Concerning direction to draw lines!

[Lorens]

When you work with line elements and draw lines between keypoints, then you have to think to draw them on the same direction, else you may get strange result when doing a torque diagram so I have been told.

That is something i found troubling especially if you have a fem model with many lines interacting with each outer, if you get a problem you may think that it may be because I have drawn the lines in the wrong direction

Any advices or tips to calm me.

Kindly Paul-Martin

 

[meshparts]

Hi,

I have the same problem, when I mesh lines with beam elements with axis offset. Since the offset direction is dependent on the line direction, some beams are offset in the wrong direction. So I have to correct this problem by manually inverting the lines (lreverse).

Regards,
Alex

 

[Mech151]

Offset direction can also be controlled by designating an orientation kp.

 

[meshparts]

Thank you Mech151! I have totally forgotten about that! That solves my problem!

Regards,
Alex

 

[meshparts]

Hi again!

I have defined an orientation keypoint and it doesn't work! I still have to manually revert the lines! I assign the orientation keypoint to the lines through latt and the orientation of the lines determine if the orientation keypoint means "positive element y coordinate" or "negative element y coordinate". If I would directly generate the beams, it would work, but that would be to complicated.

 

[Mech151]

You will need to issue the latt command at least twice, using a different orientation keypoint for the incorrectly oriented beams....

 

[meshparts]

Tat's correct, Mech151, but how can I determine automatically, if the lines are incorrectly oriented, unless I look at them? That was my problem: I have to manually invert the lines or manually assign different orientation keypoints. It seems allmost impossible to me, but I couldn't find a better way to do that... And I have a lots of lines...

 

[Mech151]

If you look more closely at the cross section definitions to understand (predict) what orientation will come from a given keypoint, then you should be able to get it all (or at least most) of the lines correct the first time. If you look at the sketches in secd in the help manual, the OKP designates the positive Z axis. In addition to changing the OKP, I normally end up having to define alternate section definitions (right hand/left hand for instance) and latt the lines with different section numbers (especially when Ibeams or c channel is modeled). If you are using section offsets, an understanding of the coordinate systems used will make things go a lot faster.

Also, based on your description of how difficult these changes are to make, I am lead to assume you are building this model in the GUI. I would recommend using command input text files. This makes these modification much easier, and facilitates the use of APDL, which is great for this type of modeling (allow parametric changes to size, increase in accuracy, etc).

 

[meshparts]

Dear Mech151,

I (allmost) never use the GUI, I have a good understanding of cross sections and coordinate systems. Nevertheless I have the described problem (see above).

It seems to me, that the orientation kp and the orientation of the line determine the offset direction of the cross sections. That means, the OKP designates not the positive Z axis but just the z axis. Whether the OKP lies on the positive or negative side of the z axis depends on the line orientation.

So the line orientation plays an inportant role. Since the line orientation is practically randomly choosen, I have to look at the lines and correct the wrong oriented ones.

I assume you have a very good understandig of this things, so perhaps you can confirm or infirm this to me...

Regards,
Alex

 

[Mech151]

Alex,
I could not look into this until I was working in ansys again. Well, now I am. If you are still watching this thread, let me know. I drew a line, issued latt, meshed, and noted orientation of my non-semetrical cross section. Then I cleared the mesh, reissued latt, remeshed, and finally noted that the two meshes looked identical. Please confirm you have seen otherwise, and provide specifics so we can figure out what is going on here. Any chance your OKP is perpendicular to the real Z axis (i.e., do you have your axis crossed).
Regards,
Mason

 

[meshparts]

Hallo again,

like I said, in my post from 3 Apr 06 3:14, the OKP determines the direction of the z axis. The sign of the sign axis is determinate by the line orientation. So in my opinion, you must allways check the mesh and, if necessary, reverse the lines.

Regards,
Alex