Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Good Frame Analysis Software? 7
- Thread starter BSE05
- Start date
Well, of course there is a lot of theory and references that I'm not going to come up with right here while eating dinner and drinking a beer 
.. but this is the general process.. And all comments are based on ETABS 9 - although it all pretty much applies to general FEA.
In my opinion, it's easier to perform this analysis/design the more detailed, and in a nutshell, the more manually meshed your model is. But that is coming from someone who has a model that cannot be auto meshed- as smart as the software is, it's not that smart. From this experience, and previous projects, I just think it's worth the effort to mesh it on your own and let the computer think hard about important stuff like eigenvalues and participating mass ratios. And you have to use your engineering judgment as to how you model your cores. For instance, if you have openings that result in really shallow link beams, it probably makes sense to model those link beams as frame elements between shell element walls. Likewise, if you have a really beefy link beam in relation to the opening, it probably makes sense to model the opening with shells. Next, I see a lot of people get sloppy or lazy with their elements. Shell elements can only transfer loads between their nodes; so if you have a shell corner framing into a bigger element, the software interpolates deflected shapes and uses some magic edge constraint algorithms to figure it out. But bottom line it's making assumptions and calculations that would be unnecessary if the wall or shells were properly meshed. I never leave a node framing into an edge without resolving it into other nodes in the model. And this also includes nodes not within tolerance. Node snap means there 'IS' a tool that can make your model perfect and error free.
So for instance, I have a model that consists of frame elements (concrete columns and on the lower levels, concrete beams), and shell elements- independent shear walls, cores, and floor plate elements. On these floor plate shell elements I have applied all of my superimposed gravity loads. I like to separate shells by function. So if I know there is a corridor, it's a corridor PT Slab; if it's a mechanical area, it's a Mech PT Slab. The advantage here is you can separately apply reducible and non-reducible live loads based on element type- and you can change the loads assignments wholesale without having to go in and find every different region in a model. Unless your model gets insanely huge there is really not a penalty for more detail.. I digress.
So next, you need your response spectrum function. You build this out of info from your geotech, IBC etc.
Next, you define response spectrum cases. Obviously, you start off with the basics. I'd define EQX and EQY. That means you have a response spectrum case with Func 1 exciting the structure at 0 degrees. Case 2 is Func 1 exciting the structure at 90 degrees. Additional cases would include those in the principal direction of the building- gotta figure that one out on your own. Finally, set your scale factors equal to 100.
Finally for the initial setup- define your load combinations. ETABS will include the vertical component of the earthquake response if you tell it to (SDS factor under special seismic provisions). Or you can include it in your Load Combos (it's just a percentage of dead).
Run it..
You'll be able to get base reactions. We calibrate the model based upon scaling the spectra results to match or be conservative in relation to the equivalent static base shear. (need to figure that one out too..) But basically you have [required Base Shear]/[Base Reaction F1] x [Scale Factor = 100] = new scale factor. It's an iterative process..
And the accidental torsion consideration is for you to figure out. That can (and does) get really proprietary as to how you handle it..
Finally, you run the design module and get your steel requirements.
I think that sums it up.
.. but this is the general process.. And all comments are based on ETABS 9 - although it all pretty much applies to general FEA.In my opinion, it's easier to perform this analysis/design the more detailed, and in a nutshell, the more manually meshed your model is. But that is coming from someone who has a model that cannot be auto meshed- as smart as the software is, it's not that smart. From this experience, and previous projects, I just think it's worth the effort to mesh it on your own and let the computer think hard about important stuff like eigenvalues and participating mass ratios. And you have to use your engineering judgment as to how you model your cores. For instance, if you have openings that result in really shallow link beams, it probably makes sense to model those link beams as frame elements between shell element walls. Likewise, if you have a really beefy link beam in relation to the opening, it probably makes sense to model the opening with shells. Next, I see a lot of people get sloppy or lazy with their elements. Shell elements can only transfer loads between their nodes; so if you have a shell corner framing into a bigger element, the software interpolates deflected shapes and uses some magic edge constraint algorithms to figure it out. But bottom line it's making assumptions and calculations that would be unnecessary if the wall or shells were properly meshed. I never leave a node framing into an edge without resolving it into other nodes in the model. And this also includes nodes not within tolerance. Node snap means there 'IS' a tool that can make your model perfect and error free.
