Influence of wall roughness and viscous total layers thickness on pressure drop

Using SimScale Fluid Flow / CFD
1

Hello,

I am doing a simulation to estimate the pressure drop through a pipe (length = 25 cm, diameter = 9.5 mm, fluid = water, velocity at the inlet = 1.5 m/s).

I chose to run a fluid dynamics analysis with an incompressible fluid and a steady-state flow with a k-omega SST turbulent model.

The meshing operation I am using is a tetrahedral meshing with prismatic viscous layers.

In order to analyse the influence of the total layers thickness and the influence of the wall roughness on the pressure drop, I performed two different studies.

1st study

In order to test different reasonable layer thickness, I used the following equation:

layer thickness = ly_plussqrt(74)*Re^(-13/14)

where l is the length scale (here the pipe diameter), y_plus lies between 30 and 300 and Re is the Reynolds number

I ran the simulation with y_plus = 30, 50, 100, 150, 200 and 300.
I did not activate the wall roughness in the simulation parameters.

The pressure drops I got are:

y_plus = 30; delta_p = 6.33 mmHg
y_plus = 50; delta_p = 6.89 mmHg
y_plus = 100; delta_p = 7.78 mmHg
y_plus = 150; delta_p = 7.89 mmHg
y_plus = 200; delta_p = 8.40 mmHg
y_plus = 300; delta_p = 8.58 mmHg

I am wondering how can I know which y+ is the optimal one and why delta_p rises in this way with the increase of y+.

2nd study

In order to analyse the influence of the wall roughness on the pressure drop, I used the same model than previously with y_plus = 30, and I activated the wall roughness.

The roughness constant is 0.5 and the different roughness height tested are:
0.0000013; 0.00001; 0.0001; 0.001.

The pressure drops I got are:

roughness height = 0.0000013; delta_p = 6.26 mmHg
roughness height = 0.00001; delta_p = 6.32 mmHg
roughness height = 0.0001; delta_p = 6.33 mmHg
roughness height = 0.001; delta_p = 6.30 mmHg

It seems that there is not a significant influence of the roughness height on the pressure drop. Moreover, the pressure drop value obtained without activating wall roughness is the same than with roughness height = 0.0001.

I have two questions:

  1. When we don’t activate the wall roughness, does simscale set a standard value of roughness height and roughness constant? If yes, what are these values?

  2. Is there an order of magnitude of roughness height for which I am supposed to detect an influence of roughness height on pressure drop?

I hope I am clear. If not, do not hesitate to ask me further details.

Thank you.

1 Like
2

Hi @corwave

Very interesting study!

For the first part, here is my possible explanation to the increase of pressure drop with the increasing y+:
Consider that with a higher y+ value, we have less resolution to capture the presence of the turbulent boundary layer. This would mean that the flow behaves “more laminar” with higher y+ values, which would increase the pressure resistance of the pipe.

From my understanding, by introducing wall roughness to the walls we are enhancing the generation of the turbulent boundary layer. With more intense turbulent layer, the flow will “slide” on it, creating the well-known turbulent profile. At that point, even if the layer becomes a bit thinner or thicker, the overall influence on the flow profile will be minimal, thus the pressure drop will not change much.

Now about the questions:
1) does simscale set a standard value of roughness height and roughness constant?
If you do not activate the roughness with a Custom boundary condition it will not be modeled.

2) Is there an order of magnitude of roughness height for which I am supposed to detect an influence of roughness height on pressure drop?
This would be quite case-dependent. Of course for low viscosity flows the influence of wall roughness will be bigger than for more viscous fluids. I would look for the ranges of roughness heights in the literature. (it is not my core area of focus)

Best!
Pawel

1 Like
3

Thank you very much for your answers. As you advised, I am going to check in the literature if I can have more details about roughness.

Best.

4

Hi,
i don’t know if this is the right place but i don’t want to open a new thread for such a stupid question.
I can’t find the option where i can activate the wall roughness in the simulation parameters. Can you please help me?
Thanks

5

Hello @ddaum,

That is a valid question, especially that you need to use Custom BC to get access to these functionality.
The “standard” SimScale conditions are simplified for ease of use. At the same time you can get access to all BC types using Custom type, but with great power comes great responsibility! You simply need to specify all the BC types for all the fields required for the simulation, and the configuration needs to make sense.

Here is a picture of a BC setup for a no-slip rough wall.
This is a setup for Incompressible, turbulent (k-omega-SST), steady state simulation:

image.png417×822 128 KB

  • velocity all to zero
  • pressure gradient to zero
  • turbulent kinetic energy (k) to Wall function and enable roughness
  • turbulent dissipation (omega) to Wall function

Hope this helps!

Best
Pawel

4 Likes
6

Hi @psosnowski ,

thank you very much, especially for the screenshot. I read in the documentation that it had to be somewhere under Custom BC / Custom type, but I searched for hours and just didnt see it.

Is it possible to get any further information about the roughness model? I am familiar with the sand roughness (here called roughness height) but what is the roughness constant?

Best

7

Hi @ddaum

The contact allows you to fine-tune the roughness model that you are using to the reference data that you have.
For example, you have the data of pipe flow made of the material of interest.
You make a simulation, set up the expected roughness height and run the job with roughness coefficient of 0.5,
If the results do not match your data, you play around with roughness coeff until you find the match.

With this setup you can go to your desired “new” simulation setup.

