Godzilla Sharpies

[mcollins07]

I’ve been studying the Sharpies with Reuel Parker’s book as a major source of information. Particularly interested in analyzing the New Haven Sharpie.
I’ve been using Michlet to do a bit of comparative analysis of hull outputs from FreeShip, so I thought I’d try Godzilla. I would like to use Michlet / Godzilla to determine the optimum B/T and L/B for this particular style of hull in flat still water.

I realize this is a highly constrained problem for Godzilla, and perhaps not Godzilla’s strength. Nevertheless, optimum B/T and L/B for this hard chine hull is the current question.

So far, I been able to constrain the problem enough to get some reasonable numbers, but not quite close enough to the Sharpie hull yet.

The hull under consideration is Lwl (LOA) of 360 inches, or 9.144m. A reasonable displacement for this hull is 4.7 m^3. I’ve held these as constraints.

Starting with a sharpie hull from Freeship, optimizing for 5 design speeds and for Objective function Rt (total resistance), Godzilla gives reasonable results of B/T = 3.216 and L/B = 5.338.
However, after 875 iterations, the hull waterlines are longitudinally symmetric. For the sharpie hull waterlines are not longitudinally symmetric, but will be somewhat blunt or squared off on the aft end. This symmetry is my primary reason for questioning the applicability of the solution. Is there a way to further constrain Godzilla to find a sharpie hull?

[daiquiri]

You should ask Leo.

[Leo Lazauskas]

Originally Posted by mcollins07

I’ve been studying the Sharpies with Reuel Parker’s book as a major source of information. Particularly interested in analyzing the New Haven Sharpie.
I’ve been using Michlet to do a bit of comparative analysis of hull outputs from FreeShip, so I thought I’d try Godzilla. I would like to use Michlet / Godzilla to determine the optimum B/T and L/B for this particular style of hull in flat still water.

I realize this is a highly constrained problem for Godzilla, and perhaps not Godzilla’s strength. Nevertheless, optimum B/T and L/B for this hard chine hull is the current question.

So far, I been able to constrain the problem enough to get some reasonable numbers, but not quite close enough to the Sharpie hull yet.

The hull under consideration is Lwl (LOA) of 360 inches, or 9.144m. A reasonable displacement for this hull is 4.7 m^3. I’ve held these as constraints.

Starting with a sharpie hull from Freeship, optimizing for 5 design speeds and for Objective function Rt (total resistance), Godzilla gives reasonable results of B/T = 3.216 and L/B = 5.338.
However, after 875 iterations, the hull waterlines are longitudinally symmetric. For the sharpie hull waterlines are not longitudinally symmetric, but will be somewhat blunt or squared off on the aft end. This symmetry is my primary reason for questioning the applicability of the solution. Is there a way to further constrain Godzilla to find a sharpie hull?

The first problem is that you are using Hull Series 1 to approximate the Sharpie. This series has three shape parameters which is not really enough to approximate a Sharpie hull. That's why Godzilla gives you a fore-aft symmetric hull.

Try using Series 9. I can't remember what the maximum number is in the old versions of Michlet. (In later versions you can use more than 42 shape parameters, so it is a relatively simple matter to "parameterise " existing hull shapes like you are trying to do).

Then, constrain some other quantities, such as CB, CP, Cwp, GMT, GML etc.
For example, if the real Sharpie CP is about 0.6 (I don't know what it is), then use a minimum CP of 0.59 and a maximum of 0.61. If you make the range much narrower it might take Godzilla a while to find the first hull that doesn't vilote the constraints.

I've been going through the same exercise as you are trying but with rowing shells, kayaks and some slender multihulls. With many shape parameters and the right constraints I can get very accurate mathematical approximations of existing hulls.

Have fun!
Leo.

[Leo Lazauskas]

Just another quick point...

If you are trying to get a "mathematical" approximation of the Sharpie, then there is no reason to use Godzilla to minimise the resistance in any way because that slows the process too much.

Constrain the Surface Area to be similar to the actual Sharpie and search for the hull with the minimum surface area that falls within all other constraints you impose, such as Cp, Cb, etc.

You should then have as a baseline hull something that has very similar principal parameters to the actual Sharpie.

Once you have that "baseline" hull, you can then start trying to minimise the resistance.

