[aerospacious]

Hey everybody,


At the moment I am trying to validate some experiments to see if SimpleFOAM can accurately describe the boundary layer of a flat plate as I hope to move on towards airfoils including surface roughness soon.
Because my target application is incompressible flow (M<0.1) and all happening at room temperature I do not include any temperature effects as in the case of a Blasius flat plate.



I performed the 3 following cases:
1. RANS + k-omega-sst turbulence model in low Reynolds form (many points for y+<1)
2. RANS + k-omega-sst turbulence model in high Reynolds form (model the viscous sub region) (nut -> nutkRoughWallFunction)
3. RANS + k-omega-sst turbulence model also in high Reynolds form but including surface roughness (nut -> nutkRoughWallFunction)

There are some things that I noticed when looking at the results that I don't get:

• Comparing the resolved (low Re) and unresolved (high Re) cases I notice that the unresolved cases show clear laminar, transitional and turbulent regimes when looking at the coefficient of friction (C_f). The resolved case on the other hand does not. Why is this as I would assume the fully resolved case to experience transition as well.

• Comparing the two unresolved cases both transition at the exact same Reynolds number, even with their difference surface roughness. Is this a criterion that can be modified? Because it transitions at 5.0e5, which is quite optimistic when I compare it to experiments as well as for my future application.

• Lastly I have the feeling that there is a mismatch between the displacement & momentum thickness with the coefficient of friction. The experiments I use for validation (a MSc. thesis that performed experiments on flat plates with varying roughness) seem to have much higher coefficients of friction (~30% higher). The reason for this, is due to the fact that the displacement and momentum thickness seem not developed enough over the flat plate. Therefor the Re_theta is not comparible with that from the experiments in my opinion. Is the k-omega-sst algoritm know for not accurately describing the momentum thickness?

I would really appreciate some suggestions as to what is happening.
Is k-omega-sst model bad at estimating the true boundary and momentum thickness? Or is it too optimistic when computing the skin friction coefficient? Or are there other factors that I'm not taking into account that I should've taken into account?


Thanks a lot in advance!

Timo

[syavash]

Originally Posted by aerospacious

Hey everybody,


At the moment I am trying to validate some experiments to see if SimpleFOAM can accurately describe the boundary layer of a flat plate as I hope to move on towards airfoils including surface roughness soon.
Because my target application is incompressible flow (M<0.1) and all happening at room temperature I do not include any temperature effects as in the case of a Blasius flat plate.



I performed the 3 following cases:
1. RANS + k-omega-sst turbulence model in low Reynolds form (many points for y+<1)
2. RANS + k-omega-sst turbulence model in high Reynolds form (model the viscous sub region) (nut -> nutkRoughWallFunction)
3. RANS + k-omega-sst turbulence model also in high Reynolds form but including surface roughness (nut -> nutkRoughWallFunction)

There are some things that I noticed when looking at the results that I don't get:

• Comparing the resolved (low Re) and unresolved (high Re) cases I notice that the unresolved cases show clear laminar, transitional and turbulent regimes when looking at the coefficient of friction (C_f). The resolved case on the other hand does not. Why is this as I would assume the fully resolved case to experience transition as well.

• Comparing the two unresolved cases both transition at the exact same Reynolds number, even with their difference surface roughness. Is this a criterion that can be modified? Because it transitions at 5.0e5, which is quite optimistic when I compare it to experiments as well as for my future application.

• Lastly I have the feeling that there is a mismatch between the displacement & momentum thickness with the coefficient of friction. The experiments I use for validation (a MSc. thesis that performed experiments on flat plates with varying roughness) seem to have much higher coefficients of friction (~30% higher). The reason for this, is due to the fact that the displacement and momentum thickness seem not developed enough over the flat plate. Therefor the Re_theta is not comparible with that from the experiments in my opinion. Is the k-omega-sst algoritm know for not accurately describing the momentum thickness?

I would really appreciate some suggestions as to what is happening.
Is k-omega-sst model bad at estimating the true boundary and momentum thickness? Or is it too optimistic when computing the skin friction coefficient? Or are there other factors that I'm not taking into account that I should've taken into account?


Thanks a lot in advance!

Timo

Hi Timo,


I haven't done RANS simulation for flat plate, but couple of things come to my mind.


1-make sure you are applying correct boundary conditions for k and omega. When I was using OF 2.3.1, I applied a constant value and a wall-function at the wall for k and omega, respectively. I do not remember if nut was defined there.


2-compare your results with DNS of flat plate. There are good researches and boundary-layer profile database on the web. See the links below:


https://torroja.dmt.upm.es/ftp/blayers/


https://www.mech.kth.se/~pschlatt/DATA/


3-compare the local velocity profile of the boundary-layer with the references. This way, you may get a sense of whereabouts the problem lies by examining the velocity profile.


4-the last but not the least, make sure you have applied the correct boundary conditions which would allow a proper spatial growth of the boundary-layer.


I hope these points will help you find the problem.


Regards,
Syavash