structural steel design with CSA S16 (Canadian code)

[netjoy]

for sway column, stability analysis should use 0.5% notional load and K=1.0 (compare with AISC 0.42% notional load). But I can't find K=1.0 for sway column in S16. any idea? Thanks.

 

[KootK]

Can you elaborate Netjoy? Are you using the direct design method or the traditional K-factor approach?

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

 

[JoshPlumSE]

Section 10.3.2
The effective length shall be taken as the actual length (k=1.0) for beam-columns that would fail by in-plane bending, provided that....

 

[TLHS]

KootK, in S16 it's not the same setup as AISC. We don't have anything called a direct design method, although he is asking about the comparable method of analysis. Factors and how it ties into everything else in the code is different.

S16 is laid out really poorly for this. Kulak's Limit States Design in Structural Steel is normally helpful, but is pretty silent on this as well.

If you read the commentary, it tells you that when you're using notional loads for stability analysis it's based on a K value of 1.

This gets confusing, because when you go to the compression capacity formulas, it tells you to use K values. However, I suggest you look at section 13.8.2 for combined bending and compression. This is what you'll have to check for anything that has a notional load in it, because it's a beam-column. All three of the checks in this clause modify the factors that relate to K.

(a) Cr uses a value of lambda equal to zero.
(b) Cr uses K=1 and, for uniaxial bending only checks buckling in the direction of bending
(c) Cr only checks weak axis buckling and torsional-flexural buckling as they affect lateral torsional buckling. You presumably use an appropriate K-Value here.

Item b is the overall strength check and it's got K=1 built into it.

As I mentioned before, the pure compression check does use K values. So a strict read of the code has you check the member for compression only using the K value method in section 13.3. Then when you check combined compression/moment the K value is set to 1 as per 13.8.2.

 

[KootK]

TLHS,

You've rightly assumed that I've got an AISC bias built in. I spent my twenties state-side. I'm back practising in Canada now, however, so I find your comments to be both relevant and helpful. Like you, I find S16 to be a bit confusing when it comes to this topic.

The reason that I asked about the analysis method being used is that the name used for the direct analysis method in Canada is the thing that confuses me the most. The commentary refers to it as the "notional load method". This, even though 8.4.1 and the commentary both require designers to use notional loads for both the K-factor method and the direct analysis method.

I also take issue with the several places in S16 where they say something to the effect of "notional loads deal with sway buckling". I think that notional loads are just loads representing initial misalignment etc. That's why we're supposed to include them in both design methods. It's the use of a second order analysis algorithm that converts buckling into amplified moments for design purposes. At best, notional loads simply provide a perturbation for load cases with no direct lateral loads.

While the code provisions are awkward to parse, I do believe that they are consistent amongst themselves with regard to the use of K=1.0 for the direct analysis method. Consider:

13.8.2 a. Length doesn't come into play for this check so neither does K. It's just a section stress check modified to account for residual stress. No conflict here.

13.8.2 b. This specifies K=1 for overall member buckling about both axes which is consistent with K=1 for the direct analysis method. All good.

13.8.2 c. For out of plane buckling modes and buckling modes involving torsion, we're to use a K value that suits the situation, as you've indicated above.

13.3 Any compression only check is, by definition, a non-sway check. And that means that K <= 1, making K=1.0 conservative.

KootK

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

 

[netjoy]

to KootK: cisc 2nd analysis VS aisc 1st amplification (the most similar approach)
to JoshPlum:10.3.2 is for strength check, 10.3.3 is for buckling [highlight #CC0000]"the effective length ... shall be based on the rotational and translational restraint afforded at the end..."[/highlight]
to TLHS: 13.8.2 (a), (b) is for strength. 13.8.2 (c) is for buckling and come back to 10.3.3.

I compared the the traditional method and the new approach, and consulted the CISC, the answer is K=1 for sway column in stability analysis. the only problem is: it is not support by CSA S16, because clause
10.3.3 notes that we shall use the effective length (K>1).
I discuss with couple engineers, most didn't realize the different between the stability and strength as they use to deal with concrete structure.

 

[KootK]

Thanks for the update netjoy. How terribly confusing. So to clarify:

1) Based on a strict interpretation of S16, there is no method equivalent to the direct analysis method (DAM) allowed in AISC. For sway column design K <> 1.0 in general.

2) CISC has given you verbal approval to use a method equivalent to DAM using K=1 for everything. This, even though we don't have any code guidance on the member stiffness reduction modifiers that need to be incorporated in such an analysis.

Have I got that right?

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

 

[netjoy]

To Kootk.
1. S16 not include DAM, but AISC include 2nd analysis (0.2% notional load with effective length) and 1st amplification (0.42% notional load with K=1). for sway column, always K>=1, and as per S16 clause, K>=1.2 (see annex F).
2. Compare to the code's requirement, K=1 is unsafe. It's OK to use a conservative method, but not the reverse. Even I got answer from CISC and I know using K=1 is no problem, it's still "do not follow the code" if I use K=1 for sway column.
The reviewer can say "You do not fllow the code and your method is unsafe comparing with code's requirement"