plot principal direction of stress

[sicheng6699]

Dear FEAP uses,

does anyone know how to plot the principal direction of stress?
Just with contour plot it is not that easy to see.
Thanks a lot for help!

Best
sicheng6699

[sicheng6699]

something like

[FEAP_Admin]

Try PRAXis.  See appendix E of the manual for options.

[sicheng6699]

Try PRAXis.  See appendix E of the manual for options.

Thank you! I didn't know that, as a user and a developer, it is still hard to known every useful command. 
I will try it soon and give my feedback.

[sicheng6699]

The plot works very well, see two figures  attached.

One further question maybe, is it possible to define a certain direction for stress plot? For example a command like
    PLOT STRE -1 x1 y1 z1
with option -1 for certain direction?

[Prof. S. Govindjee]

You would have to edit the code for PRAX to do this.  Note that PRAX already uses the maximum number of numerical arguments (3) that plot command can have, so it will be a bit tricky to do what you want.  See plot/pplotf.f plot/prax.f plot/praxp.f .

[Prof. R.L. Taylor]

This should be done as a user plot.  You can observe how the stresses are prepared for plotting by looking in ./plot/pltstr.f where we do the principal stress set up.  DO NOT edit this routine, instead make a new routine that puts things where you can plot them and define a new plot command for the uplot.

[sicheng6699]

Following your suggestions, I realize that my idea was not that correct. The indention is only to check the normal stress in certain direction. But for a stress tensor it is needed to define a rotation matrix, instead of just a vector as I mentioned. So it should be some command like

PLOT STRE -11   x1 y1 z1   x2 y2 z2

with 2 vector to determine the rotation matrix, and option (STRE -11) here to plot the stress sigma_11.
I am not sure whether I am able to implement the UPLOT subroutine, but thank you so much for pointing out the possibilities.

Best
sicheng6699

[Prof. R.L. Taylor]

We can implement something like normal stress in a certain direction quite easily, however, it is not clear how useful this is to the general user community.  Any feedback would be welcome.  Implementing stress components on a plane would require, as you note, require implementing a fuller transformation of stress but could also be done.

This would be done for a future release, not a patch to 8.4

[sicheng6699]

We can implement something like normal stress in a certain direction quite easily, however, it is not clear how useful this is to the general user community.  Any feedback would be welcome.  Implementing stress components on a plane would require, as you note, require implementing a fuller transformation of stress but could also be done.

This would be done for a future release, not a patch to 8.4

Dear Prof. Taylor,

I agree with you that such plot option is not so useful for most people.

As feedback to your interest, I'd like shortly make two commons on the contour plot with "PLOT STRE".

1) Like the Mohr-circle theory that I learned from the uni., the first step is usually to determine the principle stress, and the rotation. The following question is always to determine the stress at certain edge with arbitrary rotation.  Similar to that, if someone has a model with thin layer, one interest could be to plot the stress in certain direction, e.g. orthogonal to the layer surface, even both normal and shear stresses, just as assumption. Maybe the feature can help a lot.

2) In my case the structural member (panel) is not parallel to the x- or y-axis. Due to prestressing I am mainly interested in the "in-plane" stresses of the panels. It could be helpful, but not absolute necessary. Such structure is not popular and the feature could be helpful.


Apart from the above mentioned two points about "PLOT STRE", actually I' like to consider more features for the plot of principal stress.
In my opinion, the "PLOT PRAX" could be improved by considering the following advantages (+), and disadvantages (-).
(I read the user manuel shortly and tried out for my example, and you can also correct me if wrong.)

(+) combination with "PLOT STRE" is very very interesting and helpful (through using "PLOT STRE 1 0 1" just for outline without mesh)
(+) color can be chosen freely with "PLOT PRAX 1 +1 4" with blue for the tension part of principal stress , as you know sometimes the tension part is filled with red, which is not good to still use red for the tension.

(and sorry for the following, if wrong)

(-) plot nodal principal stress is not better than at the gauss points. (even worse for cases with singularity?)
(-) the lines has the same length, better scaled with the value, see the figure at the beginning of the discussion.
(-) with "PLOT HIDE", the principle stress can not be plotted inside or  behind the element.