Terrain type, Wind simulation

[Anonymous]

Hello,

I have start to use the new Wind load simulation in RSA 2015.

Im wondering if you can set the terrain type fore the wind loads in the wind load generator?

(Similar that you can do in the wind and snow 2D/3D generator)

Br. Jesper

[Rafal.Gaweda]

No.

[Anonymous]

okay,

But the terrain level option is the alltitude above sea level?

[Rafal.Gaweda · Accepted solution]

But the terrain level option is the alltitude above sea level?

This option has nothing to do with sea level, terein type etc. It is a virtual wind tunnel.

Terrain level - Z level (from Z=0 in GCS) from which the wind will act on model. As a default the Z coordinate of the lowest model node is displayed.

[StefanoPasquini6790]

Hi Jonsereds,

To make a compatible analysis of wind loads with rsa, compared for example to ec1, you have to build your own wind curve.

I made some tests about it and i found that this is the way:
1) take or make a spreadsheet where you can set all the parameters (except the dynamic coefficient);
2) build the pressure curve for direct wind invested panel and for 2h ( two times the maximum height of the building);
3) define from this pressure curve the velocity one, using the relationship between them;
4) set the maximum velocity curve value as 1 and scale all the curve points to this;
5) set in Robot your own velocity curve, using the maximum velocity value previously defined.

Once you have done, take some minutes to compare the obtained results to the theoretical ones, I suggest to compare the global horizzontal reactions values.

It's more difficult to explane that to do!

Good work!!!

[Anonymous]

Hello

so in this velocity factor, we dont have to include the dynamic factor... it looks logical.

do you have a possibility to share any example of your procedure ?

best regards

[StefanoPasquini6790]

Hi kasa89

this is the model that contain the comparison between code wind loads and Robot Wind Simulation loads. If you compare the restraints reactions you can see that the wind analysis results are very close to code wind loads.

The trick, as I've explained in the previous post, is to set the pressure value to 1065 N/mq (as shown in the following pic), set this value to 1 and refer all the curve value to this.

You can see all the details of the calculations in the two models attached.

[Anonymous]

Hello Stefano

thank you so much.

so the velocity factor in CFD is the multupication of exposure factor and the aerodynamic factor only, is this applicable also for truss structure like towers ?

and the most important question is : Why autodesk do not give more explanation about this issue ?

best regards

[StefanoPasquini6790]

Sorry,
This is not an issue, Autodesk published many months ago a white paper about comparison between wind tunnel and wind load simulation. The results of the tests was very close to the simulation ones. You can find this document published on forum by Bryan Frank.

I think that, before you start with the structural design of a tall building, you have to make some validation test with easy example to find the right way to make an efficent model.

Start with only one wind load case and make your own comparison between code defined loads and wind simulation loads, as I've done. In this way you can be sure of the results validation and you have find the right way to go on with the complex model. After that, you will feel proud of your validation test and you will understand that there's many ways to find the goal, not only one. Autodesk give you (for free!!) a great weapon, now you have to learn how to use it, with all the needed care.

Good luck and I want to see your model in the gallery of big project

[Rafacascudo]

I followed Stefano instructions .

 - Structure is a 25x10x40m perfectly symetrical building.

 - wind speed - 40m/s in 4 wind directions +/-Y and +/-X

 - speed curve

Variable:

0m- 0,8

40m - 0,9

80m - 1,0

Constant : all values=1,0

 Convergence 0,25%

As you can see on the picture below the reactions results are completely different for opposite directions!!!!! In the Y direction it is more than 100% difference!!!

Vento +Y and vento - Y are wind load cases created by me. Model is attached. rename to .rar

[StefanoPasquini6790]

Wow,

That's very strange....now I'm reading this post on my mobile, as soon as I can I want to investigate your model.

One question, have you run all the analysis in the same simulation session? Haven't you try to run 4 indipendent analysis?

One note, for 80 m of building's heigth 40 m/sec it sound too small, it was a testing value?

See you soon, greetings.

[Rafacascudo]

Yes, in the same simulation session.
Building is 40m high.80m is for the top speed curve value.
Anyway, just testing, not a real structure

[StefanoPasquini6790]

Hi Rafacascudo,

I've investigated your model and I have some comment or suggestion for you:

With all my testing sessions, I've understand that, for the buildings with claddings, as yours, is better to select, in the wind simulation analysis, only the panels that describe the edge of the building.

