tension simulation for networks

[microelly2]

I'm looking for a ready to run method to simulate the tension forces in a rubber band network
the force is proportional to the lenght of a segment (hooke)

I pin some points (in the vide I use arcs to mark them)
the network is a sketch object with coincidence constraints only and the fixed positions of the pinned points.

phpBB [video]

is there a fem module and a tutorial how to get this running?
thanks for a tipp.

[thschrader]

Hi microelly2,
what you wanna do is a shape optimization of a tensile structure.
For engineers this is one of the toughest stuff you can do!
Good luck and some hints...
https://www.youtube.com/watch?v=Gb5p0htTj4U
https://www.youtube.com/watch?v=P41462SbG1I
https://www.youtube.com/watch?v=i9pA_YxYkoY
https://www.b-tu.de/great-engineers-lex ... rfindungen

[thschrader]

Hi microelly,
can you give some more explanation on the video?
If I understand the process correct, you use a starting grid and then
"relax" the grid. What is the criterion? To get the global minimum of the strain energy?
There is a state at 0:06 where two grid members in my opinion are under "pressure",
not tension. Otherwise there would be no static equilibrium possible. Is it a planar grid
or 3D? regards thomas

phase1.JPG (27.66 KiB) Viewed 2594 times

[DeepSOIC]

If I understand the process correct, you use a starting grid and then
"relax" the grid. What is the criterion? To get the global minimum of the strain energy?

To me it looks like an iterative process with fixed number of iterations, where on each iteration each point is moved along the net force vector.

[microelly2]

If I understand the process correct, you use a starting grid and then
"relax" the grid. What is the criterion? To get the global minimum of the strain energy?

To me it looks like an iterative process with fixed number of iterations, where on each iteration each point is moved along the net force vector.

yes that's it at the moment. i calculate for each vertex the sum of all its edge vectors as force.
it is a 2D sketcher simulation because I want the easy way to add more constraints to the model.

[bernd]

Still I do not really understand what data do you input, and what you would like to analyse. Since it is sketch based I assume all members are in a plane. Furthermore rubber usually behaves highly nonlinear when leaving linear stress, strain range. In which range are you? Your edges are staight, but rubber usually has no bending stiffness. What do you assume?

Bernd

[microelly2]

Still I do not really understand what data do you input, and what you would like to analyse. Since it is sketch based I assume all members are in a plane. Furthermore rubber usually behaves highly nonlinear when leaving linear stress, strain range. In which range are you? Your edges are staight, but rubber usually has no bending stiffness. What do you assume?

Bernd

in the simplest case the force is the hooke force if a line is longer than mk = 2,
its a simple linear model and can be changed later. if the length is less than 2 then there is no force.

Code: Select all

	f=0.01

	for n in g2.nodes():

		nbs=g2.neighbors(n)
		v0=g2.node[n]['vector']
		r=FreeCAD.Vector()

		for nb in nbs:

			mk=2
			tf=g2.node[nb]['vector'] -v0 
			if tf.Length > mk:
				fac= 1.0*(tf.Length-mk)/mk
				tf=tf * fac
			else:
				tf= FreeCAD.Vector()
			r += tf

		(a,b)=list(conix[n])[0]

		rc=sk.movePoint(a,b,v0+ r*f,0)
		g2.node[n]['vector2']=v0+ r*f
		rc=sk.solve()

[regis]

I'm looking for a ready to run method to simulate the tension forces in a rubber band network
the force is proportional to the lenght of a segment (hooke)

I pin some points (in the vide I use arcs to mark them)
the network is a sketch object with coincidence constraints only and the fixed positions of the pinned points.

phpBB [video]

is there a fem module and a tutorial how to get this running?
thanks for a tipp.

cool stuff,
this reminds me of this
Architecture in FreeCad is promissing.

phpBB [video]

phpBB [video]

[microelly2]

cool stuff,
this reminds me of this
Architecture in FreeCad is promissing.

having some of the grasshopper features would be a nice thing.
I still look for a fast open source solution,
otherwise I will work on my slower version.

[microelly2]

I used the idea of abdullah to describe a next dimension by a circle
to get an easy to run sketch configuration.

phpBB [video]