[Kil]

Dear forumers,


Please, help me, I need your advice.
I have to simulate multiphase incompressible external flow around ship hull surfaces and to calculate the Cr ( the residual resistance coefficient Ctotal = Cf(frictional)+ Cr(residuals)) .
There are two ways to solve the problem.
The first one is to simulate at the nature ship scale and the second is to use the model scale. I think that the second way is the most fast that because It requires far fewer cells for similar yplus value. The flow turbulence will be ensured through the use of a turbulence model.


Do you think this decision will be correct? If so, why then do many commercial organizations use the nature-scale model for such tasks?
Are there any pitfalls here?

[LuckyTran]

The mesh will be identical regardless of scale and the computational cost will be identical. Use the real scale unless you are a clown.

The disadvantage of using a different scale is that not all the pi groups will be the same even when the main ones are. For example, it is quite easy to match Reynolds number and Mach number in a pipe for a 2 mm length pipe and a 2km length pipe. But the heat transfer scale is different. Flow in a 2 mm long pipe looks like adiabatic flow of an ideal gas but flow in a 2km pipe is closer to isothermal flow. If you use the real dimensions then you don't have to worry about the pi groups drifting apart, they'll just be exactly what they need to be so you can devote all the time you would have wasted to doing good CFD.


A quote from the CCO of CFD at my previous company to me, during my interview: "I don't understand why anyone would ever not use the real scale in CFD" 2014

[Kil]

Thank you for your answer. I need to study the theory so as not to make such nonsense. Because now I do not understand why the number of cells will be the same if the Reynolds numbers are different. Many examples of cases in OpenFOAM are made on the model scale (DTCHull, KVLCC) and I thought it was to save computational resources.

[FMDenaro]

Originally Posted by Kil

Thank you for your answer. I need to study the theory so as not to make such nonsense. Because now I do not understand why the number of cells will be the same if the Reynolds numbers are different. Many examples of cases in OpenFOAM are made on the model scale (DTCHull, KVLCC) and I thought it was to save computational resources.

Your issue is that the Reynolds (and other non-dimensional numbers) must be the same!
Otherwise, you have different physics in the flow.

[Kil]

Dear Filippo,
At the towing tank tests we can't to the same Reynolds number we use the zone of self-similarity by the Reynolds number. I assumed that something similar would happen here... Please, correct me if I made a mistake

[FMDenaro]

Originally Posted by Kil

Dear Filippo,
At the towing tank tests we can't to the same Reynolds number we use the zone of self-similarity by the Reynolds number. I assumed that something similar would happen here... Please, correct me if I made a mistake

I personally don’t think that…