Applying seismic load to FEA model

[GalileoG]

Applying seismic load to FEA model

This is probably a question that is specific to STAAD.Pro, but perhaps not.

I am wondering how everyone applies seismic load to their models (of buildings) when using the equivalent static force procedure. As an example, I have a large steel industrial building with a metal deck roof and horizontal roof bracing. I have seen some people use the ‘master and slave’ function to assign this type of roof as a rigid diaphragm and then apply the base shear as a point load on the center of mass of the building. I do not believe this is an accurate representation of how lateral load is being distributed by the roof diaphragm (metal deck w/ horizontal bracing.) Also, this method would not provide the load demand on any of the roof bracing, as the bracing is not being engaged in the model. The only time I see the ‘master and slave’ function being useful is when you have a reinforced concrete slab. I believe the most accurate way of applying seismic load to a model when using the equivalent static force procedure would be to apply it to each individual frame in proportion to its mass. However, with accidental eccentricities and 4 different directions, I can have 8 seismic load cases – this can be quite labor intensive. Is this latter approach the correct approach that one should use for a building of this type? Your thoughts please?

Thank you.

 

[ishvaaag]

Well, it seems quite reasonable in the purpose of extracting some EQ forces for the roof system.

Lately I have started to see some texts where when the model does not make use of rigid diaphragms, but actual stiffnesses anywhere, accidental eccentricities are forfeited. Something of the kind with proper fine-tuning will be a much welcome simplification for EQ combinations; in the end 5% of eccentricity must not make much for one good lateral resisting system and almost any theorization covering the issue should be even better than the messing eccentricities. However, the thing seems not to have filtered to the usual codes as yet.

 

[SethGuthrie]

There are ways to include the effects of accidental eccentricity when the diaphragm is treated as a semi-rigid plate, but it's not simple. It's actually something RAM Structural System automates now.

I still believe that a rigid diaphragm (or master/slave approach) is good for many roof cases where the layout of the frames is well balanced and the rigid diaphragm distribution of the loads to the frames is "good enough". Using a rigid diaphragm simplifies a lot of calculations, prevents interaction of the frame beams and the diaphragm elements, speeds up the analysis, etc.

If you are looking to design the rod braces too, I wager you could check them by hand with the time you just saved.

 

[asixth]

Just out of curiosity, what is the magnitude of the mass for the metal deck roof (50kg/sq.m?). Master-slaves are good when the diaphragm transfers the seismic force to the lateral system, if you have bracing to do that then I would be questioning why master-slave is being used.

Personally I am a response spectrum person myself. It gets the seismic forces down low.

 

[DST148]

@GallileoG : I presume the roof bracing in plan does not support gravity loads. The roof framing may be analysed independent of global analysis. As a starting point, you may convert the total diaphragm (seismic)load at the roof level as udl on the entire roof and assign point loads to the panel points based on the tributary areas. Then you may increase panel point loads on one side and decrease on the other half of the roof to account for eccentricities. Four seismic load cases and sixteen load combinations will be required to get the design member forces in the roof framing.

 

[Gumpmaster]

Can you idealize your diaphragm as flexible? If so, then there's no torsion effects right? If it's rigid (or semi-rigid) then you can just run a straight up analysis with no accidental torsion and account for the additional frame loads with some hand calcs and point loads. Pretty simple.

You can idealize your diphragm as a flat plate with the proper stiffness to match your metal deck + bracing. If it was only metal deck there is a method to determine the proper plate thickness. With the horizontal bracing, you'd need to run some hand calcs to determine the correct stiffness of the plate. If you go this route, the plate will pick up some of the gravity loads from your roof beams and girders, so watch out that those aren't under-designed (it's worth doing a hand check).

All of the lateral loads on the out of plate walls (in plane walls too) can be captured by line loads on the roof/floor level beams.

Lateral loads on the roof system and on the modeled members can be captured by the selfweight X=Cs or Y=Cs (where Cs is your lateral coefficient) command.

This is the common practice for an ELF analysis. It is very similar for a response spectrum analysis, but you need to capture your masses with reference loads.

 

[GalileoG]

Thank you all for the responses. They have been very helpful.

The National Building Code of Canada states that accidental torsion must be applied regardless of whether the diaphragm is flexible or rigid. This makes no sense to me, because a flexible diaphragm can not develop torsion. I wonder what the IBC has to say about accidental torsion and flexible diaphragms?

 

[DST148]

@GalileoG - Accidental torsion is to account for displacement of the center of mass. For flexible diaphragm, you may visualize it as a beam with non-uniform loading that would produce higher shear at one end and reduced shear at the other end. Additional shears due to torsion(displacement of center of mass) must be taken by LFRS in that direction only, LFRS in the orthogonal direction does not take shear due to torsion.