[Bollonga]

Hi everybody,

I'm trying to simulate an atmospheric boundary layer (ABL) flow around an inclined flat plate near the ground. I'm not developping the whole boundary layer but using velocity, k and epsilon profiles at the inlet with UDF's (there's no option to use turbulence intensity and turbulent length scale UDFs).

I've used a logaritmic profile for velocity:

u(h)=Uref*K*log(h/h0)

Uref=reference velocity (10 m/s)
K=constant (0.1866)
h0=roughness height (0.03 m)

I defined the turbulence intensity and turbulent length scale (TLS) profiles to get the k and epsilon profiles:

TI(h)=1/log(h/h0)

TLS(h)=Lt*(h/ht)^alpha

alpha=constant (0.4947)
ht=reference heigth (200 m)
Lt=reference TLS (300 m)

k=(3/2)*(Uref*TI)^2

TI from former TI profile equation

epsilon=C_mu^(3/4)*k^(3/2)/TLS

TLS from former TLS profile equation
C_mu=constant

Below h_min=1.5 m all profiles have a constant value equal to u(h_min),k(h_min) and epsilon(h_min).

I've initialized from the inlet but the issue is that with these profiles I get turbulent viscosity limitation from the very beginning. Even if I change the TVR limit to 10e15 limitations is reached very soon.

Does anybody have experience simulating atmospheric boundary layers?
Any advice on how to avoid this limitation?
Is it a matter of mesh density? Cause I cannot go beyond my current 2 million elements mesh... And I've seen some meshes in literature that are not very dense.

I post a picture with all the profiles involved in the simulation and some mesh screenshots.

[marcolovatto]

Hello Francisco!

I'm also beginning on ABL, let's learn together?

The order of magnitude of your TVR is too high... 1e8?? Is it an input or an output of TI and TLS input?

It seems you're using ICEM for meshing... what is your solver?

[Bollonga]

Hi Marco, sorry for a too late answer.

Originally Posted by marcolovatto

The order of magnitude of your TVR is too high... 1e8?? Is it an input or an output of TI and TLS input?

I've read that TVR of 1e8 or higher are usual in ABL flows, as there is a lot of turbulence near the ground.
TVR is derived from TI and TLS, so TVR can be considered as an output of TI and TLS or k and epsilon.

I've solved my problem by applying different boundary conditions (k and epsilon) to the ones suggested by the Eurocode (TI and TLS and then transformed to k and epsilon with Fluent users guide formulae). I've also refined the mesh specially near the ground, to get the high gradient present in the layers near the ground.

I've followed the boundary conditions form this reference:

Richards and Hoxey 1993: Appropriate boundary conditions for computational wind engineering models using the k-ϵ turbulence model

http://www.sciencedirect.com/science...67610593901247

In the next reference there's a simple example I've run to check the proposed mesh, it uses the former Richards and Hoxey BC:

Hargreves and Wright:On the use of the k –ε model in commercial CFD software to model the neutral atmospheric boundary layer

http://www.sciencedirect.com/science...6761050600136X

Originally Posted by marcolovatto

It seems you're using ICEM for meshing... what is your solver?

I'm using Fluent v14

How are you going with your ABL simulation?

Hope this helps!

[eag.10]

Originally Posted by Bollonga

Hi Marco, sorry for a too late answer.


I've read that TVR of 1e8 or higher are usual in ABL flows, as there is a lot of turbulence near the ground.
TVR is derived from TI and TLS, so TVR can be considered as an output of TI and TLS or k and epsilon.

I've solved my problem by applying different boundary conditions (k and epsilon) to the ones suggested by the Eurocode (TI and TLS and then transformed to k and epsilon with Fluent users guide formulae). I've also refined the mesh specially near the ground, to get the high gradient present in the layers near the ground.

I've followed the boundary conditions form this reference:

Richards and Hoxey 1993: Appropriate boundary conditions for computational wind engineering models using the k-ϵ turbulence model

http://www.sciencedirect.com/science...67610593901247

In the next reference there's a simple example I've run to check the proposed mesh, it uses the former Richards and Hoxey BC:

Hargreves and Wright:On the use of the k –ε model in commercial CFD software to model the neutral atmospheric boundary layer

http://www.sciencedirect.com/science...6761050600136X


I'm using Fluent v14

How are you going with your ABL simulation?

Hope this helps!

Hi, I'm trying to assign the boundary conditions that Richards and Hoxey (1993) specify but I can't achieve a homogeneous boundary layer. I don't know what I'm doing wrong, hope you can help me.