one thing connects to another and artistic placements.
If A plate sits on molded lines or all four of its edges are offset the same distance from the molded line you can get its exact shape from the plans. If a plate laps or has different offsets from the molded line, things end up inexact. One oddity is that, at the ends, the outer edges of the plates go to the molded line. The bow profile is both a molded and actual curve.
I have created:
A molded hull shell.
4-view drawings of the plating like the excerpt shown (the bottom and side views are 6-feet long)
Excel tables for all the tabular data.
As you suggest, I have been looking at scripts recently. I have the feeling that will be the ultimate solution—if there is one. I have done a few in the past but I have done a lot of programming.
What I really need is a primitive for line bending that move each part of the line relative to its distance from the fixed point in the bend.
This is an illustration I did modifying the original by (1) making it fit the hull from, rather than being vertical (2) shows both sides (3) and exaggerates plate thickness.
Is there any callout on the drawings for bend radii or anything like that?
Do you know what type of math might model those bends in your sections? The molded lines would have been laid out with splines, so basically 3rd order curves clamped or not at the ends.
And in between stations to model the shape? It sounds from your earlier question that you might have been attempting it as a ‘simple’ offset from molded for each plate.
Maybe it’s nurbs surfaced solids of degree no more than 3 of given thickness and known tangencies at the edges?
I have a feeling that some of the really bright people here can solve the CAD problem (which can then be scripted to pull from your Excel tables) once it’s translated from “How did they cut and bend plates to match lines in 1942?”
Stepping back for a second, do you actually need to be able to union and get high quality solids or is “looks good when I click raytrace” the standard you’re trying to achieve?
