[cfdnewb123]

I have difficulty replicating the general shape of the velocity profile for the rotating channel flow from DNS data. I introduce angular velocity=21 rad/s to "frame motion" and channel wall at absolute frame. I am using the Reynolds stress model with SOU for TKE & TDR.

The flow is driven by a constant pressure gradient dP/dx and because of that, I try using two different boundary conditions to match that.
1) For the first boundary conditions, I use translational periodic boundary conditions between inlet and outlet.
2) For the second boundary conditions, I use pressure inlet & pressure outlet with the gage pressure for the inlet set based on the constant pressure gradient.

https://imgur.com/gkTz2HU

Edited: However, both methods are to no avail as my velocity profile is quite off especially along the initial slope. I try increasing the velocity but the slope of the initial profile did not become less steep at all. Please help as I am so stuck at this

[vinerm]

The profile from your simulation is actually steeper (except with RSM, pressure inlet and outlet) than it should be, at both ends. You may be able to improve it with fine tuning of mesh, RSM model options, and numerical schemes, but I doubt that would be very fruitful. This might be the best you get with RSM. Though being a relatively expensive model, it is still RANS and suffers from many limitations. Except being able to model anisotropy, it can perform worse then EVMs under many conditions.

[cfdnewb123]

Thanks for the response. I am mistaken in my earlier post as I want the initial profile to be less steep not steeper.

Does it mean that I cannot get a better profile with RSM? May I ask what do you mean by "it can perform worse than EVMs under many conditions"? I did test the problem using the k-epsilon model but I still get the same result. I thought I should be able to get a good profile using RSM as the paper "Modeling rotational effects in eddy-viscosity closures" by Peterson uses a k-epsilon variant and his result seems quite good.

[vinerm]

What I implied is that RSM is not always better than EVMs, such as, or . The only addition it has is anisotropy. That is useful in certain circumstances, such as, highly swirling flows. But EVMs can predict better than RSM in most cases. So, try with SST or RNG ; you might get better predictions.

[cfdnewb123]

Thanks for the suggestion. I try using the two solvers that you suggested but the difference is not much and the initial slope of the profile did not become less steep. Is it because of my wall friction velocity? Because the papers that I read all prescribed wall friction velocity and I did not do anything like this as I do not know how to set this on Fluent.

[vinerm]

If a friction velocity has been prescribed in the work you are trying to validate against, then it is certain that you won't match the results until you do the same. To prescribe friction velocity, convert it into shear stress and apply at the wall.

[cfdnewb123]

Thanks for the response. I try applying the udf for shear stress but I have some issue with it. I narrow down the problem to the pressure gradient which I used as C_P_G(c,t)[0]. When I set pressure gradient with a constant, the udf works fine but not when I used the macro. Is there something wrong with the pressure gradient?

Edit: I resolved my udf, so sorry for the trouble.

[cfdnewb123]

Originally Posted by vinerm

If a friction velocity has been prescribed in the work you are trying to validate against, then it is certain that you won't match the results until you do the same. To prescribe friction velocity, convert it into shear stress and apply at the wall.

Does prescribing the friction velocity in terms of shear stress enforce the prescribed friction velocity? I try defining the shear stress tau in terms of friction velocity using this equation u*=sqrt{tau/rho}. However, the result does not produce the intended friction velocity of 0.141.

For non-rotating channel, the friction velocity was ten times than prescribed and it goes in the negative direction.
https://imgur.com/OljnC4Y

For rotating channel, the friction velocity are so off and both ends are not of the same magnitude. However, the velocity profile does looks better.
https://imgur.com/fdcH6wH

[vinerm]

What you are plotting is velocity-magnitude, not friction velocity. Friction velocity is given as



where is wall shear stress.

[cfdnewb123]

Thanks so much for the response and patience. I confuses velocity with friction velocity. The friction velocity is indeed what I prescribed. However, the result is still weird as shown in my earlier post. I think the issue is the pressure gradient which is supposed to be equal to -1 and drive the flow. I enforce this pressure gradient by setting the inlet & outlet to periodic.

However, when I plot my static pressure, it shows that the pressure is zero!
https://imgur.com/6OMdPrj
Or does the pressure gradient option under periodic condition refers to total pressure (=static + dynamic)? However, the total pressure is just a horizontal line which means pressure gradient is zero. Is there a way to enforce this fix pressure gradient between inlet and outlet?