CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   Multiphysics type (https://www.cfd-online.com/Forums/main/235877-multiphysics-type.html)

StudentIB May 3, 2021 11:49

Multiphysics type
 
Hi,

I wonder which type of multiphysics analysis would be needed to account for all these effects:
- fluid flow around the object is modeled explicitly (so we definitely have CFD here)
- this flow causes deformations of a solid body (the body is not rigid so we must add solid mechanics)
- these deformations cause sound that propagates within surrounding air (so there are also acoustics involved)

The first two effects would normally form FSI analysis but what about the last effect - acoustics. How to combine them all ? Is that simply acoustic-structure interaction or vibro-acoustics analysis or something else ? And which software is capable of something like that ?

Thanks in advance for your help

LuckyTran May 3, 2021 12:29

It depends...

Sound propagation also falls under the category of CFD if you do it well enough. A lot of people do steady RANS and want to model sound propagation, and here enters acoustic models. If the body is generating sound then this is not flow-generated acoustics but acoustic waves interacting with solid bodies which is much simpler. But then you have to consider whether this sound interacts with the flow-sound or not. And so on.

Simply saying solid deformation and fluid flow (+ acoustics) is like saying you want to model physics. Yes... that is implied... You need to get much more specific about exactly what models you will be using for the solid part and fluid part to get constructive feedback about what phenomena is or is not being modeled correctly. As usual, DNS solves everything.

LaminarFlow May 3, 2021 12:51

Quote:

Originally Posted by LuckyTran (Post 803072)
It depends...


As usual, DNS solves everything.


Well but DNS for industrial problems is often not possible. The required floating point operations is proportional to Re^3. For usual problems with big dimensions it needs years or hundreds of years to solve the problem once on a actual HPC. But in some terms you are also right. RANS is often not reliable. So I would recommend to look into LES if you have access to a big Cluster. Even at our university aeroacoustics is solved with LES. Regardless of that, I would first check if the available hardware is fast enough for LES. If not then you have to consider faster hardware or good understanding of the RANS model and its parameters. And always keep in mind that you should validate your results with measurements.

All comments are intended for industrial use. For a private project to learn something I would start with RANS anytime, because LES is nothing else than the combination of DNS (big vortices) and RANS (small vortices).

StudentIB May 3, 2021 12:53

Thanks for replies. As an example let’s consider wind chimes. The sound they generate is induced by deformations caused by the wind flow. Regular FSI wouldn’t be enough since it can’t account for sound wage propagation in the air. So how could I model this in commercial software such as Ansys - which modules or analysis type to choose ?

LuckyTran May 3, 2021 14:05

FSI is merely a wrapper that passes the fluid simulation onto the solid simulation and back-n-forth. Of course it does not simulate acoustic waves in air. That is the job of the CFD. FSI is merely an interface for passing boundary conditions back-n-forth.

But are you actually simulating a wind chime? And what kind of wind chime are we talking about?

In wind chimes the perturbation is generated predominantly by Karman vortex structures and these interact with the wind chime. The acoustic waves generated by the wind chime is much much longer wavelength than the length scale of the largest vortex and they are mostly non-interacting. You could resolve the vortex shedding with simple unsteady laminar or URANS, couple this with two-way FSI. Conveniently, unsteady laminar and URANS can also simulate propagation of acoustic waves generated by the wind chime.

What is missing that cannot be modeled easily is if the acoustic waves generated by the wind chime interact with the flow structures. Or you have shear layers that spontaneously generates sound which interacts with the wind chime. In these situations, then you need say LES/DNS style calculations or need to resort to aeroacoustic models. But this depends on the frequency of the wind chime and the particular flow.

So again... it depends... Is this a small wind chime in a light breeze or an organ pipe in a tornado? And is the wind chime made out of plastic or metal?


All times are GMT -4. The time now is 19:54.