[Tom12345]

Hi all,
I want to simulate the effect of golf ball dimples on a car spoiler in different angles and different velocity.

I created a wind tunnel with velocity Inlet and pressure Outlet(relative Pressure =0 Pa) and a Wall boundary.

The aim is to reduce the drag force of the spoiler. The golf ball effect works with very high turbulence above the dimples, but I don't have many experience with the different turbulence models in ANSYS (I use 18.0 Workbench), so I don't really know, which turbulence model I should use.

Can anyone help me? Is the option "High (Intensity =10%) fine? Or should I use a "k and Epsilon" model?

Thanks in advance

[ghorrocks]

More turbulence = more drag. Why do you think increasing turbulence will reduce drag? The reason dimples on a golf ball reduce drag is because a golf ball in flight just happens to lie just on the laminar flow side of the laminar-turbulent transition, and the turbulence generated by the dimples triggers the flow to go turbulent. In a sphere the drag coefficient just on the turbulent side of the transition point is less than just on the laminar side. And as you increase the turbulence from there the drag just gets higher.

So the only time dimples will help is when your wing is right on turbulence transition, AND the laminar flow does not stay attached but turbulent flow would stay attached if turbulent. In all other conditions the dimples will just increase drag.

So I am saying it is unlikely to work

Back to your question - how do you model this:

Directly modelling the dimples will be very challenging as that will require LES. That is unlikely to be practical unless you are a CFD expert with a supercomputer handy. You can approximate the dimples with source terms on the face but this will be very approximate. You suggest increasing the turbulence levels at the inlet, this is another approach which is simple but is even less realistic.

[urosgrivc]

Arent dimples there to triger the flow to go from laminar to turbulent so there is a clerly defined transition and wortices as smal as posible.
If there is no dimples flow behind the ball creates some kind of 3D von carman like wortex which pulls the bal left,rigt,up and down a bit.
I thought they were there because of more steady flight, as a plane ball shape is unsteady midair and wobles a bit, and in golf you want acuracy so problems like this are best eliminated.

Nice cfd focused video about it:
https://www.youtube.com/watch?v=GHOoZYhF6r4

I suspect puting dimples on the under side of the spoiler will decrese dovnforce as you dont want flow seperation on the underside, but would make les of a diference on the uper side
and what good is it if you manage to decrese the drag if you dont have dovnforce anymore than it is beter to just get rid of the spoiler if it doesent work and save some mass in the proces Drag to dovnforce ratio is what matters

about the simulation I think that wery fine elements will be needed near the dimples ant vith fine inflation layers to obtain Y+ of les than 1 for turbulence model I would go for SST at first and I suspect that maybe thransient will be needed if the flow will not converge in steady, hope you have lots of ram and cores otherwise you will somehow have to reduce the model size to just a small section of the wing otherwise you will lose resolution.

[Tom12345]

Originally Posted by ghorrocks

More turbulence = more drag. Why do you think increasing turbulence will reduce drag? The reason dimples on a golf ball reduce drag is because a golf ball in flight just happens to lie just on the laminar flow side of the laminar-turbulent transition, and the turbulence generated by the dimples triggers the flow to go turbulent. In a sphere the drag coefficient just on the turbulent side of the transition point is less than just on the laminar side. And as you increase the turbulence from there the drag just gets higher.

You are right, I just simplified it. Its not just the drag I want to improve, the combination with the drop is important. (less drag with almost same drop OR, what will happen more likely, almost same drag with more drop)

Originally Posted by ghorrocks

Directly modelling the dimples will be very challenging as that will require LES. That is unlikely to be practical unless you are a CFD expert with a supercomputer handy. You can approximate the dimples with source terms on the face but this will be very approximate.

What is the problem with Large Eddy Simulation? Long simulation duration? Or the setting?

And what about the other turbulence models?

The simulation is for a student project, so it should not be too complicated.

