Cable Element

[dfish67]

Hello,
I am trying to understand the modeling of cables in FEA. Do any of you know of any good reading/references. I am interested in creating a dedicated cable element in some inhouse software.

Thanks

 

[EngForm78]

You could try the NAFEMS.org web site. They have alot of info and resources.

EngForm78

 

[GBor]

One way to model cables is using a non-linear truss element. Basically input the stress-strain curve of a truss element such that when you have positive strain, you have positive stress (hence, a modulus of elasticity). When you have negative strain, you have a flat (virtually zero slope) stress-strain curve (hence, little or no modulus).

I believe Constantine Spyrakos wrote a book on non-linear analysis and he may have included this. Sorry, don't have the reference at my fingertips. Algor Publishing promoted the book several years ago.

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com

 

[EdR]

You can also use a curved beam element with a compression cut-off on the material definition....(assuming you have that type of capability in the material routines)

Ed.R.

 

[dfish67]

Thanks for the replies...we already have a zero compression truss. Specifically, what I would like to model is the tension effect of weight in a cable. Let's see...imagine a horizontal cable suspended from two cantilever beams. If you were to swing from the cable the two beams would tend to move towards each other. We are starting to model systems where this effect is becoming more consequential. I have done some research on catenaries and such but am still unsure how it all applies to FEA. If someone could suggest a definitive source of info....heck even an "indecisive maybe" would be appreciated.

Thanks again

 

[GBor]

I'm having a little trouble understanding the difficulty. You can apply mass and gravity to your current truss elements I'm assuming. Does your zero compression truss allow for a definable modulus when it goes into tension?

I worked a problem with a commercially available package of a car impacting a telephone pole...electrical cable vibration. The poles and cable are what I'm picturing when you discuss a catenary between two cantilever beams. Is this image marginally correct?

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com

 

[Denial]

Many years ago I had a similar need. The best paper I came across for the rigorous treatment of a cable was by an electical engineer who was interested in the sags & tensions in transmission lines. "Precise Sags and Tensions in Multiple Span Transmission Lines, by J.Barrien, appeared in 1975 in "Electrical Engineering Transactions", published by the Institution of Engineers, Australia. Unlike most papers on cables, this one includes the effect of the cable's stretch.

For how to go about incorporating this behaviour into an FE program, look at the classic book by R.K.Livesley, "Matrix Methods of Structural Analysis". In particular its chapter 10 "The Analysis of Non-linear Structures", where section 10.4 gives some useful expressions.

If your only problem is that of the horizontal cable between two cantilevers, and the load is applied at the middle of the cable, it would be relatively easy to set up a spreadsheet to do the calculation. By allowing for symmetry you reduce the problem to one with only two degrees of freedom: the horizontal movement of the support and the vertical movement of the loaded point.

 

[EdR]

As with gbor I am also having difficulty understanding the problem....You can include the weight of the member on a beam element which is what you are looking for...As for the catenary shape, that is specified when defining the coordinates of the beam element nodes...The axial load developed will be properly accounted for by the beam.....

A cautionary note....Unless you have done special corrections DO NOT use a truss element. Truss elements have no bending capability and multiple truss elements connected as in a beam actually form a mechanism that will not solve...Use beam elements....

Ed.R.

 

[GBor]

Ed.R. makes a valid point about truss elements, but only depending upon your software (hence, I imagine, the "special considerations" that Ed.R. mentions). Beam elements are appropriate in certain software packages, but you have to be careful how the beams consider the rotational degrees of freedom.

If I think about it, I guess it depends on whether you are trying to use a single beam element to represent the catenary, or a series of smaller elements. For a single element, the truss clearly would NOT work, but for a series of much shorter elements, the lack of rotational degrees of freedom in the truss elements may better represent the catenary. If you go with beam elements, make sure to release the Rotational DOF's if you are trying to use a series...at least, that's the thoughts off the top of my head.

 

[dfish67]

The effect I am looking for is the tension in the cable. Which is directly related to the unit weight of the cable. The tension at either end of the cable (in the simple example I gave) will have a component horizontal as well as vertical. If I modeled it with a truss element or a beam element there would only be a vertical component.

I have not tried it but I am sure that if I tried to model it with multiple truss or beam elements that it would not solve. I think that is approaching a mechanical mechanism which our solver, although being non-linear, will not solve.

Thanks

 

[rob768]

Can't you use a spring element? or a series of spring elements and mass elements?
Just set the stiffness to sero in the case of compressive load (perhaps using a gap element, if your models allows it)or remove when compressive forces occur.

 

[crisb]

Use truss elements, apply a mass proportional load to the cable, then subdivide the mesh into enough elements to get a good approximation to the catenary. Use nonlinear analysis and apply an initial strain to the cable to get the correct tension.

I have done this on numerous occasions and confirm it works, no mechanisms form.

 

[EdR]

dfish67:

What you said in your last post is not correct...A 3-D beam element has 6 dof and will provide both an axial and two transverse resisting forces as well as 3 moments...(Do not constrain these rotational dof, rather let them connect to each other and probably leave them free at the ends, depending on the end conditions) Thus you will get both an axial and shear forces in the element...If you want these in some direction other than comes out of the code it is a simple matter to rotate the vectors....

Note that this is different than a truss element which carries only an axial load...Think of modeling a simple horizontal beam with transverse (vertical) loads by a series of truss elements...Since the elements can only carry loads in a direction along their length (horizontally) there is no stiffness in the transverse (vertical) direction to resist the loads....

As noted earlier you only need include the weight of the cable in your definition of the elements or loads, depending on the specific code input requirements.

crish....The element you were using was probably not a typical 3 dof truss element but rather a specialized cable element.....

Ed.R.

 

[crisb]

dfish67 and edr

The vertical component results from the deformation of the cable that occurs during the nonlinear analysis.