problem with turbulent model in free surface
I use k-e and RNGk-e for my free surface flow model, but the result of k-e is better than RNGk-e is the possible? because most papers said RNGk-e is more accurate than k-e in swirl and high curvature flow.
the second question how can I calculate Fr (v/(g y)^0.5) in CFX post in a section where v is the average velocity and y average hight at that section. thanks: |
please can anyone help me in that problem
|
There is not much difference between k-e and RNG k-e for most applications. And I would expect them to both have similar problems with swirl and high curvature flow - you need RSM models to handle that.
The equation you quote should be easy enough to process in CFD-Post. What is your problem with it? |
Quote:
the problem in that equation is, how can I express average velocity for example at plane 1, is that right (avgvelocity@plane1) or something like this about y or height of water, how can I express y at VF 0.5. this is my problem. thanks |
If you have a plane, then the areaInt(water.vf)@plane/area()@plane is the proportion of the plane which has water. There's a starting point for you.
|
Quote:
about turbulent model, I think your purpose is, in spite higher computational time of RNGk-e it is result almost similar to k-e, so which kind of turbulent model of Reynolds stress model (RSM) is useful to handle that case, (LRR Reynolds Stress, QI Reynolds Stress, SSG Reynolds Stress, Omega Reynolds Stress, Baseline (BSL) Reynolds Stress model, Explicit Algebraic Reynolds Stress model) are available in CFX |
Quote:
|
I do not understand your question.
Are you asking what turbulence models are available in CFX? (the answer is: it is all in the documentation, read it) Or are you asking what turbulence model do you recommend for your application? (the answer is: I always recommend using SST unless you have a good reason not to. You have not stated a good reason not to, so I recommend SST) |
Quote:
yes, my question about best turbulence model for swirling and high curvature flow like hydraulic jump. but I think SST at the same class like RNGk-e, standard k-e (which is eddy viscosity model), I test RNGk-e, standard k-e almost have the same result. according to your previous post that RSM is the best to handle that high flow curvature, my question is what specific model in RSM in CFX there are many model as I posted previously. |
No, I do not agree. The curvature I see in the hydraulic jumps I am familiar with is more a turbulent effect, that is it is random in space and time. You do not need things like RSM models to handle this a standard model like SST should work. Well, at least I would try SST first.
The flows which definitely have high curvature which mean 2-equation models like SST are unsuitable (at least without modification for the curvature) are things like industrial cyclone flows. The streamlines have extreme accelerations on them in this case. But at the end of the day if you get good results with RSM and poor results with a 2-equation model then that is convincing to me. |
Quote:
|
You suggested RSM models. Have you tried that?
|
Quote:
|
All the models have strengths and weaknesses, so none of them are "the best". If you read the documentation the SSG RSM model is probably the one to try. If want to try a omega based model try BSL RSM.
|
Quote:
at the end of channel there is 8 cm difference with experiment, which is very high, I dont know why. is it possible to say CFX (RANS, with turbulent model) does not work in hydraulic jump. my mesh and all BC are correct only Iam not sure about outlet B.C I used outlet BC at the end of channel (at drop) with super critical, and at bottom of channel with static pressure = 0 https://drive.google.com/open?id=0B3...0h4VzBLUGJEVlk https://drive.google.com/open?id=0B3...kM3T3QzZUdqaWs https://drive.google.com/open?id=0B3...EtpVWZseGVTdUE |
CFX can model hydraulic jump. You have not set it up correctly. But the hard bit is finding what you have done wrong.
These results also suggests the turbulence model is not causing the problem, as you say the problem is more likely in you fundamental simulation setup. From your images I can see lots of problems straight away: * Your mesh is way too coarse * Your inlet is too close * Your outlet is too close So they are some things you need to fix before you look at anything else. |
Quote:
https://drive.google.com/open?id=0Bz...DhVYTNjLWZLVnc |
Regarding the coarse mesh issue, have you attempted to run a finer mesh and see if the solution is sensitive to the mesh resolution ?
The use of coarse, medium, fine, etc meshes is just relative terminology for a given problem. What is coarse for a case may be just fine for another regime ? There is no generality, and sensitivity studies must always be done to some extent. |
Quote:
|
Quote:
|
Hydraulic jump in incompressible flow is equivalent to a shock wave in compressible flow. That means, it is a very thin region in the flow direction with rapid changes (if not sudden) of the solution variables.
How good are the current results to predict the start and end of the jump region ? Predicting the height is a very sensitive function of the mesh and the free surface model being used. Hope the above helps, |
Quote:
imagine, 7 cm error in jump is very, if only one or two cm error I can say due to mesh size, I think related to other things, in most free surface flow in CFX I think elurian multiphase (homogeneous with standard free surface is the best choice). |
Quote:
Hello. I want to simulate an spillway with different turbulent models in Fluent but I wanted to know the differences of volume fraction of water and air with changing the turbulence models, in advance. Are there any difference in water surface when you change turbulence models? I must say I use VOF multiphase model. Thank you |
Note this is the CFX forum. Try the Fluent forum for questions on Fluent. Having said that I note your question is a generic question on turbulence models which is applicable to either code.
|
Quote:
|
Quote:
|
Quote:
I do not mean the accuracy of flow surface. I want to know that is there any difference for flow free surface for a specific section of model when you change turbulence model? For example when you change from RNG to STl or any other turbulence models, is there any difference in volume fraction for that section? Or they are exactly the same? My aim is to compare free surface of flow for different turbulence models. |
You seem to be talking at crossed purposes. Yaseen said that in his opinion RANS turbulence models cannot predict the free surface at the hydraulic jump accurately (I am no expert on hydraulic jump modelling but this sounds likely)
Whereas Saeed said if you change the turbulence model does it change the free surface location (I am sure the answer to this is yes, especially for grossly different models like RNG versus Reynolds Stress models) |
Quote:
|
If you want to be "quite sure" then why are you asking people you have never met on an internet forum? How can you trust their answers? Isn't the only way to be "quite sure" to run the models yourself, or to read about them in a respected journal?
|
Quote:
Did I have any other choice when I cannot start the simulation except other people's experience? If we do not trust the research conducted in literature by other people, we would not continue to make progress the current knowledge. I am sorry if I chose uncommon way. Maybe I should review the way I look at people. |
I don't mean to cause disrespect and certainly there is nothing to apologise for. I was just trying to point out that answers given on online forums can be wrong and there are no checks and balances to correct mistakes. Sometimes forums work well and you get good answers, sometimes you don't - and it can be hard to tell at first glance whether you got a good answer or not.
|
All times are GMT -4. The time now is 13:14. |