Old steel truss design 2

[jayrod12]

Hey everyone.

I have this steel roof truss comprised of double angles for all members with a 10' bay spacing for the bottom chord and verticals at the intermediate 5' marks for the top chords. This truss was designed and constructed in the late 60's under Canadian design standards.

I've checked a few of the components and it appears as though the original designer neglected combined action for the design of the top chord members. As far as I know your top chord must be supported at max 24" in order for you to neglect combined action or you must provide sufficient argument that the roof deck can transfer the load to the panel points.

When I run the existing members under the original loading (and using the applicable codes from the day for both loading determination and strength determination) for combined action they fail. Am I missing some provision?

I've used a slenderness of 0.9L/r where L is the panel length (5') and r of the 6x4x3/8" double angles (1.93 in) and the steel is 44ksi yield so I used 24.75ksi for allowable axial stress and 26.5ksi for allowable bending stress. Do these seem correct?

Granted it is not all of the members failing this way just the odd top chord member (I haven't got to checking the webs yet).

Would you consider continuous beam conditions for the top chord when analyzing for the moment? Something like wl^2/12?

 

[jayrod12]

In all fairness the outcome of this thread will make only a small impact on my design of the reinforcing for new loading. Like I said this is for my own knowledge now. I thought maybe there was something I missed.

 

[CELinOttawa]

No! This MUST be to current code or get upgraded, and those upgrades designed with current methods. That was the first thing I said. I was against this as unnecessary from the start.

 

[jayrod12]

I agree with you CEL which is why I said the results of this thread really don't impact the final output. I will be designing it to meet current standards in full but I want to know why it doesn't appear to work originally. everything else about the building seems to be up to snuff. Just not this

 

[SlideRuleEra]

Would you consider continuous beam conditions for the top chord when analyzing for the moment? Something like wl^2/12?

I agree with others that the top cord can be considered continuous... more or less. Instead of assuming "perfect" continuity (wl^2/12), in the past it was common to down rate continuity to a value >8 (simple span) but <12. For example, wl^2/10 was commonly used for cofferdam design. Suggest taking a look (engineering judgment) at the truss connections as assign a number <12 to the denominator of the equation, and use that.

http://www.SlideRuleEra.net

http://www.VacuumTubeEra.net

 

[CELinOttawa]

Sorry jayrod, just responding to Kootk. Should have been more clear.

 

[KootK]

Ah, I see. Then yeah, pretty safe bet the original design didn't include combined stresses in a manner consistent with what we'd do nowadays.

What is the particular mode of failure for the top chord Jayrod? Strong axis buckling? Overall section stresses?

I sketched some ideas in the attached PDF. Compound trussing probably isn't the cheapest way to go for this application, just the coolest way. If it was exposed architecture, I'd definitely consider it. Note that I didn't reinforce the right panels on your sketch. You'll get the idea.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[KootK]

@Jayrod: how do your webs connect to your chords? Gusset plates? Laps between vertical angle legs? The best reinforcement strategy may be dictated by that condition.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[BAretired]

My my my... No. The concern you're even investigating was a nearly unthought of condition in the 60s. Trusses were designed as a set of two force members without combined actions. I have seen a separate check on the the bending actions of the chords separately from the truss analysis, the continuous beam condition you suggested at the end of your original post.

We worry about far more than our predecessors, and as far as I am aware, combined actions were rarely applied before the 1970s. Many steel trusses I have seen where I had access to calcs (industrial buildings for the most part) had ONLY a graphics statics analysis as the basis of their design. That was it, that was all. After that they added some robust panel detailing able to take combined shear forces and occasionally couples.

I would be very keen to hear from some of the older members on this issue. I'm always keen to learn, particularly about older methods.

In the 1950's in Toronto, Ontario, I designed trusses as pin connected members. Normally, for a span of 140', the truss would have been quite a bit deeper than 7'-0" and would usually have been spaced considerably farther apart than 12'-6", possibly as much as 25' or 30'.

Normally, there would have been open web steel joists spanning the distance between trusses. They were spaced at about 6' or 7' centers and the truss panel points would correspond to the joist spacing thereby eliminating bending in the top chord. This was done so that the steel deck could be 1.5" deep 22 ga and the flutes ran parallel to the truss.

It was the policy in our office to use Horizontally Braced Frames (HBF) spaced at about the quarter points of the trusses. These were rigid frames of the same depth as the trusses which kept the trusses vertical, similar to bridging used on open web steel joists.

Offhand, I do not recall designing a steel truss where the load was uniformly distributed along the chord, but if that situation had arisen, I would have taken into account bending using a coefficient of, perhaps wL²/16 for positive and negative moments in the central region of the truss and a little more near the ends.

In Toronto, as I recall, the plan checkers required a graphical stress analysis to be shown on the drawing for trusses. To keep them happy, we included a stress diagram but I always analyzed each truss using recognized non-graphical methods. IMO, analyzing a parallel chord truss by hand methods was much easier and less time consuming than preparing the stress diagram.

BA

 

[CELinOttawa]

Very good... So other than being a non-graphical preference, that is in line with what I understood.

Anyone else able to add to the historical record?

 

[KootK]

@BA: those sound like elegant projects. I don't quite get the HBF's though. Were they square portal frames or cross bracing? Were they installed between all trusses or intermittently with struts connecting them like OWSJ?

