Large models in RSAPRO

[rbruins]

I disagree for using walls as bars specially for high rise buildings

I've done a test on a 250mmx2000mm wall with a height of 100m. And a beam with the same dimension. (Concrete C20/25) A bracing load of 1kN/m1

The displacement for the panel is 10008.9mm
the Bar 10009.2mm

The Eigen modes are panel : 0.0565 hZ
Bar : 0.0566 hZ

There are some differences. But very minimal!
Why not use a bar for a wall?

See the attachment for the model.

[romanich]

Dear rbruins,
You see mode shapes in your model?
Why not use a bar for a wall? Because bar that is bar and wall that is wall. We use approach of structural mechanics for bars and theory of elasticity & theory of plasticity for wall.

[rbruins]

Dear rbruins,
You see mode shapes in your model?
Why not use a bar for a wall? Because bar that is bar and wall that is wall. We use approach of structural mechanics for bars and theory of elasticity & theory of plasticity for wall.

Firstly, This only applies to slender structures.

I've examined the mode shapes to be identical (Each being seperate, as I have a separte structure).

structural mechanics = simplified theory of elasticity.
FEM = simplified theory of elasticity.

The result from simple shells and bars should always lead to similar results.
And robot has no plasticity for shells only Bars. So still no difference.

Not that I don't want to use Panels. But I don't use panels for simple problems where bars are far better suited to solve the problem.

[mecheil2000]

I think it is better way to use Etabs the king of software for buildings
no more headeach,Just draw floors, walls columns in no second
and everything is runnig properly
no bugs, no problems for dynamic analysis no prblem for seismic analysis
easy way to apply wind and static seismic load

[troy_the.best]

I've done a test on a 250mmx2000mm wall with a height of 100m. And a beam with the same dimension. (Concrete C20/25) A bracing load of 1kN/m1

The displacement for the panel is 10008.9mm
the Bar 10009.2mm

The Eigen modes are panel : 0.0565 hZ
Bar : 0.0566 hZ

There are some differences. But very minimal!
Why not use a bar for a wall?

See the attachment for the model.

Are you an engineer, Rrubin?
not as simple as you say. The structure solved by the FEM, because the logic of the load transfer from single point if the beam passes into the colon, for instance, the behavior of shear walls is not the right way. also what is right under vertical loads, already expected to be different. Modeling of shear walls in high rise buildings, but that simple logic can not be seated.
Please be careful.

best regards
falcoando_turkey

[rbruins]

Are you an engineer, Rrubin?
not as simple as you say. The structure solved by the FEM, because the logic of the load transfer from single point if the beam passes into the colon, for instance, the behavior of shear walls is not the right way. also what is right under vertical loads, already expected to be different. Modeling of shear walls in high rise buildings, but that simple logic can not be seated.
Please be careful.

best regards
falcoando_turkey

Yes, I'm an engineer.

Let me clarify a few things:
I believe that large FE-model are bad! They let you lose grip on your analysis and are far more prone to errors! You should always split the different types of calculation in Global/local models. And you should always make then as simple as possible. Because you can check the simple models far easier that complex ones, and they are faster. (1minute is far beter than waiting 30min). And they're easier to understand for other people.
The problem only is, that the thinking is done by yourself and not left to the computer.

For global analysis of (slender)shear walls, the simple test that I've made shows that can actualy be simple. There is close to no difference. You can even account for excentric loading using moments (obtained from local influences). This was done all the time in the pre-FEM era.

Local influences (wall/floor interaction) should always be dealt on small models (e.a. single floors). Even support displacement can be investigated.

Sorry for making such a big thing out of this. But seeing the problems with large models (in robot/esa/whatever) modelling, calculationtime and the dreaded output. That I tend to preach for not doing it and start doing it like it was done for 20 years back.

But hey everyone loves the big models for their client so they can impres them!

[troy_the.best]

Yes, I'm an engineer.

Let me clarify a few things:
I believe that large FE-model are bad! They let you lose grip on your analysis and are far more prone to errors! You should always split the different types of calculation in Global/local models. And you should always make then as simple as possible. Because you can check the simple models far easier that complex ones, and they are faster. (1minute is far beter than waiting 30min). And they're easier to understand for other people.
The problem only is, that the thinking is done by yourself and not left to the computer.

For global analysis of (slender)shear walls, the simple test that I've made shows that can actualy be simple. There is close to no difference. You can even account for excentric loading using moments (obtained from local influences). This was done all the time in the pre-FEM era.

Local influences (wall/floor interaction) should always be dealt on small models (e.a. single floors). Even support displacement can be investigated.

Sorry for making such a big thing out of this. But seeing the problems with large models (in robot/esa/whatever) modelling, calculationtime and the dreaded output. That I tend to preach for not doing it and start doing it like it was done for 20 years back.

But hey everyone loves the big models for their client so they can impres them!

Hi friend,

When the robot comes to these solutions seem to work normally. but the panel and the vertical loading bar behavior certainly is not the same as outside. Please pay attention to this issue.

Sorry if I said anything offensive. my brother. see later.

best regards
falconado_turkey

[length zero]

rbruins

Regarding to your scope, you are right in some cases and you are wrong in som cases.
You Can't concluse generally that modeling of panels and shear walls can be made by bars, and it is better solution.
There are more papers writen to this scope, but they give more specifics for modeling analysis and design purpose, and in most of cases they are proposed to use only in advanced analysis like push over, ore NL time history, because there are no more explained mechanism of hinge behaviour on shells, and for this purpose bar models have advantage from shell models, and in this contest your proposal is right.
Especially in robot this way of concepting is not good choice.
I tell the reason why:
If u have walls 5x0.5 m , and beams 0.5x1m and if you discretize your model with bars, and wen you make model by axis of bar elements, you have length of beams and walls(bars) greater than reality, because that part of beam which is into wall must be modeled as rigid beam, and also wall(bar) which interfree with beam must be also modeled as rigid bar(wall), and this you cant doo in robot.
This is main scope

Regards
Z.G.

[troy_the.best]

rbruins

Regarding to your scope, you are right in some cases and you are wrong in som cases.
You Can't concluse generally that modeling of panels and shear walls can be made by bars, and it is better solution.
There are more papers writen to this scope, but they give more specifics for modeling analysis and design purpose, and in most of cases they are proposed to use only in advanced analysis like push over, ore NL time history, because there are no more explained mechanism of hinge behaviour on shells, and for this purpose bar models have advantage from shell models, and in this contest your proposal is right.
Especially in robot this way of concepting is not good choice.
I tell the reason why:
If u have walls 5x0.5 m , and beams 0.5x1m and if you discretize your model with bars, and wen you make model by axis of bar elements, you have length of beams and walls(bars) greater than reality, because that part of beam which is into wall must be modeled as rigid beam, and also wall(bar) which interfree with beam must be also modeled as rigid bar(wall), and this you cant doo in robot.
This is main scope

Regards
Z.G.

Hi friend,

you right absolutly. my english isnt enough explain this case. but have do it. thanks. which are you using robot version? im using 2011, but this fully bugs and error. dosent works with it. see later.
regards

falconado

[length zero]

Currently im using RSA 2011,
Post the bugs in forum, and we will
regards
Z.G.