Building of Aluminium Boat . Mounting of hull framing structure

[an2reir]

Thank you Barry, your feedback is indeed most helpful . I just had a long conversation with my builder, we will find a way and fix it right for this first prototype . What we are doing is we are just working to develop an aluminium boat company - and this is a prototype. I will probably redesign the hull bottom for the series production boat. I do tend to keep my thinking however that the scale model does prove the hull shape to be buildable because the scale model is 500mm LOA and was made in steel plate 3mm thick. I am inclined to think if the steel hull bottom did form an acceptable curvature on the bottom , together with the scale model's topsides and internal framing not showing some striking flaw this is acceptable partial validation. Together with the gaussian analysis in Rhino not showin irregularities and together with the surface unrolling into a flat cutting file by both Rhino and Paneling Tools; that makes more validations. I think the reason for the hull bottom plate to be more curved is the method of the bottom having been assembled and welded together and to chines but without a keel. there was a continuos welding of the two bottom plates on center inside. I think there has been contraction from welding of the two bottom plates on center inside

 

[Barry]

A question to those familiar with Rhino

Does the program generating a cut file, allow for the thickness of the material or does it just deal with the sheet profile as if it is infinitely thin?

 

[an2reir]

Just NURBS surface no thickness. Now that you mentioed barry - I will talk to Rhino assistance people ; have the hull bottom developable surface check by the Rhino people

 

[DCockey]

I did again check now the gaussian curvature and it appears like the attached.

Is your file in meters or millimeters? If the file units are meters than for a 5.8m hull a Gaussian curvature plot which is the all green with a scale from -0.01 to 0.01 does indicate that it is "developable" or close to "developable". However if the Rhino file is in millimeters then all green with a scale from -0.01 to 0.01 does not indicate that it is close to "developable". I would expect Gaussian curvature of 0.000o1 mm^-2 or less.

 

[Ad Hoc]

However if the Rhino file is in millimeters then all green with a scale from -0.01 to 0.01 does not indicate that it is close to "developable". I would expect Gaussian curvature of 0.000o1 mm^-2 or less.

David,
I don't use Rhino so no idea.
However, is this then not a failing or short coming of the software?
Since one assumes this metric for measuring, if a shape is developable or not, should be non-dimensionalized as such?

 

[Mr Efficiency]

Looking at the attachment in post #3, and using the Mk.1 Eyeball, no way that is developable, as drawn with those straight sections, what says you, Ad Hoc ?

 

[DCockey]

David,
I don't use Rhino so no idea.
However, is this then not a failing or short coming of the software?
Since one assumes this metric for measuring, if a shape is developable or not, should be non-dimensionalized as such?

My question was intended for Andrei, the original poster. Rhino lets the user select the length units to use, anything from angstroms to lightyears. Gaussian curvature is dimensional with units of inverse length squared. The numerical value of Gaussian curvature will differ by a factor of 1,000,000 depending on whether meters or millimeters are used. Use millimeters and the numerical value may appear small, for instance 0.0o1, even though the surface would not be considered close to developable for many applications. Switch to meters and the Gaussian curvature of the same surface will be 1000 which no longer looks like a small number. By the way, a sphere with a Gaussian curvature of 0.001 mm^2 has a radius of 32 mm. A sphere of 3200 mm radius has a Gaussian curvature of 0.00001 mm^2

The use of Gaussian curvature for assessing if a surface is "close enough" to "developable" is very common and not peculiar to Rhino users. Gaussian curvature is zero for an exact developable surface, and that fact leads to the practice of looking at Gaussian curvature and assuming if it's numerical value is a "small" number then the surface is close enough to developable. As my examples above illustrates the seemingly small numbers such 0.001 or even 0.00001 can be misleading if the length units used are small. My preference for determining how close a surface is to developable is to examine the principal curvatures which can be done in Rhino using the Curvature command. If a surface is exactly developable then one of the principal curvatures will be zero. If a principal curvature is not zero (within floating point arithmetic accuracy) then the user can look at the curvature circle and/or the numerical value of the principal curvature (inverse of radius) and decide if the curvature is small enough.

 

[an2reir]

Hi there, thank you all for your feedback. David Cockey, my file is in millimeters and I did set the Gaussian scale from -0.00001 to +0.00001. Same green no distortion. When I set the Gaussian cuvature range from -0.000001 to + 0.000001 mm the image is the attached, there is an area of light blue at the tip and light blueish at center. My assumption was that together with the validation of having made a model in steel plate thickness 3mm and together with the validation that the hull bottom is unrolled into flat surface by two different unrolling procedures that do not work the same with many other files; the Rhino unrolling and the Paneling Tools plugin software unrolling , I did consider the above as validation enough. And that perhaps I will need to push the plate at the tip . In the first build the hull bottom has been put together by welding at center the two bottom panels without a keel first at points on the inside, then welding the chines then welding the center on the inside with continouous weld. I beleive the metal has contracted at center on the inside and thus the hull bottom has slightly curved up and therefore the side bottom panel now has more curvature than expected. I think the hull bottom should have been welded on short welds followed by space 500 mm followed by short weld and on the outside f the hull. In this case I assume there might still be a curvature of hull bottom but that would be at the order of mm. I will ask now the Rhino people if the bottom is in real life supposed to unroll or not

 

[DCockey]

Andrei, use the Curvature command in Rhino and click at assorted locations on the bottom. Watch the reported minimum principal curvature. What is the largest number you see? The inverse of that number will be the radius of curvature of the ruling lines in mm.

