Modeling and shortening open web steel joist

[McSEpllc]

Hi Everyone,

I am working on a 1952 office building (originally a school, but has been used as an office building for many years), in which there are existing open web steel joists to be shortened. The existing open web steel joists have a custom top and bottom chord in a WT-configuration. The top and bottom chords are made of the 1/8" (deep) !!! x 3 3/8" (wide) flange with a 3/8" (wide) x 3/4" (high) stem. The web is a continuous 5/8" diameter rod zigzagging and welded to the stem. These joists span over 24.5-ft, are spaced 2-ft o/c, and support a 2.5-in deep unreinforced concrete slab.

I did a RISA model, modeling the flanges as 7/32" x 3.375" plates (which has the same cross section area as the actual WT-members.) There are three rows of bridgeing, which I used for the unbraced length of top and bottom flanges, plus the web node spacing in the other direction. For allowable steel stresses I am using the then common A7 steel. For the top flange the RISA model comes up with "compressive stress fa exceeds F'e (Euler buckling)", and the bottom chord stressed at 465% of its allowable stress in its original condition. Even reducing the unbraced length of the top chord to 3" in the transverse direction (the existing slab undulates and locks in the joist from side way movement), the joist is not working.

Looking at joist performance, I did not observe signs of distress. There are areas of the slab cracked though, but without a general pattern.

I am hesitant to just modify the existing truss, even though the net-stresses are less (still over-stressed in RISA) even after shortening the joists by 7-ft. I am also hesitant removing the joists, as the existing slab is unreinforced and has a 65 year history of tension and compressive stresses within the slab based on the current joist layout. My inclination is to add new W8x18 joists side by side with the existing joists, plus adding the new joist web and bottom chord at the cut end.

Do you think this is reasonable, or excessively conservative?
Thanks Guys!
Eric McDonald, PE
McDonald Structural Engineering, PLLC


Eric McDonald, PE
McDonald Structural Engineering, PLLC

 

[MikeHalloran]

I get the impression that RISA would be happier or less confused if you used a more exact model of the chords. Is that possible?


Mike Halloran
Pembroke Pines, FL, USA

 

[JAE]

Yes, RISA will not give you the same capacity of top chord when using a flat plate vs. a "real" WT shape. They have different axial capacities independent of just their areas.

Check out Eng-Tips Forum's Policies here:
faq731-376

 

[Once20036]

Agree with the above comments, but I`m more interested in the bottom chord.

This is a tensile member, so to see it loaded to 465% capacity is a red flag. Are you getting compression down there, possibly from a pin-pin connections instead of pin-roller connections?
Sounds to me like there's something wrong with the model. Run a quick calc on chord T/C and compare that to the model. Make sure that your bottom chord is always in tension and your top chord in always in compression.
Get this tension chord figured out, then worry about the top chord.

I would expect that the top chord is braced by much more than just the bridging. Even if you model with a proper WT, that's still a huge unbraced length. Typically weak axis is considered braced at a relatively close spacing (pins or welds to the deck above) and it's only strong axis buckling resisted by the web members.

 

[Lomarandil]

A couple of things:

The euler buckling strength of the WT should be higher than your "equivalent flange", due to the stem increasing r in the flange weak axis direction. This will help your euler buckling stress in that direction (with Lb being the web node spacing as you mentioned). You should either model the flanges more precisely (see my post in the RISA subforum) or perform some hand calcs to determine your allowable capacity.

By 1952, you may be able to justify 33ksi A7 rather than 30ksi A7 (if you aren't already). That won't help you much, but maybe a little.

I presume you don't have any extra continuity or indication of composite action to help you.

Clearly, the existing joists have worked for the past decades -- given the discrepancy between that physical result and your model output, I'd dig a little deeper into your analysis before prescribing a retrofit.

----
The name is a long story -- just call me Lo.

 

[wannabeSE]

Has a quick check been done to validate the model.

Assume 200plf load and 26' span, M= 16.9 kip-ft
Assume 12" deep joist, T=C= 16.9/1' = 16.9 kips
With 0.7in[sup]2[/sup] bottom chord f[sub]a[/sub] = 16.9/0.7 = 24 ksi

 

[TLHS]

This is a dumb question, but who knows... Did you actually pin all your members, or is there fixity in there and you're getting a bit of moment that's killing your bottom chord?

 

[bhiggins]

Can you post a picture of your RISA model with loadings? Post an isometric image of the plates and members rendered, post an image of the axial load diagrams, and a deflected shape diagram to make sure everything is working properly.

 

[McSEpllc]

Thanks Everyone for the feedback and suggestions!

In the RISA forum of Eng-tips I got the question posted how to model a custom shape, but have not gotten a response yet. Now I am in process working my way through with it.

The topchord in the RISA model did not even work when braced at 3" o/c by being confined by the slab drooping on both sides of the flange. That is using the simplified plate shape.

The bottom chord is in tension in the model. I did the webs both as pinned and fixed connection to the chords. The members at the joist seat I modeled as fixed, as there is no triangulation.

Fy with the A7 is 30 ksi.

I did a reaction check, but not a hand calc check on chord forces - thanks for making me do that wannabeSE. It turns out the RISA's tensile force and stress is correct, but also have a large moment at the ends, due to how it modeled it.

This is my first time posting, so not sure they'll come through:

Eric McDonald, PE
McDonald Structural Engineering, PLLC

 

[UcfSE]

In talking with joist fabricators' engineers over the years and doing what you're doing, most joists are designed as simply-supported, pin-roller. It looks like you have a pin-pin condition. Changing to pin-roller will affect how the joist behaves usually to better match the original intent. The pink deflected shape also looks odd at each end, as though there is some rotational fixity in the joint restraints.

Also, have you looked at any of the SJI specs from that time? They may give different allowable stresses than what RISA is using, which I assume is based on a current code.

Are the intermediate horizontal rollers in your model for the unbraced length?

 

[JLNJ]

It looks like you used a fixed-ended condition. A pretty grave error which makes a mess of your model results.

 

[bhiggins]

It looks like your top and bottom chords are taking some bending moment due to fixed supports which will increase the stress ratios. Like others suggested, I would idealize the support as a triangle and model it as a pin-roller. I would also model the top and bottom chords as custom WT shapes to gain more capacity and to be more accurate- more moment of inertia to increase Pcr, should increase your capacity significantly.