Leo.

[mcollins07]

Originally Posted by Leo Lazauskas

Just another quick point...

If you are trying to get a "mathematical" approximation of the Sharpie, then there is no reason to use Godzilla to minimise the resistance in any way because that slows the process too much.

Constrain the Surface Area to be similar to the actual Sharpie and search for the hull with the minimum surface area that falls within all other constraints you impose, such as Cp, Cb, etc.

You should then have as a baseline hull something that has very similar principal parameters to the actual Sharpie.

Once you have that "baseline" hull, you can then start trying to minimise the resistance.

Leo.

Ok, thanks Leo, I'll consider that approach too.

I’m using version 8.07 which does not have the Series 9, but the manual does show it has a Series 7 with fore/aft asymmetry. This time I started with your in.mlt example for a kayak. However, with this in.mlt file, Godzilla runs, but does not give me the asymmetry. Perhaps I have a problem in the in.mlt file?

Interestingly, I tried a constraint of Cp = 0.60 with the Series 0 offsets just as in the previous in.mlt file. It ran, but converged to Cp=0.5333 and did not result in any asymmetry.

[Leo Lazauskas]

Originally Posted by mcollins07

Ok, thanks Leo, I'll consider that approach too.

I’m using version 8.07 which does not have the Series 9, but the manual does show it has a Series 7 with fore/aft asymmetry. This time I started with your in.mlt example for a kayak. However, with this in.mlt file, Godzilla runs, but does not give me the asymmetry. Perhaps I have a problem in the in.mlt file?

Interestingly, I tried a constraint of Cp = 0.60 with the Series 0 offsets just as in the previous in.mlt file. It ran, but converged to Cp=0.5333 and did not result in any asymmetry.

If you look at the end of the out.mlt it says that the beam constraint is violated.

Also, I noticed a few odd "tabs" in your input file. You might not be able to see them in some text editors.

Leo.

[mcollins07]

Ok, I'm having fun now.

I found the problem with the beam constraint and had to change shape factors also. But now I’m getting results that I can correlate with my sharpie model.

I have not tried using the surface area as a primary constraint yet, but intended to try it. However, getting any mathematical model for a sharpie is not my goal.

I have a highly defined model in excel. I'm currently using chord/camber angles for chine’s as design parameters, then let Freeship or Michlet tell me what the Cp, Cb, Cm, … are. So, this current method maybe closer to traditional boat builder’s view than a modern NA’s approach. Michlet gives me feedback on only the one aspect of the design. Based on that feedback, along with many other design consideration, I attempt to "Analyze" and optimize the design. The goal is to get a quick easy feedback loop to my model, which is in excel. Godzilla’s automated iterations may facilitate that by making many iterations on it’s own.

For me, and I suspect many others, the design parameters that Godzilla uses, are not the same as the ones that I use. A powerful future feature for the tool, might be to accept input of several sets of offsets which I produce from excel or FS, and a tool then provides automated analysis of those, for min, max and averages of the parameters which Godzilla is using to specify hulls. If I could input several sets of offsets, and then tweak the constraints, the tool might be more intuitive to use. … Just a raw idea from one user.

Thanks again Leo.

~ Michael

[mcollins07]

Leo,
Can you share with us how you turn the “Shape Factors” into offsets? I’m interested in Series 7 equations in particular. Is this a standard method reported in literature or something you created? If it is proprietary, I understand you might want to keep it proprietary, but I thought I’d ask. If you have references you can share, that would be great.

~ Michael

[Leo Lazauskas]

Originally Posted by mcollins07

Leo,
Can you share with us how you turn the “Shape Factors” into offsets? I’m interested in Series 7 equations in particular. Is this a standard method reported in literature or something you created? If it is proprietary, I understand you might want to keep it proprietary, but I thought I’d ask. If you have references you can share, that would be great.

~ Michael

There's nothing secret about the hull series. I have just set up the series in such a way that rectangular, parabolic, and elliptical shapes arise fairly naturally. That's why the same series can be used to generate Wigley hulls, hemispheres, rectangular blocks, elliptical struts etc.

There is some explanation of the formulas for the various hull series in the manual.

See also: Section 2.3 of the thesis at:
http://digital.library.adelaide.edu..../1/02whole.pdf

Good luck,
Leo.