[Jinjolee]

In the motorBike tutorial case, the turbulence kinetic energy (k) and specific turbulence dissipation (omega) are 0.24 and 1.78 respectively. Using https://www.cfd-online.com/Tools/turbulence.php, we can find that the turbulence length scale is about 0.275m. However, in https://en.wikipedia.org/wiki/Turbulence_kinetic_energy, the turbulence length scale is calculated using:

l = 0.07*L

, where L is the characteristic length.

In this case, the motorBike has a characteristic length of 1.42m. So its turbulence length scale should be 1.42*0.07 = 0.0994m. How is the turbulence length scale of the motorBike calculated?
And if I need to calculate one for my own model, should I stick with the equation l=0.07*L? Thanks a lot.

[Jinjolee]

Does anyone know how to calculate it?

[RobertHB]

The characteristic length is something strange. It changes from case to case, from author to author and is seems the most important factor is to be consistent with what you call your characteristic length throughout your work. For channel or duct flows L is the hydraulic diameter.

I use to calculate the turbulent length scale as , which, if i'm not mistaken, is a formula if found somewhere on this website.

The turbulent kinetic energy i calculate as with I, the turbulence intensity being and Re the Reynolds number. At which point we are back at the question of "what is the characteristic length?".

[Jinjolee]

Originally Posted by RobertHB

The characteristic length is something strange. It changes from case to case, from author to author and is seems the most important factor is to be consistent with what you call your characteristic length throughout your work. For channel or duct flows L is the hydraulic diameter.

I use to calculate the turbulent length scale as , which, if i'm not mistaken, is a formula if found somewhere on this website.

The turbulent kinetic energy i calculate as with I, the turbulence intensity being and Re the Reynolds number. At which point we are back at the question of "what is the characteristic length?".

For that characteristic length, I used the data provided in the motorBike/system/forceCoeffs. The lRef in the dictionary should be corresponding to the characteristic right? I just took that lRef as characteristic length and sub in the equation, please tell me if it is wrong.

Also, the equation you gave is using the hydraulic diameter, so it seems referring to internal flow. Yet, the motorBike tutorials is an external flow, so I am not sure if the equation is applicable or not.

[Tobi]

Maybe you were referring to my thread: turbulent length scale

For codes using a turbulence length-scale based on the mixing-length (Fluent, Phoenics and CFD-ACE for example) replace 0.038 and 3.8% with 0.07 and 7%.

[Jinjolee]

Originally Posted by Tobi

Maybe you were referring to my thread: turbulent length scale

Thanks for your reply. Actually my calculation uses l=0.07L, which is shown at the top. Yet, using either 0.038 or 0.07, the turbulence length scale is still not equal to 0.275m. In other words, using l=0.038L or l=0.07L, the k and omega obtained is different from the one shown in the motorBike tutorial initialConditions file. So I do not how that k and omega are obtained.

[LuckyTran]

What's important is the value of k and epsilon or k and omega. Sometimes this is calculated from a turbulence length scale (if it is known), many times it is not.

k, epsilon/omega, and the length scale are a property of the flow, not necessarily of the geometry.

The l=0.038L is an empirical correlation coming from pipe flows, and only valid for the centerline. I forget the exact reference, but it is a popular turbulence text (probably something like Schlicting or Batchelor or something). The characteristic length L is either the pipe diameter or hydraulic diameter of the pipe or pipe-like thingy. You cannot just put in any arbitrary length here, like the length of the bike. The flow arriving coming at the bike is not a pipe flow... So you should not be using this method to estimate k and epsilon anyway.

The real question is where does 0.24 and 1.78 come from, and that I'm not sure either. It is a valid question with a non-trivial answer. Being a flow property, it requires you to know something about the flow. Turbulence boundary conditions are not so straightforward.

[Jinjolee]

Originally Posted by LuckyTran

What's important is the value of k and epsilon or k and omega. Sometimes this is calculated from a turbulence length scale (if it is known), many times it is not.

k, epsilon/omega, and the length scale are a property of the flow, not necessarily of the geometry.

