Structural Modeling of Untopped Metal Roof Deck

[StructEng20]

What are other engineer's takes on diaphragm modeling of a typical untopped metal roof deck?

Consider 1 1/2" Vulcraft Type B, 20 GA (design thickness = 0.0358in). I use RAM Structural System's Frame module for my lateral analysis of steel framed buildings with the following parameters for this type of roof deck:

Effective Thickness: 0.0358 in
Poisson's ratio: 0.30 (same as steel)
Elastic Modulus: 1820 ksi

Note that the Elastic Modulus is an Effective Elastic Modulus (E') based on recommendations by Bentley:

https://communities.bentley.com/pro...iki/3348/ram-frame---semirigid-diaphragms-faq

https://communities.bentley.com/pro...ive-elastic-modulus-for-a-semirigid-diaphragm

I model this as a semi-rigid diaphragm in RAM Frame, but I'm uncertain whether this is accurate or whether untopped metal roof deck should be modeled as a Flexible/None or Pseudo-Flexible diaphragm where the diaphragm/frame loads are assigned by the user? Load distribution for a Semi-Rigid diaphragm is determined by the parameters listed above through FEA. However, by definition, this qualifies as a Flexible diaphragm per Building Code and loading would distribute based on tributary areas of each frame.

Thought I'd poll the audience to see other's perspectives and practices.

Thanks.

 

[Mjkkb2]

I always model it as Pseudo Flexible in RAM. I found that if modeled semi rigid following Bentley's instructions, the results were sometimes bothersome-not making sense. Based on the building code definition untopped deck is allowed to be modeled as flexible. It takes an extra step to input the percentages of the lateral load for each 'frame' but seems to me like it's the right thing to as you would have done it, say 20 ears ago by hand.
I have also started to rethink using the semi-rigid floors in RAM. I have been getting strange out of plane deflections recently and if justifiable by code I model them as rigid and forget the FEM altogether.

I think the problem with using semi-rigid is the programs assumption of negletcing the presence of all the gravity framing (which I believe is what's been causing my haedache). You can also model all framing as lateral and just pin everuthing which would give you a more acurate model using RAM Frame but that's a pita also.

SO pseudo-flexible for deck and rigid for concrete or concrete on deck is what I have been doing lately....

 

[StructEng20]

Thanks Mjkkb2 - I have had the same thoughts.

However, I question how dated the Flexible diaphragm approach is. The reason being is that the deck has stiffness, it has load-carrying capacity, it has capability of transferring load right?

 

[MrHershey]

No diaphragm is perfectly 'rigid' or perfectly 'flexible'. These are analytical simplifications, you'll always be somewhere in the middle. The question is when you're close enough to one end or the other and can go ahead and use the simplification while still maintaining an acceptable level of conservatism. That's where the codes come in to draw the lines for us. If you meet the code definition for a flexible diaphragm, I'd tend to run with it. Your actual situation may not be perfectly flexible (again, no situation is), but it's been deemed to be close enough.

You're of course always permitted to run things semirigid if you wish to. At least US codes allow you to model actual stiffness whenever you wish to (and requires it if you don't meet rigid/flexible definitions). But I'd be real hesitant to model metal deck as semirigid unless I had a real good reason to. The stiffness of metal deck diaphragms depends on A LOT of factors, including support spacing, fastener spacing, fastening type, deck profile, and deck thickness, among others. That's a lot of things to potentially get wrong and end up further from actual behavior than if you would have just assumed flexible.

 

[Mjkkb2]

I agree with the above. I sometimes use the pseudo flexible approach recommended by Bentley but like I said, unless you model all the supports (gravity) the results will include large out of plane deflections which are not accurate. Like commented above, any system will have some rigidity/flexibility and one just have to be ok with assuming either end of the spectrum. I often (mostly using RAM software) will have a separate model (one with rigid and one with semi-rigid) diaphragms and compare load distribution to my lateral system members. IF I get large differences between the two I just design the member using the highest load.