how find fluctuating velocity
Hi
I'am using ansys cfx 10, and I'am making a simulation of jet flow at low Re numbers. I used a SST model and now I don't know how find the fluctuating velocity of the flow. Someone can help me. Thanks |
Re: how find fluctuating velocity
You can calculate the magnitude of the fluctuating component from the Turbulence Kinetic Energy. If you look it up in the documenation, you'll see that the turbulent kinetic energy (k) is defined as:
k=1/2*u^2 where u is the fluctuating component of velocity. Based on this you can estimate the fluctuating component as: u = sqrt(2k) In CEL that would be: Fluctuating Component = sqrt(2*Turbulence Kinetic Energy) -CycLone |
Re: how find fluctuating velocity
Thank you so much
|
Re: how find fluctuating velocity
For anisotropic flow , the kinetic energy is defined as k = ½ * (u'^2 + v'^2 + w'^2) = ½ * (3 * u'^2) = 3/2 u'^2 where (') denotes fluctuating velocity component. Agree?
|
Re: how find fluctuating velocity
Yes, good point. You're right, since u is a vector, the turbulent kinetic energy should be 3/2u'^2. My bad.
-CycLone |
Hi guys,
What do I do if I want to find u', v' and w' (the fluctuating velocity in each direction)? Using this method would only allow me to find u'. Or am I understanding this wrong? And what exactly do you mean by anisotropic flow? |
If you are using an isotropic turbulence model (eg the 2-eqn models) then u', v' and w' is the same by definition. Only the RSM model allows for a difference between u', v' and w'.
Look at a turbulence modelling textbook for more details, "Turbulence Modelling for CFD" by Wilcox is the one I recommend. |
Hi!
I'm using the BSL Reynolds Stress model and I want to know how can I find u', v' and w' (the fluctuating velocity in each direction). Can they be calculated with the "Reynolds Stress uu, vv..." variable or with another variable? How can I obtain this terms? Thanks |
See a turbulence modelling textbook eg Wilcox.
|
But in the CFX Documentation it is written explicitly that Reynolds Stresses may be used by means of CCL!! But How?
Jessica |
The CFX reference manual lists the availabel CEL variables. I think Reynolds Stresses are available if you have an appropriate turbulence model enabled.
|
yeah i have chosen one of them and i am awaiting the reply of ANSYS-Support
|
Right click in the expression window and all available functions and expressions are shown.
e.g. massFlowAve(Reynolds Stress uu + Reynolds Stress uw)@outlet |
I need them in Pre and not only at Boundaries...
|
You can use them in pre and monitoring them while running, this can also be done at specified points or max/min/ave of the entire domain
You can also use variables while specifying the boundaries. for example make the velocity or temperature dependent of time or other variable. question is if you would get the time fluctuationg part from abstractin u.trans avg from u p, li { white-space: pre-wrap; } Velocity u - Averaged Velocity u of that I am not sure. |
I know that i am working with it since the begin of my phd but i know i tried to use RSM and want to calibrate the rs.vv to zero (another topic) by introducing some damping.. but it still showing me that the variables are not available..
jessica |
How to find the Reynolds stresses uu, vv, ww, uv, uw and vw in CFD post?
@Zaktatir: Any luck in finding the method to specify RS in CFX-Pre? |
If you are doing a RSM model it will be in the results file by default (unless you took it out :) ). You will find it under the puzzlingly labelled "..." button.
|
I found it. Just made the contour plot and found them as variables.
One more thing. As we know that linear eddy viscosity models assumes that the turbulence stress is linked to mean rate of strain or mean velocity gradient. So in that case uu, vv and ww are equal. While in RSM model they have markable difference in magnitude. But still I have good results with these two equation models. Why? |
The RSM and 2-eqn models use very different approaches, as you point out. If uu, vv and WW are quite different and therefore the turbulence is anisotropic you would think that means the 2eqn models should have problems. Your report that they match suggests:
1) Turbulence is not significantly affecting the flow - for instance an airfoil simulation will predict very similar lift regardless of the turbulence model. The drag will be quite different though. 2) It could be the anisotropy is not significantly affecting the flow. 3) It could be just luck, and at a different operating point results are quite different |
Thats secret behind success of eddy viscosity turbulence models for many types of engineering flows to find out mean engineering quantities. Thanks.
|
Once the equation are averaged over time and space (ensemble average, please correct me, if it wrong). then I find two situations very interesting.
1. We are trying to find out the fluctuating components through RSM model. In this case it is only giving you only the RSM values of them instead of true turbulence. Well I admit that RANS only model the large eddies. 2. Running the unsteady simulation to find out the transient solution from the averaged qualities. I also find it out it very confusing situation that when we make the Reynolds averaging, we get the NS in terms of mean quantities except Reynolds stresses. Then to find out these components, we need to express them in terms of mean quantities. Whats the point? :confused: PS. May be I am confusing the concepts. So this may be the good chance that I get corrected by some expert. |
Doesn't the post you quoted show you how to do it?
|
All times are GMT -4. The time now is 00:47. |