Designing for Torsion?

[ParabolicTet]

What are some general tips for desigining a structure for torsion resistance? Are hollow tubes the most efficient shapes? How about welded box sections with some type of internal stiffener arrangement like cardboard uses? Pros and cons of either?

Thanks!

http://web.peoriadesignweb.com/eos

 

[mloew]

ParabolicTet,

Torsional stiffness is governed by the following formula:

K = (J G)/L

For a member of fixed geometry (L can not vary) you will want to maximize J (the polar inertial moment) and G (torsional or shear modulus; itself related to E, the elastic modulus)

Best regards,

Matthew Ian Loew
"I don't grow up. In me is the small child of my early days" -- M.C. Escher

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[prex]

Well, any (mechanical) engineer knows that a circular hollow section is the most efficient one for resisting in torsion. This also holds as far as torsional stiffness is concerned.
Also a tube with a higher diameter and the same unit weight as another one, will be stronger and much stiffer than the smaller one.
Any closed hollow section with major and minor dimensions not too far apart will work well in torsion, and internal stiffeners will not be of much help: however the stiffeners could be required to insure the stability with thin sections.
However all this is very basic stuff: can you be more specific on your concerns and your goal?

prex

http://www.xcalcs.com

Online tools for structural design

 

[feadude]

see this thread

http://eng-tips.com/viewthread.cfm?SQID=70555&SPID=679&newpid=679&page=2

I hope this helps.

 

[mloew]

feadude,

The link you posted does not work for me. try posting again using the thread number format that appears below the subject line. For example this thread is thread727-95155.

Best regards,

Matthew Ian Loew
"I don't grow up. In me is the small child of my early days" -- M.C. Escher

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[feadude]

MLoew:
Here it is:
thread679-70555

 

[GregLocock]

It is very important to use diaphragm bulkheads in rectangular sections, and at changes of axis in tubular sections.

Square tube is not a terrible choice for torsion, but as you move to a rectangular shape the dominant dimension is the smallest of the width and the depth. A rule of thumb for more complex sections is that it is roughly as effective as the largest circle that can fit inside the section, of the same wall thickness. Strongly reentrant shapes are a nightmare (a square section is weakly reentrant, torsionally).







Cheers

Greg Locock

 

[imagitec]

ParabolicTet wrote:

Torsional stiffness is governed by the following formula:

K = (J G)/L
...
J (the polar inertial moment)

This equation is valid only for circular cross-sections, for which the torsional constant, K (not the same as your K, above), is equal to the polar moment of inertia. Consider a closed tube versus a tube split along its length. They have virtually the same polar moment of interia, but the split tube has a much lower torsional stiffness.

FEA can be used to determine K for noncircular cross-sections. Extrude the cross-section into a beam and apply a torque. K can then be calculated from the deflection:

K = (T*L)/(G*theta)

Rob Campbell

http://www.livejournal.com/users/robcampbell

 

[imagitec]

The quote in my previous post was from MLoew, not ParabolicTet. Sorry for the mistake.

Rob Campbell

http://www.livejournal.com/users/robcampbell

 

[magicme]

the torsional properties of a single (simple) cross section are not difficult to work up, but for any "structure" made up of simple tubes, etc. welded into an assembly, i don't imagine how you could hand calculate the structure's torsional effectiveness.

a "cheap and easy" method to evaluate a few structures would be to make a first-cut finite element model of each by using simple beam or pipe elements connected in different ways and load it.

in ANSYS, you could make these graphically on the screen in very little time, and delete and relocate elements, load it, run it and change it.....if this is linear static analysis, you could knock these out at fast nearly as you can draw.

once you settle on some credible concept structure, refine the model as desired (change beam element properties or switch them to plate elements, etc etc).

I am sure this applies to any other FE software as well.

daveleo