Cross Section Optimization for real project, beam mesh system

Hi guys,

Good morning to you.

I am doing a real pavilion project the dim of it is 12M * 8M with Karamba.

However, when I doing the cross section optimization, the results(optimized beams) is quite confusing.

Problem 01: the requested in-put of 'CroSecs' from 'OptiCroSec' component, based on introduction of itself, "a list of preferable structural thickness", and the algorithm will automatically find the sufficient number as the beam thickness. BUT, when I plugged 3 ranges, 1 to 15 CM, 2 to 15 CM, 5 to 15CM, the algorithm will always choose the finest number as the beam thickness, which turns out the beam thickness is, 1CM, 2CM, 5CM. 5CM in reality 'maybe' correct, but I just cannot image 1 and 2 CM beam will work. That`s very confusing.

Problem 02: the 'MaxDisp' from 'OptiCroSec' component, no matter how I slide the number, shape will remain still always. what is the fucntion of this input, should it be like a control to deformation of the shape?

Thanks in advance for any insight you can give!!

I attached 2 images to show the problem 01, and in the file rhino and gh, i marked out the problem part with red panels.

Best wishes,

Lei

Attachments:

Replies to This Discussion

Permalink Reply by Karamba3D on August 26, 2016 at 1:12pm

Hi Lei,

ad 1.) There can be many reasons why the resulting cross sections are not what you expect. Check your load-cases: Did you include live-loads (e.g. people climbing on the structure) besides gravity? In case of shells, asymmetric live-load often governs the design. Did you take account of horizontal loads (wind, people pushing against the structure)? Did you consider imperfections (deviation between ideal and built geometry)? Did you calculate the buckling load factors for the above load-cases and did you set the buckling lengths of the beams accordingly? Did you include safety factors for the material and the loads?
ad 2.) There are two criteria for the cross section design of structures: the ultimate limit state (ULS - the structure should not fail) and the serviceability limit state (SLS - the structure should not deflect and vibrate too much). The 'MaxDisp'-input of the 'OptiCroSec'-component sets the allowable maximum deflection for the SLS. In case that the ULS governs the design the 'MaxDisp'-input has no effect.
Best,

Clemens

Permalink Reply by Lei Yang on August 28, 2016 at 7:50pm

Hi Clemens,

Thank you for the fast reply.

For ad 1., I ONLY considered the gravity as the load at the very begining stage of design. The point is for finding initial geometry divisions and deformation strategically. If so, how can I get the expected correct beam cross section, let`s say, i want 3 different size of beams which present 3 different load status(at the hinge support point the beam will be thicker while at the top thinner)?

Thank you.

Lei

Permalink Reply by Lei Yang on August 29, 2016 at 1:22am

Hi,

I upgrated a GH file from KARAMBA website.

I changed the script a bit and marked as black oval shape.

And the problem is just as same as mine, which, the optimized result will only choose the smallest number from the in-put list of 'CroSecs' in component 'OptiCroSec' instead of choose beam thickness by the load.

any idea why is that?

Thanks in advance!

Best,

Attachments:

160829_04_LargeDeformationTriangularGridshell_debug.gh, 50 KB

Permalink Reply by Karamba3D on August 30, 2016 at 8:53am

Dear Lei,

the cross sections only look as though only the smallest cross section is chosen. By enabling the display of cross section names one can see that different cross sections result from optimization (see attached file).

Best,

Clemens

Attachments:

160829_04_LargeDeformationTriangularGridshell_debug_CP.gh, 45 KB

Permalink Reply by Lei Yang on August 31, 2016 at 1:08am

Dear Clemens,

Thank you again for the useful explanation.

I have a screen uploaded on website. The red underscore is to show under which number, is the real gravity? Obviously I could get larger CroSecs by changing load capacity. If I aim to simulate the reality, which number should I put in to get the real gravity?

Thanks a lot!

Lei

Permalink Reply by Karamba3D on August 31, 2016 at 7:15am

Dear Lei,
you get the acceleration of gravity for earth by supplying a vector of unit length. Instead of increasing the external loads one could as well decrease the target value for the uilization at the cross section optimization component (see manual).

Best,

Clemens

Permalink Reply by Lei Yang on September 5, 2016 at 1:27am

Dear Clemens,

Thank you so much for the patient explanation.

And here i have to bother you again.

The initial question: since i know the unit vector could simulate the real gravity by input initial cross section from range of 2 to 10 CM, i could still get a very very very thin beam. The environment is only with gravity, no wind load, etc., so does the result saying with ONLY gravity, the very thin beam is rational? Now the beam is around 2CM and its length is about 50 CM, really slender.

Thank you so much!

Best,

Lei

Permalink Reply by Karamba3D on September 6, 2016 at 1:39pm

Dear Lei,

a slenderness of 50/2=25 is not impossible for a steel beam with full circular cross section. However without doing an in-depth check of your support and load assumptions it is impossible to say whether your structure will stand up or not.

Did you take into account that the full cross section will probably not be available at the element connections?

One should never use a finite element program as a black box. Do approximate manual calculations for the beam forces. Then check whether the cross sections proposed by Karamba can be realistic: either via simplified design formulas or a second program for steel design.

Best,

Clemens

Permalink Reply by Lei Yang on September 6, 2016 at 8:47pm

Dear Clemens,

Thank you so much!

I`ll double check with the structure engineer.

Best,

Lei

RSS