[harris]

Hi everyone,

I have a modeling problem in ansys cfx 13 and im not sure on how to define the roughness of the problem's walls. In wall roughness, it asks me to give "sand grain roughness". I read the cfx help on this but I didn't understand if I have to give the original roughness height of my wall or an equivalent sand grain height. Unfortunatelly in the help, it uses an image of an older version of cfx in which asks for "rougnhess height", not "sand grain roughness".

Thanks in advance

[ghorrocks]

The CFX Theory guide has an extensive discussion about this. It refers to to some references for converting from roughness height to sand grain roughness.

[harris]

i cant find any references to make the conversion. the materials that i use are road asphalt and bulding cement (im modeling the wind flow in a specific neighbourhood). I've googled to find info but nothing has been found. As far as i know sand grain roughness is caclulated by experimental observation of the material. Is thar right, and if it is were can i collect info from?

[ghorrocks]

I don't have CFX here right now, but seem to recall it is in the theory guide under turbulence modelling/Wall treatment/Rough walls.

[harris]

I searched a bit more on the subject and I found out that the sand grain size is experimental result indeed. I found some data for cement pipes and I will use them as guide. There are not the exact Im looking for but I'll make a compromise.

[monkey1]

Hi Harris!
As I suppose I had the same problem as you do, here the Info I got from the CFX support. The Sand Grain Roughness k_s that has to be defined for rough walls on a wall boundary is linked to the real or the aerodynamic roughness length k_0 by the following relation:
k_s approx. equal to 30*k_0

Found in the Article:
BLOCKEN B, STATHOPOULOS T, CARMELIET J: CFD simulation of the Atmospheric Boundary Layer – wall function problems. Atmospheric Environment 41(2): 238-252, 2007

So in my case I am investigating the (atmospheric) flow over different terrains. In order to avoid modelling each building I want to use the roughness length defined for different topografies (from flat plains up to built up inner city areas). In different publications roughness lengths can be found ranging from k_0 = 0.01m for flat terrain up to k_0 = 5m for inner city areas. To get the corresponding CFX sand grain roughnesses I have to multiply thos values by 30, getting k_s = 0.3 m for flat and k_s = 150m for rough terrain.

I hope this is some kind of helpful and related to your problem.

[evcelica]

Those are some big sand grains, I'd hate to get one of those in my eye!

[monkey1]

You're right evcelica...I was kind of "shocked" too. But if that is what the theory and the Ansys support is saying...then I gotta stick to it

[monkey1]

After talking to the support again here some more infos about wall roughness:
For CFX it is definitly a must to respect the following rule. When you set up a roughness of let's say k_0=0.05m and get a resulting k_s=1.65m following the rule mentioned above, you have to have a mesh in wich the wall adjacent cell (the first cell on the wall with the specified roughness) with a height equal to at least the DOUBLE of the roughness length k_s. This leads to some problems as to have a good wall resolution. In the mentioned example the first cell would have to be 3.1m high! If you are not investigating atmospherical problems with a scale of several kilometers but rather the flow around buildings and its effect on e.g. people then a wall resolution of >3m is too big!
If you do not stick to this rule cfx will still give you results, but as I understood the support it will kind of neglect all cells below the height of 2*k_s regarding a fully resolved solution of the flow equations. What it means exactly for a specific case can only be deucted by carrying out simulations with different roughnesses. Even the support couldn't give me an EXACT answer wether the solution will be correct or not. The only thing they said is: For your specific problem we don't see a solution except for not using a roughness specification.

[ghorrocks]

That goes without saying - the roughness model is a sub grid model, so if the feature you are using it to model are bigger than the grid then obvioulsly its validity is going to be tenuous. It will still give you results but no guarantees on how good it will be.

Not sure what you mean about CFX ignoring cells. That is not my understanding. It simply applies the roughness model equations to the wall boundary regardless. I think you will find it does not ignore cells.

If you are modelling roughness larger than an element or two in size then you really should draw, mesh and directly model the feature.

[harris]

how can a roughness heigth be k_0=5m? isnt it enormous? the real rougness height magnitude order isnt in mm or cm? in my problem the tallest bulding is 16m tall, so its unacceptable to use grain size of the scale of 30*k_0 if k_0=5m

[ghorrocks]

I agree that using the roughness model to model large features such as buildings may be inappropriate. I do not know as this is not an area of my expertise. I would do a literature search and look at the basic equations to see if the wall roughness equations are still valid at this larger scale.