Calculating J values for non-standard sections 1

[NJonesUK]

I have some non-standard sections that I am trying to use in a model however I need to work out the torsional properties of the members and am having trouble finding an accurate way of calculating them either by hand or by computational methods (without trying to get my director to shell out on proper shape-modelling software).

I was wondering if anyone here had done anything like this before or had come across some way of being able to analyse these things?

To give you an idea of what I'm looking at, I have for example an I-section with 12mm plates welded top and bottom and 10mm plates welded to the sides of the section and then a triangular section welded to one side of the member. I believe that it will be sensible to ignore the triangular section as it's only really there to make the connection to another member easier, however I can't use simple theories to work out the torsional properties since the section isn't 'thin walled', and assuming it was would be incorrect.

Any help would be greatly appreciated. You guys have been very good to me so far and I look forward, when I've got your experience, to be able to impart my help on others on this forum!

 

[ash060]

J can be estimated with SUM(kbt^3) for each piece. "k" is a constant that depends on the b/t ratio, but my reference says that using 1/3 for k is all right for most shapes.

The wraping constant Cw for I shapes is approx. [d^2(Iy)]/4

I hope this helps

 

[JD2]

If you can get hold of a copy of "Steel Structures, Design and Behavior" by Salmon and Johnson. Chapter 8 will pretty much tell you more than you want to know. My copy is copyrighted 1971 so newer editions might be formated slightly different.

 

[Denial]

Ash060's approach is valid only if the individual elements being summed do not form any closed loops.

 

[GregLocock]

The accurate way of doing it is in FEA. Calculix is free and sufficiently powerful.

ash060's method is likely to give a low estimate if I understand it correctly (eg consider the result of 4 small vs 1 large square section).

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[NJonesUK]

Thanks Greg and everyone else for your posts. Very useful, especially the Calculix program, amazingly powerful for free!

 

[Lion06]

I have the most recent copy of S & J, but I haven't looked at it for this particular issue. I've used Blodgett, Section 2-10. The k factor Ash mentions is referred to as Beta in Blodgett and varies from 0.141 to 1/3 depending on the aspect ratio of the piece.