Elastic buckling analysis

[Catjacob]

Hello,

I am doing the design check of a multi-span steel composite box girder viaduct but struggle to get elastic buckling analysis done.

To check the stability of the bare steel section under self weight of fresh concrete (i.e a steel trough supporting wet conrete), I set up a 3D model formed of shell elements and 3D beams. The 3D beams are used to model stiffeners and steel narrow top flanges, see the first screenshot.

What I want to see is lateral (torsional) buckling of the whole span. However, strutural analysis program shows that at least the first 30 buckling modes are local failure of the web plates between plan bracing, the second and third picture refer. no overall buckling modes are found.

Typical span of the vaiduct is 50m and ALL bracing members are 200mm diameter circular hollow sections.

Does anybody think the web plates are too thin or bracings are not close enough?

http://files.engineering.com/getfile.aspx?folder=7a7e3685-e19d-4cb2-b177-86448316cb10&file=Flesh.jpg

http://files.engineering.com/getfil...d20-ae3d-cf8723b25590&file=Buckling-Mode1.jpg

http://files.engineering.com/getfil...e32-b02b-15b6bdc7ad8b&file=Buckling-Mode2.jpg

 

[ukbridge]

About to rush out, but check more failures past the first 30 eigenvalues. If you have a class 4 section you'll typically find many local buckling failures of the web and flange plaates before you come across the first local buckling model. These modes are typically accounted for in effective section properties, and do not therefore need to be considered.

 

[Catjacob]

Thanks for your reply.

I knew local buckling has not to be considered and it worries me. It is because I am not able to check whether the steel trough would fail in overall buckling until I got the first global buckling mode and its load factor…

Funny thing about it is that if I assign the correction section properties to bracing members which is in fact a bit larger than 200mm in diameter, the first 40 failure modes are local buckling of webs. Increasing web thickness just makes more local web buckling modes.

 

[ukbridge]

How did you get on Cat? Found some sunday reading for you

http://www.atkinsglobal.com/~/media...n-of-the-olympic-park-bridges-h01-and-l01.pdf

Have a reaad of SCI Advisory Desk note AD 330, “Open top box girders for bridges”, New Steel Construction, February 2009 if you can find it on the CIS if you're practicing inn the UK.

After having a think about it, I don't think you're going to get any global buckling failure modes because of the top flange plan bracing.
Plan bracing is a very effective means of restraining LTB as it provides fully effective lateral restraint to the TF, so in theory you shouldn't see any global buckling failure modes. The only exception is in the hogging region which does not have the benefit of the bracing but this (shouldn't) be anywhere near as critical as at midspan due to the concreted span 1.


If you skim through the many modes (a very tedious approach in LUSAS) the first non local buckling mode will (in theory) have a wavelength equal to the spacing of the diaphragms. Remove the plan bracing and you should see a global buckling mode within the first 0-30 eigenmodes. Interesting how the designer used plan bracing, which is not particularly popular with contractor as it interferes with the fixing of reinforcement etc.

I'm also a bit surprised as to your articulation, I'd definitely check the AIP to make sure your boundary conditions are correct. From what I can see all your supports are fixed in translation in all three directions, this is going to attract a lot of locked in stresses. The bridge needs to be able to breathe longitudinally and transversely.

In any case try these things:

1. Remove the top flange plan bracing, and calculate Mcr. This should satisfy you that the plan bracing prevents any global buckling failure modes. Experiment with reducing the thickness of the top flange too, although this will probably turn up even mor
2. Perhaps do a check using the old school BS5400 BEF method, modified for open top box-girders. This is typically very conservative but if it works then at the very least you can move on.
3. Check for any errors in the eigenvalue buckling analysis output, I find LUSAS to be a good software package but sometimes its easy to not pick up problems.

Hopefully some of these things have helped. Another pointer, the strength and stiffness of the bracing has typically very little effect on the buckling resistance if you do a bit of reading about so wouldn't get too bogged down with that.

 

[ukbridge]

*even more flange buckling moddes.

 

[Catjacob]

uxbridge,
i was busy with something else last week and didn't look at my problem further...But I will pick it up next week.

I start to agree to what you think: the plan bracings are very effective to stabilize the steel troughs against global torsional buckling. I remember I read a book a few years ago about the use of plan bracing to prevent global buckling failure. However, that book doesn't give me more details about min stiffness requirement of an "effective" plan bracing system. (Of course, to derive min stiffness requirement is not easy...)

Regarding the articulation, it is not the real arrangement. I deliberately fixed all bearings after the first run/analysis failed to get any results. I wondered whether the bridge deformed too much and iteration failed to converge.

Anyway, thank you for you help. I will try removing plan bracing members to see whether LTB occurs.

 

[ukbridge]

I'm not sure what code you're checking your bridge to. However, PD 6695-2 might be useful to you, which is gives recommendations for the design of steel bridges in accordance with the Eurocodes in accordance with the UK National Annex.

It gives a simplified criterion to check whether or not the plan bracing system is fully effective i.e. whether your effective length is equal to the spacing of the plan bracing joints. It would then be a simple matter to calculate the lateral-torsional slenderness, Lamdaa_LT and then determine the LTB resistance of your box girder section. This would be a simpler way of checking for LTB without relying on your LUSAS model.

Interestingly I designed a warren truss bridge a year or so ago which has plan bracing along the top chord - this caused me and the CATII checker no end in grief getting LUSAS to give us the expected buckling deflected shapes. I recommend calling LUSAS tech support to get some assistance, it might be worth assigning releases at the end of the plan bracing to make sure no moment is taken at the joints.

If the plan bracing will not be removed after concreting (which I imagine it will not) make sure to perform a check to sure its OK when the bridge is open to traffic.

Good luck!

 

[pikku14]

Hi,

I haven't had a chance to read through all the responses yet, so hopefully I'm not doubling up. You might find some use in the paper "EN1993 Practice Paper: Buckling Analysis of Steel Bridges" found in this link:

http://www.ice.org.uk/Events-conferences/Private/Bridge-Design-to-Eurocodes----UK-Implementation

There's a section talking about the pitfalls of elastic buckling analyses, and suggests using simpler models using beam elements instead for flanges to eliminate the flange torsional buckling modes