sideway unprevented steel column per Canadian code (CISC) 1

[netjoy]

it's no nomograph for unbraced column in CISC handbook, but appendix F give K for some unbraced columns. I wonder if "effective length" method is allowed to design unbraced column as other code (eg AISC), or had to use P-delta method (not sure where I saw this).
Thanks.

 

[JoshPlumSE]

I'm looking at an older version of the CSA code (1994), but my belief is that these provisions are largely unchanged.

Clause 8.6 (Stability Effects) does require a P-Delta analysis.

Clause 13.3 (Axial Compression) does assume that you will be using K factors.

In this sense, I believe the code is more similar to AISC's LRFD 2nd or 3rd edition codes than it is to the 13th or 14th edition codes (which almost always use K=1.0).

 

[jayrod12]

Clause 8.4 in the CISC S16-09 Talks about stability effects, Clause 9 discusses stability of members. Clause 10 talks about design lengths and slenderness.

Your column's slenderness (KL/r) cannot exceed 200 for compression members. If you have a cantilever column your K=2.0 (from Annex F). K is also included in the determination of your Cr from Clause 13.3.

Part 4 of the steel design manual (particularly Table 4.4) gives factored compressive resistances based on KL/r and Fy. And Table 4.8 gives amplification factor U to use in clause 13.8.2 and 13.8.3.

I believe this amplification factor and K account for the p-delta but I could be wrong (it's been known to happen just don't tell my wife).

Hopefully that helps,

 

[dik]

Jay... You are correct... you can tell your wife...

Dik

 

[jayrod12]

Thanks Dik,

I probably won't tell her that either. It's always better when she thinks she's right but I know different and keep my mouth shut.

 

[netjoy]

Thanks.
I saw an explain why CISC handbook no include nomograph for unbraced column, it said K method is not suitable for unbraced column and second-order method is required. But I can't remember where is it and can't find it again. So I try to get some Authoritative explain.

 

[JoshPlumSE]

AISC has a good presentation from Shankar Nair on their website about the design of structures for stability. As part of that presentation, he goes into the limitations of the K factor method.

http://www.aisc.org/UploadedContent/OnlineVideoCourseFiles/Set_2/part1.html

It's a fairly lengthy presentation and is pretty dense. But, if you can get through it, then you should have a good understanding of the benefits of a 2nd order analysis over just assuming a K value.

In particular, Part 2 slides 40 through 53 discuss some of the problems with the calculation of K and relying on nomographs and such.

 

[netjoy]

Thank you JoshPlum.

I had been studied "the theory of stability" and knew some of this thing. But back to the code, when use CISC Clause 13.8, we need calculate Cr.

Second-order can find out the more exact force/stress in the member and when the member will fail, but can we use the loading if the stucture was not fail in analysis? I don't think so. Because this loading capacity is based on the theoretical section capacity. and we need the the design section capacity to complete the design.

In one word, my point of view is: second-order in a way to calculated the force/stress, we need an indication how to get the section resistance.

 

[IDS]

Thanks Josh, excellent presentation.

Also useful to have the steel design perspective for someone who works mainly in concrete.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[JoshPlumSE]

Netjoy -

One of the points of the presentation is that term "buckling load" is inaccurate. The real behavior of most steel structures is really one of axial force amplifying the existing bending moments. At some point (the point of "buckling") this amplification becomes infinite.

If you do a second order analysis, then you will account for this moment magnification in the demand side. Your connections and splices and such will be designed to more realistic forces and moments than if you relied exclusively on the K value to protect you from buckling.

The AISC code (and that presentation) goes farther than the Canadian code does. It goes into the "Direct Analysis Method" which allows you to use a K value equal to 1.0 for all columns... So, long as you account for initial imperfections (that the reason for the notional loads) and some material non-linearity due to residual stresses and such (this is the reason for the reduced stiffnesses).

To my knowledge, the Canadian code does not have either of these caveats. So, we have to use our engineering judgment, of course. IMHO, using K=1.0 for all members would be a stretch. However, it is clear that a 2nd order analysis already takes into account much of what the K factor is intending to do. Therefore, using both means that you are liking double dipping a bit and may be a bit over-conservative.

No definitive solution from me! But, if I have confidence in the 2nd order analysis that I'm doing and I've got decent lateral loads, then I don't feel the need to be extra conservative with my K values. I'm looking for reasons which would justify using a lower K value to take out some of that conservatism.

 

[netjoy]

Thank you so much, JoshPlum,

per the video you post, AISC have 3 methods for this situation, K=1 or K>1 based on analysis methods and notional load apply on.

CISC used similar method as AISC with high notional load, but AISC indicate K=1 for this method and people can't find indicate in CISC. More confuse is the code give example for K=2 and note "K=0.65 to 2.0"

Now I pretty sure that no K>1 should be used if follow CISC S16-05 (S16-09)

 

[JoshPlumSE]

For AISC, there are indeed multiple methods. There is a method which involves running a P-Delta analysis, using K = 1.0 and applying notional loads. We'll call this method the "design by 2nd order analysis" method. One important restriction of this method is that AISC only allows it if the ratio of 2nd order drift to 1st order drift is less than 1.1. Meaning that the drift amplification you get from your P-Delta analysis is less than a 10% increase. If it is greater than that, then AISC requires that you go back to user higher K values.

Since most moment frames result in about a 10% increase in drift from P-Delta you don't really know whether you're going to be allowed to use the K=1.0 or not until after you've gotten a good way into your design.

Extending this out to the CISC code (which is not so explicit as the AISC code) makes some sense. But, the decision becomes less clear cut than what you suggest in that last posting.

 

[netjoy]

AISC use k=1 when 1)second-order analysis w/ notional load =0.2% and EI & EA reduced; 2) first-oder + amplified w/ 0.42% notional load and limited delta and axial load
CISC use K=1 for second-order analysis w/ notional load 0.5%.