I suggest to don't increase the accuracy of the convergence over the asdk default value (0,5%), this will increase your calculation time, with the remote possibility to create an incorrect analysis. For example, if you look the colormap of pressure of your wrong Y analysis, you can see that all the panels are red pressured, ABSOLUTELY NOT CORRECT
If you use 0.5% you will see that the colormap of pressure is the same of the opposite wind direction analysis and the results are very close.

There is an error message in your model about the load distribution, I've not investigated yet.

To get a global better accuracy and best evaluation of localized effect, split your claddings by story for all the building's facade. This is important for the global convergence of the aeroelastic equilibrium.
I'm starting with a lot of tests between Robot wind load simulation and autodesk simulation cfd, obviously I will make it on the free time after working hour, so I can't give you more information about the duration time of this comparison and the next results evaluation..

Keep in touch on the forum for all tips and tricks about wind load simulation and I invite all the users to post all its personal comments/tricks/suggestions about it.

Greerings at all and FORZA ROBOT.

[Rafacascudo]

Hi Stefano,

thx

I followed your new suggestions.Created 2 models . One with claddings and the other with panels with 1m mesh size

-Divided the claddings or panels (1 cladding/panel between 2 stories).

-0,5%

-1 wind direction per simulation

- Selected Only  claddings/panels that get the frontal wind for each of the 4 directions.

-no load distribution warnings

As you can see  , very different results using claddings compared to panels and still different results for wind in contrary directions for both models.

CFD with claddings

CFD with panels

models attached

[StefanoPasquini6790]

Hi Rafacascudo,

I made some tests in comparison between Robot Structural Analysis wind load generation and Autodesk Flow Design.

I've understood that it's impossible to get the same results for the Y+ direction and the opposite one. As you can see in the following video, during the wind simulation the drop value is variable. The good news is that the global drop value (average) is very close to Robot reaction along wind direction, and the Robot value are a little bit higher, then the analysis can be judged accettable for the structural dimensioning.

Try to use Autodesk Flow Design, demo version, in this way you can better understand what I'm telling.

Greetings

[Rafacascudo]

Thx Stefano,

I´m not sure if I understood it correctly , but anyway , if it´s impossible to get , at least , similar values for opposite directions , what direction should I trust , + or - , when using Robot CFD??

I was also puzzled with the different results I got using claddings and shell panels. I never thought it would be relevant for the generated wind pressure value. So I did some further tests and found out that:

- the resultant force is different ,depending on the panel thickness that gets the frontal wind.

- The resultant force is the same if the panel material is different. It doesn´t matter if it´s steel, concrete, glass or paper.

Can you confirm this on AFD??

[Rafacascudo]

Where can I dload AFD demo? Couldn´t  find it in autodesk page

[StefanoPasquini6790]

Hi Rafacascudo, this is becoming a one by one discussion. Is there noone interested to this topic?

However, probably in your country AFD is not present yet, tomorrow I will try to send you the link.

About the differences between claddings and panels, I don't know the reason of this strange behaviour but keep in mind that the claddings, for the aeroelastic problem solving, are hidden meshed, probably this double meshing can cause calculation issues.

One more test: try to move your building with the base center placed in GCS origin 0,0,0. The little eccentricity of the structure can cause different distances from the tunnel bonduary and then different results for equilibrium solution.

Answering to your question, with AFD is not possible to evaluate different kind of surfaces (panels or claddings). They are simply 3d objects.

To be continued....

[Rafacascudo]

Hi Stefano , thx for your answers!!

   I find this a very interesting and important discussion. I think  CFD in Robot must be discussed for the simple cases  like this simple building , so we can have confidence to use it in the complicated cases like curved  , complex geometry structures and also for structures with obstructions.

Back to discussion...

   If claddings are not recommended ,at least not yet , what panel thickness should we use? For this same building , If I use 1.4cm or 14cm or 140cm thickness for the panel that gets the front wind I get very different results for the total reaction.

And the results are not linear!!!

40m/s , 0,5% convergence, 0,8-0,9-1,0 speed profile , 1m mesh size

1.4cm - 155,1 tf  (very fast convergence time, less than 1min real time)

14cm -  90,6 tf

140cm - 116,5 tf

1,4cm would be like a glass wall. 14cm a normal wall and 140cm an extra thick wall.

If the wall thickness is really relevant , should we trust these numbers??

in a 2nd test ,as you suggested ,I moved the strucuture  up. Support level is now on Z=10m.

Results are:

1.4cm - 155,1 tf  (again ,very fast convergence time, less than 1min real time)

14cm -  100,5 tf

140cm - 116,6 tf

So... I´m lost