Principal Stress Trajectories

Hello Everyone,

I've been looking into deriving lines of principal stress on curved surfaces and have come across some really interesting stuff but I'm hoping to spark a discussion that might create a clearer picture of the processes involved. So far I've come accross:

- Odysseas's Principal Stress Grid

- Jon Mirtschin's post on importing FEA data from GSA to Grasshopper

- and Christina Doumpioti's thesis on The Adaptive Growth of Fibre Composite Structures

From these sources I gather one must:??

- use a finite element analysis program to determine the magnitude and directions of the principal stresses at all the mesh vertices

- then bring that data into grasshopper

- write a script that generates a vector field from the FEA data and interpolates a series of evenly distributed curves that follows the principal stress vectors

Would this be the most sensible way to go about this? I just want to make sure I have decent course of action laid out as my scripting/solid mechanics knowledge is rather limited so it's going to be quite the undertaking...lol

Any comments or reference suggestions would be greatly appreciated. Thanks kindly in advance for any and all help.

All the best,

Matt

www.mattschmid.com

Tags: GSA, analysis, composite, element, fibre, finite, principal, stress

▶ Reply to This

Replies to This Discussion

Permalink Reply by msgrom on November 9, 2010 at 5:44pm

Hello Jon,

I had a question about how the streamlines are generated and whether you had a suggestion about how to approach the problem Im having.

The attached files show the 3d vectorfield, the same vectorfield baked as lines (plan, red lines) and the resulting streamlines for those lines as outputted from structDraw (plan, purple crvs).

How would you go about the problem so that the contourlines (closed contourlines) from the vectorfield represent the values on the surface (like the topographic lines)?

Im thinking about measuring the values of the vectorfield on a grid of points across the surface and then creating the contour lines from the closest values on that grid of points but im also having frustrating results with that. Any further help would be very much appreciated. Thanks so much for the work so far with StructDraw by the way.

Best,

M

Attachments:

Permalink Reply by Jon Mirtschin on November 9, 2010 at 10:34pm

Hi Mateo,

Great, nice to see it's giving plausible results, even if they could do with a of improvements (there's not enough hours in the day).

The GSA output actually produces two vectorfields (maxima and minima) that are perpendicular, so you can run the procedure for the streamlines you require.

Any chance you can zip the rhino model with the vector lines and email it to me to use as a test case? I can try to write a command to rotate the vectors 90%. I can also use your file as a test case to improve the results you're getting.

Cheers,

Jon

Permalink Reply by Matt Schmid on October 5, 2010 at 10:52am

Hi Jon,

Amazing! a few hours?! I've been pulling my hair out trying to do this, you're an inspiration. I would love to do anything I can in evaluating and testing this. Just let me know how I can help.

All the best,

Matt

▶ Reply

Permalink Reply by Jon Mirtschin on October 6, 2010 at 5:07am

A few hours in a couple of sessions (so maybe exaggerating slightly).

I did spend some time when I took a trip to father in law's remote farm away from the computer (it often helps solve problems not being near it) thinking about some of the papers I've read and how I'd apply openNurbs to do it.

I'm sure my algorithm is slower than most, and the quality might be not as good, but it is providing answers and there's lots of scope for improvements.

I'll email you with how to access this feature while it's under development.

Cheers,

Jon

▶ Reply

Permalink Reply by sivam krish on March 17, 2011 at 1:38am

I have started an online digital design training course on intuitive structural design. http://www.digitaldesignacademy.com/?page_id=146

I would like to use a package that can help students visualize principle stress lines. If there is one, I will be happy to use if for my course and introduce it to the students.

Cheers

Sivam

▶ Reply

Permalink Reply by sivam krish on October 4, 2010 at 8:19am

I too have been searching for ways of doing this, for teaching. I get stundents to draw principle stress lines. Found some papers in japanbese on how to do it, but none in english.

▶ Reply

Permalink Reply by zhushunlai on June 28, 2011 at 6:07am

dear sivam can you list the japanese paper on the topic how to draw stress lines.hope you can share it ...it may be very useful for us

▶ Reply

Permalink Reply by sivam krish on June 28, 2011 at 6:05pm

Automatic Plotting of Principle Stress Trajectories by Direction Finding Contouring, by Mischisuke JO - The Japan Society of mechanical Engineering 2002_

▶ Reply

Permalink Reply by zhushunlai on June 28, 2011 at 7:55pm

thank sivam so much. i am so glad to recieve your reply so quick!!

▶ Reply

Permalink Reply by dani bas on December 12, 2010 at 12:12pm

i am following this theme with absolut atention.

i am working on my diplon project (architecture) and i am designing a voronoi 2.5D based structure and want to control the density and position of vertice points throught FEA.

▶ Reply

Permalink Reply by Daniel Hambleton on April 23, 2013 at 9:35am

I haven't updated the plugin in some time, but the SPM Vector Components was designed to do exactly this kind of thing...

We tried to incorporate interpolation between data points, different integrators, direction switching (if you're only interested in finding streamlines for both the principal stress directions), orbit finding, etc...

Not sure if it even works in the new GH versions...! If there is interested, I'd be happy to update it.

EDIT: just reread your post - the SPM components only deal with the vector field integration, you'd have to get the principal stresses in to Rhino first (via GeometryGym or Millepede or something).