For the answers: (1) I am not very familiar with the implicit/explicit LES, I am very new to LES... (2) I see the description should be what you have pointed out here, but when I look at the code, it seems that owndelta is replaced by neiDelta if neiDelta > maxDeltaRatio * ownDelta. But as the description says, I think neiDelta should be replaced by maxDeltaRatio * ownDelta. I have replied to you in the #726377664213942323 module... Many thanks....

If I would run a boundary layer simulation for example and use a recycle and re-scaling approach for my inlet condition. How would I set my Initial condition to minimise the time the boundary layer would take to reach "true turbulence" for the lack of a better word?

you mean fully developed boundary layer flow (wall-bounded flows), or a "developing" boundary layer?

fully developed BL in the first half and then my device of interest after that

add a "buoyancy" term in the region where you expect to generate your channel flow. Instead of gravity, you put a random forcing, which you'll have to scale accordingly so you don't crash your simulation

** Buoyancy term in the momentum equation

you turn it off once you got "turbulence"

That's the "poor-man's" whay

way

the "true" way should be to put a tollmien-schlichting instability wave (not as difficult as it sounds), and let it grow/develop

The problem with the latter is that you need a good grid, little dissipation from PISO, and a MATHEMATICALLY CONSISTENT LES MODEL

The latter two, in particular, may be problematic in OpenFOAM

Very interesting. So there is not really a shortcut to initialise the flow to something that has the same statistics of fully developed boundary layer, kinda like a synthesised turbulence inlet, but for 3D dimensions?

some statistics of turbulence are a consequence of so-called ejections/sweeping processes in the near-wall region, brought by the bursting of hairpins, which is the growth of many unstable modes. Some others are due to "outer" motions (coherent structures), with very large modes. These phenomena is inertial, hence resolved by LES. To model these means doing RANS, or it means you have all relevant modes (spectral non-linear scales, via SVD) which you can superpose to your 'steady' solution, from which LES would pick up...

In short, no

Thank you. We covered inlet and wall conditions in our course quite well but i suppose those are more important than your IC because you can always just let it run longer and the influence of the IC will disappear if it wasn't totally bad.

the influence of the IC (inlet condition, i imagine) doesn't "dissapear" in time, but in space

the farther you are from the IC, your results is less "correlated" to the flow passing through said IC

hi all, this isn't an LES question but it is a turbulence question using RAS so I think this section is the best fit for now. I'm using Spalart-Almaras for aerofoil flow and I noticed in the OpenFOAM documentation the ft2 term is omitted. What is the significance of this term, and how might it affect my solution?

@User it might be a good idea to open a RAS/RANS channel

Have you checked LARCS-NASA website?

Usually the ft2 term models transition to turbulence

if you have transition to turbulence (think of flow in a flat plate) then it may be of importance

Done

Better to have both topics on the same place, in any case this channel doesn't move much

Thanks heaps @slopezcastano Since I'm taking this aerofoil to high AOA I might need to model it, so I guess I'll have to add the ft2 term myself

hi all, I'm attempting to run a mesh convergence analysis and am running into issues. I am using simpleFoam to solve for steady turbulent flow over an aerofoil using Spalart-Almaras. I have a mesh with an outer section that I have made (I believe) suitably fine, and a boundary-layer mesh where I define the boundary layer thickness, the expansion ratio and the first cell thickness. I have set my first cell size to make y+ =1 and this result converges. When I refine this first cell thickness (keeping BL thickness and exp ratio the same) my results vary dramatically, and I am not seeing mesh convergence at all. What is happening here? My understanding is that refining a mesh should result in convergence, but I also don't fully understand the y+=1 criteria, and I'm also concerned about my first-cell aspect ratio, but checkMesh says the mesh is good for all refinements

Are we talking RANS? if yes, you must be careful of the wall functions you use for k/omega/etc, since some wall models are only valid for high Reynolds number (y+>100) and others not

as you refine, then you need to change the wall functions

Other discussion is the aspect ratio of the cells