Strand7

[Junior1234]

Hi Eng-tips community,

Does anyone know how to model a curved beam on strand7

Thanks.

 

[canwesteng]

Probably as a series of straight beams.

 

[bugbus]

Tools > Points and Lines > This will insert a series of straight beam elements along a circular or parabolic path.

If you want the elements themselves to be curved, you will have to look at quadratic plate or brick elements.

As a side note, is anyone aware of any commercial FE packages that include curved beam elements and whether there are any major advantages of modelling a curved beam in this way?

 

[WARose]

As a side note, is anyone aware of any commercial FE packages that include curved beam elements and whether there are any major advantages of modelling a curved beam in this way?

STAAD does. But I've never trusted it to do that. (Not sure I'd trust any FEA package on that.) So I have typically just done it with a bunch of short straight beams.

 

[IDS]

It would be interesting to compare STAAD curved beam results with the straight beam approximation, but really using a reasonable number of straight beams gives a perfectly good approximation for practical purposes, and using the procedure given by gusmurr it's quick and easy to generate however many beams you want.

p.s. Junior1234 - did you see my response to your AS3600 question, and did it answer your question?


Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[Junior1234]

Thank you everyone for your answers. IDS your answer in my other thread was helpful thank you very much.

The answers that i am generally seeing is that you should model out a series of straight beams. In that case my question would be how many before id expect an accurate answer. Answer i am expecting would be in a percentage of the diameter/length.

Thanks again eng-tips community

 

[bugbus]

The more beams, the more accuracy, but with diminishing returns. You would have to experiment a bit to see how sensitive the results are to the number of elements.

It will depend on the change in angle along the beam, the tighter the curve the more elements you will need to properly model the geometry.

I would think that you would want something at least about one beam element for every few degrees change of angle but that's just a guess.

 

[WARose]

The answers that i am generally seeing is that you should model out a series of straight beams. In that case my question would be how many before id expect an accurate answer. Answer i am expecting would be in a percentage of the diameter/length.

I can't give you a answer in terms of percentage.....but I have typically used (approx.) 1 foot segments. Considering how fast FEA software knocks out something with beam elements (in a static analysis) these days......the number isn't that big of a deal.

 

[canwesteng]

I've always used enough so that it looks curved to my eyes in the GUI. They are just simple beam elements, so don't really eat up processing power as WARose noted

 

[WARose]

Good point canwesteng. In this case, it "looking" right is pretty close to what you need. The rule of thumb I talk about above comes from when I use to do angle arches at spans of about 20-30 feet.

 

[Junior1234]

In that case, considering everyone has agreed that putting a certain number of straight elements to form a curve is a satisfactory way to model out the curved member my follow up questions are
- what type of restraint would the nodes between the straight members be (how would i make sure strand 7 considers it as one member)
-what is the most efficient way to model out a large mount of straight members between to node points to create that curve

Thanks everyone for their responses

 

[WARose]

- what type of restraint would the nodes between the straight members be (how would i make sure strand 7 considers it as one member)

I'm not familiar with "strand" enough to say.....but in STAAD, I wouldn't have any releases (assuming we are talking about a continuous member here).

-what is the most efficient way to model out a large mount of straight members between to node points to create that curve

I use to have a spreadsheet that would generate the coordinates (for circular shaped curves). I'd then plug them into STAAD and connect the joints (with beam elements) manually within.

 

[IDS]

Beam members meeting at a node are treated as having a rigid connection by default, so any continuous line of beams will be treated as a continuous beam.

For creating a series of beams on a curve, see the post from gusmurr (3rd post in the thread). The tools-points and lines menu lets you create a series of beams, with any number of segments, along a circular arc, parabola, ellipse, or varying radius, by clicking on three points.

If you have a current support licence you should also have a look at the web-notes at the Strand7 web site, which has detailed notes and examples on a wide range of different topics.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[Mohanlal0488]

Just something small to mention, GNL would be required to develop any additional loads due to the curvature of the geometry.

 

[Settingsun]

Also bear in mind that the number of straight segments may need to be different for out-of-plane loading than in-plane.

 

[IDS]

Just something small to mention, GNL would be required to develop any additional loads due to the curvature of the geometry.

Why would it?

If you define the geometry of the curved beams, the analysis will find the resulting actions of that geometry.

You will need a non-linear analysis to take account of the actions resulting from the deflections due to the initial loads, but the same applies to a structure with straight beams.


Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/