Optimization using StarCCM+ (How to use Pressure Distritribution results as deformers of the igs hull shape?)

[Guest Ms. Neykov Svetlozar]

Hello in 3D visualization for example Rhino you have displacement map, I was wondering if this idea can be applied in vessel optimization in CAESES.

 

For example ships so not to vary with different curves deforming the shape (guessing which will be the best optimization), but to put whole 3d vector  (point direction and magnitude) field of pressure (coming from starccm+) using some additional factor for multiplying the deformation, so the user can choose how much to deform the shape (by changing values of the multiplier). Because the best shape will be with the lowest pressure?

 

If not Is there such an option in CAESES to deform whole object by picture for example to map (gray-scale image from the results) and to brush how much i want the deformation like Zbrush?

 

Thank you.

[Jörg 29]

Hi Neykov,

 

Thanks a lot for this good input, we've had similar discussions in our team a long time ago. Somehow, we haven't followed up on this. So far, there is no direct option to realized such a task in CAESES. But I will pick your idea for our next discussions and give you some feedback on this. Both concepts (vector field & picture map) seem to be attractive to me in some situations.

 

I'm not sure whether it helps, but we have the option to deform a shape based on the adjoint sensitivities, e.g. as calculated by STAR-CCM+.

 

Cheers

Joerg

 

BTW: We have just recently opened a new category for these kinds of questions and suggestions:

 

https://www.caeses.com/forum/index.php?/forum/22-ideas-and-suggestions/

[Guest Ms. Neykov Svetlozar]

Maybe this also will be useful to implement - soft modification tool from Maya , based on this gray scaled alpha 0 - 255 ... So user not only can implement the grayscalized pressure results from StarCCM (or other CFD,FEM) but also to have smooth deformation with nearby shape - to blend with it in the areas where the picture is over (like falloff blend from the displacement texture.)  See Example here: http://download.autodesk.com/global/docs/maya2014/en_us/index.html?url=files/Deformer_Tools_Soft_Modification_Tool.htm,topicNumber=d30e363975

 

Another useful think will be 2 shapes and blend between them. Final very optimised, and beginning (and user can dial a slider).

 

 

 

Thank you!

[Jörg 29]

Thanks for this great input again!

 

Note that we have a spot transformation, as well as a cartesian shift transformation that allow you to create similar results. I'll check whether we have a good example for it that I can provide to you.

 

Cheers

Joerg

[Guest Ms. Neykov Svetlozar]

Thank you I will wait for input better yet I have another tedious task in the pipeline: Integration plugin needed for example of inputing shape from Rhino to Cases and later to  StarCCM. Maybe plugin like automation tool that makes polysurfaces to parametric models for example explode and dupborder then defining border curves as futures for shape deformation. Because I have ship/vessel shapes using (Rhino modeling tools I am used to it) which I don't want to remodel (It will take me lot of time to learn modeler of Cases and if I can reuse the geometry and curves from Rhino as a modeling tool and with some additional plugin connector like a convertor it will be great to parametrize rhino hull forms it will be great. There was iges import and openNurbs but need more info I want pipeline Rhino- Cases-StarCCM

[Jörg 29]

I mean there are many options and ideas from my side, depending on how much time and effort you want to invest. In customer projects, we sometimes convert IGES geometry to a parametric model by using some fitting mechanisms. I guess this is not what you are looking for, you rather want to deform the surface patches, right?

 

Another option might be to work with STL data from Rhino. This data can be easily used in CAESES, and also deformed with existing transformations (shifts, spots, lackenby, ...). The connection to STAR-CCM+ is then set up in just a few minutes. CAESES also takes care of the coloring for boundaries, if required. Sure, then you will need the way back to math. surfaces - this might also be directly possible when applying the same transformations to the IGES import, but not 100% sure, this depends on e.g. the surface quality.

 

Anyway, I think your ideas are quite nice.

 

Cheers

Joerg

[Mr. Carlo Pasquinucci 1]

Hi Neykov,

 

I think that you can easily deform your geometry applying the Free Form Deformation in Caeses directly from your import geometry.

With Free Form Deformation you don't need to remodel the geometry, but you can easily work directly on the imported Iges.

I was thought for this kind of problems.

If you need some ideas, you can share me (through email) your geometry.

I'm thinking a solution that can be useful for you.

 

Cheers,

Carlo

[Jörg 29]

Yes, Carlo, thanks you are right. Just notice that Free Form Deformations are part of the Pro Version.

