Why is the maximum sagging moment being used in the LTB checks when I have restrained the top flange? 1

[BA123]

Hi,

I'm posing this question, as it seems that Tekla Structural Designer is using the maximum moment when checking the LTB of the bottom flange, for a fixed ended beam, this max moment in the section can be a sagging moment (i.e. bottom flange in tension), if it has the greater magnitude.
I would have thought the logical moment to use for checking LTB of the bottom flange is the hogging moment, where the bottom flange is in compression. There is a tekla support article (which can be found by googling my thread title), that references "Annex G for BS 5950 and Section 6.3.2.4 of EN 1993", but I'd like to know why the max moment is used regardless of which flange is in tension/compression and not just "this is the code".

Can anyone explain the reasoning behind this?

Thank you

 

[human909]

Maybe this is all dealt with somehow by the code. It would be nice to get to the bottom of it one day.

I'm not sure it is, I'm not sure when I'll find time to get to the bottom of it. Certainly not these next two months. I'm too busy!

I originally called out @Smoulder to substantiate his comments. But I'm certainly not ignorant to the imperfect nature of code (AS4100 or others).

As noted in this thread here, I don't trust AS4100 treatment of stiffeners as restraint for deep beams. I will be having a look at this in the next month or so as I intend to provide a much more robust twist restraint on my portal beams. (This is for the same structure I mentioned earlier supporting 900T.)

 

[Tomfh]

I'm not sure it is, I'm not sure when I'll find time to get to the bottom of it.

Read the link I posted above. It provides a clear summary of the L restraint issue and includes a clear example showing how an elastic buckling analysis with L restraints can yield significantly lower capacity compared to the simplified "effective length" method, which is kinda weird if the simplified method is conservative. It'd be like if the column design equation came out giving higher buckling loads than Euler buckling predicts. As far as I can tell it's the same thing Smoulder is talking about.

Why don't you trust the twist detail, assuming it satisfies the 2.5% rule?

 

[human909]

Read the link I posted above.

I have read it previously. And I was heavily involved in the discussion that lead to that good blog post by @Agent666.

My statement above was a reply to your "Maybe this is all dealt with somehow by the code". I said I'm not sure it is. So I believe we are in agreement.

Why don't you trust the twist detail, assuming it satisfies the 2.5% rule?

Well for starters the 2%/2.5% rule is a cheat itself. It is about stiffness not about strength. Secondly AS4100 doesn't explicitly require the 2.5% rule for stiffeners. Thirdly all this is the assuming the column is perfectly laterally restrained is also a big assumption. Lastly, rigour. I don't mess around with buckling. I've dealt with all sorts of funky structures prone to buckling that I know enough to know that I don't know enough! Cylinders under external pressure (or vacuum) are a fun buckling experience.

So no I don't trust the twist detail, so I'll investigate it.

 

[Smoulder]

@Tomfh Thanks for link to engineers vs sheep. Almost same analysis including beam size. I shouldn't have bothered.

I still think it's misworded in AS4100. Seems that people on or around code committee didn't think it worked with moment reversal when written but no question it is correct for single curvature. Also don't think there has been new research. Would have to ignore foreign literature including Yura in 2010 and be lucky that it turned out to be right. But does it come up often? I'm not really into steel much but haven't come across it governing in my work. Just something filed away in case one day.

@human909 I used to have derivation for kt factor for P restraints but have deleted it. Would be relevant for your stiffener issue. Pretty sure it was Distortional Buckling of I Beams by Bradford.

 

[human909]

But does it come up often? I'm not really into steel much but haven't come across it governing in my work. Just something filed away in case one day.

Not really. For starters most steel construction still has flooring roofing that would do a decent job restrain the member and fly bracing where it does not.

I DO deal alot with steel which has no roofing and often no flooring or flooring that I won't count upon. And I still don't find myself obviously pushing the behaviour beyond AS4100.

Though when it comes to members without lateral restraint especially cantilevers or columns subject to bending AS4100 leaves me high and dry. I have to delve elsewhere for reassurance that my designs are suitable.

 

[Tomfh]

But does it come up often?

It depends what you mean by how often it comes up. Theoretically, it happens every time someone uses an L restraint to reduce effective length this way—like in that extract from the Portal Frame book I linked.

But like you said, there’s usually enough safety factor and redundancy that it doesn’t actually lead to buckling. So in that sense, it rarely comes up.

 

[human909]

I'd suggest that many lateral restraints do also a decent twist restraint and you don't need much to when you have regular spaced members.

I've mentioned it before, but I think it is worth repeating that twist restraint without lateral restraints doesn't count for anything in AD4100 which I've always found odd.

 

[human909]

Why don't you trust the twist detail, assuming it satisfies the 2.5% rule?

You got me thinking.... And this would be an easier enough detail to test using FEA eigen value buckling approaches. So I crunched some numbers. 610UB 9200mm effective span. 200kN top centred load with no restraints except at the beam ends. These were the results

3.7073 with fixed ends 9200 LONG (HOGGING MOMENT)
3.2001 bearing with fixed ends 10000 LONG (400 bearing each end) 9200 effective span (HOGGING MOMENT)
1.7385 bearing with 3 stiffeners pairs 10000 LONG (400 bearing each end) 9200 effective span
1.3284 bearing with 1 stiffeners pair 10000 LONG (400 bearing each end) 9200 effective span
0.9528 bearing with no stiffeners 10000 LONG (400 bearing each end) 9200 effective span
0.7371 pinned @ center 9200 effective span

A single stiffener pair wasn't much more effective than none at all.

The bearing connection would have provided some out of plane rotational restraint. The fixed ends would have provided full out of plane rotational restraint.

Overall a quick comparison with AS4100 results all lie within expectations with the exception that stiffeners really don't magically improve the lateral bracing of the critical flange for a long 620UB.

 

[Smoulder]

@human909 I don't understand your results. Not clear what axis is being fixed and which give hogging moment. All cases? Just first two? What is pinned @ centre in the last one?

Sounds like you are comparing F restraints to P restraints. Only significant in AS4100 for short beams not long ones.

 

[human909]

@human909 I don't understand your results. Not clear what axis is being fixed and which give hogging moment. All cases? Just first two? What is pinned @ centre in the last one?

Sounds like you are comparing F restraints to P restraints. Only significant in AS4100 for short beams not long ones.

The comparison wasn't about L restraints. It was a specific reply to address the influence of stiffeners in providing a suitable restraint to the critical flange. I believe this analysis shows that is not necessarily the case.

The only analysis's giving hogging moment are the ones were it is explicitly stated. (The first two)
The next two have a fixed bearing connection. In theory this should give some hogging moment but in practice fixing just the bottom flange struggles to give a full moment connection. Semi-rigid would be the best description.
The last two are clearly pinned. In behaviour. And the stiffener only gives a minor influence in restraining the top flange.

 

[Smoulder]

@human909 Ok would have to see details. Sounds like lots of variables interacting. First two with hogging but full fixity show 9200 effective with bearings is quite different from 9200 idealised. Same for last two. Is three stiffeners giving warping restraint or changing bending moments? Any difference there means twist restraint through stiffeners is minor variable for this geometry. F vs P (kt=1.0 vs kt=1.08 approx) not expected to be significant for long beam.