[Tom12345]

Originally Posted by urosgrivc

...Drag to dovnforce ratio is what matters

Right, that is what I want to improve, read the last answer

Originally Posted by urosgrivc

about the simulation I think that wery fine elements will be needed near the dimples ant vith fine inflation layers to obtain Y+ of les than 1 for turbulence model I would go for SST at first and I suspect that maybe thransient will be needed if the flow will not converge in steady, hope you have lots of ram and cores otherwise you will somehow have to reduce the model size to just a small section of the wing otherwise you will lose resolution.

Ok, thank you for the answer, it is just a student project I want to do on my laptop, so not too much ram and cores :/

SST is a combination of the k-Epsilon and k-Omega models, isn't it? Where do I set to this in Workbench 18.0? In the set menu of the Inlet, I dont have SST or k-Omega models, there are just "high (Intensity= 10%)", "k-Epsilon", "k-Eddy Viscosity Ratio"...

[urosgrivc]

If you dont know where to set boundary conditions and models jet => than LES is probably way too complicated for you at this moment, you realy have to have years of knowladge to use more comlicated models like les as there is a lot of factors you will probably get wrong at first, plus laptop is a problem, LES is a transient model which will be computationaly wery demanding for the hardware so I think that there is no meaning in waiting a week for laptop to end the job just to find out that results are foolish becouse simulation was not set up the wright way it can be moths or years before a cfd novist can actualy use this software in the proper way not to mention expert stuff.

I am speaking from my own experience I am using CFX for about 6 years now last 3 on a daily basis, i hawe used LES like 3 times mainly just to test it out a bit not to get actual results.
I dont mean to scare you, but CFD is hard to get it wright from the begining.

Just try a normal wing without dimples, learn on it than try to add them. You will observe that to get good results for a wing like structure is hard because of flow separation and fluctuations you will find convergence dificult and I think this is why glenn mentioned LES, which I agree is the proper way to do this job.

And be ALWAYS objective and sceptical obout the results you are geting you need to test and check your simulation if the numbers you are geting as a result have meaning and are independant of other stuf like BCs, mesh... It can take time before one is comfortable with results and simulation setup

[ghorrocks]

I am no expert on golf ball dimples but my understanding is that the turbulence tripping causes the flow to stay attached further around the ball compared to laminar flow, and that reduces drag from the additional pressure recovery. This is why a dimpled golf ball goes further than a smooth one - less drag. I do not think your accuracy comment is correct

What is the problem with Large Eddy Simulation?

Difficult, technical validation and setup; massive grids are required; tweeny weeny time steps; results in massive run times and requires big computers. Definitely not to be considered for a student project.

Personally I think your project is too complex for a student project. Turbulence enhancement from dimples is a very challenging topic.

To make it tractable the only way I can suggest is to use a turbulence source term on the dimpled face. You are going to have to do some research to determine what is a suitable quantity of turbulence to add at the dimples.

You could use the turbulence transition model in CFX to handle the turbulence enhancement, but this is quite dodgy. It was not designed to handle dimpled surfaces so I would not trust the accuracy of its results. But it does mean you do not need to determine the intensity of the turbulence source term.

[urosgrivc]

Well a lot of papers actualy state that magnus efect is enhanced on the ball because of dimples and that the drag (is incresed in some papers and decresed up to 50% in others) but the ball goes further as it has aditional lift from rotating backvards.
So I am not convinced and sure what is the main driving mechanism. It is definitly an interesting topic

Glenn do you think that if there is a lot of dimples and these are relatively small that this could be simulated by a rough wall (equivalent roughnes height).

[ghorrocks]

Yes, a rough wall is one way of simulating it - provided the dimples are small and there are lots of them as you say. Getting it accurate, so that the applied roughness is equivalent to a given number of dimples, might be challenging. But all these approaches are challenging.

[Qkarl]

hey thomass, Im currently doing simulation regarding the dimples on golf ball too. May i know is there any outcome from you right now?

[Qkarl]

hi glenn, can you give me some guide on yplus calculation? I am confused with it right now. Thanks in advanced

[ghorrocks]

Please start a new thread for a new question.