Is 3D analysis/design modeling really necessary 5

[mtu1972]

As background information I am an "experienced" structural engineer but a relative newcomer to my present firm. We design industrial facilities in various locations accross the US.

Younger engineers/supervisors feel that it is essential to create the complete 3D building model to design any new structure. This means 8 directional wind load cases, additional wind load cases for various roof pressures and suctions, ditto for seismic (whether it will govern or not), secondary members such as girts, equipment loadings, floor openings, etc.

We do not work on essential and/or highly irregularly shaped buildings.

We do not do BIM or 3D CAD drafting unless specifically requested by the customer.

Am I among a dying breed in thinking that the lateral load resisting elements can be done in 2D frames in the two perpendicular directions and the remaining infill framing can be designed using only gravity loads and conventional design methods?

GJC

 

[msquared48]

On second thought, it may have been New York.

Mike McCann
MMC Engineering

 

[rowingengineer]

I do believe this has been the subject of both T.V and movies scripts.

Are you talking about the citigroup center, or what ever it is named these days. 45 degree's was the problem but it wouldn't have been found in the 3D model I don't think as this was a detailing problem. As they say the devil is in the detail.




Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that them like it

 

[ChadV]

http://en.wikipedia.org/wiki/Citigroup_Center

"In 1978, prompted by a question from a Princeton University engineering student, LeMessurier discovered a potentially fatal flaw in the building's construction: The original design's welded joints were changed to bolted joints during construction, which were too weak to withstand 70-mile-per-hour (113 km/h) quartering winds."

 

[apsix]

http://en.wikipedia.org/wiki/Citigroup_Center

 

[rapt]

Mike, Rowingengineer, etc,

I remember the story, but, unfortunately, not the building or its location.

And I agree, these days, 3D analysis is essential for multi-storey horizontal load analysis and also for horizontal load analysis for any complicated structure.

The problems come in when inexperienced engineers are left in control. Normally the young engineer in the office is given the task of doing the analysis, because the experienced ones are too slow with the computers and graphical input programs, but he does not understand a lot of the practicalities to do with construction, connections etc so does not really know how to set the boundary conditions for the structure.

When it comes to vertical load effects and concrete buildings, we really start to have problems with this. People who have no idea how to design and what the answer should be are running programs that produce structural drawings from a BIM input. The computer has done the complete design and detailing!! This is where I have a problem. The computer is a calculator, not an engineer. While it is used as a calculator by experienced engineers, they can be very useful. They know what settings to make to model real conditions and know what to look at in the results to make sure the design is suitable. But, these days of low fees and short deadlines, all of this is being ignored and untrained inexperienced engineers are putting out final designs with inadequate supervision and checking, beacuse the computer will get it right!

I seem to be seen to have an anti-FEM leaning on this list. That is incorrect. I have an anti "untrained engineer using computers unsupervised" leaning.

Computer software for engineers has created a gigantic pool of engineers who are expert in everything because they can push the buttons to get a result for any analysis and many design situations. Unfortunately, it has not increased the understanding of design for many of those, because they are not inquisitive enough and /or do not have time to learn from the programs. And design engineers need to understand design before they can be counted as that. Being able to run a computer program does not make someone a design engineer.

I do not apply this to just FEM software, it is all software including RAPT which we produce. A large number of the people using RAPT have never done PT design by hand. Many have never questioned its results or interrogated it to try to get a better design. They just accept what it gives out. This is dangerous. I commented to someone the other day that when we started selling RAPT in the mid80's, there were probably 15-20 really experienced building PT designers in Australia (all designing using hand calculators). Now there are 3000-5000 who say they are but there are probably really still only probably 50, possibly less who could be put in that class. The remainder can produce a design. That was not our aim in developing RAPT but it is an unfortunate consequence of its and other program's availability. The rest of the world has the same or worse problems. At lease our design code is logical for PT slab design. Anyone who designs PT slabs to ACI318 code does not even have that benefit.

And RAPT is 2D. Multiple this by 1000 and you can see the problem with FEM design software.

 

[hokie66]

Now that's telling it like it is. And you've got to be able to think in 3D in order to benefit from 3D analysis programs.

 

[rowingengineer]

amen




Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that them like it

 

[asixth]

"The computer is a calculator, not an engineer"

This is one of the best quotes I have ever heard (structures-wise).

 

[haynewp]

This may be of some assistance.

http://www.johnmartin.com/publications/ExactAnalysis/2D-3D-Paper-As-Published.pdf

 

[seny]

"And you've got to be able to think in 3D in order to benefit from 3D analysis programs" 35 years ago (i do not know what was before) all the structural who was using 2d needed to think 3d too. structural always needed to think 3d to have ability to prepare more or less real load schematic.
i'd like to see somebody who currently involved to high rise design and will be able to prepare good model using 2d (torsion is killer of irregular structures). in good prepared FEM 3d at some point we can be surprised w/ something like shear walls supported by floor diaphragm - what was almost impossible to catch in 2d at all using typical tributary and distribution of industry practice. I am voting for 3d.

 

[rapt]

Seny,

I agreed above about 3D analysis for complicated buildings.

But how do you design it?

Do you go through the entire building and design every element/member for combined flexure + shear + torsion. or do you use Etabs and assume that it has done all of this correctly for you!

Do you reduce torsional stiffnesses where you do not think they are appropriate or cannot really be developed.

Do you downgrade slab/beam connections to the core because they cannot be detailed to generate the connection assumed?

Do you allow for creep and shrinkage shortening effects in your vertical members?

If there are transfer beams, do you use an overall 3D model to predict the structure response.