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[mtu1972]

I agree with all of what BAretired outlined above. I started work in 1973 with a slide rule. Moved onto calculators but did not have computer access until the late '80's. Trusses were detailed so as to not have loads between the panel points. If the Architect wanted to support something from the truss chords at other locations, then two additional web members were added to create the new/required panel point.

Since everything was done with allowable loads, you could include all of the governing loads (DL, LL, SL, etc.) and check the governing condition. Most roofs were built-up roofs with gravel, so we never really worried about wind uplift, etc.

Through the '70's, it was common to include a Maxwell Stress Diagram on the design drawings so that the fabricator could graphically get the member forces. My guess is this is the graphic analysis noted above.

Were we overly simplistic. Sure! But trusses had been successfully designed that way for decades. Railroad and highway bridges, as well as buildings. Were we missing some secondary stresses? Probably, but we met the standards that were set by AISC, etc.

In my later years, I would often have to check an existing truss for new/added loads or load points. The easiest way was to plug them into a computer. Often the continuity would show overstresses in the chords. I could remove the self weight and make sure the joint loads included the total dead load. Then checking the chords as completely pinned or continuous to the points of chord splices would often give adequate results.

The lateral bracing of the trusses was done with what we called "sway frames". They included struts at the top and bottom chord elevations along with chevron or V bracing. They were placed at the third or quarter points as required to make satisfy the out-of-plane conditions for the trusses.

Most of us lived in 2-D worlds. Worked the plans and the respective cross-sections and elevations with 2-D structural solutions. I realize that it is now a 3-D world and everyone wants to model everything in one model and then hit Enter and see the results. Then whenever something is > 1.000 (and turns red on the model), regardless of how close to OK it is, they just go bonkers.

I often joked that 2D and 2D and 2D equals 6D which is > 3D. I know you aren't going back to the 2D world, but things were really so much simpler then and I feel the added complexities introduced to the Codes are because of what computers can do for the designer, but it seems to be an out-of-control spiral.

And while I am ranting, don't get me started on FEA. It is an approximate method and seeing some of the plate elements turn red does not necessarily mean failure. It may mean that your model needs refining, or you need to take a closer look at those isolated conditions with other methods of design and analysis.

gjc

 

[BAretired]

KootK: The HBFs were rectangular frames having the same depth as the trusses and fitting between them. Web members were usually 'V' shaped with a central vertical, although this may have varied with individual designers.

BA

 

[CELinOttawa]

Beautiful post MTU1972... I prefer 6D myself!

What is your confortable level of overstress? Depending on the analysis method, the application, the material at hand, and what I have made in assumptions, I am often happy at 1.02 and 1.05.

 

[mtu1972]

I've always been conservative with my load assumptions. Min. Live Loads are per the codes; Dead Loads are based on the materials of construction and were generally rounded up. Snow Loads were always a concern up here in MI, MN, & WI and I generally rounded up, especially where drifting was in play.

So I would have no problem with 1.05. Some of my latest supervisors could not accept ANY overstress. We always joked about one in particular "seeing red".

gjc

 

[Triangled]

OP: I've checked a few of the components and it appears as though the original designer neglected combined action for the design of the top chord members. As far as I know your top chord must be supported at max 24" in order for you to neglect combined action or you must provide sufficient argument that the roof deck can transfer the load to the panel points.
i think the 24" may be from the current SJI specification for steel joists based upon testing for their manufacturers, at least, that's the only place I've heard that number in this context.

BA: Normally, there would have been open web steel joists spanning the distance between trusses. They were spaced at about 6' or 7' centers and the truss panel points would correspond to the joist spacing thereby eliminating bending in the top chord. This was done so that the steel deck could be 1.5" deep 22 ga and the flutes ran parallel to the truss.
i agree with BA on this, but/and is this the way your system is? In other words, is there a mechanism for the direct introduction of bending into your top chords?... or are all loads introduced at panel points?

OP: Would you consider continuous beam conditions for the top chord when analyzing for the moment? Something like wl^2/12?
i might. But it seems to me that if it is a system as BA described above, alternate panels may deflect upwards and other panels deflect downwards creating a different KL/r and hence a different wl^2/?. In other words, how is the top chord braced/restrained?

double angles are fun, especially if z-axis buckling is overlooked.

 

[jayrod12]

Wow, lots of action since I was on last. Hopefully I respond to most.

The method of failure that I get for even strictly axial load (neglecting bending) is flexural buckling.

The webs and chords are all connected by ~1/2"(12mm) gussets.

These trusses do have some form of uniform load as the roof deck is 3" 20ga decking that bear directly on these trusses.

There is bottom chord bridging at every second bottom chord joint and vertical cross braces on the bridging lines every 5th truss. Top chord bracing 6 bays out of 44 evenly spaced.

The 24" that I mention existed in CSA-S16-61.

Maybe someone can explain this to me. In my RISA model, when I run the truss (loading at joints only, all members pinned) using the AISC 14th edition ASD and allowable loading I get the 5th and 6th top chord member fail at 1.06 (I could live with that probably).

However when I run the same truss using the CSA S16-09 (my ideal code check) and factored loading the member fails by 150%. When I hand checked the same member using the CSA S16-09 and factored loading I get a code check of 117% in flexural buckling. I've attached the risa model in case any of you want to take a crack at it.