[Jing]

Hi everyone,

I have some confusion between these basic concepts-Laminar flow, turbulent flow, steady flow and time-dependent flow.

I thought laminar flow was steady flow, while turbulent flow was connected with time-dependent flow before, but I just found it should be wrong.

Can I understand those concepts in the following way now:
Laminar flow and turbulent flow are distinguished in the scale of space, while steady flow and time-dependent flow are distinguished in the aspect of time; both laminar and turbulent flow could be either steady or time-dependent?

And another question is: For turbulent flow,"time-averaged" properties are used in RANS equations, what is the scale of that time?

Any discussions are appreciated.

Regards,
Jing

[michujo]

Hi Jing.

Just to clarify: you're right in that laminar flow can be either steady or unsteady. However, turbulent flow is always unsteady. Turbulence is an inherently unsteady process since it involves rapid variations of the thermo-fluid properties. Turbulent flows can, nevertheless, be statistically steady, in the sense that the mean flow features do not vary over time.

In RANS you are modelling all the turbulent scales so I think the time scale of the averaging procedure should be the characteristic time associated with the slowest eddy.
Maybe some expert around here can tell you more about this stuff.

Cheers,
Michujo.

[Aeronautics El. K.]

Laminar is a flow in which the fluid flows in parallel layers while turbulence is a stochastic phenomenon. Steady is a flow where the properties reach a steady state after some time and they do not vary any more while in unsteady flow the properties vary in time although there might be a periodicity in the variation.
Solving a turbulent flow using RANS models means that you're solving a steady flow.

[Jing]

Originally Posted by michujo

Hi Jing.

Just to clarify: you're right in that laminar flow can be either steady or unsteady. However, turbulent flow is always unsteady. Turbulence is an inherently unsteady process since it involves rapid variations of the thermo-fluid properties. Turbulent flows can, nevertheless, be statistically steady, in the sense that the mean flow features do not vary over time.

In RANS you are modelling all the turbulent scales so I think the time scale of the averaging procedure should be the characteristic time associated with the slowest eddy.
Maybe some expert around here can tell you more about this stuff.

Cheers,
Michujo.

Hi Michujo, thank you for your clear explanation for the concepts, and I think I get what you mean of "Turbulence is always unsteady, but can be statistically steady".

Cheers,
Jing

[Jing]

Originally Posted by Aeronautics El. K.

Laminar is a flow in which the fluid flows in parallel layers while turbulence is a stochastic phenomenon. Steady is a flow where the properties reach a steady state after some time and they do not vary any more while in unsteady flow the properties vary in time although there might be a periodicity in the variation.
Solving a turbulent flow using RANS models means that you're solving a steady flow.

Hi, why you say "Solving a turbulent flow using RANS models means that you're solving a steady flow",I didn't get it and couldn't agree with this point.

Cheers,
Jing

[FMDenaro]

Originally Posted by Jing

Hi, why you say "Solving a turbulent flow using RANS models means that you're solving a steady flow",I didn't get it and couldn't agree with this point.

Cheers,
Jing

RANS average is intrinsically steady when defined by

<f>(x) = lim (1/T) Int [0, T] f(x,t) dt
T-> inf

[Nimisha]

RANS equation stands for Reynold's Average Navier Stokes equation and is essentially a time averaged mathematical model a.k.a representing steady flow.
To solve for turbulence, instead of employing the complete Navier Stokes equations, the simplified RANS model is used with additional equations to support it by adding additional parameters which empirically model turbulence effects (eg : k-epsilon turbulence model). These addition equations model the fluctuations in flow properties with time.
I hope my understanding is correct but please correct me if I am wrong.

[LuckyTran]

Originally Posted by Nimisha

RANS equation stands for Reynold's Average Navier Stokes equation and is essentially a time averaged mathematical model a.k.a representing steady flow.
To solve for turbulence, instead of employing the complete Navier Stokes equations, the simplified RANS model is used with additional equations to support it by adding additional parameters which empirically model turbulence effects (eg : k-epsilon turbulence model). These addition equations model the fluctuations in flow properties with time.
I hope my understanding is correct but please correct me if I am wrong.

It is totally 100% incorrect to say that RANS solves for turbulence (maybe not exactly 100%, at least 99%). The variables you are solving for are the mean velocity, the time-averaged velocity which is defined as containing no turbulence in it. You are solving for the mean velocity which has been influenced by turbulence (which you model). But never do you solve for the fluctuating velocities. You have to model them because you cannot solve for the mean flow without them.

You can also imagine that I have a laminar flow in where there are no fluctuations but I apply Reynolds-averaging and solve the RANS equations anyway. How can you solve for turbulence in this situation? There isn't any...


The issue is you need to differentiate between a turbulent flow (i.e. a flow that is turbulent and contains turbulence) vs actually addressing the turbulent fluctuations themselves. RANS solves for the mean variables, which are influenced by the turbulence. But the mean variables cannot possibly be any farther from the turbulent fluctuations.

[FMDenaro]

Originally Posted by LuckyTran

It is totally 100% incorrect to say that RANS solves for turbulence (maybe not exactly 100%, at least 99%). The variables you are solving for are the mean velocity, the time-averaged velocity which is defined as containing no turbulence in it. You are solving for the mean velocity which has been influenced by turbulence (which you model). But never do you solve for the fluctuating velocities. You have to model them because you cannot solve for the mean flow without them.


You can also imagine that I have a laminar flow in where there are no fluctuations but I apply Reynolds-averaging and solve the RANS equations anyway. How can you solve for turbulence in this situation? There isn't any...

Actually, the issue is much more complex. This is because you can get a RANS solution without solving the RANS equations supplied by some approximate turbulence model. Just consider to have a DNS solution and perform explicitly the time-averaging. You get the "exact" RANS solution and it totally contains the contribution of turbulence (i.e., the fluctuations) to the averaged field.