Oasys GSA

A discussion thread for posting questions and examples relating to Analysis model generation in Grasshoper to Oasys GSA.

Wouter Brok raised the question here http://www.grasshopper3d.com/xn/detail/2985220:Comment:619778?xg_so... about how to generate an element capable of tension resistance only.

Tie elements are available in GSA to emulate cables and tensile rods. They can be generated in the Geometry Gym plugin be assigning the Beam type in the Beam Attributes component. Model attached.

Note this will force the GSA solver to use a non linear solver to enforce this. If you're confident your loading scenarios will only induce tensile forces, you can still use the static solver by setting a bar element (and then verifying manually no compression force exists).

Attachments:

120618 gsa beam type.gh, 3 KB

Replies to This Discussion

Permalink Reply by Michael van Telgen on June 18, 2012 at 10:19am

Nice, a GSA topic.

While we're on the subject, I was wondering how I can do a modal buckling analysis?

Or is this already supported in one of the newer versions of ssigsa?

Permalink Reply by Jon Mirtschin on June 18, 2012 at 10:27am

I've enabled a few components to generate non linear (and I seem to recall modal) analysis and controls, but I'll have to double check modal buckling. I'll see if I can add it ASAP.

Permalink Reply by Wouter Brok on June 19, 2012 at 10:01am

Dear Jon,

Today I run my calculations with the 4 tie tension elements (see image 2 and 3, 4 corner points). The calculation duration was a half hour and no output because the max cycles were reached (1000000). When I select "link" instead of tie (inside GSA) and select only tension in the link properties (see image 1), the calculations are done in 30 seconds.

Is there a way to define the "link properties" (as shown on the image) in GH?

Attachments:

Permalink Reply by Jon Mirtschin on June 19, 2012 at 11:16am

Hi Wouter,

If you aren't getting convergence (ie models running forever) there's something amiss (either the problem is unstable or ill conditioning is occurring).

It's difficult to predict the cause of the problem without the model. You could be getting a tension yield if the stress is too high, or perhaps there's an initial compression in an element (once it it is disregarded it's ignored). The analysis log should report this.

I'm not sure about the link element, it's a rigid element but that might be acceptable (always be wary of adjacent very short elements next to proportionally long). I'd suggest applying a pretension, perhaps running with bar elements to determine required level or prestress.

Happy to advise further,

Jon

Permalink Reply by Michael van Telgen on June 20, 2012 at 12:00am

It looks like you are trying to make an anticlastic shape? Is that right? In that case you can just make pins at the corners and use 1D force density.

The thing you have to remember when using a link is that it is infinitely stiff in its own direction (EI=infinite). Ties are not. So your tie stiffness may be too small (in that case just increase I to find out). One million cycles is much too long for this simple problem. I'd say this needs maybe 2000 cycles max? Maybe less.

If you need some more help, let me know. I'm happy to help

Permalink Reply by Wouter Brok on June 22, 2012 at 9:54am

Dear Micheal,

Thanks for your help!!, I already got the anticlastic shape working (image is already a anticlastic shape but with very little double curvature). The link element on the corner are used to fix the corner points of the membrane to the supporting structure:

So I'm going to try to increase the I in Grasshopper or try Cable elements like Peter is proposing. Now trying to get the model running

Thanx

Wouter

Permalink Reply by Michael van Telgen on June 20, 2012 at 12:13am

Permalink Reply by Peter Debney on June 22, 2012 at 3:03am

For tension-only you can also use Cable elements. These are a little trickier than Ties as they act like wires running over pulleys: cables elements with the same property number will all have the same tension. This has the knock-on effect that the nodes are not fixed along the length of the cable.

They are sensitive to their set-up. All the cables with the same property must be in a single chain with all the elements going in the same direction: all cable elements with the same property form one cable. There can be no breaks or forks in this chain.

If used correctly they can be very powerful; if used incorrectly then the results can be quite unexpected.

I have put an example model along with a webinar presentation that includes these super-elements at http://www.oasys-software.com/webinar?id=55/

Permalink Reply by Lukas Nordström on April 26, 2013 at 2:06am

Hi Jon!

We are tying to develop a structural optimization using a fem mesh, galpagos and Oasys GSA. Do you have any experience and examples relating to this issue?

Permalink Reply by Lukas Nordström on April 26, 2013 at 2:07am

Hi!

We are testing the grasshopper functions parametric beam and truss optimizer on the geometric gym web page. We have them running on one computer but once loaded to the other computer we get an error in the SSiBEAM component saying "Request for generated beam has coincident points and is ignored at node 1 0,0,0. We have same grasshopper version 9.14 and same gsa version 8.6 build 23.

Any suggestions what to do?

Thanks for respons

Permalink Reply by Jon Mirtschin on April 26, 2013 at 5:03am

I'll see if I can automate checks for this (and force Rhino to load the plugin).

This error occurs if the rhino plugin isn't loading (recent installers should always register it). Try using rhino command pluginmanager or drag drop .rhp files as outlined at http://www.geometrygym.com/installing

Permalink Reply by Lukas Nordström on April 29, 2013 at 11:23am

Hi! Thanks a lot Jon for the answer!

Now I have got it all started...Now a new question ;)

I have set up a simple mesh mesh with applied force at the left boundary and two fixed nodes at the right. My question is when I move the fixed points (marked with black) with the slider the "old" points (marked in red) are still fixed. So by moving the sliders I just add fixed nodes not just move them. Do you have any clue what this could depend on? See attached picture

All the best

Lukas

Attachments: