[pamstad]

Hello everybody,

my goal is to get an appropriate total height of the prism layer, such that I can capture most (ideally the total boundary layer) of the turbulent boundary layer with my prism elements.

My idea was that I could do that with the help of the Eddy viscosity ratio = turbulent viscosity/dynamic viscosity. For that I plotted the Eddy viscosity ratio (see attached picture)

Now my question is: My prism layer (see also attached picture) lies completely in the dark blue region, where the Eddy viscosity ratio is actually quite small. Wouldn`t it make sense if it has a larger total height? Does it make sense what I have done or how do you determine the total height of your prism layer?


Thanks in advance for your answers and have a nice day!

[ghorrocks]

My advice is to just try making it bigger and smaller, run the simulation and see what happens. The best way to learn is to do it yourself.

Also read the CFX documentation on numerical considerations for near wall modelling. It has some useful tips as well.

[pamstad]

Thanks Mr. Horrocks for your answer.

I have read the CFX documentation and several other documents about modelling near the wall.

What bothers me about your suggested approach is the following:

1. Assuming I run some simulations with bigger and smaller Prism layers and obtain some results (assuming all of them fulfil my convergence criterias)

2. Then one of the simulations matches the best, let's say with a measurement I have. Then someone ask me, why I have chosen this specific arrangement of prism layers? And then my answer is because this matches the best with my measurements? Doesn't really make sense to me.


Just to clearify, all my results I've got with different prism layers are in the range of 3 %. I just have a hard time to justify why one would be better than the other and as mentioned again, just because one matches the measurements better than the other is not a real argument to me...


Then coming to my next question: When you run simulations with different parameters, based on what do you decide which results is the "best"? Of course assuming that all the simulations fulfil convergence criteria.


Thanks a lot for your help and have a nice day. I hope you know what I mean.

[ghorrocks]

Very good question, it is good to hear that you are thinking about these things. Simulation verification and validation are very important and address the exact question you ask.

Some links to look at:
FAQ: https://www.cfd-online.com/Wiki/Ansy..._inaccurate.3F
The key bit on the FAQ is the sensitivity analysis. This allows you to determine if you are converging to an accurate result or if it is just luck that one of your runs in accurate.

Journal of Fluids Engineering has a policy on CFD accuracy: https://journaltool.asme.org/Templat...umAccuracy.pdf

Works by Celik or Roache are excellent on this topic.