Double Web Plate Girder Bending Capacity 2

[andriver]

Ladies and gents,

I am in the process of designing a double web plate girder, the girder tapers towards the ends. At the center of my plate girder, where the flanges are the widest is where my questions applies.

I have been treating my double web plate girder as a box girder, using section F7 in AISC. At the center though, where my flange is the widest, the flange sticking outside of my "box" girder are non-compact. I had originally used equation F7-2 for Flange Local Buckling, using the entire section/plastic modulus of the shape. When my calculations were first checked, a more senior engineer said I should be using section F3, where I used EQ F3-1. This lowered my capacity significantly.

I checked my results with a Bentley structural analysis software, that has a double web plate girder section. When I run through it's printout, it uses Section F7 EQ F7-2 for the bending capacity. This jives with my original approach, however I noticed the capacity was significantly higher in the model output. Digging through the numbers, it is because on EQ F7-2, it uses my entire girder flange width as b in the b/tf term. I however had only used the flange width between my box girder as my b in the b/tf term.

My questions are, how would you approach solving this problem. If you agree with treating it as a box girder (even though the flanges outside the box are non-compact), which value would you use for b?

Thanks for any help. Attachment hopefully helps clear up any confusion.

 

[KootK]

I believe the cisc and I assume the aisc propose doing exactly as jiang has prescribed?

I'll track it down when I get back to Canada

So here's what I've found. I read the sections and commentaries of AISC and CISC that I thought would contain relevant provisions. I didn't comb either manual front to back.

CISC. For slender (as opposed to non-compact) flanges, they give you two methods. One is to adjust your stress limit so that you don't reach the elastic buckling stress of the flanges. Per my comments above, that makes intuitive sense to me. Granted, that which appeals to my intuition does not represent the whole of structural engineering dogma.

The other method is to calculate an effective section modulus assuming that only the portions of flanges that would meet the b/t ratios are present. Note that, in the CISC manual, this method is used with Sx (My) rather than Zx (Mp). Nothing in the section is going plastic. I've attached what I believe to be the source document for this method. It's typed and it's long. I didn't spend much time on it but thought that others might be interested in picking up the mantle.

AISC. For slender flanges, they give you F3-2 which, like the one CISC method, is simply limiting your stresses to values which would not result in elastic buckling of the flanges. The don't explicitly provide an alternate method by which an effective, compact section can be calculated. Perhaps something is buried elsewhere in the manual or in another document such as the Stability Criteria of Metal Structures. Interestingly, F7-4 does give an effective section modulus method for tubes kept elastic.

I'm drifting away from OP's question a bit here, obviously, as his flanges are non-compact rather than slender.

As for everyone else's comments, I will roll with F3 because that is the most conservative value, although I was just trying to understand the code better. I did not want to leave any strength on the table.

I would also like to understand the code better. My impression is that both F3-1 and F7-2 are attempting to prevent any form of flange buckling by keeping the flange stresses below the critical elastic stress values via linear interpolation. As such, I propose this as a design method:

1) Calculate Mn using F3-1 and the properties of the entire section (lambda based on flange overhang). However, change the 0.7 to 1.0. My understanding is that the 0.7 accounts for residual compression stresses in the flange. You (hopefully) won't have those in your built up shape.

2) Calculate Mn using F7-2 and the properties of the entire section (b based on distance between webs).

3) Use the smaller of the values in #1 and #2.

This procedure should get you as liberal as possible while still preventing any of the flange elements from buckling locally. The switch from 0.7 to 1.0 should get you an appreciable bump.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[jayrod12]

The cisc two methods result in the same final answer, essentially everytime. So it's really potato-potato to me. I seem to find the lower yield stress easier to do. But understand the concept behind the effective modulus.