Surface Roughness
Hello to all.
I am trying to accurately simulate fluid flow through a "real" pipe fitting and surface roughness is part of that.
I am given from the pipe manufacturer a single value for roughness of different pipes in dimensions of distance. i.e 0.0028 inches.
I see that there are several options for surface roughness of materials but units are now shown which complicates matters since there is more than one value I can enter. Where do I go to get full definitions and units of these input options so I can correctly enter the data?
Thanks
Ken Martin
Select Language
Free 30-Day Trials
Feedback
Questions? Ideas? Problems?
Rough wall functions
Configuring the rough wall functions is not a precise procedure.
First I would suggest you run your simulation with smooth (default) walls as a baseline for comparisons. In doing so you need to ensure that your y+ values on your walls are within the limits 30-300 (assuming you are using wall functions, default). y+ is a function of the flow conditions (primarily the Reynolds number) and the distance of the first mesh cell center to the wall. For a given flow condition you will need to run a simulation to generate the y+ values and then use the Accuracy tool to control the cell sizes at the walls to get your y+ values within bounds.
The rough wall functions are configured by:
Cs - Tuning parameter (0-1)
Ks - Sand-grain roughness height (m), where smooth = 0m
To calibrate your simulation you will need well regarded results or experimental data for rough walls.
Assuming your pipe roughness value to be equivalent to sand-grain roughness (it may not be, you will have to check with the original data) then you will need to convert it to meters and set Ks. Then use Cs to tune your simulation against your rough wall baseline data.
Between a smooth wall and rough wall simulation you should see only second order differences, i.e., there should only be fractional differences (if any) in your results.
Surface Roughness
Thanks for the very helpful points. I had opted to using the kABL wall function for the surface roughness. It asks for a single dimension in inches (since I have converted using the to the IP system since the values make more sense to me)
I did as you recommended. There is pressure drop data for a 12 inch round radius "smooth" elbow posted by ASHRAE for which gives a C value for various turn radii.
I generated a 12 inch round 90 degree elbow with 6 feet of ducting on both ends so as to take into account inflow and outflow effects created by the elbow in the turn. Using the pressure result I can see down the duct on both sides of the 90 degree elbow where it was no longer making an effect and narrowed the results bars to just so show the pressure drop associated with the elbow (a small amount went up the inlet duct and a larger amount when down the outlet duct as would be expected)
The program results were slightly lower than the given measured values so I adjusted the kABL wall function dimension to a larger value until the calculated and the given values where the same for pressure drop across the elbow. The duct surface kABL value that worked was found to be 0.005 inches. (which makes sense as a smooth metal surface roughness value)
I am hoping that now that I have established a roughness factor for the kABL wall function that I can use the same value for any kind of ducting arrangement of the same material.
I haven't yet but will model some other types of given duct fittings to see if the surface roughness of 0.007 inches will work in other conditions.
Thouhts?
Ken
Use U Rough Wall Function
Your geometry configuration sounds good with the positioning of your inlet and outlet.
However, I would recommend you use the U Rough Wall Function, rather than the k ABL Wall Function. The reason being is that ABL stands for Atmospheric Boundary Layer. The roughness lengths for ABL are generally of the order of 1-10 meters. Also it is intended for use in conjunction with the Inlet->Type = ABL.
The next version of Caedium will cater for the U Rough Wall Function using units dependent length scales, so you will be able to enter values directly in inches. In the mean time multiple your inches value by 0.0254 to convert to meters.
To make sure your results are accurate you also need to control the Scalar Fields->y+ values on your walls to be between 30-300. y+ is dependent on the flow conditions and the first cell height spacing. You will need to run a series of iterations for these values to settle down. Once they do will get an idea of how much smaller your cells closest to the walls need to be. Use the Accuracy tool to control the size of elements on your walls. I suggest you focus only on the values in the region locally around the elbow.