So for instance, I have a model that consists of frame elements (concrete columns and on the lower levels, concrete beams), and shell elements- independent shear walls, cores, and floor plate elements. On these floor plate shell elements I have applied all of my superimposed gravity loads. I like to separate shells by function. So if I know there is a corridor, it's a corridor PT Slab; if it's a mechanical area, it's a Mech PT Slab. The advantage here is you can separately apply reducible and non-reducible live loads based on element type- and you can change the loads assignments wholesale without having to go in and find every different region in a model. Unless your model gets insanely huge there is really not a penalty for more detail.. I digress.
So next, you need your response spectrum function. You build this out of info from your geotech, IBC etc.
Next, you define response spectrum cases. Obviously, you start off with the basics. I'd define EQX and EQY. That means you have a response spectrum case with Func 1 exciting the structure at 0 degrees. Case 2 is Func 1 exciting the structure at 90 degrees. Additional cases would include those in the principal direction of the building- gotta figure that one out on your own. Finally, set your scale factors equal to 100.
Finally for the initial setup- define your load combinations. ETABS will include the vertical component of the earthquake response if you tell it to (SDS factor under special seismic provisions). Or you can include it in your Load Combos (it's just a percentage of dead).
Run it..
You'll be able to get base reactions. We calibrate the model based upon scaling the spectra results to match or be conservative in relation to the equivalent static base shear. (need to figure that one out too..) But basically you have [required Base Shear]/[Base Reaction F1] x [Scale Factor = 100] = new scale factor. It's an iterative process..
And the accidental torsion consideration is for you to figure out. That can (and does) get really proprietary as to how you handle it..
Finally, you run the design module and get your steel requirements.
I think that sums it up.
mattmlm
Structural
- Oct 13, 2005
- 18
- 0
- 0
jen4950
How long should ETABS take to run the analysis of a 30 story, 2 million sq ft, concrete building? I'm not very familiar with the program, but it is used in our firm. During a conversation the other day someone mentioned the model taking a day or two to run. Can this be right?
MLM
How long should ETABS take to run the analysis of a 30 story, 2 million sq ft, concrete building? I'm not very familiar with the program, but it is used in our firm. During a conversation the other day someone mentioned the model taking a day or two to run. Can this be right?
MLM
I'm running a 5 million S.F., extremely complex and irregular building- flat PT plates, 3 independent cores, 3 conventionally framed concrete lower levels with additional and generally undefined steel braced frame structure framing into it; building splits off into two towers about mid height; and the 2 independent towers are different heights and have different intrastory heights- so they fight eachother, and finally 2 additional wings have drastically different geometries; and no expansion joints- 900 feet tip to tip. All in a seismic zone. Sleep at night will be much easier in about 4 years after it's proven itself.
That building takes 9 hours to run- about 60,000 manually meshed shells (number sticks in my head, I manually meshed ALL of them LOL! - with the help of Copy-Paste between levels). And I don't know the number of frame elements or DOF's off the top of my head.
That building takes 9 hours to run- about 60,000 manually meshed shells (number sticks in my head, I manually meshed ALL of them LOL! - with the help of Copy-Paste between levels). And I don't know the number of frame elements or DOF's off the top of my head.
- Thread starter
- #30
Thanks for all of the great input. After playing with the demo, I recently purchased Ram Advance and am very pleased with what it can do over a wide variety of building material types. Its retaining wall design feature is also a plus. Also relatively low cost.
I was more impressed with the retaining wall module inside Advanse than Advanse itself. That is the program I use for retaining walls now.
Insanity in individuals is something rare - but in groups, parties, nations and epochs, it is the rule.
-Friedrich Nietzsche
Insanity in individuals is something rare - but in groups, parties, nations and epochs, it is the rule.
-Friedrich Nietzsche
EKARITSIRACHAINAN
Structural
MIDAS/Gen is another choice for structural engineer.
for more informaion, please goto
for more informaion, please goto
- Status
Similar threads
- Locked
- Question
- Locked
- Question