advice on how meshare tube with flange

[Ned112]

Good morning, I'm not new to the forum but it's a lot I don't write, I've been following the forum for a long time and I've always found many useful directions.

Now I am here for a problem that afflicts me in the use of patran, software that I approach from self-taught (those tutorials and theoretical basis of finished elements made at university).

are taken with an analysis of a tool, consisting of more solid, imported in its definitive 3d form through a step file generated by solidworks.
the analysis will be various, static, sinusoidal forces, random loads through psd and shock.

the object in question has many features and components that I have provided in part to delete/simplify/replace with simpler forms to speed up the analysis.

But now I'm struggling with a stage that I find a little ostica.
I have to meshare the main component, a thin wall tube (0.5mm) flanged with flanges that cannot be considered thin.
In fact I would like to reconstruct a geometry in which the flanges are meshate with solid elements and the walls of the tube with shell elements.
I think it is also the most correct way given how thin the wall is.

here the first question:
What is the theoretically and numerically correct way to "unite" the cga knots meshano shells on the tube to the solid knots that meshano the flanges?
the degrees of freedom of the two elements (shell and solid) are different and I don't know which modeling is the best?
Do I connect them with mpc? with rbe2/3 elements with other elements?
What do you think is the best way?

the second question is more practical:
how can I eliminate the solid by leaving the "traces" (magic of constructive curves) to remesh without having to redesign the entire geometry of the tree?
Is there any way to "shake" the solid in 3 parts? tube type and 2 flanges to be able to meshare easier?

I thank everyone who will help me

 

[Onda]

Good morning, I'm not new to the forum but it's a lot I don't write, I've been following the forum for a long time and I've always found many useful directions.

Now I am here for a problem that afflicts me in the use of patran, software that I approach from self-taught (those tutorials and theoretical basis of finished elements made at university).

are taken with an analysis of a tool, consisting of more solid, imported in its definitive 3d form through a step file generated by solidworks.
the analysis will be various, static, sinusoidal forces, random loads through psd and shock.

the object in question has many features and components that I have provided in part to delete/simplify/replace with simpler forms to speed up the analysis.

But now I'm struggling with a stage that I find a little ostica.
I have to meshare the main component, a thin wall tube (0.5mm) flanged with flanges that cannot be considered thin.
In fact I would like to reconstruct a geometry in which the flanges are meshate with solid elements and the walls of the tube with shell elements.
I think it is also the most correct way given how thin the wall is.

here the first question:
What is the theoretically and numerically correct way to "unite" the cga knots meshano shells on the tube to the solid knots that meshano the flanges?
the degrees of freedom of the two elements (shell and solid) are different and I don't know which modeling is the best?
Do I connect them with mpc? with rbe2/3 elements with other elements?
What do you think is the best way?

the second question is more practical:
how can I eliminate the solid by leaving the "traces" (magic of constructive curves) to remesh without having to redesign the entire geometry of the tree?
Is there any way to "shake" the solid in 3 parts? tube type and 2 flanges to be able to meshare easier?

I thank everyone who will help me

to combine solids with shells there are various systems. Solids use only 3 gdl of knots while shells use 6.
then, to pass a moment from solid to shell you have to connect the knots in solid thickness to the shell, so that the compression traction forces at the solid side turn into the shell.
Please note that in the connection zone the values can be wrong, so this type of model is done when it does not affect the passage zone between shell/solid.
the simplest method is perhaps to make sure that the shell mesh shares the external faces of the solid and that the knots are in common between solid and shell.
This poses a problem about the type of elements to be used, since typically for solids you use 8 knot tetrahedra while for quad4 shells.
If you can meshare the brick solid, I think it's the best option. otherwise you could use rbe3 there is a great utility in general ->rbe3 node connection.
However, I repeat, stress in the passage area and absolutely inaccurate. I don't understand what a difference in thickness you have, but I suppose you'd have better result in meshare all shell, even if the flanges aren't exactly subtle.

As for the second question, I suggest you prepare geometry in another cad, like soldworks or rhino, create all geometries divided and then import into patran. even in Patran you do everything, but with much greater difficulty.
to break the solid you can use planes, surfaces, etc. but often the time spent in patran is much higher than what you use by making a well-prepared model in a cad.

 

[Ned112]

thanks to the info.


they recommended me femap saying that for meshare it is much faster and intuitive. But it was a student opinion. What do you think?

 

[Onda]

I agree,
femap is developed and advanced, patran a little less. .
the best for meshare remains hypermesh,
However, we need to see what the company is doing.
making an investment from the beginning, probably femap has the best relationship between cost and usability.

 

[Ned112]

Hello, thank you again for the info. Maybe I try the hypermesh student license because I work at the unit.

looking for something I needed, I noticed that you gave an answer a few days ago, we talked about climbing the size of a geometry to pass from m to mm.

to retransform everything in mm, make a group with all the fem and all the geometries and then scale it of 1000 on the three axes. (make a copy first). Patran does not have a system of units of measures, so as long as they are consistent with each other.
Are you sure you can only import parasolids? normally is an additional license. try to change in the import tab where parasolid is written to put iges.
the units of measurement always change from the import tab from the key where 'parasolid xmt option' is written
However changing units of measurement or scale the model within patran is the same thing.

to measure the distance under elements->show->node->distance

to display multiple elements connected to the first plotting go under utilities->display->auto plot/erase fem
attention that this utility adds elements to the group, I recommend using it from a group where you have all the fem
the sol 200 makes optimizations.

Now can you explain to me a little bit more about the procedure?
but above all, I have already provided to replace some pieces of the assembly to which I am working with models 2d (superfici).
If I stop, will you scale me? even surfaces that are not solid?
and the mesh can be scaled? Or do I have to do it again?

If you could give me some extra details about how?

 

[Onda]

Hello, thank you again for the info. Maybe I try the hypermesh student license because I work at the unit.

looking for something I needed, I noticed that you gave an answer a few days ago, we talked about climbing the size of a geometry to pass from m to mm.



Now can you explain to me a little bit more about the procedure?
but above all, I have already provided to replace some pieces of the assembly to which I am working with models 2d (superfici).
If I stop, will you scale me? even surfaces that are not solid?
and the mesh can be scaled? Or do I have to do it again?

If you could give me some extra details about how?

as said to scale you have to put everything inside a group, geometries and fem and then scale the group.
first make a copy of back up of the file, since patran has only one undo.
this operation scales both geometries and fem elements and therefore also works for surfaces.
Just create a group and add to the entities. then go to become-scale and apply the desired scale factor. Keep the word on move, otherwise you find yourself with two different size models.

 

[Ned112]

thanks for the info! Very precious!