If there are transfer beams, do you use the reactions from your 3D model to design the transfer beam? This significantly underestimates the real load on the beams.

For the last 3 above, you need to do a construction sequence analysis to get it right. I bet you do not!

A 3D model of a completed building constructed from an elastic homogenous material bears no resemblence to what you are building and how you are building it. So is your model any more accurate than a 2D model? Which is more conservative? Who knows! For reactions on a transfer beam, a manual load takedown is slightly conservative but a 3D model is unconservative by about the same amount. The average is about right.

 

[80smetalfm]

Ive modelled 3D buildings esp for seismic using ETABS. My comments are as follows and revolve around ETABS:

- Understanding mass source, response spectrum functions, no. of modes, load case & combinations is an issue with 3D modelling esp for Seismic areas.
- Detailing of moment connections, torsion issues can be an issue esp when replicating.
- Making assumptions is crucial hence the graduate Engineer really requires supervision. e.g. When to put springs at base of buildings & against rock batters to basement of high rise buildings!!
- GIGO will always rule. garbage in garbage out!!
- Normally with ETABS it takes me about 1 week to model a 20-30 stories building & from 2-6 weeks to annalyse the whole structure!!

 

[seny]

rapt,
thank you, for comment.
my short answer can be found in my previous post - critical members i always double-check by hand calculation (and it is not necessary to discuss that lintel above the door opening is important in structural field as everything else) and i do my best to prepare the most realistic model in any kind of soft - 2d or 3d.
i agree 100% - "The computer is a calculator, not an engineer", but it's about any kind of soft - 3d and 2d.
i always think that structural engineer who is carrying so much responsibilities must learn every day everything new what exists on the market, does not matter how experienced he / she is. i use to do calculation and design manually, in 2d and 3d. use to use different soft starting from main DOS soft to soft family of RISA, STAAD, VA, CSI and many other.
Can tell you, and for sure you know it as well, no one who is really involved into real design never trust some one calculation software, it is always number of checks and double-checks, using different soft and by-hand as well.
Soft 2d or 3d is only tool.
Finally, back to subject, counting everything I personally still vote for 3d.
One more time thanks for your respond.

 

[271828]

rapt, in your last post, it seems to me that you're not really describing an issue with 3D vs 2D modeling. Instead, I think you're describing depth of knowledge of concrete building behavior. The same guy who won't do all those things correctly in a 3D model will also miss most of them in a 2D model.

Granted, a guy will catch a few things in a 2D model that he won't catch in a more elaborate model. My favorite example is in steel braced frame design. One of my pals owns a big steel detailing firm. He says that they regularly see braced frames with HSS12x12 braces and W12x14 or W14x22 beams. Of course, the beam design is wrong and the connections are ridiculous. The automated system performs the calcs as if the beam has no axial load because it has both ends of the beam attached to the floor diaphragm. Any minimally competent engineer running a 2D model of such a frame will catch that every time because he *mentally registers* every size in every frame.

OTOH, if a guy sets up his automated system model, hits "Run" then "export to dxf" and doesn't inspect the elevation on the drawings, he's likely to miss that. Sounds negligent as if only a doofus would do that, but it apparently happens with some regularity.

To me, the main issue with 3D models is more an issue with automated systems: the fact that lots of member sizes make it onto the drawings without ANY human EVER mentally registering the size and whether it makes sense. I think any office using an automated system should have a HARD requirement that every plan and elevation be printed out and have a highlighter mark on every size. This indicates that the responsible PE or SE actually looked at every size in a format that forces him or her to mentally register what will be used. I really think 99% of problems with computer usage would evaporate if people did that.

I've been assigned to work on jobs that had automated system design errors, one of which made it to the field, catastrophically (money wise), I might add. Every single error I've seen would've been prevented with this practice.

 

[rowingengineer]

“e” as you correctly point out you do need to know how to treat your poison in any design for you to be a good modeller either 2D or 3D. I also agree checking procedures are important and should be compulsory. This is one area I believe most offices when under pressure let slip.
However I think rapt has hit the nail on the head with dangers 2 & 3 for 3D with concrete; These are problem I see every day with people using finite element/3-D.
1. “Torsional stiffness’s where you do not think they are appropriate or cannot really be developed”.
With regards to torsional stiffness’s, in 2-D you normally rely on compatibility torsion (when you don’t have equilibrium torsion of course) in design, this is conservative but quick and efficient. In 3-D normally the programs will include torsion stiffness in members as a default, this is un-conservative if the torsion gets above cracking strength of the member.

2. “If there are transfer beams, do you use an overall 3D model to predict the structure response.

If there are transfer beams, do you use the reactions from your 3D model to design the transfer beam? This significantly underestimates the real load on the beams.”

With regards to transfer beams, here you have a situation where the 3D model will move the loads up and down columns across floors and all round the place, so you can have a member not just a transfer beam undersized but columns, when they are subject to higher loads during construction than final design, because the slabs above are supporting this column or transfer beam but they haven’t been constructed yet. This is very hard to do in 2D design.

I would say it is easier to not have any idea what is happening in a 3D model than a 2D for inexperienced engineers. Because the outputs look so nice and glittery, yes there is a moment there and there, guess it is right, I did a quick wl2/8 and the moment was -20% sounds good. Look there is a high moment over that support guess it will disappear when the concrete cracks in that area, but the moment around this won’t increase, or alternatively I will added heaps of reo here as that is where the moment is highest.
But the crux of the problem as Rapt points out is not the program but the experience, training level and to many simple assumptions.


Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that them like it