CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Verification & Validation (https://www.cfd-online.com/Forums/openfoam-verification-validation/)
-   -   validating simpleFoam for mixinig of Non-Newtonian fluids (https://www.cfd-online.com/Forums/openfoam-verification-validation/165325-validating-simplefoam-mixinig-non-newtonian-fluids.html)

Ali Blues January 14, 2016 06:14

validating simpleFoam for mixinig of Non-Newtonian fluids
 
Hi,
I was wondering if anyone has validated OpenFOAM’s simplefoam solver for simulating mechanically agitated mixing of Non-Newtonian fluids, i.e. combination of MRF and Herschel-Bulkley model. I’ve been trying to compare the results against the following article (http://www.sciencedirect.com/science...5527010700387X) which I think serves as a very good validation test case as includes both experimental and numerical results (using Fluent), but unfortunately the results are not adding up, both with regards to power consumption (evaluated from computed impeller torque) as well as velocity profile. In the following link I’ve included the case for Re=21.5.
http://we.tl/IrtS3CQZI9
I used openfoam 2.4 for the computations, so simplefoam with MRF in system/fvOptions and Herschel-Bulkley in constant/transportProperties. I initially performed computations assuming a Newtonian fluid with an effective dynamic viscosity computed from available correlation equations (using equations 4 and 5 in the above article). Then switched to non-Newtonian model. The velocities more or less converge to 1e-6, but pressure stagnates earlier around 1e-4. The mesh was generated using snappyHexMesh.
For the power number (Np=P/(rho*N^3*D^5), where P=2*pi*N*T is power, N impeller speed in rev/sec, T is torque (N.m), D is impeller diameter) I obtain a value of Np=5.2 while it should be 0.57.
The axial velocity (fig 4.b) does show a similar profile but it is way off, so instead of being [-0.036 0.036] (after taking into account that the results are scaled by vtip=D/2*omega =0.18/2*8= 0.72 m/s in the figure), it is [-0.005 0.005].
I’ve gone over anything that I could think of that may cause this, but just can’t figure out where things are going wrong. I very much hope someone with experience would give their thoughts on this.

Thanks,
Ali

Ali Blues January 18, 2016 10:52

Hi,
Just a few more comments. I checked the definition of the Herschel-Bulkley model in OF 2.4 and it has been modified from the old "Fluent" like one (which has actually been commented out). Now if I take the converged solution from the new definition and carry iterations with the old definition, the simulations will diverge! why?

Also I took into account that the model coefficients are relative to the density. So that's taken care of.

In addition for measuring the force/torque, I included the fluid density in the force functionObject. So the combined values from viscous and pressure don't need to get multiplied by density.

Having said all that, again I would greatly appreciate it if some one would respond and if one has any suggestions as to why such discrepancies are present relative to the power number and velocity profile from the article (http://www.sciencedirect.com/science...5527010700387X). I'm sure there are quite a few of you OpenFOAM users out there which use MRF and non-newtonian models. If not Herschel-Bulkle, at least the power law model. Even in that case I haven't seen any validated test case. So I think this is a very good test of the validity of the implementation in OpenFOAM.

Thanks
Ali

Ali Blues January 20, 2016 06:59

Come on fellows :(
no one has any comments/suggestions?

Best
Ali

wurst March 31, 2016 10:05

MRF and Herschel Bulkley
 
I'm also working with simpleFOAM, MRF and Herschel bulkley models (for fruit purees). I wanted to have a closer look at your case but it is not available anymore. Would you mind making it available one more time and then I can have a closer look? Thank you!

Ali Blues April 6, 2016 06:31

Hi Jonas,
Thanks for the reply. Attached is the case:
https://we.tl/qBwqVNMJrd

I still haven't managed to resolve the issue. I've tried a case of a turbulent mixing of a power law fluid with Rushton impeller as well, since I thought there is perhaps something wrong with the Herschel bulkley model. But the results don't seem to be promising either. I've reduced the problem all the way down to a agitating newtonian flow in laminar range using a helical ribbon. But again results are off at least by a factor of 2.

if you agree, I think it is best if i send you the simpler case as well (laminar newtonian flow), and we start from there. I'll post the case sometime later this week. Thanks for the help

Best,
Ali

wurst April 6, 2016 09:56

Validation Case Coeffs (MRF Herschel-Bulkley simpleFOAM)
 
Hi Ali,

thank you for posting the case again! I also think it will be a great paper to validate the model with.

I cannot figure out how you derived the Herschel Bulkley-Coeffs for the transportProperties file. Where did you get rho from? Did you use the rheological properties from the paper?

Have you looked at radial and tangential velocity profiles, yet? I think it's a little trickier with OF...

Best,
Jonas

Ali Blues April 11, 2016 09:11

Hi Jonas,

Ya, the paper doesn't give all the property details, I found them in a master thesis from the same group: http://digital.library.ryerson.ca/is...ect/RULA%3A844
page 56 (68 of pdf) table 5.2 gives the respective rho and page 58, table 3 gives the mu0 for 1.5 wt % solution. They are of course not from the same sample, but if you check other quantities such as K and n and tauY, both report quite similar values. So I think it is safe to use the rho and mu0 from the MSc thesis.

Concerning the radial and tangential components, yes it doesn't seem to be a direct/quick way to get them. But in any case, the provided axial component at (r = 145 mm and θ = −45) already is sufficient for comparison, specially that it features a more complex profile than the radial and tangential.

Best
Ali

Tobi April 15, 2016 08:29

Dear all,

!!please!! ... this sub-forum is not for questions. Its for comparisons between OpenFOAM solvers (or new shared solvers) with analytical solutions or measurements as denoted by the sticked post:

Quote:

This forum is for comparing the solvers of OpenFOAM with experimental or analytical results. Thread titles should have the solver that is to be validated and a unique name for the case in the title (for instance "simpleFoam - backward facing step measured by Pitz and Dailey"). It is NOT a forum for questions for help - there are other Forums for that. Threads for which this Forum is not appropriate will be moved to a more appropriate thread to keep the forum focused on its topic. Apart from this the Forum is unmoderated
Thanks in advance,

blais.bruno July 8, 2016 09:28

I have worked a lot with mixing and OpenFOAM. If this is still an issue, send me your case and I can take a look at it ( I realize it has been a year now)

Quote:

Originally Posted by Ali Blues (Post 581728)
Come on fellows :(
no one has any comments/suggestions?

Best
Ali



All times are GMT -4. The time now is 18:22.