The l=0.038L is an empirical correlation coming from pipe flows, and only valid for the centerline. I forget the exact reference, but it is a popular turbulence text (probably something like Schlicting or Batchelor or something). The characteristic length L is either the pipe diameter or hydraulic diameter of the pipe or pipe-like thingy. You cannot just put in any arbitrary length here, like the length of the bike. The flow arriving coming at the bike is not a pipe flow... So you should not be using this method to estimate k and epsilon anyway.

The real question is where does 0.24 and 1.78 come from, and that I'm not sure either. It is a valid question with a non-trivial answer. Being a flow property, it requires you to know something about the flow. Turbulence boundary conditions are not so straightforward.

Many thanks for your explanation. However I still do not get the calculation of turbulence length scale. And before this, I have a question about the wind tunnel. Although I have heard of wind tunnel experiment, i have never carried out once(I don't have much fluid course). So I would like to know which of thefollowing cases is preferred to be carried out in CFD, in case of trying to simulate the fluid flow and calculate drag of a car when it is driving outdoor:
1. Typical wind tunnel (front=inlet; back=outlet; all other four faces=wall)
2. Front=inlet; bottom=wall; all other faces=outlet

So for the preferred case, what parameter and formula is suitable for calculating turbulence length scale? Thank you very much!

[Jinjolee]

Actually I am wondering how to obtain the turbulence length scale in the motorBike tutorial. That is not an internal flow, and has no hydraulic diameter, so how can we estimate the length scale?

[Tobi]

Hi,


that is not correct. The motorbike tutorial has an hydraulic diameter and it is simply the projected area and its perimeter. I just refer to József Nagy's christmas community competition. Here, we analyzed the 'Suzanne's head' (Blender). The hydraulic diameter can be estimated as the projected area and its perimeter in that case too.


However, I never investigated into the motorbike tutorial and its values for k and epsilon. At the end we should always keep in mind, that the turbulence modelling is an approach to handle and simplify the complexity of the turbulence itself. Therefore, the values for k and epsilon / omega should be approximately in a good range but at the end there are so many other parameters that influences the results too - such as discretization schemes, mesh quality, and so on and so on.

[Jinjolee]

Originally Posted by Tobi

Hi,


that is not correct. The motorbike tutorial has an hydraulic diameter and it is simply the projected area and its perimeter. I just refer to József Nagy's christmas community competition. Here, we analyzed the 'Suzanne's head' (Blender). The hydraulic diameter can be estimated as the projected area and its perimeter in that case too.


However, I never investigated into the motorbike tutorial and its values for k and epsilon. At the end we should always keep in mind, that the turbulence modelling is an approach to handle and simplify the complexity of the turbulence itself. Therefore, the values for k and epsilon / omega should be approximately in a good range but at the end there are so many other parameters that influences the results too - such as discretization schemes, mesh quality, and so on and so on.

Thanks Tobi, this looks reasonable .
And one more question, is the turbulence length scale same as the characteristic length used in calculation of drag coefficient?
Thanks a lot!

[steddy]

Hi Joe, have you ever found it out? I'm in the same situation.

[HPE]

might help this function object?: turbulenceFields.

[VUBracing_Jasper]

I found a calculator in the following link: http://ichrome.com/blogs/archives/342

This calculates the Turbulence length scale with the formula: l=0.07*L

Below in the "HOW TO USE" it is stated that L equals:
- pipe's diameter
- or a vehicle's length

Does anyone think this is incorrect?

Best regards.

[vjpugliese91]

In my case, I simulated slug flow in a pipe, and I used the grid size as turbulent length to calculate the boundary conditions for interFOAM - LES- Smagorinsky solver, and got good results compared with experiment results.
According the Large Eddy Simulation (LES) theory, we need to estimate the turbulence inside the cell (the unresolved scale).

[Sakun]

Originally Posted by vjpugliese91

In my case, I simulated slug flow in a pipe, and I used the grid size as turbulent length to calculate the boundary conditions for interFOAM - LES- Smagorinsky solver, and got good results compared with experiment results.
According the Large Eddy Simulation (LES) theory, we need to estimate the turbulence inside the cell (the unresolved scale).

Hi,


can you explain bit further about grid size value for turbulent length scale ?



much appreciate