Question on Using Karamba for Designing Statically Feasible Structures

Replies to This Discussion

Permalink Reply by Karamba3D on February 8, 2014 at 12:25am

Dear William,

the deflection limit is usually assumed to be L/300 for simply supported beams and L/150 for cantilevers. These limits are chosen so that the (unavoidable) deflections go noticed by the users. There are use-cases where the maximum deflection needs to be smaller than L/300: For some systems of internal partition walls the floor deflections need to be smaller than L/500 so that no cracks in the walls develop; bridges for high speed trains must not deflect more than  L/800;...

In Karamba you can get the maximum nodal deflection directly from the Analyze-component. When calculating your tables response under dead weight plus live load (someone dancing on top of it) take this value and compare it to L/300 of it span. Be aware that for a table also horizontal loads play an important role (they produce bending moments at the connections of legs and plate).

All nodal displacements and displacements along beams can be retrieved via the 'Nodal Displacements'- and 'Beam Displacements'-cmoponents. Its then up to you to calculate the local deformation limit for each structural element.

The reason why you get such large displacements can be manifold: too large loads, too small Young's Modulus of the material. Try to start with a simple system for which you can calculate the expected deformation by hand. Then increase its complexity gradually.

Best,

Clemens

Permalink Reply by Will on February 8, 2014 at 7:28am

I get it. Yes, as you said, i got something wrong on floor slab loads (mesh load), which is to big. And then because i have no structure background, rather than calculating the deflection, i prefer 'optimize cross section' which i can see that the structure is not deformed visually. And i should use the gravity load with factor 1.5 to ensure it is a feasible structure right? Any other things i should take note?

I am ordering the student license right now. As if it's okay, i will use this in my thesis for spinning tower structure.

Well, thanks Clemens

Permalink Reply by Karamba3D on February 9, 2014 at 11:57pm

For realistic deadweight and live-load assumptions the partial safety factors against failure on the side of the loads are 1.35 and 1.5 (if there is only one live-load) respectively. To ensure usability (e.g. limit deflections) the factors are 1.0.

In an approximate analysis it is possible to use gravity load only and apply a factor to include live-load. This factor depends on the type of structure (steel and concrete) and on the type of usage.

You are welcome to use Karamba for your thesis.

Best,

Clemens

Permalink Reply by Will on February 12, 2014 at 11:04am

You have my sincere thanks.

By the way, these few days i've been exploring a lot and try to model a full tower structure. (been enrolling all the beams and nodes very carefully), cause i want to know visually how will a spinning tower with diagrid structure is likely to collapse.

There are additional questions:

1. Regarding my simple purpose on using Karamba. How to model a core support which is a big concrete cylinder. Will it be enough by defining all endpoints of floor beams which end on the core as support node?

2. Can we change the node tags size?

Permalink Reply by Karamba3D on February 12, 2014 at 12:00pm

ad 1.) If you support the endpoints of the floor beams then the supports take the load instead of the concrete core. You could try to use stiff springs (or beams with a fictitious material of zero weight) to model the offset between the middle axis of the core and its wall and thus attach the floor beams indirectly to the core. Try different stiffnesses because very stiff elements may decrease the accuracy of the numeric solution. Use hinges at the floor beams ends to model the connection to the core.

ad 2.) In the 'Karamba' folder of the installation directory there is the file 'karamba.ini'. It can be opened with any text editor. There you will find the entry 'node_text_height = 20' change the number to the desired value and restart Grasshopper.

Best,

Clemens 

Permalink Reply by Will on February 12, 2014 at 8:35pm

great, i'll try it immediately. The problems are solved by instance.

By the way, this afternoon, my thesis mentor and i will send you the machine.id file. Thanks.

Regards,

William S.

Permalink Reply by Will on February 14, 2014 at 1:00am

Hi Clemens,

I have used the spring system for the core and it worked well. Thanks.

Btw, i have problem with optimize cross section. It worked well too if i used only for optimizing from crosecs in CSTable for all beams. The problem is i want some beams to get optimized by list of circular hollow cro-sec and some others from CSTable. How do i get this? Now, there are two OptiCroSec panel in the formula and i don't know where to connect the 'model' outputs of them. Should they be reassemble and reanalyze? Do you have any similar formula that contains case like this so i can study from it.

Thank you