Best,
Pawel

8

A post was split to a new topic: Atmospheric Boundary Layer (ABL)

9

@psosnowski

I’m bit struggling to Get Right Y+. I’m performing a few Simulation Test to see the Y+ effects on the Roughness.

So, For the simulation performance, I’ve added the Roughness height 0.01 m. So From my understanding, the First Layer thickness or the Estimated wall distance has to be twice or thrice more than the Roughness Height.

So What If we failed to obtain that Distance(Twice bigger than than Roughness Height). what will be the effects on the results in terms of the velocity or pressure Head?

The Reason I’m asking. As Y+ depend on the few main factors Shear Velocity, Friction Reynolds number, Boundary layer length and the Free stream Velocity. When I tried to calculate the Estimated wall distance (By Considering the Y+ In between 30 to 300), I’m getting low first layer thickness, Which is not even close to the Roughness Height.

Few Question before moving forward.

  1. What If we fail to obtain the Y+ In between 30 to 300, What Would be the Solution to overcome this issue?

  2. Is it necessary to Include Y+ or Create Layer in order to Detect the Roughness over the Surface, What if we don’t use the layers and add the roughness directly is my result will change and are those results would be accurate?

  3. I’ve few confusion about how the OpenFOAM works in terms of Roughness and Y+ , It would be grate if Someone can guide through any form of documentation or notes?

In addition to this. to overcome this I tried two different simulation Please provide your thought over it.
—I’ve decided to use Relative LAYER as True. and Add the Final thickness as the Ratio of the Final Layer Thickness/ Level Grid Size
—So here I tried to the Grid Size should be approximately Equal to the Final layer thickness and thats how I created the layers.
The point I’m making here, Can I use the similar method and from the reverse calculation, I can estimate what will be the Appropriate wall distance which is twice more than the Roughness Height.

Basically, If you use relative layer sizing as true then we can actually estimate the Final Layer thickness from the refinement level. So we can add the appropriate Final layer thickness and from this value, we can see what is the Estimated Wall Distance or First layer thickness is. I’m not sure whether is the right way to calculate the relative layer would be grate if someone can let me know whether its an appropriate way to do it or not.

Again the main objective here, I want my First layer thickness twice more that the Roughness Height, But from the Y+ calculation, I can certainly see It’s not possible If I consider the Y+ in between 30 to 300.

I hope you got the question It would be grate if someone can help me understand this.

Sumedh

10

Hi @sumedhsurve!

Let me do some research on this to see what we can do with OpenFOAM to overcome those problems. Will get back to you as soon as I know more.

Best,

Jousef

11

@jousefm thank you very much for your response

12

When you make your First Layer thickness >twice the Roughness Height, what y+ values are you seeing?

13

@DaleKramer @jousefm
It’s more than 500. and that not valid It should be less than 300

14

So, if you make first layer thickness to get into y+ 30-300 what ratio of 1st layer to roughness have you?

15

image
image1874×555 67.7 KB

@jousefm @psosnowski
you check the Image above. I’ve created my own Y+ calculator. So at the Right hand side you can see all the values I’ve use Y+ 200 and according to this My First Layer thickness is 0.010792 and My roughness Height is 0.01 m

So In order to catch the roughness your First Layer thickness has to be twice or thirse bigger than the roughness height. and my ratio is somewhere around 1.0792.

Let me know if you need any additional information.
SUmedh

16

I use calculated meshing parameters for layering as only a rough guide (especially for relative layers).

With the results of a converged simulation, if you map your y+ to the surface geometry in the old post processor with the ‘internal mesh’ OFF, then you see what you really get for y+ and where you get it. If you set your parameters so that most of this visual surface mapping is between 30-300 (ignoring 1-30 areas because they are simply overrefined and simply cost computation time and not reduced accuracy) (also, I have been informed that there are reasons to even keep that to y+ 30-100 to increase accuracy), what is your first layer to roughness ratio in that majority area (sometimes I have measured this with a ruler on my screen of a 2d mesh clip as a ratio of a known refinement level cell in the same image , and I understand in Paraview you can measure things)?

17

It’s 1.0792

I’m using Old Postprocessor, In addition, I’ve calculated my Y+ in Paraview. but the same value.

What I’m thinking that there are some issues with the boundary layer length. (But not sure)

and also I want to know Is it really essential to make your roughness twice bigger than the roughness height. Because I performed few simulations by taking the less height which three times smaller than the First Layer thickness. and compared with those results whose roughness is not much small than the estimated wall distance but I haven’t see that much difference in results.

but I haven’t received any substantial answer on this that Is it really necessary to have the wall distance two times more than the roughness

18

But it looks like that is just a calculation…

So, if you have confirmed that in the results, then I am not too sure that you need to worry about it since you have too thin a 1st layer vs what you want. This should only be a cost of computation time and not results accuracy (I think :wink:, we may need confirmation on that) Just make sure that on your last layer, it blends into the further away volume cells smoothly (From my research that means at ~0.5 2D area ratio??).

Dale

19

@sumedhsurve,
i know its an “old” post, but maybe u find out anything new ?

I got the same “problem”, my roughness is 0,01-0,02 m and my first layer for a y+ around 30 should be 0,0011 m.

I don’t really know if this is a problem ?

Best regards
Kai

20

@sumedhsurve never showed us the actual surface mapping I asked for so I do not know what happened…

If you give me a project link or share, I don’t mind taking a look at the situation…

1 Like