for the same cell, high aspect ratios may lead to dispariging fluxes across faces

very high for the very small faces

and the converse for the big faces

problem becomes worse for curved BCs where high aspect ratio cells are present

then interpolation of fluxes at the "small" faces must be corrected

I was under the impression Spalart-Allmaras was RAS so I think so. I was told to aim for a y+ of 1 by my supervisor but I suppose neither of us considered that I should take a look at exactly what wall functions are being used And while I haven't checked the aspect ratio of my small faces I'm sure that the aspect ratio is very large (first layer is ~3 microns tall and 0.125mm long) because I'm working with a relatively low-resolution aerofoil coordinate file It sounds like I'm experiencing literally everything you're saying is a problem

Well, do you need wall functions at all if y+<1, for S-A? the tilda-viscosity is zero at the resolved walls....

Spalart-Almaras can be made DES rather easily

and is very robust

Anyway

1rst suggestion: fix the aspect ratio

try to go somewhere around 500-900

more than a 1000 in curved surfaces might be a problem

if not, you need to limit your interpolations

or use central reconstruction schemes (somewhat unstable)

fair enough, thanks so much for the input and suggestions I'll have to set up some sort of splining to fix the aspect ratio so if I haven't done something wrong I should be able to put together everything you've said

hi guys, I am trying to calculate the free stream k and omega values for SST turbulence model, is there any technique or formula to do so?

I am only able to find formulas for calculation at the inlet and not in the free stream

The inlet should fundamentally resemble the free stream (in external aerodynamics and appropriate domain size)

oh ok, so as an initial condition i could just give the same values and try to simulate?

Are you talking about IC or BC now?

in of, its the same 0 folder containing u, p, k, omega and nut files... where we define these right

There is a fundamental difference between BC and IC. Your BC will be present for the entire simulation, or need to be defined for the entire simulation. Your IC is the state of the system at t=0 or Iteration 0.

okay, understood. Thanks!

Good "rules of thumb" are given in the LARCS-NASA website... I don't suggest cfd-online's wiki though

this is the same one right, will look into it

yeah

Hi everyone! In a lid driven cavity, could someone please tell me what the boundary conditions would be for k and omega (or k epsilon)? Would k be 0 at the moving wall too?

I haven't found anything which talks about it for moving walls 🙂