[Guest Ms. Neykov Svetlozar]

Thank you for quick response Mr. Palluch yes I want to deform the surface patches. In Rhino I am doing it with shifting few control points F10 then using smooth command. But still the idea is that CFD knows more than naval architect for small and complicated flow physics so let it it deform them. I can deform just main shapes but small details let CFD do it. More or less technique shown in the video: "https://www.youtube.com/watch?v=hgjxO5yuRxg"  . Sometimes I need to pass trough exact points then I use loft or surf network commands of few curves but usually there are bumps (and surfaces are not faired) the shape doesn't fit very well so I use and the rebuild command (smoothing mechanisms in Rhino to loose the points for which the surface need to pass: "https://www.youtube.com/watch?v=3XkdIsleAqY" 

 

In your way with fitting surface like Maxsurf guys do "https://www.youtube.com/watch?v=KgHmvT-Zbyk" is nice but I don't know how to do it in Rhino. So I am still adopting the way do do something similar with Rhino. To have clean surface passing trough points I make clean Nurbs patches with Sweep2 rails, one by one. Then I match them also use filletsrf or blend for some and join them together to make watertight G2 (tangent continuity surface)... 

 

About the other topic lets say after I have the clean polysurface (in the example I give 8 small patch Nurbs surfaces) then I need it parametrisized in such a way that the deformation deform the whole patches together. So lets say we divide the deformations on 2 ways

1. Globally: As you said with changing main dimensions (L,B,D,d,Cp,Cb,Cm, Lakenby) this can be done with  the control points of all additional deformational curves or FFD passing trough whole the ship stretching it in one direction Scale1d, for changing length, breadth etc.. (For example on the level of the waterline for changing the waterplane area) - for that I was thinking "dupborder command"  

2. Locally:  

2.1 By using additional deformational  curves for example frames of the ship (soft deforming few of their control points) controlling locally the patch deformation (in the bow and stern area) 

2.2 By using the control points of the surface patch(my suggestion with soft mod) 

 

So about the local optimization of the patches 2.2. 

Maybe STL is the way to go here to mesh all the patches in one mesh and deform it (ether with FFD, cage in Rhino), and see the results in StarCCM+.  But still mind that after all this nice activity I will have mesh (.stl) triangulated in CASES or in StarCCM+ but I need Nurbs at the end. So my question is can I still keep my initial NURBS surface geometry(the 8 nurbs patches) after the results of StarCCM+ which show how to deform the geometry (hypothetically after the addition of the future) deformed the STL object and return new hull form to Cases. And after this CASES can do the another iterations to deform few more times Stl.mesh based on the Vector map pressure results from StarCCM+. After it is finished and I liked (trade of of 3-4 iteration between streamlined (optimized)shape and space loosing (because I need to keep in mind placing engines, outfitting, accommodation etc...). So when I liked the shape can I say yes return me to original 8 patches Nurbs form (but deform it with this optimized stl shape)? So at the end to have optimized Nurbs form (same as input form, but optimized in Rhino), which can then continue further down the production line or I will be left with .stl which will be triangulated but optimized?  Sorry for the long post needed to clear the things in my head.

 

Mr. Pasquinucci here is the example Nurbs Geometry to play. lets start it simple;).

http://www.filedropper.com/examplegeometry

 

 

 

Is there example of boat going from Cases trough StarCCM

[Mr. Carlo Pasquinucci 1]

Hi Nevkov,

if I good understood, you want to coupling Caeses, Rhino, StarCCM+.

The idea is to start from Rhino, go to Caeses, modify the geometry in Caeses, export in Star, reimport in Rhino, modify in Rhino.

 

In Caeses you can easily import your Iges file from Rhino as Nurbs, deform as you want and then I created a feature that approximate the deformed surface to a bsplinesurface that you can export in Star and in Rhino, but not always it works good and you have to customize for your geometry.

 

It's very complex and I suggest to use another strategy.

 

Sorry, but I don't understand why you need to use also Rhyno.

Another question: the pressure in Star are exported in a text file?