 

[an2reir]

Andrei, use the Curvature command in Rhino and click at assorted locations on the bottom. Watch the reported minimum principal curvature. What is the largest number you see? The inverse of that number will be the radius of curvature of the ruling lines in mm.

Ok, I am starting to do " My preference for determining how close a surface is to developable is to examine the principal curvatures which can be done in Rhino using the Curvature command. If a surface is exactly developable then one of the principal curvatures will be zero. If a principal curvature is not zero (within floating point arithmetic accuracy) then the user can look at the curvature circle and/or the numerical value of the principal curvature (inverse of radius) and decide if the curvature is small enough." I will try to check like you wrote DCockey. I attach as well another Gaussian check this time wit the curvature range from - 0.000001mm to + 0.000001mm. Same light blueish at the tip and towards center. Reading there values I would have thought that the aluminium panel 5mm would unroll and would be mountable on the hull setup with perhaps some tensioning.

 

[DCockey]

Ok, I am starting to do

In the command line area there will be a report of several quantities including the principal curvatures for each time you click. The value of interest is the minimum principal curvature. You can expand the command line area by pulling down on the bottom of it.

 

[Mr Efficiency]

Mr Efficiency, using your Mk.1 Eyeball, is the hull bottom in the illustration below developable? It's a design I just created.

It could be, and likely is. I do seem to detect some curvature in the sections, and there is a gentle transition in deadrise, that would not require much convexity to be developable. A developable shape with all straight sections, would need parallel sections.

 

[an2reir]

DCockey , I am doing as per your above . I did the check for the moment at one point seen on the screenshot. Minimum principal curvature
4.05951e-05 (0.331064, -0.816085, 0.473711). None of the values are zero. Values appear to me to be small. In real life that would translate in my opinion the hull bottom plate would require some tensioning but would be buildable. If I double check and surface unrolls and if I make a steel model and it does work I would totally go for it. I do understand however and I did take note that one of the values should be zero. I did a number of the check at five points and values are as below:
Surface curvature evaluation at parameter (4410.61, 44.5596):
3-D Point: (88232.7, -16169, 333.202)
3-D Normal: (-0.266546, -0.562455, -0.782686)
Maximum principal curvature: -0.000349357 (0.905179, 0.132853, -0.403733)
Minimum principal curvature: 4.05951e-05 (0.331064, -0.816085, 0.473711)
Gaussian curvature: -1.41822e-08
Mean curvature: -0.000154381
Select point on surface for curvature measurement. Press Enter when done ( MarkCurvature=Yes ):

Surface curvature evaluation at parameter (4575.2, 50.2976):
3-D Point: (88075.7, -16101.5, 342.082)
3-D Normal: (-0.309122, -0.570576, -0.760846)
Maximum principal curvature: -0.000360902 (0.905725, 0.0673392, -0.418483)
Minimum principal curvature: 3.11371e-05 (0.290011, -0.81848, 0.495969)
Gaussian curvature: -1.12375e-08
Mean curvature: -0.000164883
Select point on surface for curvature measurement. Press Enter when done ( MarkCurvature=Yes ):

Surface curvature evaluation at parameter (4771.61, 54.4997):
3-D Point: (87894.3, -16043.8, 380.249)
3-D Normal: (-0.371062, -0.572532, -0.731109)
Maximum principal curvature: -0.000400831 (0.906698, -0.0533638, -0.41839)
Minimum principal curvature: 1.58172e-05 (0.200527, -0.818144, 0.538915)
Gaussian curvature: -6.34001e-09
Mean curvature: -0.000192507
Select point on surface for curvature measurement. Press Enter when done ( MarkCurvature=Yes ):

Surface curvature evaluation at parameter (4913.07, 45.0129):
3-D Point: (87772.2, -16076.1, 473.391)
3-D Normal: (-0.424165, -0.566079, -0.706851)
Maximum principal curvature: -0.000404618 (0.896367, -0.151358, -0.416675)
Minimum principal curvature: 7.57634e-06 (0.128883, -0.810336, 0.571615)
Gaussian curvature: -3.06552e-09
Mean curvature: -0.000198521
Select point on surface for curvature measurement. Press Enter when done ( MarkCurvature=Yes ):

Surface curvature evaluation at parameter (5065.96, 39.7019):
3-D Point: (87643.3, -16067.6, 550.36)
3-D Normal: (-0.478314, -0.552903, -0.682286)
Maximum principal curvature: -0.000414511 (-0.877911, 0.281514, 0.387326)
Minimum principal curvature: 4.47942e-06 (0.0220804, -0.78425, 0.620052)
Gaussian curvature: -1.85677e-09
Mean curvature: -0.000205016

 

[DCockey]

DCockey , I am doing as per your above . I did the check for the moment at one point seen on the screenshot. Minimum principal curvature
4.05951e-05 (0.331064, -0.816085, 0.473711). None of the values are zero. Values appear to me to be small. In real life that would translate in my opinion the hull bottom plate would require some tensioning but would be buildable. If I double check and surface unrolls and if I make a steel model and it does work I would totally go for it.

A minimum principal curvature value of 4.0595e-05 means the ruling "line" through that point has radius of over 24 meters. Seems close enough for me.

 

[DCockey]

It could be, and likely is. I do seem to detect some curvature in the sections, and there is a gentle transition in deadrise, that would not require much convexity to be developable.

It is very close to developable. The minimum principal curvature corresponds to a radius of curvature of several kilometers. The sections are very slightly curved.

A developable shape with all straight sections, would need parallel sections.

Agree, but if the "exact" section has very little curvature then building with straight sections may be feasible.