Results from numerical methods vs. structural codes

[SandorR]

Hello!

I'm studying structural design, to Eurocodes in particular. I was reading the section on calculating wind loads on structures. The analysis provided by the codes simply calculates the wind dynamic pressure from wind speed and multiplies it by pressure coefficients depending on the surface geometry of the building to obtain the pressure on the surface. This kind of analysis seems to be quite simple and based on a lot of empirical data.

Another method of calculating fluid pressure on a surface is through Computational Fluid Dynamics simulations. These kind of modern methods sound more accurate to me (although I don't have much experience with CFD in practice), while the methods in Eurocode seem quite conservative as they originate from time before computer analysis was possible.

Let's say I calculate the wind pressure on a structure using CFD and the result obtained is less conservative than what I would get from simply applying pressure coefficients from codes. Which pressure should I use for the design? Is it possible/common to get this kind of disagreement between methods?

I hope my question is clear, as I said I don't have much experience yet in structural design as I am merely a student.

 

[Guest0527211403]

There's an inherent risk in having too much trust in a computational model, especially when still learning about the tools and methods. I think it's certainly possible to have disagreement between a numerical analysis and codes built on years of empirical evidence.

That said I think it's commendable to try to build a model of your own and to question why this might differ from what's stated in the codes, and what the rationale for more conservative values might be (ie ask yourself if there's more to the picture than you have considered up to this point). Approaching it with a mindset of exercising your critical thinking skills will hold you in good stead.

Good luck!

 

[BAretired]

It is quite conceivable that the CFD method gives more accurate values than the method given in the code. However, the building code adopted in your area is the minimum standard you must meet by law. If you do not, you may have trouble defending your design to the authority having jurisdiction or in a court of law in the event of problems deemed to be caused by wind pressure.

BA

 

[SandorR]

BAretired,

So I cannot simply calculate characteristic values of the wind loads in a software like Autodesk Robot and use those if they give me less conservative values than calculating exactly by the codes?

I wonder what is done when the architectural design of the building is something very unusual which is not covered by the codes? It seems at least Eurocodes are limited to relatively simple shapes of buildings. For example something like Sydney opera house, how was that designed to building codes (I don't know the Australian ones but let's say something like that were to be built in Europe)? I would imagine that kind of complex shape requires analysis by CFD, then again CFD was probably not common in the 70s when it was built..

 

[BAretired]

SandorR,

You can use your best judgment in determining design wind load on a particular building. If it turns out to be much less conservative than the code values, I would be inclined to err on the side of safety. You must use your best engineering judgment.

Something like the Sydney opera house would be very difficult to analyze with precision. I don't know what the engineers on that project actually did, but it would not surprise me to learn that they performed wind tunnel tests. I agree that the CFD method may not have been common during the design phase. Usually, the less you know about something, the more conservative you have to be. I believe that determining wind pressure on a complex shape is not something that can be precisely pinned down.

BA

 

[jhardy1]

Sydney Opera House was indeed designed using wind tunnel testing:

https://collection.maas.museum/object/12041

"Arup also noted that the wind load on the roof 'will be considerable' and concluded that 'extensive model tests will be required to arrive at a true distribution of stresses under varying loads'."
CFD was not really an option in the late 1950s / early 1960s.

The current Australian wind loads code (AS 1170.2) makes several references to the use of wind-tunnel testing "or similar studies" (presumably including validated CFD modelling) for structures with unusual geometries not generally covered by the Standard. "Validation" is very important - it is very easy to build a CFD model which might capture the overall flow patterns and global pressure distributions reasonably well, but may completely miss the dynamic fluctuations and peak pressure zones associated with turbulence, vortex-shedding, aero-elastic coupling, etc. (Just look at how much time and money Formula One teams spend on both wind-tunnel testing and CFD modelling to understand the loads and pressure distributions on their cars.)

http://julianh72.blogspot.com

 

[Zenaku]

I think you should use the calculation method that you are told to use, although it is very interesting to do the calculations in two ways and compare them