[Falu]

Hey People,

for my Master thesis i want to influence the place of the transition from laminar to turbulent. After my literature research i found out i can do this by producing turbulent inflow in the inlet with specific attributes. Some methods requiere a modification of the internal Calculation, if i am not mistaken. For example in A digital filter based generation of inflow data for spatially developing direct numerical or large eddy simulations Klein wants to comupute specific two-point correlations, in Random Flow Generation Technique for Large Eddy Simulations and ParticleDynamics Modeling (2001) Smirnov wants to compute turbulence via an inverse fourier transformation and in Simulations of bypass transition (2001) Jacobs and Durbin want to go a similar way like Smirnov, but compute so called Orr-Sommerfel-Modes instead of the Fourier equations.

Is someone here who has developed codes for this cases or can give me some ideas how to do it myself? I am very grateful for any help!

Cheers,
Fabian

[ghorrocks]

Can you explain what you are doing? Is this transition on a flat plate? Or an airfoil? Or something else?

Are you using the turbulence transition model? Or LES? Or something else?

And finally: Why are you doing this analysis? What are you trying to learn from it?

[Falu]

Originally Posted by ghorrocks

Can you explain what you are doing? Is this transition on a flat plate? Or an airfoil? Or something else?

Are you using the turbulence transition model? Or LES? Or something else?

And finally: Why are you doing this analysis? What are you trying to learn from it?

Im modelling a free jet stream. At the moment im using the LES method, but im free to change my turbulence model (yet i dont want to use RANS or DNS).

For now i just want to be able to control the place of transition of my model. I read that critical indicators for the transition point are the turbulent length scale and the turbulent intensity, so im trying to compute methods which give me the option to change those variables.

Edit: To be more specific, i use LES in the main region. At the edges of my model i use RANS (in the regions where no fluid is to be expected). In the nozzle im using LES too, but here im totally free to use something else.

[ghorrocks]

Modelling turbulence transition in CFX is challenging - it does not have a LES based transition model, so I suspect you will find the mesh and time step requirements to get it accurate are going to be prohibitive.

And turbulence is modelled in LES as a sub-grid model with the large turbulent eddies resolved. So if you want to specify where transition occurs you need to have energy in the sub-grid model (easy) and specify the large turbulent eddies (very hard).

In other words, I would think it a major research project to implement a turbulence transition model in the LES model in CFX, even if you intend to specify in advance where transition occurs.

Reading your edit: Does that mean you are actually using one of the hybrid RANS/LES models like SAS, DES, SBES or one of those? Or are you using pure LES? The way turbulence is specified is quite different in each approach.

And finally: you did not answer the most important question in my previous post: Why are you doing this analysis? What are you trying to learn from it? I am only guessing if I do not know what you are trying to achieve.

[Falu]

Originally Posted by ghorrocks

Modelling turbulence transition in CFX is challenging - it does not have a LES based transition model, so I suspect you will find the mesh and time step requirements to get it accurate are going to be prohibitive.

And turbulence is modelled in LES as a sub-grid model with the large turbulent eddies resolved. So if you want to specify where transition occurs you need to have energy in the sub-grid model (easy) and specify the large turbulent eddies (very hard).

In other words, I would think it a major research project to implement a turbulence transition model in the LES model in CFX, even if you intend to specify in advance where transition occurs.

Reading your edit: Does that mean you are actually using one of the hybrid RANS/LES models like SAS, DES, SBES or one of those? Or are you using pure LES? The way turbulence is specified is quite different in each approach.

For now I am currently free to change the turbulence method. The important aspect is, that my main simulation area is using a numerical method, (like LES and DES and so on) and the averaged methods are just used for unimportant parts of the volume to reduce simulation time.

Originally Posted by ghorrocks

And finally: you did not answer the most important question in my previous post: Why are you doing this analysis? What are you trying to learn from it? I am only guessing if I do not know what you are trying to achieve.

My goal is to change the place of the transition from laminar to turbulent. Thats it and im more or less free how to accomplish this. Ideally, the user will be able to make a simulation. in which the transition occurs after (for example) 2m. In another simulation the user wants a transition after 1m, so he change a few variables and its done.

I know its very hard to realize something this and the case above is just the ideal case. For now I just want to be able to change the place of transition and I thought the best way to do so is by creating turbulence with specific properties in the inlet. For this i need to find a way how to implement the above mentioned papers in Ansys CFX.

My task is just a small Task in the bigger picture, in which it is analyzed how to accomplish to influence the onset of the transition in technical challenges and use it for your advantage.

[ghorrocks]

If you want to model turbulent transition in CFX then the turbulent transition model add-on to the SST model is what most people use. In that model you can directly specify turbulent transition as a built-in part of that model, or you can let the model predict the transition location. Is this model suitable?

Also, what about the DES, SAS or SBES models? Have you considered them? Those models already have approaches which deal with the transition from a RANS model to a LES model.

If you are trying to combine a turbulence transition model with a DES/SAS/SBES model then I hope you are expecting your project to be a major piece of work taking months to years to get working. This is research stuff, not a model you just have to click on and it just works.

[Falu]

Thank you, these models looks useful on the first sight. I will have a closer look at the functions about "Transitional Turbulence".

Im new to Turbulence Modelling and i dont really understand what you mean by "Those models already have approaches which deal with the transition from a RANS model to a LES model". I want to model the Laminar-Turbulent Transition, the RANS-LES Transition is of no intereset for me? My Model shall mostly be resolved numericaly and just the unimportant parts are modelled with RANS.

I know you can import a velocity vector, which you can set up to your inlet of your model. I thought its possible to generate a velocity vector, for example with a Fourier Transformation, which has an increased turbulence intensity and therefore move the onset of the laminar-turbulent-transition. That was my first idea of how to handle this problem.

[ghorrocks]

Always look at what models are already around before you think of your own. So it is definitely worth your time to research the SAS, DES and SBES models. Even if they are not appropriate for you, the knowledge of turbulence behaviour incorporated in the models are important in your line of work.

My comment about RANS to LES transition is because you stated that you are using RANS in "unimportant" bits of your flow, and LES in the "important" bits. This implies that you need a transition from RANS to LES. Any transition from RANS to LES is not a trivial exercise - read the SAS, DES and SBES models to see the complexity they have to handle it in their cases.

You are correct that the laminar to turbulent transition is completely different to the RANS to LES transition.

Have you read any fundamental turbulence modelling textbooks? "Turbulence modelling for CFD" by Wilcox is my go-to text. Make sure you have a solid understanding of turbulence modelling theory before you consider any advanced topics like laminar to turbulence transition or RANS to LES transitions.

I am getting the feeling you are severely underestimating the difficulty involved in what you are proposing. Did you read the final line of my last post?

I hope you are expecting your project to be a major piece of work taking months to years to get working. This is research stuff, not a model you just have to click on and it just works.