Computer modeling

[SteelPE]

I have a somewhat complex structure (at least for me) that is about 300’ long x140’ wide. It is basically a 2 story structure with multiple roof elevations with a portion of the structure is curved (see attached sketches). There are braces and moment frames all over the structure (which you may not be able to see).

After consulting a mentor it was decided that the structure should be run as one piece w/o an expansion joint. The building was modeled using a common program. We usually use the program to design gravity members and calculate lateral loads. Lateral frames are checked but are also run in another 2D program for verification.

During the frame analysis, the program was telling me that I had a pdelta problem on the left side of the building. This would lead me to believe that the structure is unstable however; initial deflections are around 1/4” (from the 3D analysis ignoring pdelta) and my masses are small which would lead me to believe that I don’t have a stability problem.

I did everything I could to try and eliminate the problem… even put in dummy bracing to see if I could eliminate the error. Nothing worked. Then I decided to split the building at the location shown in the sketch (where I had wanted to split the building in the first place before talking to my mentor). With the building split, the building no longer has any stability problems.

I can either physically place an expansion joint at the shown location (which would be a pain architecturally)…. Or I could continue with my modeling and then apply the loads as I see fit throughout the structure in my 2D analysis keeping the structure in one piece (considering pdelta effects in the 2D analysis). Any other ideas?

 

[gobsmacked]

If you could mark up page 1 of your sketch to show the positions and extent of the vertical bracing, that might help understand why you might be getting a pdelta prob. Or maybe not. If cutting the structure makes the problem go away, doesn't that suggest that your first computer input might have an error? What checks have you done? In general, I would expect that cutting the structure would increase its flexibility, not reduce it.

 

[SteelPE]

Well, since you asked for it, here are the frame types and their relative locations. We are still in the preliminary design phase so the frame locations may change.

 

[steellion]

Make sure that the P-delta error isn't just a localized failure of a member, like if you apply a uniform lateral load to the weak axis of a member.

My typical M.O. for computer design is to design without P-delta until the sizes get close to their final size and I have a stable structure, then switch to P-delta. It relieves a lot of headaches early on, because if you have a member that is way undersized, it will give you an error every time.

 

[dik]

I'd probably put an expansion joint to the right of the line drawn.

Dik

 

[msquared48]

I assume the long frame to the left of the joint has a short drag link to get the forces to it due to the apparent opening in the diaphragm?

Mike McCann
MMC Engineering
Motto: KISS
Motivation: Don't ask

 

[SteelPE]

msquared,

yes, there is a beam across the entire from of the building at the floor elevation to act as a drag strut at the moment frame. There is also a drag strut at the brace frame at both the front and back of the building (at the change in roof elevation).

steelion

The analysis works W/O pdelta. However the buildings drifts at the elevation in question appear to be small 1/4" with the mass at this portion being about 100k (not what I would consider very large). Maybe it will show up in the 2D analysis... but I have a suspicion it will not.

The odd thing is that it doesn't show up when I split the building in two and use the same frame sizes.... I could go back to tech support... but I don't have a week.

Roof load distribution will not matter since there are so many changes in elevation. When I get down to the floor there may be some problems because the floor will want to act as a rigid diaphragm and redistribute the forces according to the frame stiffnesses. I can try to fudge this into the 2D analysis because I don't believe this floor will transfer forces in this manner.

 

[a2mfk]

First, interesting building, what is it for? The second floor layout and roof almost looks museum-ish...

I thought the same as Dik when I looked at your building. Regardless of your software modeling issues, I think I'd want an expansion joint there, maybe one in both directions to isolate them structurally and allow for movements. Especially since architecturally and ultimately structurally this thing is two buildings connected by a small bridge of roof and floor from what I can tell.

I'm not 3-D software literate, but seems like you need to get your mentor to look over your shoulder at everything you are doing to see if the two of you can figure it out. My RISA/RAM experience has shown me it can be so nit-picky you need to run through it with someone that can see all your assumptions and box checks, etc.

Great article I found since I have a similar roof plan I am working on that I am putting in expansion joints, and your question got me thinking:

http://www.modernsteel.com/Uploads/Issues/April_2005/30740_fisher.pdf

 

[SteelPE]

a2mfk

Thanks for the article, I will look at it when I get to work tomorrow. The building is a church. Actually it will be three buildings (if I decide to put in the 2nd joint which is likely) I just don't show the other structure behind this one (you can see the some of the grid lines though).

Unfortunately when it comes to my mentor, I am his mentor for computer modeling in both 2D and 3D. I deciphered all of the stability provisions in the AISC 2005 spec, then I had to figure out how to properly implement them in the programs that we have. Then I had to teach him how to properly use them in the programs. It took a while.

 

[enko]

I agree with a2mfk
I am convinced that one building is much more stiffer than the other.
Afterwards they have to be connected with each other at a single point.

Check the P-delta options of the software, maybe initial conditions (just a suggestion)