Experimental method for turbulent flow research

[Sangho Ko]

Hello guys~!


I'm interested in turbulent flow and I'm finding out turbulence stuff these days to apply PhD course.
While finding those, I have 5 questions about those so I'm writing about it at here.
This post is not asking fundamental knowledge of turbulent flow, but rather about methods, including PIV for turbulent flow

1. Dominant method in turbulent flow research
While searching turbulent flow things, I found something that is different from what I have knew.
As far as I know, the majority of turbulent flow research is carried out by DNS(Direct Numerical Simulation) rather than experiment.
I've heard that the reason for this is due to limitation of experiment, low resolution of camera.
So, I thought DNS is kind of mainstream method in turbulence research and I have been digging into CFD(computational fluid dynamics) things until recently.
But I found that many laboratories conduct turbulence research using experimental approaches, which is totally different from what I knew.
Even in some turbulence research groups consisting of numerous teams, more than half of them adopt experimental methods rather than DNS.
So I'm confused about which one is mainstream method in turbulence research: experiments and DNS.
I know that experiments and DNS complement to each other, but I want to know whether what's going on in real?
Is the majority of turbulence research conduct by DNS or experiment? Or are both of them used equally frequently?

2. Does method selection play a major role in turbulence physics?
Additionally, I'm actually much more interested in turbulence physics than in methodologies such as imaging processes for experiments and numerical schemes for CFD(DNS). Also I'm not confident in both numerical method and image processing in my opinion although it is due to my short experience. I see methodology as just a tool to obtain velocity and pressure values at every single node point.
From the perspective of interest in turbulence physics rather than methodology, does it really matter whether the method is experimental or CFD?
Does methodology play a major role in research for turbulence physics?

3. Camera performance to capture turbulent flow accurately
Subsequently, I'm wondering whether camera performance is enough to capture small eddies in turbulent flow or not.
Actually, I have a really short experience of experimental fluid dynamics.
But at that time, I tried to capture the motion of a region close to the viscous sublayer, and I thought that the camera resolution was absolutely insufficient to capture such small motions of turbulence eddies. This is the reason why I thought that turbulence research should be carried out using DNS.
How is camera performance nowadays? Is it sufficient?

4. Real-world application
Finally, I heard that DNS is suitable for fundamental turbulence structure and scaling law, while experimental research is more appropriate to high-Reynolds-number turbulent flow which corresponds to the majority of the real-world application. Is it right? It sounds correct but I want to be sure from you who are experts in experimental fluid dynamics.

5. Communities for broad experimental fluid dynamics
If there are any other proper communities to upload this kind of questions(experimental method for fluid dynamics), can you let me know?
If this group is specialized for use of PIVlab only, I think it is better to upload these questions to other communities.


Thank you for reading this long post.

[Ivan Nepomnyashchikh]

Hello,

   - this is a community for OpenPIV, not for PIVlab,

   - yes, probably, ResearchGate will be a better place to post your question. I am not aware of any other fluid mechanics specific forums.

I don't have extensive experience to address your questions in-depth, but here's what I think.

1) Looks like both CFD and experiments are used approximately equally frequently. But bear in mind that all the CFD simulations (DNS included) must be validated and verified. American Society of Mechanical Engineeris (ASME) even have an official document detailing the validation and verification process which we are supposed to follow. So, experimental studies are always superior in this sense.

2) From the perspective of the turbulence physics, it doesn't.

3) Viscous sublayer is a notoriously difficult case for experiments. CFD might be the only feasible approach in many such cases. When planning a study, you need to pick your method: experiments, CFD or a combination of them. That involves a lot of factors and the resolution of the camera is just one of them. If you don't have experience, you need to talk to your research advisor about the method suitable to your particular case. Nowadays, cameras have sufficient resolution. The resolution of the camera matters for the estimation of the turbulence energy dissipation term primarily. Another major thing is the laser. Lasers can be time resolved and non time resolved. The former ones cost a fortune, the later ones are somewhat affordable. To study turbulence "in full" you need a time resolved laser, the cheapest of which costs on the order of $100000.

4) It's not correct. You can use either CFD or experiments for both real-world applications and fundamental studies. The problem with real-world applications is that they involve complex flow geometries some times. DNS might not be able to handle such geometries and experiments can be the only way to study them.

Ivan

[Sangho Ko]

Thank you for your quick response!

2025년 4월 2일 (수) 오전 12:38, 'Ivan Nepomnyashchikh' via openpiv-users <openpi...@googlegroups.com>님이 작성:

--
You received this message because you are subscribed to a topic in the Google Groups "openpiv-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpiv-users/zHpcuphhjj8/unsubscribe.
To unsubscribe from this group and all its topics, send an email to openpiv-user...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/openpiv-users/07fe9314-bc08-48f9-b14a-dfa942cc74can%40googlegroups.com.