FE analysis of steel connection

[WITT55]

I am doing FEA for a complicated connection in a steel framed building. I am in search of a code or reference that would provide me with an allowable stress or strength of the steel. It seems that the AISC manual does not apply as I do not have simple bending, tension or compression forces. I want to have an appropreiate factor of safety at the same time not being over comservative, as it has already been designed by my client and is used numerous times. Have you done a FEA of steel before, what did you uses as an allowable stress and what was your basis? Any help or direction would be greatly appreciated.

Thank You,

 

[feadude]

look at AISC spec for allowable stresses for combined loading. there are two sections : one for bending plus compression and another for bending plus tension. most of AISC spec have analytical/FEA and/or testing documents to back them up. if you are looking for the least weight go with LRFD method. Personally I use ASD because I have used it for years and I know it better than LRFD.

 

[JStephen]

I think you'll find the building codes just really aren't set up for FEA of components like that.

Look back in AISC-ASD at the section dealing with bearing in bolt holes, and you'll find that the allowable bearing stress can exceed the yield stress in that situation. But that doesn't mean the part has "failed". I suspect that if you start looking at fillet welds in great detail with FEA, you'll find very high stresses there at times, too.

With the ASD, I'd probably try to show that the attached beams would fail in bending (or compression) before my connection failed. (I think you'll find some of the safety factors in the commentary.)

 

[WITT55]

Thank you for your replies. I am fully aware of the AISC spec and combined loading. However, I would agree that AISC and other building codes are not applicable to this type of analysis. I was somewhat vague in my initial post. What I have are four built up beams that tie into a tube column. The beams are fully welded to the column, so the beam connection is not of great concern. I am doing a FEA of this situation, my main concern being the tube column at this loacation. I am having trouble determining what the allowable stress are. And do I look at the principle stress or do I compare Von Mises stress to the allowable.

 

[Qshake]

Von Mises to the yield stress so that you'd have to use a safety factor to compare with the allowable.

Regards,
Qshake

Eng-Tips Forums:Real Solutions for Real Problems Really Quick.

 

[crisb]

I work from the UK and have done a number of connections like this.

Firstly I apply factored loads to an elastic model, and compare with the yield stress. If there are a few places over the yield stress, I consider a nonlinear run with plastic material properties and look at the plastic strain. In Europe hand designs for tubular connections are based on CIDECT rules and its unlikely that you can match these capacities (which were based on test results) without accepting a degree of plasticity.

 

[feadude]

In US (AISC) allowable stresses for tubing for tension, compression,shear, or combination is well below yield (linear) and I fully agree with this philosophy having seen some structures fail using european standards that have less factor of safety.

 

[trainguy]

You should consider running plates through slots (1 plate thru 2 slots) in your tubular column. If it were me, I would do whatever is needed to avoid out of plane local bending in your column wall.

There's something to be said for connection designs that absolutely require FEA - they may need design mods that make them easier to treat with manual calcs, thereby increasing not only the margin of safety, but also your margin of confidence!

Thats my 2 cents ($CAN)

tg

 

[kasper27]

WITT55:

It sounds like you have a problem where you are evlauating bi-planar stresses (different than bi-axial). I have a similar problem where I have strong-axis bending + axial compression + flange prying on a steel pile. I am trying to evaluate if my flange is locally overstressed due to the bi-planar stresses and shearing stresses. I do not believe that the AISC Manual of Steel Construction addresses this behavior.

I am considering comparing the principal plane stresses OR von Mises stress to ......(here's where I need help or a reference).....and the maximum shearing stress to .....(again, need help here). I am working w/ service-level loads. Do you have any references you could suggest I review that provide recommended allowable normal and shearing stresses?.....Would you recommend a different analysis approach?

Thanks in advance for your thoughts.

 

[tonycola]

I suggest that you look at ASME section VIII, division 2. This code uses stress intensity to define allowables. As you know this code is for pressure vessels and not for building structures. But, the allowable stress intensities which are tabulated ASME section II are applicable to the stress intensities that are calculated by FEA software.

Have fun!

 

[14159]

I have yet to find a connection that can't be designed using rational engineering analyses. You might have to do some yield-line analyses or take a hard look at the AISC prying action stuff, but you should not have to resort to FEA.

As a designer in the USA, with some experience with FEA and steel connection design, I wouldn't use FEA for this, PERIOD. Unless you have a tremendous amount of expertise in FEA, you will get A result and not have a clue whether it's reasonable. It might even look reasonable but be garbage.

Years ago, I did some of this for part of my thesis and was astounded at the difference in results from choosing different types of shell or solid elements. Much of this had to do with advanced parameters that most designers know nothing (or at least nowhere near enough) about, such as the type of integration, etc. I could change a seemingly benign parameter and get answers 30% different, either higher or low.

I also don't know what kind of program you're using, but I would want something like ansys or abaqus for serious modeling. I would not use whatever generic element that the typical structural packages throw in there. Who knows what it'll do.

Just my $0.06.
DBD