esys orientation

[pizzaf40]

Hello, everyone.

I'm doing some composite simulations. Now I'm treating modeling a tube with the volumes of the individual layers, so not as layered. the problem is that, having elastic orthotropic characteristics, it is necessary that x is the axial direction for each element, z is the radial and y the tangential (on all elements).

to solve the problem I created a cylindrical system, but I managed to create it only with the axial z. therefore not respect the elastic characteristics.
I then tried to create a Cartesian with x in axial direction, but so I obviously got the z and y that do not respect the radial and tangential directions...and I couldn't find how to reorient them.
(I can't change the input elastic data because it tells me that the system matrix is no longer defined positive, so the x-inversion in material data can't do it)

Can someone recommend me how to do it?


the elements with solid186 in ansys 12.1, the cylinder is composed of 9 concentric cylinders, the thicknesses of each in general are different.

 

[pizzaf40]

solved... the so-called systems of the elements are oriented respecting the direction of the x, and then the normal z is oriented with veorient...in fact, they are set esys and the veorient before meshare, but the effect is that.

 

[pizzaf40]

Not really!

I would correct...the solution worked, but to work requires the keyopt(3)=1, i.e. setting solid layered. and this for my taste is annoying for displaying results. in the absence of that keyopt, the veorient command is ignored. So I'm at the starting point if I don't want to use the layered option. I can't put the radial axis in a cylindrical system.

Can anyone suggest a solution or hypothesis that can solve my problem?

 

[stefano.garbin]

If you are interested in laminations, stress visualization and strain on each layer, interlaminary efforts etc. I highly recommend you work with shell181 or shell281.
if you design your model as surfaces and then meshi you already have the orientation z of the already arranged elements, that is, orthogonal to the surfaces.
in the case of a solid tube you can use as an element the solsh190 (solid shell), which produces solid elements with normal z axis to the base surfaces; eye that solsh190 can only use it if you make a sweep mesh type from internal to external surface or vice versa.

the problem is that, having elastic orthotropic characteristics, it is necessary that x is the axial direction for each element, z is the radial and y the tangential (on all elements).

Why does that have to be? x y and z choose them!
I solto if I have a fabric use z as direction out of the plane and x and y in the plane. in the case of unidirectional usually use indifferently x, y or z as direction of the fiber, depending on how I cooda better.

 

[pizzaf40]

In fact, I was fossilized on solid186 because of the tests I had done pre-project in which I compared tensional trends for the effects of edge using shell, solid186 monovolume layered, solid186 with multivolume (one per layer). of course it was superior the last possibility and the difficulties of modeling were not serious.
I've seen the solsh around, but I've never read it in the helo... you're making me curious.

the problem of the orthotrope on a x axis (or rather, not on a z axis) arises from the fact that giving the elastic characteristics with the main in z miscarriage. this because it checks that the matrix of the system is defined positive with a formula dependent on the values of poisson and young...but the value of this verification changes a lot by changing the axes, it is not independent from these in short. So the verification equation is negative (then the matrix is not defined positive with certainty) and does not make me proceed with the solution. I wondered why the verification is dependent on x,y,z orientation but I didn't go looking for a stuff that I could lose 2 days maybe. I took for good and tried to get around him. Moreover I do not understand well why they made the veorient usable only with keyopt layered and cylindrical systems set only with a z axis... but maybe they were implementation complications. So I've got them, too.

However, for now, I'm kind of informing myself about the solsh and see if there's anything interesting

 

[stefano.garbin]

I wondered why the verification is dependent on x,y,z orientation but I didn't go looking for a stuff that I could lose 2 days maybe. I took for good and tried to get around him. Moreover I do not understand well why they made the veorient usable only with keyopt layered and cylindrical systems set only with a z axis... but maybe they were implementation complications. So I've got them, too.

for what I know, but I might be wrong, the positive definite matrix verification does not depend on how to insert x, y and z values (I think if you look in the help you find the correct formula). It seems strange that veorient depends on whether the solid is layered or not, but I will check. in case you don't care, make a solid mesh with the elements oriented like csys,0 (cartesians) and then change the orientation with emodif.
other thing, there should be cylindrical coordinate systems with axis other than z, as long as you define them (see help command cswpla)
If you're having trouble, do you know what? ? ?

 

[Dmaz]

Hello, you're at the University of Padua?
I'm working ank on a project that involves the use of a composite laminate and I had to orientate all the reference systems of the laminate elements using a series of local systems and a macro. .

 

[pizzaf40]

for what I know, but I might be wrong, the positive definite matrix verification does not depend on how to insert x, y and z values (I think if you look in the help you find the correct formula).

eh, even to me it stinks what...and even gives me the error only by inverting the data of the axes, and the formula that puts me on the error message does not seem absolutely independent.

It seems strange that veorient depends on whether the solid is layered or not, but I will check. in case you don't care, make a solid mesh with the elements oriented like csys,0 (cartesians) and then change the orientation with emodif.
other thing, there should be cylindrical coordinate systems with axis other than z, as long as you define them (see help command cswpla)
If you're having trouble, do you know what? ? ?

I solved with the layered... I'm sure that veorient doesn't go if not-layered (with solid186 at least) because when I send the apdl of commands with that only difference, I don't orientate the local coordinate systems.

Hello, you're at the University of Padua?
I'm working ank on a project that involves the use of a composite laminate and I had to orientate all the reference systems of the laminate elements using a series of local systems and a macro. .

Yes, yes, I'm in thesis at Padova! I had already done some simulations in its time, but basically shell that did not give me problems having the normal z to the surface. the macros I can not use them, however in the end I solved and it seems to work well