Thankyou

Carlo

[Guest Ms. Neykov Svetlozar]

The idea is to start from Rhino (with shape that I have or I can model fast because I know it), go to Caeses connect the form with STARCCM and run to have basic solution, then come back to CASES to make different variations modifications based on the pressure results of STARCCM (or other CFD soft it can be free also like OpenFOAM) (it will be good if I can use Vector field of pressure to modify the hull form) and for example to have 4 different versions of model in Cases deformed with example 25%factor of pressure results, 50%factor of pressure results, 75 %factor of pressure results 100%factor of pressure results deforming the shape more and more (so 4 different shapes) . Then come back to StarCCM get the solutions again for the 4 shapes - and return back to CASES to optimize plot the results with Pareto using some genetive algorithm  ( and compare which on I want) based on some controversy criteria for example min resistance (on x axis) and max space for arrangement (B  of ship on y axis). So to have easy graph in which I can choose design, and judge the ship characteristics. After this when I choose the compromise design which I want  for example  maybe 25% factor of pressure results deformation I want to extract the shape in clean NURBS and go back to Rhino.  

  

About Rhino I need the geometry after to use it as sections for making drawings, and continue further with software like Nupas, Shipconstructor etc so can make the details of the ship for production.

 

Otherwise to have nice optimized shape but to be mesh(.srl), or ugly Nurbs (with a lot of patches, not stiched (watertight) ) it is usless, because I need to make sheets after this, so to cut the shape for shell production etc..

 

Yes STARCCM can give you txt, pressure results... If it is more easy or have already connection of CASES with other but free CFD soft like OpenFoam, also is a solution...

[Mr. Carlo Pasquinucci 1]

Hi Svetlozar,

 

in Caeses it's possible to export NURBS surface, if you don't care about the polygon of control, but they are Nurbs.

After your modellation, you have to select each Imagesurface and create a Nurbs from it. Then, you can export as Nurbs and also as 3dm. But the poligon of control is not more the original one (it has more points).

 

If you use the SurfaceDeltaShift or Deltashift or other Shift function, the total surface remain watertight (if the original are watertight),because the same transformation is applied to each surface. You have to keep attention only on the keel

 

In order to deform the geometry using the pressure field, my idea is to use SurfaceDeltaShift with a Bspline surface and the move its points in y direction in according to the pressure field.

 

If you have an example outpfile from Star, I can explain how to do.

 

In the attache picture, the grey surfaces are the original ones, the green, the deformed, the red the deforming bsplinsurface

and in gold the control points that you have to link to the Star file.

 

Cheers,

Carlo

 

 

 

[Guest Ms. Neykov Svetlozar]

Ok I think it can give similar results like: Excel table " pressure values(scalar value), x,y,z coordinates of point" Something like the example file, at least what I have seen from tutorials...

 

http://http://www.filedropper.com/121

 

 

So my question is, can the shape be deformed by this scalar values, and not to make separate curves with controlled cv which deformed the shape? But more like texture displacement map deformation based on the numbers (with position the points and magnitude of deformation the pressure results)?

 

I will try to find result with Open Foam so I can give u exact output...

 

I will start learning Ceases...

[Guest Ms. Neykov Svetlozar]

Dear Mr.

I mean there are many options and ideas from my side, depending on how much time and effort you want to invest. In customer projects, we sometimes convert IGES geometry to a parametric model by using some fitting mechanisms 

 

Tell me more about this option because I am following this tutorial:

"https://www.youtube.com/watch?v=DG0w_YBRQKo" 

 

And I've reached the point Model in Cases: Export to iges (to go to Rhino because OpenNurbs somehow didn't work) Then from Rhino I tried to come back with .Iges import in Ceases and it works.  But now the surface losed the quality of beeing remodified (its parametrization)

 

So how to convert it back to parametric model? Is there link to any tutorial where I already have a surface and then convert it (or make new one but parametrized)?

Here I attached before rhino file made by tutorial where I can deform the shape as I want in Ceases, and after in Rhino the same, but when I go back in Ceases I cannot do nothing to shape..

 

http://www.filedropper.com/showdownload.php/rhinoimport

 

Thank you...

[Jörg 29]

Hi,

 

I've just noticed your post... Happy New Year!

 

You need to have a parametric model in CAESES ready, in order to start the fitting. This fitting can be done manually, by simply trying to tune the model parameters as best as possible, to get closer to the imported geometry. Or automatically, which is a bit more complex (as said before, we've mostly done it in customer projects).

 

So the main work is to set up a parametric model in CAESES that gives you the general characteristics and required details of your imported geometry. I.e. there is no generalized model that can be readily fit into your imported data, of course.

 

Cheers