I have two questions about a simulation using the kOmegaSSTIDDES model.
I want to have ABL inflow conditions, and I have already coded an ABL inflow boundary condition for Omega.
Q1) Is it sufficient to use the mean velocity profile set out by Richards and Hoxey (1), or should I use “time dependent inflow conditions” as stated in the COST best practice guidelines (2) for URANS/LES? I’m guessing maybe some sort of cyclic condition.
Q2) Given that I am using the kOmegaSST model I should have y+ < 1 everywhere on the walls, can I still use wallfunctions in this case, or should I employ fixed value or zero gradient conditions depending on the variable? Basically, will it make a difference if I just leave the boundary conditions as wallfunctions given that it’s so far within the viscous sublayer?
Thank you, and any advice would be happily received.
Good questions and I would forward them to our expert @dlynch in this case who can give you more tips on how to achieve optimal performance in the case of DDES simulations.
Hi @morgan_evans21, quite a big topic, unfortunately. So typically, weather using OpenFOAM based Solvers or PaceFish (LBM) based solvers we are defining inlets as mean profiles, both for velocity and turbulence variables. Time-dependent inlets are not very mainstream at this moment and for traditional CFD use cases are not required. For example, in pedestrian comfort, we are mainly interested in mean velocity fields and for building surface pressures we can get pressure coefficients and scale to whatever reference velocity we need.
I am sure in the near future CFD will become closer to wind tunnels by incorporating these fluctuating boundaries, it’s just not quite here yet. Please get in touch with our sales staff if you are interested in using our transient PaceFish based solvers (to use the K-omega SST DDES models).
In OpenFOAM the K-omega SST model is capable of handling Y+ approaching 1, as expected, however, it can also operate in K-epsilon mode around Y+> 30 so it’s not too critical to resolving the Y+ in the ABL, however, it is more necessary on the building surfaces.
I thought that might be the case with the inflow conditions.
As for the wall conditions, I also figured that a value Y+ > 30 for the ABL and Y+ < 1 on the building surfaces would be the ideal situation. Thank you for confirming this.
With regards to scaling the reference velocity, I have read that this is applicable to blockages with sharp corners, such as a rectangular building. Can this technique also be used for smoother blockages, take for example an aerofoil?
My poor little laptop can’t handle these simulations with its 2 cores. I may get in touch soon.
Hi @morgan_evans21, yes very true, you sound very familiar with the litterature. I am actually just writting a piece about reynolds scaling, when it is done I will share it. In summary however, for bluff smooth bodies, like cylinders, then you can apply scaling to some degree, being careful not to take the reynolds number into the transition region. However, for slender aerodynamics (a 2D aerofoil) I think the drag will be strongly influenced, so I think this is not an ideal application for scaling but I think it is still done to some degree (thinking about wind tunnels for aerospace applications). A am using Tom Lawsons book ‘Building Aerodynamics’ as a solid source, I strongly recomend it to get a full explaination.
Brilliant, I will have a read, and please do link me to your paper.
My background is with laminar viscoelastic fluid flow, but given the lockdown situation I thought I’d have an attempt at learning about turbulent flows.
